2021 Scientific Research in School by Barker College

Science Science

Scientific Research in School Volume 3 Number 1 September 2021

Extension Journal

Honor HonorNon NonHonores Honores

Mission Mission

An Anglican community inspiring An Anglican community every learner inspiring every learner every experience every experience every day every day

Vision Vision

To be a leader in Christian education To be a leader in Christian education that is characterised by a global vision that is characterised byhope a global vision that inspires that inspires hope

Values Values Commitment Commitment Compassion Compassion Courage Courage Integrity Integrity Respect Respect

We acknowledge the Dharug, Darkinjung, Wonnarua and Yolŋu peoples who are the traditional custodians of the land on which Barker College, Darkinjung Barker, Ngarralingayil Barker and Dhupuma Barker stand. We pay respect to the Elders past, present and emerging of the Dharug, Darkinjung, Wonnarua and Yolŋu nations and extend that respect to other Indigenous people within the Barker College community.

Senior Editor Dr Matthew Hill Creative Direction Mrs Susan Layton Dr Matthew Hill Research Supervisors Dr Alison Gates Dr Katie Terrett Dr Matthew Hill

About the Scientific Research in School Journal When the New South Wales Education Standards Authority announced a new course “Science Extension” to commence in 2019 we were thrilled that there was an opportunity for a formally-assessed capstone experience in Science for our students. From the perspective of the Barker Institute it was an exciting chance to support students doing academic research, alongside other subjects such as History Extension, Music Extension and English Extension 2.

Where many capstone project courses fail is the at final step of the research process – dissemination. Research is not merely the process of conducing an investigation and writing a report, but sharing it with the wider community so that people can learn, critique, have other student researchers at multiple schools build on the projects published. I am so glad to be able to publish this journal each year now celebrating 44 articles each representing genuine contributions to science.

Dr Matthew Hill Director of The Barker Institute

Introduction

Science seeks to give reason to our perceptions in a way that is coherent and beautiful.

The wonderful student authors in this journal and their outstanding research articles are shining lights in a difficult period for Sydney with heightened COVID-19 in the 2021 winter. We are living in a time where many are trying to understand what is going on around them; what the right response is, and, how they can contribute to thriving amidst and beyond a global pandemic. How proud am I that Barker continues to develop research-literate young adults who have completed an apprenticeship in science academia. The world needs them and their voice will be heard. While this academic journal is a testament to a wonderful Science Department at the School, the skill and guidance of the three Science Extension research supervisors is unparalleled. In addition, the unsurpassed opportunities for Barker students to excel in their chosen fields are boundless. However, the highest recognition must go to the twenty one student researchers who have asked questions of the universe, analysed the results and shared them with the world. Bravo to all in this amazing program! I hope you enjoy and learn from these articles just as I have done.

Mr Phillip Heath AM Head of Barker College

Science Extension invites students to become practising scientists as they think deeply about the history, philosophy and methods of science.

I am proud of and grateful for the vibrancy and determination of students and staff in the Science Extension program. This rigorous and challenging course is breaking ground in preparing students for future scientific endeavours. It is astounding to consider the depth and breadth that our projects cover in only a few short years of offering this subject. The opening of the new laboratories in 2021 has provided a dedicated space for Science Extension students to develop their projects and to gain experience in collaborative laboratory research. We look forward to the fruit that this will bear over many student cohorts. I add my congratulations to each of these students for these excellent reports. Many of our Science Extension students intend to pursue careers in science and we wish them every success with their future studies.

Mrs Virginia Ellis Head of Science

Dr Matthew Hill Director of The Barker Institute

Dr Katie Terrett Chemistry Teacher

Dr Alison Gates Agriculture & Science Teacher Assistant coordinator STEAM

Undertaking proper scientific research takes curiosity, capacity and commitment. These students found research questions that they were passionate about and worked hard to implement the scientific research process to answer them. They demonstrated a high capacity for scientific thinking, inquiry and communication resulting in these high-quality journal articles. It was a joy and a privilege to work with these fine young scientists. We are incredibly proud of them and we are excited to share their work with you in this journal.

Contents Part 1: Biology and Envionmental Science The relationship between methylglyoxal concentration in Manuka Honey and the growth of certain bacterial pathogens Juliet Iraninejad

Static Magnetic Fields … The Solution to Sustaining Future Global Food Production? Ailish King

Growth of SCOBY in Kombucha Using Different Tea Types Madison McIntyre

Are we flushing money down the fish tank? An investigation of the extent to which commercially available microbes effectively colonise aquaria Anais Tomlinson

To bean or not to bean... Will Vanilla grow more microbes after different processing techniques? That is the question… Nathan Finikin

Part 2: Physics The Boundary of Chaos: An Investigation into the Length Ratio Dependent Chaotic Dynamics of a Planar Double Pendulum Harry Breden

How small can you go? How Iron powder can mitigate inefficiencies of a transformer’s soft iron core Francessca Buffa

Bias of a Coin Toss Tomo Bower

The influence of text structure on learning Physics Caleb Swanson

Could you be any more random? A study of written and spoken Random Sequence Generation Manxi Zhang

Contents Part 3: Chemistry Synthesis of the 3,4-methylenedioxy analogue of Pyrimethamine Thomas Abbott

Concentration of Allicin in Garlic Brianna Lollback

Influence of ultraviolet light on the stability of allicin in aqueous garlic extract James Wilson

105

Concentration of Lycopene in Different Varieties of Tomatoes Charlie Scholefield

115

Effect of time after harvest on chlorophyll concentration in spinach leaves Kyle Scholtz

121

As simple as making hot dogs: Using analogies to teach limiting and excess reagents in high school chemistry Cleo Christie-David

127

Synthesis of Pyrimethamine Jess Samuelson

135

Antibacterial activity and degradation pathways of methallyl isothiocyanate Ollie Bacon

143

Optimising the synthesis of the 4-iodo analogue of pyrimethamine Maxine Wu

151

Scientific Research in School Volume 3 Issue 1 2021

10 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Biology & Environmental Science Biology is the study of living things, astounding in the breadth of opportunities it offers for student investigation. From human biomechanics to agricultural production, our students have explored questions from across the spectrum of the discipline.

We are thrilled with the efforts of our students that undertook research in the domains of Biology and Environmental Science. It is deeply inspiring, as teachers, to watch students dig deep and produce work of this standard. Ailish’s project on the influence of a magnetic field on plant growth is the epitome of a well defined and controlled experiment executed well. On account of her meticulous and thorough approach she has produced excellent results. Similarly, Juliet’s project about the relative anti-microbial properties of Manuka honey offers a fascinating insight into the marketing of premium agricultural products. Anais’ project about aquarium supplements was also driven by a consumer question: are these additives worth what we pay for them? Madi’s project on the yield of SCOBY from different kombucha substrates presented many laboratory challenges and it was a testament to her creativity and resilience that she was able to pivot her project in response. Nathan’s interesting project about vanilla was another really well designed project that explored how processing, storage and packaging can lead to microbial contamination on vanilla (but doesn’t have to). In our third year of teaching Science Extension we have been blessed by the opening of the new laboratory facilities at school. The breadth and depth of our student projects is testament to the generous resources of the school. We commend our students for their outstanding achievements and wish them well as they take the next step in their science careers.

Science Science Extension Journal • 1

Scientific Research in School Volume 3 Issue 1 2021

2 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

The relationship between methylglyoxal concentration in Manuka Honey and the growth of certain bacterial pathogens. Juliet Iraninejad Barker College Manuka honey is known to be a more effective antimicrobial agent than honey without a ‘Manuka Honey’ label (Sutherland, 2020). An increase in antimicrobial effectiveness of certain types of honey is caused due to the acidity, sugar content, hydrogen peroxide content and other components such as the floral source the honey is derived from. This research paper aims to explore the relationship between methylglyoxal (MGO) concentration in manuka honey and the growth of certain bacterial pathogens. Producers of Manuka honey claim that the Manuka honey with the highest concentration of MGO will have the most potent antimicrobial activity, obtaining a larger zone of inhibition during this experiment. In an attempt to test the proposed hypothesis, two different brands of Manuka honey with the same varying MGO concentrations are inserted into an agar plate using the well diffusion method and tested against two bacteria, Staphylococcus epidermidis and Escherichia coli. The study concluded that there was a significant difference in the zones of inhibition against S.epidermidis, showing that the Manuka honey with the lower concentration of methylglyoxal (30+) had an increased zone of inhibition, rejecting the null hypothesis. Furthermore the E.coli-inoculated agar plates showed that there was no significant difference between the zones of inhibition, accepting the null hypothesis. Therefore, this experiment denies the promises made by the brands as a decrease in MGO rating increases the zones of inhibition produced. Literature review Antibiotic resistance is a global health and development threat and a significant medical challenge for the modern age (WHO, 2020). The rapid consumption of antibiotics has led to the allowance of certain species of bacteria to develop adaptations allowing them to become resistant to said antibiotics. Honey is a natural substance that has been known to possess certain antimicrobial properties inhibiting much of bacterial growth. As well as this, honey was discovered to inhibit the growth of an already antibiotic-resistant strain of bacteria, Methicillinresistant Staphylococcus aureus (MRSA) (Alvarez, 2014) establishing the substance’s importance in an environment with increasing amounts of antibiotic resistant bacteria. Honeybees use the nectar of flowers to produce a natural substance known as honey. This naturally occurring antimicrobial product has been used by humans since ancient times, dating back to 5500 years ago, to prevent

infection and soothe wounds (Sarmar, 2017). The antibacterial properties present in honey are mainly caused by four factors, acidity, osmolality, hydrogen peroxide content as well as other components. Acidity Honey’s pH is approximately 3.5 - 4.5 making the substance reasonably acidic (Alvarez, 2014). The acidity of honey is mainly generated due to the production of gluconic acid (Figure 1), providing the honey with a lower pH. “Gluconic acid is produced from glucose through a simple dehydrogenation reaction catalysed by glucose oxidase” (Ramachandran et al., 2006, p1). This acidity in itself enables the honey to eliminate bacteria that cannot grow in harsh, acidic environments. Osmolality Honey has a high sugar content, meaning that it has an extremely high osmotic force due to the concentration of sugars within the substance. Honey’s composition is roughly made of:

Science Extension Journal • 3

Scientific Research in School Volume 3 Issue 1 2021

• • • •

38.19% Fructose 31.28% Glucose 17.20% Water 13.33% other compounds and substances

Figure 3: Action of honey sugar molecules on microbial growth. (Source: Lay-flurrie, 2008) Figure 1: The structural formula for Gluconic acid (Source: Ramachandran et al., 2006)

Considering honey is made of approximately 70% sugar, when encountered with a bacteria, the water will be displaced from the inside of the bacteria - killing it and relocated across the cell membrane, towards the honey, as shown in Figure 2. This is the reason why it is famously known that ‘honey does not spoil’, because the organisms responsible for spoilage - can not grow on the surface of fully-ripened honey (Molan, 2015). However, if one was to dilute this honey, there would be a higher likelihood of bacterial and fungal colonies forming on the substance, as the sugar content lowers and the osmotic pressure is reduced (Molan, 2015). Hydrogen peroxide Hydrogen peroxide is generated by an enzyme called glucose oxidase (see Figure 3), this enzyme is activated when honey is diluted (Molan, 2015). This substance, hydrogen peroxide, is known to be the major antibacterial factor in honey, however is inactivated by heating honey and exposure to UV radiation (Molan, 2015).

Other components It has been discovered that the antimicrobial properties present in honey did not originate from the bee species that processed and produced it, but instead these antimicrobial properties originated from the flowers that store the nectar bee’s use to make said honey (AlvarezSuarez et al., 2014). Therefore, the medicinal variance of certain honey’s is also extremely dependent on the floral source the honey is derived from. For example, Manuka honey is a monofloral honey derived from the Manuka tree, “Leptospermum scoparium, of the Myrtaceae family, which grows as a shrub or a small tree throughout New Zealand and eastern Australia” (Alvarez, 2014, p421). This particular honey has been recognised by researchers to contain a higher level of bacterial prevention than other honey’s from differing floral sources. Consumer claims Manuka honey is also known to contain a special compound called methylglyoxal. This compound, scientists believe, is directly responsible for the specific antibacterial activity that Manuka honey retains (Atrott et al., 2009).

2 Figure 3: The reaction scheme of glucose oxidation catalysed by glucose oxidase (Source: Suzuki et al., 2020, p2)

4 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Because of Manuka honey’s antibacterial properties, it has been widely marketed to have a more potent antimicrobial effect dependent on the concentration of methylglyoxal within the honey. Therefore, as the antimicrobial effect is supposedly increased, the brand of the honey can increase the expense of the product. The consumers trust that as they pay a premium for Manuka honey they are getting increased health benefits, therefore, the experiment tests two different brands of honey of identical methylglyoxal concentration to see if the customers are getting honey with corresponding antimicrobial effects. This results in a test of the consumer claims of the two brands of Manuka honey. Bacterium used Bacteria are universally classified based on Gram staining to be either positive or negative. These responses are connected to the thickness and composition of the cell wall of the bacterium (Sutherland, 2020). Staphylococcus epidermidis is a topical gram-positive skin bacteria that has the capabilities to cause opportunistic infections. This bacteria does not produce aggressive toxins (Otto, 2009) and when combined with the bacteria being mostly present on the surface of human skin, this bacterium is a safe choice for the school laboratory. Escherichia coli is a rod-shaped gram negative bacterium that is known for living in the gut of humans. Some strains of this bacterium are known to be regularly living within the human complex, providing their existence to be favourable in the breaking down of food compounds. However, there are some strains of this bacterium that are responsible for causing a variety of common diseases, mostly associated with the consumption of contaminated food and water.

Scientific research question Is there a significant difference between the concentration of methylglyoxal in Manuka Honey and its inhibitory effect on bacterial growth, specifically against Staphylococcus epidermidis and Escherichia coli?

Scientific hypothesis There will be a significant difference between the inhibitory effects of Manuka Honey with varying methylglyoxal concentrations. Specifically, the Manuka honey with the highest concentration of methylglyoxal will have a larger zone of inhibition than the Manuka honey with a lower concentration of methylglyoxal.

Methodology Cultures of K12 Escherichia coli and Staphylococcus epidermidis were purchased from a laboratory supplier, cultured in a nutrient broth and incubated for 48 hours. Using aseptic technique, the surface of the nutrient agar plates were inoculated with a prepared broth of S. epidermidis and a prepared broth of E.coli with a sterile spreader to form a lawn. 14 plates were prepared, 5 of E.coli and 9 of S. epidermidis (as shown in Table 1). The increase in testing for S.epidermidis was largely due to the focus on skin-related bacteria. Wells, 8.0 mm in diameter, were cut from the culture media using the reverse end of a sterile glass pasteur pipette, and then filled with the appropriate honey. Table 1: Table representing the conditions of each nutrient agar plate. Plate Brand of MGO Bacteria number Honey value (+) 1 S.epidermidis Ab’s 30 2 S.epidermidis Ab’s 30 3 S.epidermidis Ab’s 100 4 S.epidermidis Ab’s 100 5 S.epidermidis Capilano 30 6 S.epidermidis Capilano 30 7 S.epidermidis Capilano 100 8 S.epidermidis Capilano 100 9 E.coli Ab’s 30 10 E.coli Ab’s 100 11 E.coli Capilano 30 12 E.coli Capilano 100 13 (Cont.) S.epidermidis No honey 0 14 (Cont.) E.coli No honey 0

The plates were incubated at 37ºC and observed after 24 hours for clear zones of inhibition around the wells. The diameter of the zone was measured using a vernier calliper and the results were recorded in a table. Plates were observed again at 48 hours and the zones of inhibition showed no differentiation to that of the initial measurements.

Results Table 2 shows all measurements taken of the zones of inhibition, presented in a single table for easier viewing. Tables 4 and 5 show the results produced from the ANOVA calculator in relation to the Tukey HSD test for S.epidermidis and E.coli. Tables 3 and 5 show the one way ANOVA of the four treatments (each table representing a different bacteria) inputted, drawing specific attention to the p-value collected.

Science Extension Journal • 5

Scientific Research in School Volume 3 Issue 1 2021

Raw data

Table 2:Raw data collected from measuring the zone of inhibition around the wells Honey MGO Zone of Inhibition data (mm) brand (units) Capilano

30+

S. epidermidis 1

100+

AB’s

30+

100+

Control

2 2 2 2 0 0 0 0 3 5 5 5 2 3 3 2 0

S. epidermidis 2

0 0 0 0 0 0 0 0 6 6 4 4 2 1 2 2

E. coli 1

3 2 1.5 2 1 2 1 1 2 2 1 1 2 2 3 2 0

ANOVA calculations for Staphylococcus epidermidis treatments

Table 3: Tukey HSD values for Staphylococcus epidermis treatments Treatment Q stat P-value Inference pairs A vs B

3.4915

0.0873248

Insignificant

A vs C

13.0931

0.0010053

**p<0.01

A vs D

3.9279

0.0450609

*p<0.05

B vs C

16.5846

0.0010053

**p<0.01

B vs D

7.4194

0.0010053

**p<0.01

C vs D

9.1652

0.0010053

**p<0.01

Table 4: One-way ANOVA of the 4 independent treatments Staphylococcus epidermidis FP-value Source SS DoF MS stat Treatment 100.5 3 33.53 51.0 1.8E-11 Error 18.3 28 0.65 Total 118.9 31

ANOVA calculations for Escherichia coli treatments Table 5: Tukey HSD values for Escherichia coli treatments Treatment Q stat P-value Inference pairs A vs B 3.1569 0.1698022 Insignificant A vs C 2.2549 0.4186357 Insignificant A vs D 0.451 0.8999947 Insignificant B vs C 0.902 0.8999947 Insignificant B vs D 3.6079 0.1015578 Insignificant C vs D 2.7059 0.2733552 Insignificant

6 • Science Extension Journal

Table 6: One-way ANOVA of the 4 independent treatments – Escherichia coli FP-value Source SS DoF MS stat Treatment 2.79 3 0.932 3.03 0.07 Error 3.68 12 0.307 Total 6.48 15

Discussion The research hypothesis predicts that the MGO rating would be broadly predictive of the zone of inhibition (or antimicrobial effect) of the honey. That is, a honey with a high MGO rating should be more antibacterial (have a larger zone of inhibition) than one with a lower rating. S.epidermidis In relation to the proposed hypothesis, a significant difference was displayed for S.epidermidis as demonstrated by Table 4, but was not the significant difference that was expected. The unexpected difference shows that the Manuka honey with the lower concentration of methylglyoxal was the most effective when it came to inhibiting the growth of the bacterial pathogen, S.epidermidis. For this section of the results, the p-value corresponding to the F-statistic of one-way ANOVA is lower than 0.05. This suggests that one or more treatments are significantly different (this result is shown in Table 3). Furthermore, because the p-value (1.7628e-11) is lower than the alpha value (which was set to 0.05) the researcher can conclude that the null hypothesis is rejected and the initial hypothesis is accepted. This results in the statement that, because of the 0.05 alpha value, there is a 95% chance that there is a significant difference between the zones of inhibition between the two different concentrations of methylglyoxal present in each honey sample. However, when the results were calculated, the values generated show that, although the alpha value for this experiment is 0.05 (as is the standard for scientific investigation), the p-value discovered is less than 0.01. These results conclude that although the alpha value 0.05 was used throughout the experiment and because the p-values are less than 0.01, there is a 99% chance that these zones of inhibition are significantly different. Due to the p-value being incredibly low (1.7628e-11) the researcher is able to conclude that the result produced from the ANOVA calculator is reliable as it is far lower than 0.01, increasing the certainty of the significant difference.

Scientific Research in School Volume 3 Issue 1 2021

E.coli From the p-value generated via ANOVA calculations for the E.coli treatments in Table 6, there is the acknowledgement that this value is higher than 0.05, which is the chosen alpha value for this experiment. From this, the researcher is able to conclude that the null hypothesis is accepted and the initial hypothesis is rejected, meaning that the treatments are not significantly different. Furthermore these experimental results show that, due to there not being a significant difference between the honey products used, both honey products were less antimicrobial against E.coli. The most likely reason for this occurrence is that honey may not be as antibacterial for gut organisms (E.coli), than it is for topical bacteria (S.epidermidis). This result then assumes that honey increases in antimicrobial properties if used on topical bacteria present on the surface of the skin rather than on internal bacterial organisms. Significance of results The results obtained from this experiment show that although it is expected that the zone of inhibition should increase along with the increased concentration of methylglyoxal, the opposite results were collected as a result of the experiment. This result could be caused due to experimental error or poor consumer claims. The two brands used in this experiment were AB’s and Capilano. These brands place a consumer claim that with an increase in the methylglyoxal concentration of a honey, the antimicrobial activity is also increased. Due to this promise different brands are able to charge premium prices for higher concentrations. However, if the honey with the lower concentration increases antibacterial activity, as shown in the experimental results, then this is the honey that should have an increased price. Furthermore, the Manuka honey sourced from the Capilano brand shows that there is no significant difference between the varying concentrations of methylglyoxal. This increases the concerns surrounding the brand as Manuka honey should show a significant difference to say the least. This significant difference however, was that the lower concentration of methylglyoxal seemed to retain more productive antimicrobial properties. This draws the conclusion that marketing strategies are not always (or very rarely) based on scientific evidence. Future improvements In papers mentioned previously, there was the notion of diluting the honey to activate the hydrogen peroxide

within the honey substance. However, when this is done, it manipulates the intended research question as there is mainly a focus on honey, without manipulation. Due to this however, there is the chance that the high viscosity that honey retains may make the substance more difficult to work with, thus making the investigation retain more experimental errors. Honey is a substance also known for its high viscosity, making it a reasonably thick substance and difficult to manipulate. Furthermore, during the experiment there was a challenge in dispersing small and equal amounts of honey into wells that were 8.0mm in diameter. Unfortunately, the viscosity of the honey rendered it difficult to be put into a pipette and inserted into the well, so instead, a small amount of honey was picked up using the needle-like end of a glass pipette, and the honey fell into the well. However, by using this inaccurate method, it was noticed that too much honey had been inserted into the well and spread across the edges of the well due to overflow. This affected the experimental accuracy and therefore rendered the experiment unreliable as the measurements of honey were not equal or consistent. Due to limited supply there are a few sources of error that could be a result from the source of honey being from a single jar. For example, honey is known to “not have a shelf life” due to the inability of organisms to grow on the surface, as mentioned previously. However, considering consumer behaviour, it is more likely that the less expensive honey (Capilano) would be bought more frequently when compared with the more expensive honey (AB’s). Therefore, it is understood that that particular brand of honey could have initially been on the shelf of the store longer (prior to purchasing) than that of the Capilano brand. Furthermore, due to this increase of shelf life, there is the possibility that the honey could decrease in MGO effectiveness over time or be exposed to UV degradation, potentially damaging the MGO effectiveness.

Conclusion Honey has been used and valued by humans for centuries due to its medicinal qualities. These antimicrobial properties that honey retains has caught the attention of many researchers globally, attempting to discover the causative agent responsible for its natural medicinal properties. Honey has been discovered to retain medicinal qualities due to its acidity, sugar content (high osmolality), hydrogen peroxide content and other components such as floral source. Specifically in Manuka honey is a compound methylglyoxal (MGO). Producers of Manuka honey state that as MGO rating

Science Extension Journal • 7

Scientific Research in School Volume 3 Issue 1 2021

increases (methylglyoxal concentration) antibacterial potency increases.

the

To test this claim, an experiment was conducted measuring the zone of inhibition produced from Manuka honey with 30+ and 100+ MGO rating against S.epidermidis and E.coli. The study concluded that there was a significant difference in the zones of inhibition against S.epidermidis, showing that the Manuka honey with the lower concentration of methylglyoxal (30+) had an increased zone of inhibition, rejecting the null hypothesis. Furthermore the E.coli-inoculated agar plates showed that there was no significant difference between the zones of inhibition, therefore the null hypothesis was accepted. Therefore, this experiment denies the promises made by the brands as a decrease in MGO rating increases the zones of inhibition produced.

Acknowledgements I would love to acknowledge Dr Alison Gates for her constant supervisory support during all areas of my experiment. I would also like to extend this gratitude to Dr Matthew Hill for guiding me through a suitable results layout in the results portion of the scientific report. Furthermore, I would also like to thank Barker College and the lab staff for providing me with the adequate equipment needed in order to complete this experiment.

References Alnaimat, S, Wainwright, M & Al’Abri, K 2012, ‘Antibacterial potential of honey from different origins: a comparison with Manuka honey’, Microbiology, Biotechnology and Food sciences, pp. 1328–1338.

8 • Science Extension Journal

Alvarez-Suarez, JM, Tulipani, S, Romandini, S, Bertoli, E & Battino, M 2009, ‘Contribution of honey in nutrition and human health: a review’, Mediterranean Journal of Nutrition and Metabolism, vol. 3, no. 1, pp. 15–23. Atrott, J & Henle, T 2009, ‘Methylglyoxal in Manuka Honey – Correlation with Antibacterial Properties’, Czech Journal of Food Sciences, vol. 27, no. Special Issue 1, pp. S163–S165. Berg, HC 2012, E. coli in motion., Springer. Hammond, EN & Donkor, ES 2013, ‘Antibacterial effect of Manuka honey on Clostridium difficile’, BMC Research Notes, vol. 6, no. 1. Lay-flurrie, K 2008, ‘Honey in wound care: effects, clinical application and patient benefit’, British Journal of Nursing, vol. 17, no. Sup5, pp. S30–S36. Mandal, A 2019, Manuka honey could be useful in treating cystic fibrosis lung infection, News-Medical.net, viewed 14 June 2021, <https://www.newsmedical.net/news/20190530/Manuka-honey-could-be-usefulin-treating-cystic-fibrosis-lung-infection.aspx>. Otto, M 2009, ‘Staphylococcus epidermidis — the “accidental” pathogen’, Nature Reviews Microbiology, vol. 7, no. 8, pp. 555–567. Ramachandran, S, Fontanille, P, Pandey, A & Larroche, C 2006, ‘Gluconic Acid: Properties, Applications and Microbial Production’, Food Technol. Biotechnol., pp. 185–195. Suzuki, N, Lee, J, Loew, N, Takahashi-Inose, Y, OkudaShimazaki, J, Kojima, K, Mori, K, Tsugawa, W & Sode, K 2020, ‘Engineered Glucose Oxidase Capable of Quasi-Direct Electron Transfer after a Quick-and-Easy Modification with a Mediator’, International Journal of Molecular Sciences, vol. 21, no. 3, p. 1137. World Health Organization 2020, Antibiotic resistance, World Health Organization.

Scientific Research in School Volume 3 Issue 1 2021

Static Magnetic Fields … The Solution to Sustaining Future Global Food Production? Ailish King Barker College This study investigated whether the presence and intensity of a Static Magnetic Field (SMF) would enhance and increase the germination rate, height, and root weight of lettuce. This was carried out through having 2 treatment groups with a lower intensity SMF of 0.003 tesla and a higher SMF of 0.006 tesla and a control group with no SMF present. This study was conducted over 4 weeks. The data was analysed through initial observations and measurements and a One-way ANOVA test was carried out to determine if the results were statistically significantly. The results, for germination rate, height and root weight were all found not to be statistically significant therefore suggesting the presence and respective intensity of an SMF doesn’t affect or enhance the growth of a lettuce seedling. However, from the initial data, an enhanced difference can be observed amongst the different treatment groups, where the treatment group with the high SMF had the most enhanced measurements. Conclusively, these results did not align with the hypothesis or previous literature and studies, therefore suggesting that further, more specific research needs to be carried out in this field of science. Additionally, a more controlled environment and a larger sample size should be used in future studies of this investigation. Literature Review In accordance with the World Population Prospects (WPP), the United Nations (UN) is expecting for a 95% possibility that the population will exponentially expand from 7.7 billion in 2019 to 8.5 billion in 2030 and then towards 9.7 billion in 2050 (United Nations Department of Economic and Social Affairs, 2019) which in turn places an overwhelming demand on the agricultural sectors of the world to sustain food and water production to fulfil demands globally. In conjunction, an expected increase in global wealth over the next 30 years (Piesse, 2020) and therefore a shift in global diets, is expected to raise the demand for meat and dairy consumption, consequently requiring a greater amount of crop resources to be produced than other food products. As a result, the sustainable intensification of agricultural practices and the lifting of food production in underutilised regions of the world will have to be implemented as a way to sustainably produce enough food to satisfy the ever-growing population. This also places an emphasis on the idea that this intensification and exploring sustainable methods to accommodate for change should be implemented imminently Prior Research into the Effects of an SMF Research into the Geomagnetic Field (GMF) also referred to as the Earth’s static magnetic field (SMF),

and its inescapability as an important environmental stress factor overtime in conjunction with the ability for plants to respond to environmental stimuli, has allowed for a contemplation of its ability as an SMF to be a driving force in the growth, speciation, and diversification of plant species (Wu, 2019). Thus, this further has led to the notion that magnetic fields and the presence of such may be of positive consequence for plant quality and quantity. The existing literature and research into the reactions and perceptions in changes of plants in an SMF is sparse as a result of investigations being unsystematic and devoid of testable theoretical predictions (Maffei, 2014). However, as time progressed into the 1960s and onwards where technology and knowledge of science grew exponentially, the investigation into the use and effect of magnetic fields within an agricultural context emerged, where a magnetotropic was proposed, where an increase in auxin, a plant hormone which contributes to the elongation of cells in shoots, and therefore involved in regulating plant growth was found (Maffei, 2014). This was further supported by UJ Pittman in 1963, where he stated that ‘magnets were having a stimulating effect on plant initiation and growth within 48hours’ of placement. This research involved less systematic errors and more testable predictions in contrast to experiments conducted in the past.

Science Extension Journal • 9

Scientific Research in School Volume 3 Issue 1 2021

Based on this minimal research and the popular emergence of Precision Agriculture, a slim number of crop producers in the late 20th and early 21st century endeavoured onto the utilisation of magnets within irrigation water, known as Magnetic Water Technology (MWT) to increase the yields, and the quality/quantity of their crops (Saraff, 2020). Additionally, with further, specific research and real-life trials into the phenomenon of the use of magnetic fields for agriculture to further consolidate the pre-existing knowledge, this could be implemented into the underutilised areas of the world for the purpose and solving of increasing the crop yield to satisfy the growing global demand for food towards 2050 within a greenhouse agricultural context. Further Studies ‘Magnetism and Plant Growth, The Effect on Germination and Early Growth of Cereal Seeds’ from the Canada Agriculture Research Station in 1963 contributed to the construction of a hypothesis that the presence of a SMF will enhance plant growth which is illustrated in the root weight results, where the treatment group had results of up to 15mm in average primary root length, which was much greater than the average lengths achieved in the untreated groups. Pittman’s investigation also contributed to a more suitable method which could be carried out, where the notion of multiple replicates and random allocation of seedlings and pot placement were drew up upon for increased fairness and to decrease chances of bias. Further, this investigation measured root growth and plant height, which would be suitable dependent variables for the purpose of further investigation towards the effects and whether the use of SMFs in glasshouse agriculture is advisable. ‘The Effects of Magnetic Fields on Plant Growth and Health’ (2012) into SMF intensity affecting germination rate influenced this research project and in particular the hypothesis and methods. It deduced that the treatment groups under a higher SMF of 0.49 Tesla had higher germination rates and increased plant heights as it plays a role in increasing photosynthesis rates. His research further kept the growth period of 4 weeks, the soil type, and the allocation of water the same, all of which were variables to be kept the same in this project.

Scientific research question “How does the presence and intensity of a static magnetic field impact on the germination rate, height and root growth of a lettuce seedling?”

10 • Science Extension Journal

Scientific hypothesis The presence of a static magnetic field will have a positive effect on the germination rate, height, and the root growth of lettuce. The higher the intensity of the MF, the more of an effect it will have, where the rate of germination, height, and root growth of the lettuce will be enhanced/increased

Methodology Acetate Template Using scissors and a protractor, a 21cm diameter circle was drawn and cut out from a sheet of 2mm thick plastic. A 5cm diameter circle was then drawn using a protractor and cut using scissors in the middle of the 21cm diameter circle for the purpose of the magnets. 3 lines horizontally numbering 4, 2, 4 with 2cm gaps in between each hole, were drawn going left to right across the 21cm diameter plastic as seed holes. These were cut out where they had a 1cm diameter. Refer to figure 1.

Figure 1: Birdseye view of the acetate template

Control Group Using a 10cm long spade, 16 heaped scoops of Hortico All Purpose Blend Potting Mix was poured into each ceramic pot. The pots were ¾ full. The acetate template was placed on top of the soil. Using a metal skewer, the holes on the acetate template were poked 15cm deep into the pot for the seedlings to be placed in. 2 of Mr Fothergill’s Australian Yellow Leaf Lettuce seeds were randomly allocated into each hole using a skewer for precision to release and push them down 15cm deep. The holes were then covered with the soil already in the pot using a smaller, 5cm long spade. Refer to figure 2 and 3 for reference.

Scientific Research in School Volume 3 Issue 1 2021

down 15cm deep into the soil. The holes were then covered with the soil previously dug out again with a 5cm long spade. Refer to figure 4 and 5 for reference.

Figure 2– cross-section of the control group set-up.

Figure 4: cross-section of the treatment group set-up.

Figure 3: birds-eye view of the control group set-up

220mls of water was evenly poured into the pot after being planted. 25ml of diluted Seasol Complete Garden Health treatment was poured evenly into the pot. These steps were repeated for the other two pots in the control group. The control group acts as a comparison for the two treatment groups with an SMF present. This control group with no SMF present will act as the usual conditions a plant typically experiences as it grows in a greenhouse situation. Static Magnetic Field Groups Using a 10cm spade, 16 heaped scoops of Hortico All Purpose Blend Potting Mix was poured into each ceramic pot. The pots were ¾ full. The acetate template was placed on top of the soil. Using a 6cm spade, the 5cm diameter hole in the acetate template was emptied 18cm deep into the soil. The Boyle 22mm x 4mm neodymium magnets (3 magnets for group 2 at 0.003 TESLA collectively and 6 magnets for group 3 at 0.oo6 TESLA collectively. Tesla was recorded using a EMF recorder iPhone app) were wrapped in two layers of Glad clingwrap to reduce the chance interference with the surroundings and were then placed carefully into the hole. The soil that was previously dug out was used to fill the hole using a 5cm long spade. Using a metal skewer, the holes on the acetate template were poked 15cm deep into the pot for the seedlings to be placed into. 2 of Mr Fothergill’s Australian Yellow Leaf Lettuce seeds were randomly allocated into each hole using the skewer for precision to release and push them

Figure 5– birds-eye view of the treatment group set-up.

220mls of water was poured into each pot evenly. 25ml of diluted Seasol Complete Garden Health treatment was poured evenly into the pot. These steps were repeated for the other 4 pots in the SMF groups with the appropriate amounts of magnets for each treatment group. All pots were then randomly assorted, as seen in figure 6 and 7, into a Naturallife 69 x 49 x 160cm 4 tier greenhouse. The greenhouse was placed in direct sunlight and all plants were exposed to 10 hours sunlight per day. All pots were watered with 220ml of water every second day. 25ml of diluted Seasol Complete Garden Health treatment was evenly allocated to each pot at the start of every second week. All pots were exposed to a humidity of 56% in the greenhouse, measured with a hygrometer.

Science Extension Journal • 11

Scientific Research in School Volume 3 Issue 1 2021

floor in direct sunlight in an attempt to keep the standardised conditions the same. Then, through observing the pots, the number of sprouts were counted and recorded into a table. Conclusion Germination Measurements On the day after the conclusion of the experiment, the ceramic pots were removed from the greenhouse and placed onto the ground in direct sunlight in an attempt to keep the standardised conditions the same. The number of sprouts in each ceramic pot were then counted and recorded into a table.

Figure 6 random allocation of pots in the greenhouse set-up

The differences which would later be inputted into an ANOVA test and possibly a post-hoc test to measure whether there was a statistical significance in difference would be measured by the difference between the number of sprouts at the conclusion of week 4 minus the number of sprouts at week 0. Measuring Root Weight The root weight was measured by uprooting the lettuce sprouts from each pot the day after the experiment. The roots were cut about 1cm above into the stem to avoid including the leaves and rest of the stem into the measurement. The total weight of all individual sprouts in each pot was measured using an electric balance and recording it into a table. The total root weight of each pot was recorded. The average root weight in each group will then be processed through a possible ANOVA test and a posthoc test if necessary to determine whether there was a statistical significance in difference.

Figure 7: random allocation of pots in the greenhouse set-up

Measuring Height In this experiment, height is measured at the conclusion of the experiment and classified as the measurement of the stem when pulled gently up straight.

Analytical Method Within this experiment, germination, root weight and height were measured.

The sprouts were gently held up to their maximum height, and using a 10cm ruler, the height was measured and recorded for each pot and written down in a table.

Measuring Germination In this experiment, germination was measured as the number of sprouts weekly on the 7th day and at the conclusion of the experiment. Weekly Germination Measurements On the 7th day of the week, each ceramic pot was brought out of the greenhouse and placed consecutively on the

12 • Science Extension Journal

This was completed for each pot in each group. After all sprouts in each pot for each group were measured and recorded, the total amount of the height measurements for each group collectively were averaged. The averages of each pot in each group will then be inputted into an ANOVA test and possible a post-hoc test to determine whether the results are statistically significant different.

Scientific Research in School Volume 3 Issue 1 2021

Results After the conclusion of the experiment, the results of germination, root weight and height were placed into tables labelled 1-5. Height Table 1 shows the average height of the lettuce seedlings in each pot, followed by the average lettuce seedling height for the particular treatment group, all calculated to 2 significant figures. Table 1: The average accumulative height of the sprouts in each pot, and then the average height for each group

Treatment Group

Pot Number

1 2 3

Average Height – Pot (Cm) 1.3 2.6 0.0

1 2 3 1 2 3

1.8 1.9 3.3 3.8 3.3 1.3

Average Height – Group (Cm) 2.4

1.9

3.2

A One-way ANOVA was completed to see if there was a significant difference between the growth heights of any of the three different magnetic field strengths. The results indicated there was no significant difference between the growth heights (F=1.2800, p=0.3444). Therefore, this experiment suggests that magnetic field strength does not impact the growth height of lettuce seedlings. Germination Tables 2-4 show the germination rate over 4 weeks and the differences observed. Table 2– The germination rate of the lettuce seedlings for the control treatment group.

Germination rate (no. of sprouts) WEEK 0 WEEK 1 WEEK 2 WEEK 3 WEEK 4 Difference (B-A)

POT 1

POT 2

POT 3

0 0 0 1 1 1

0 2 4 5 9 9

0 0 0 0 0 0

Table 3: The germination rate of the lettuce seedlings for the second, lower SMF treatment group with 3 magnets present per pot

Germination rate (no. of sprouts per week)

POT 1

POT 2

POT 3

WEEK 0 WEEK 1 WEEK 2 WEEK 3 WEEK 4 Difference (B-A)

0 0 0 2 3 3

0 0 2 2 2 2

0 1 2 3 4 4

Table 4: The germination rate of the lettuce seedlings for the third, higher SMF treatment group with 6 magnets present per pot presented.

Germination rate (no. of sprouts per week) WEEK 0 WEEK 1 WEEK 2 WEEK 3 WEEK 4 Difference (B-A)

POT 1

POT 2

POT 3

0 2 3 5 6 6

0 3 5 7 8 8

0 1 1 1 2 2

A One-way ANOVA was completed to see if there was a significant difference between the average germination rates of any of the three different magnetic field strengths. The results indicated there was no significant difference between the average germination rates (F=0.4135, p=0.6789). Therefore, this experiment suggests that magnetic field strength does not impact the germination rates of lettuce seedlings. Root Weight Table 5 shows the accumulative root weight of each pot, followed by the average accumulative root weight for each group. This was calculated to 3 significant figures. Table 5 The accumulative root weight of each pot, followed by the average accumulative root weight for group

Treatment Group

Pot Number

Weight – Pot (g)

Average Weight – Group (g)

1 2 3

0.125 0.992 0.000

0.372

1 2 3 1 2 3

0.231 0.127 0.364 0.943 1.30 0.232

0.241

0.825

Science Extension Journal • 13

Scientific Research in School Volume 3 Issue 1 2021

A One-way ANOVA was completed to see if there was a significant difference between the root weight of any of the three different magnetic field strengths. The results indicated there was no significant difference between the root weight results, (F=0.1.4055 p=0.3158). Therefore, this suggests that magnetic field strength does not impact the root weights of lettuce seedlings. There may have been an interference with the availability of sunlight distributed amongst the three levels in the greenhouse, which may have affected or skewed results by limiting the growth of certain pots and therefore groups. There may have been a systematic and random error, due to the small size of the lettuce seeds as well, where there may have been more than one seed dropped per hole which may have given different pots a growing advantage over another.

Discussion The key research question was to investigate the effect of an SMF’s presence on plant growth, in terms of germination, height and root weight. Throughout the experiment there was 2 treatment groups that experienced the presence of an SMF, and one group with received no SMF as a control. In the endeavour to test the hypothesis – that the presence and increased intensity of an SMF will enhance plant growth (seen through increased germination rate, height, and root weight) – would suggest that if any of the groups with an SMF present showed signs of significantly greater growth, then the presence and greater intensity of the SMF can be seen to have a positive effect on plant growth and would therefore require more exploration in this field of science and agriculture. All measurements including germination rate, height and root weight showed there was no significant differences between any of the pot replicates in each group as seen in the results section. The ANOVA tests conducted for each area of measurement showed than none of the three groups or their individual pot replicates were significantly different to one another as seen in the results section. The initial data however suggests that there is a difference when an SMF is present. This can be seen: Height The initial data from the height area of measurement can draw that treatment group 3 which consisted of the higher SMF (0.006 TESLA) had the highest average lettuce seedling height (3.2cm) than the 2nd treatment group and the control treatment group (which the average seedling heights of 1.9cm and 2.4cm) respectively.

14 • Science Extension Journal

Germination The initial data suggested that treatment group 3, with the higher SMF of o.oo6 TESLA, had a higher germination rate with each pot having a larger number of sprouts in differences (6,8,2), than the 2nd treatment group and the control treatment ((3,2,4) and (1,9,0)). This initial data concludes that the higher SMF had the highest germination rate, followed by the control and the second treatment group with only a difference of 1 sprout. Root Weight The initial data shows that treatment 3, with the higher intensity of an SMF (0.006 tesla) had the highest average root weight (0.825g) than the treatment two with a lower intensity SMF (0.003 tesla) and the control treatment (0.241g, 0.372g) respectively. From this initial data, it can be seen that the treatment group 3 had the highest germination rate, height measurement and root weight which can contribute to the presence and the higher intensity of an SMF enhancing overall plant growth. Also, in each area of measurement, the control treatment which was absent of an SMF had higher germination rates, height measurements and root weight than the 2nd treatment group which had a lower intensity SMF present. These initial data results align with Massimo Maffei (2014) who explored ‘Magnetic Field Effects on Plant Growth, Development and Evolution’ and concluded that lower intensity SMFs will not affect germination rates and dry root weights in comparison to higher SMF intensities, that affected both in an enhanced way. However, some pots across the different treatment groups can be deemed to be affected by the random allocation in the green house, where the pots located on the middle level of the green house were seen to have lower germination rates and height measurements than the rest of the pots. In the future, this should be minimised as it can skew results and not present values that are due to the presence or lack of an SMF. These initial data results can therefore not be claimed as there was not a clear consistency as a result of the disadvantage seen from being randomly allocated to the middle level of the greenhouse. Furthermore, within all treatment groups for all areas of measurement there is no evidence that there is a beneficial element for the presence of an SMF. Therefore, there is much more study needed to draw any firm conclusions around this data as previous literature has suggested an improvement, however this data suggests the opposite.

Scientific Research in School Volume 3 Issue 1 2021

In further studies, the environment should be as standardised and controlled as possible, with the trials ideally taking place in a laboratory glasshouse where temperature, air pressure and sunlight distribution, such as the use of artificial light can be more carefully monitored. This can be difficult to achieve at a highschool level, but in this investigation all variables such as humidity, temperature, water allocation etc. were kept the same to the best ability. Also, more precise seed collection and allocation equipment may be used in the future to avoid inconsistent amounts of seeds being allocated to different pots, where an unfair advantage or disadvantage may occur. Further, again, this trial is hard to draw conclusions from due to the small sample size and the notion that its statistical results do not align with prior literature and their promising results. In this investigation, there was approximately 90 lettuce seeds. In a more a confident investigation to either support or reject the hypothesis, around 200-300 lettuce seeds would be used to increase the sample size.

Conclusion This research project explored whether the presence and intensity of an SMF would enhance plant growth. Data was collected through 3 areas of measurement, germination rate, height, and root weight. The analysis of the data was carried out through differences and averages and was further analysed using multiple OneWay ANOVA tests and the observations seen in the initial data. The data concluded that the presence and the level of intensity of an SMF had no effect on germination rate, height or root weight which did not align with the hypothesis that all 3 of these would be enhanced or previous literature such as Uj Pittman’s investigation (1963) or Massimo Maffei’s project (2014) that SMFs cause an enhanced plant growth result. This further suggests that this data may be insufficient, and that there were random errors within this project. Therefore, thus concludes that more research needs to be done within this field of science and agriculture. Future projects regarding this phenomenon and research should have larger sample sizes and more monitoring and certainty of controlled variables and standardisation.

Acknowledgements I would like to thank Dr Alison Gates from Barker College Hornsby for her ever-helpful assistance in the development of this research project idea and data collection methods.

I would also like to thank Dr Matthew Hill from Barker College Hornsby for his helpful assistance in the research into static magnetic fields and his thorough suggestions with statistical analysis. Finally, I would like to thank my parents and friends for their constant support and help with my project through the construction and data collection processes.

References Ali, Y, Samaneh, R & Kavakebian, F 2014, ‘Applications of Magnetic Water Technology in Farming and Agriculture Development: A Review of Recent Advances’, Current World Environment, vol. 9, 6 December, pp. 1-5, viewed 20 February 2021, Cdnsciencepub.com, Ahvaz Jundishapur University of Medical Sciences, Golestan Blvd., Ahvaz, Iran Bertea, C, Narayana, R & Agliassa, C 2015, ‘Geomagnetic Field (GMF) and Plant Evolution: Investigating the Effects of Gmf Reversal on Arabidopsis thaliana Development and Gene Expression’, Journal of Visualised Experiments, 30 November, pp. 1-2, viewed 11 February 2021, US National Library of Medicine National Institutes of Health, NCBI Fu, E 2012, ‘The effects of magnetic fields on plant growth and health’, Young Scientists Journal, no. 11, viewed 13 December 2020, https://www.ysjournal.com/wpcontent/uploads/Issue11/The-effects-of-magnetic-fields-on-plantgrowth-and-health.pdf Maffei, ME 2014, ‘Magnetic Field Effects on Plant Growth, Development and Evolution’, Frontiers in Plant Science, vol. 5, 4 September, viewed 10 February 2021, https://www.frontiersin.org/articles/10.3389/fpls.2014.00445/full Piesse, M., 2021. Global Food and Water Security in 2050: Demographic Change and Increased Demand - Future Directions International. [online] Future Directions International. Available at: https://www.futuredirections.org.au/publication/global-foodand-water-security-in-2050-demographic-change-and-increaseddemand/ [Accessed 11 June 2021]. Pittman, U 1963, ‘Magnetism and plant growth i. Effect on germination and ea:rly growth of cereal seeds'’, Magnetism And Plant Growth, 11 February, pp. 1-5, viewed 19 February 2021, Cdnsciencepub.com, Shanghai Library UN DESA | United Nations Department of Economic and Social Affairs. 2021. World population projected to reach 9.8 billion in 2050, and 11.2 billion in 2100 | UN DESA | United Nations Department of Economic and Social Affairs. [online] Available at: https://www.un.org/development/desa/en/news/population/world -population-prospects-2017.html [Accessed 9 June 2021]. Wu, T 2019, Static Magnetic Fields, ICNIRP, viewed 18 2021, February https://www.google.com/search?rlz=1C5CHFA_enAU834AU83 4&ei=WsU0YK-5MsyV4EP2dmXwA8&q=5+basic+methods+of+statistical+analysis&oq =method+of+statistical+a&gs_lcp=Cgdnd3Mtd2l6EAEYATIHC AAQRxCwAzIHCAAQRxCwAzIHCAAQRxCwAzIHCAAQRx CwAzIHCAAQRxCwAzIHCAAQRxCwAzIHCAAQRxCwAzI HCAAQRxCwA1AAWABgoxVoAXACeACAAaoBiAGqAZIB AzAuMZgBAKoBB2d3cy13aXrIAQjAAQE&sclient=gwswiz&safe=active&ssui=on

Science Extension Journal • 15

Scientific Research in School Volume 3 Issue 1 2021

16 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Growth of SCOBY in Kombucha Using Different Tea Types Madison McIntyre Barker College Kombucha is a fermented beverage produced from the fermentation of sweet tea solutions by a microbial culture known as the symbiotic culture of bacterias and yeasts, or a SCOBY. The kombucha drink is becoming more popular for consumers due to its proposed benefits of increased probiotics and antioxidants, increasing heart and liver health and manage diabetes as well as proven anticarcinogenic, antihypertensive, antidiabetic, and hepatoprotective properties. The SCOBY also has industrial purposes such as a sustainable leather alternative or other fabrics, sustainable packaging or medical and culinary purposes. Making the SCOBY an important commodity and not only a bi-product of the kombucha fermentation process. The investigation conducted aims to determine whether the tea type used within kombucha has an effect on the growth of the SCOBY and the mass of the microbial mat produced. The experiment concluded that there is no significant difference between the tea substrates used (green, rooibos, black, white, oolong). However, this then broadens the possibilities for Kombucha production and the production of SCOBY grown commercially as an independent product of kombucha. Literature Review Kombucha is a drink produced through the fermentation of a sweet tea using bacteria and yeasts. Widely grown domestically for consumption, kombucha has become increasingly popular for its proposed health benefits of increased probiotics and antioxidants, increasing heart and liver health and manage diabetes as well as proven anticarcinogenic, antihypertensive, antidiabetic, and hepatoprotective properties. The SCOBY is an abbreviation of a symbiotic culture of bacteria and yeast, and is the microbial mass that is created during the fermentation process of lactic acid bacteria, acetic acid bacteria and yeasts. Literature relating to Kombucha/SCOBY: Multiple studies have shown the broad range of microbes within the kombucha SCOBY/tea fungus and their ability to vary between fermentations/cultures (Chakravorty et al., 2016; Coton et al., 2017; Reva et al., 2015) and especially between industrial uses that are not limited to kombucha fermentation (Lopez, Beaufort, Brandam, & Taillandier, 2014; Nehme, Mathieu, & Taillandier, 2008). Although the composition differs between cultures there are some main microbial components that are present in most cultures, including; Saccharomyces cerevisiae, Acetobacter xylinoides, Bacterium gluconicum, Acetobacter aceti, Acetobacter

pasteurianus, and Gluconobacter oxydans (Jayabalan et al., 2014). This could have had an effect on the results of the experiment through the reactions of different microbial compositions within different SCOBY starter cultures and therefore the result may not be common throughout all cultures. Research on microbes for sustainable industrial purposes has been a current topic as consumer views and priorities shift towards sustainability. Some examples of these industrial uses include multiple applications of the dried microbial mat including a leather alternative once flattened and dried or a biodegradable packaging. Kruk et al. (2021) and Domskiene, Sederaviciute, and Simonaityte (2019) explored these applications. This has an important relationship to the significance of my research as it applies the findings and data into a real world application. The growth rate of the SCOBY within kombucha has a direct relationship with the volume of tea solution. (Dutta, & Paul, 2019). This suggests there is an optimal ratio for growth of the SCOBY that should be aimed to be achieved, for industrial applications and within my experiment. Tea has been found to be the best substrate for Kombucha to date as Kombucha tea recorded the

Science Extension Journal • 17

Scientific Research in School Volume 3 Issue 1 2021

highest specific growth rate (Ahmed, Hikal, & AbouTaleb (2020) with studies being done to determine which concentration of sugar is optimal for microbial and antioxidant growth proving black tea with a sugar concentration of 26% has the greatest antioxidant growth (Vohra, Fazry, Sairi, & Babul-Airianah (2019) and allows for an estimated improved/optimal concentration for increased growth efficiency.

One litre of boiling water was added via a measuring cylinder to a 1500mL Pyrex beaker. 200g of white sugar was weighed and added to the boiling water along with a magnetic stirring pellet. 3 tea bags were suspended from the top of the beaker and the mixture was stirred on a magnetic stirrer for 2 hours until the water cooled and the tea infused. Cooled tea was refrigerated prior to use.

Temperature has a large effect on microbial diversity, a solution containing a SCOBY culture that is kept at a constant temperature of 30 degrees will have a much higher level of biodiversity within the culture than a replicate of the same culture kept at 20 degrees (StPierre, 2019) and this means that my experiment will have to be kept at the same temperature to keep the test results reliable.

40mL of the tea solution was measured in a measuring cylinder and added to a 100mL glass test tube. This was repeated to make a total of four test tubes. A piece of SCOBY was carefully cut to exactly 3.00g and added to each test and a latex balloon was stretched over the mouth of the test tube to provide a simple sealed chamber. The test tubes were labelled with sequential numbers (1=1 oolong tea, 2= black tea, 3 = 3 rubios, 4 = 4 green tea and 5=5 white tea) and each of the four replicates for each tea type were labelled A,B,C,D.

The kombucha fermentation process remains constant however significant markers such as sugar levels, PH, antioxidant properties and alcohol content change throughout the different tea types/solutions (Jakubczyk et al. 2020). Literature relating to experimental design: The method development process of my project was largely influenced by the project conducted by Wagner et al. (2013), where the effect of essential oils in tea types was measured on the microbial mass. The experiment conducted made the tea solution more acidic through the addition of 35mLs of vinegar, this step was not included within my experiment however could affect the final growth of SCOBY. The method allowed for an understanding of optimal quantities of ingredients (sugar, tea concentration ect.) and the separation of the solution into smaller trail batches.

Scientific research question Does the type of tea affect the final mass/growth rate of the symbiotic culture of bacteria and yeasts (SCOBY) in kombucha fermentation?

Scientific hypothesis The type of tea used in kombucha fermentation will change the rate of growth of the SCOBY as it provides different ingredients for the process of fermentation.

Methodology Teabags of five different tea varieties (Oolong, Rubios, Green, Black, White) were purchased from a supermarket along with a standard 2kg bag of white granulated sugar.

18 • Science Extension Journal

The entire test tube chamber was weighed on a scale using a precise and calibrated balance by taking a beaker and placing the test tube and balloon on the balance (see Figure 1). Results were recorded in a table. Each apparatus was re-weighed on days 4, 7 and 11 (the final day of the experiment). On the eleventh day of the experiment the SCOBY was carefully removed from the test tube and placed on the same scale (this time on a tared petri dish - see Figure 2). The weight of the SCOBY was recorded. The weight of any residu al tea that was left on the petri dish following removal of the SCOBY was recorded and subtracted from the microbial mass.

Results The results of the SCOBY mass are shown in table 1. There was no significant difference between the mass of any of the SCOBYs within the respective substrates. However, there are differences between the average dry weight difference of all the tea types

Scientific Research in School Volume 3 Issue 1 2021

was completed with an Anova (analysis of variance) test, where the P value of the data was discovered to be 0.0531, which is higher than the standard p-value of 0.05, as shown in Table 2. Therefore this data is statistically insignificant. Tukey post hoc tests revealed that the differences between each combination was insignificant with the exception of treatment D and E (Tukey HSC Q-statistic = 4.7702, p=0.029). However, the overall insignificance of the results requires the hypothesis that the type of tea used in kombucha fermentation will have an effect on the growth rate of the SCOBY has to be rejected and the null hypothesis must be accepted.

Figure 1: Apparatus for measuring mass of the contents of the test tube.

Table 2: One-way ANOVA of the 4 independent treatments – Escherichia coli PSource SS DoF MS F-stat value Treatment 0.952 4 0.238 2.994 0.053 Error 1.193 15 0.098 Total 2.145 19

Some possible reasons for the results of this experiment are that the oxidisation process, which is what differs black tea with the other types of tea used within the experiment, has no significant effect on the fermentation process or the growth of the SCOBY and therefore an insignificant result was achieved. The effect of oxidisation between black, green and rubios teas has shown that there is no significant difference in bacterial composition of the SCOBY within kombucha, however the yeast communities did show a difference between the tea types (Gaggìa et al., 2018), these composition differences have shown to have an insignificant effect on the mass of the microbial mat.

Figure 2: The weighing of the final mass of SCOBY outside the apparatus

(not statistically significant) as shown in Table 1. The replicates of the Kombucha solutions of all tea types lost weight through the experiment due to the chemical reaction that occurred to allow for fermentation by the microbes for their growth. However the SCOBY increased in mass when removed from the solution and weighed separately.

Discussion The statistical analysis of the data collected with the data used and treatment (k independent variable) groups to

Some key limitations of the experiment conducted was the time in which the experiment had to be completed, the kombucha fermentation process takes an average of 14 days, where the experiment was only run over a 11 day period meaning that a longer fermentation time may have an effect on the growth of the SCOBY. Another limitation was the precision of the scales when measuring the initial three grams of SCOBY, the precision was measured to 0.01 and therefore it cannot be ensured that all of the mass was exactly the same at the beginning of the experiment, however this has a reduced effect as the precision and scale were kept constant through the experiment. To improve the experiment the SCOBY could have been acclimated to the tea type weeks in advance to conducting the experiment, and then remeasuring the three grams of SCOBY for each replicate of the experiment from the acclimated and differentiated SCOBY mother cultures.

Science Extension Journal • 19

Scientific Research in School Volume 3 Issue 1 2021

Table 1: Mass data of SCOBY Tea type

Tu be

Mass (g)

Oolong Black Rooibos

Green White

Initia l

Day 4

Day 7

Day 11

Change

1A 1B 1C 1D 2A 2B 2C 2D 3A 3B 3C 3D 4A 4B 4C 4D 5A

82.53 83.41 82.74 83.75 82.09 82.68 83.38 82.97 81.57 82.69 83.5 82.95 81.33 82.17 82.75 83.46 84.14

82.41 83.29 82.49 83.63 82.02 82.62 83.32 82.91 81.49 82.61 83.41 82.87 81.27 82.08 82.7 83.39 84.07

82.38 83.26 82.46 83.58 81.97 82.56 83.27 82.86 81.45 82.57 83.38 82.84 81.23 82.05 82.65 83.37 84.03

82.34 83.24 82.43 83.55 81.93 82.54 83.25 82.84 81.44 82.56 83.34 82.81 81.21 82.06 82.66 83.36 83.99

-0.19 -0.17 -0.31 -0.2 -0.16 -0.14 -0.13 -0.13 -0.13 -0.13 -0.16 -0.14 -0.12 -0.11 -0.09 -0.1 -0.15

82.57

82.52

82.48

82.44

-0.13

82.81

82.76

82.71

82.69

-0.12

84.31

84.25

84.23

84.18

-0.13

Mean change

-0.218

-0.14

-0.1

-0.133

SCOBY weight

SCOBY change

4.2 4.58 4.71 3.94 4.47 4.34 4.2 4.31 4.22 4.09 4.6 3.94 3.99 3.47 4.28 4.12 4.81

1.2 1.58 1.71 0.94 1.47 1.34 1.2 1.31 1.22 1.09 1.6 0.94 0.99 0.47 1.28 1.12 1.81

4.79

1.79

4.29

1.29

4.66

1.66

Mean SCOBY change 3.6225

4.33

4.2125

3.965

4.6375

The SCOBY mass is increased within the kombucha through the growth of a daughter colony from the original microbial mat/mother culture, Figure 3 shows this process. The paper cloth within the diagram was replaced with a rubber balloon to prevent the loss of gasses produced within the fermentation reaction, consequently this limited the oxygen available for the reaction to occur and may have had an impact on the results of the experiment/growth of the SCOBY.

Figure 3: The SCOBY growth process with the multiplication of microbial mats. (Source: Dutta, 2019)

20 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

These results of my experiment mean that there are broader industrial applications than previously thought and a SCOBY can be grown in multiple tea types without affecting the growth rate of the microbial mass, which can be used for sustainable packaging or fabrics (Kruk et al., 2021). The growth of SCOBY affects not only the food and agricultural industry but also the biotechnological processes and biomedicine, the findings of this experiment allow for different tea substrates that allows for a cheaper production of SCOBY without affecting the growth or production of the kombucha. This experiment could be improved in the future by allowing the full 14 days of fermentation and accounting for the loss in weight of the apparatus in a way that captures the losses more effectively to have an overall growth of the day weights rather than a loss. Another direction for future research could include changing the independent variable of the experiment to be the amount of sugar and test the effect of different sugar levels of the SCOBY to final an optimal concentration of the tea solution for microbial growth of the SCOBY, and then test the new discovered optimal level with the same conditions as the experiment conducted to test whether certain tea types growth changes over the two sugar concentrations after comparing the results with the data discovered by this experiment. Another improvement would be to change the dependent variable and measure the composition of the SCOBY between the different tea types as an extension of the experiment completed by the Department of Agricultural and Food Science in 2019 (Gaggìa et al., 2018). When weights of the entire apparatus were recorded on days 4 and 6 it was determined that each test tube was losing weight. This was presumably the mass of carbon dioxide and the energy that was lost as heat. This was unexpected as it was initially assumed that the growth of the microbial mat would exceed the loss of the other products. Several experts were consulted at this point to determine possible next steps. Accordingly the method was adjusted to weigh the microbial mat by physically removing it from the test tube. This is less than ideal because it is imprecise (owing to the mass of tea which clings to the SCOBY) and also because it does not account for any microbial mass that is growing in the tea suspension and not adhering to the microbial mat formed on the surface of the tea solution. However since it was microbial mass that was being measured in this experiment, this was determined to be the best solution to the problem.

Conclusion In conclusion, the type of tea used as a substrate has no significant effect on the growth rate of the microbial mass/symbiotic culture of bacteria and yeasts. My research project explored whether the types of tea used as a substrate affects the rate of microbial growth within kombucha. I used multiple test tubes to create smaller batches of kombucha and grew them for eleven days with four replicates of each tea substrate (black, green, white, oolong and rooibos tea). Data was collected by weighing the mass of the microbial matt on scales as the difference in weight of the whole apparatus began to decrease throughout the experiment due to the chemical reaction of fermentation which was not originally accounted for in the method and required consultation of experts to adjust the method and data collection of the experiment, to be able to effectively measure the increase in mass of the microbial mat. The data analysis involved completing a one way ANOVA test with a post HOC Tukey test and found that the differences were statistically insignificant between the different tea solution substrates, with a P value of 0.531. The value of an insignificant result means that industrially where SCOBY is produced for purposes other than kombucha, such as for sustainable packaging, food or a textile fabric, the tea type substrate will not affect the growth of the mass and therefore other aspects can be prioritised and broadens options for producers, such as limiting cost or other influencing factors within production of the SCOBY as a commodity.

Acknowledgements I would like to thank everyone who has assisted me in the process of completing my experiment and report writing process. Dr Alison Gates, my supervisor, helped greatly through the whole process and her knowledge and suggestions when the method needed to be adapted were of great assistance and accepted with gratitude. I also wish to show my appreciation for my mother and the lab staff at Barker College who supplied the materials in order for me to make the tea solution and the apparatus needed within the experiment. The assistance and second opinions provided by Dr Matthew Hill were also extremely valuable to me and the experiment as they provided outsider opinions on areas for improvement as well as additional knowledge in the scientific process.

References Gaggìa, F., Baffoni, L., Galiano, M., Nielsen, D. S., Jakobsen, R. R., Castro-Mejía, J. L., Bosi, S., Truzzi, F., Musumeci, F.,

Science Extension Journal • 21

Scientific Research in School Volume 3 Issue 1 2021

Dinelli, G., & Di Gioia, D. (2018). Kombucha Beverage from Green, Black and Rooibos Teas: A Comparative Study Looking at Microbiology, Chemistry and Antioxidant Activity. Nutrients, 11(1), 1. https://doi.org/10.3390/nu11010001

Domskiene, J., Sederaviciute, F. and Simonaityte, J. (2019), "Kombucha bacterial cellulose for sustainable fashion", International Journal of Clothing Science and Technology, Vol. 31 No. 5, pp. 644-652. https://doi.org/10.1108/IJCST-022019-0010

Dutta, H., & Paul, S. K. (2019). Kombucha drink: production, quality, and safety aspects. In Production and management of beverages (pp. 259-288). Woodhead Publishing.

Ahmed, R. F., Hikal, M. S., & Abou-Taleb, K. A. (2020). Biological, chemical and antioxidant activities of different types Kombucha. Annals of Agricultural Sciences, 65(1), 3541.

Jakubczyk, K., Kałduńska, J., Kochman, J., & Janda, K. (2020). Chemical profile and antioxidant activity of the kombucha beverage derived from white, green, black and red tea. Antioxidants, 9(5), 447. Wagner, A., Geerts, C., Sondi, N., & Wu, P. (2013). Effects of flavoured tea on the products of kombucha fermentation. Vohra, B. M., Fazry, S., Sairi, F., & Babul-Airianah, O. (2019). Effects of medium variation and fermentation time on the antioxidant and antimicrobial properties of Kombucha. Malaysian Journal of Fundamental and Applied Sciences, 15(2-1), 298-302. Kruk, M., Trząskowska, M., Ścibisz, I., & Pokorski, P. (2021). Application of the “SCOBY” and Kombucha Tea for the Production of Fermented Milk Drinks. Microorganisms, 9(1), 123. St-Pierre, Danielle L., "Microbial Diversity of the Symbiotic Colony of Bacteria and Yeast (SCOBY) and its Impact on the Organoleptic Properties of Kombucha" (2019). Electronic Theses and Dissertations. 3063. Jayabalan, R., Malbaša, R. V., Lončar, E. S., Vitas, J. S., & Sathishkumar, M. (2014). A review on kombucha tea— microbiology, composition, fermentation, beneficial effects, toxicity, and tea fungus. Comprehensive Reviews in Food Science and Food Safety, 13(4), 538-550. Soares, M. G., de Lima, M., & Schmidt, V. C. R. (2021). Technological aspects of Kombucha, its applications and the symbiotic culture (SCOBY), and extraction of compounds of interest: A literature review. Trends in Food Science & Technology.

22 • Science Extension Journal

Coelho, R. M. D., Almeida, A., do Amaral, R. Q. G., da Mota, R. N., & de Sousa, P. H. M. (2020). Kombucha. International Journal of Gastronomy and Food Science, 100272. Laavanya, D., Shirkole, S., & Balasubramanian, P. (2021). Current challenges, applications and future perspectives of SCOBY cellulose of Kombucha fermentation. Journal of Cleaner Production, 126454.

Scientific Research in School Volume 3 Issue 1 2021

Are we flushing money down the fish tank? An investigation of the extent to which commercially available microbes effectively colonise aquaria Anais Tomlinson Barker College In 2013 fish were found to be the most popular pets in the world, today over 8.7 million fish are owned as pets in Australia (Thompson, 2013). The microbiome of the fish tank is important for nutrient cycling and bacteria are responsible for converting toxins (e.g., ammonia) to inert compounds. Microbial enhancers are commercially available liquid aquarium additives that are intended to “enhance” the microbiological community within the aquarium. They are costly additives (approximately AUD$25 for a 250mL) that prescribe weekly addition to the tank. Given the expense of the enhancer, this research aims to determine whether the addition of the product results in any change in the biodiversity of the microbiome in the aquarium. Water samples were taken from four aquaria of relatively small volumes and cultured onto malt marmite plates. The enhancer was dosed to two of the tanks according to the manufacturer's instructions and all our tanks were resampled a week later. Plates were incubated at 25 degrees for a period of 48 hours and then a grid was used to count coliform forming units (CFU’s). An ANOVA. resulting in a p-value of 0.4184, was used to determine that the biological enhancer had no significant effect on the microbiome of the aquaria. Literature review In 2013 fish were found to be the most popular pets in the world, today over 8.7 million fish are owned as pets in Australia (Thompson, 2013) which motivates the research into whether or not commercially available aquarium microbes are useful in improving the biodiversity of substrate microflora?

of fish the ammonia levels can quickly reach toxic levels thus harming the fish. The Osmoregulation of fish living in fresh water means that the fish must excrete dilute urine in large volumes. Osmosis allows a large volume of water to enter the body fluid from surrounding hypotonic freshwater (VetSci, 2010). Osmoregulation incorporates homeostatic mechanisms essential for life. Osmoregulation is the active control of the cellular water balance (Klipp et al., 2005).

Nitrogen cycle The nitrogen and ammonia within fish tanks are involved within the nitrogen cycle. This cycle introduces ammonia through the waste of uneaten food, which bacteria called Nitrosomonas will develop and oxidize the ammonia present in the tank, and thus eliminate it. The product produced by this oxidation reaction is Nitrites. Figure 1 illustrates this process. Fish excrete nitrogenous waste as ammonia. Considering ammonia is highly toxic for fish the ammonia can pose a risk if stored in the body. Due to a fish’s environment – large volumes of water, they are able to continuously excrete ammonia directly into the water, this dilutes ammonia to a non-toxic level. However, in a closed system depending on the quantity

Figure 1: The Aquarium Nitrogen Cycle (Source: Karonen, 2020)

Science Extension Journal • 23

Scientific Research in School Volume 3 Issue 1 2021

How it affects fish Ammonia and nitrite can be trapped within fish tanks due to the nitrogen cycle, the fish swim around in the ammonia and nitrite. The fish then eat and absorb this causing harm to themselves. Fluval biological enhancer is a product which claims to be formulated with live bacteria to help establish a safe aquarium environment, eliminate ammonia and nitrite, and ultimately prevent fish loss. Ammonia can be very harmful to fish as well as the environment within an aquarium as ammonia causes stress and damages the gills and other tissues of fish. If a fish is exposed to a small amount of ammonia it can lead to them being more prone to bacterial infections and poor growth (Francis-Floyd et al., 2009). Nitrites affect the oxidation of haemoglobin, compromising blood oxygen transport. This oxygen depletion can lead to elevated lactate concentrations (Jensen, 2003). Thus, a biological enhancer is needed to eliminate these nutrients in an enclosed fish habitat. This project although unique is similar to many projects already done an example of this is Patin’s article (2018) about Microbiome dynamics in a large artificial seawater aquarium relates to my project as it discusses how such fluctuations in the physical and chemical parameters of the water column affect microbial function may inform our understanding of animal health in closed aquaculture systems (Patin et al., 2018). This project however just focuses on fish and their artificial habitat and the ammonia and nitrates within the fish tank. The purpose of biological enhancer A biological enhancer for aquariums is a responsive biological aquarium supplement that inoculates aquarium water with beneficial bacteria. The copious amounts of bacteria eliminate the ammonia and nitrites and thus generating a biologically well-balanced aquarium where fish can flourish.

intensive aquaculture systems.” This technology has many stated benefits including: the prevention of the introduction of disease from incoming water, improved biosecurity, improved water quality and water-use efficacy, and the reduction of sensitivity to light fluctuations (www.sciencedirect.com, n.d.). Figure 2 shows the process of BFT.

Figure 2: An illustration representing Biofloc technology (BFT). (Source: unknown, 2012)

Consumer Claims Figure 3 shows Fluval’s biological enhancer which states that this product is infused with a powerful team of beneficial bacteria that immediately inoculates aquarium water therefore we can test the statements made by plating samples from aquarium water to see if there is any difference in the bacteria apparent. If there is then it is evident that the biological enhancer has effectively inoculated the water however if there is no statistically significant difference in the CFUs grown, then the product has not done what it claims to do. These microorganisms should produce a biological flora that instantly metabolizes ammonia and nitrite. The company does also say that regular application helps to completely exclude establishment of undesirable bacteria (Fluval USA, n.d.)

Fluval’s biological enhancer’s description states that “. It quickly establishes safe and essential conditions in new aquarium setups, so that you can introduce fish to new aquariums immediately.” This enhancer has a unique Biofloc technology (BFT) that eliminates product instability, inefficacy, and a short shelf life. (Pet Stock, n.d.) The advantages of BFT BFT is defined as “the use of aggregates of bacteria, algae, or protozoa, held together in a matrix along with particulate organic matter for the purpose of improving water quality, waste treatment and disease prevention in

24 • Science Extension Journal

Figure 3: A 250mL bottle of Fluval’s biological enhancer that costs AUD$23.49. (Source: Fluval USA, n.d.)

Scientific Research in School Volume 3 Issue 1 2021

Scientific research question Are commercially available aquarium microbes useful in improving the biodiversity of substrate microflora?

Scientific hypothesis There will be a statistically significant difference in the number and diversity of microflora before and after treatment with the biological enhancer.

Methodology Four similar fish tanks (similar volume and position) were chosen with a school laboratory setting. A water sample from within the gravel of each tank was taken using a sterilised 120ml sample bottle, the water was taken four times to obtain replicate samples. The liquid was then cultured onto commercially prepared agar using a sterile inoculating method. These plates were then left at room temperature for a period of 48 hours. Both nutrient and malt marmite agar plates were trialled however it was concluded that the malt marmite plates grew more CFUs with a high diversity thus coming to the decision that the malt marmite plates would work better for this experiment. After the samples were taken, 50ml of Fluval was administered to tank 1 and 4 leaving the other two as controls. After 1 week I extracted a water sample from within the gravel of each tank using a sterilised 120ml sample bottle. The liquid was then cultured onto commercially prepared malt extract agar plates using a sterile inoculating method. These plates were then left at room temperature for a period of 48 hours. After the samples were cultured and left to grow, the growth on the commercially prepared agar plates was compared and studied to determine the effectiveness of the biological enhancer, the controls were also studied and evaluated to see if there was a statistically significant difference in the amount of CFUs grown.

Analysis Methodology Once 48 hours have passed the petri dishes were put on a grid and counted for the different types of CFU’s. The CFU’s count was through a random sampling method then compared using an Analysis of variance (ANOVA) test to see whether there was a statistical difference or not.

Results Table 1 contains the raw data collected through the random sampling method in order to count the quantity of the CFUs grown and the different types in each of the four fish tanks used from the agar plates (Figure 4). Table 1:Raw data on bacterial growth

Fish tank: Untreated 1

Treated

Untreated

Treated Untreated 3 Treated Untreated 4

Treated

Inoculation Methodology In a fume cupboard a Bunsen burner was lit to increase sterilisation and the table was cleaned with methylated spirits. I then took 2 drops from the 120ml sample bottle using a sterilised pipette and dropped them on the malt agar plate. The liquid was then spread using a sterilised L-shaped hockey stick spreader. The Agar plate lid was then placed on the cultured plate, labelled, and sealed. The sealed agar plates were then left at room temperature for a period of 48 hours.

Fluval

N/A

Bacteria A B D F B H J A A B C F D A D I D A F D I B J G E D E B A B C

Bacteria in Bacteria sample in whole 1 16 1 16 6 96 5 80 2 32 2 32 5 80 3 48 1 16 1 16 40 640 5 80 6 96 1 16 1 16 1 16 77 1232 1 16 6 96 7 112 40 640 35 560 43 688 1 16 1 16 5 80 66 1056 6 682 15 240 47 752 84 1344

The results are presented graphically in figure 5 Indicating no obvious trend. An ANOVA test was used to further understand these results and the findings for that test show that the P-value is above the accepted value of 0.05 which is shown in Table 2.

Science Extension Journal • 25

Scientific Research in School Volume 3 Issue 1 2021

.Figure 4: These are all the agar plates plated on malt marmite agar.

CFU counted

Bacterial growth on agar plates 1800 1600 1400 1200 1000 800 600 400 200 0

Type of CFU total treated control group

total treated experimental group

total untreated control group

total untreated experimental group

Figure 5: This is a graph showing the results from the raw data. Table 2: The results from the ANOVA test

Source

DoF ν

Treatment

732,736

732736

Error

1,433,216

Total

2,165,952

P value 0.4184

Turkey HSD – Q statistic 1.4300

716608

Discussion The key research question was whether commercially available aquarium microbes are useful in improving the biodiversity of substrate microflora. During the phase of the experiment the tanks were all treated as they normally would thus increasing the validity of the results. During this experiment only two of the four fish tanks received the biological enhancer. Testing the hypothesis - There will be a statistically significant difference in the number and diversity of microflora before and after treatment with the biological enhancer. – this would suggest that if there was a statistical difference then the applications of this product could be endless.

26 • Science Extension Journal

F statistic 1.0225

In order to interpret the data collected in a way to either prove or disprove the hypothesis an ANOVA test was performed on the data to see whether there were any statistical differences found in the CFUs counted on the agar plates. An ANOVA test uses the means of the given samples to check the impact of one or more factors by comparison. (Analytics Vidhya, 2018). As seen in table X the results of the ANOVA test showed that the p-value was 0.4184 which is above the accepted value of 0.05 which shows that the results were statistically insignificant. The results demonstrate that when the tanks were inoculated with the biological enhancer (Tanks 1 and 4) there was no statistically significant difference (P-value 0.4184>0.05) in the bacterial communities present in the gravel when compared to the two control tanks (tanks 2 and 3) as can be seen in table 1. The statistically insignificant result from this

Scientific Research in School Volume 3 Issue 1 2021

ANOVA test can lead to the conclusion that this product does not do what it claims to do for the microbial communities within the fish tank. When collecting samples from the bottom of the fish tanks a small plastic sample bottle was used however a more reliable method may have been to use bio films. The sampling method used in this experiment however took less time however it may have produced results that do not reflect the correct diversity and number of bacterial communities. Originally a nutrient agar plate was used to culture the bacteria but found that the CFUs grown was not appropriately representing the diversity within the bacteria. Figure 6 shows the CFUs grown on a nutrient agar plate and incubated at 40oC while figure 7 shows the CFUs grown on a malt marmite agar plate and left in the lab at room temperature (25oC). There is a clear difference in the diversity and amount of CFUs grown between figure 6 and figure 7.

Figure 4: This is an image of a plate grown before any biological enhancer was added to a fish tank when grown on nutrient agar.

of the area must have an equal chance of being chosen. Random sampling with quadrats is used to examine differences between contrasting habitats within a habitat.” (“Sampling methods,” 2016). This method was used to calculate the amount of CFUs on each plate. The same sixteen squares were chosen from each plate and then counted. This method is not the most reliable method of counting and decreased the reliability of this experiment as seen in figure 8 the sixteen squares chosen does not appropriately represent the CFUs grown. This decreased reliability could be increased by counting each individual CFU grown on the whole plate however due to time restrictions this was not possible. If this experiment were to be replicated to increase reliability the random sampling method would not be suggested as it does not represent the CFU grown in a way to produce the most accurate results.

Figure 6: This image is of one of the plates grown on the malt marmite agar with a square attached to represent the random sampling method.

Limitations Some limitations from this experiment included:

Figure 5: This is an image of a plate grown before any biological enhancer was added to a fish tank when grown on Malt marmite agar.

Random sampling is “Random sampling is used to select a sample that is unbiased. Within each area, every part

Minimal range of data – if more fish tanks or plates of CFUs were grown the results may have been more reliable leading to a more accurate result. Currently the lack of repeated data found in this experiment may not lead to an accurate result.

Equipment – due to the minimal number of aquaria’s available no replication was able to happen instead to combat this half of the tanks used were inoculated to try and achieve the most reliable results. In a standard university microbiology lab, they have a machine that can distinguish how many different types of DNA are present in a water sample. This machine would give a more accurate representation of the microflora present in the aquaria.

Science Extension Journal • 27

Scientific Research in School Volume 3 Issue 1 2021

Improvements or interesting additions which could be done If someone wanted to expand on this experiment, there would be a few different paths they could go down to explore this product in more detail these include: -

Anaerobically incubating the agar plates – this could show more bacteria from within the gravel as they would be more used to an anaerobic setting. Saltwater tanks – this biological enhancer claims to work in both salty and freshwater. This would be interesting to see the difference in the efficacy of the product.

Conclusion In conclusion the results of this experiment lead to the rejection of the hypothesis that there is a statistical difference in the biological diversity of microbes in the fish tank before and after treatment with the biological enhancer. According to this project the product fails to meet its consumer claim that it significantly changes the microflora of the aquarium and on this basis, the product is not recommended. However, a number of steps could be taken to validate the research further and to undertake further investigation. Firstly, anaerobically incubating the agar plates could show more bacteria from within the gravel as they would be more used to an anaerobic setting. Secondly, by testing the same product and seeing the effects it has on saltwater tanks. My research project explored whether or not commercially available aquarium microbes are useful in improving the biodiversity of substrate microflora. I grew CFUs from samples collected from four different aquaria within a school laboratory setting and observed the difference in the amount of CFUs grown before and after putting in the biological enhancer. Data was collected through a process of random sampling (the same squares on each plate). The data analysis involved an ANOVA turkey HSD test.The results from my data showed a P value of 0.4184 which is greater than the accepted value of 0.05 leading me to reject my hypothesis that there will be a statistically significant difference in the number and diversity of microflora before and after treatment with the biological enhancer.

Acknowledgments I would like to thank Dr Alison Gates for her consistent help throughout my project assisting me with understanding bacteria and microbiology and applications of previous scientific knowledge. I

28 • Science Extension Journal

would also like to thank Dr Katie Terret and Dr Matthew Hill, for their help in understanding the processes for the writing of a report, and more explicitly Dr Matthew Hill for his knowledge on statistical analysis. Finally, I would like to thank the school’s laboratory staff for their help.

References Analytics Vidhya. (2018). Analysis Of Variance (ANOVA) | Introduction, Types & Techniques. [online] Available at: https://www.analyticsvidhya.com/blog/2018/01/anova analysis-of-variance/#: ~:text=ANOVA%20checks%20the%20impact%20of. Field Studies Council. (2016). Sampling methods. [online] Available at: https://www.biology fieldwork.org/alevel/fieldwork-techniques/introduction-tosampling/sampling-methods/ [Accessed 15 Apr. 2021]. Fluval USA. (n.d.). Fluval Cycle Biological Enhancer, 8.4 fl oz (250 mL). [online] Available at: https://fluvalaquatics.com/us/product/cyclebiological-enhancer-3/ [Accessed 1 Jun. 2021]. Francis-Floyd, R., Watson, C., Petty, D. and Pouder, D. (2009). Ammonia in Aquatic Systems. [online]. Available at: https://edis.ifas.ufl.edu/pdf%5CFA%5CFA03100.pdf [Ac cessed 24 May. 2021]. Jensen, F.B. (2003). Nitrite disrupts multiple physiological functions in aquatic animals. Comparative biochemistry and physiology. Part A, Molecular & integrative physiology, [online] 135(1), pp.9–24. Available at: https://www.ncbi.nlm.nih.gov/pubmed/12727546 [Access ed 29 Jan. 2020]. Klipp, E., Nordlander, B., Krüger, R., Gennemark, P. and Hohmann, S. (2005). Integrative model of the response of yeast to osmotic shock. Nature Biotechnology, [online] 23(8), pp.975–982. Available at: https://www.nature.com/articles/nbt1114 [Accessed 7 Dec. 2020]. Patin, N.V., Pratte, Z.A., Regensburger, M., Hall, E., Gilde, K., Dove, A.D.M. and Stewart, F.J. (2018). Microbiome Dynamics in a Large Artificial Seawater Aquarium. Applied and Environmental Microbiology, [online] 84(10). Available at: https://aem.asm.org/content/84/10/e00179-18 [Accessed 7 Feb. 2021]. Pet Stock. (n.d.). Fluval - Biological Enhancer - Live Bacteria for Aquariums. [online] Available at: https://www.petstock.com.au/product/fish/fluvalbiological-enhancer-live bacteriafor aquariums/53706#:~:text=Fluval%20Biological%20Enha ncer%20is%20a,biological%20habitat%20for%20your%20fis h.&text=Regular%20application%20helps%20to%20competi tively ,of%20undesirable%20bacteria%20in%20aquariums.. Thompson, A. (2013). What’s the Most Popular Pet? [online] Live Science. Available

Scientific Research in School Volume 3 Issue 1 2021

at: https://www.livescience.com/32415-whats-the-mostpopular-pet.html. unknown (2012). Scheme of biofloc technology (BFT) System. [online] ResearchGate. Available at: https://www.researchgate.net/figure/Scheme-of-biofloctechnology-BFT System-Avnimelech2009_fig1_264547908 [Accessed 12 Jun. 2021].

www.sciencedirect.com. (n.d.). Biofloc Technology - an overview | ScienceDirect Topics. [online] Available at: https://www.sciencedirect.com/topics/agricultural-andbiological sciences/biofloctechnology#:~:text=Biofloc%20technology%20(BT)%20is% 20defined [Accessed 8 Jun. 2021].

VetSci. (2010). Comparative Nitrogen Excretion. [online] Available at: http://vetsci.co.uk/2010/05/15/comparativenitrogen-excretion/# [Accessed 7 Jun. 2021].

Science Extension Journal • 29

Scientific Research in School Volume 3 Issue 1 2021

30 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

To bean or not to bean... Will Vanilla grow more microbes after different processing techniques? That is the question… Nathan Finikin Barker College Vanilla is one of the world’s most expensive spice crops and there are three major commercially available species of vanilla: V. tahitensis, V. pompona and V. planifolia (Correll, D. S. 1953). This report focused on V. planifolia as it is the most widely available variety to consumers, however different manufacturers use different processes resulting in wide variations in the moisture content of the beans. It was hypothesized that the higher the moisture content of the bean, the greater the microbial contamination on the exterior surface. Commercially available V. planifolia beans from four different manufacturers were tested to determine the diversity of flora that could be cultured from swabbing the exterior surface of the vanilla bean and then inoculated onto agar plates, two regular nutrient agar and two malt marmite agar. These were incubated at 37 degrees and 25, respectively. It was determined that the vanilla with the highest moisture content in fact had the least amount of contamination while the brand with the least moisture had the most, thus leading to the rejection of the hypothesis. One possible explanation for this unexpected result is that vanillin has antimicrobial properties.

Literature review Vanilla Gallage et al. (2018) provides significant information on what vanilla is used for as it is a spice that provided a significant portion of the economy for South American and Caribbean countries. The report also shows ideal growth conditions as vanilla’s native location is from Mexico and thus vanilla grows best in humid warm-temperate conditions. Bory et al. (2008) highlights different species of vanilla, however in the list V. planifolia is the most widely grown as well as commercially grown variant of vanilla. It also goes into detail about the methods of reproduction of vanilla and the labour-intensive implications of its fertilisation. It also highlights some historical significance. The articlw informed my choice of which variety of vanilla to purchase, and as such I will use V. planifolia. Correll (1953) goes into detail about the distribution of vanilla across the globe, including Indian ocean islands Latin America, French Oceania, Australasia, and the South Pacific. It also mentions the different species of vanilla beans the three main being Vanilla planifolia, Vanilla pompona and Vanilla tahitensis. The article

goes into detail about horticulture, especially the fact that vanilla production is very labour intensive and requires high management during propagation and fertilization. Processing Technologies Brillouet et al. (2010) goes into significant detail about the sugars and chemicals found in vanilla (βglucosidase and glucovanillin), the extraction methods of these chemicals as well as how to identify them. It discusses the concentrations of the chemicals at different points on a vanilla bean. It also goes into detail about the anatomy of a vanilla bean. I will build on this article as I will look at microbial contaminants after sale of vanilla. Krishnakumar (2011) highlights some of the storage methods for vanilla, along with drying and curing beans. It also has an experiment based on the different amounts of time after sweating occurred after harvest (in this case it is most effective at 2 days). The report shows that vanillin concentration is highest when the moisture in the bean is at its highest post-processing. This is relevant to my rationale as the higher the moisture content and vanillin, the more money you can sell vanilla for. This is relevant as storage is the main Science Extension Journal • 31

Scientific Research in School Volume 3 Issue 1 2021

factor in which microbial contamination can occur so therefore highly significant to take it into account. Microbial contamination of vanilla Röling et al. (2001) looks at the microbial biology in different curing methods of vanilla. Higher temperatures inhibit the growth of fungi and addidtionally during packaging it is ensured that the environment was sterile. It also looks at different DNA sequences and uses cultivation-independent profiling of bacterial communities, by using PCR to determine the concentrations and types of bacterial growth in a vanilla bean. This allows for an educated guess on what species of fauna might be present after the inoculation. Ranadive et al. (2011) talks of the different grading systems of vanilla before it goes to market, it also talks of the different aromas of vanilla and how these change across geographical locations. It also looks at the different abundances of volatiles in vanilla beans. The authors examine moisture content and storage conditions (the fact that vanilla beans with a moisture content of higher than 25% can easily develop mould if under poor conditions/handling). They used spectrometers and colorimetry to determine the concentration of vanillin. The HPLC (High performance liquid chromatography) method was used to give the best estimate. The article also looked at the possibility of microbial contamination and discusses the recommended limits of microbial contaminants. This builds on the previous article as it allows for assisted identification of the microbes that might be present after the investigation is carried out. Van Dyk et al. (2010) talks of the influence of curing procedures on vanillin, it talks of the change in chemicals during the process (glucovanillin to free vanillin). They used three different curing methods: blanching, curing, and sweating. They then hydrolyse the glucovanillin and determine the concentration of vanillin after hydrolysis, this was done through colorimetry. This investigation will test the microbial contaminants after storage and sale of vanilla. Silva et al. (2011) goes into detail about the chemical compounds found in traditional vanilla curing, all with aromatic properties. It then discusses the different concentrations of these aromatic chemicals. This test used a standard curing procedure and the vanilla beans were tested for moisture. This investigation will look at the microbial growth after storage, which happens after the drying process, as the drying process leads to the final concentration of moisture in a vanilla bean.

32 • Science Extension Journal

This leads to the formulation of a research question and hypothesis around the impact of moisture content on the microbial contamination of beans.

Scientific research question Does the moisture content of a vanilla bean effect the diversity of microflora that can be cultured from the exterior surface?

Scientific hypothesis The higher the moisture concentration in the vanilla bean, the greater biodiversity of microbes on the vanilla bean.

Methodology Four different quality brands vanilla beans: Spice & Co, Hoyts, Queen and BestGrow were selected and tested to see which vanilla bean had the most microbial growth after time periods after a week. The vanilla beans were visually appraised and observed to be profoundly different in appearance. Moisture content was originally determined by a web search of the analysis provided by the manufacturer on each website. This was recorded in Table 1. The reported values did not appear to be consistent with the visual appraisal (see Figure 1) and so a proxy measure was calculated as described in the results.

Figure 1:Different brands of vanilla bean Table 1: Moisture content as claimed by processor (data obtained from website).

Brand BestGrow Queen Spice & Hoyts

Claimed moisture percentage 30% 20-25% 15-20% 10-15%

Scientific Research in School Volume 3 Issue 1 2021

In a fume hood and using a Bunsen burner to create a convection shield and therefore reducing sources of exterior contamination, each vanilla bean was swabbed with a sterile cotton swab for around 30 seconds. This was done twice and then the swabs were individually inoculated onto two regular agar plates and two malt marmite plates. The regular agar plates were placed into an incubator at 37 degrees and the malt marmite plates were placed in an incubator set at room temperature (25 degrees). Results were recorded visually over the time periods of 12, 24, 42, 54, 70 and 120 hours. The marmite plates were recorded up to 300 hours after swabbing. After the plates growth had finished the results were recorded by counting the number of coliform forming units (CFUs) of each microorganism present.

Results Experimental Results As described in the methodology, the moisture content data provided from the manufacturer was considered unreliable based on the visual appraisal (see Table 1 and Figure 1). Accordingly, bulk density would normally be calculated by a measure of mass/volume, however since the volume of the bean can only be determined by immersion in a fluid (Lacoma, T. (2018), this would alter the moisture content and could affect any surface contamination that is intended to be tested. Therefore, a proxy was used calculating the length to mass ratio of the beans and this was determined to be an adequate measure based on consultation with colleagues. Owing to the challenges of accurately calculating the moisture content of the beans in a school laboratory, the vanilla beans were weighed and their lengths were measured allowing an approximate moisture content to be calculated by calculating weight over length. This is recorded in Table 2.

presented in a graphical form in Figure 2 (agar plates) and Figure 3 (malt marmite plates). The qualitative observations of the plates were recorded in the laboratory notebook and these observations are shown in Table 6. Table 3: Average growth per vanilla brand.

Agar Nutrient agar Malt marmite

Best Grow 3.5 0

Queen 11.5 1

Spice & Co 20.5 8

Hoyts 43.5 21

Statistical Analysis An initial statistical analysis was undertaken to determine whether there was a statistical difference between the amount of microbial growth recorded in the different brands of vanilla bean. Data were analysed using the online platform ASTATSA and the output is recorded in Figures 4-7. A post-hoc Tukey test and ANOVA test was used to determine whether a significant difference existed between the vanilla brands. An alpha value of 0.05 was used and the data from nutrient agar plates and malt marmite plates were treated in two separate tests. The ANOVA test showed that the Hoyts vanilla (group D) is significantly different to all other brands (Tables 7 and 8). This was also true for the malt marmite plates where Hoyts was significantly different to all other brands of vanilla (Tables 9 and 10). It was found that one or more of the data points were significant for both the nutrient agar and malt marmite plates which had P-values of 0.0018 as seen in Figure 4, and 0.0014.

Table 2 Approximate moisture contents of each brand of vanilla

Vanilla Brand

Weight (g)

Length (mm)

BestGrow Queen Spice & Co Hoyts

6.22 2.3

15.4 13.7

Approximate moisture content (%) 40.38961 16.788321

2.76

22.3

12.376682

1.28

16.8

7.6190476

The raw data from the counting of the plates is contained in Tables 3 and 4. The average data for each brand were calculated and tabulated in Table 5. These data, showing the number of CFUs counted, was Science Extension Journal • 33

Scientific Research in School Volume 3 Issue 1 2021

Table 4: Nutrient agar plate analysis

Organism

Description

Orange with circular dot in centre

Yellow squarish shape

Cloudlike, orange colour

Treelike (straight with branches), white/yellow

White with yellow base, powdery

White with slit in middle

Yellow, translucent

White, opaque, reddish dot in centre

Yellow/white with orange circle

Yellow, ridges

White, cross in centre

Large area white yellow splotches

Brown bloblike

Total number of colonies

N/A

Hoyts Hoyts Spice & Spice & Queen 1 Queen BestGrow 1 2 Co 1 Co 2 2 1

F-Statistic = 41.2299, P-value = 0.0018

34 • Science Extension Journal

BestGrow 2

Scientific Research in School Volume 3 Issue 1 2021

Table 5: Malt Marmite Agar Analysis

Organism

Description

Hoyts 1 Hoyts 2 Spice & Spice & Queen 1Queen 2 BestGrow Co 1 Co 2 1

BestGrow 2

Large grey fuzzy

White opaque, round margin, pinched

Opaque, round, higher

Round coneish, white

Straight edge, ridges

Straighter including lines, white

Bloblike, not round, white, opaque

Powder-like on top of opaque with pinch

Large, brownish, raised centre

Total Number of colonies

N/A

F-Statistic 46.8333, P-value = 0.0014

Table 6: Qualitative observation notes of plates

Time 0-12 hours:

24-54 hours: 70 hours: 70 hours: 120 hours: 300 hours:

Observation Notes This was the period with the most significant growth. Specifically, the Hoyts had the most growth followed by the Spice &Co. bean, then the Queen bean and finally the Best Grow bean. During this period there was minimal growth on the malt marmite plates. The agar plates rate of growth had significantly decreased. First signs of fungal growth are seen on the malt marmite plate on the queen bean. First signs of fungal growth are seen on the malt marmite plate on the queen bean. Fungal growth was present on all plates apart from the Best Grow plate with the queen plate having a single large colony while other plates had smaller colonies with multiple species. Around this time the plates had also dried, and the results were recorded. Fungal growth had significantly slowed and thus results were recorded.

Science Extension Journal • 35

Scientific Research in School Volume 3 Issue 1 2021

50 45

40 35

30 25

20 15

10 5 0

0 BestGrow

Queen

Spice & Co

Hoyts

BestGrow

Queen

Spice & Co

Hoyts

Figure 2: Average number of colonies of microorganisms on Nutrient agar plates.

Figure 3:Average number of colonies of microorganisms on Malt Marmite Agar Plates.

Importantly the ANOVA test is not being used to interrogate the overall hypothesis for this project but rather to determine whether there is a statistical difference between the groups of vanilla. In this case, the statistical analysis reveals a statistical difference between the Hoyts vanilla and the other brands. There was also a small statistical difference between the BestGrow group and the Spice & Co group.

Table 10: Statistical difference between testing groups for malt marmite plates.

Table 7: ANOVA statistical test for nutrient agar plates. Source SS DoF MS Fstat pvalue Treatment 1794 3 597.8 41.2 0.0018 Error 58 4 14.5 Total 1852 7

Discussion

Table 8: Statistical difference between testing groups for nutrient agar plates.

Treatments pair A vs B A vs C A vs D B vs C B vs D C vs D

p-value

Inference

0.2922745 0.0369978 0.0016263 0.2267710 0.0038190 0.0129559

Insignificant Significant Significant Insignificant Significant Significant

Table 9: ANOVA statistical test for malt marmite agar plates. Source SS DoF MS Fstat pvalue Treatment 562 3 187.3 46.8 0.0014 Error 16 4 4 Total 578 7

36 • Science Extension Journal

Treatments pair A vs B A vs C A vs D B vs C B vs D C vs D

p-value

Inference

0.8999947 0.0529249 0.0016304 0.0799671 0.0019676 0.0099325

Insignificant Insignificant Significant Insignificant Significant Significant

Initially beans were weighed and measured for their length; these weights were then put over their length values and an approximate moisture content was determined. Following the original assumption that the most expensive vanilla would have the most moisture it was found that the cheapest brand, Hoyts, had the lowest moisture content. BestGrow which was the most expensive had the highest moisture content. This is seen in tables 1 and 2. Originally my hypothesis stated that the vanilla with the highest moisture content would have had the most microbial growth. This corresponds to the generally assumed rule that the higher the moisture content of an object, the greater the microbial contamination on the exterior surface. (Lacoma, T. (2018)). Yet this investigation found the inverse of this was true. The findings of this investigation suggest that vanilla beans contain some compound that is anti-microbial. The most obvious candidate is the active ingredient which is the vanillin.

Scientific Research in School Volume 3 Issue 1 2021

Ahmad, et al. (2020) states that vanilla has many pharmaceutical implications including Antibacterial, Antibiofilm, Antigenotoxic and Antifungal. All these properties would imply that vanilla has some sort of application that could inhibit the growth of microbes on a vanilla bean. This is due to vanilla beans with more moisture containing more vanillin (Dyk et al. 2010). Vanillin is seen to have mild anti-fungal properties and thus the results were skewed in the inverse direction to what was expected. Xu et al. (2021) showed that vanilla was particularly effective at combatting and inhibiting the growth of microbes present after the fermentation of milk, some of these pathogens are common thus suggesting that it might have more widespread effects. Li, Q., & Zhu, X. (2021) investigated that although vanillin had some antimicrobial properties, isomers of this molecule which occur naturally were much more effective for example, 2-hydroxy-5methoxybenzaldehyde. Thus, it is seen that vanilla has antimicrobial properties and this has skewed the results. Vanilla beans with higher moisture content were seen to contain higher amounts of chemicals including vanillin and 2hydroxy-5-methoxybenzaldehyde. It is found that these would have impacted results, this is shown by the fact that the BestGrow plate which had the highest moisture content had the least amount of growth. Consumers are generally willing to pay more for a safer product. Because vanilla with higher moisture content is seen to have less microbial growth it should be a main goal for the processor/manufacturer to attempt to produce vanilla beans which have higher moisture content. This can be used as a marketing incentive and thus with education consumers will look for vanilla beans which have a higher moisture content. However, this method is more expensive to produce. If this investigation were carried out again, more plates would have been made so that replication could be increased and thus reliability of results would have been increased. If equipment that allowed for accurate moisture content to be determined this would have been done. This would produce more accurate results as the moisture content assumed in this experiment does not take volume into account as there was no opportunity to use water displacement. This would impact the exterior environment of the bean and thus impacting results.

Other researchers could build on my project by looking into the anti-microbial properties of vanilla. One experiment would be to make a lawn of microbes and then place samples of vanilla beans of these different brands onto the lawn. Alternatively, paper disks could be placed into a bag with the BestGrow vanilla which had the highest moisture content. These disks could then be placed onto a lawn of microbes and the zones of inhibitions be measured by a vernier calliper.

Conclusion This investigation hypothesised that the higher the concentration of moisture in a vanilla bean the greater the biodiversity of microbes found on the exterior surface of a vanilla bean. However, on the basis of experimentation this hypothesis was rejected as the bean with the highest moisture content was actually found to have the lowest microbial contamination. Additionally, it was found that the vanilla which had the least amount of moisture was found to have the most growth. The most likely explanation for this is that the higher concentration of vanillin present in the higher moisture bean, is responsible for the antimicrobial effect. Further research is warranted on the antimicrobial properties off vanillin. It is also conceivable that the different processing procedures may contain a step that retards microbial growth. If this is the case it is likely important to consider possible contamination that may occur during transport and storage. In my scientific investigation I selected four different brands of vanilla from the same species. I then hypothesized that the higher the moisture of a vanilla bean the higher the levels of microbial growth. The hypothesis was found to be incorrect, and it was seen that null-hypothesis that the more moisture in a vanilla bean the more colonies of microbes were present after swabbing was accepted. An ANOVA test was used to compare the significance of variance within the data populations. This showed that there was a large statistical difference between the cheapest vanilla brand and the rest. Thus, further research must be done to determine what impacted the results. I suggest that the likely cause for the lower microbial growth in the vanilla with higher moisture content is a chemical found naturally in vanilla. This investigation is valid as it fulfilled the research question. As it is found that moisture content does somewhat effect the microbial growth on the exterior surface of a vanilla bean. It was slightly unreliable due to a low amount of replication of testing groups. In this investigation there was a sample size of 2 per test of vanilla. The vanilla beans were selected randomly from within the brands and were also Science Extension Journal • 37

Scientific Research in School Volume 3 Issue 1 2021

inoculated onto random agar plates. This shows that the results were unlikely to be skewed due to bias within testing groups.

Acknowledgements I would like to thank David Soo for providing inspiration behind the project and sharing some insightful papers that he has used during his own work also for providing the “Best Grow” vanilla which was of the highest quality. I would also like to thank Dr Alison Gates for assisting in carrying out my experiment along with helping with data collection and analysis.

References Ahmad, H., Khera, R. A., Hanif, M. A., Ayub, M. A., & Jilani, M. I. (2020). Vanilla. In Medicinal Plants of South Asia (pp. 657-669). Elsevier. Andromeda, R. (2017). How to Calculate Density by Water Displacement. Sciencing. Available at: https://sciencing.com/calculate-density-water-displacement7373751.html

Gallage, N. J., & Møller, B. L. (2018). Vanilla: the most popular flavour. In Biotechnology of natural products (pp. 324). Springer, Cham. Krishnakumar, V., Bindumol, G. P., Potty, S. N., & Govindaraju, C. (2011). Processing of vanilla (vanilla planifolia andrews) beans-influence of storing fresh beans, killing temperature and duration of killing on quality parameters. Journal of Spices and Aromatic Crops, 16(1). Lacoma, T. (2018). Factors That Affect the Growth of Microorganisms. Sciencing. https://sciencing.com/factorsaffect-growth-microorganisms-5299917.html. Li, Q., & Zhu, X. (2021). Vanillin and its derivatives, potential promising antifungal agents, inhibit Aspergillus flavus spores via destroying the integrity of cell membrane rather than cell wall. Grain & Oil Science and Technology. Ranadive, A. S., Havkinfrenkel, D., & Belanger, F. C. (2011). Quality control of vanilla beans and extracts. Handbook of vanilla science and technology, 163-183. Röling, W. F., Kerler, J., Braster, M., Apriyantono, A., Stam, H., & van Verseveld, H. W. (2001). Microorganisms with a taste for vanilla: microbial ecology of traditional Indonesian vanilla curing. Applied and Environmental Microbiology, 67(5), 1995-2003.

Bory, S., Grisoni, M., Duval, M. F., & Besse, P. (2008). Biodiversity and preservation of vanilla: present state of knowledge. Genetic Resources and Crop Evolution, 55(4), 551-571.

Silva, A. P., Gunata, Z., Lepoutre, J. P., & Odoux, E. (2011). New insight on the genesis and fate of odour-active compounds in vanilla beans (Vanilla planifolia G. Jackson) during traditional curing. Food Research International, 44(9), 2930-2937.

Brillouet, J. M., Odoux, E., & Conejero, G. (2010). A set of data on green, ripening and senescent vanilla pod (Vanilla planifolia; Orchidaceae): anatomy, enzymes, phenolics and lipids. Fruits, 65(4), 221-235.

Van Dyk, S., McGlasson, W. B., Williams, M., & Gair, C. (2010). Influence of curing procedures on sensory quality of vanilla beans. Fruits, 65(6), 387-399.

Correll, D. S. (1953). Vanilla-its botany, history, cultivation and economic import. Economic Botany, 7(4), 291-358

38 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Physics Physics underpins all that we observe, and this year four students took very different approaches to produce unique developments in our understanding of the physical world.

Five students tackled a diverse range of projects grouped broadly in the physics or numerical analytics domain. Each project demonstrated sophistication in complex areas. To highly analytical projects were conducted by Harry and Manxi. Harry sought to understand computational modelling of a double pendulum and use this to pinpoint the precise conditions under which the pendulum’s motion switched from being defined as periodic to chaotic. Manxi also wrestled with highly technical definitions where she sought to quantify and therefore compare randomness of strings of numbers. Both projects involved drawing out analytical methods from the literature, understanding them, critiquing them, and developing their own methods for analysis with very impressive logic. Tomo’s project was impressive particularly in scale, involving flipping a coin 2000 times using a contraption of his own design for consistency to investigate whether changing the height above ground would result in any bias in the coin toss. The results, and comparison with previous investigations into unexpected bias are worth investigating. Fran’s project also involved construction of her own apparatus to be used in testing. She produced three transformer cores made not out of solid or laminated soft iron, rather soft iron particles of only micrometres in diameter. Her literature and discussion describe how the optimum size of particles results from a balancing of various competing effects including minimizing eddy currents and impedance. Finally, Caleb asked an intriguing research question where he sought to take known principles and apply them in a new context. Building on work from YouTube sensation Derek Muller (Veritasium) in physics education instructional design, Caleb sought to investigate student understanding of misconceptions in learning physics from written text with implications of how to think about and improve instruction in the classroom and beyond.

Science Science Extension Journal • 39

Scientific Research in School Volume 3 Issue 1 2021

40 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

The Boundary of Chaos: An Investigation into the Length Ratio Dependent Chaotic Dynamics of a Planar Double Pendulum Harry Breden Barker College A planar double pendulum is defined by attaching two point masses together, with one of the point masses being connected to a pivot point. It is an interesting dynamic system because of its tendency to exhibit chaotic motion. Chaotic motion can be quantified using the Lyapunov exponent. If the Lyapunov exponent is positive, the system is considered chaotic. If the Lyapunov exponent is negative, instead of being chaotic, the system produces periodic motion. Extant research into the planar double pendulum indicates that as the length ratio of a pendulum increases, the Lyapunov exponent increases. Previous research has determined that at length ratio 1: 1 (comparing the length of the upper arm to the lower arm) the pendulum’s motion is periodic, while if the length ratio is increased to 1: 3, the pendulum’s motion is chaotic (Gupta et al., 2014). Building on Gupta et al.’s results, this research aimed to increase the precision of the measured length ratio representing the transitional point between periodic and chaotic motion. A computational simulation that provided a numerical solution to the Euler-Lagrange equations of the pendulum was used to determine the Lyapunov exponent for differing length ratios. The results demonstrated that the transitional length ratio lies between 1: 2.34375 and 1: 2.375, an increase in precision by a factor of 64 compared to the current bound established by extant research.

Literature Review

The Planar Double Pendulum System A planar double pendulum is defined by attaching two point masses with a rigid, weightless rod, with the top point mass connected to a pivot point with a second rigid, weightless rod as seen in Figure 1 (Levien & Tan, 1993). The length ratio of a pendulum is expressed as 𝐿𝐿𝐿𝐿1 : 𝐿𝐿𝐿𝐿2 . A pendulum is a Hamiltonian system, meaning its gravitational potential energy and kinetic energy is constantly exchanged and conserved throughout its motion (Biglari & Jami, 2016). Most importantly, the system has tendencies to produce chaotic motion (Richter & Scholz, 1984; Safitri et al., 2020). The Phase Space The phase space is an important mathematical tool that is used when describing a system’s motion. In this research, the computational simulation was defined within a phase space coordinate set. A system’s phase space is the graphical interpretation of the canonical coordinates that encode all possible physical states of the system (Nolte, 2010). As the system moves with time, a path is ‘traced’ within phase space, known as the phase space trajectory (Nolte, 2010). Every degree of freedom of the system is represented as a dimension of

Figure 1: Diagrammatical representation of a pendulum system. (After 'Double Pendulum', 2020).

the multidimensional phase space (Nolte, 2010). In the case of a pendulum, these dimensions are 𝜃𝜃𝜃𝜃1 , 𝜃𝜃𝜃𝜃2 , 𝜃𝜃𝜃𝜃1̇ , 𝜃𝜃𝜃𝜃2̇ (Levien & Tan, 1993), using the convention 𝜕𝜕𝜕𝜕 𝑓𝑓𝑓𝑓̇ = [𝑓𝑓𝑓𝑓(𝑡𝑡𝑡𝑡)]. 𝜕𝜕𝜕𝜕𝜕𝜕𝜕𝜕

Science Extension Journal • 41

Scientific Research in School Volume 3 Issue 1 2021

Chaotic Motion and the Lyapunov Exponent The Lyapunov exponent (𝜆𝜆𝜆𝜆 in Equation 1) has proven to be the most useful quantification of chaos (Guan, 2014), and as such was used to quantify chaos in this research. A system is chaotic when 𝜆𝜆𝜆𝜆 is > 0, and is periodic (motion repeated at set intervals) when 𝜆𝜆𝜆𝜆 is < 0 (Wolf et al., 1985). Qualitatively, chaos is the physical phenomenon where a dynamic system is highly dependent on its initial conditions, and its motion is seemingly random (Gupta et al., 2014). 𝜆𝜆𝜆𝜆 is defined as the average exponential rate of divergence of infinitesimally close orbits in phase space (Wolf et al., 1985). Infinitesimally close orbits within phase space correspond to nearly identical physical states, hence an exponential divergence of these orbits implies a rapid loss of predictability of the system (Shivamoggi, 1997). 1 ‖𝛿𝛿𝛿𝛿Ζ0 (𝑡𝑡𝑡𝑡)‖ 𝜆𝜆𝜆𝜆 = lim � lim � ln �� 𝜕𝜕𝜕𝜕→∞ ‖𝛿𝛿𝛿𝛿Ζ0 ‖→0 𝑡𝑡𝑡𝑡 ‖𝛿𝛿𝛿𝛿Ζ0 ‖

(1)

Formal definition of the Lyapunov exponent for a dynamic system. (Source: Wolf et al., 1985).

Danforth’s algorithm[1] (2017) which determines 𝜆𝜆𝜆𝜆 (summarised by Equation 2) has been used to quantify the chaos of a pendulum (Gupta et al., 2014; Levien & Tan, 1993; Skokos, 2009). Despite the studies’ use of Danforth’s algorithm, none of them presented a complete and easily repeatable method for the algorithm. Therefore, this paper includes a repeatable summary of Danforth’s algorithm for calculating 𝜆𝜆𝜆𝜆 of a pendulum in Part 3 of the Methodology, with the algorithm being generalised to any dynamic system in Appendix 1. 𝜕𝜕𝜕𝜕

1 𝜆𝜆𝜆𝜆𝑖𝑖𝑖𝑖 (𝑡𝑡𝑡𝑡) = � ln�𝑦𝑦𝑦𝑦⃑𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 � 𝑡𝑡𝑡𝑡

(2)

𝑛𝑛𝑛𝑛=1

Equation of the i-th largest Lyapunov exponent as a function of time. (Source: Danforth, 2017).

A visual interpretation of this algorithm in two dimensions can be seen in Appendix 2 (Figure 4, 5 and 6). This provides an intuition for the mathematical processes that are undergone. As per the details of Danforth’s algorithm in Appendix 1, Galloway & Macaskill (2009) suggests that in order

for 𝜆𝜆𝜆𝜆 of a pendulum to be calculated, 𝛼𝛼𝛼𝛼 = 𝜃𝜃𝜃𝜃1 , 𝜃𝜃𝜃𝜃2 , 𝜃𝜃𝜃𝜃̇1 , 𝜃𝜃𝜃𝜃2̇ , 𝛼𝛼𝛼𝛼 ��⃑ with the set �𝜖𝜖𝜖𝜖𝑦𝑦𝑦𝑦 0 � being the set of column vectors in Equation 3.

𝜖𝜖𝜖𝜖 0 0 0 0 𝜖𝜖𝜖𝜖 0 0 lim �� , � � , � � , � �� 0 0 𝜖𝜖𝜖𝜖 0 𝜖𝜖𝜖𝜖→0 0 0 0 𝜖𝜖𝜖𝜖

(3)

Column vector set for displacement vectors at the limit 𝜖𝜖𝜖𝜖 → 0

The importance of Gram-Schmidt orthonormalisation is to ensure that the displacement vectors do not collapse onto the dominant eigenvectors of the system, which increases the uncertainty of the 𝜆𝜆𝜆𝜆 calculation (Danforth, 2017; Galloway & Macaskill, 2009). The Principle of Least Action Rather than utilising Newton’s second law of motion 𝑑𝑑𝑑𝑑𝑝𝑝𝑝𝑝⃑ (𝐹𝐹𝐹𝐹⃑ = ), the principle of least action was used to 𝑑𝑑𝑑𝑑𝜕𝜕𝜕𝜕

formulate the pendulum’s simulation and EulerLagrange equations (Gray, 2009). Feynman’s (1963) definition of the principle of least action is “the average kinetic energy less the average potential energy is as little as possible for the path of an object going from one point to another”. The action functional 𝑆𝑆𝑆𝑆𝑖𝑖𝑖𝑖 of a pendulum is: 𝜕𝜕𝜕𝜕1

�

𝜕𝜕𝜕𝜕0

1 2 𝑚𝑚𝑚𝑚𝜃𝜃𝜃𝜃𝚤𝚤𝚤𝚤̇ − 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝜃𝜃𝜃𝜃𝑖𝑖𝑖𝑖 𝑑𝑑𝑑𝑑𝑡𝑡𝑡𝑡 2

(4)

Action functional between the time period 𝑡𝑡𝑡𝑡0 and 𝑡𝑡𝑡𝑡1 for a pendulum system. (Source: Gray, 2009).

This functional is relatively simple to compute numerically compared to the forces and acceleration of the masses. The actual path that is taken by the masses is that which minimises the action integral (Grey, 2009; Feynman, 1963). One consequence of this is the Euler-Lagrange equation:). 𝑑𝑑𝑑𝑑 𝜕𝜕𝜕𝜕ℒ 𝜕𝜕𝜕𝜕ℒ − =0 𝑑𝑑𝑑𝑑𝑡𝑡𝑡𝑡 𝜕𝜕𝜕𝜕𝜃𝜃𝜃𝜃𝚤𝚤𝚤𝚤̇ 𝜕𝜕𝜕𝜕𝜃𝜃𝜃𝜃𝑖𝑖𝑖𝑖

(5)

The Euler-Lagrange equations for a pendulum system. (Source: Deyst, 2003)

________________

This algorithm has been adapted from Danforth’s lecture Numerical Calculation of Lyapunov Exponents and his lecture notes 5.2 Numerical Calculation of Lyapunov Exponents. However, the algorithm had been well-researched before the lecture was presented in 2017, and has been used to calculate 𝜆𝜆𝜆𝜆 previously.

[1]

42 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Extant Research Testing the Chaos of the Planar Double Pendulum Extant research into the pendulum has predominantly been through the use of a computational simulation. This is due to the pendulum’s sensitivity to its initial conditions, hence making it very challenging for a built model to undergo a valid testing method that can be reliably repeated. Biglari & Jami (2016) provide information regarding the Kolmogorov–Arnold–Moser theorem. This theorem must be considered as it suggests that for certain initial conditions, the system may exhibit quasi-periodic motion, which is neither periodic nor chaotic. The theorem states that at low energies (in the region of length ratio 1: 1 to 1: 3), the pendulum system’s EulerLagrange equations may be integrable, meaning that if the phase space trajectory is subjected to a weak nonlinear perturbation, a portion of the invariant torus survives. This torus is the topological surface on which the phase space trajectory is bounded. Hence in this investigation, the motion of the pendulum near the transition point was investigated for possible quasiperiodicity, which can be seen if 𝜆𝜆𝜆𝜆 falls within the approximate range of 0 ± 0.05.

A study into the pendulum system by Stachowiak & Okada (2006) analysed the chaos of the system through the Lyapunov exponent. The study chose to investigate the dynamics of a pendulum in regard to its total energy 𝐸𝐸𝐸𝐸, and provided the knowledge that there is a clear boundary between periodic and chaotic motion at 𝐸𝐸𝐸𝐸 ≈ 4.46. This suggests that there are specific characteristics of a pendulum that makes it chaotic. Levien & Tan’s 1993 research provides valuable information on 𝜆𝜆𝜆𝜆 as the initial angle increases. It was found that the system is chaotic if 𝜃𝜃𝜃𝜃1 (0) is > 𝜋𝜋𝜋𝜋/3. This again showcases a specific characteristic of the pendulum system that makes it chaotic.

Gupta et al. (2014) explores the chaotic behaviour of a pendulum numerically. The simulation used by Gupta et al. was a MATLAB simulation, allowing them to measure how the mass and length ratios influenced the chaos of the system. It found that 𝜆𝜆𝜆𝜆 increases when the mass ratio is increased. It was also found that 𝜆𝜆𝜆𝜆 increases when the length ratio is increased, with the system being periodic at length ratio 1: 1 and chaotic at 1: 3. However, the researchers did not find a more precise length ratio at which the system transitions from periodic to chaotic motion. This research was designed as a follow on to Gupta et al.’s paper, with the goal being to increase the precision of the measured length ratio representing the transitional point between periodic and chaotic motion, referred to as the ‘transitional length ratio’ in this research.

This research is important in regard to controlling and improving dynamic systems that are derived from pendulums, such as double-armed robots (Behera & Kar, 2009). With the increase in robotics in industry, double pendulums have become a critical facet of manufacturing. Knowing when a double pendulum can produce chaotic motion will prove to be important in understanding and optimising pendulum-based robotic manufacturing systems.

Scientific Research Question As the length ratio of a planar double pendulum increases (with initial small angle displacements), at what precise length ratio does the system transition from periodic to chaotic motion?

Scientific Hypothesis That the length ratio at which a planer double pendulum system transitions from periodic to chaotic motion can be more precisely determined within the bound of 1: 1 and 1: 3 as established by extant research.

Methodology

Part 1: Modelling the Dynamics of a Pendulum The reasoning for this modelling was to determine the Euler-Lagrange equations of a pendulum system. These two equations (one for each mass) govern the dynamics of the masses, and formed the basis of the computational simulation. Some simplification steps have been omitted in the modelling for the sake of brevity, but all equations are accurate to the dynamics of the pendulum system. The key initial conditions that must be defined for this system are the length of the pendulums’ arm (𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 in metres), the point masses’ mass (𝑚𝑚𝑚𝑚𝑖𝑖𝑖𝑖 in kilograms) and the angular displacement from the vertical of the two masses (𝜃𝜃𝜃𝜃𝑖𝑖𝑖𝑖 in radians), where 𝑖𝑖𝑖𝑖 = 1, 2 indexing the two point masses. These are labelled in Figure 1. The Lagrangian ℒ for a system is known to be equal to: ℒ = 𝐾𝐾𝐾𝐾1 + 𝐾𝐾𝐾𝐾2 − 𝑈𝑈𝑈𝑈1 − 𝑈𝑈𝑈𝑈2 1 1 2 2 ∴ ℒ = (𝑚𝑚𝑚𝑚1 + 𝑚𝑚𝑚𝑚2 )𝐿𝐿𝐿𝐿1 2 𝜃𝜃𝜃𝜃1̇ + 𝑚𝑚𝑚𝑚2 𝐿𝐿𝐿𝐿2 2 𝜃𝜃𝜃𝜃2̇ 2 2 +𝑚𝑚𝑚𝑚2 𝐿𝐿𝐿𝐿1 𝐿𝐿𝐿𝐿2 𝜃𝜃𝜃𝜃1̇ 𝜃𝜃𝜃𝜃2̇ 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝜃𝜃1 + 𝜃𝜃𝜃𝜃2 ) + 𝑚𝑚𝑚𝑚(𝑚𝑚𝑚𝑚1 + 𝑚𝑚𝑚𝑚2 )𝐿𝐿𝐿𝐿1 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝜃𝜃1 + 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚2 𝐿𝐿𝐿𝐿2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝜃𝜃2

(6)

Using the Euler-Lagrange equation presented in Equation 5, the equations of motion of the two masses can be obtained:

Science Extension Journal • 43

Scientific Research in School Volume 3 Issue 1 2021

3. 𝑆𝑆𝑆𝑆1 = 𝐿𝐿𝐿𝐿1 �𝜃𝜃𝜃𝜃2̈ 𝐿𝐿𝐿𝐿2 𝑚𝑚𝑚𝑚2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝜃𝜃1 − 𝜃𝜃𝜃𝜃2 ) 2

+ �𝜃𝜃𝜃𝜃2̇ � 𝐿𝐿𝐿𝐿2 𝑚𝑚𝑚𝑚2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠(𝜃𝜃𝜃𝜃1 − 𝜃𝜃𝜃𝜃2 )

+ (𝑚𝑚𝑚𝑚1 +𝑚𝑚𝑚𝑚2 )�𝑚𝑚𝑚𝑚 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃1

4. (7)

+ 𝐿𝐿𝐿𝐿1 𝜃𝜃𝜃𝜃1̈ � � = 0 2

𝑆𝑆𝑆𝑆2 = 𝐿𝐿𝐿𝐿2 𝑚𝑚𝑚𝑚2 �−�𝜃𝜃𝜃𝜃1̇ � 𝐿𝐿𝐿𝐿1 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠(𝜃𝜃𝜃𝜃1 − 𝜃𝜃𝜃𝜃2 ) + 𝜃𝜃𝜃𝜃1̈ 𝐿𝐿𝐿𝐿1 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝜃𝜃1 − 𝜃𝜃𝜃𝜃2 ) + 𝜃𝜃𝜃𝜃2̈ 𝐿𝐿𝐿𝐿2

(8)

+ 𝑚𝑚𝑚𝑚(𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃2 )� = 0

Due to their non-linear nature, there is no known method that solves the Euler-Lagrange equations analytically. However, they can be computed numerically using a computational program, one example being the dsolve{ } function in the Maplesoft computational simulator (Salisbury & Knight, 2002) which was used in this research. This method, however can provide some uncertainty within the Lyapunov exponent calculation as a numerical solution is not an exact solution to the differential equations. Part 2: Computing the Transitional Length Ratio Using the Bisection Method In order to find precisely the transitional length ratio (where the pendulum transitions from periodic to chaotic motion), the bisection method was used. This method has not been used previously in research into a pendulum’s dynamics, but is a common method for finding the zeros of polynomials. For this research, this method can be thought of as trying to find the length ratio that makes 𝜆𝜆𝜆𝜆 as close to 0 as possible ie. the length ratio’s zero. During Test 1, the known transitional length ratio bound is between 1: 1 and 1: 3 as established from extant research (Gupta et al., 2014). The length ratio halfway between this bound (ie. 1: 2) will be tested and determined to be either chaotic or periodic. This will set a new bound for the transitional length ratio. The length ratio halfway between the new bound will then be tested, ‘telescoping’ the transitional length ratio to its precise value after repeating multiple times.

Part 3: Steps Taken to Calculate The Lyapunov Exponent for Differing Length Ratios Using Danforth’s Algorithm 1.

Maplesoft computational program was generated to simulate the motion of a double pendulum system, using the Euler-Lagrange equations 𝑆𝑆𝑆𝑆1 and 𝑆𝑆𝑆𝑆2 , the initial conditions in Table 1 and the dsolve{ } function. ��⃑0 was defined as the vector Within the program, 𝑣𝑣𝑣𝑣 representing the initial conditions of the pendulum in the phase space of the pendulum system.

44 • Science Extension Journal

6. 7. 8.

Within the program, ��⃑ 𝜀𝜀𝜀𝜀𝑦𝑦𝑦𝑦𝑛𝑛𝑛𝑛𝛼𝛼𝛼𝛼 (𝛼𝛼𝛼𝛼 = 𝜃𝜃𝜃𝜃1 , 𝜃𝜃𝜃𝜃2 , 𝜃𝜃𝜃𝜃̇1 , 𝜃𝜃𝜃𝜃2̇ ) was defined as the basis displacement vectors of the four dimensions of the phase space at the limit as 𝜖𝜖𝜖𝜖 → 0. Using the simulating program, the five conditions 𝛼𝛼𝛼𝛼 ��⃑ (𝑣𝑣𝑣𝑣 ��⃑, 𝑛𝑛𝑛𝑛 𝜀𝜀𝜀𝜀𝑦𝑦𝑦𝑦𝑛𝑛𝑛𝑛 ) were iterated for a small time step (0.01 𝛼𝛼𝛼𝛼 ��⃑ seconds), generating ��⃑ 𝑣𝑣𝑣𝑣𝑛𝑛𝑛𝑛+1 and the four 𝑘𝑘𝑘𝑘 𝑛𝑛𝑛𝑛+1 . The largest magnitude of the difference between the 𝛼𝛼𝛼𝛼 ��⃑ ��⃑ was recorded, ie. vectors 𝑣𝑣𝑣𝑣 𝑛𝑛𝑛𝑛+1 and 𝑘𝑘𝑘𝑘𝑛𝑛𝑛𝑛+1 𝛼𝛼𝛼𝛼 𝛼𝛼𝛼𝛼 ��⃑ ��⃑ �𝑣𝑣𝑣𝑣 ��⃑ � = �𝑦𝑦𝑦𝑦 �. 𝑛𝑛𝑛𝑛+1 – 𝑘𝑘𝑘𝑘 𝑛𝑛𝑛𝑛+1

𝑛𝑛𝑛𝑛+1

𝛼𝛼𝛼𝛼 The four ��⃑ 𝑘𝑘𝑘𝑘𝑛𝑛𝑛𝑛+1 were orthonormalised using GramSchmidt orthonormalisation, generating the four 𝛼𝛼𝛼𝛼 ��⃑ 𝜀𝜀𝜀𝜀𝑦𝑦𝑦𝑦 𝑛𝑛𝑛𝑛+1 . Steps 4 to 6 were repeated for 150 seconds (ie. 𝑠𝑠𝑠𝑠 = 15 000) and 𝜆𝜆𝜆𝜆(𝑡𝑡𝑡𝑡) was calculated utilising Equation 2. If 𝜆𝜆𝜆𝜆 was positive (ie. the system is chaotic), Steps 1 to 7 were repeated for the length ratio halfway between the tested ratio and the closest known ratio that produces periodic motion; if 𝜆𝜆𝜆𝜆 was negative (ie. the system is periodic), Steps 1 to 7 were repeated for the length ratio halfway between the tested ratio and the closest known ratio that produces chaotic motion. Steps 1 to 8 were repeated six times, changing the length ratio each time as outlined in Step 8.

Table 1: Initial conditions used for the computational simulation.

Description

Length of the top pendulum arm Length of the bottom pendulum arm Initial angular displacement of Mass 1 (in radians) Initial angular displacement of Mass 2 (in radians) Initial angular velocity of Mass 1 Initial angular velocity of Mass 2 Mass of Mass 1 Mass of Mass 2 Local acceleration due to gravity

Symbol

Value

𝐿𝐿𝐿𝐿1

1𝑚𝑚𝑚𝑚

𝜃𝜃𝜃𝜃1 (0)

0.2

𝐿𝐿𝐿𝐿2

2𝑚𝑚𝑚𝑚

𝜃𝜃𝜃𝜃2 (0)

0.2828

𝜃𝜃𝜃𝜃2̇ (0)

0𝑐𝑐𝑐𝑐 −1

𝜃𝜃𝜃𝜃1̇ (0) 𝑚𝑚𝑚𝑚1

𝑚𝑚𝑚𝑚2 𝑚𝑚𝑚𝑚

0𝑐𝑐𝑐𝑐 −1 1𝑘𝑘𝑘𝑘𝑚𝑚𝑚𝑚

1𝑘𝑘𝑘𝑘𝑚𝑚𝑚𝑚

−9.8𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 −2

Results The Lyapunov exponent time series of each length ratio was generated within the Maplesoft simulation, producing the plots in Figure 2A-G. These graphs show the value of 𝜆𝜆𝜆𝜆 on the 𝑥𝑥𝑥𝑥-axis plotted against time 𝑡𝑡𝑡𝑡 on the

Scientific Research in School Volume 3 Issue 1 2021

A Test 1 1: 2

B Test 2 1: 2.5

C Test 3 1: 2.25

D Test 4 1: 2.375

E Test 5 1: 2.3125

F Test 6 1: 2.34375

G Test 7 1: 2.35937

𝜆𝜆𝜆𝜆: Negative

Interpretation: Periodic Action: Increase 𝐿𝐿𝐿𝐿2 𝜆𝜆𝜆𝜆: Positive

Interpretation: Chaotic Action: Decrease 𝐿𝐿𝐿𝐿2 𝜆𝜆𝜆𝜆: Negative

Interpretation: Periodic Action: Increase 𝐿𝐿𝐿𝐿2 𝜆𝜆𝜆𝜆: Positive

Interpretation: Chaotic Action: Decrease 𝐿𝐿𝐿𝐿2 𝜆𝜆𝜆𝜆: Negative

Interpretation: Periodic Action: Increase 𝐿𝐿𝐿𝐿2

𝜆𝜆𝜆𝜆: Negative

Interpretation: Periodic Action: Increase 𝐿𝐿𝐿𝐿2

𝜆𝜆𝜆𝜆: Undetermined

Interpretation: QuasiPeriodic Action: Stop Tests

Figure 2 (A-G): Lyapunov exponent time series for initial length ratio of 1: 2 (Figure 2A) with iterative steps to length ratio 1: 2.359375 (Figure 2G), where the sign of the Lyapunov exponent is undetermined. Science Extension Journal • 45

Scientific Research in School Volume 3 Issue 1 2021

Table 2: Lyapunov exponent values for each length ratio evaluated through the Maplesoft computational simulation. Unlike Figure 2, these are arranged in order of increasing length ratio, depicting the change in 𝜆𝜆𝜆𝜆 occurring between 1: 2 and 1: 2.5.

Length Ratio

1:2

𝒕𝒕𝒕𝒕 = 𝟒𝟒𝟒𝟒𝟒𝟒𝟒𝟒

1 : 2.25

1 : 2.3135

1 : 2.34375

1 : 2.359375

1 : 2.375

1 : 2.5

𝜆𝜆𝜆𝜆(𝑡𝑡𝑡𝑡)

-3.03

-1.42

-0.26

-0.47

-0.14

-0.45

0.52

𝒕𝒕𝒕𝒕 = 𝟏𝟏𝟏𝟏𝟏𝟏𝟏𝟏𝟒𝟒𝟒𝟒

-3.22

-1.31

-0.32

-0.28

-0.02

0.56

0.49

Average

-3.18

-1.40

-0.29

-0.24

0.37

0.51

Interpretation

Periodic

0.05 QuasiPeriodic

Chaotic

𝑦𝑦𝑦𝑦-axis for each of the seven tested length ratios. While for each length ratio 𝜆𝜆𝜆𝜆 initially fluctuated (even between positive and negative values) it then settled to a more consistent value which was observed and recorded to characterise the motion of the system. More specifically, it was determined whether the motion is periodic (𝜆𝜆𝜆𝜆 < 0) and therefore if the length of the second arm was to be increased, or chaotic (𝜆𝜆𝜆𝜆 > 0) and therefore the length of the second arm was to be decreased. The plots follow the chronological order of the length ratios which were tested, and demonstrate the process by which the transitional length ratio was ‘telescoped’ to a more precise measurement. This bisection process of varying the length ratio based on 𝜆𝜆𝜆𝜆 continued until there was uncertainty in whether 𝜆𝜆𝜆𝜆 was positive or negative, ie. the system was producing quasi-periodic motion. Visual Analysis It can be observed that Figure 2A, 2C, 2E and 2F (on the following page) have a negative Lyapunov exponent (𝜆𝜆𝜆𝜆). In contrast, Figure 2B and 2D show 𝜆𝜆𝜆𝜆 to be positive. At length ratio 1: 2.359375 (Figure 2G) it cannot be determined whether 𝜆𝜆𝜆𝜆 is positive or negative with certainty. Numerical Analysis The average value of 𝜆𝜆𝜆𝜆 between the time period of 40s to 150s was found within the Maplesoft computational program, using the integral in Equation 9. The time period of 40s to 150s was chosen as 𝜆𝜆𝜆𝜆(𝑡𝑡𝑡𝑡) becomes relatively stable at 𝑡𝑡𝑡𝑡 = 40, and the Maplesoft simulation could not compute 𝜆𝜆𝜆𝜆(𝑡𝑡𝑡𝑡) for values > ~150. The average value, the value at 𝑡𝑡𝑡𝑡 = 40 and the value at 𝑡𝑡𝑡𝑡 = 150 have been summarised in Table 2. 150

� 𝜆𝜆𝜆𝜆(𝑡𝑡𝑡𝑡)𝑑𝑑𝑑𝑑𝑡𝑡𝑡𝑡 ÷ 110

(9)

Expression to find the average value of 𝜆𝜆𝜆𝜆(𝑡𝑡𝑡𝑡) between the time period 40𝑐𝑐𝑐𝑐 to 150𝑐𝑐𝑐𝑐 46 • Science Extension Journal

Using the data in Table 2, Figure 3 shows the relationship between the average Lyapunov exponent and the length ratio, expressed as the fraction 𝐿𝐿𝐿𝐿2 ÷ 𝐿𝐿𝐿𝐿1 .

Figure 3: A plot of the average Lyapunov exponent as a function of the length ratio (expressed as the fraction 𝐿𝐿𝐿𝐿2 ÷ 𝐿𝐿𝐿𝐿1 ). It is important to note that 𝜆𝜆𝜆𝜆 and 𝐿𝐿𝐿𝐿2 ÷ 𝐿𝐿𝐿𝐿1 are both dimensionless quantities, hence no units are required.

Discussion As per the literature review, if 𝜆𝜆𝜆𝜆 is < 0, the system is periodic, and if 𝜆𝜆𝜆𝜆 is > 0, the system is chaotic (Wolf et al., 1985). From the visual and numerical analysis of the Lyapunov exponent times series, between the length ratios 1: 2 to 1: 2.34375, 𝜆𝜆𝜆𝜆 was negative and so the pendulum system was periodic. Furthermore, it was shown both visually and numerically that at length ratios between 1: 2.375 and 1: 2.5, the pendulum system was chaotic as 𝜆𝜆𝜆𝜆 was > 0. It can be inferred that the transitional length ratio lies between the length ratio of 1: 2.34375 (the upper bound of periodic motion) and 1: 2.375 (the lower bound of chaotic motion). This represents an improvement in precision of determining the transitional length ratio by a factor of 64 times in comparison to extant research (Gupta et al., 2014). As there was uncertainty in whether 𝜆𝜆𝜆𝜆 was positive or negative in Figure 2G, it was concluded that at the length ratio 1: 2.359375, the pendulum produced quasiperiodic motion.

Scientific Research in School Volume 3 Issue 1 2021

From the plot in Figure 3, it can be observed that there was a positive association between the Lyapunov exponent and the length ratio, however an exact linear correspondence between the two variables was not evident from the data. One reason this could arise is due to errors within the Lyapunov exponent calculation. However, this was not likely to be the cause of this nonlinear correspondence, as the computation simulation provided a numerical solution of the Euler-Lagrange equations that were accurate to 1 part per 106 (Maplesoft, 2012). The exact uncertainties of the Lyapunov exponent calculations are quite hard to derive, however they could be investigated in future research. Another explanation for this non-linear correspondence is that the two variables (length ratio and 𝜆𝜆𝜆𝜆) are correlated through a third variable, which may cause a change in both 𝜆𝜆𝜆𝜆 and the length ratio.

One possible candidate of this third variable is the total energy 𝐸𝐸𝐸𝐸 of the system, which is increased as the length ratio increases (Stachowiak & Okada, 2006). Furthermore, there is evidence that there is a clear boundary between periodic and chaotic motion at 𝐸𝐸𝐸𝐸 ≈ 4.46 (Stachowiak & Okada, 2006). 𝐸𝐸𝐸𝐸 is generally expressed as ∑𝑖𝑖𝑖𝑖[𝑈𝑈𝑈𝑈𝑖𝑖𝑖𝑖 + 𝐾𝐾𝐾𝐾𝑖𝑖𝑖𝑖 ] (Boundless Physics, n.d.). As the system is a Hamiltonian (Assencio, n.d.), 𝐸𝐸𝐸𝐸 stays constant throughout the motion of the pendulum. Furthermore within this research, at 𝑡𝑡𝑡𝑡 = 0, 𝐾𝐾𝐾𝐾 = 0. Because

𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑

𝑑𝑑𝑑𝑑𝐿𝐿𝐿𝐿2

= −𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚2 cos 𝜃𝜃𝜃𝜃2 = 9.41 > 0, it is clear

that when the length ratio is increased, the total energy of the system also increased. It is proposed that the more energy the pendulum system has, the more likely it will be chaotic. This is because the phase space velocity will have a larger magnitude, and hence a slight perturbation to the phase space trajectory will have a larger proportional influence on the system. This may cause the phase space trajectory to diverge from its original path, ie. produce chaotic motion (Wolf et al., 1985).

Further research into the planar double pendulum might investigate the total energy of the system in two ways. The length ratio could be varied while ensuring that the total energy of the system stays constant throughout the tests. This can be achieved by changing a variety of variables (𝑚𝑚𝑚𝑚, 𝜃𝜃𝜃𝜃, 𝜃𝜃𝜃𝜃̇). If 𝜆𝜆𝜆𝜆 remains constant when the length ratio is changed and the total energy of the system is kept constant, it can be proposed that the length ratio is not the cause of the changing 𝜆𝜆𝜆𝜆 observed in this research. However, a relationship between 𝐸𝐸𝐸𝐸 and 𝜆𝜆𝜆𝜆 would also need to be investigated. This can be done by keeping the length ratio constant and varying 𝐸𝐸𝐸𝐸. A proposed method would be to provide one mass with differing initial angular velocity, rather than the zero initial angular velocity that was used in this research.

It is most likely that the reason for a pendulum’s chaotic motion is a combination of all the factors discussed above, however this is not yet clear from known research (Chen, 2008). Finally, this research only focused on length ratios between 1: 1 and 1: 3. However at the limit as 𝐿𝐿𝐿𝐿2 → ∞, the planar double pendulum system can be thought of as a planar pendulum system (ie. only one mass on one rod), which is a periodic system (Parks, 2000). This suggests there is another transitional length ratio, where the pendulum transitions from chaotic to periodic motion. A conclusion that can be drawn from this is that there may be a finite range of length ratios of a double pendulum system that produce chaotic motion, which could be investigated in future research. To summarise, it is proposed that the increase in length ratio may not solely be the cause of the increase in 𝜆𝜆𝜆𝜆. Other qualities of the system, specifically the total energy 𝐸𝐸𝐸𝐸, should now be researched in order to determine if there are additional factors influencing the system’s chaotic motion.

Conclusion My research project explored the transitional length ratio between periodic and chaotic motion of a planar double pendulum system. Through the use of a computational simulation of a double pendulum, the chaos of the system was quantified and analysed through calculating the Lyapunov exponent (the accepted measure of chaotic motion). Previous research determined that the transitional length ratio lies between the bound of 1: 1 and 1: 3. The bisection method was used to increase the precision of this measurement, and it was determined that the transitional length ratio occurs between 1: 2.34375 and 1: 2.375, improving the precision of this measurement by a factor of 64 times. In doing so, my hypothesis was supported, that is, the precision of the transitional length ratio value can be increased from the bound established in current research.

Acknowledgements I would like to thank Dr Matthew Hill for providing assistance and advice throughout the duration of my project. I would also like to thank Professor Nusantara of the State University of Malang for sharing his code that generated the displacement initial condition, without which the calculation of the Lyapunov exponent would not have been possible. I would like to thank Boyd Carruthers and Jeremy von Einem for reviewing my paper and providing valuable feedback.

Science Extension Journal • 47

Scientific Research in School Volume 3 Issue 1 2021

References Assencio, D n.d., ‘The double pendulum: Lagrangian formulation - Diego Assencio’ [WWW Document], https://diego.assencio.com/?index=1500c66ae7ab27bb01064 67c68feebc6. Biglari, H & Jami, A 2016, ‘The Double Pendulum Numerical Analysis with Lagrangian and the Hamiltonian Equations of Motion’, S. International Conference on Mechanical and Aerospace Engineering, London, United Kingdom. Boundless Physics n.d., ‘Potential Energy and Conservation of Energy’ [WWW Document], https://courses.lumenlearning.com/boundlessphysics/chapter/potential-energy-and-conservation-ofenergy/. Chen, J 2008, ‘Chaos from Simplicity: An Introduction to the Double Pendulum’, Department of Mathematics and Statistics, College of Engineering, University of Canterbury. Danforth, C 2017, Lecture 21: Numerical calculation of Lyapunov exponents, online video, 7 April, https://youtu.be/YL4twBVKNK0. Deyst, H 2003, ‘Lecture #7 Lagrange’s Equations’, Massachusetts Institute of Technology. ‘Double Pendulum’ 2020, Wikipedia, wiki https://en.wikipedia.org/wiki/Double_pendulum.

article,

Feynman, R, Leighton, R & Sands, M 1963, The Feynman lectures on physics Vol II. Reading, Mass, Addison-Wesley Pub. Co. Gray, C 2009, ‘Principle of least action’, Scholarpedia, 4(12):8291. Guan, K (n.d.). Important Notes on Lyapunov Exponents. Arxiv. [Preprint]. arXiv:1401.3315 [nlin].

48 • Science Extension Journal

Gupta, M, Bansal, K & Singh, A 2014, ‘Mass and length dependent chaotic behavior of a double pendulum’, IFAC Proceedings Volumes, vol. 47, pp. 297-301. Levien, R & Tan, S 1993, ‘Double pendulum: An experiment in chaos’, American Journal of Physics, vol. 61, pp. 1038– 1044. Maplesoft 2012, ‘Differential Equations in Maple 16’, Waterloo Maple Inc.. Nolte, D 2010, ‘The tangled tale of phase space’, Physics Today, vol. April 2010, pp. 33-38. Parks, J 2000, ‘The Simple Pendulum’, Department of Physics and Astronomy, The University of Tennessee. Richter, P & Scholz, H, 1984. ‘Chaos in classical mechanics: The double pendulum’ Stochastic Phenomena and Chaotic Behaviour in Complex Systems, vol. 21, pp. 86–97. Safitri, A, Nusantara, T & Chandra, T 2020b. ‘Factors of length ratio and mass at chaos of double pendulum system’, AIP Conference Proceedings, pp. 070016, DOI 10.1063/5.0000582. Salisbury, K & Knight, D 2002, ‘The multiple pendulum problem via Maple®’, International Journal of Mathematical Education in Science and Technology, vol. 33, pp. 747–755, DOI 10.1080/002073902320602905. Stachowiak, T & Okada, T 2006b ‘A numerical analysis of chaos in the double pendulum’, Chaos, Solitons & Fractals, vol. 29, pp. 417–422. Wolf, A, Swift, J, Swinney, H & Vastano, J 1985, ‘Determining Lyapunov exponents from a time series’, Physica D: Nonlinear Phenomena, vol. 16, pp. 285–317, DOI 10.1016/0167-2789(85)90011-9.

Scientific Research in School Volume 3 Issue 1 2021

Appendix 1 Summary of Danforth’s (2017) Algorithm for Computing the Lyapunov Exponent

Define ��⃑ 𝑣𝑣𝑣𝑣0 as the vector representing the initial condition of the system in an 𝑖𝑖𝑖𝑖-th dimensioned phase space (ie. ��⃑0 ∈ ℝ𝑖𝑖𝑖𝑖 ). 𝑣𝑣𝑣𝑣 �01 , 𝑤𝑤𝑤𝑤 �02 , 𝑤𝑤𝑤𝑤 �03 , … 𝑤𝑤𝑤𝑤 �0𝑖𝑖𝑖𝑖 �[2] with centre ��⃑ 𝑣𝑣𝑣𝑣0 (see Figure 5). Take a unit ball 𝑈𝑈𝑈𝑈0 in ℝ𝑖𝑖𝑖𝑖 defined by the orthonormal basis set �𝑤𝑤𝑤𝑤 𝛼𝛼𝛼𝛼 ��⃑ � (𝛼𝛼𝛼𝛼 = 1,2,3, … 𝑖𝑖𝑖𝑖) as the sum of ��⃑ 𝑣𝑣𝑣𝑣0 and 𝑤𝑤𝑤𝑤 �0𝛼𝛼𝛼𝛼 . Define the set �𝜖𝜖𝜖𝜖𝑦𝑦𝑦𝑦

𝛼𝛼𝛼𝛼 ��⃑ ��⃑1 and the set �𝑘𝑘𝑘𝑘 𝑈𝑈𝑈𝑈0 into an ellipsoid with centred at 𝑣𝑣𝑣𝑣 1 � lying on the ellipsoid’s surface (see Figure 4 and Figure 6).

1. 2. 3.

6. 7. 8. 9.

𝛼𝛼𝛼𝛼 𝛼𝛼𝛼𝛼 ��⃑ ��⃑ ��⃑, 𝑣𝑣𝑣𝑣1 and the set �𝑘𝑘𝑘𝑘 Iterate the 𝑖𝑖𝑖𝑖 + 1 conditions �𝑣𝑣𝑣𝑣 0 �𝜖𝜖𝜖𝜖𝑦𝑦𝑦𝑦0 �� for a small time step, generating ��⃑ 1 �. This will transform

𝛼𝛼𝛼𝛼 𝛼𝛼𝛼𝛼 𝛼𝛼𝛼𝛼 ��⃑ ��⃑ ��⃑ ��⃑1 and �𝑘𝑘𝑘𝑘 ��⃑1 – 𝑘𝑘𝑘𝑘 Record the of the difference between the vectors 𝑣𝑣𝑣𝑣 1 �, ie. �𝑣𝑣𝑣𝑣 1 � = �𝑦𝑦𝑦𝑦1 �.

𝛼𝛼𝛼𝛼 ��⃑ �1𝛼𝛼𝛼𝛼 }. This transforms the Orthogonalise the set �𝑦𝑦𝑦𝑦 1 � using Gram-Schmidt orthonormalisation to generate the set {𝑤𝑤𝑤𝑤 ellipsoid into a unit ball (see Figure 6). �1𝛼𝛼𝛼𝛼 } with centre ��⃑. 𝑣𝑣𝑣𝑣1 Take a unit ball 𝑈𝑈𝑈𝑈1 in ℝi defined by the orthonormal basis set {𝑤𝑤𝑤𝑤 𝛼𝛼𝛼𝛼 𝛼𝛼𝛼𝛼 ��⃑� as the sum of ��⃑ 𝑣𝑣𝑣𝑣1 and 𝑤𝑤𝑤𝑤 �1 . Define the set �𝜖𝜖𝜖𝜖𝑦𝑦𝑦𝑦 1

Repeat Steps 4 to 8 𝑡𝑡𝑡𝑡 times, and utilise Equation 2 to calculate 𝜆𝜆𝜆𝜆𝑖𝑖𝑖𝑖 (𝑡𝑡𝑡𝑡).

________________ Subscript indexes iteration; superscript indexes dimension of decreasing expansion direction in the system’s phase space vector space [2]

Science Extension Journal • 49

Scientific Research in School Volume 3 Issue 1 2021

Appendix 2 – Visual Interpretation of Danforth’s Algorithm in Two Dimensions

Figure 4: Danforth’s Lyapunov exponent numerical calculation algorithm in ℝ2 after one iteration. (After: Danforth, 2017).

Figure 5: Unit ball 𝑈𝑈𝑈𝑈0 in ℝ2 . (After: Danforth, 2017). 50 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

𝛼𝛼𝛼𝛼 ��⃑ Figure 6 :Red ellipsoid represents the transformed 𝑈𝑈𝑈𝑈0 unit ball into an ellipsoid with centre 𝑣𝑣𝑣𝑣 ��⃑1 and set �𝑦𝑦𝑦𝑦 1 � lying on the ellipsoid’s surface. 𝛼𝛼𝛼𝛼 ��⃑ Blue ellipsoid represents the ellipsoid with minor and major axes as the orthogonalised set �𝑦𝑦𝑦𝑦 1 �. 𝛼𝛼𝛼𝛼 2 Black unit ball represents 𝑈𝑈𝑈𝑈1 in ℝ with orthonormal basis set {𝑤𝑤𝑤𝑤 �1 }. (After: Danforth, 2017).

Science Extension Journal • 51

Scientific Research in School Volume 3 Issue 1 2021

52 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

How small can you go? How Iron powder can mitigate inefficiencies of a transformer’s soft iron core Francesca Buffa Barker College Transformers are devices used to distribute electrical power from power plants to our homes and further provide adequate voltage for everyday appliances. Transformers have a soft iron core to improve their efficiency, however, power is still not completely conserved, at least in part due to losses of energy in the core. Powdered soft iron transformer cores have been considered as alternative to solid laminated soft iron cores as they decrease the area over which eddy currents can flow (a key source of inefficiency in a core). However, hysteresis effects mean that if particles are too small, they can also be inefficient. In this research project three powdered soft iron cores were created by suspending three different sized powders (with diameters of 45μm, 150μm and 425μm) in an epoxy resin to insulate the iron particles. Voltage and current input and output for a transformer with each of the three cores was measured allowing for calculation of the efficiency of each core. It was found that the intermediate size (150μm) had the maximum efficiency of the three sizes supporting the hypothesis that larger and smaller particle sizes could have effects that reduce efficiency. More testing of diameter values between 45μm and 450μm would allow for a more predictive model to be fitted proposing an optimal powdered size for maximal efficiency Literature Review Electrical Transformers Transformers are devices used to step up or step-down the voltage in a circuit, without introducing a new power source. It can be explained using Faradays law of induction which states that, ‘If a terminated wire is moved so as to cut a magnetic curve, a power is called into action which tends to urge an electric current through it’ (Al-Khalili, 2015, p.6). In other words, a change in the magnetic environment of a coil of wire will cause a voltage to be induced in the wire due to electromotive force (EMF) acting on the free moving electrons in the conductor. Thus, voltage can be altered by a specific magnetic field induction. Therefore ideal transformers function with the use of a primary and secondary coil of conductive material as can be seen in Figure 1. The primary coil induces a magnetic field when a current is passed through it and when the secondary coil is placed in this magnetic field, a current will be induced in that coil. This law only applies to alternating current (AC), as it produces a constantly fluctuating magnetic field when passed through a coiled wire. When a wire is coiled multiple times, a greater magnetic field strength is produced and furthermore, the efficiency can be increased by introducing a high

Figure 1: A diagrammatical representation of a transformer which transfers power from the primary coil to the secondary coil supported by a laminated soft iron core. (Source: Hoult, 2020)

permeability material, such as iron, as a core. The magnetic field produced by the primary coil magnetises the soft iron core by aligning the magnetic domains, increasing the flux density near the secondary coil and inducing a current with great efficiency (Hurley, Wolfle & Breslin, 1998) according to Faraday-Lenz’s law, 𝜀𝜀𝜀𝜀 = −𝑁𝑁𝑁𝑁

∆𝜙𝜙𝜙𝜙 ∆𝑡𝑡𝑡𝑡

where,

𝜀𝜀𝜀𝜀 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸, 𝜙𝜙𝜙𝜙 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑

(1)

Faraday-Lenz’s Law: (Young & Freedman, 2019 p.999)

Science Extension Journal • 53

Scientific Research in School Volume 3 Issue 1 2021

Ideal transformers function with the relationship, 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = 𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆 or 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝐼𝐼𝐼𝐼𝑃𝑃𝑃𝑃 = 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝐼𝐼𝐼𝐼𝑆𝑆𝑆𝑆

(2)

where, 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆 = 𝑆𝑆𝑆𝑆𝑑𝑑𝑑𝑑𝑆𝑆𝑆𝑆𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝 = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑 𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑣𝑣𝑣𝑣𝑑𝑑𝑑𝑑 𝑉𝑉𝑉𝑉𝑠𝑠𝑠𝑠 = 𝑆𝑆𝑆𝑆𝑑𝑑𝑑𝑑𝑆𝑆𝑆𝑆𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑣𝑣𝑣𝑣𝑑𝑑𝑑𝑑 𝐼𝐼𝐼𝐼𝑃𝑃𝑃𝑃 = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐼𝐼𝐼𝐼𝑠𝑠𝑠𝑠 = 𝑆𝑆𝑆𝑆𝑑𝑑𝑑𝑑𝑆𝑆𝑆𝑆𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆

Figure 2: How laminations reduce the area over which Eddy currents can form, minimising energy loss from Eddy currents. However, as the 2018) powder particles reach a nanoscale, the (Source: Collins,

Further, the change in voltage is dependent on the ratio of coil turns in the primary coil and the secondary coil. If the secondary coil has the same number of turns as the primary, theoretically, the induced voltage will be the same, but if the secondary coil has less rotations, the voltage will drop according to the ratio so, 𝑉𝑉𝑉𝑉𝑠𝑠𝑠𝑠 =

𝑁𝑁𝑁𝑁𝑠𝑠𝑠𝑠 𝑉𝑉𝑉𝑉 𝑁𝑁𝑁𝑁𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝

(3)

where, 𝑁𝑁𝑁𝑁𝑠𝑠𝑠𝑠 = 𝑆𝑆𝑆𝑆𝑑𝑑𝑑𝑑𝑆𝑆𝑆𝑆𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑇𝑇𝑇𝑇𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓, 𝑁𝑁𝑁𝑁𝑝𝑝𝑝𝑝 = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑑𝑑 𝑇𝑇𝑇𝑇𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓

Major Transformer Losses due to Iron cores Iron cores are not perfectly efficient due to various power losses including heat, localised joints, stacking holes, slitting, hysteresis and eddy currents (Nyenhuis, Girgis & Mechler, 2001). In powdered iron cores, the major losses are hysteresis and eddy current related (Dekker, 2004).

Hysteresis losses are a result of molecules in the core resisting magnetisation from the alternating magnetic field, resulting in friction within the core and therefore heat loss. A major way to reduce this type of loss is to reduce the frequency of the alternating current (Lee1 et al., 2015).

hysteresis losses increase due to permeability of iron relating to the number of magnetic domains in the material which decreases as the size of the material decreases, (Lee1, et al., 2015) as seen in Figure 3. Therefore this project is important because while smaller particles increase efficiency by reducing eddy currents, other contributing losses increase so investigating various small particles will increase our understanding of optimal powder sizes. Eddy Currents in more detail While eddy current losses can be reduced by increasing the frequency of the power source (Leon & Semlyen, 1993) further eddy current losses are directly related to the size of the magnetic material (Collins, 2018), according to, 𝑃𝑃𝑃𝑃𝑒𝑒𝑒𝑒 =

𝐶𝐶𝐶𝐶𝐵𝐵𝐵𝐵2 𝐹𝐹𝐹𝐹2 𝑑𝑑𝑑𝑑 2 𝑝𝑝𝑝𝑝

where, d = thickness

(4)

Therefore a decrease in material size in the core will increase the efficiency of the transformer (Shokrollahi & Janghorban, 2007). Once a magnetic particle is in the nanometre size, the eddy current produced within the particle is negligibly small, however, as previously mentioned, other losses such as hysteresis increase (Trasimond & Lawrence, 1986).

Eddy current losses are caused when the induced magnetic field produces small circulating currents in the Iron core, reducing the electromotive force induced in the secondary coil and releasing energy as heat (Moses, 1998). Laminations of iron in the core reduces eddy current loss as smaller circulating currents are produced, producing less heat and therefore less energy, compared to solid iron cores, as seen in Figure 2. Therefore, the reduction of size from laminations to powder particles will reduce the eddy current loss in the core further.

54 • Science Extension Journal

Figure 3: As the particle diameter of the soft iron material decreases, the number of magnetic domains within the particle decreases and the effect of this change on the coercivity of the particle is shown. As coercivity is inversely related to the permeability, a point of increased hysteresis loss is indicated. (Source: Lee1, et al., 2015)

Scientific Research in School Volume 3 Issue 1 2021

Hysteresis losses in more detail Soft magnetic materials have high permeability. Permeability is the ease in which a magnetic substance is magnetised (Dekker 2004). Magnetic permeability is given by the ration, B/H, where B=magnetic induction, and H=magnetic force necessary to produce magnetic induction. Iron has a high permeability so produces low hysteresis losses, making it a desirable choice for transformer cores (Trasimond & Lawrence, 1986). Powdered iron cores utilise particulate iron with insulation, however when the size of the iron is reduced greatly, the material’s flux carrying ability reduces, indicating a reduction in permeability, increasing hysteresis loss (Dekker, 2004). Permeability is inversely related to coercivity and is therefore highly sensitive to size variation (Xue, 2008). As seen in figure 3, the coercivity of an iron powder particle gradually increases to a maximum value as the size decreases but as it decreases further, the coercivity rapidly decreases towards zero. This is because as a particle’s diameter decreases, a magnetic multidomain state converts to a single domain state. In the multidomain region, the increasing coercivity indicates a decreasing efficiency as it is more difficult to reverse magnetisation due to alternating current, increasing friction within the core and therefore increasing hysteresis loss (Lee1, et al., 2015). Therefore the larger particles are more desirable when producing an efficient transformer core. However once the particle reaches a single domain state, the coercivity decreases and permeability increases rapidly. Although this means that less force is required to magnetise the core, the smaller particles can only carry a very small magnetic flux, meaning that the EMF produced by the transformer is small and a greater power input, or increased number of coils, is required to fully transfer sufficient voltage. This results in wasted force, lost as hysteresis loss (Dekker, 2004). Therefore the multidomain particles are a more efficient choice, despite having a higher permeability than the particles in the single domain region. Transformer frequency Frequency is the oscillations per second of an AC current and will change depending on the load and supply in the circuit. The frequency will increase when the amount of current needed is less and decrease when the amount of current needed is more (Dixon, 2003). Transformers cannot change the frequency. The three commonly used frequencies in transformers are 50Hz, used in North America and Australia, 60Hz, used in Europe and 400Hz, used for high power applications. The frequency selected

will affect the efficiency of a transformer in different ways. When a high frequency is chosen, the hysteresis losses in the core reduce and when a low frequency is chosen the eddy current losses in the core reduce (Dixon, 2003). Hysteresis losses can be minimised by selecting the appropriate frequency for the powder size as higher frequencies of alternating current reduce hysteresis losses (Trasimond & Lawrence, 1986). As solenoid values operate at high frequencies in industry, in order to meet demands for high operating speeds, powder cores are often more efficient than laminated cores (Ueno, et al., 2016). Other Contributing Factors to Power Loss An important consideration to be made when conducting the experiment is the result of impedance due to the transformer. Impedance reduces the overall current due to an unwanted back EMF being produced by the induced magnetic field. Since AC current through a coil of wire forms a fluctuating magnetic field, magnetic field lines cut across the conductor carrying current into the transformer, producing an EMF against the electron flow in the wire. Therefore, when AC current flows through a conductor a small back EMF is induced, reducing overall current (Miller & Rabinovici, 1994) In all electric circuits, power is lost to heat as a result of resistance in the circuit. Resistance occurs due to loading in a circuit and resistance in the wire. To minimise safety concerns as well as heat loss, appropriate current, voltage and resistance must be used when testing transformer cores. Further, a low resistance wire, such as copper, should be used (Dekker, 2004). The density of the soft iron core in transformers also effects the efficiency of the core. The denser the core is, the more iron there is present in a constant volume. Higher density dictates an increased flux density in the core, resulting in a more efficient transfer of power (Shokrollahi and Janghorban, 2007). In powdered iron cores the particles must be insulated with a 1% - 3% spacing (Trasimond and Lawrence, 1986). When the iron is properly powdered, the density is 2% below the true density of solid iron, meaning that the powdered iron density must be compressed to 90% to ensure the insulation. This reduction in density reduces the efficiency of the transformer (Trasimond and Lawrence, 1986). 10% of power losses in distribution transformer cores are a result of Localised Joint losses, Stacking Holes and slitting (Nyenhuis, Girgis & Mechler, 2001). In

Science Extension Journal • 55

Scientific Research in School Volume 3 Issue 1 2021

laminated transformer cores, a minor power loss occurs due to localised joints. Loss occurs where the iron laminations form joints, a result of the longitudinal and cross loss components of the entire core. (Bernacki, Rymarski & Dyga, 2017). Another minor loss is caused by stacking holes. Holes are used to stack laminations, causing flux crowding in that area and therefore an energy loss through heat. (Nyenhuis, Girgis & Mechler, 2001). As steel is produced in 800nm-1000nm width, a steel slitter is used to cut the appropriate width for laminations, slightly reducing the magnetic properties of the material and reducing the magnetisation of the core, resulting in power loss (Nyenhuis, Girgis & Mechler, 2001). Flux distributions within an assembled laminated soft iron core can be used to predict the total losses as increased flux density increases eddy current loss (Moses, 1998). The fluxes to be monitored are rotational flux, harmonics, stress and interlaminar flux, with interlaminar flux causing the greatest concern, occurring throughout the core. As interlaminar flux depends of the coating, grain size, lamination size, resistivity and permeability, quality insulation should be utilised. This is greater emphasised as rotational flux and harmonic flux increases in the joint regions, or air gaps, of the core, further increasing power loss through eddy currents (Moses, 1998). Powdered Iron Cores in Industry Powder Iron transformer cores are in early stages of development as further research is required to create a core with optimal efficiency. However, many of their properties are already favoured over traditional laminated cores in industry, currently being commonly used in power conversions, line filter applications and radio frequency applications. The construction of Powdered Iron cores evenly distributes a mass of iron in a constant volume. This even distribution reduces the risk of a hot spot forming, and therefore improves the efficiency and safety of the transformer device. Further, Powder Cores are typically a less expensive option due to their simpler construction method. The method of construction currently used in industry also allows the shape of the core to be manipulated into various configurations, effecting the overall efficiency of the transformer, and allowing manipulation of the devices position in electrical circuits. Typically, Iron Powder cores use a range of different sized powders, with 75𝜇𝜇𝜇𝜇m diameter particles having the highest concentration. The standard Iron powder particle sizes commercially available are 425μm diameter (300 mesh), 150μm (100 mesh) 45μm diameter (40 mesh).

56 • Science Extension Journal

Scientific Research Question What effect does particle size have on the efficiency of a powdered iron core in a transformer?

Scientific Hypothesis The most efficient powder particle size will be the intermediate size of 150μm. This size has been predicted as it lies between possible outliers. According to the literature, when particle size increases losses are increased due to eddy currents and when particle size decreases, hysteresis loss increases. Therefore the intermediate particle size should perform the best.

Methodology Production of powdered Iron Cores In this experiment three different powder iron cores were made and tested, each with different sized iron powder particles. The three particle sizes used had a 425μm diameter (300 mesh), 150μm (100 mesh) 45μm diameter (40 mesh). These sizes were selected as they are the standard iron powder sizes available, resulting in a straightforward method of production of the cores. Further, as these are standard sizes, future mass production of Iron cores will be simplistic once the optimal standard is known. A standard transformer used for teaching the Physics of transformers was used for this experiment as seen in Figure 4, with the generic soft iron core used as the prototype for a mould to be produced. The core is cylindrical with a diameter of 2.0cm and length of 14cm. To construct the different cores a mould was made from plaster such that an identically shaped and size cores could be made to the generic core. Then, 50 grams of the desired powder was mixed with epoxy resin and set in the mould. A small stirring rod was used to mix the particles in the epoxy until evenly distributed throughout and the mixture.

Figure 4: (a) A student transformer with an air core which involves the secondary coil nested inside the primary coil, with room for a soft iron core to be inserted in the centre (b) the standard soft iron core used to create a mould for the powdered cores.

Scientific Research in School Volume 3 Issue 1 2021

The process of producing the cores was challenging as the mass of the iron caused issues with dispersion of the powder throughout the epoxy. The powder sank to the base of the mould due to a downward force acting due to Earth’s gravitational field. However, three cores with an equal mass of iron were produced and are presented in Figure 5, as an appropriate method was discovered through multiple tests which will be outlined below. It took three trials to achieve sufficient distribution of the iron powder throughout the core. Construction Trial 1: After the mixture was stirred resulting in an initially even distribution, the mixture was left for 24 hours to harden. When the mould was broken, it was revealed that the iron powder had sunk to the bottom and was not evenly distributed. Construction Trial 2: Once the mixture was stirred until as evenly distributed as possible, a permanent magnet was placed above the mould to prevent the powder from sinking to the bottom. However, once the mixture had dried, the powder had divided in the hardened epoxy, with half of the mass sinking to the bottom of the mould and half set at the top, not evenly distributed. Construction Trial 3: Before mixing the iron powder into the epoxy resin, the epoxy was heated with a heating source. This increased the viscosity of the epoxy when the powder was added as the hardening process had already begun. The increased viscosity prevented the iron powder from sinking to the bottom of the mould, producing a core with evenly distributed powder.

Construction Trial 3 was successful and was used to create the three moulds used in this experiment as seen in Figure 5. Data collection and analysis The efficiency of a transformer is easily quantified through data collection in experimentation. Efficiency relates to the power loss of a transformer which can be found by comparing the total power in the primary circuit and the power in the secondary circuit. Since, 𝑃𝑃𝑃𝑃 = 𝐼𝐼𝐼𝐼𝑉𝑉𝑉𝑉

𝐸𝐸𝐸𝐸𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 =

𝑃𝑃𝑃𝑃𝑝𝑝𝑝𝑝 × 100% 𝑃𝑃𝑃𝑃𝑠𝑠𝑠𝑠

(5)

(6)

When the voltage and the current in both circuits are known, the power loss and therefore the efficiency can be calculated and analysed for each test. In this experiment, four different voltages from a power supply were selected from 2-10 Volts. The primary and secondary currents and voltages were measured with an ammeter and voltmeter respectively, and the input and output power were calculated. Efficiency can be found from the gradient of the graph of input vs output power. Then a graph of efficiency vs particle size will be used to address the hypothesis that decreasing particle size only increases efficiency to a point. The frequency that has been selected for testing is 50Hz, the common output of Australian power points, providing a result with relevancy.

Results Power inputs and outputs Tables 1,2 and 3 record the measured primary and secondary voltages and currents and therefore the calculated input and output power for each of the cores. The power values are then plotted in figures 6, 7 and 8 allowing for a linear regression model to be fitted where the x-coefficient is the determined efficiency of the core. The trendlines are compared on figures 9.

Figure 5: The standard soft iron core and the powdered iron cores with powder particle diameter sizes, 45𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 (labelled 40), 150𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 (labelled 100) and 425𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 (labelled 300). Although the method did not produce exactly equal sized cores, an equal mass of insulated iron powder is in each core.

Science Extension Journal • 57

Scientific Research in School Volume 3 Issue 1 2021

Table 1: Power input and output of the 425𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle core, calculated from the primary and secondary current and voltage

1.562 3.145 4.708 6.490

0.80 1.62 2.55 3.59

1.250 5.095 12.005 23.299

0.076 0.163 0.244 0.332

0.023 0.059 0.094 0.129

0.00175 0.00962 0.02294 0.04283

0.04

Vp 1.833 3.290 4.844 6.583

425𝜇𝜇𝜇𝜇 𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 Diameter Particles

Power output (W)

0.05

y = 0.0019x - 5E-05

0.03 0.02 0.01 0

Table 3: Power input and output of the 45𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle core, calculated from the primary and secondary current and voltage

0.06

Pp 1.741 5.593 12.498 22.843

Vs 0.102 0.185 0.362 0.484

Is 0.036 0.069 0.105 0.142

Ps 0.00367 0.01276 0.03801 0.06873

45𝜇𝜇𝜇𝜇 𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 Diameter Particles y = 0.0023x - 0.0001

0.04 0.02 0

Power input (W) 0

Power input (W) Figure 6: Efficiency curve for 425𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle powder core where the gradient of the fitted line is the calculated efficiency of the core

Figure 8: Efficiency curve for 45𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle powder core where the gradient of the fitted line is the calculated efficiency of the core

Power Inputs Vs Power Outputs

Table 2: Power input and output of the 150𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle core, calculated from the primary and secondary current and voltage

Ip 0.80 1.62 2.55 3.59

Pp 1.612 5.377 12.000 21.154

Vs 0.101 0.183 0.264 0.348

Is 0.034 0.069 0.101 0.139

Ps 0.00343 0.01263 0.02666 0.04837

0.08 0.07

Power Output (W)

Vp 1.791 3.259 4.762 6.590

Ip 0.95 1.70 2.58 3.47

0.06 0.05 0.04 0.03 0.02

Power output (W)

0.01 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0

150𝜇𝜇𝜇𝜇 𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 Diameter Particles

Power Input (W)

y = 0.0031x - 0.0028

Linear (45𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 dimeter) Linear (150𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 dimeter) Linear (425𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 dimeter)

Power input (W) Figure 7: Efficiency curve for 150𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle powder core where the gradient of the fitted line is the calculated efficiency of the core

58 • Science Extension Journal

Figure 9: Efficiency curve for 45𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle powder core, 150𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle powder core and 425𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃 diameter particle powder core where the gradient of the fitted lines is the calculated efficiency of the cores

Scientific Research in School Volume 3 Issue 1 2021

Efficiency Table 4 and Figure 10 show the final determined efficiencies for the three cores. Table 4: The efficiency of the different iron particle sizes, found using the gradient of the trendlines in Graphs 1,2 and 3.

Iron Particle diameter in core (𝝁𝝁𝝁𝝁𝝁𝝁𝝁𝝁)

Efficiency

425

0.0019

150

0.0031

0.0023

The reliability of the results is strong as the plotted data on graphs 6, 7 and 8 fall closely on a linear projection with a very slight deviation from a y-intercept of zero. Graph 1’s y-intercept was -0.00005W, figure 7’s intercept was -0.0028W and figure 8’s intercept was 0.0001W. This indicates a high reliability of the measurements for the constructed cores. If another power measurement was taken for one of the three constructed cores, one would be quite confident that it would sit on the same line as already plotted meaning that measurement of four power values for each core is sufficient in determining a reliable result and further testing of these same cores is not required.

Efficiency Vs Particle Diameter 0.0035 0.003

Efficiency

0.0025 0.002

However, if the cores were re-made with the same particle diameters, there is uncertainty that the data would display similar results as slight variations in the distributions could alter some measurements. Future research into the distribution of iron powder should be conducted to rectify this uncertainty.

0.0015 0.001 0.0005 0

with size (Xue, 2008). This understanding of powder sizes and core losses resulted in a hypothesis predicting that ideal powder particle size does not have a linear relationship but rather a curve with negative concavity, a hypothesis supported by the results in figure 9.

100

200

300

400

500

Particle diameter (𝜇𝜇𝜇𝜇𝑃𝑃𝑃𝑃) Figure 10: The efficiency of the different iron particle sizes, found using the gradient of the trendlines in Figures 6, 7 and 8.

Discussion Table 4 and Figure 10 depicted results satisfying the hypothesis that while efficiency initially increases as particle size decreases, the efficiency eventually decreases again suggesting an optimal powder particle size. The highest efficiency was resulted from 150μm diameter particles followed by 45μm and 425μm. If a parabolic model could be fitted to the three data points, it appears that there is an ideal size between 200μm and 250μm that yields optimum efficiency, however the model is applied only tentatively. With more certainty it can be said that the optimal powder particle diameter size is between the two extremes of 45μm and 425μm. This is consistent with the literature that depicts that a decreasing particle size causes efficiency increase due to eddy current loss decrease (Shokrollahi & Janghorban, 2007), although nett losses only decrease to a point when particle size decreases because of hysteresis losses. As powder particle size gets smaller, hysteresis loss increases due to iron’s permeability relationship

Further, testing an increased distribution of cores with particle diameter sizes between 30μm and 500μm would result in a more accurate curve to fit the data, as the three plotted data points allow a level of uncertainty in the results. Therefore, an increased number of data points produces a parabolic trendline that represents the results with greater accuracy. By far the most important issue is that the powdered iron cores constructed and tested in the experiment were highly inefficient compared to regular laminated iron cores. Although, the aim of the experiment to compare the effect of a changing powder particle size was successfully achieved, the cores with powder particles only produced an efficiency less than 0.005, whereas the standard laminated iron core that is sold with the student transformer used in this experiment produced an efficiency of 0.0279. The lower efficiency was not surprising due to the substantially lower mass of iron used to construct the powdered iron cores, as the mass of iron in the powder cores was 50.0g and the mass of iron in the laminated core was 300g. An increased mass of iron produces a greater efficiency as more material becomes magnetised by the first coil and therefore transfers more voltage to the secondary coil. Future research should involve greater masses of iron particles, however, this poses a

Science Extension Journal • 59

Scientific Research in School Volume 3 Issue 1 2021

future concern with the use of iron powdered cores, as the insulation required to separate the powder obtains a greater capacity than laminated iron, causing an inefficiency (Trasimond & Lawrence, 1986). Once optimal powder particle size has been determined, further research could investigate the maximum density of powder before the epoxy fails to insulate the particles from each other resulting in greater eddy current losses. The frequency used in this experiment was a result of the standard voltage output of Australian power points, 50Hz. This frequency was appropriate for testing as it is the standard used in the country, however a change in frequency will affect the efficiency of the cores tested. To conclude a result with increased validity, the experiment should be tested with different frequencies, allowing an increased understanding on the effect of particle size on the efficiency of powdered transformer cores (Dixon, 2003). Further research could repeat the test at higher frequencies which is common for various applications (Ueno, et al., 2016) which may result in a smaller particle size being more efficient as higher frequencies reduce hysteresis losses (Trasimond & Lawrence, 1986).

Conclusion In my research I have investigated powdered iron transformer core efficiency. Powdered iron transformer cores produce power loss due to various aspects of its construction, with the major losses a result of eddy currents and hysteresis loss. This article explores the effect of particle size on the nett losses, as increased particle size reduces hysteresis loss but increases loss due to eddy currents. By construction and efficiency testing of three cores with distinct particle sizes, it was determined that a particle diameter size with optimal efficiency between lies between 200μm and 250μm, but with a higher certainty between 45μm and 450μm.

Acknowledgments

I wish to thank Dr Matthew Hill for assisting me with development of my research question, his consistent guidance throughout the research project and for proofreading the report. I wish to thank Mr Cameron Dearn for helping me to select the appropriate procedure for testing of the cores and for supplying reliable technology to record the data.

References Al-Khalili J, 2015, ‘The birth of the electric machines: a commentary on Faraday’ Experimental researches in electricity, http://dx.doi.org/10.1098/rsta.2014.0208

60 • Science Extension Journal

Bernacki K, Rymarski Z and Dyga Ł, 2017, ‘Selecting the coil core powder material for the output filter of a voltage source inverter’, Electronics Letters’, vol. 53, pp.1068–1069, https://ietresearch.onlinelibrary.wiley.com/doi/pdf/10.1049/el.201 7.1534 Dekker M, 2004, ‘Magnetic Materials and their characteristics’ Transformer and Inductor Design Handbook https://coefs.uncc.edu/mnoras/files/2013/03/Transformer-andInductor-Design-Handbook_Chapter_2.pdf Dixon L, 2003, ‘Eddy Current Losses in Transformer Windings and Circuit Wiring’, Texas instruments, http://educypedia.karadimov.info/library/slup197.pdf Hurley G, Wolfle H and Breslin G, 1998, ‘Optimized transformer design: inclusive of high-frequency effects’ IEEE Trans. Power Electron, vol. 13, pp. 651–659. https://doi.org/10.1109/63.704133 Lee1 J, Cha J, Yoon H, Lee J and Kim Y, 2015, ‘Magnetic multigranule nanoclusters: A model system that exhibits universal size effect of magnetic coercivity’, Scientific Reports, vol. 5 Leon F and Semlyen A, 1993, ‘Time Domain Modeling of Eddy Current Effects for Transformer Transients’ Transactions on Power Delivery, vol. 8, https://tspace.library.utoronto.ca/bitstream/1807/9972/1/Semlyen_ 9842_2827.pdf Moses J, 1998, ‘Comparison of transformer loss prediction from computed and measured flux density distribution’, IEEE Transactions on Magnetics, vol. 34, pp. 1186–1188, https://doi.org/10.1109/20.706473 Nyenhuis E, Girgis S, Mechler F, 2001, ‘Other factors contributing to the core loss performance of power and distribution transformers’, IEEE Transactions on Power Delivery, vol. 16, pp. 648–653, https://doi.org/10.1109/61.956752 Shokrollahi H and Janghorban K, 2007, ‘Soft magnetic composite materials (SMCs)’, Journal of Materials Processing Technology, vol. 189, pp. 1–12, https://www.sciencedirect.com/science/article/abs/pii/S09240136 07001756 Trasimond A and Lawrence W, 1986, ‘Powdered iron core magnetic devices’, United States Patent Soileau, https://patentimages.storage.googleapis.com/7f/c1/26/42b9dab236 67a5/US4601765.pdf T.J.E Miller & R. Rabinovici, 1994, ‘Back-emf waveforms and core losses in brushless dc motors’, IEE Proc. B, vol. 141, https://ieeexplore.ieee.org/document/477581 Ueno T, Tsuruta H, Saito T, Watanabe A, Ishimine T and Yamada K, 2016, ‘Practical and Potential Applications of Soft Magnetic Powder Cores with Superior Magnetic Properties’, Sei Technical Review, vol. 7, https://global-sei.com/technology/tr/bn82/pdf/8202.pdf Young H, Freedman R, 2019, ‘Sears and Zemansky’s University Physics with Modern Physics’ Pearson international edition, vol. 12 Xue D, Chai G, Li X and Xiaolong F, 2008 ‘Effects of grain size distribution on coercivity and permeability of ferromagnets’ Journal of Magnetism and Magnetic Materials, vol. 320, pp. 1541– 1543 http://nano.lzu.edu.cn/chai/style/images/papers/Guozhi_2008_JM MM.pdf

Scientific Research in School Volume 3 Issue 1 2021

Bias of a Coin Toss Tomo Bower Barker College We analyse the process of flipping a coin from four different heights (ground height, 10cm, 20cm, and 30cm) to see if there are any variances from the expected 50/50 due to this change. A coin flipping contraption was made to ensure there were as little errors as possible in the flipping of the coins. Through flipping coins 500 times from each of the four different heights using the contraption, we show that at different heights, flipped coins tend to have a difference from the expected 50/50. Though this was the case, the results cannot be used as conclusive evidence for the difference in heights affecting the coin flip as the 𝜒𝜒𝜒𝜒 2 tests that were done came back as statistically insignificant. With a larger sample size, a conclusive result could be found, and the hypothesis could be confirmed. Introduction A coin flip has the same probability of landing on either heads or tails and this has been accepted as a fundamental statement in the classical probability theory (Strzalko et al, 2008). But what if that was not so? What if a coin flip was not 50/50 and could be biased in the favour of a side? This may in fact be the case because it is undeniable that the coin toss follows the laws of mechanics (ignoring air resistance) and with careful adjustment could, at least in theory, be flipped and always land on the same side (Diaconis, Holmes, & Montgomery, 2007). This study aimed to test a different variable to the past experiments done on coin toss bias, being the height above a surface from which the coin is tossed.

Literature Review Coin tosses are widely accepted as a random phenomenon, seen to have an equal probability of landing on either heads or tails. Unpredictable to humans and computers and thus is used in many games, casinos for example realflipacoin.net, and sporting fixtures (Horridge, 2017). Many people over the years have come up with their own theories on how to rig a coin toss with whole blogs and websites dedicated to finding ways to artificially beat the system (Johnson, 2020; Gammon, 2010; Mansur, 2008), however there are little scientific studies providing rigour to these theories.

and the distribution of the initial conditions affect the coin toss including Diaconis, Holmes, and Montgomery’s study (2007) on the dynamical bias in the coin toss, Strzalko and his colleague’s: Understanding coin tossing (2010), and Hou Young and Mahadevan’s study (2011) on the dynamics of a thick coin. These studies have looked in-depth at the physics behind how a coin lands being tossed, focussing especially close to how the initial conditions affect the outcome. These papers looked at how the angle ψ between the angular momentum vector M and the normal at the time t=0 changes the outcome and they concluded that coin tosses were affected by ψ and thus were not random phenomena but followed the laws of physics. This is backed in Diaconis’ study by how when the coin tossing contraption that was used (which can be seen in figure 1), could be carefully adjusted it was possible to have the coin always land on the face of the coin that started up (Diaconis, Holmes, & Montgomery, 2007). Thus, the conclusion can be made that coin tosses obey the laws of mechanics (ignoring air resistance) and that their flight paths are (at least in part) predetermined by the initial conditions of ψ, M, and n. There have also been studies that have investigated the randomization of dice throwing which has also looked at the initial conditions of the dice and how it affects the outcome of the dice throw (Nagler & Richter, 2008; Strzalko et al, 2009; Strzalko et al, 2012).

The study of classical randomization devices has dated back to Poincare’s study on the randomization of roulette (Poincare, 1896). There have been a few previous studies conducting tests in how the mechanics

Science Extension Journal • 61

Scientific Research in School Volume 3 Issue 1 2021

be necessary to avoid the possible human bias when testing a variable such as height about the ground as in this present study. A paper which did use a mechanical contraption found that vigorous flipping could also bias a coin towards the original face-up side (Diaconis, Holmes, & Montgomery, 2007). The researchers flipped a coin 50 times of which 27 were considered and found that there was a slight bias of 51% towards the face-up side when a coin was flipped vigorously. Figure 1: Contraption that was used in Diaconis and his colleague’s study. Table 1: Results from Clark and Westerberg’s study.

The research presented in this present study aims to increase the sample size of Diaconis, Holmes, & Montgomery using a new contraption and flip coins from various heights to quantify the effect of height on coin-flipping bias.

Scientific research question Does the height from which the coin is flipped affect the bias of a coin toss?

Scientific hypothesis The lower the coin is flipped from the more likely it is to be biased towards the face-up side when flipped.

Methodology

Clark and Westerberg’s study on coin toss randomness takes a different approach. They investigated the possibility of humans successfully manipulating the result of a coin toss. Their research involved 13 people toss 300 times and try to have as many heads land from the toss as possible. By introducing an incentive, a $20 and $10 coffee voucher to the first and second most biased respectively, it helped to replicate real life situations for example a sports game or gambling. They reported results of the 13 participants between 51-68% of the 300 coin tosses landing on heads (Clark & Westerberg, 2009) results can be seen in Table 1. They propose that these results humans do have the ability to create a bias on a perceived ‘random phenomenon’. The participants were only given a few minutes to practice the toss before the 300 tosses and were only given simple instructions. Even then, more than half of the participants were able to create a statistically significant bias with p values ranging from 0.77 down to less than 0.001 to the coin toss. This implies that a bias is possible, and that using a non-human contraption would

62 • Science Extension Journal

Design and construction of the coin-flipper In order to flip a coin from various heights, a contraption was constructed that could flip a coin in a similar manner every time. The starting design for the contraption was taken from a previous study which used a metal ruler as a spring to flip the coin (Diaconis, Holmes, & Montgomery, 2007). When their design was replicated it was found that the coin would often land on the ruler after being flipped. Therefore, the design was changed slightly so that the ruler was parallel to the surface, flipping the coin in the opposite direction, in a manner that meant that it missed the ruler and would therefore constitute a fair coin flip. After this modification was made, another problem that was encountered with my contraption during testing was the deformation of the metal ruler. After around only 50 flips using the ruler it deformed and bent almost a centimetre from the original set up.

Scientific Research in School Volume 3 Issue 1 2021

Figure 6: Diagram of contraption at 20cm height.

Figure 2: Prototype contraption

Figure 7: Diagram of contraption at 30cm height

Figure 3: Prototype contraption ruler deformation from original set up.

Figure 4: Diagram of contraption at “ground height”.

Figure 8: Final contraption frontal view.

Figure 5: Diagram of contraption at 10cm height.

Figure 9: Final contraption side view

Method of Experimentation The contraption (Figures 4, 5, & 6) was used to flip an Australian 20 cent coin from four designated heights

Science Extension Journal • 63

Scientific Research in School Volume 3 Issue 1 2021

(Ground height, 10cm, 20cm, and 30cm) 500 times at each. 1. The contraption was set up at “ground height” and the contraption was measured with a spirit bubble ruler to ensure there were no changes in the angle at which the coins were launched. 2. Coins were placed on the contraption with tails up and flipped using the contraption 500 times and each flip was recorded with which side came up after the coin landed. 3. Steps 1-3 were repeated with the contraption set at 4 other heights: 10cm, 20cm, and 30cm. 4. After all results were tallied and measured the difference in results between heights was observed and a𝜒𝜒𝜒𝜒 2 test (chi-squared test) was used to show if the number of tails that came up was significantly different from the expected 50% or not.

Results Table 2 shows the counted heads at each height, along with a chi-squared test result. The alpha value to determine significance was chosen to be 0.0125. As four chi-squared tests were conducted, a Bonferroni correction was applied to the significance threshold reducing it from p=0.05 to p=0.05/4 = 0.0125. Figure 10 shows a visualisation of this data from Table 1. Table 2: Column graph of the difference from 50% for different heights

Release height (cm) 0 10 20 30

# Heads

% Heads

χ2 statistic

p value

261 257 244 242

52.2 51.4 48.8 48.4

0.968 0.392 0.288 0.512.

0.32 0.53 0.59 0.47

Figure 10: Column graph of the difference from 50% for different heights

64 • Science Extension Journal

Discussion Statistical Significance of Results The column graph that is shown in table 2 shows a trend that came about in the tests whereas the heights changed the difference from 50% lowered from 2.2% to -1.6%. Though these values showed a trend the 𝜒𝜒𝜒𝜒 2 tests that were done to see the statistical significance showed that from the different values gathered in the experiment none of the tests were statistically different from the expected 50/50 chance of the coin toss. The 𝜒𝜒𝜒𝜒 2 tests that were done on the results gathered from the experiment gave a null result for the experiment and therefore the hypothesis that the coin toss will be more likely to land on the face-up side when tossed from a lower height could not be confirmed even though the results pointed towards the hypothesis. Comparisons to previous literature Previous literature has stated that the coin toss would be 51% more likely to land on the face-up side when tossed vigorously (Diaconis, Holmes, & Montgomery, 2007). This was confirmed in their study through a series of 50 coin flips of which 27 were considered. My study has come up with some very interesting results as the differences that were found in the coin flips from the different heights came back all 4 times to have a bias of more that 1% from the expected 50/50. From this I have concluded that either there may be slight errors in my study or the Diaconis and his colleague’s study found a wrong value or a value specific to the coin they used which was not a 20 cent piece like what I have used in this study. Improvements to the Study To improve my study, I believe that a larger sample size could be taken to gain a statistical significance for the results. The data that was collected showed a trend but the 500 flips on each side seemed to not be enough to show a statistical significance. With more tests it could be concluded if the results stay the same, become a higher difference from 50/50, or get closer to the expected 50/50. This would give a proper conclusion to the study and gain a proper answer to the question of if height affects the bias of a coin toss. There were also some sources of error in my study in how every ~20 flips done there would be a flip that is wildly different in height and/or number of flips it makes. Though these flips were discounted from the results there is the issue that milder/smaller differences could have been occurring that could not be detected with the naked eye causes slight variances with the results. Another source of error that could be attributed to the tests done were the stability of the contraption when flipping the coins. In order to make sure the flips remained constant and

Scientific Research in School Volume 3 Issue 1 2021

had a vigorous flip the plank of wood was pulled back all the way to the same point each time. When the plank was released, it caused a decent jolt to occur with the contraption. This was decently negated through holding the contraption steady with the other hand while flipping but there may be slight inconsistencies in the data due to the contraptions jolting. To fix this issue that contraption could be held in place by screwing it to the platform that the contraption was placed on or though the use of clamps to keep the contraption as bolted to the platform as possible. Future Areas of Study Some future areas that could be researched in this topic would be to do the same test but with a larger sample size in order to gain statistical significance as stated above, continue with the research but at different heights for example 40cm, 50cm, 60cm, and 70cm, or to research another variable in the flipping of a coin. This could include the rotational velocity, the angle from the normal of the coin, or the angle the coin is flipped from.

Conclusion Despite the non-significant results, the trend that occurred in the tests were fascinating. The nonsignificant results also show just how difficult it can be in studying random phenomena. Though there were differences and trends that were found in the tests that were carried out due to the non-significant results for this study the following conclusion can be made: For coin tosses, the classical assumption that they are a 50/50 chance is decently solid. In the future if more tests are carried out and the sample size is increased further this conclusion may be disproved. In this study I have looked at the bias of coin tosses when they have been flipped from different heights. A coin tossing contraption was used to ensure that flips were similar and ensure no other variables changed. The contraption was set to 4 different heights and 500 flips were done from each. The results were collected, and the percentages found had a difference from the expected 50/50 of 1.2 to 2.2%, though using a 𝜒𝜒𝜒𝜒 2 test these results were found to have non-significant results and thus could not be used as conclusive evidence to say that the height at which the coin is tossed from affects the probability of the flip.

thank Michael Bower for his help in building and planning the contraption used in the experiment.

References Clark, M.P.A. and Westerberg, B.D. (2009). How random is the toss of a coin?: Holiday Review, vol. 181, no. 12. Diaconis, P., Holmes, S. and Montgomery, R. (2007). Dynamical Bias in the Coin Toss. SIAM Review, [online] vol. 49, no. 2. Available at: http://statweb.stanford.edu/~susan/papers/headswithJ.pdf Gammon, K. (2010). Cheat With Science: Win a Coin Toss. Available at: [online] Wired. https://www.wired.com/2010/11/st-cheatscience-cointoss/ [Accessed 14 Jun. 2021]. Horridge, K. (2017). 2 Up Coin Toss: Australia’s Favorite Illegal Game. [online] Casino.org Blog. Available at: https://www.casino.org/blog/2-coin-toss-game/ [Accessed 14 Jun. 2021]. Hou Young, E. and Mahadevan, L. (2011). Probability, geometry, and dynamics in the toss of a thick coin: American Journal of Physics, vol. 79, no. 12. Johnson, N.J. (2020). How To Rig A Coin Toss. [online] at: Nicholas J Johnson. Available https://www.conman.com.au/post/7-ways-to-rig-a-cointoss#:~:text=Rest%20the%20coin%20on%20the [Accessed 14 Jun. 2021]. Mansur, R. (2008). How to Force the outcome of a coin toss with a simple trick. [online] WonderHowTo. Available at: https://cons.wonderhowto.com/how-to/force-outcome-cointoss-with-simple-trick-263840/ [Accessed 14 Jun. 2021]. Nagler, J. and Richter, P. (2008). How random is dice tossing?: APS physics, vol. 78, no. 3. Strzalko, J., Grabski, J., Stefanski, A., Perlikowski, P. and Kapitaniak, T. (2009). Can the dice be fair by dynamics?: World scientific, vol. 20, no. 4. Strzalko, J., Grabski, J., Stefanski, A., Perlikowski, P. and Kapitaniak, T. (2008). Dynamics of coin tossing is predictable: Elsevier B.V, vol. 469, no.2. Strzalko, J., Grabski, J., Stefanski, A., Perlikowski, P. and Kapitaniak, T. (2012). The three-dimensional dynamics of the die throw: American institute of physics, vol. 22, no. 4. Strzalko, J., Grabski, J., Stefanski, A., Perlikowski, P. and Kapitaniak, T. (2010). Understanding Coin-Tossing: Springer Science+Business Media, vol. 32, no. 4.

Acknowledgements I would like to thank Dr Matthew Hill for proofreading the report and for help in the statistical analysis of the data results from the experiment. I would also like to

Science Extension Journal • 65

Scientific Research in School Volume 3 Issue 1 2021

66 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

The influence of text structure on learning Physics Caleb Swanson Barker College Education relies on high quality resources that are easily distributable, such as textbooks and videos to provide the backbone of content delivery. In Physics, these must be optimised to address misconceptions held by students. Previous research has demonstrated that for physics video instruction, a dialogical teacher-student based conversation was far more effective in changing physics-related misconceptions than a traditional exposition. This research seeks to explore whether this trend is also evident for printed, written physics material such as that found in textbooks. 19 students from a beginner and 18 students from an intermediate level of Physics learning experience at a co-educational high school in Sydney, Australia were given one of two sets of sheets containing instruction on basic Newtonian Mechanics; a scripted dialogue between a teacher and student, or a traditional textbook-like exposition, and their gain in understanding was measured with pre and posttests. The exploratory study suggested that the beginner level student found the exposition more beneficial, while the intermediate student found the dialogue more beneficial. Literature Review Introduction to Physics Education Nations around the world are seeking to increase the number of students and quality of teaching in the sciences, especially Physics (ACOLA, 2013; The Royal Society, 2014). The student who studies Physics should leave the course with a greater knowledge on scientific problem solving, inquiry, mathematical skills and information synthesis, skills integral to careers in science, engineering and other areas (Landau, 2006). Physics also has a reputation of being an interesting, yet difficult discipline for students in both high schools and university (Ornek, 2008; Angell, 2002), often resulting in students avoiding taking the subject entirely (DeWitt, 2018). Perhaps the clearest result from Physics Education research is on the efficacy of teaching that has moved from traditional instructional methods to interactive methods (Redish, 2004), as these interactive lectures yield significantly greater learning gains (Sharma, et al., 2010). One aspect of said traditional instructional methods is a dependence on expository information dumps, to try and incite a correct understanding in the mind of the student (whereby the information is presented in a spoken or written format). Research shows these traditional methods are not as effective in inducing Physics-related conceptual change unless interaction is introduced such that a student observes refutational, misconception-based exchangesS (Tippett, 2010; Muller, 2007).

Misconceptions and Physics Education As students attempt to make sense of challenging concepts, they develop personal knowledge structures which often contain misconceptions (Tippett, 2010). Due to their influence, these misconceptions interfere with learning expert concepts (Smith, diSessa & Roschelle, 2009). Troublingly, students can doggedly hold onto mistaken ideas even after receiving instruction designed to dislodge them (Smith, 2009). Thus, Physics educators must either address and disprove these misconceptions before they come into fruition and become a hinderance to further learning, or must deconstruct already existing misconceptions. While it may be possible to give complete information as students are learning a concept for the first time, minimising the chances of students developing misconceptions (Kuczmann, 2017), Redish (2004) argues that misconceptions will arise through casual observation of the world and will then be brought into the science classroom and mis-applied to the difficult concepts of Physics. Therefore the deconstruction and replacement of existing misconceptions needs to be a significant tenet of Physics Education, and should not be undervalued. Since the 1980’s, science researchers have investigated the conceptual change potential of refutational text, a text structure that refutes common misconceptions and that has been shown to be one of the most effective textbased means of correcting student misunderstandings (Hynd & Alvermann, 1986; Guzzetti, 2000). The key difference however, between this heavily researched refutational text and the dialogical text employed in this study, is that in dialogical text, misconceptions are Science Extension Journal • 67

Scientific Research in School Volume 3 Issue 1 2021

stated without warning that they are incorrect, while in refutational text they are introduced as incorrect (Tippet, 2010). The key components of refutational text are shown in figure 1.

Figure 1: The key components of refutational text (Source: Tippet, 2010).

Using traditional media to deliver non-traditional instruction There is a utility for what might be considered traditional modes of media (lectures, online videos, worksheets, textbooks etc.) in education. They are accessible and scalable in a way that experimentation, groupwork and problem-based-learning are not. Research from The University of Sydney has advocated for nontraditional Physics instruction through the traditional medium of lectures (Sharma, 2010; Georgiou, 2014) along with using cognitive principles to improve the efficacy of linear (YouTube) educational videos (Muller, 2008). Each of these papers articulates the importance of non-traditional methods such as interactive, engaging lectures to help address and challenge student misconceptions. In particular, Dr Derek Muller (of YouTube’s ‘Veritasium’) is a recognised expert in designing effective multimedia for Physics Education, whereby he specifically focused on improving Physics Education delivered through the video format. One area of Muller’s research that targets misconceptions in Physics involved replacing a lecturer speaking directly to cameraproviding an expository collection of information (very much like a traditional lecture)- with a filmed discussion between a teacher and student in a dialogue based discussion on misunderstandings the student possessed (Muller, 2006 & 2007). Muller’s study involved 678 first year Physics students at The University of Sydney including students studying the Fundamental course (little prior physics instruction), the Regular course (senior high school Physics experience), and the Advanced course (excelled at high

68 • Science Extension Journal

school Physics). They watched one of four videos on Newtonian mechanics which included either: a. Exposition of subject matter (modelling a lecture) b. Refutation (an exposition in which alternative conceptions were stated and then refuted by the speaker) c. Dialogue (similar to the refutation, but the misconceptions were raised by an actor playing a student, such that the video took on a format similar to a Socratic dialogue) d. Extended exposition (exposition with added interesting material not relevant to the course, such that the instruction would take the same length of time as the refutation and dialogue treatments) As mentioned before, key to the unique dialogue video was that the students’ misunderstandings were stated without warning that they were false, whereby only through refutation by an expert did the truth emerge. Only in the dialogue and the refutation treatment were misconceptions addressed. Pre-post testing using a 26 question, validated multiple choice test, the Force and Motion Concept Evaluation (FMCE) (Thornton & Sokoloff, 1998), revealed Students who watched the misconception containing dialogue and refutation treatments had a greater gain in understanding than other treatments. Muller found the gain was highest for the dialogue treatment. Why was dialogue most effective in improving student conceptual understanding? By asking students to report on their mental effort while watching the video, Muller found that the vast majority of students who watched the student-teacher dialogue found the process ‘confusing’ and ‘hard to understand,’ reporting a high level of mental effort during instruction (6.0/9). However, exposition video viewers mostly found their video ‘simple’ and ‘clear,’ rating a lower mental effort necessitated (5.4/9) (Muller, 2007). Muller concluded that ‘the increased cognitive load incurred with misconception-based treatments was germane rather than extraneous on the average for (university physics) students with all levels of prior knowledge’ (Muller, 2007, pp.150), as this increased cognitive load resulted in more mental effort being invested into the video, leading to greater gains. These findings are supported by previous misconception research which suggests that cognitive conflict (i.e. mental effort) is essential for conceptual change (Guzzetti, et al., 1993; diSessa, 2014). Muller (2007) suggested that since students who watched the misconceptions treatments identified with the student so

Scientific Research in School Volume 3 Issue 1 2021

were able to recognise their knowledge being challenged, such that they were forced to invest more mental effort. Do the findings translate to a written format? As identified earlier, traditional methods of teaching such as lectures, videos and text/textbooks are not in themselves outdated and are likely to still be widely used for decades to come. Consequently, the question is raised: Is the inclusion of dialogue-based material only effective in a video format, or does it too translate to a written/reading format?

Scientific research question Do students reading a scripted dialogue learn more effectively than students reading an expository text of the same content?

Scientific hypothesis That students who read a refutational dialogue between a student and teacher on basic Newtonian mechanics will learn more effectively than students who read an exposition from a teacher.

Methodology To parallel Muller’s study on instructional video design (2007), this present research tested the same independent variable (the structure of the instructional material) choosing to focus on the two most salient options: dialogue vs exposition. Instead of using online videos, this research changed the medium to a written, hard-copy, paper format to see if a similar benefit of dialogue over exposition could be observed. Both treatments consisted of lesson content identical to a snippet of the video scripts Muller had written and used for his experiments, and also used elements of the measurement tool (FMCE) which he used. There were other differences due to the logistical limitations of a study occurring within the constraints of a high school academic course rather than a PhD research program including: The number of participants The age and academic level of participants The length of the instructional material, as approximately a third of the video transcripts were used in this study, and The length of the measurement tool, as only one set of question from the FMCE could be used in this study rather than five sets of questions.

These were necessary not only due to having limited access to participants, but also due to limitations in accessing the participants’ time. Despite these differences, the similarities allow for this research to be a helpful exploration into the corresponding aspects of textual instructional design. Participants Two classes of high school students were selected from a co-educational independent school in Sydney, Australia. These included a class of 19 Year 10 Science students (approximately 16 years old) and a class of 18 Year 11 Physics students (approximately 17 years old). Both of these classes had differing levels of exposure to physics (specifically Newtonian Mechanics). While difficult to compare, it is proposed that the Year 10 students were classified at an equal or lower level of understanding than Muller’s fundamental students, and the Year 11 students could be classified as equal to, or maybe slightly higher than, the same fundamental students. The two levels were chosen because of their differing levels of Physics Education experience such that interaction with the two treatments could be observed across a diverse population of students, to inform choices of future researchers of less exploratory and more in-depth studies. All students were informed that their participation in the study was voluntary and that their responses would be de-identified once the pre-test, treatment group, and post-test were matched and collated. All students from both classes provided consent to be a part of the research. The methodology was submitted to and received approval from The Barker Institute before recruitment of participants commenced. Design of instructional material The first third of the content scripts of both the dialogue and exposition treatments used by Muller were included in this study. This included; an overview of Newton’s laws, definitions of speed, acceleration and velocity in the context of graphs, and an analysis of forces acting on a book being pushed across a table. This was selected over the remaining content as the remaining examples used did not specifically apply to the FMCE questions selected, and also were similar to the earlier content in terms of containing or not containing refutational text so the impact of removing them was minimal. In Figures 2 and 3, the differences between the dialogue and exposition text are clear. For example, “Ah, yeah. I just thought things always tend to lose energy, slow down and go to rest” compared to “all objects like to keep their motion unchanged, going the same direction with the same speed”.

Science Extension Journal • 69

Scientific Research in School Volume 3 Issue 1 2021

Tutor: Well zero is a constant velocity isn’t it? Student: Yeah I guess so. If you gave it a push and let go, it would slow down and stop though. Tutor: That’s because of friction. It’s easier to see if we take friction out of the equation. If I put this slider on an air track, you can see that whatever its motion, without any unbalanced forces it continues at constant speed. Student: Ah, yeah. I just thought things always tend to lose energy, slow down and go to rest.

Figure 3: An excerpt from the dialogue paper demonstrating the nature of the dialogue textual structure. This dialogue can be easily distinguished from both exposition (Figure 3) and refutational text (Figure 1). (After: Muller, 2008)

One of the interesting insights Newton had was that all objects like to keep their motion unchanged, going in the same direction with the same speed. This is Newton’s first law of motion. Stated more formally, it goes; an object will continue with uniform velocity unless acted on by an unbalanced force. Of course, a special, pretty important case of this is that if an object is not Figure 2: An excerpt from the exposition paper that encapsulates the nature of the expository text (After: Muller, 2008)

Design of data collection instruments Like the instructional material, the five FMCE (Thornton & Sokoloff, 1998) questions used for the pre and post-test were a subset of the test used in Muller’s research (2007). The test asked questions about what kind of force (left, right, increasing in magnitude, constant in magnitude, etc.) would have to be applied to a sled on ice, to achieve a certain velocity or acceleration (constant, decreasing, etc.). The questions were multiple choice and addressed the general basics of Newton’s first two laws, as did the instruction, meaning that the analysis would allow for generalised conclusions about the efficacy of the treatments on teaching the concepts associated with Newton’s first two laws.

70 • Science Extension Journal

On the reverse of the post-test sheet, students were asked to rate their mental effort expenditure on a 9 point scale and were asked to comment on the process. Administering the intervention and data collection Students from each class were randomly assigned to one of two groups which would read either the exposition or discourse instructional material. Table 1 shows the sample size relating to each year and treatment group. Table 1: The distribution of participants across year and treatment groups

School Year Year 10 Year 11

Exposition 10 10

Discourse 9 8

Total 19 18

The classroom teacher introduced the author of the paper, and the author briefed the class on the experiment, drawing attention to the fact that those surrounding the student would have a different reading sheet than themselves. They were told that the research was testing the efficacy of different written forms, and so to not look onto others’ sheets, or other student answers to the pre or post-tests. Other than information relating to informed consent, no other details of the research purpose were given to the students before the commencement of the task, preventing subconscious bias. Both classes sat the pre-test for five minutes. After the pre-test, the random allocation occurred where alternate students would read the different treatment material. They were given six minutes to read the pages. These were then collected before the post-test (to parallel the participants in Muller’s study not having access to the video when completing the post-test) and each student was given the post-test and reflection questions, having 7 minutes to complete before collection. Analysis Methodology The treatment group, year level, mental effort and answers to the pre-test and post-test questions were entered for each student into Microsoft Excel, and then the students results were tabulated based on their given treatment. Interesting comments were noted, and averages and normalised gain from pre-test to post-test (Gain =

Post%−Pre% 100%−Pre%

(Hake 1998)), were compared. As

this was purely an exploratory study based on the observed effectiveness of a dialogical compared to an expository text structure in educational Physics multimedia, the decision was made to test two different levels of Physics learning experience (obtaining a diverse sample with potentially different responses) rather than trying to get a large sample size at one level of Physics learning experience. Unlike Muller’s study

Scientific Research in School Volume 3 Issue 1 2021

seeking confirmatory evidence for the theory involving approximately 90 students per treatment group, this study included approximately 10 student per treatment group, per level of Physics learning experience. While technically inferential statistical tests could be administered, it would be inappropriate to consider them as strong evidence due to the small sample size and short pre-/post-test. Future research can be based on this exploratory study which would involve focusing on one level of Physics learning experience and a larger sample size.

Results Table 2 summarises all quantitative results from both classes across both treatment groups.

Discussion Observation 1: Year 11 students had a better initial understanding of Newton’s laws than Year 10 students. From comparing pre-test scores between Year 10 (mean score 0.47/5) and Year 11 (mean score 2.2/5) (See Table 2), it can be seen that Year 11 had a greater initial understanding. This can be explained as the Year 11 students had completed an additional mechanics unit, while the Year 10 students had less than a term of exposure to senior Physics and Newton’s laws. Observation 2: Exposition was more beneficial for Year 10 students, Dialogue was more beneficial for Year 11 students. The second-last column of Table 2 reveals the Year 10 students who read the exposition had a larger normalised gain (+0.27) than those who read the dialogue (+0.07). This was inconsistent with the hypothesised result that dialogue would result in greater gains, and so is inconsistent with Muller’s research on video instruction. However, this was the opposite for the Year 11 class. The Year 11 students who read the dialogue had a larger normalised gain (+0.29) compared to the exposition (-

0.03). This result is consistent with Muller’s research on video instruction and the hypothesis of this research. Supposition 1: Exposition may be better for students with limited prior understanding, but dialogue may be better for students with some level of understanding (and misconceptions to address). From observations 1 and 2, we can suppose a unifying principle for the seemingly contradictory results between Year 10 and Year 11 students. With regard to the Year 10 group, their lower prior understanding of basic mechanics may have resulted in the exposition treatment having a greater influence on their post-test scores because new information was presented simply and clearly. Meanwhile, the Year 11 group had a greater initial understanding, and found the dialogue more influential to changing their pre-existing misconceptions. This supposition has potentially major implications for Science Education, suggesting that in order to best fit the needs of students with differing prior levels of understanding, a teacher perhaps should give seemingly different and contrarily structured instruction to their students. In relation to Muller’s results, this could affirm that the Year 11 students were approximately equivalent to his fundamental university student group, and such reaped similar benefits to them. This supposition also suggests that the Year 10 group possessed less knowledge than the fundamental university group. Observation 3: Mental effort did not positively correlate with normalised gain. Regarding the Year 10 class, the last column shows mental effort was the same (5.1/9) for both treatments, while their average normalised gain was greater for exposition (+0.27) than dialogue (+0.07). For Year 11 there is a similar lack of positive correlation, despite the higher gains for the dialogue treatment, those students reported a lower mental effort than the exposition treatment.

Table 2:Quantitative data from the experiment including sample size, average pre- and post-test scores, average mental effort and normalised gain from pre- to post-test.

School Year

Treatment Group

Sample size

Year 10 Year 10 Year 11 Year 11

Exposition Dialogue Exposition Dialogue

10 9 10 8

Average Pre-test Score (/5) 0.60 0.33 2.0 2.4

Average Post-test (/5) 1.8 0.67 1.9 3.1

Average Normalised Gain1 +0.27 +0.07 -0.03 +0.29

Average mental effort (out of 9) 5.1 5.1 5.6 4

Science Extension Journal • 71

Scientific Research in School Volume 3 Issue 1 2021

Observation 4: In Year 11, the exposition treatment required more mental effort than the dialogue treatment. Comparing mental effort in Year 11, the expositional treatment had an average mental effort score (+5.6) significantly greater than dialogue (+4.0). This is the opposite to what occurred amongst the students in Muller’s university study where dialogue students reported a higher level of mental effort. Supposition 2: Written format may require a different application of mental effort than video format. Muller’s previous research reported a clear result that students watching a dialogue video used more mental effort (and this was used as an explanation of why they had greater learning gains). Not only did mental effort not correlate with learning gains (Observation 3), Year 11 students used more effort reading the exposition than the dialogue (Observation 4). Both results were surprising. It is important to consider that a written format is vastly different to a video format. Perhaps the passive nature of a video with exposition (which necessitated low mental effort in Muller’s study) is not the same for written exposition. As a student needs to read the text and construct meaning it may be that the written exposition takes more mental effort to follow the flow of the argument, whereas the dialogue is easier to read as it replicates a real conversation requiring less mental effort. More research including interview questions would be needed to verify this. Further research There are a number of areas for further research to verify the supposition that novice students may benefit from exposition whereas comparatively experienced students benefit more from dialogue. First, this exploratory study could be extended to more classes of students in Year 10 and Year 11 to see if the results with a larger sample are consistent. It should also be considered if longer treatments and pre-post tests can be used to more closely replicate elements of Muller’s original study. If the supposition that in traditional instructional media, novice students benefit from exposition while more advanced students benefit from dialogue is true, a similar relationship could exist in the video format. Using Muller’s dialogical and expository videos as treatments for Year 10 and Year 11 students, research could be done to see if such a relationship exists. If these steps are verified the next logical step would be to investigate a clearer distinction between what makes someone a novice or expert with regards to being better suited to dialogue than exposition. Interviews of

72 • Science Extension Journal

students and analysis of other work samples may offer insights into this question.

Conclusion My research is in an important area of Physics (Newtonian mechanics) and the misconceptions that Physics students hold on this topic. The research builds upon the ideas and findings of a PhD thesis completed at The University of Sydney (Muller, 2008) which would become the basis of the popular YouTube channel “Veritasium”. Muller’s thesis included a number of published papers focusing on how YouTube videos can teach Physics, specifically how a teacherstudent dialogue was far more effective in conveying challenging Physics concepts to students than a more traditional expository information lecture. My work translates this to a different paradigm, exploring what is the best way to teach challenging Physics concepts in a printed, hard-copy, format. In a controlled study, I used five questions from a validated Physics diagnostic instrument, the FMCE (Thornton & Sokoloff, 1998), as a pre- and post-test to measure students increase in understanding after reading either a misconceptions containing dialogue or a non-misconception exposition. To analyse this data, I compared the mean net increase in marks for each set, finding that the more experienced group found the dialogue treatment more beneficial, while the less experienced group found the exposition more helpful. Ultimately, through considering the key tenets of what makes Physics-based misconceptions so proliferate amongst the student body, and the previous research into the benefit of dialogue when presenting students with videos, my final supposition was that the dialogue treatment was more useful for those students with greater experience with Physics and therefore a greater number of misconceptions to be uprooted, while the exposition treatment was more useful for students still in the developmental stages of understanding Physics.

Acknowledgements I would like to thank Dr Matthew Hill for offering invaluable supervisory support throughout the developmental and writing stages of this report. Thank you to Mr Matthew Arnot for allowing me access to his Year 10 and 11 classes to gather data, and thank you to his students for giving me their time.

References Angell, C., Guttersrud, Ø., Henriksen, E.K. and Isnes, A. (2004). Physics: Frightful, but fun. Pupils’ and teachers’ views

Scientific Research in School Volume 3 Issue 1 2021

of physics and physics teaching. Science Education, 88(5), pp.683–706.

Knowledge in Transition. Journal of the Learning Sciences, 3(2), pp.115–163.

Australian Council of Learned Academies (2013). STEM: Country Comparisons. Melbourne: Australian Council of Learned Academies.

The Royal Society (2014). Vision for Science and Mathematics Education Summary. In: Vision for Science and Mathematics Education. [online] Available at: https://www.voced.edu.au/content/ngv:63943.

DeWitt, J., Archer, L. and Moote, J. (2018). 15/16-Year-Old Students’ Reasons for Choosing and Not Choosing Physics at a Level. International Journal of Science and Mathematics Education, 17(6), pp.1071–1087. DiSessa, A.A. (2014). A History of Conceptual Change Research. The Cambridge Handbook of the Learning Sciences, pp.88–108. Guzzetti, B.J. (2000). Learning Counter-Intuitive Science Concepts: What have we Learned From over a Decade of Research? Reading & Writing Quarterly, 16(2), pp.89–98.

Thornton, R.K. and Sokoloff, D.R. (1998). Assessing student learning of Newton’s laws: The Force and Motion Conceptual Evaluation and the Evaluation of Active Learning Laboratory and Lecture Curricula. American Journal of Physics, 66(4), pp.338–352. Tippett, C.D. (2010). Refutation text in Science Education: A Review of two Decades of Research. International Journal of Science and Mathematics Education, [online] 8(6), pp.951– 970.

Guzzetti, B.J., Snyder, T.E., Glass, G.V. and Gamas, W.S. (1993). Promoting Conceptual Change in Science: A Comparative Meta-Analysis of Instructional Interventions from Reading Education and Science Education. Reading Research Quarterly, 28(2), p.116. Hake, R.R. (1998). Interactive-engagement versus traditional methods: A six-thousand-student survey of mechanics test data for introductory physics courses. American Journal of Physics, [online] 66(1), pp.64–74. Hynd, C. and Alvermann, D.E. (1986). The Role of Refutation Text in Overcoming Difficulty with Science Concepts. Journal of Reading, 29(5), pp.440–446. Kuczmann, I. (2017). The Structure of Knowledge and Students’ Misconceptions in Physics. AIP Conference Proceedings, 1916(1). Landau, R. (2006). Computational Physics: A Better Model for Physics Education? Computing in Science & Engineering, 8(5), pp.22–30. Muller, D. (2008). Designing Effective Multimedia for Physics Education. PhD Thesis. Muller, D.A., Bewes, J., Sharma, M.D. and Reimann, P. (2007). Saying the wrong thing: improving learning with multimedia by including misconceptions. Journal of Computer Assisted Learning, 24(2), pp.144–155. Ornek, F., Robinson, W.R. and Haugan, M.P. (2008). What Makes Physics Difficult? International Journal of Environmental & Science Education, 3(1), pp.30–34. Redish, E.F. (2004). A Theoretical Framework for Physics Education Research: Modeling Student Thinking. Physics Education. Sharma, M.D., Johnston, I.D., Johnston, H., Varvell, K., Robertson, G., Hopkins, A., Stewart, C., Cooper, I. and Thornton, R. (2010). Use of interactive lecture demonstrations: A ten year study. Physical Review Special Topics - Physics Education Research, 6(2). Smith III, J.P., diSessa, A.A. and Roschelle, J. (1994). Misconceptions Reconceived: A Constructivist Analysis of

Science Extension Journal • 73

Scientific Research in School Volume 3 Issue 1 2021

74 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Could you be any more random? A study of written and spoken Random Sequence Generation Manxi Zhang Barker College Randomness is a state which lacks orderliness and predictability. Its abstract nature has remained elusive to philosophers and mathematicians for centuries, leading to research into the human ability for generating randomness. Previous studies differ on whether humans can consciously generate randomness. This study further investigates the variables which affect the human ability for randomness through comparing speaking and writing as methods for Random Sequence Generation (RSG). 21 high school students aged 16-18 wrote 200 digits and spoke 200 digits attempting to achieve a randomised sequence. The data was analysed through the methods proposed by Barbasz et al., (2008) – Shannon Index of Entropy and Correlation Function – against pre-existing random sequences such as 𝜋𝜋𝜋𝜋 and computer-generated sequences. Unexpectedly, two two-tailed t-tests indicated that writing produced more random results than speaking for both of metrics, Entropy (t=-2.253, p=0.0356) and Correlation Function (t=4.68, p=0.00014). Further analysis revealed simple training that could help humans achieve greater randomness allowing for further investigation questioning the impact of speaking or writing.

This research investigates human ability to create random strings of numbers, comparing randomness in written and spoken forms. However it is essential to first understand what is meant by randomness, which as a concept lacks representation and exploration in literature (Eagle, 2005); specifically in noting that having the appearance of randomness does not necessarily mean that a sequence of numbers is genuinely random.

next number from the previous numbers (e.g. with the roll of a dice, the previous number/pattern has no influence on the next number). ii. Kolmogorov Complexity (Kolmogorov, 1933) defines finite sequences with descriptive complexity; complexity is achieved when the sequence cannot be expressed in a simpler way than repeating the whole sequence. iii. Martin-Löf randomness (Martin-Löf, 1966) defines infinite sequences with measure theory. 2

Defining Randomness Mathematically, the theory of randomness, also known as algorithmic randomness, is underlaid by three different intuitions (or principles) (Twerijn, 2016): i. Mises–Wald–Church randomness (von Mises, 1919; Wald, 1936, 1937; Church, 1940) illustrates randomness as a relative notion, therefore formalising random sequences as unpredictable 1; unpredictability is achieved if a human or computer is unable to guess the

The key to defining the randomness of a finite sequence such as the ones conducted in this experiment is unpredictability, however this is difficult to quantifiably measure. We might ask a person or computer to predict the next number after seeing the first 10 and see how accurate they are, but this is just as much a measure of human/computer pattern recognition as the randomness of the sequence. Therefore, the key to quantifiably measuring randomness lies in descriptive complexity.

1 Mises-Wald-Church randomness under the modern theory of randomness is flawed, thus sequences defined by it are noted as stochastic rather than random.

Literature Review

As Martin-Löf randomness is only applicable to infinite series, and thus does not have a direct link to determining randomness in pre-generated sequences, refer to Twerjin (2016) for further explanation on randomness in infinite sequences. Science Extension Journal • 75

Scientific Research in School Volume 3 Issue 1 2021

Maximum descriptive complexity occurs when for a string of digits the most simplistic way to retell or describe the sequence would be to list the string of digits in its entirety. In the example provided by Twerijn (2016) 1000 zeros cannot be random as its description is far shorter than the sequence (i.e. “1000 zeros”). Thus, Kolmogorov Complexity, formalizes the intuition that for a sequence to be considered Kolmogorov random it cannot be compressed into a regular pattern. However, these definitions are rooted in theoretical scenarios and as such specific scientific tests for randomness that can be applied to strings of numbers need to be established. Some are evident in research into Random Sequence Generation. Significance and use of Random Sequence Generation Random Sequence Generation (RSG) is the process in which a series of digits are generated with the intention for randomness. The significance of reliable RSG lies in its implications in computational sciences (Figurska et al., 2008). Similarly, research in areas of human generated randomness may contribute to furthering understanding of the "cognitive science of decisionmaking" (Persaud, 2005, p.211) as well as improving computer generated codes or cryptography in the simulation of artificial intelligence, a mimic of the human brain. Human ability for RSG As the geneticist J. B. S. Haldane put it, ‘man is an orderly animal’ who ‘finds it very hard to imitate the disorder of nature’ (Haldane, 1941. In: Pandit, 2012); in analysing the human ability for RSG, there is the need for an understanding of why we deviate from randomness in those ways. Barbasz et al. (2008) propose two reasons for this; the structural limitations in short-term memory, i.e. working memory capacity (Baddeley, 2001), and the control processes of ‘strategies’ used in human RSG, that is the limitation that stems from the need for humans to have a RSG strategy or plan (which is essentially antithetical to randomness). Lack of knowledge or understanding of randomness as a deficiency was investigated in studies conducted by Lopes & Oden (1987). Their results showed an improvement to the degree of randomness following the provision of feedback. 3Barbasz

et al. (2008) give the example of “1, 3, 7, 5, 2, 4” as having a turning point at 7 and 2.

76 • Science Extension Journal

However, in another study they were unable to prove professional statisticians as ‘better’ generators of randomness, therefore concluding that there was a limit to which human RSG could be improved. Persaud (2005) conducted research into conscious RSG of the digits (0-9) and concluded that humans can consciously generate a random sequence of numbers. In direct contradiction, Figurska et al, (2008) published a polemic study which reconducted Persaud’s research with a larger sample group, concluding that humans could not generate a random sequence. While there is some research that supports this claim that humans cannot generate a random sequence (Kareev, 1992; Brugger et al., 1997) the matter is by no means settled. In particular, the literature suggests a lack of study in the variables which may improve or hinder RSG. Hence this study attempts to build upon this gap by comparing a new variable (speaking the digits vs writing them down), and slightly adjusting the methods of quantifying randomness. Analysis of RSG Across the literature of previously conducted experiments, various indexes are used in the analysis of random sequence generation. Yet there is little coherence between papers, thus presenting a difficulty in collating information and building upon previous research. There is a similar problem within the analysis of data where determining randomness relies on multiple factors of randomness deviations. Barbasz et al. (2008) categorises these indexes of analysis into one of three groups. (i) Equality of distribution or frequency Measured by an index of R (redundancy) (Towse & Neil, 1998; Barbasz et al., 2008) based on the Mathematical Theory of Communication (Shannon, 1948), it takes the assumption that a maximum amount of information is received with equal distribution; an unequal distribution, results in increased predictions of the likelihood of events. (ii) Consecutive responses Measured by analysis of paired elements, Random Number Generation (Evans, 1978; Barbasz et al., 2008) and RNG2 include indexes such as ASC, DESC (ascending/descending pairs), TPI (Turning Point Index, Towse & Neil, 1998) 3 and RUNS which look for tendencies towards forward or backward counting.

Scientific Research in School Volume 3 Issue 1 2021

(iii) Repetitions over various distances Measured by the Phi index (Towse & Neil, 1998) it analyses bias in repetition of the same phrase over different distances. This is further explained in the proposed Correlation Function.

another numerical tool can be utilized that looks for repeated digits which therefore exposes patterns. Table 1: Values of Entropy for different levels of randomness

Hence, due to the difficulty of placing the importance of any one variable over another, Barbasz et al. (2008) propose two methods of analysis – a measure of entropy H(x) (Shannon, 1948) and a Correlation Function Cf to be used in conjunction with one another in lieu of the various uncorrelated indexes illustrated above.

Description

Shannon Index of Entropy A random sequence of numbers will most likely have an even number of each digits. E.g. a random string of 200 numbers has the highest probability of including 20 of each digit (0, 1, 2, 3, 4, 5, 6, 7, 8, 9) and so if the distribution of digits substantially varies from approximately 20 of each (e.g. 60+ “ones” rather than approximately 20), it is likely that the sequence is non-random.

Nonrandom (ordered)

This condition for randomness can be quantified by calculating the Entropy of the sequence. Entropy decreases if any digit (or multiple digits) appears too often in the string of numbers. Entropy increases to a maximum if there are an even number of each digit. Shannon’s Index of Entropy is one method of calculating the entropy (Equation 1) though an alternative calculation of entropy can be found in Persaud (2005). 𝑛𝑛𝑛𝑛

𝐻𝐻𝐻𝐻(𝑥𝑥𝑥𝑥) = − � 𝑝𝑝𝑝𝑝(𝑖𝑖𝑖𝑖) log2 𝑝𝑝𝑝𝑝(𝑖𝑖𝑖𝑖)

(1)

𝑖𝑖𝑖𝑖=1

Equation 1: Shannon’s Index of Entropy (random sequences are likely to have a higher entropy) Where H(x) is entropy, p(i) is probability of ith events, n is number of possible events.

Maximum entropy for a sequence of 10 different events is 3.32193. Thus, a higher value for entropy suggests greater randomness. The entropy tool becomes more reliable the longer the string of numbers is and so this is an appropriate tool for one measure of randomness for the 200-digit strings constructed by participants in this experiment. It is clear however that a non-random but ordered sequence can give a high entropy score. While a brief visual inspection would pick up the particular string in the table above as non-random,

Random**

Semirandom

Sequence 86117 38193 26117 93105 37433 78247 51359 62377 01234 56789 01234 56789

Shannon’s Entropy (max = 3.32193)*

Comments

2.9086950

A more even spread of digit frequencies.

2.8609640

A less even spread of digit frequencies.

3.3219281

This matches the max entropy as it has an even spread of the digits

11111 11111 0 11111 11111 *For ten different ‘choices’ of digits **20 consecutive digits of 𝜋𝜋𝜋𝜋 Nonrandom (repeated)

Each digit in the sequence is the same.

Correlation Function Correlation considers the repetition of digits. The correlation function is the sum of how often digits get repeated immediately after itself, or two digits later, three, four all the way up to 9 or 10 digits. For a truly random sequence each individual probability of repetition should be 0.1. While previous use of the Correlation Function was limited to 9 stages (Twerjin, 2016) the method employed in this experiment was extended the function to the 10th stage to detect patterns of 10 digits, taking into consideration human tendencies to think in tens as a result of our base ten number system. Furthermore, where Twerjin (2016) suggested a low correlation function value as more random, we believe the distance from a Cf value of 1 is a better representation, as the theoretical possibility of a number repeating in any stage is 1/10. The correlation function for each position (𝑖𝑖𝑖𝑖) can be found using Equation 2 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) =

𝑋𝑋𝑋𝑋(𝑖𝑖𝑖𝑖) 𝑛𝑛𝑛𝑛

(2)

Equation 2: Correlation Function

The correlation function which is the fractional occurrence of repletion at the 𝑖𝑖𝑖𝑖 th position. In this experiment, the total correlation function is the sum of correlation functions from 𝑖𝑖𝑖𝑖 =1 to 𝑖𝑖𝑖𝑖 =10. X(i) is the number of pairs of an identical event separated by i positions at the sequence with the n number of Science Extension Journal • 77

Scientific Research in School Volume 3 Issue 1 2021

elements. A value of Cf closer to 1.0 suggests greater randomness.

sequence has failed (Cf = 7.15 which is a great distance from 1) as 100% of the time the digit 1 is repeated. Similar to the entropy, a non-random but ordered sequence gives a value for the correlation function is similar to a ‘random’ sequence. Inspecting the graph, for i=1-9 it never has any repetition, however the 10th digit is repeated on many occasions. This suggests that inspecting the graph can confirm whether the correlation function is a valid method for determining randomness. Just as for Entropy, when there are 200 digits involved in a sequence (rather than just 20 in the examples above), the difference between semi-random and the digits of 𝜋𝜋𝜋𝜋 become more obvious.

This can be exemplified with four strings of numbers of differing randomness. Table 2 presents four sequences of numbers and the Correlation functions for each. Table 2: Example of Cf values for example sequences Description

Sequence

Random

86117 38193 26117 93105 37433 78247 51359 62377

Semirandom Nonrandom (ordered)

Correlation Function 0.90

Further from predicted value of repetition.

0.50

This value shows a lack of repetition. However close analysis of the sequence and Figure 2 This value is significantly larger as it shows a pattern and over repetition.

7.25

11111 11111 11111 11111

Closer to predicted value of repetition.

0.75

01234 56789 01234 56789

Nonrandom (repeated)

Comments

Summary Drawing on the literature, as there remains ambiguity on whether humans can do randomness, we will specifically probe whether humans are more proficient at producing randomness over a certain variable.

In this study humans attempted to generate a random number sequence in two ways, by speaking and by writing as there is no previous literature to suggest either variable to be of greater benefit when performing RSG. The prediction is that RSG when speaking may be easier at writing as previous digits or strings may have less influence on later digits which is a key tenet of randomness.

While the total correlation function (summing over i =1-10) provides a measure of randomness it can be more deeply explored and understood by graphing the correlation function for each value of i (Figure 1).

Should humans struggle with randomness, this study sought to provide concrete areas where they struggled and whether that is consistent over the variable of speaking and writing.

The visualized patterns in Figure 1 are a clear indicator of how a more ‘random’ sequence behaves. Unsurprisingly, the non-random (repeated) 1 0.9

0.8 0.7

Random

0.6

Semi-Random

0.5

Non-Random (Ordered)

0.4

Non-Random (Non-Ordered)

0.3 0.2 0.1 0

i positions Figure 1: Visual representation of Cf by stages for levels of randomness listed in Table 2 78 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Scientific research question Can humans generate random sequences and does the methods of writing and speaking act as a variable in the randomness of the sequence generated?

Scientific hypothesis That, humans are more effective at generating a string of random numbers when they are spoken aloud rather than when they write them on a piece of paper.

Methodology Ethics Statement Each participant was informed that their involvement in the study was voluntary, and that the results of their participation would not be traced or impact their relationship with researcher and school. The research was deemed to have minimal impact to participants with approval from the Barker Institute. Preparation A pilot test, of 6 participants who were asked to recite or write 200 digits in an order they considered to be random, was conducted to determine how to best conduct the actual experiment. The results concluded that providing 200 digits did not prove too difficult for the participants, however we do not believe participants would be comfortable giving over 300 digits as they were already struggling by the last 20 digits in the 200. Some participants took longer than other and the entire process for each individual averaged around 10 minutes, with the recording of each set of digits taking 3-4 minutes and the supplementary questions being 1 - 2 minutes. As each of these participants only did one, their results were only used for the pilot study and not used in the 21 described below. Data Collection A sample group of 21 student subjects (min. age: 15, max. age: 18) with no previous knowledge of the experiment were asked to provide a string of 200 from the digits (0,1,2,3,4,5,6,7,8,9) via two different methods (written and spoken); a larger sample size than Persaud (2005) as argued by Figurska et al. (2008). Each participant was asked to write down the digits in a 20x10 grid or asked to speak out loud and have their sequence recorded by the researcher. They were then asked to complete the other activity via the other method; the order of method s was alternated between participants (see Table 4). The participant was then asked supplementary

questions, asking them to comment and reflect on their own performance. This was to ensure data from participants were from an “engaged attempt”. The data was typed and analysed in excel (see Table 3: Results). Data Analysis With the independent variable speaking or writing the numbers, the dependent variable is how random the generated sequence is. This was calculated with the suggested method (Barbasz et al., 2008) of two indexes used in conjunction with one another, with the slight adjustment to the correlation function over 10 steps rather than only 9 to ensure repetition is caught and for ease of interpretation The results were analysed in three ways: 1.

Two t-tests were used to compare speaking and writing on each quantitative metric, providing the simplest answer to the research question. If both metrics return p < 0.05 the result will be considered significant as there is still only a one-in-twenty probability that a false positive would be returned from a non-different sample. However, if only one returns p < 0.05 then a higher threshold will be required (α=0.025), as there is a one-in-ten probability that a false positive would be returned for just one of the two metrics if α=0.05.

The quantitative measures were compared to the Shannon’s Entropy and Correlation Function for a known random sequence (𝜋𝜋𝜋𝜋) and classified categorically as either highly random (if it is closer to the ideal value than 𝜋𝜋𝜋𝜋), random (if it is within twice the distance of 𝜋𝜋𝜋𝜋, nonrandom (within three times the distance of 𝜋𝜋𝜋𝜋), or highly non-random. Histograms were produced indicating the randomness across each of these functions. Results for speaking and writing were then compared with two X2 tests of difference.

Finally, case studies were investigated including closer inspection of the distribution of the frequency of each digit (related to Entropy) and distribution of the Correlation Function to explore what was preventing individuals (if anything) from achieving randomness.

Science Extension Journal • 79

Scientific Research in School Volume 3 Issue 1 2021

Results The following (Tables 3 and 4) are the processed values for Entropy and Correlation Function. Table 3: Resultant values for H and Cf Spoken

H (ideal = 3.322) 3.283

0.690

3.263

3.158

0.810

3.035

1.00

3.275

0.770

3.273

0.72

Subject Number

Written First

Spoken First

Written

Cf (ideal = 1.000)

H (ideal = 3.322)

Cf (ideal = 1.000) 0.80

3.284

0.740

3.302

0.70

3.274

0.675

3.274

0.76

3.306

0.730

3.228

0.82

3.273

0.740

3.257

0.87

3.284

0.725

3.293

0.74

3.220

0.760

3.199

0.86

3.297

0.720

3.271

0.76

3.243

0.830

3.268

0.94

3.240

0.760

3.241

0.83

3.286

0.685

3.277

0.74

3.288

0.800

3.282

0.79

3.144

0.795

3.105

0.93

3.243

1.130

3.282

1.19

3.284

0.885

3.267

0.84

3.307

0.675

3.287

0.72

3.286

0.720

3.274

0.78

3.305

0.670

3.292

0.80

3.299

0.615

3.246

0.78

Average

3.283

0.690

3.248

0.82

The first 200 digits of 𝜋𝜋𝜋𝜋 have an entropy of 3.29416 and correlation function of 0.895. This establishes a threshold for considering the randomness of a sequence. If the distance from the ideal Entropy and Correlation Function for any 200-digit string is no more than the distance of 𝜋𝜋𝜋𝜋 from the ideal, then by the measures used in this paper it can be classified as “highly random”. If it is no more than twice the distance of 𝜋𝜋𝜋𝜋 from the ideal, it is classified as “random”. Any further it is non-random with more than four times the distance of 𝜋𝜋𝜋𝜋 from the ideal being classified as “Highly non-random”. These criteria are shown in Table 5. Table 6 is a reproduction of the full results table (Table 3) with the colour coding in relation to 𝜋𝜋𝜋𝜋. Table 5: Classifications for levels of randomness H

Ideal

3.32193

1.000

𝜋𝜋𝜋𝜋

3.29419

0.895

|Ideal - 𝜋𝜋𝜋𝜋 |

0.02774

0.105

Highly random

0.055

0.210

random

0.083

0.315

non-random

4*|Ideal- 𝜋𝜋𝜋𝜋 |+

0.111+

0.420+

Highly non-random

2*|Ideal- 𝜋𝜋𝜋𝜋 | 3*|Ideal- 𝜋𝜋𝜋𝜋 |

Table 6: Values of H and Cf for each sequence, colour coded by classifications of randomness determined from Table 5. Spoken

Table 4: Summary Table Spoken H

Subject Number

Written H

Spoken first

3.268

0.774

3.255

0.837

Written First

3.263

0.745

3.242

0.812

Because both tests returned values less than 0.05, it can be reported that there was a significant difference in both metrics. RSG of 200 digits when writing was quantitatively more random then when speaking. Analysis 1. T-tests Two two-tailed t-tests (for dependent means) were performed comparing the mean Entropy for speaking and writing (t=-2.253307, p=0.03562) and mean Correlation Function for speaking and writing (t=4.683825, p=0.00014). 2. Comparing Entropy and Correlation function to that of a sequence often accepted as showing traits of randomness – 𝜋𝜋𝜋𝜋

80 • Science Extension Journal

If the values for a particular string are equal or closer to the ideal value of 𝝅𝝅𝝅𝝅, then

Written First

Spoken First

Written Cf

(ideal = 1.000)

(ideal = 3.322)

(ideal = 1.000)

0.69 0.81 0.77 0.74 0.675 0.73 0.74 0.725 0.76 0.72 0.83 0.76 0.685 0.8 0.795 1.13 0.885 0.675 0.72 0.67 0.615

3.263 3.035 3.273 3.302 3.274 3.228 3.257 3.293 3.199 3.271 3.268 3.241 3.277 3.282 3.105 3.282 3.267 3.287 3.274 3.292 3.246

0.80 1.00 0.72 0.70 0.76 0.82 0.87 0.74 0.86 0.76 0.94 0.83 0.74 0.79 0.93 1.19 0.84 0.72 0.78 0.80 0.78

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

(ideal = 3.322) 3.283 3.158 3.275 3.284 3.274 3.306 3.273 3.284 3.22 3.297 3.243 3.24 3.286 3.288 3.144 3.243 3.284 3.307 3.286 3.305 3.299

Average

3.283

0.69

3.248

0.82

Highly random Random nonrandom Highly nonrandom

Scientific Research in School Volume 3 Issue 1 2021

Shannon’s Entropy Index From the last four rows of Table 6, figures can be produced to compare the categories of randomness for speaking and writing. Figure 2 presents a histogram for the degree of randomness (as measured by Entropy) for both speaking and writing,

Discussion

12 Count

Further explanation of the result can be seen in the following analysis.

10 8 5

3 3

Highly random

random

Entropy Spoken

2 0

non-random Highly nonrandom Entropy Written

Figure 2: Distribution of levels of randomness in terms of entropy

It can be seen in Figure 2 that six times students were able to have at least the same entropy as 𝜋𝜋𝜋𝜋 and so are classified as “highly random”. This occurred five times when speaking, and only once when writing. However, a X2 test for independence was performed and found no significant difference between the randomness of speaking and writing (X2 = 3.2952, p = .348306). Correlation Function Figure 3 presents a histogram for the degree of randomness (as measured by Correlation Function) for both speaking and writing. 12

Count

4 0

6 2

given was under speaking. However, a X2 test for independence was performed and found no significant difference between the randomness of speaking and writing (X2 = 7.338, p = .062). This is supported by Figure 4 where the average Cf graph for both methods is a similar curve.

0 Highly random

random CF Spoken

As shown in Figure 4, Cf in the first three stages typically was visibly lower than later stages which fluctuate about 0.1 (the predicted value). This suggests that a reason why some subjects scored further from the ideal correlation function was that they were too unlikely to repeat digits within one, two or three places in the sequence. In a truly random sequence, a higher amount of repetition should occur than did for human RSG. This is the first clear improvement that can be made to human RSG. Shown in Table 7, the digit 0 had the greatest difference from the expected value. This reflects the overall distribution of digits, revealing 0’s infrequent use. Table 7: Frequency of digits from combined data of all sequences Digit

Frequency

Expected

682

840

847

840

903

840

941

840

828

840

819

840

805

840

824

840

883

840

868

840

As entropy is a calculation of distribution of digits, again, this suggests a simple ‘lesson’ can be made to improve human RSG; a reminder that the digit “zero” should not be neglected.

non-random Highly nonrandom CF Written

Figure 3: Distribution of levels of randomness for correlation function

It can be seen in Figure 3 that the only three times a “highly random” was achieved was under writing and the only four times “high non-random” was Science Extension Journal • 81

Scientific Research in School Volume 3 Issue 1 2021

Figure 4: Graph of Cf value in stages for averages of spoken and written sequences

Connections to Literature Though the literature pointed towards a human inability to consciously generate randomness (Figurska, et al. 2008; Barbasz, et al. 2008) and hence suggesting spoken results to have yielded higher numbers in the Highly Random’ and ‘Random’ category, in accepting the null hypothesis, these results suggest that there is no clear indication in humans generating ‘better’ randomness in either metric. This study thus concludes three statements: 1. That humans can produce a more random sequence when writing rather than speaking 2. That humans are not wholly bad at randomness 3. That there is potential for humans to improve randomness to a certain extent with simple advice.

Non-Random, where if H and Cf were separate results, the distributed ratio is 44:36 favouring Highly Random – Random. Although the category boundaries are arbitrary and the sample size on the smaller side, along with Persaud (2005) it suggests that though imperfect, humans are not ‘dreadful’ at conscious RSG. The third statement relates to the ‘lessons’ which this study believes are possible for improving RSG from humans. Participants’ results were shown to be held back by a lack of the digit 0 and hesitancy in repetition consecutively and up to 3 digits apart. This is supported by Lopes & Oden (1987) who concluded randomness as ‘improvable’, to an extent.

The first statement is supported by the t-tests that showed written sequences as significantly closer to the ideal value on both metrics of Entropy and Correlation function than spoken sequences.

Hesitancy for repeating numbers immediately as a significant factor detracting from ‘potential randomness’ is evident in the lower values in the first three stages of the Cf graphs (Figure 4) while ‘forgetting’ the digit 0 over other digits is noted in its significantly smaller frequency in the data over all (Table 7). They are also both supported by the participant answers to the supplementary questions.

The second statement is supported by the distribution of the sequences in the categories Highly Random, Random, Non-Random, Highly

However, the identified ‘mistakes’ had an equal chance of occurrence under both spoken and written circumstances, suggesting for future research, a

82 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

second round of data collection should be conducted with the same participants after being provided adequate feedback (Lopes & Oden, 1987) in order to analysed ‘improved’ results, before the investigating the relationship between spoken and written RSG. This study thus proposes that further research should have sequences resemble ‘optimal’ randomness before conducting research into various factors that may affect RSG.

Conclusion My research project was on models to quantify the randomness of a sequence of numbers. It involved extracting (and slightly modifying) specific randomness metrics including Entropy (equal distribution of digits) and Correlation function (an appropriate amount of repetition) and justifying the complex mathematical models for each. These were then applied in the context of the human ability for conscious Random Sequence Generation, with focus in particular on the differences between spoken and written sequences. I asked 21 participants to give what they believed to be a random sequence of 200 digits (from 0-9). Data was collected from the participant for spoken and written. The data analysis involved performing two two-tailed t-tests (for dependent means) on the means of Entropy and Correlation Function for speaking and writing, as well as two X2 tests on their categorical proximity to the randomness of 𝜋𝜋𝜋𝜋. The results on my data analysis showed the difference in both Entropy and Correlation Function to be significant according to the t-tests, with a higher level of randomness for writing rather than speaking, contrary to my initial hypothesis.

Acknowledgements I wish to extend my special thanks to Dr Matthew Hill for the extensive supervisory support throughout the whole project, both in data collection and in writing up the report, as well as for the advice and encouragement provided. Thank you also to the students from Year 10 - 12 who participated in the data collection and the teachers who facilitated it.

References

Brugger, P. (1997). Variables That Influence the Generation of Random Sequences: An Update. Perceptual and Motor Skill 84, 627–661. Church, A. (1940). On the concept of a random sequence. Bulletin of the American Mathematical Society, 46, 130– 135. Eagle, A. (2005). Randomness is Unpredictability. British Journal for the Philosophy of Science. Evans, F. J. (1978). Monitoring attention deployment by random number generation: An index to measure subjective randomness. Bulletin of the Psychonomic Society, 12, 35-38. Figurska, M., Stan ć zyk, M., Kulesza, K. (2008). Humans cannot consciously generate random number sequences: Polemic study. Medical Hypotheses 70, 182–185. Haldane, J. B. S. (1941) ‘The Faking of Genetical Results’, Eureka 6 (8) 21-24 Kareev, Y. (1992). Not That Bad After All. Generation of Random Sequences. Journal of Experimental Psychology: Human Perception & Performance 18 (4), 1189-1194. Kolmogorov, A. N. (1933). Grundbegriffe Wahrscheinlichkeitsrechnung. Springer.

der

Lopes, L. L. & Oden, G. C. (1987). Distinguishing between random and nonrandom events. Journal of Experimental Psychology: Learning, Memory and Cognition, 13(3), 392-400. Martin-Löf, P. (1966). The definition of random sequences. Information and Control, 9, 602–619. Pandit, J. J. (2012) ‘On Statistical Methods to Test If Sampling in Trials Is Genuinely Random: Editorial’, Anaesthesia 67 (5), 63. Persaud, N. (2005). Humans can consciously generate random number sequences: A possible test for artificial intelligence. Medical Hypotheses 65, 211–214. Shannon, C. (1948). A Mathematical Theory of Communication, Reprinted with corrections from The Bell System Technical Journal, Vol. 27, pp. 379–423, 623–656 Terwijn, S.A. (2016). The Mathematical Foundations of Randomness, in: Landsman, K., van Wolde, E. (Eds.), The Challenge of Chance: A Multidisciplinary Approach from Science and the Humanities, The Frontiers Collection. Springer International Publishing, Cham, pp. 49–66. Towse, J. N. & Neil, D. (1998). Analysing human random generation behavior: A review of methods used and a computer program for describing performance. Behavior Research Methods, Instruments & Computers. 30(4), 583591. von Mises, R. (1919). Grundlagen der Wahrscheinlichkeitsrechnung. Mathematische Zeitschrift, 5, 52–99.

Baddeley, A.D. (2001). Is working memory still working? American Psychologist 56, 851–864.

Wald, A. (1936). Sur la notion de collectif dans la calcul des probabilités. Comptes Rendus des Seances de l’Académie des Sciences, 202, 180–183.

Barbasz, J., Stettner, Z., Wierzchoń, M., Piotrowski, K., Barbasz, A. (2008). How to estimate the randomness in random sequence generation task? Polish Psychological Bulletin 39.

Wald, A. (1937). Die Wiederspruchsfreiheit des Kollektivbegriffes der Wahrscheinlichkeitsrech-nung. Ergebnisse eines Mathematischen Kolloquiums, 8, 38–72. Science Extension Journal • 83

Scientific Research in School Volume 3 Issue 1 2021

84 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Chemistry This year’s chemistry projects were focused in the area of medicinal chemistry, a vitally important research area with the potential to unlock new pathways to treat diseases.

Thomas, Maxine and Jess were part of the Breaking Good project; a citizen science project targeing the availability and cost of essential medicines. Building on previous projects, Thomas and Maxine were able to synthesise new analogues of the antimalarial drug, pyrimethamine. Jess focused on assessing the reliability of the synthesis of this original drug of interest, pyrimethamine, which has been developed by Sydney Grammar School. New areas of research this year were explored by James, Brianna and Kyle. James and Brianna tackled projects investigating the concentration of allicin in garlic cloves. Allicin is a potent natural compound with impressive anti-inflammatory and antioxidant benefits. James made new discoveries about the effects of UV light on the stability of allicin extracted from garlic, an area of research that has yet to be reported on and has potentially vast applications. Brianna’s attention was focused on the effect of temperature on the concentration of allicin extracted. Kyle was interested in spinach and the concentration of chlorophyll measured after harvest. All of these projects took inspiration from nature and a desire to maximise our intake of these medicinally important compounds present in common foods.   Both Charlie and Ollie built on previous research published in this journal. Charlie’s project aimed to investigate the effect of tomato variety on the lycopene concentration extracted and Ollie chose to combine his interests in chemistry and biology by investigating the antibacterial properties of isothiocyanate compounds. By making small changes to the chemical structure of simple compounds, the antibacterial properties can be dramatically altered. Finally, Cleo’s chemistry project probed a very different aspect of the field, how the difficult and abstract concepts in chemistry can be represented using analogies and the effect this has on student learning.

Science Science Extension Journal • 85

Scientific Research in School Volume 3 Issue 1 2021

86 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Synthesis of the 3,4-methylenedioxy analogue of Pyrimethamine Tom Abbott Barker College Malaria remains one of the most pervasive and impactful diseases in tropical regions. This can be attributed to the development of resistance to popular anti-malarial agents through point mutations in the dihydrofolate reductase domain. There is significant potential to overcome such resistance by replacing the substituents on the phenyl ring of Pyrimethamine, a widely used anti-malarial agent. The 3,4-methylenedioxy analogue of Pyrimethamine was successfully synthesised in a high school laboratory following a simplified synthetic procedure compared to that reported in the literature. Time constraints did not allow for biological testing to be undertaken, and further steps must be taken to purify the compound in both steps 2 and 3 of the synthesis. Literature Review Malaria, a treatable and preventable disease, continues to be a significant public health concern, with over 400,000 deaths occurring yearly because of the parasite (WHO, 2020). Due to insufficient access to cheap and effective antimalarial drugs, the problem of parasite resistance (particularly in malaria) has posed significant humanitarian and health issues to impacted regions. In sub-Saharan Africa, there were 380,000 deaths due to Malaria in 2019, with 99.7% of these deaths being directly caused by the Plasmodium Falciparum strain of the parasite (Weiss et al., 2019). While the global incidence rate of malaria has been decreasing, there has recently been a dramatic decrease in the rate of reduction of both its morbidity and mortality in the sub-Saharan African region. While this could be a result of many factors - such as increasingly turbulent weather leading to optimised mosquito breeding conditions - the leading theory for this alarming change is that drug resistance in the parasite, manifesting through rapid mutations in the plasmodial dihydrofolate reductase (DHFR) enzyme, has led to dramatic decreases in the efficacy of several well-established antimalarial agents which act by inhibiting this particular enzyme (Endo et al., 2017). DHFR is an essential target for most antifolate classed anti-malarial drugs. It catalyses the reaction of dihydrofolate to tetrahydrofolate which is part of the thymidylate cycle (Figure 1), making it responsible for the production of thymidylic acid (dTMP), which is an essential nucleotide for DNA synthesis. Inhibition of DHFR thus leads to a deficiency of dTMP, and subsequently DNA, which is crucial for enzyme reproduction (McKie, 1998). Therefore, the efficacy of anti-malarial agents lies in their ability to inhibit the folate biosynthetic pathway, stopping the proliferation

Figure 1: The biosynthesis of tetrahydrofolate. After: (Tropak et al., 2015).

and causing the death of the parasite (Chon, Stover, & Field, 2017). Pyrimethamine (1) is a potent inhibitor of P. Falciparum DHFR and was effectively used to treat P. Falciparum malaria until the widespread appearance of anti-folate resistance significantly diminished its efficacy as an antimalarial agent (Gatton et al., 2004). This resistance was first reported in rural Tanzania during the 1970’s, and in 2006 the enzyme mutations were declared as significant threats to the efficacy of Pyrimethamine and other DHFR inhibitors (Ebel, 2021; Mharakurwa, 2011). Sequencing of the gene from Pyrimethamine resistant strains indicates that resistance has ensued from widespread point mutations in DHFR (Sirawaraporn, 1997). In wild-type DHFR, the critical interaction with

Figure 2 A) Pyrimethamine, B) Pyrimethamine’s position and hydrogen bonding in plasmodial DHFR. Science Extension Journal • 87

Scientific Research in School Volume 3 Issue 1 2021

Figure 2:Evolutionary tree for the development of resistance in malaria. The red text highlights mutations that have been observed, and solid arrows indicate the anticipated pathways by which resistance develops. (After: Sirawaraporn, 2003).

Pyrimethamine (1) is the hydrogen bonding that occurs at the active site between the amine groups in the drug, and the active site amino acids (Figure 2B), allowing the drug to inhibit the tetrahydrofolate biosynthetic pathway (Tropak, 2015; Gatton, 2004). The correlation between Pyrimethamine resistance and mutations in the DHFR domain has been demonstrated in vitro via mutagenesis of synthetic P. Falciparum DHFR genes (Durland et al., 2021). Mutagenesis involves the transformation of specific malarial parasites with gene constructs carrying individual mutations. Moderate levels of Pyrimethamine resistance initially occur from a single point mutation (S108N) in the DHFR domain, in which a small structural change to the serine residue causes new structural-interference interactions with Pyrimethamine’s benzene ring, prohibiting successful docking and inhibition (Xu et al., 2013). However, double and triple mutations impart much higher levels of resistance. The most common resistant strains contain the double mutations (S108N + C59R) or (S108N + N51), or the triple combination of these mutations (S108N + C59R + N51R) (Figure 3) (Tarnchimpoo et al. 2002). Recently, treatment failure has been reported due to the quadruple mutation (S108N + C59R + N51R + I164L) (Ahmed, 2006; Das et al., 2013). This is of concern to antimalarial efforts as the presence of such mutations places pressure on anti-folate drug availability and efficacy. The exact evolutionary pathways of such mutations are difficult to determine, but it is generally accepted that they develop sequentially according to the pathway outlined in Figure 3 – with single mutant strains predicating the existence of double, triple, and then quadruple mutant variants. The most common mutation occurs at the serine 108 residue (Kamchonwongpaisan, 2004). Resistance follows from a steric clash with Pyrimethamine (1) at the active site due to the compound’s inability to form hydrogen bonds between the NH2 groups on the heterocyclic pyrimidine component and the active site amino acids. While the mutations do not directly interact with Pyrimethamine, they change the structural features of DHFR, creating 88 • Science Extension Journal

spacial and polarity differences in the substituents of the active site, and therefore the nature of the active site bonding (Figure 4) (Nattee et al., 2017).

Figure 4: Position of mutations in DHFR in relation to Pyrimethamine (1) complexed with the enzyme.

The kinetics of inhibition suggests that the introduction of multiple mutations leads to a lower affinity between the enzyme and Pyrimethamine. In biochemistry, the enzyme inhibition constant (Ki), is an important indicator of the efficacy of a specific substrate against specific enzymes. The value of the Ki constant is numerically equal to the substrate concentration at which the reaction rate is half of its maximum, which occurs at its saturating point. Triple and quadruple mutant DHFR enzymes demonstrate a decreasing range of Km values that are 2% - 40% of wild-type, indicating that a significantly higher concentration of the substrate is necessary for successful enzyme inhibition. (Sirawaraporn, 1997). While this is the leading hypothesis for increasing levels of drug resistance, the exact structure of the bio-functional enzyme is unknown, and further research must be completed to determine Pyrimethamine’s interactions with both mutant and wild-type DHFR. The most common approach to designing novel medicines is through the continuous re-development of analogues of drugs that have lost their potency. Furthermore, the development of new Pyrimethamine derivatives contributes to our understanding of how

Scientific Research in School Volume 3 Issue 1 2021

molecular structure affects specific interactions with wild-type and mutant DHFR. The majority of analogues that have been synthesised to date consist of various halogen substitutions at positions R2-R6 on the phenyl ring (Figure 5A) (Kamchonwongpaisan, 2017). The focus of this study will be to synthesise and test the 3,4methylenedioxy analogue of Pyrimethamine (2). This analogue shows promising signs of being an effective antimalarial agent, with molecular models indicating an enzyme inhibition constant (Ki) of 1.1 ± 0.3 against S108N single mutant plasmodial DHFR (Nattee et al., 2017) Furthermore, the relatively large size of the methylenedioxy group situated at the R3 and R4 positions on the phenyl ring (Figure 5B) significantly changes the general shape of the molecule, which will provide interesting structure-activity information that may contribute to a more sophisticated understanding of Pyrimethamine’s interactions with DHFR. While molecular modelling has suggested that a large atom in the R4 position may lead to lower inhibition rates due to a spacial clash with the substituents in the enzyme active site, this is not the case in experimental analysis, possibly due to the conformational flexibility of the DHFR enzyme (Chu, 1996). The following research aims to gather data about the methylenedioxy analogue’s antimalarial activity by testing it in vitro against single and double mutant DHFR.

The synthetic pathway for Pyrimethamine (1) developed by Sydney Grammar students will be the focus of this methodology (Figure 6) (OSM, 2016). Students have successfully synthesised Pyrimethamine from 4chlorophenylacetontrile in collaboration with the Breaking Good Project and Sydney University. Furthermore, in 2020 a Barker College student successfully used this method to synthesise the iodoanalogue of Pyrimethamine for the first time, which consists of an iodine substitution at the R4 position (Figure 3) (Barker College, 2020). When considering the methylenedioxy analogue (2), the synthetic pathway - outlined in Figure 6 - remains the same; however, the initial reactant will be changed to 3,4(methylenedioxy)phenylacetonitrile (3) to reflect the structural changes in the final molecule.

Scientific research question Can the methylenedioxy analogue (2) of Pyrimethamine be synthesised from 3,4-(methylenedioxy) phenylacetonitrile (3) in a school laboratory, and its effectiveness as an anti-plasmodial agent be subsequently tested?

Scientific hypothesis That the methylenedioxy analogue (2) of Pyrimethamine can be synthesised from 3,4(methylenedioxy) phenylacetonitrile (3) in a school laboratory, and be tested as an anti-malarial agent against single and double mutant Plasmodium Falciparum.

Methodology General experiment details H NMR spectra were recorded at 300K with a Bruker Avance DRX400 NMR spectrometer. Residual chloroform (δ 7.26) was used as an internal reference for 1

Figure 5: Position of mutations in DHFR in relation to Pyrimethamine (1) complexed with the enzyme.

Figure 6: The proposed synthetic pathway. Science Extension Journal • 89

Scientific Research in School Volume 3 Issue 1 2021

the 1H NMR spectra. The data is reported in terms of chemical shift (δH ppm), relative integral, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet) and assignment. Atom labels on structures are to illustrate 1H NMR spectral assignments and do not necessarily correspond to the IUPAC identifiers provided. Mass spectra were recorded by the Mass Spectrometry Unit of the School of Chemistry at the University of Sydney using an amaZon SL mass spectrometer, and analysed using Bruker Compass DataAnalysis 4.2 software. The molecular ion ([M + H+] or [M – H+]) is listed. Throughout the reaction process, analytical Thin Layer Chromatography (TLC) was conducted in order to gauge the progress of the reaction, and determine the point of completion. TLC was performed using Merck Kieselgel 60 F254 pre-coated aluminium sheets (0.2 mm), and visualisation was enabled by inspection under UV light at 254 nm. The TLC was conducted with either, 50:50 Dichloromethane (DCM) : Hexane, 100% DCM, or 100% ethyl acetate. The eluent for each specific TLC is assigned in the following methodology. Step 1: Synthesis of 3,4-(methylenedioxyphenyl)-3oxopentanenitrile (4)

Figure 7:3-4-(methylenedioxy)phenylacetonitrile

3,4-(methylenedioxy)phenylacetonitrile (10.03 g, 0.062mol, 1 equiv.), ethyl propionate (6.67g, 0.065 mol, 1.05 equiv.) and potassium tert-butoxide (13.97g, 0.12mol, 2 equiv.) were combined in THF (100 mL) at room temperature, stirred in a round bottom flask. The reaction mixture turned a dark red and heated up rapidly. When the mixture appeared homogeneous stirring was turned off. The reaction was sealed and left to sit for 6 hours. The reaction mixture was worked up by pouring onto a 1.0 M HCl in a separating funnel (100 mL). The aqueous layer was extracted with DCM (3 x 50 mL). The combined organic layer was washed with saturated aqueous NaHCO3 (100 ml) and brine (100mL), dried with anhydrous sodium sulfate, filtered, and concentrated in vacuo to afford 3,4(methylenedioxyphenyl)-3-oxopentanenitrile (4) (13.25 g, 0.61 mol, 98%) as a reddish oil. TLC was conducted with 50:50 DCM : Hexane as the eluent. The crude 3,4(methylenedioxyphenyl)-3-oxopentanenitrile was not purified further before being used in step two. 90 • Science Extension Journal

Step 2: Synthesis of 2-(benzo[d][1,3]dioxo-5-yl)-3isobutoxypent-2-enenitrile (5)

Figure 3: 2-(benzo[d][1,3]dioxo-5-yl)-3- isobutoxypent-2enenitrile

3,4-(methylenedioxyphenyl)-3-oxopentanenitrile (4) (13.25 g, 0.61 mol) was dissolved in a mixture of toluene (65.0 mL) and 2-methylpropan-1-ol (6.50 mL). 18M H2SO4 (2.00 mL) was added, and the mixture was refluxed for 10 hours in a Dean Stark apparatus. The reaction mixture was poured onto a saturated sodium hydrogen carbonate solution (100 mL) in a separating funnel and the aqueous phase was extracted with DCM (2 x 50 mL). The combined organic extracts were dried over anhydrous sodium sulfate. Addition of 2.5 mL of triethylamine to the reaction mix converted the unreacted starting material to its very polar triethylammonium enolate salt. Chromatography silica (25g) was added to the organic phase, which was made up to 200mL with DCM and stirred for two hours. The organic phase was then filtered under vacuum and rinsed with 1M HCl (100 mL) and deionised water (50 mL) to remove all traces of triethylamine. The solvent was removed in vacuo to yield 2-(benzo[d][1,3]dioxo-5yl)-3-isobutoxypent-2-enenitrile (5) (12.03g, 0.044 mol, 73%) as a reddish oil. The product was used without further purification in the next step of synthesis. Step 3: Synthesis of 3,4-methylenedioxy pyrimethamine analogue (2)

Figure 9: 3,4-methylenedioxy pyrimethamine analogue

2-(benzo[d][1,3]dioxo-5-yl)-3-isobutoxypent-2enenitrile (12.03 g, 0.044 mol, 1 equiv.) was dissolved in DMSO (225 mL). Guanidine hydrochloride (9.07 g, 0.095 mol, 2.2 equiv.) was stirred into the solution followed by sodium methoxide powder (5.95 g, 0.11 mol, 2.5 equiv.). The solution became dark red in colour on addition of the sodium methoxide, which dissolved into the solution within an hour. No precipitation of sodium chloride was observed.

Scientific Research in School Volume 3 Issue 1 2021

The solution was allowed to stand at room temperature for 24 hours before being poured onto water and extracted with DCM. An emulsion formed in the separating funnel and was left to settle over 48 hours.

Crystals formed in the DCM/water mixture and were isolated to afford 2.7g of sandy orange organic compound. TLC was conducted in 100% ethyl acetate to confirm the existence of a polar compound.

Results Step1: Synthesis of 3,4-(methylenedioxyphenyl)-3-oxopentanenitrile

Figure 10: : 1H NMR spectrum after step 1. 1H NMR (500 MHz, chloroform-d): δ 1.05 (t, J = 7.2 Hz, 3H), 2.62 (m, 2H), 4.57 (s, 1H), 6.01 (s, 2H), 6.84 (m, 3H)

Figure 41: Mass spectrum after step 1. A Compound in positive ESI zoomed around peak at 345.05 Da. B Compound in negative ESI zoomed around peak at 216 Da.

Science Extension Journal • 91

Scientific Research in School Volume 3 Issue 1 2021

Step 2: Synthesis of 2-(benzo[d][1,3]dioxo-5-yl)-3-isobutoxypent-2-enenitrile

Figure 125: 1H NMR spectrum after step 2. 1H NMR (500 MHz, chloroform-d): δ 1.05 (t, J = 7.2 Hz, 3H), 0.95-0.90 (d, 6H), 1.98 (m, 1H), 1.28 (t, 3H), 2.71 (q, 2H), 3.78 (d, 2H), 5.59 (s, 2H), 6.78 (s, 2H)

Step3: Synthesis of 3,4-methylenedioxy pyrimethamine analogue

Figure 13 : 1H NMR spectrum after step 3. 1H NMR (500 MHz, chloroform-d): δ 1.05 (t, J = 7.2 Hz, 3H), 1.07 (t, 4H), 2.32 (m, 6H), 4.49 (s, 2H), 4.71 (s, 2H), 4.88 (s, 2H), 6.01 (s, 2H), 6.69 (s, 3H)

92 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

TLC for steps 1,2 and 3

Figure 14: Thin Layer Chromatography (TLC) performed with 100% DCM for steps 1 (A), 2 (B) and 3 (C) as labeled.

Discussion Step 1: Synthesis of 3,4-(methylenedioxyphenyl)-3oxopentanenitrile

Figure 15: Overview of the reaction in step 1, forming 3.4(methylenedioxyphenyl)-3-oxopentanenitrile.

Step 1 of the synthetic pathway (Figure 15) involved a condensation reaction between 3,4-(methylenedioxy) phenylacetonitrile (3) and ethyl propionate, forming compound 4. Initially, addition of potassium tertbutoxide, a strong base, enabled the deprotonation of the CH2 group of compound 3 (Figure 16). Subsequently, reaction with the highly volatile ethyl propionate led to elimination of an ethoxide group, enabling the condensation reaction in Figure 17 to afford 3,4(methylenedioxyphenyl)-3-oxopentanenitrile (Compound 4).

Figure 15: Visualising the deprotonation of the CH2 group in 3,4-(methylenedioxy)phenylacetonitrile.

Figure17: Mechanism of the reaction between 3,4(methylenedioxy) phenylacetonitrile and ethyl propionate.

The yield for this reaction was 98.5%, significantly higher than both Sydney Grammar’s 2016 Pyrimethamine synthesis which had a yield of 90%, and the Barker College 2020 iodo-Pyrimethamine analogue synthesis, which had an estimated yield of 68% (OSM 2017; Barker College 2020). This can be attributed to more optimised reaction conditions, as well as an increased quantity of the starting material (0.06 mol) in comparison to previous synthesis attempts using the same pathway. Furthermore, the highly electronegative nature of the oxygen atoms at the R3 and R4 positions could have led to a more polar, and thus more soluble compound, creating a more homogenous reaction mixture and further optimising the reaction process. Supporting this high yield, the 1H NMR confirmed the existence of the desired compound, and only indicated the existence of some slight impurities, with a 1H NMR spectrum comparable to that which was already available in the literature (Havel et al., 2018). The multiplet at 2.62ppm was assigned to the ethyl CH2 protons, which is expected to appear as a quartet due to coupling with the neighbouring methyl group. This signal may have been complicated due to a slight impurity such as the presence of the enol tautomer in the reaction mixture, or due to restricted rotation of the molecule. The singlet further downfield at 4.59ppm was assigned to the CH proton H4 (Figure 1). This was likely shifted downfield from the ethyl group signals due to the electron withdrawing nature of the nitrile group causing a deshielding of the hydrogen atom. Similarly, the singlet at 6.01ppm was assigned to the methylenedioxy hydrogens due to even more pronounced deshielding from the two adjacent oxygen atoms. Meanwhile, the multiplet at 6.84ppm was assigned to the aromatic protons H1 – H3, which likely appeared downfield due to the deshielding effect of the benzene ring. Science Extension Journal • 93

Scientific Research in School Volume 3 Issue 1 2021

Mass spectroscopy (Figure 13) supported the formation of the desired compound, with a peak in the negative mode at m/z 215.71, which is close to the desired M-H peak molecular weight. With a high yield, indicative of a sufficiently optimised reaction, and the production of a sufficiently clean crude compound which was confirmed to be compound 4, step 1 of the synthesis provided a good baseline for further reactions. Step 2: Synthesis of 2-(benzo[d][1,3]dioxo-5-yl)-3isobutoxypent-2-enenitrile

Figure 18: Overview of the reaction in step 2, forming 2(benzo[d][1,3]dioxo-5-yl)-3-isobutoxypent-2-enenitrile.

Step 2 of the synthesis (Figure 18) involved a dehydration reaction performed under reflux between 3,4-(methylenedioxyphenyl)-3-oxopentanenitrile and 2methylpropanol, forming compound 5. Initially, addition of 2-methylpropanol to compound 4 and subsequent protonation afforded intermediate ß. Spontaneous elimination of water and regeneration of the H+ catalyst afforded the desired compound 4. (Figure 20)

impurity in the compound, possibly left over starting material (Compound 4). The singlet at 6.78ppm was assigned to the aromatic protons, primarily because of their proximity to the deshielding benzene ring pulling them downfield. Similarly, the protons on the methylenedioxy group were assigned to the signal at 5.95ppm. The doublet at 3.78ppm is characteristic of the isopropyl CH2 protons, while the ethyl CH2 and CH3 protons were assigned to the multiplets in the 1-2.5ppm range due to their location in the molecule. These signals characteristically appear as quartets, but may not appear as expected due to impurities. Mass spectrometry was not able to be performed, however the placement and integration of the 1H NMR signals were sufficient in confirming the existence of compound 5. Further purification of the reaction product would be required to obtain full characterisation data for this compound as it has thus far not been reported in the literature. Although work needs to be carried out to improve the purity of compound 5, it was decided that there was a sufficient quantity of the desired product to proceed with the next step of the synthesis. Step 3: Synthesis of 3,4-methylenedioxy pyrimethamine analogue

Figure 8:Synthesis of 3,4-methylenedioxy pyrimethamine analogue Figure 6: : Mechanism of the substitution reaction between 3,4-(methylenedioxyphenyl)-3-oxopentanenitrile and 2methylpropanol.

Perhaps the most challenging aspect of step 2, was the production of a relatively large amount of water (Figure 20), which was removed in a Dean Stark apparatus under high temperatures, driving the equilibrium reaction in its forward, endothermic direction in order to maximise yield.

Figure 7: Mechanism of the substitution reaction resulting in the formation of water, explaining the need for reflux.

The yield for this step was 73%, which was higher than Sydney Grammar’s step 2 yield of 58%, but also does not account for impurity and residual toluene in the reaction product. Due to the difficulty of this step, likely caused by the equilibrium nature of the reaction, the 1H NMR spectrum was difficult to assign as there was some 94 • Science Extension Journal

Step 3 of the synthetic pathway (Figure 21) involves a complex reaction which is initiated by the deprotonation of guanidine hydrochloride by sodium methoxide. Following this, a reaction occurs between the nitrile carbon of compound 5 and the guanidine which results in an electron rearrangement that facilitates the elimination of the 2-methylpropanol group, allowing the formation of the heterocyclic pyrimidine component of compound 2. The product was difficult to isolate, and 72 hours after the reaction no crystals had come out of the solution. Because of this, the residual solid at the bottom of the round bottom flask was separated and run through a vacuum filter affording a highly insoluble compound. 1 H NMR analysis of this solid indicated no 1H peaks in the NMR spectrum, suggesting that this product was an inorganic salt. After leaving the remaining mixture in a separating funnel with water for several days, 1.94g of another solid was isolated. While this appears to be a relatively small yield of <30%, it is likely that there is additional product in the remaining solution which

Scientific Research in School Volume 3 Issue 1 2021

could be isolated at a later time and purified. Gratifyingly, 1H NMR analysis indicated that this isolated solid was the methylenedioxy analogue, compound 2. As with step 2, the 1H NMR spectra for this compound was complex and hard to assign, likely due to minor impurities. The appearance of polar baseline material in the TLC (Figure 14C) confirmed the existence of a highly polar compound, which is characteristic of Pyrimethamine analogues such as 5 due to the highly polar amino groups on the pyrimidine ring. The appearance of two broad singlets at 4.49 and 4.71 ppm are characteristic of these NH2 groups. Furthermore, the triplet at 1.07 ppm was indicative of the CH2 on the ethyl group. Two more singlets at 6.01 and 6.69 ppm were likely caused by the aromatic protons and the methylenedioxy protons respectively, these signals being pulled downfield due to the highly electronegative oxygen atoms and deshielding benzene ring. The signal at 1.75 ppm was characteristic of water, which is difficult to remove from the reaction product.25 Furthermore, a large singlet at 2.62 ppm indicated that a significant volume of solvent (DMSO) remained in the reaction mixture, due to the high boiling point of this solvent. As the relevant signals were clearly apparent in the 1H NMR spectrum alongside minor impurities, the evidence was strong enough to confirm the formation of compound 2. Time constraints and an impure final compound did not allow for biological testing of the compound as an anti-malarial agent. Compound 2 is currently at Sydney University awaiting further purification. If this results in a pure sample of this compound, biological testing will be carried out very soon. Future Research The successful high school synthesis of Pyrimethamine analogues helps to widen our understanding of the synthesis process developed by Sydney Grammar, and also provides new interesting structural-activity information in regards to Pyrimethamine’s interactions with the DHFR enzyme. With regard to this report, the synthesis of the 3,4methylenedioxy analogue needs to be revised and subsequently revisited. Particular areas of interest include: • Step two of the synthesis, in which compound 3 needs to be recreated with a higher purity such that further 1H-NMR and 13C-NMR can be conducted, and the spectra for this compound confirmed before being included in the Breaking Good database. • Step three of the synthesis, in which further optimisation needs to be completed in order to

obtain a higher yield and subsequent purification, allowing for biological testing to be undertaken. It is imperative that further research is done in order to ‘stay ahead’ of mutations in the virus. Generally, this can be done by continuing to develop and synthesise new analogues of Pyrimethamine in an attempt to develop both affordable and accessible medicines. If not, we could observe alarming exponential growth in both the mortality and morbidity rates of P. Falciparum, particularly in the Sub-Saharan Africa region.

Conclusion The research described in this report resulted in the successful synthesis of the 3,4-methylenedioxy analogue (2) of Pyrimethamine using the synthetic pathway developed by Sydney Grammar School. The formation of the appropriate product after each step was confirmed using 1H NMR spectroscopy and mass spectroscopy. Being the second analogue synthesised using this pathway, this synthesis confirms the feasibility of using the Sydney Grammar pathway to create affordable analogues of Pyrimethamine, particularly those which include various non-halogenic substitutions at the R3-5 positions on the phenyl ring. However, due to a relatively small yield because of impurities and isolation difficulties introduced in steps 2 and 3 of the synthesis, the analogue was not able to be submitted for biological testing against P Falciparum and to obtain its enzyme inhibition data. The reaction pathway should be refined, and the product purified after each step to allow for biological testing in the future. Doing so will allow more structural-activity information to be collected regarding Pyrimethamine’s interactions with the DHFR enzyme.

Acknowledgements I would like to thank Dr Katie Terrett. Throughout this project, she has invaluable guidance and assistance, providing insights into my report, and making herself available to explain concepts and help me understand the intricacies of the research. This would not have been possible without her. I would like to thank Dr Michael Tropak, for providing assistance and guidance about using PyMol, as well as directing me towards the RCSB structure of DHFR complexed with Pyrimethamine. I would also like to thank collaborators of the Breaking Good project at Sydney University, who ran NMR and mass spectroscopy analysis of my compounds throughout the reaction process.

References Ahmed, A. (2006). Quadruple Mutations in Dihydrofolate Reductase of Plasmodium falciparum Isolates from Car Science Extension Journal • 95

Scientific Research in School Volume 3 Issue 1 2021

Nicobar Island, India. Antimicrobial Agents and Chemotherapy, [online] 50(4), pp.1546–1549. Available at: https://dx.doi.org/10.1128%2FAAC.50.4.1546-1549.2006 [Accessed 6 Jun. 2021]. Chon, J., Stover, P.J. and Field, M.S. (2017). Targeting nuclear thymidylate biosynthesis. Molecular Aspects of Medicine, 53, pp.48–56. Chu, E. and Allegra, C.J. (1996). The role of thymidylate synthase in cellular regulation. Advances in Enzyme Regulation, [online] 36, pp.143–163. Available at: https://pubmed.ncbi.nlm.nih.gov/8869745/. Das, S., Chakraborty, S.P. and Hati, A. treatment failure with novel mutation in falciparum dihydrofolate reductase (pfdhfr) West Bengal, India. International Journal Agents, 41(5), pp.447–451.

(2013). Malaria the Plasmodium gene in Kolkata, of Antimicrobial

humans and Anopheles mosquitoes. Proceedings of the National Academy of Sciences, 108(46), pp.18796–18801. Nattee, C. and Khamsemanan, N. (2017). A novel prediction approach for antimalarial activities of Trimethoprim, Pyrimethamine, and Cycloguanil analogues using extremely randomized trees. Journal of Molecular Graphics and Modelling, 71, pp.13–27. Open Source Malaria. (2016). Sydney Grammar School Synthesis. [online] Available at: http://malaria.ourexperiment.org/daraprim_synthesis/15813/P yramethamine_synthesis_Status_at_the _end_of_2016.html [Accessed 6 Jun. 2021]. Ouellette, M., Leblanc, É. and Kündig, C. (1998). Antifolate Resistance Mechanisms from Bacteria to Cancer Cells with Emphasis on Parasites. Resolving the Antibiotic Paradox, pp.99–113.

Durland, J. and Ahmadian-Moghadam, H. (2021). Genetics, Mutagenesis. [online] PubMed. Available at: https://www.ncbi.nlm.nih.gov/books/NBK560519/# [Accessed 6 Jun. 2021].

Quan, H. (2020). High multiple mutations of Plasmodium falciparum-resistant genotypes to sulphadoxinepyrimethamine in Lagos, Nigeria. Infectious Diseases of Poverty, 9(1).

Ebel, E.R., Reis, F. and Petrov, D.A. (2021). Historical trends and new surveillance of Plasmodium falciparum drug resistance markers in Angola. Malaria Journal, 20(1).

Sardarian, A. (2003). Pyrimethamine analogs as strong inhibitors of double and quadruple mutants of dihydrofolate reductase in human malaria parasites. Organic & Biomolecular Chemistry, 1(6), pp.960–964.

Endo, N., Yamana, T. and Eltahir, E.A.B. (2017). Impact of climate change on malaria in Africa: a combined modelling and observational study. The Lancet, 389(Special Issue), p.S7. Gatton, M.L., Martin, L.B. and Cheng, Q. (2004). Evolution of Resistance to Sulfadoxine-Pyrimethamine in Plasmodium falciparum. Antimicrobial Agents and Chemotherapy, 48(6), pp.2116–2123. Gottlieb, H.E., Kotlyar, V. and Nudelman, A. (1997). NMR Chemical Shifts of Common Laboratory Solvents as Trace Impurities. The Journal of Organic Chemistry, 62(21), pp.7512–7515. Havel, S., Khirsariya, P. and Akavaram, N. (2018). Preparation of 3,4-Substituted-5-Aminopyrazoles and 4Substituted-2-Aminothiazoles. The Journal of Organic Chemistry, 83(24), pp.15380–15405. Hyde, J.E. (2007). Drug-resistant malaria − an insight. FEBS Journal, 274(18), pp.4688–4698. Issuu. (2020). Barker College Science Extension Journal. [online] Available at: https://issuu.com/barkercollege/docs/2020_science_ext_journ al_in_pdf_ [Accessed 6 Jun. 2021]. Kamchonwongpaisan, S. (2004). Inhibitors of Multiple Mutants ofPlasmodiumfalciparumDihydrofolate Reductase and Their Antimalarial Activities. Journal of Medicinal Chemistry, 47(3), pp.673–680. McKie, J.H. (1998). Rational Drug Design Approach for Overcoming Drug Resistance: Application to Pyrimethamine Resistance in Malaria. Journal of Medicinal Chemistry, 41(9), pp.1367–1370. Mharakurwa, S., Mkulama, M.A.P. and Musapa, M. (2011). Malaria antifolate resistance with contrasting Plasmodium falciparum dihydrofolate reductase (DHFR) polymorphisms in 96 • Science Extension Journal

Sirawaraporn, W., Sathitkul, T., Sirawaraporn, R., Yuthavong, Y. and Santi, D.V. (1997). Antifolate-resistant mutants of Plasmodium falciparum dihydrofolate reductase. Proceedings of the National Academy of Sciences, [online] 94(4), pp.1124– 1129. Available at: https://www.pnas.org/content/94/4/1124. Tarnchompoo, B. and Sirichaiwat, C. (2002). Development of 2,4-Diaminopyrimidines as Antimalarials Based on Inhibition of the S108N and C59R+S108N Mutants of Dihydrofolate Reductase from Pyrimethamine-Resistant Plasmodium falciparum. Journal of Medicinal Chemistry, 45(6), pp.1244– 1252. Tropak, M.B. (2015). Pyrimethamine Derivatives: Insight into Binding Mechanism and Improved Enhancement of Mutant βN-acetylhexosaminidase Activity. Journal of Medicinal Chemistry, 58(11), pp.4483–4493. Weiss, D.J. (2019). Mapping the global prevalence, incidence, and mortality of Plasmodium falciparum, 2000–17: a spatial and temporal modelling study. The Lancet, 394(10195), pp.322–331. World Malaria Report 2020: 20 years of global progress and challenges. (2020). [online] Geneva: World Health Organisation. Available at: https://apps.who.int/iris/rest/bitstreams/1321872/retrieve [Accessed 4 Jun. 2021]. Xu, M. and Zhu, J. (2013). Novel Selective and Potent Inhibitors of Malaria Parasite Dihydroorotate Dehydrogenase: Discovery and Optimization of Dihydrothiophenone Derivatives. Journal of Medicinal Chemistry, 56(20), pp.7911–7924.

Scientific Research in School Volume 3 Issue 1 2021

Concentration of Allicin in Garlic Brianna Lollback Barker College The medicinal benefits of garlic have been extensively studied and attributed to the medicinally active thiosulfinate, allicin. This report outlines an investigation into whether temperature can be used to maximize the concentration of allicin in garlic, potentially maximising the medicinal benefits an individual can receive from consuming raw garlic. This investigation observed the effect of temperature on the concentration of allicin in raw garlic cloves, by using a spectrophotometric method that is sensitive enough to measure concentration of allicin in the micromolar range. The garlic cloves were subjected to different temperatures before the garlic was dehydrated and crushed into a fine powder and placed in solution. The data was recorded by first reacting the garlic extract with excess Lcysteine and then 5,5’-dithiobis-2-nitrobenzoic acid which measured the decrease in cysteine concentration. The results of the experiment found that temperature did not have a significant effect on the concentration of allicin measured when garlic cloves were stored over a period of four days. Literature review Garlic (Allium sativum L.) has been applied to culinary and medicinal purposes in modern and ancient practices (Peyman et al. 2013; Gaber et al., 2020; Azene, 2015). Garlic contains a rich source of organosulfur compounds responsible for its flavour and aroma as well as its health benefits (Oregon State University, 1985). Specifically, allicin, an organosulfur compound in garlic is responsible for the majority of the pharmacological activity of crushed raw garlic cloves (Lawson and Hunsaker, 2018), as it is the most biologically active compound in garlic (Rahman, 2007). Allicin, most commonly found in raw garlic, is known to reduce inflammation and offer antioxidant benefits (Bahare, 2019). Additionally, research has focused on allicin’s antimicrobial properties, which has found that allicin in its pure form exhibits antibacterial activity against a wide range of Gramnegative and Gram-positive bacteria, antifungal activity and antiparasitic activity against major human intestinal protozoan parasites such as Entamoeba histolytica and Giardia lamblia and antiviral activity (Ankri and Mirelman, 1999). Allicin is formed immediately in raw garlic as a self defence mechanism when the clove is damaged by worms, fungi, bacteria, or by physically crushing the clove (Leontiev et al. 2014). Allicin is produced by the precursor molecule, alliin (an amino acid) being converted into allicin by action of the alliinase enzyme (Figure 1) (Chhabria and Desai, 2018). The alliinase enzyme is located in the space between garlic cells, whilst alliin is located within the garlic cells themselves. This means that alliin and the alliinase enzyme can only

interact to form allicin once the cell walls of the garlic have ruptured (Janská et al., 2021).

Figure 1: Formation of allicin from alliin catalysed by the alliinase enzyme.

Allicin is known to be a highly unstable and volatile organosulfur compound due to the presence of the thiosulfinate functional group on the molecule (Abe, Hori and Myoda, 2019). This makes allicin heat sensitive, as it rapidly decomposes in the presence of air and water into an abundance of volatile thiosulfinate derivatives (mainly vinyl dithiines, ajoenes and allyl sulfides) (Figure 2) (Cheewinworasak et al., 2018). This is why there is an inability to ensure a certain abundance of allicin is present within a garlic clove. The effect of pH, concentration and light on the stability of allicin Research into the factors that influence the concentration of allicin have been vital in understanding the unstable nature of the allicin compound. A study by Wang et al. (2015) as well as a study by Lawson and Hughes (1992) investigated the influence of pH, concentration and light on the stability of allicin in garlic after crushing. Both

Science Extension Journal • 97

Scientific Research in School Volume 3 Issue 1 2021

Figure 2: Degradation of allicin (After: Oregon State University, 2021)

studies found that at room temperature, allicin in aqueous extract was most stable at pH 4.5-6, however, allicin exposed to pH levels outside this range began to degrade within 30 minutes and were undetectable within 2 hours when the pH was higher than 11 and lower than 3.5. Additionally, both studies found that allicin extract was sensitive to pH and temperature but not to visible light. The paper by Wang et al. (2015) observed that at these temperatures, higher concentrations of allicin in water could be kept for up to five days without obvious degradation. Temperature Research has also looked at the thermal degradation of allicin and the implications on its bacteriostatic properties. Canizares and co-workers (2004) established a link between effective inhibition of the bacterium, Helicobacter pylori (Hp), and the temperature the aqueous garlic extracts were stored at (6 °C, 19°C, 21°C and 26°C) after crushing of the raw garlic material. The research concluded that allicin extract stored at lower temperatures (6 °C) showed the greatest inhibition of the in-vitro growth of Hp and the bacteriostatic properties remained active for up to 10 months of storage at this temperature. Moreover, research by Mansor et al. (2016) has looked at the effect of different storage temperatures of garlic powder on the degradation of allicin within the range of 30°C - 85°C. This paper has shown that allicin is most stable at temperatures around 30 °C, only having a slight

98 • Science Extension Journal

reduction over time. However, at higher temperatures around 70°C-85°C, the organic compound allicin decomposes rapidly (as seen in Figure 3). In addition, a study by Mathialagan et al. (2017) ‘obtained result(s) in accordance with the findings from Mansor et al. (2016) on the thermal stability of Allicin’ (Mathialagan et al. 2017 p.g.1750). Both found that after raising the extraction temperature over 35 °C there was a significant amount of allicin deteoration. In another study by Fujisawa et al. (2008) allicin in an aqueous extract degraded stoichiometrically in proportion to its temperature over the range 4° C - 42° C, and from this the half-lives were estimated to be a year at 4°C (degrading from 1.8 mg/ml to 0.9 mg/ml), 32 days at 15°C and 1 day at 37°C (degrading from 2.0 mg/ml to 1.0 mg/ml).

Figure 3: Allicin content vs. time at various temperatures (Source: Mansor et al., 2016)

Scientific Research in School Volume 3 Issue 1 2021

Maximising alliin concentration There have been no previous reports investigating how a change in the conditions the cloves are exposed to before crushing affects the concentration of allicin after the garlic is crushed. This paper investigates whether a change in storage temperature of the garlic cloves would affect the concentration of alliin precursor, which would result in a change in allicin concentration after crushing. This would aid in understanding how simple adjustments to the storage conditions the garlic cloves are subjected to can enhance the allicin concentration. Thus, maximising the medicinal benefits an individual can receive from consuming garlic. Therefore, this paper investigates the effect of temperatures achievable in a domestic setting on the concentration of allicin. Literature methods for the analysis of allicin content in garlic rely on preparing a stable powder and then extracting the allicin for analysis. Commonly, convective hot air-drying is used to dehydrate garlic. By dehydrating garlic, allicin stability is improved as the reduction in water content considerably minimizes physical, chemical and microbiological degradation during storage which allows for the preservation of allicin content once the garlic has been crushed to form the allicin (Papu et al., 2014). There are several methods reported in the literature for quantifying the allicin content in garlic powder (Ranitha, 2016; Bernhard et al. 1990; Bose, 2014). The simplest method, first reported by Han et al. (1995) is the spectrophotometric method, where allicin is converted into a coloured compound and subsequently analysed using a colorimeter to determine the absorbance and hence the concentration in solution. A benefit of this method is that it does not require an allicin standard to quantitate allicin, instead it relies on the Beer-Lambert law which states that there is a linear relationship between the absorbance and concentration for substances which are able to absorb within the UV/visible region of light (Figure 4). The relationship is expressed as A = εlc, where A is the absorbance, ε is the molar absorptivity (M-1cm-1), l is path length (cm) of the cuvette and c is concentration (mol/L) (Figure 5). Methods which make use of High Performance Liquid Chromatography (HPLC) (Ranitha, 2016) and Gas Chromatography (GC)(Koichi, 1989) for analysis are also commonly reported in the literature, however, we do not have access to the required equipment to carry out this analysis.

Scientific research question How does the storage temperature of garlic effect the concentration of alliin, and hence the concentration of allicin present in garlic?

Figure 3: Calibration curve which relies on the Beer Lambert law demonstrating the relationship between concentration and absorbance (After: UNC Eshelman School of Pharmacy, 2021)

Figure 4: Principles of the Beer-Lambert Law (Source: Brian McNamara, 2018)

Scientific hypothesis That temperature will have an effect on the concentration of allicin extracted after storage at different temperatures.

Methodology Preparation and storage of cloves 5,5’-dithiobis-2-nitrobenzoic acid (DTNB), 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and L-cysteine were purchased from Sigma Aldrich. Five bulbs of garlic and five small plastic bags were purchased from The Veggie Patch (local grocery store). The papery skin of the garlic bulbs was peeled and the cloves were separated from the bulbs. The cloves were placed in a bowl and mixed together. Ten cloves were randomly selected and placed in each of the five plastic bags (50 cloves in total). Each bag was assigned a label: Freezer, Fridge and Room temperature. These bags were then stored at their respective temperatures and were exposed to the same light conditions for four days. Science Extension Journal • 99

Scientific Research in School Volume 3 Issue 1 2021

Dehydration and production of garlic powder Both ends of the garlic cloves were chopped. The cloves were peeled and the peels were discarded. The garlic cloves were then sliced into uniform widths of 2mm. The sliced garlic was then spread onto a lined tray (aiming for as close to a single layer as possible), with sliced garlic from each assigned bag being kept separate. The tray was placed in the oven at 50 °C for 4 hours. The dehydrated garlic was then removed from the tray and allowed to cool at 25 °C for 10 minutes, then pulverized in a mortar and pestle until it was a fine powder. Preparation of solutions Preparation of 5.0 mM DTNB solution 0.404g of DTNB powder was weighed using a mass balance. The weighed powder was transferred into a 200 mL glass bottle and dimethyl sulfoxide (DMSO) (200 mL) was added. The container was wrapped in foil and put into a dark cupboard. Preparation of HEPES buffer To make the 1.0 M HEPES buffer at pH 7.6, a pH-meter was calibrated by placing it into a pH 7.0 buffer solution. Then 23.8g of HEPES buffer powder was weighed into a clean, dry beaker using an electronic balance. Deionised water was used to dissolve the HEPES and the solution was transferred to a 100 mL volumetric flask with the final desired volume made up to 100mL with distilled water. A magnetic stirrer was used for approximately 10 minutes to dissolve the HEPES, with sodium hydroxide slowly added to the solution via a glass pipette whilst constantly measuring the pH using the pH meter until 7.6 was measured.

Preparation of 50 mM HEPES buffer 5.0 mL of 1.0 M HEPES buffer is introduced into an empty 100 mL volumetric flask and the final volume was made up to 100 mL with deionized water. Preparation of 2.0 mM L-cysteine 0.096 g of L-cysteine was weighed using an electronic balance. The crystals were then transferred into a 500 mL glass bottle and 400 mL of deionized water was added into the flask. The mixture was stirred until a solution was obtained.

Analysis 10 g of garlic powder was dissolved in 300 mL of deionized water. The mixture was stirred at room temperature using a magnetic stirrer for 1 hour. Finally, insoluble solids were filtered under vacuum using a sintered glass funnel to collect the dissolved allicin in the filtrate. Then 0.5 mL of garlic extract was added to 1.2 mL of 2 mM L-cysteine and allowed to sit at room temperature for 10 minutes. Then, 3 mL of 50 mM HEPES buffer (pH 7.6) and 1 mL DTNB was added to the solution where it was then stirred and allowed to sit for 2 minutes at room temperature. Absorbance of the samples were read at 450nm with a colourimeter to obtain the concentration of excess L-cysteine remaining in the sample (seen in Figure 6).

Figure 6: Allicin reacting with L-cysteine and DTNB to form a yellow compound

100 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Since the absorbance of the solute varies due to its concentration the relationship is expressed as: A = εlc, where for this specific experiment - A = absorbance measured by the colourimeter at 450nm (maximum absorbance wavelength for the yellow compound) - ε = molar absorptivity which is 14150 M-1cm-1 (The constant value specific for absorption of yellow compound) for this experiment - l = path length of the cuvette (1 cm therefore equals to 1 in the experiment) - c = concentration in mol/L

Results Since the concentration of the excess L-cysteine is twice the concentration of allicin (allicin reacts with L-cysteine in a 1:2 ratio). Concentration of allicin is therefore defined as <the concentration of yellow compound> subtracted from <the initial concentration of L-cysteince> all divided by two. Table 1: Mean and standard deviation for each group temperature

Temp. Control (25°C) Room Temp (25°C) Fridge (3°C) Freezer (-18°C)

Mean Conc. (mol/L)

Mean Conc. (mg/g)

StDev.

2.2925x10-4

1.1274

0.0012

2.2946x10-4

1.1284

0.0011

2.2960x10-4

1.1291

0.0014

2.2948x10-4

1.1285

0.0003

The mean concentration allicin (mg/g) was calculated by multiplying the mean concentration (mol/L) by the molecular mass of allicin (162.28 g/mol) and by a factor of 1000 to get mg/L. This value given in mg/L was then divided by 33g (as there was 33g of powder in each litre of water) to give a value in mg/g.

Table 2: ANOVA test output for comparing the concentration of allicin after being stored at selected temperatures

H0= There is not a statistically significant difference between the concentration of allicin extracted from garlic cloves stored at different temperatures HA= There is a statistically significant difference between the concentration of allicin extracted from garlic cloves stored at different temperatures F-stat 2.1135 α 0.05 P-value 0.1387 Analysis The result is not significant as p > 0.05

Discussion The ANOVA statistical analysis (Table 2) has a p-value of 0.1387, since the p > 0.05, the null hypothesis (H0= There is not a statistically significant difference between the concentration of allicin extracted from garlic cloves stored at different temperatures) was accepted. This indicates that temperature of garlic cloves does not play a major role in influencing allicin concentration, in contrast to the effect of temperature on the degradation of allicin. The standard deviation for each of the four groups is small, indicating that there is not a large amount of variation from the mean in each group, indicating the high reliability of the data. In the original method (Han et al, 1995) 0.5 mL of garlic extract is added to 1.2 mL of 2 mM L- cysteine and allowed to sit at 30 °C. However, in my experiment it was not kept at 30°C but instead at room temperature (approx. 25°C). This slightly lower temperature may have slowed down the reaction rate. Additionally, my analysis took place over two days, with the garlic extract (garlic powder in solution) being kept in a cupboard overnight. This may have impacted the results as the second day displayed more stable absorbance readings. This is likely due to the additional storage time facilitating further settling of any fine suspension present in the extract. Samples with fine suspension will likely affect the absorbance readings obtained by the colourimeter due to light scattering. This suggests that garlic extract should be left for a period of time to allow all debris to collect on the bottom of the container to gain a better set of results. Due to time constraints, we couldn’t repeat this part of the experiment. Additionally, there was no access to a centrifuge which would be another way of mitigating the presence of any fine suspension on the extract. Since there is not a certain abundance of garlic in each clove, this was accounted for by mixing the garlic cloves together and then randomly selecting 10 to place into each bag. In the original method the absorbance readings were read at 412 nm, however my absorbance readings (Appendix 1.1) were read at 450 nm as this was the closest wavelength the colourimeter could be set to.

Science Extension Journal • 101

Scientific Research in School Volume 3 Issue 1 2021

Previous studies have used a much larger range of temperatures when investigating allicin concentration. However, these values were not practical for this project due to the incubation conditions available in school. Moreover, temperatures which would be more applicable to household settings were chosen to enhance the applicability of any results obtained. The effect of higher or lower garlic clove storage temperatures cannot be ruled out and may be a subject for future investigations. These results cannot be compared to other results in the literature as there are no literature reports on the effect of temperature on alliin concentration, or the effect of temperature before the extraction of allicin. It can be speculated that a similar experiment has been conducted and not reported in the literature because it was found that temperature did not have an effect. There are not many reports on the stability of alliin, however it can be speculated that the sulfoxide functional group on the alliin molecule allows for its relatively high stability compared to allicin, as the sulfoxide functional group is extremely stable (Figure 7) (ScienceDirect Topics, 2021). This may be why temperature did not have an effect on the allicin concentration obtained after altering the storage temperature of garlic cloves, and hence the concentration of alliin before crushing. Conversely, allicin is heat sensitive due to its thiosulfinate functional group located on the molecule, since the disulfide bond typically has a bond dissociation energy of 250 kJ/mol it can be easily broken (Figure 7). In addition, since the disulfide bond is 40% weaker than C – C and C – H bonds (Mansor et al., 2016), it is the most susceptible to cleavage at higher temperatures and is therefore the weakest link in the structure of allicin. This is why temperature has a substantial impact on allicin concentration but not alliin concentration. The alliinase enzyme is usually 80% stable over the pH range 6-8. Additionally, alliinase is 80% stable at temperatures below 40°C and has a sharp decrease in enzyme activity above this temperature. This indicates that alliinase is relatively stable at the temperatures that were tested in this paper (Chhabria and Desai, 2018). This further supports the experimental results as the temperatures investigated in this study are likely not extreme enough to affect alliin or alliinase concentration, hence explaining why no significant difference in concentration was measured for allicin extracted after storage of the garlic (Appendix 2.1).

Carbon sulfoxide functional group

Thiosulfinate functional group Disulfide bond

Figure 5: The structure of alliin compared to the structure of allicin

Conclusion This research project explored the effect of storage temperature before extraction on the concentration of allicin extracted from garlic, to investigate whether small changes in storage conditions would allow for the maximization of allicin concentration and hence the medicinal benefits pf garlic. The garlic cloves were stored at selected temperatures: -18°C, 3°C or 25°C before the garlic cloves were dehydrated to produce a garlic powder. Data was then collected by first reacting the powder with excess L-cysteine and then DTNB, which measured the decrease in L-cysteine concentration. The data analysis involved an ANOVA statistical test to compare if there was a statistically significant difference between the means of the four different groups. The results of my analysis showed that temperature did not have a statistically significant effect on the concentration of allicin in garlic, possibly due to the stability of both alliin and alliinase present in the garlic cloves before crushing.

Acknowledgements I would like to acknowledge and thank Dr Katie Terrett for her continuous supervisory support. I would also like to acknowledge Caitlin Tedesco who helped me develop potential research ideas and understand key concepts.

References Abe, K., Hori, Y. and Myoda, T. (2020) ‘Volatile compounds of fresh and processed garlic (Review)’, Experimental and Therapeutic Medicine, 19(2), pp. 1585–1593. Ankri, S. and Mirelman, D. (1999) ‘Antimicrobial properties of allicin from garlic’, Microbes and Infection, 1(2), pp. 125–129.

102 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Brian McNamara (2018) Introduction to UV-Vis Spectroscopy 03 Beer-Lambert Law. Available at: https://www.youtube.com/watch?v=VWHt4MBWBmc (Ac cessed: 5 June 2021). Cañizares, P. et al. (2004) ‘Thermal Degradation of Allicin in Garlic Extracts and Its Implication on the Inhibition of the inVitro Growth of Helicobacter pylori’, Biotechnology Progress, 20(1), pp. 32–37. Cheewinworasak, T. et al. (2018) ‘Effect of drying condition of Thai garlic (Allium sativum L.) on physicochemical and sensory properties’, International Food Research Journal, 25, pp. 1365– 1372. Chhabria, S. and Desai, K. (2018a) ‘Purification and characterisation of alliinase produced by Cupriavidus necator and its application for generation of cytotoxic agent: Allicin’, Saudi Journal of Biological Sciences, 25(7), pp. 1429– 1438. Chhabria, S. and Desai, K. (2018b) ‘Purification and characterisation of alliinase produced by Cupriavidus necator and its application for generation of cytotoxic agent: Allicin’, Saudi Journal of Biological Sciences, 25(7), pp. 1429– 1438. El-Saber Batiha, G. et al. (2020) ‘Chemical Constituents and Pharmacological Activities of Garlic (Allium sativum L.): A Review’, Nutrients, 12(3), p. 872. Fujisawa, H. et al. (2008) ‘Thermostability of Allicin Determined by Chemical and Biological Assays’, Bioscience, Biotechnology, and Biochemistry, 72(11), pp. 2877–2883. Iberl, B. et al. (1990) ‘Quantitative Determination of Allicin and Alliin from Garlic by HPLC*’, Planta Medica, 56(3), pp. 320– 326. Janská, P. et al. (2021) ‘Effect of physicochemical parameters on the stability and activity of garlic alliinase and its use for insitu allicin synthesis’, PLOS ONE, 16(3) Lawson, L. D. and Hughes, B. G. (1992) ‘Characterization of the Formation of Allicin and Other Thiosulfinates from Garlic’, Planta Medica, 58(4), pp. 345–350. Lawson, L. D. and Hunsaker, S. M. (2018) ‘Allicin Bioavailability and Bioequivalence from Garlic Supplements and Garlic Foods’, Nutrients, 10(7).

Compounds’, Iranian Journal of Basic Medical Sciences, 16(10), pp. 1031–1048. Oregon State University (1985) Garlic, Linus Pauling Institute. Available at: https://lpi.oregonstate.edu/mic/foodbeverages/garlic (Accessed: 5 June 2021). Rahman, M. S. (2007) ‘Allicin and Other Functional Active Components in Garlic: Health Benefits and Bioavailability’, International Journal of Food Properties, 10(2), pp. 245–268. S, Bose, B, L. and S, Banerjee (2014) ‘Quantification of allicin by high performance liquid chromatography-ultraviolet analysis with effect of post-ultrasonic sound and microwave radiation on fresh garlic cloves.’, Pharmacognosy Magazine, 10(Suppl 2), pp. S288-93. Saito, K. et al. (1989) ‘Determination of Allicin in Garlic and Commercial Garlic Products by Gas Chromatography with Flame Photometric Detection’, Journal of Association of Official Analytical Chemists, 72(6), pp. 917–920. Salehi, B. et al. (2019) ‘Allicin and health: A comprehensive review’, Trends in Food Science & Technology, 86, pp. 502– 516. Singh, A. (2014) ‘Effect of Drying Characteristics of Garlic-A Review’, Journal of Food Processing & Technology, 05(04). Sulfoxide - an overview | ScienceDirect Topics (no date). Available at: https://www.sciencedirect.com/topics/chemistry/sulfoxide ( Accessed: 8 June 2021). Tesfaye, A. (2015) ‘Traditional Uses, Phytochemistry and Pharmacological Properties of Garlic (Allium Sativum) and its Biological Active Compounds’, International Journal of Scientific Research in Science, Engineering and Technology, 1, pp. 142–148. UNC Eshelman School of Pharmacy (2021) The Pharmaceutics and Compounding Laboratory. Available at: https://pharmlabs.unc.edu/labs/spectrophotometry/beers.htm (Accessed: 5 June 2021). Wang, H. et al. (2015) ‘Influence of pH, concentration and light on stability of allicin in garlic (Allium sativum L.) aqueous extract as measured by UPLC’, Journal of the Science of Food and Agriculture, 95(9), pp. 1838–1844.

Leontiev, R. et al. (2018) ‘A Comparison of the Antibacterial and Antifungal Activities of Thiosulfinate Analogues of Allicin’, Scientific Reports, 8(1), p. 6763. Mansor, N. et al. (2016) ‘Quantification and Characterization of Allicin in Garlic Extract’, Journal of Medical and Bioengineering, 5(1), pp. 24–27. Mathialagan, R. (2016) Extraction of Allicin from Garlic Using Ultrasonic-Assisted Method. Mathialagan R. et al. (2017) ‘Optimization of ultrasonicassisted extraction (uae) of allicin from garlic (allium sativum l.)’, Chemical Engineering Transactions, 56, pp. 1747–1752. Mikaili, P. et al. (2013) ‘Therapeutic Uses and Pharmacological Properties of Garlic, Shallot, and Their Biologically Active

Science Extension Journal • 103

Scientific Research in School Volume 3 Issue 1 2021

Appendices Raw Data for Absorbance Readings Table 1: The effect of temperature on Absorbance

Temperature

Absorbance Absorbance Absorbance Absorbance Absorbance 1 (nm) 2 (nm) 3(nm) 4(nm) 5(nm)

Control (25°C)

0.044 (90.4%)

0.053 (88.5%)

0.044 (90.3%)

0.035 (92.5%)

0.039 (91.5%)

Room Temperature (25°C) 0.032 (92.9%)

0.036 (92.1%)

0.044 (90.3%)

0.044 (90.4%)

0.030 (93%)

Fridge (3°C)

0.022 (95%)

0.042 (90.75%)

0.029 (93.5%)

0.038 (91.7%)

0.035 (92.3%)

Freezer (-18°C)

0.034 (92.5%)

0.038 (91.6%)

0.035 (92.2%)

0.037 (91.9%)

2.1 Raw Data for concentration of Allicin based on Absorbance Table 2: Absorbance and concentration of Allicin for Control (25°C)

Absorbance (nm) 0.044 0.053 0.044 0.035 0.039

Concentration of Allicin (mol/L) 0.000229214 0.000228896 0.000229214 0.000229532 0.000229391

Table 3: Absorbance and concentration of Allicin for Room Temperature (25°C)

Absorbance (nm) 0.032 0.036 0.044 0.044 0.03

Concentration of Allicin (mol/L) 0.000229638 0.000229497 0.000229214 0.000229214 0.000229709

Table 4: Absorbance and concentration of Allicin for Fridge (3°C)

Absorbance (nm) 0.022 0.042 0.029 0.038 0.035

Concentration of Allicin (mol/L) 0.000229992 0.000229285 0.000229744 0.000229426 0.000229532

Table 5: Absorbance and concentration of Allicin for Freezer (-18°C)

Absorbance (nm) 0.034 0.038 0.038 0.035 0.037

104 • Science Extension Journal

Concentration of Allicin (mol/L) 0.000229568 0.000229426 0.000229426 0.000229532 0.000229462

Scientific Research in School Volume 3 Issue 1 2021

Influence of ultraviolet light on the stability of allicin in aqueous garlic extract James Wilson Barker College Garlic has been recognised and exploited as a medicinal panacea for thousands of years and is widely utilised as a flavouring agent for foods. Since its discovery in 1944, extensive research has been dedicated to investigating the nature and stability of allicin, the primary organosulfur accredited for garlic’s therapeutic properties. However, the extent to allicin’s photostability remains undefined. The present study examines allicin’s potential for photodecomposition by investigating whether exposure to artificial ultraviolet light of wavelength 366 nm in storage affects allicin’s rate of degradation in an aqueous garlic extract. Two aqueous garlic extract samples were stored over a 25 hour period; one exposed to UV of wavelength 366 nm, and the other to darkness as a control. Utilising an established indirect colourimetric assay, the allicin concentration of the two samples was monitored and content loss was analysed using an ANOVA test and PostHoc Tukey HSD. The discovery of significant differences in allicin content loss between the two samples demonstrated that allicin exposed to UV light degraded at a faster rate, thus providing evidence for its susceptibility to photodecomposition. Literature review Garlic (allium sativum) has occupied a prominent position amongst bulb vegetables for millennia due to not only its utility as a condiment, but its exploitable therapeutic potentials (Borlinghaus et al., 2014). Its health benefits have been attributed to the antioxidant activity of a variety of organosulfurous compounds, however in 1944 the sulfur-containing defence molecule allicin (diallyl thiosulfinate) (1) was isolated and accredited as the primary organosulfur of the thirty-three identified compounds responsible for crushed garlic’s pharmacological potentials (Cavallito and Bailey, 1944) (Table 1). Table 1: Types and concentrations of prominent thiosulfinates in garlic extract (After: Herng, 2014)

Thiosulfinate (TS) Allyl-2-propenTS (allicin) (1) AllylmethaneTS trans-1-propenyl-2-propeneTS methyl-2-propeneTS Allyl-trans-1-propeneTS methylmethaneTS trans-1-propenylmethaneTS methyl-trans-1-propeneTS

%mol 50 – 90 3 – 20 5 – 18 1.5 – 8 1.5 – 2 1–2 1–2 0.5

The organosulfur presence within fresh garlic is approximately four times greater (per gram of weight) than that of other food sources such as cruciferous vegetables (Prati et al., 2014), with 1 g of fresh garlic containing between 11 and 35 mg of organosulfurous compounds, of which, allicin constitutes approximately 70% (Mansor et al., 2016). Allicin is not present in raw

garlic, but is produced in an enzymatic reaction catalysed by the crushing of garlic cloves. Upon cell lysis, the odorless non-proteinogenic amino acid alliin, (S-allylcystein sulfoxide) (2) (Figure 1), originally compartmentalised in mesophyll cells, reacts with the enzyme alliinase that is released from the cell vacuoles, thus creating allicin (Prati et al., 2014) (Figure 1). Hence, in vivo and most dietary applications for allicin require the garlic to be in an aqueous extract form. Allicin has been discovered to possess a range of antimicrobial activities that are responsible for crushed garlic’s therapeutic potentials. These antimicrobial properties include antibacterial activity against Gramnegative and Gram-positive bacteria including strains of Escherichia Coli and salmonella, as well as a broad spectrum of bacterial isolates, many of which are resistant to antibiotics (Ankri and Mirelman, 1999). Allicin has also shown antiparasitic activity, including but not limited to in vitro inhibition of the major human intestinal protozoan parasite, Entamoeba histolytica (Mirelman, Monheit and Varon, 1987), which is responsible for 100,000 global deaths per year (Gunther et al., 2011). Additionally, this thiosulfinate demonstrates antifungal activity, particularly against low concentrations of species of Candida, Cryptococcus, Trichophyton, Epidermophyton and Microsporum in vitro, and antiviral activity, with influenza B and herpes simplex viruses showing sensitivity to crushed garlic (Tsai et al., 1985).

Science Extension Journal • 105

Scientific Research in School Volume 3 Issue 1 2021

Figure 1: Chemical formation of allicin (1). Alliin (2) hydrolyses with alliinase to form allylsufenic acid, which spontaneously condenses to produce allicin (Source: Wallock-Richard et al., 2014

Allicin stability Unfortunately, these therapeutic properties are limited due to the unstable nature of allicin. The organosulfur is known to readily decompose into a variety of organosulfur compounds including diallyl trisulfide, diallyl disulfide and diallyl sulfide (Levy, 2019) (Figure 2) in the presence of air and water, and more rapidly in adverse temperatures or pH solutions, accounting for the characteristic odour of garlic present when cooking or after ingestion (Prati et al., 2014).

or boiling will cause almost all the allicin to decompose and hence diminish the garlic’s total therapeutic potential (Jiménez-Monreal et al., 2009). In regard to allicin’s pH tolerance, both Mansor et al., (2016) and Wang et al., (2014) concluded that the optimal environmental pH for allicin is 5-6, with it being very unstable at pH levels lower than 1.5 and greater than 11, due to the disulfide bond’s susceptibility to break under acidic conditions (Han et al., 1995). Considering allicin isn’t formed in whole garlic, heat-dried garlic can still produce allicin once crushed. Light exposure Whilst allicin’s sensitivity to temperature and pH has been heavily scrutinised, literature on its tolerance to light is limited. Only Wang et al., (2014) has investigated allicin’s photosensitivity to visible light in storage and concluded no difference to storage in darkness (Figure 3). Moreover, no ultraviolet (UV) light experimentation on allicin has been performed, and therefore allicin’s stability under the influence of UV radiation is unknown.

Figure 2: Chemical structures of diallyl sulfide, diallyl disulfide and diallyl trisulfide (Source: Bansal et al., 2018)

Thus, purified allicin isn’t commercially available (Levy, 2019), but is instead a constituent in powdered garlic supplements. Hence, expanding the literature upon the stability of allicin is of great interest to scientific researchers in the pursuit of advancing allicin’s medical implementation and versatility. Previous studies have identified that allicin in an aqueous garlic solution is most stable at storage temperatures between -20 ⁰C to 30 ⁰C (Wang et al., 2014). However, it is very susceptible to decomposition at higher temperatures due the disulfide bond’s dissociation energy of 250 kJ/mol being overcome easily by increased heat energy in the system (Mansor et al., 2016), and thus, cooking with garlic involving frying 106 • Science Extension Journal

Figure 3: Allicin (1) decomposition in an aqueous garlic extract under light and dark conditions. Light is of a moderate intensity of 903𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇𝜇−2 𝑠𝑠𝑠𝑠 −1 (Source: Wang et al., 2014)

Scientific Research in School Volume 3 Issue 1 2021

Ultraviolet light UV radiation appears beyond visible light in the electromagnetic spectrum, and exhibits shorter wavelengths and higher frequencies. Additionally, since the energy of a photon is inversely proportional to the wavelength, UV waves have higher energy than visible light (Figure 4). UV light is categorised into three spectral bands: UVA (315 – 400 nm), UVB (280 – 315 nm) and UVC (200 – 280 nm) (Paul et al., 2011). Any intensity of electromagnetic radiation including UV can catalyse the degradation of organic matter through a process known as photolysis or photodecomposition (Paul et al., 2011) which is a reaction whereby chemical bonds are broken due to the transfer of photon energy. A compound’s susceptibility and rate of photolysis is influenced by factors including the chemical’s reactivity, light absorption properties and bond energy, as well as the intensity of the applied radiation (Speight, 2018).

Figure 4: The electromagnetic spectrum (Source: NASA, 2013)

Quantification methods For the quantification of thiosulfinates such as allicin, conventional methods include the use of High Performance Liquid Chromatography (HPLC) and Gas Chromatography (GC) to directly determine concentration. However, both procedures require the use of external standards, and cannot be performed in a high-school laboratory due to the need for specialised equipment. Instead, an indirect spectrophotometric assay developed by Han et al., (1995) and reused by Mansor et al., (2016) can be utilised in the absence of this equipment. The method quantifies the concentration of allicin through a reaction with L-cysteine (3) with the proven basis that one mole of allicin, as well as the other thiosulfinates, reacts with two moles of L-cysteine (3) to produce two moles of S-allyl mercaptocysteine (Han et al., 1995) (Figure 5). Unreacted L-cysteine (3) is reacted with 5,5’-dithiobis-(2-nitrobenzoic acid) (DTNB) (4) to form a yellow-coloured compound, 2-nitro-5thiobenzoate (NTB) (5) in a 1:1 stoichiometric ratio. NTB (5) has a molar absorptivity (𝜀𝜀𝜀𝜀) of 14150 𝑀𝑀𝑀𝑀 −1 𝑐𝑐𝑐𝑐𝜇𝜇𝜇𝜇−1 and an optimal absorbance (𝐴𝐴𝐴𝐴) at 412 nm in a colourimeter. Beer-Lambert’s law (Equation 1) can then be used to relate the NTB (5) absorbance readings to its concentration (𝑐𝑐𝑐𝑐) and that of the unreacted L-cysteine (3), and the amount of allicin reacted is backcalculated using stoichiometry.

Figure 5: Chemical diagram for the quantification of allicin using colourimetry or spectrophotometry

𝐴𝐴𝐴𝐴 = 𝜀𝜀𝜀𝜀𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐

Equation 1: Beer-Lambert’s law

Hence, this report has chosen to investigate the influence of ultraviolet light due to the absence of reports in this area. UVA light was specifically chosen due to its prevalence in sunlight, and it being only slightly more intense than visible light, thus allowing for an extension of the literature upon the threshold of allicin’s photostability. A wavelength of 366 nm was selected due to its availability in a school laboratory. The hypothesis formed stems from the fact that UV waves emit more energy than visible light waves, and thus the potential for the organosulfur compounds to undergo photolysis is increased.

Scientific research question How does exposure to artificial ultraviolet light of wavelength 366 nm in storage affect the degradation rate of allicin in aqueous garlic extract?

Scientific hypothesis That aqueous garlic extract exposed to ultraviolet light of wavelength 366 nm in storage will exhibit a greater rate of allicin concentration loss than extract not exposed to ultraviolet light.

Methodology General experimental details For the determination of thiosulfinate concentration at each time interval of this procedure, the spectrophotometric indirect method for quantifying garlic thiosulfinates used by Han et al., (1995) was adopted and modified for this experiment. Prepared garlic powder can be stored at room temperature without loss of allicin content for at least one year (Han et al., 1995). Solutions and materials were also prepared in accordance to procedures outlined in Herng, (2014), and all chemicals were purchased from Sigma-Aldrich.

Science Extension Journal • 107

Scientific Research in School Volume 3 Issue 1 2021

Preparation of solutions and materials Preparation of 5.0 mM DTNB solution DTNB (4) powder (0.3964 g) was weighed using an electronic balance. The powder was introduced into a 200 mL conical flask and dimethyl sulfoxide (DMSO) (200 mL) was added. The solution was magnetically stirred until the powder visibly dissolved. Preparation of 50 mM HEPES buffer at 7.6 pH A pH-meter was calibrated by placing in a pH 7.0 buffer solution. HEPES buffer powder (23.8 g) was weighed using an electronic balance, and then added to a 250 mL beaker. 70 mL of distilled water was introduced to the beaker and the resulting mixture was magnetically stirred for 10 minutes. Sodium hydroxide (10 M solution) was slowly added dropwise using a glass pipette whilst the pH of the solution was being measured using the pH meter, until a pH of 7.6 was measured. The solution was transferred to a 100 mL volumetric flask and distilled water was added to reach the 100 mL mark, resulting in a 1.0 M solution. To dilute to 50 mM, 5.0 mL of 1.0 M HEPES buffer was transferred into a separate 100 mL volumetric flask. Distilled water was added until the solution reached the graduation mark. The solution was hand-shaken for 30 seconds for homogenisation. Preparation of 2.0 mM L-cysteine L-cysteine (3) crystals (0.09696g) were weighed using an electronic balance. The crystals were introduced into a 500 mL conical flask and 200 mL of distilled water was added to the flask. The solution was magnetically stirred until the L-cysteine (3) visibly dissolved, and then another 200 mL of distilled water was added. Preparation of garlic powder Cloves from fresh garlic bulbs commercially purchased from the same batch were peeled and sliced into approximately 3 mm thick slices. The slices were dried in an oven at 60 °C for 12 hours, and then grinded into a fine powder with a mortar and pestle. The powder was stored in an air-tight container in the dark. Experimental procedure Preparation of garlic extract To create the stock garlic extract solution, 5 g of garlic powder was mixed with 150 mL of distilled water in a 250 mL conical flask. The mixture was homogenised using a magnetic stirrer for 1 hour at room temperature. A vacuum pump was used to filter larger insoluble particles from the mixture, and the resulting filtrate was left to sit in the dark for 18 hours to allow the fine suspension to settle.

108 • Science Extension Journal

Storage of garlic extracts 20 mL of the garlic extract was placed in two separate petri dishes. One dish was placed in a light-proof 366 nm UV cabinet and the other dish was placed in a separate light-proof cabinet in the dark as a control. Measurements were taken immediately before storage, and after storage times of 0.75 hours, 8.75 hours, and 25 hours under 366 nm. In addition, thiosulfinate content was also quantified after 25 hours in the dark. For each measurement, five 0.5 mL samples were extracted from each dish and the allicin concentration for each was quantified using the colourimetric assay outlined below, with the room lights turned off to minimise the effect of exposure to visible light. Analysis of thiosulfinate content by colourimetry 0.5 mL of garlic extract was transferred into five separate 10 mL beakers and 1.2 mL of 2 mM L-cysteine (3) was added to each beaker. The beakers were swirled and left to sit at room temperature for 10 minutes. 3 mL of 50 mM HEPES buffer (pH 7.6) and 1 mL of 5 mM DTNB (4) were added to the beakers, and the solutions were magnetically stirred for 30 seconds, before sitting at room temperature for a further 2 minutes. Each sample was transferred into separate cuvettes and absorbance readings for each were taken at 400 nm using a PASCO colourimeter, which was limited to readings at increments of 50 nm.

Results Upon creation of NTB (5) and determination of its concentration through Beer-Lambert’s law (Equation 1), allicin concentration could be back-calculated with the formula in Equation 2.

𝑐𝑐𝑐𝑐𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 =

𝑐𝑐𝑐𝑐𝐿𝐿𝐿𝐿−𝑎𝑎𝑎𝑎𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑐𝑐𝑐𝑐 (𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎) − 𝑐𝑐𝑐𝑐𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 2

Equation 2: Stoichiometrically derived formula for the concentration of allicin

Time = 0 results from both samples are the same due to them being measured from a stock solution. Due to proportionality between absorbance and concentration in Beer-Lambert’s law (Equation 1), greater absorbance readings corresponded to higher concentrations of NTB (5) (Appendix, Table A1, A2, A3 and A4). Since NTB (5) forms from reaction of L-cysteine (3) and DTNB (4) with a 1:1 molar ratio, the calculated NTB (5) thus conveys the excess L-cysteine (3) from the reaction with allicin (Appendix, Table A3 and A4). Hence, greater absorbance readings corresponded to greater residual Lcysteine (3) and thus less allicin present in the initial reaction. Considering that crushed garlic thiosulfinates and L-cysteine (3) react in a 1:2 stoichiometric ratio, the allicin content in each extract was therefore back

Scientific Research in School Volume 3 Issue 1 2021

calculated, and is shown in Tables 2 and 3. Figure 6 graphically captures the degradation of allicin content as a function of storage time for each sample. Similar absorbance values in Tables A1 and A2 for time A and B as well as graphical results indicate no conclusive allicin photodecomposition within the first 8.75 hours. However, the results show a greater decrease in allicin concentration over the 25 hour period in the sample exposed to UV.

Statistical test results A one-way analysis of variance (ANOVA) statistical test of the calculated thiosulfinate concentration indicated significant differences between the means for various time points with a p value < 0.00001, which was less than 𝛼𝛼𝛼𝛼 = 0.05 (Table 4). The pairs of storage time points that demonstrate significant mean concentration differences were identified using a Post-Hoc Tukey test and the differences are outlined in Table 5.

Table 2: Average results values for UV sample

Storage time (hours)

Absorbance

A B C D

0.0235 0.0274 0.0245 0.0398

0 0.75 8.75 25

Excess Lcysteine (3) (𝝁𝝁𝝁𝝁𝝁𝝁𝝁𝝁) 1.66077 1.93639 1.73144 2.81272

Thiosulfinate concentration (𝒎𝒎𝒎𝒎𝝁𝝁𝝁𝝁)

Estimated allicin concentration (𝒎𝒎𝒎𝒎𝝁𝝁𝝁𝝁)

Thiosulfinate concentration (𝒎𝒎𝒎𝒎𝝁𝝁𝝁𝝁)

Estimated allicin concentration (𝒎𝒎𝒎𝒎𝝁𝝁𝝁𝝁)

0.20969593 0.20955812 0.20966059 0.20911995

0.146787151 0.146690684 0.146762413 0.146383965

Table 3: Average results values for controlled sample

Storage time (hours)

Absorbance

A E

0.0235 0.0333

0 25

Excess Lcysteine (3) (𝝁𝝁𝝁𝝁𝝁𝝁𝝁𝝁) 1.66077 2.35335

0.20969593 0.20934964

0.146787151 0.146544748

Allicin content over time Allicin Content (mM)

0.2098 0.2097 0.2096 0.2095 0.2094 0.2093 0.2092 0.2091 0.209

Hours UV

Controlled

Figure 6: Graph of the allicin content in the UV and controlled sample throughout storage Table 4: ANOVA test output for comparing the mean thiosulfinate concentration for each storage time. H0 = there is no difference in the means; HA = the means are not all equal

Mean Standard deviation f-statistic Alpha value P value Analysis

A B C 0.20969593 0.20955812 0.20966059 0.0001 0.0001 0.0001 36.6308 0.05 < 0.00001 p < 0.05, significant difference in means

D 0.20911995 0.0001

E 0.20934964 0.0001

Science Extension Journal • 109

Scientific Research in School Volume 3 Issue 1 2021

Table 5: Post-Hoc Tukey Test p values for comparing the thiosulfinate concentration for each storage time.

concentration in garlic extract would require the HPLC method. However, since it is known that allicin constitutes 60-80% of total garlic thiosulfinates (Han et al., 1995; Mansor et al., 2016) (Table 1), and that the method used for this experiment quantifies total present garlic thiosulfinates, an estimation can be made for allicin concentration by multiplying the value for total thiosulfinates by a factor of 0.7 (Table 2 and 3). Moreover, since this report aims to determine the relative rate of allicin degradation, obtaining a specific value for allicin content is not a requirement for a valid conclusion.

Discussion

Due to the nature of a high-school laboratory, additional adjustments to the broader method attained from Mansor et al., (2016) were required which increased uncertainty. Instead of homogenising the garlic solution with an incubated shaker and then filtering off insoluble solids with a centrifuge in preparation of the aqueous extract, homogenisation was achieved less effectively with a magnetic stirrer, and a vacuum pump was used for filtration. Large fluctuations in absorbance values during preliminary testing due to slight cloudiness in the mixture substantiated this uncertainty, however it was found that a clear mixture could still be generated if the solution was left for approximately 18 hours after the filtering to allow the suspension to settle. Despite this, minor fluctuations in absorbance readings still occurred, which may have accounted for the apparent initial decrease then increase in the UV sample’s allicin content over the first two measuring points (Figure 6).

Pair A:B A:C A:D A:E B:C B:D B:E C:D C:E D:E

Q-stat 2.52 0.00 15.12 7.56 2.52 12.60 5.04 15.12 7.56 7.56

p-value 0.41073 0.000001 0.00000 0.00027 0.41073 0.00000 0.1486 0.00000 0.00027 0.00027

Analysis Insignificant Significant Significant Significant Insignificant Significant Insignificant Significant Significant Significant

The significant pairwise comparisons (Table 5) not only demonstrate allicin degradation across both samples during storage, but verify a greater loss of allicin content in the UV sample as shown with the D:E comparison, therefore implying the occurrence of UV direct photolysis. Hence, the alternate hypothesis (HA ) is accepted as it is shown that allicin decomposed at a faster rate when exposed to UV light in storage. To ensure validity, multiple preliminary tests were conducted to identify key limitations in the method appropriated from Mansor et al., (2016). Upon initial testing of the colourimetric quantification process, it was discovered that the reaction times of at least one of these two reactions significantly affected the concentration of the NTB (5) formed. After having allowed the L-cysteine (3) and allicin to react for 10 minutes and then allowing the DTNB (4) and excess Lcysteine (3) to react for multiple hours instead of two minutes, the solutions produced an absorbance reading approximately 3.5 times greater than that of the previous trial when mixed for the prescribed time. This suggested that the NTB (5) concentration, and thus the final allicin concentration calculation was dependent on the time the system was allowed to react, implying that this method may not be as suitable for calculating a definite value for allicin concentration in an aqueous extract at particular time points. However, as long as the reaction times during all trials were kept consistent, the relative rate of allicin decomposition could still be determined, thus satisfying the scientific research question. Hence, modifications were made to the final procedure to ensure consistency in reaction times, with greater detail applied to the method of reaction, such as substituting hand-shaking for magnetic stirring, to ensure a valid experiment. However, in the absence of HPLC, greater attention should be applied to eliminating ambiguity in the reaction method and timing for greater precision in future research. Another limitation of this method is its inability to determine exact allicin concentration, but rather the total thiosulfinate concentration. Assay of exact allicin 110 • Science Extension Journal

Moderately low standard deviation between the absorbance value measurement trials (Appendix, Table A1 and A2) indicates high precision through repetition, therefore reflecting a high reliability of the experiment. However, there did exist some uncertainty due to the limitation of the colourimeter only being able to read up to increments of 50 nm, therefore the absorbance readings weren’t taken at the optimal wavelength (412 nm) for NTB (5). In the absence of significant uncertainties, it can be reasoned that the allicin in the UV sample dissociated faster due to the presence of UV radiation through photodecomposition. Whilst a molecule’s light absorption properties determine its susceptibility to photochemical processes as quantified under its molar absorptivity coefficient (𝜀𝜀𝜀𝜀), organic compounds will decompose when subjected to a high enough intensity of applied radiation (Speight, 2018). The photolytic process is initiated through the molecule’s absorption of a quantum of light energy from a photon, causing it transform into a transient state. In this exited state, molecules are subject to undergo a number of photochemical processes including photodissociation, intramolecular rearrangement (photoisomerisation),

Scientific Research in School Volume 3 Issue 1 2021

reaction with other molecules, or deexcitation through emittance of luminescent radiation. Photodissociation occurs when the exited molecule deexcites, and results in the molecule dissociating in a chain reaction (Speight, 2018). Compounds more susceptible to photolysis are characterised by higher molar absorptivity coefficients (𝜀𝜀𝜀𝜀) and have a higher threshold wavelength for photodecomposition. However, allicin’s light absorptivity properties are unknown. Hence, the acceptance of the alternate hypothesis indicates that 366 nm surpasses the threshold for photodecomposition in allicin, and that this threshold may lie anywhere between the UVA and low visible light range, depending on the visible light’s luminous intensity. Future research Whilst the results captured significant differences between the degradation rate of allicin in aqueous garlic extract when stored under UV than in darkness, this conclusion was founded upon only 4 measurement points due to equipment availability and time constraints. Hence, a greater storage period with increased measurement points for both samples similarly to that of Wang et al., (2014) (Figure 3) would facilitate a more comprehensive conclusion and understanding of allicin’s photodecomposition under 366 nm UV. Furthermore, this experiment investigated exposure to waves of a minor increase in intensity from visible light, UVA, in an attempt to discern the upper threshold of allicin’s photostability. Further experimentation and analysis with higher frequency electromagnetic waves should be considered as to map allicin’s rate of degradation in more extreme conditions and to complement our understanding of its thermostability. Moreover, additional testing with wavelengths between 366 nm and the lower range of the visible light spectrum is recommended for identifying a more exact threshold frequency for allicin photostability.

Conclusion This experiment investigated whether allicin decomposed faster when exposed to artificial ultraviolet light of wavelength 366 nm, hypothesising a greater rate of allicin content loss in UV storage conditions than in darkness. Through an indirect colourimetric assay, the allicin content of the sample exposed to ultraviolet light was monitored, and compared to the control at the end of the time period. The data analysis involved an ANOVA and Post-Hoc Tukey test of the allicin concentrations at each time point of measurement. Significant differences between the final data points demonstrated a difference in allicin degradation rates, leading to an acceptance of the hypothesis that allicin exposed to UVA in storage decomposes at a faster rate compared to the control experiment. Future

experimentation should look to investigating a greater range of frequencies and period of storage.

Acknowledgements I would like to thank Dr Katie Terrett for her tremendous assistance and invaluable support with all aspects of the project. This report would not have been possible without her guidance in developing an area of research, and aid in appropriating a method for data collection. I wish to extend my appreciation to the Barker College Science Department for providing the necessary chemicals and allowing me flexible access to the laboratory to complete the time-dependent data collection process. Finally, I would like to thank my colleague, Mr Charles Scholefield, for his support and enthusiasm throughout the duration of the work.

References Ankri, S. and Mirelman, D. (1999). Antimicrobial properties of allicin from garlic. Microbes and Infection, [online] 1(2), Available at: pp.125–129. http://www.bashaar.org.il/files/6130.pdf [Accessed 16 Oct. 2020]. Bansal, M., Singh, N., Pal, S., Dev, I. and Ansari, K.M. (2018). Chemopreventive Role of Dietary Phytochemicals in Colorectal Cancer. Advances in Molecular Toxicology, 12, pp.69–121. Borlinghaus, J., Albrecht, F., Gruhlke, M., Nwachukwu, I. and Slusarenko, A. (2014). Allicin: Chemistry and Biological Properties. Molecules, [online] 19(8), pp.12591–12618. Available at: https://www.mdpi.com/14203049/19/8/12591/htm [Accessed 17 Dec. 2020]. Cavallito, C.J. and Bailey, J.H. (1944). Allicin, the Antibacterial Principle of Allium sativum. I. Isolation, Physical Properties and Antibacterial Action. Journal of the American Chemical Society, 66(11), pp.1950–1951. Gunther, J., Shafir, S., Bristow, B. and Sorvillo, F. (2011). Amebiasis-Related Mortality among United States Residents, 1990–2007. The American Journal of Tropical Medicine and Hygiene, [online] 85(6), pp.1038–1040. Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3225148/#:~ :text=Amebiasis%2C%20caused%20by%20the%20intestinal [Accessed 11 Oct. 2020]. Han, J., Lawson, L., Han, G. and Han, P. (1995). Spectrophotometric Method for Quantitative Determination of Allicin and Total Garlic Thiosulfinates. Analytical Biochemistry, 225(1), pp.157–160. Herng, H.J. (2014). Quantification and Characterization of Allicin in Garlic Extract. [Dissertation] core.ac.uk. Available at: https://core.ac.uk/reader/301117439. Jiménez-Monreal, A.M., García-Diz, L., Martínez-Tomé, M., Mariscal, M. and Murcia, M.A. (2009). Influence of cooking methods on antioxidant activity of vegetables. Journal of food science, [online] 74(3), pp.97–103. Available at: https://www.ncbi.nlm.nih.gov/pubmed/19397724. Science Extension Journal • 111

Scientific Research in School Volume 3 Issue 1 2021

Levy, J. (2019). The Healthiest Part of Garlic? [online] Dr. Axe. Available at: https://draxe.com/nutrition/allicin/#Best_Way_to_Obtain_It [Accessed 17 Apr. 2021]. Mansor, N., Herng, H.J., Samsudin, S.J., Sufian, S. and Uemura, Y. (2016). Quantification and Characterization of Allicin in Garlic Extract. Journal of Medical and Bioengineering, 5(1), pp.24–27. Mirelman, D., Monheit, D. and Varon, S. (1987). Inhibition of Growth of Entamoeba histolytica by Allicin, the Active Principle of Garlic Extract (Allium sativum). Journal of Infectious Diseases, 156(1), pp.243–244. NASA (2013). Electromagnetic Spectrum - Introduction. [online] Nasa.gov. Available at: https://imagine.gsfc.nasa.gov/science/toolbox/emspectrum1.h tml. Paul, A., Dziallas, C., Zwirnmann, E., Gjessing, E.T. and Grossart, H.-P. (2011). UV irradiation of natural organic matter (NOM): impact on organic carbon and bacteria. Aquatic Sciences, 74(3), pp.443–454. Prati, P., Henrique, C.M., Souza, A.S. de, Silva, V.S.N. da and Pacheco, M.T.B. (2014). Evaluation of allicin stability in processed garlic of different cultivars. Food Science and Technology (Campinas), 34(3), pp.623–628. Speight, J.G. (2018). Reaction mechanisms in environmental engineering : analysis and prediction. Kidlington, Oxford, United Kingdom ; Cambridge, Ma, United States: Butterworth-Heinemann, An Imprint Of Elsevier, pp.231–267. Tsai, Y., Cole, L., Davis, L., Lockwood, S., Simmons, V. and Wild, G. (1985). Antiviral Properties of Garlic:In vitroEffects on Influenza B, Herpes Simplex and Coxsackie Viruses. Planta Medica, 51(05), pp.460–461.

112 • Science Extension Journal

Wallock-Richards, D., Doherty, C.J., Doherty, L., Clarke, D.J., Place, M., Govan, J.R.W. and Campopiano, D.J. (2014). Garlic Revisited: Antimicrobial Activity of Allicin-Containing Garlic Extracts against Burkholderia cepacia Complex. PLoS ONE, 9(12), p.e112726. Wang, H., Li, X., Liu, X., Shen, D., Qiu, Y., Zhang, X. and Song, J. (2014). Influence of pH, concentration and light on stability of allicin in garlic (Allium sativumL.) aqueous extract as measured by UPLC. Journal of the Science of Food and Agriculture, 95(9), pp.1838–1844.

Scientific Research in School Volume 3 Issue 1 2021

Appendix Raw data Tables A1 and A2 convey the recorded absorbance values for each sample throughout storage, whilst the concentration of the yellow compound NTB (5) is shown in Tables A3 and A4. Tables A3 and A4 thus correspond to the excess Lcysteine (3) from the reaction due to the 1:1 stoichiometric ratio between L-cysteine (3) and NTB (5). Table A1: Absorbance readings for UV sample throughout storage

Absorbance Storage time (hours) 0 0.75 8.75 25

Mean

0.0201 0.0307 0.0250 0.0409

0.0232 0.0264 0.0239 0.0385

0.0203 0.0235 0.0218 0.0393

0.0260 0.0296 0.0266 0.0380

0.0281 0.0266 0.0252 0.0422

0.0235 0.0274 0.0245 0.0398

Standard deviation 0.0035 0.0029 0.0018 0.0017

Table A2: Absorbance readings for controlled sample at time = 0 and time = 25

Absorbance Storage time (hours) 0 25

Mean

0.0201 0.0358

0.0232 0.0343

0.0203 0.0328

0.0260 0.0336

0.0281 0.0299

0.0235 0.0333

Standard deviation 0.0035 0.0022

Table A3: Concentration of NTB (5) and excess L-cysteine (3) for UV sample calculated by Beer-Lambert’s law

Concentration (𝝁𝝁𝝁𝝁𝝁𝝁𝝁𝝁) Storage time #1 (hours) 0 1.42049 0.75 2.16961 8.75 1.76678 25 2.89045

Mean

1.63957 1.86572 1.68904 2.72084

1.43462 1.66077 1.54063 2.77738

1.83745 2.09187 1.87985 2.68551

1.98586 1.87985 1.78091 2.98233

1.66077 1.93639 1.73144 2.81272

Standard Deviation 0.202564 0.164666 0.103221 0.100560

Table A4: Concentration of NTB (5) and thus excess L-cysteine (3) for controlled sample calculated by Beer-Lambert’s law

Concentration (𝝁𝝁𝝁𝝁𝝁𝝁𝝁𝝁) Storage time #1 (hours) 0 1.42049 25 2.53003

Mean

1.63957 2.42402

1.43462 2.31802

1.83745 2.37455

1.98586 2.11307

1.66077 2.35335

Standard Deviation 0.202564 0.126250

Science Extension Journal • 113

Scientific Research in School Volume 3 Issue 1 2021

114 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Concentration of Lycopene in Different Varieties of Tomatoes Charlie Scholefield Barker College Tomatoes are an important source of antioxidants, involved in reducing the risk of many chronic diseases such as cancer. This is because they contain a large number of carotenoids. Lycopene is one of these, an organic compound that gives red fruits and vegetables their bright pigment. This report will outline the effect of different tomato varieties on the concentration of lycopene extracted. Data was collected on three different varieties: Truss, Roma, and Cherry tomatoes, and analysed with an ANOVA and a PostHoc Tukey test to determine whether or not there was a significant difference in the means of the three groups. The results supported the hypothesis, suggesting that there is a significant difference in the concentration of lycopene from different varieties of tomatoes. Literature review Lycopene (C40H56) is a naturally occurring organic compound found in many red fruits and vegetables such as tomatoes, watermelons, pink grapefruits, red carrots, and papayas (May, 2020). It is the molecule that gives these compounds their bright red pigment. It is a longchain hydrocarbon molecule with alternating single and double carbon-carbon bonds, as seen in figure 1, referred to as a conjugated structure (de Montemas, 2020). This makes it part of a group of molecules called carotenoids, which give colour to red, yellow, or orange plant parts. β-carotene, also found in tomatoes, is another member of this group.

Figure 1: Chemical structure of lycopene.

Free Radicals Free radicals are molecules capable of independent existence that contain at least one unpaired electron (Agarwal et al., 2014), which they can either donate or accept. This makes them both oxidants and reductants, causing them to be highly reactive (Lobo et al., 2010). They can be produced through normal metabolic processes, or exposure to external sources such as Xrays, ozone, cigarette smoking, air pollutants, and industrial chemicals (Lobo et al., 2010). When functioning normally, free radicals are useful in helping fight off pathogens, protecting the body from infections. However, when there is an imbalance between free radicals and antioxidants, excess free radicals can cause oxidative stress. This means they begin to react with

fatty tissue, proteins, and DNA (Agarwal and Rao, 2000) causing chain reactions that slowly damage cells. This can lead to harmful, chronic diseases such as atherosclerosis, asthma, diabetes, and cancer. Antioxidants Antioxidants are extremely beneficial in the way they prevent excess oxidation of free radicals from occurring. They effectively balance out the oxidative stress caused, preventing major problems from occurring in the body. This makes it important to maximise antioxidant consumption. Carotenoids are known for their antioxidant properties in humans. This means they can prevent slow damage to cells that occur as a result of excess free radicals. This is important in preventing harmful chronic diseases, such as cancer. Lycopene is one of the most potent antioxidants, often referred to as a free radical scavenger (Fish, Perkins-Veazie and Collins, 2002). Because of its structure, lycopene (and other antioxidants) can easily react with free radicals, essentially ‘cleaning up’ the number of them in the body when there is an excess. This helps prevent oxidative stress damage caused to cells, therefore reducing the risk of diseases like cancer. In particular, cancer is one of the leading causes of death in the western world. Lifestyle and diet are considered as major risk factors, with about 50% of cancers related to diet (Agarwal and Rao, 2000), resulting in 35% of cancer mortalities (Williams, 1999). Producers will therefore often look to increase lycopene concentration in their tomatoes, because of the antioxidant property the compound holds. Tomatoes with higher levels of lycopene have greater health benefits, appealing to consumers as a better choice. Tomatoes are one of the most widely grown vegetables on Earth, second to potatoes (Baliyan and Rao, 2013), Science Extension Journal • 115

Scientific Research in School Volume 3 Issue 1 2021

meaning it is important to maximise the health benefits from their consumption. On top of this, studies have found that fresh tomatoes account for 50% of a person’s total lycopene intake (Rao, Waseem and Agarwal, 1998). Other Reports Many other reports have investigated ways to maximise the concentration of lycopene in tomatoes. Common variables include the storage temperature and the use of processed tomato products. It has been found that processed tomato products, such as tomato paste, contain higher amounts of lycopene than fresh tomatoes (Gärtner, Stahl and Sies, 1997). It was found that, on a dry weight basis, cherry tomatoes contained 124.0 mg of lycopene per 100 g, while tomato paste contained 204.6 mg of lycopene per 100 g (Toma et al., 2008). This suggests processing tomatoes leads to an increase in lycopene levels, possibly due to increased concentration from water loss (Story et al., 2010) and/or the use of heat and oils in cooking. There have also been many studies into the proposed health benefits of lycopene, such that it improves male fertility. Trials have been conducted reporting improvements in sperm parameters and pregnancy rates with the daily supplementation of lycopene for 3-12 months (Agarwal et al., 2014). Furthermore, lycopene levels in the body have been found to be inversely related to the incidence of cancers, including breast cancer and prostate cancer (Agarwal and Rao, 2000). Its proposed prevention of prostate cancer is extremely important, since it is the most common cancer and the fifth driving reason for death in men (Soares et al., 2019).

Beer Lambert law (Figure 2) states that the light absorbed is proportional to the concentration of the compound absorbing the light, making it possible to calculate the concentration of lycopene in solution.

Figure 2: The Beer Lambert Law

Problems with this method involve the inability to accurately calculate the concentration of lycopene, because of interference of other carotenoids, such as βcarotene, in the colourimetry process. The light absorption spectrum for β-carotene is shifted slightly more to the blue end than lycopene, because of small differences in molecular shape (May, 2020). The closeness of each molecule’s respective spectra makes it important to choose a wavelength of light that not only absorbs the most lycopene, but the one with the least interference. This can be seen through the graph in Figure 3.

Lycopene content is predicted to be affected by variety of tomatoes since it has been found in studies that there is a relationship between the amount of lycopene and redness of tomatoes (Toma et al., 2008). Extraction of Lycopene A majority of methods used for lycopene extraction are not possible in a school lab, as they are time consuming, expensive, and use hazardous organic solvents (Davis, Fish and Perkins-Veazie, 2003). An example of this is high performance liquid chromatography, HPLC. This is a technique used that separates and identifies each component in a mixture, making it easier to single out and measure the amount of lycopene. The methodology in this report involves the use of colourimetry to analyse a sample of tomato puree dissolved in a solvent mixture. Colourimetry works by emitting a specific wavelength of light through a solution. The compound absorbs some light and the light transmitted is measured and related to absorbance. The 116 • Science Extension Journal

Figure 3: Light absorption spectra for lycopene and βcarotene (Source: de Sousa, 2014)

Lycopene has three absorbance peaks, at 444, 471, and 503 nm. The peak at 503 nm is the most effective since the absorbance of β-carotene (and other carotenoids) at this wavelength is relatively low, leading to minimal interference. A wavelength of 500 nm was chosen for this experiment, as this is the closest setting to match the peak shown. What will this report involve? This report will investigate the change in lycopene concentration across different varieties of tomatoes. This will be carried out with a more simple, accessible method than previous reports have used, allowing also for the testing of this method’s reliability for future use.

Scientific Research in School Volume 3 Issue 1 2021

A large amount of literature is focused on the effect of processing on lycopene concentration and the health benefits of lycopene. It is therefore important to join these together in research, by studying the most efficient ways to boost lycopene concentration for human consumption. One of these is the growing of different varieties of tomatoes, something that is simple for producers to implement. This report therefore aims to find any difference in concentration of lycopene from different tomato varieties. Education on the importance of tomato variety in human consumption and agricultural production can improve the amount of lycopene in the western diet, reducing the risk of diseases such as cancer.

Scientific research question How does the specific variety of tomato affect the concentration of lycopene extracted?

Scientific hypothesis That the variety of tomato affects the concentration of lycopene extracted.

Methodology Preparation of Chemicals A solvent mixture of 2:1:1 Hexane: BHT: acetone was used to extract the lycopene, and it was made as follows: Butylated hydroxytoluene (BHT) (0.10 g, 0.0454 mol) was dissolved in 200 mL of absolute ethanol. The resulting solution was protected from light by covering the bottle with aluminium foil. BHT in ethanol (100 mL) was combined with acetone (100 mL) and hexane (200 mL) to produce the solvent mixture. Preparation of Tomatoes It is important to note that the following procedure was carried out with the lights turned off and the blinds closed, to minimise light exposure. Three groups of tomatoes were made, each containing roughly 300 g of a different variety of tomato. The groups were selected as follows: • Truss Tomatoes • Roma Tomatoes • Cherry Tomatoes

Tomatoes in group 1 were blended in a Thermomix to form a uniform puree. This was placed into a 500 mL Schott bottle and covered with aluminium foil to protect the sample from light. This bottle was marked with the group number. These steps were repeated for each group, resulting in three separate Schott bottles and three varieties of puree. Extraction of Lycopene Tomato puree (0.6 g) was accurately weighed into a 20 mL volumetric flask on an analytical balance. This was repeated five times for each variety, resulting in fifteen flasks with puree. Each flask was filled to the 20 mL mark with the Hexane: BHT: acetone solvent mixture using a clean, dry, glass pipette. Each stoppered flask was shaken briefly, and magnetic stirrer bars were added. The mixtures were stirred magnetically for fifteen minutes. The flasks were then filled to the top with distilled water and stirred for an additional five minutes. Finally, the flasks were shaken and allowed to stand until two distinct layers were visible. This was made up of a coloured organic layer on top and a colourless aqueous layer underneath. Colourimetry A colourimeter with Spark Vue data logging software was used to determine the absorbance of each sample. A small sample of the Hexane: BHT: acetone solvent mixture was placed into a clean, quartz cuvette. This was used to calibrate the colourimeter at a wavelength of 500 nm. The coloured, organic layer from the top of each flask was removed with a plastic pipette and used to fill a cuvette. This was placed into the colourimeter and the absorbance at 500 nm was recorded. These absorbance values were converted into lycopene concentration using the Beer Lambert Law. The absorbance value was calculated through the colourimeter. The molar absorption coefficient for lycopene is 17.2 ∗ 104 𝑀𝑀𝑀𝑀−1 ∗ 𝑐𝑐𝑐𝑐𝑚𝑚𝑚𝑚−1 (Fish et al., 2002). The optical path length was the length of the cuvette, in this case 1 cm. Using this law, an equation was constructed (Equation 1) to calculate the milligrams of lycopene per gram of tissue: (1)

Equation 1: Calculating the milligrams of lycopene per gram of tissue. (After: Fish et al., 2002) Science Extension Journal • 117

Scientific Research in School Volume 3 Issue 1 2021

Results

Table 5: Post-Hoc Tukey test p-values for comparing the mean concentration of lycopene extracted from each variety.

Table 1: Absorbance values for each sample Variety A B C D Truss 0.419 0.422 0.394 0.393 Roma 0.732 0.662 0.626 0.609 Cherry 0.348 0.280 0.310 0.286 Table 2: Concentration of lycopene in tissue) Variety A B C Truss 21.788 21.944 20.488 Roma 38.064 34.424 32.552 Cherry 18.096 14.560 16.120

E 0.389 0.604 0.305

each sample (mg/g D 20.436 31.668 14.872

E 20.228 31.408 15.860

Table 3: Mean and standard deviation for each variety.

Mean concentration (mg/100 g)

Standard deviation

Truss Roma Cherry

20.9768 33.6232 15.9016

0.8194 2.7495 1.3897

Mean concentration of lycopene (mg/100 g)

Variety

Mean concentration of lycopene for each variety 40 30 20

33.6232 20.9768

15.9016

10 0

Truss

Roma

Cherry

Variety of Tomato Figure 4: Bar graph representing mean concentration of lycopene for each tomato variety.

H0 = There is not a significant difference in the average concentration of lycopene in each variety. HA = There is a significant difference in the average concentration of lycopene in each variety. Table 4: ANOVA test output for comparing the concentration of lycopene extracted from different varieties.

Average Standard Deviation F statistic Alpha Value P value Analysis

118 • Science Extension Journal

Truss

Roma

Cherry

20.9768

33.6232

15.9016

0.8194

2.7495

1.3897

122.93912 0.05 < 0.00001 p < 0.05, significant difference in means

Treatment Pairs

Tukey HSD Q statistic

Tukey HSD pvalue

1:2

15.36

< 0.000001

1:3

6.17

0.0247

2:3

21.53

< 0.000001

Tukey HSD analysis Significant (*p < 0.05) Significant (*p < 0.05) Significant (*p < 0.05)

Discussion Analysis of results The extraction of lycopene from each variety of tomato showed quite concise results. From inspection, the mean concentration of lycopene was different between each variety. Roma tomatoes were found to have the highest lycopene concentration, followed by Truss tomatoes, and lastly Cherry tomatoes. The low standard deviation values for each group supported these results, indicating a low variance in the data points. This suggests there is a difference in the concentration of lycopene between different tomato varieties, but further analysis was required to determine this. An ANOVA test was conducted to determine whether or not there was a significant difference between the means. This also involved a Post-Hoc Tukey test, which indicated, if any, which two groups had a significant difference. ANOVA results generated a p-value < 0.00001, which was less than the alpha value of 0.05. This allowed for the rejection of the null hypothesis, supporting the alternate hypothesis that there was a significant difference in the average concentration of lycopene in each variety. The Post-Hoc Tukey test also indicated a significant difference in concentration between all 3 groups, each pair showing a p-value of less than the alpha value of 0.05. This research agrees with previous reports, which also suggested different concentrations of lycopene in different tomato varieties. In 2008, Toma et al. also found that on a dry weight basis Roma tomatoes contained the highest amount of lycopene, while Truss tomatoes contained the least. This compares very similarly to the results of this report. These results could be due to a range of factors. Each variety of tomato is grown under different conditions, and the shape and size differ significantly. To gain a further understanding on the topic much more research could be done into the specific elements of the varieties that lead to this difference in lycopene concentration, as the results show this is quite a significant amount.

Scientific Research in School Volume 3 Issue 1 2021

One research area is the concentration difference of lycopene in the skin of the tomato versus the flesh. This could have a major impact between varieties, since a key difference between each is the size and shape. Cherry tomatoes are generally smaller, and therefore have a higher ratio of skin to flesh than larger tomatoes. This could be experimented through similar methods outlined in this report. Furthermore, the growing conditions of each variety may have an effect on the lycopene concentration. This includes things such as irrigation, access to sunlight, location, and time until harvest – all which fluctuate with the growth of different varieties. Processing methods could also severely impact the lycopene content, including time from farm to supermarket, packaging, and additional treatments. It has been found that one way to maximise lycopene concentration is the processing of raw varieties into products such as tomato paste and tomato juice. This had been investigated by many reports, indicating that this is because of the high heat conditions in some processing methods. Also, the interactions between lycopene and fats enhance its bioavailability, suggesting cooking processes using oils to form tomato sauces and paste are what lead to this increase (Soares et al., 2019). Although exposure to temperature was not studied in this report, this research suggests lycopene concentration is also affected by the variety of tomato, indicating that this should be considered when forming commercial tomato products. It is therefore important for producers to consider the type of tomato they are growing, especially if they are aiming at maximising the antioxidant properties of their produce. This is also an important factor for consumers to consider when aiming at increasing their antioxidant intake as part of achieving a more balanced diet. Although this can be an important thing to consider in some cases, it is often recommended to consume any variety of tomato for its antioxidant properties, and this research still indicates that all tomatoes possess high levels of lycopene. Possible sources of error Although the methodology produced clear and concise results, there were some potential sources of error that could be minimised in further research. It was fairly difficult to properly blend the tomatoes, leaving moderately sized pieces of skin in the mixture instead of a uniform puree. This was found to be quite problematic, since it was difficult to take a small sample of each puree to test that was a true reflection of the tomato. For example, some groups had flakes of skin while others had none. There was, however, a large effort in ensuring

each group’s puree was similar in consistency, and multiple samples were taken to reduce the impact of this error margin. Since the tomatoes were store bought, it was difficult to know whether all variables were fully controlled. Differences in lycopene concentration therefore could be due to other factors, such as the time harvested, or processing and packaging methods. It would have been ideal to grow the tomatoes for this experiment, but time was a major issue. Therefore, for future experimentation, to improve reliability of results the tomatoes should be grown instead of purchased. Furthermore, the colourimetry method of extraction posed some sources of error that were ultimately unavoidable. This mainly involved the interference from other carotenoids, such as β-carotene in the absorbance of light. To minimise this impact, a wavelength of 500 nm was chosen, aimed at maximising the absorbance potential of lycopene and minimising interference from other compounds. This should not significantly impact the difference in means between varieties though, since all samples were tested at the same wavelength. This reflection of concentration would not be entirely accurate, compared to HPLC methods. It was, however, important to apply this method, since more advanced HPLC methods are impossible to do in a school lab, as the equipment is expensive and not easily accessible. It could therefore be very beneficial for this research to be conducted with a more advanced method, aiming at refining and confirming the results found. Areas of further research This report provides a starting point for further research to be conducted, as there are many improvements that could be made to ensure all other variables are controlled. It does, however, suggest a difference in concentration exists, allowing for more research to be built off this. More research could also be conducted on why this relationship exists, such as an investigation of the effect of growing conditions on tomato lycopene, or the difference in concentration of lycopene in the skin versus the flesh of the tomato. Also, research could be conducted in investigating the effect of different variables on the concentration of lycopene in tomatoes. One example is a further look at processed tomato products, gaining a deeper understanding why they contain higher concentrations of lycopene.

Conclusion This report has investigated the effect of different varieties of tomato on the concentration of lycopene extracted. Three groups were created, each consisting of Science Extension Journal • 119

Scientific Research in School Volume 3 Issue 1 2021

either Truss, Roma, or Cherry tomato varieties. These were each mixed with a solvent Hexane: BHT: acetone mixture, and absorbance at 500 nm was recorded using a colourimeter. These absorbance values were converted into concentration of lycopene using Beer Lambert’s Law. The data analysis involved an ANOVA test to compare whether any of the mean concentrations were significantly different, followed up with a Post-Hoc Tukey test to determine which groups were different. The results of this showed all groups had significantly different mean concentrations of lycopene, leading to the acceptance of the hypothesis that the variety of tomato affects the concentration of lycopene extracted.

Acknowledgements I would like to thank Dr Katie Terrett for her assistance in formulating an idea, help in setting up the method, guidance throughout the data collection process, and for her support in writing and editing my report. Also, thanks to James Wilson, for suggesting valuable edits and supporting me in my data collection.

References Agarwal, A., Durairajanayagam, D., Ong, C. and Prashast, P. (2014). Lycopene and male infertility. Asian Journal of Andrology, 16(3), p.420. Agarwal, S. and Rao, A.V. (2000). Tomato lycopene and its role in human health and chronic diseases. Canadian Medical Association, 163(6), pp.739–744. Baliyan, S.P. and Rao, M.S. (2013). Evaluation of Tomato Varieties for Pest and Disease Adaptation and Productivity in Botswana. International Journal of Agricultural and Food Research, 2(3), pp.20–29. Davis, A.R., Fish, W.W. and Perkins-Veazie, P. (2003). A rapid spectrophotometric method for analyzing lycopene content in tomato and tomato products. Postharvest Biology and Technology, 28(3), pp.425–430. de Montemas, A. (2020). Storage temperature and its effect on the concentration of Lycopene extracted from tomatoes.

120 • Science Extension Journal

Fish, W.W., Perkins-Veazie, P. and Collins, J.K. (2002). A Quantitative Assay for Lycopene That Utilizes Reduced Volumes of Organic Solvents. Journal of Food Composition and Analysis, 15(3), pp.309–317. Gärtner, C., Stahl, W. and Sies, H. (1997). Lycopene is more bioavailable from tomato paste than from fresh tomatoes. The American Journal of Clinical Nutrition, 66(1), pp.116–122. Lobo, V., Patil, A., Phatak, A. and Chandra, N. (2010). Free radicals, antioxidants and functional foods: Impact on human health. Pharmacognosy Reviews, 4(8), p.118-126. May, P. (2020). Lycopene. [online] www.chm.bris.ac.uk. Available at: http://www.chm.bris.ac.uk/motm/lycopene/lycopeneh.htm [Accessed 8 Feb. 2021]. Rao, A.V., Waseem, Z. and Agarwal, S. (1998). Lycopene content of tomatoes and tomato products and their contribution to dietary lycopene. Food Research International, 31(10), pp.737–741. Soares, N. da C.P., Elias, M. de B., Lima Machado, C., Trindade, B.B., Borojevic, R. and Teodoro, A.J. (2019). Comparative Analysis of Lycopene Content from Different Tomato-Based Food Products on the Cellular Activity of Prostate Cancer Cell Lines. Foods (Basel, Switzerland), 8(6), pp.1–14. Sousa, F.A. de, Neves, A.A., Queiroz, M.E.L.R. de, Heleno, F.F., Teófilo, R.F. and Pinho, G.P. de (2014). Influence of Ripening Stages of Tomatoes in the Analysis of Pesticides by Gas Chromatography. Journal of the Brazilian Chemical Society, 25(8). Story, E.N., Kopec, R.E., Schwartz, S.J. and Harris, G.K. (2010). An Update on the Health Effects of Tomato Lycopene. Annual Review of Food Science and Technology, 1(1), pp.189–210. Toma, R.B., Frank, G.C., Nakayama, K. and Tawfik, E. (2008). Lycopene content in raw tomato varieties and tomato products. Journal of Foodservice, 19(2), pp.127–132. Williams, G. (1999). Diet and cancer prevention: the fiber first diet. Toxicological Sciences, 52(90001), pp.72–86.

Scientific Research in School Volume 3 Issue 1 2021

Effect of time after harvest on chlorophyll concentration in spinach leaves Kyle Scholtz Barker College Chlorophyll-a and chlorophyll-b are chemical compounds found in photosynthetic plants and its application in medicine is becoming widely accepted. This report aimed to reveal how time after harvest affects chlorophyll-a and chlorophyll-b concentration. Spinach leaves where stored over different time periods and dimethylsulfoxide was used to extract chlorophyll from leaf tissue. A colourimeter was used to determine chlorophyll-a and chlorophyll-b concentration. The results demonstrated that chlorophyll-a concentration had no significant difference after each harvest time, but concentration of chlorophyll-b did have a statistical difference after each harvest time. Concentration of chlorophyll-b was found to be higher 48 hours after harvest compared to instantly after harvest which resulted in the rejection of the null hypothesis. Abbreviations Chlorophyll-a (Chl-a), Chlorophyll-b dimethylsulfoxide (DMSO).

(Chl-b),

Literature review Chlorophyll is a chemical compound or green pigment which is found within the cells of the thylakoid membrane of the chloroplast. Chl-a and Chl-b are present in higher plants whereas chlorophyll c, d and e are found in photosynthetic algae. Chlorophyll reflects and absorbs certain wavelengths of light; its primary role is to absorb light to use for photosynthesis. Plants use the energy collected from the chlorophyll to convert carbon dioxide and water into glucose and oxygen. The glucose produced through photosynthesis is used for energy and the oxygen produced is released into the atmosphere. Without chlorophyll, plants would be unable to undergo photosynthesis and hence would be unable to synthesise carbohydrates.

heads. The structure of a chlorophyll molecule includes a porphyrin ring in the centre of the molecule, as seen in Figure 1. Chl-a and Chl-b differ by one atom in a side chain on the third carbon, Chl-b contains an aldehyde group (-CHO) whereas Chl-a contains a methyl group (CH3), as seen in Figure 1. Chl-a is the primary pigment of photosynthesis whereas Chl-b is an accessory. Chl-a absorbs light from the orange-red and violet-blue regions of the visible spectrum whereas Chl-b absorbs light from blue areas of the visible light spectrum.

The process of photosynthesis is written in the equation: 6CO2 + 6H2O → C6H12O6 + 6O2.

Figure 1: Structure of chlorophyll (Source: May, 1999).

The two main types of chlorophyll are Chl-a and Chl-b, Chl-a is generally found in higher concentration then Chl-b by a 3:1 ratio but it varies between species and can be influenced by a number of factors including pre- and post-harvest treatment and agroclimatic conditions (Ferruzzi & Blakeslee 2007). Chl-a and Chl-b have different roles in the process of photosynthesis, and they have different structures. Both Chl-a and Chl-b are similarly shaped with hydrophobic tails and hydrophilic

It is known that as plants lose the green colour, otherwise known as degreening, it is a result of the degradation of chlorophyll. None of the genes which encode for the catabolic enzymes responsible for the breakdown of chlorophyll have been isolated but chlorophyll degradation has been proven to have multiple degradative pathways (Matile, Hörtensteiner & Thomas, 1999). A study conducted by Yamauchi and Watada (1991) concluded that the degradation of

Science Extension Journal • 121

Scientific Research in School Volume 3 Issue 1 2021

chlorophyll in spinach was regulated through the Peroxidase-hydrogen peroxide pathway which opens the porphyrin ring of the chlorophyll molecule as seen in Figure 2. The opening of the porphyrin ring results in the colour loss of spinach leaves. Chlorophyll has many pathways of degradation and it is not the same in every plant, some plants rely on chlorophyllase which results in the release of the phytol chain in chlorophyll to form chlorophyllide (Yamauchi & Watada, 1991). The main chlorophyll degradative reactions and the derivatives formed are summarised in Figure 3.

Figure 2: Peroxidase-hydrogen peroxide pathway. (Source: Kaewsuksaeng, 2011)

The benefit of chlorophyll to humans has been extensively researched in the medical field and it has

Figure 3: Derivatives of chlorophyll. (Source: Ferruzzi & Blakeslee, 2007)

122 • Science Extension Journal

been found that chlorophyll and its derivatives, as seen in Table 1, have many medical benefits. When chlorophyll is digested, it is exposed to an acidic environment resulting in conversion to metal free pheophytins. Chlorophyll derivatives are then absorbed by intestinal cells and enter blood circulation which allows the derivatives to act through a wide range of mechanisms, including being a modifier for genotoxic effect and antioxidant activity. A genotoxin is an agent or chemical which can cause chromosomal or DNA damage. A study conducted by Waters et al. (1996) assessed the antimutagenicity profile for chlorophyll and their research concluded that chlorophyll is antimutagenic to range of direct- and indirect-acting mutagens (Waters et al. 1996). A study conducted by Osowski et al. concluded that consumption of chlorophyll in food does not significantly protect against mutagenic compounds but a derivative of chlorophyll, chlorophyllin, which does not occur naturally, can act as a binding agent against mutagenic compounds if it is used as a supplement (Osowski et al. 2010). Chlorophyllin was found to have applicable therapeutic measures for individuals exposed to aflatoxin, as chlorophyllin was found to be effective in preventing liver cancer (Egner, Kensler & Muñoz, 2003).

Scientific Research in School Volume 3 Issue 1 2021

Table 1:Chlorophyll and its derivatives used in medicine (Source: Mishra, Bacheti & Husen, 2011).

Chlorophyll Category Natural chlorophyll Metal free chlorophyll derivatives

Metallochlorophyll derivatives

Derivatives

Chlorophyll a, b, c, d, e Pheophytin, Pyropheophytin Zn-Pheophytin Zn-pyropheophytin Chlorophyllide Pheophorbide Cu(II)chlorin e 4 Cu-chlorin e 6 Cu-chlorin e 4 ethyl ester

A study conducted by Barnes et al. (1992) concluded that the use of Dimethyl Sulfoxide (DMSO) as a solvent is an effective way of extracting chlorophyll from plants due to its amphiphilic properties. After incubation in DMSO, further extraction resulted in no additional chlorophyll and there was no green colour left in plant tissue which concluded that through incubating plant tissue in DMSO, the complete extraction of chlorophyll can take place. The use of DMSO as a means to extract chlorophyll proved to be more effective than previous methods such as using 80% acetone as incubation in DMSO as it resulted in no chance in the ratio of Chl-a and Chl-b (Barnes et al. 1992). Through knowing how time after harvest affects chlorophyll concentration, the degradation pathways of chlorophyll can be better understood. The degradation of chlorophyll results in a variety of derivatives which are becoming widely accepted for their application in the medicinal field. Through understanding how time after harvest affects chlorophyll concentration, a better understanding of when chlorophyll is most useful can be formed

Scientific research question How does time after harvest affect the concentration of chlorophyll-a and chlorophyll-b extracted from spinach?

Scientific hypothesis As time after harvest concentration will decrease.

increases,

chlorophyll

March 2021. The plant was grown outdoors under a netting to protect against animals and environmental stresses. The plant was grown in favourable conditions and was provided adequate sunlight, water and cooling when necessary. Four sets of three spinach leaves were harvested and stored in complete darkness at room temperature. Each set of leaves were stored over different time periods, 0 hours, 24 hours, 48 hours and 96 hours. After each set of leaves were stored for the required time period the extraction of chlorophyll occurred. Extraction of chlorophyll-a and chlorophyll-b (carried out for all time points) Plant tissue from each leaf was cut into small pieces, approximately 5mm x 5mm, using a sterilized scalpel. 0.1g of the plant tissue, one sample from one leaf and two samples from the other two leaves (5 samples) were weighed and then placed in separate test tubes. 10ml of DMSO solvent was added into each test tube which was shaken for five seconds. The test tubes were placed into a test tube rack and incubated in a water bath with a constant temperature of 600-650C for 60 minutes which allowed for decolourisation of leaf tissue. After the 60minute incubation period, when the leaves were fully decolourised, the test tubes were removed from the water bath and cooled at room temperature for 30 minutes. This method was completed for each time point. Colourimetry (Carried out for all time points): A PASCO colourimeter was used alongside Spark Vue data logging software. A clean quartz cuvette was filled with DMSO solvent and the colourimeter was calibrated at 650 nm. A sample from each test tube was removed using a dropper the solution was placed in separate, clean quartz cuvettes. Each cuvette was placed in the colourimeter and absorbance was measured at 650 nm for each sample and results were recorded. Beer-lamberts law (Equation 1) was used to calculate Chl-a and Chl-b concentration. The extinction coefficient (ε) used for Chl-a was 18.47 and the extinction coefficient used for Chl-b was 50.81. (Inskeep & Bloom 1985).

A = εlc Equation 1: Beer-lambert law.

Methodology Preparation A spinach plant (swiss chard variety) was grown from the beginning of December 2020, up until the end of

Science Extension Journal • 123

Scientific Research in School Volume 3 Issue 1 2021

Results Table 2: Mean concentration of Chl-a and standard deviation for each group.

Group

Time after harvest (hours) 0 24 48 96

1 2 3 4

Mean concentration (mg/L)

19.1364 18.768 22.4336 21.091

2.3678 1.5768 3.391 3.6041

Table 3: Mean concentration of Chl-b and standard deviation for each group.

20 15

Analysis

Chl-a concentration (mg/L)

10 5 0

Time after harvest (Hours)

Figure 4: Mean concentration of Chl-a versus time after harvest. Table 3: Mean concentration of Chl-b and standard deviation for each group.

Group

1 2 3 4

Time after harvest (hours) 0 24 48 96

Mean concentration (mg/L)

6.618 7.0702 8.4186 7.4936

0.2066 0.5078 0.9074 1.5612

Chlorophyll-b 25 20 15 10 5 0

Time after harvest (Hours)

Figure 5: Mean concentration of Chl-b versus time after harvest.

H0 = That there is no statistical difference between the mean chlorophyll concentration after each harvest time.

124 • Science Extension Journal

19.136

24 hours (T2) 18.768

48 hours (T3) 22.433

96 hours (T4) 21.091

2.3678

1.5768

3.391

3.6041

0 hours (T1) Average Standard deviation F-statistic Alpha Value P value

Chlorophyll-a

Chl-b Concentration (mg/L)

HA = That there is a significant difference between the mean concentration of chlorophyll after each harvest time.

1.81412 0.05 0.185179 p value > Alpha value, therefore values are not statistically significant.

H0 = That there is no statistical difference between the mean chlorophyll concentration after each harvest time. HA = That there is a significant difference between the mean concentration of chlorophyll after each harvest time. Table 4: ANOVA test output for comparing concentration of Chl-b extracted after different times after harvest.

Average Standard deviation F-statistic Alpha Value P value Analysis

0 hours (T1) 6.618

24 hours (T2) 7.0702

48 hours (T3) 8.4186

96 hours (T4) 7.4936

0.2066

0.5078

0.9074

1.5612

3.307 0.05 0.047164 p value < Alpha value, therefore values are statistically significant.

Table 5: Post-Hoc Tukey test p-values for comparing the mean concentration of Chl-b extracted at each time point after harvest Tukey Tukey HSD QTreatment Pairs HSD analysis stat p-value M1 = 6.62 T1:T2 1.07 0.872 insignificant M2 = 7.07 M1 = 6.62 significant T1:T3 4.27 0.036 M3 = 8.12 (p < 0.05) M1 = 6.62 T1:T4 2.07 0.479 insignificant M4 = 7.49 M2 =7.07 3.20 0.150 insignificant T2:T3 M3 = 8.42 M2 = 7.07 T2:T4 1.00 0.891 insignificant M4 = 7.49 M3 = 8.42 T3:T4 2.19 0.433 insignificant M4 =7.49

Scientific Research in School Volume 3 Issue 1 2021

Discussion This report explored how time after harvest affects Chla and Chl-b concentration in spinach. As seen in Tables 2 and 3, the ratio of Chl-a to Chl-b was approximately 3:1 which aligns with previous studies, (Ferruzzi & Blakeslee, 2007). An ANOVA statistical analysis was completed for Chla, as seen in Table 4. The p-value calculated for Chl-a was 0.185179 and the alpha value used was 0.05. The calculated p-value was greater than the alpha value, hence the null hypothesis cannot be rejected, (Ho = There is no significant statistical difference between the means of each group.). This suggests that time after harvest has no significant affect on Chl-a concentration in spinach. However, due to time pressure and limited recourses the longest time period after harvest where Chl-a concentration was measured was 96 hours which decreases the reliability of the experimental results. The extinction coefficient used to calculate Chl-a concentration was 18.47 which is the extinction coefficient used when absorbance is measured at 647 nm (Inskeep & Bloom 1985). The colourimeter used in the experiment was unable to measure absorbance at 647 nm but absorbance was measured at 650 nm and 18.47 was still the value used as the extinction coefficient which decreased the accuracy of the results. An ANOVA statistical analysis was also completed for Chl-b, as seen in Table 5. The calculated p-value for Chl-b was 0.047164 which is smaller than the alpha value used (0.05). p-value < alpha-value, hence the null hypothesis can be rejected, and the alternative hypothesis can be accepted, (HA = There is a significant statistical difference between the means of each group.). This suggests that time after harvest does affect Chl-b concentration in spinach. The calculated f-ratio was 3.307 which indicated that there was an overall difference between the sample means. A Post-Hoc Tukey test, as seen in Table 6, was conducted and it concluded that the variance between T1 (0 hours) and T3 (48 hours) were statistically different. This was evident because the p-value calculate was 0.03696 and the value for Q was calculated as 4.27 which indicates a statistically significant difference between the two groups. Chl-b concentration increased between T1 and T3 which was different to Chl-a concentration which had no statistically significant differences between groups. A possible reason for these results is that the ratio of Chl-a to Chl-b is a 3:1 ratio, hence there is much lower concentration of Chl-b. This can possibly lead to slightly different results as changes in Chl-b concentration can be more statistically significant due to the lower concentration. It is also possible that this result is an outlier but due to the large sample size (20 samples), it

is unlikely to be an outlier. Similar to Chl-a, the coefficient used to calculate the concentration of Chl-b (50.81) is the value used when absorption is measured at 647 nm. This source of error has the same affect on Chl-b as it has on Chl-a. With the exception of the value calculated for Q between T1 and T3 for Chl-b, the majority of the results concluded that time after harvest does not affect chlorophyll concentration. The concluded results where unexpected as the hypothesis stated that time after harvest would affect chlorophyll concentration. It is possible that the duration of the experiment was a source of error as chlorophyll concentration was only calculated at four time points. Through increasing the length of the experiment and adding more time points, a better understanding of how time after harvest affects chlorophyll concentration can be formed. It is possible that results did not align with literature because chlorophyll may take longer to degrade then what time allowed for in the experiment. Through increasing time after harvest, a larger dataset could have been collected and more reliable results could have been collected. It was due to limited time that a larger number of time points could not have been recorded. In the extracted solutions, other pigments may have been present, hence the pigments present may have absorbed some of the light. This is a possible source of error which could have impacted the accuracy of the experimental results. Multiple sources of error such as not having an accurate colourimeter and limited time resulted in inaccuracies throughout the experimental results. By measuring absorbance at 650 nm instead of 657 nm, inaccurate results were recorded. Limited time resulted in a small data set which negatively impacts the reliability of the experimental results. In order to develop a deeper understanding of how time after harvest affects chlorophyll, further studies must be conducted. To extend on results concluded from this report, a wider range of time points will be needed, alongside a more accurate colourimeter to ensure accurate results. Through understanding how time after harvest affects chlorophyll levels, yield of natural chlorophyll can be increased through knowing when chlorophyll is at its highest concentration. This knowledge can be applicable to the medicinal field as chlorophyll is becoming increasingly accepted in medicine. Chlorophyll is becoming increasingly significant in science because medical applications of chlorophyll are increasing and through a better understanding of chlorophyll and how it works, its application to medicine can soon become better understood.

Science Extension Journal • 125

Scientific Research in School Volume 3 Issue 1 2021

Conclusion

Acknowledgments

There has been significant potential found in the use of chlorophyll in medicine. This experiment investigated how time after harvest affects Chl-a and Chl-b concentration in spinach leaves. DMSO was used to extract chlorophyll from spinach leaf tissue which allowed for absorption of the solutions to be recorded. The concentration of Chl-a recorded at each time point did not reveal any statistically significant differences which resulted in the null hypothesis being accepted. Chl-b revealed a statistically significant difference between Chl-b concentration measured at T1 and T2. Chl-b concentration increased between T1 and T2 but further research will need to be conducted to further understand how time after harvest affects chlorophyll concentration.

I would like to thank Dr Katie Terrett for her extensive assistance in the completion of the report. I would also like to thank Mr Robert Paynter for his supervision when the experiment was completed.

The results collected revealed that between 0 hours after harvest and 96 hours after harvest, Chl-a concentration did not significantly change. Chl-b however did reveal a statistically significant difference between timepoints as chlorophyll concentration was recorded to be higher 48 hours after harvest compared to instantly after harvest. The results addressed the research question and revealed how time after harvest affects chlorophyll concentration, but only at four different time points. Due to limited time the longest period spinach was stored at was 96 hours which only allowed chlorophyll concentration to be recorded up until 96 hours. There is need for further research in multiple areas surrounding this report including further research on the degradation pathway of chlorophyll, finding more accurate equations for determining chlorophyll concentration and recording how chlorophyll concentration is affected by time after harvest through using a larger data set. Chlorophyll has the potential for widespread medical use however further research must be conducted before it can be better understood.

126 • Science Extension Journal

References Barnes, J. D, Balaguer, L, Manrique, E, Elvira, S & Davison, A., 1992, ‘A reappraisal of the use of dmso for the extraction and determination of chlorophylls a and b in lichens and higher plants’, Environmental And Experimental Botany, vol. 32. Egner, P., Kensler, T., & Muñoz, A., 2003, ‘Chemoprevention with chlorophyllin in individuals exposed to dietary aflatoxin’, Elsevier Science B.V. Ferruzzi, M., & Blakeslee, J., 2007, ‘Digestion, absorption, and cancer preventative activity of dietary chlorophyll derivatives’, Nutrition Research, vol. 27. Inskeep, W., & Bloom, P., 1985, ‘Extinction Coefficients of Chlorophyll a and b in N,N-Dimethylformamide and 80% Acetone’, Plant Physiol, vol. 77. Kaewsuksaeng, S., 2011, ‘Chlorophyll Degradation in Horticultural Crops’, Walailak J Sci & Tech, vol. 8. Matile, P, Hörtensteiner, S., & Thomas, H., 1999, ‘Chlorophyll Degradation’, Plant Physiol., vol. 50. May, P., 1999, Chlorophyll, Bris.ac.uk. Mishra, V., Bacheti, R., & Husen, A., 2011, ‘Medicinal Uses of Chlorophyll: A Critical Overview’, Chlorophyll: Structure, Function and Medicinal Uses. Osowski, A., Pietrzak, M., Wieczorek, Z., & Wieczorek, J., 2010, ‘Natural Compounds In The Human Diet And Their Ability To Bind Mutagens Prevents Dna–Mutagen Intercalation’, Toxicology and Environmental Health, vol. 73. Waters, M., Stack, H., Jackson, M., Brockman, H., & De Flora, S., 1996, ‘Actvity profiles of antimutagens: in vitro and in vivo data’, Mutation research, vol. 350. Yamauchi, N & E. Watada, A 1991, ‘Regulated Chlorophyll Degradation in Spinach Leaves during Storage’, J. Amer. Soc. Hort. Sci., vol. 116.

Scientific Research in School Volume 3 Issue 1 2021

As simple as making hot dogs: Using analogies to teach limiting and excess reagents in high school chemistry. Cleo Christie-David Barker College Analogies are commonly used to communicate complex, and often abstract, topics in science education. There is limited research into the effect of analogy on how students respond to exam-style questions. This research investigates the potential benefit of using analogy as a teaching model in chemistry with a controlled study involving 34 Year 10 students at a co-educational independent high school in Sydney, Australia. One video used the analogies of hot dogs and burgers, and the other video relied solely upon traditional chemistry examples to explain the concept of limiting and excess reagents. The students were asked exam style questions which required them to recall and explain limiting and excess reagents qualitatively and answer practical questions quantitatively. Although the literature suggested that analogy would have more positive effects, the results failed to enhance this claim. Whilst there were somewhat higher levels of sophistication in responses from the analogy treatment, it was not enough to suggest that analogical reasoning was more beneficial than traditional methods but presents avenues for further research

Literature Review Analogy is an effective tool used in education which involves giving real-life examples to enable students to understand abstract ideas, especially in the sciences. Analogies can be tools of both discovery and the transfer of knowledge as they evoke mental images that are concrete for unsophisticated thinkers (Aubusson, Treagust & Harrison, 2009; Harrison & Treagust, 1993). For example, the recipe of a hot dog from the components of a sausage and bun can become an analogy for the reactants and products in a chemical reaction. Analogies can be valuable tools in conceptual learning by facilitating a cognitive understanding of the abstract to welcome higher-order thinking (Duit, 1991; Richland & Simms, 2015; Sutula & Krajcik, 1988). Successful use of analogy For an analogy to be successful, there must be correspondence between the analogy and the abstract idea involving similar features relating to either “concepts, principles or formulas” (Glynn et al., 1989, p.383). The correspondence is a form of mapping one idea to another with a systematic comparison, either verbally or visually between common and uncommon features (Aberšek, 2016; Aubusson et al., 2009). Through this mapping, analogies allow what is familiar to be used to make the “unfamiliar accessible and

understandable” (Aubusson et al., 2009, p.212; Richland & Simms, 2015). For example, the winds of a bird can be used as an analogy for how the wings of a plane work or the human eye is analogous to how a camera operates. Unsuccessful use of analogy However, analogies can also create misconceptions where there are differing and misleading features between the abstract and target knowledge (Champagne et al., 1985; Dilber & Bahattin, 2008; Thiele & Treagust, 1992), or where students are unfamiliar with the analogy (Gentner & Gentner, 1983; Nagel, 1961). Orgill and

Figure 1: Incorporation of analogy in new knowledge (Source: Thiele & Treagust, 1992)

Science Extension Journal • 127

Scientific Research in School Volume 3 Issue 1 2021

Bodner (2004) suggest that the transfer of irrelevant components, i.e. analogies, between the existing and target knowledge domains can cause the development of misconceptions, which sees students incorporate the analogy into their response. This leads to the argument that analogies are not as useful in the classroom as the literature suggests and as expected (Duit, 1991; Harrison & Treagust, 1993).Thiele and Treagust (1992) have modelled the distinction between a desired and undesired effect of analogies in supporting the formation of target knowledge from existing knowledge (Figure 1). Dilber and Bahattin (2008) elaborate on the undesirable effects when students are unable to separate the analogy from the target and conceptual knowledge becomes incorrect. To exemplify the difference between the desired and undesired effect, if you were to use the analogy of ‘time is money’; the desired effect would be to find the similarities between time and money i.e. that it is valuable and can be wasted; the undesired effect would be to not distinguish the differences and assume that time itself has a monetary value. Harrison and Treagust (1993) explain that these alternative conceptions cause students to visualise the concept in a different matter to what is intended by the teachers, and due to the visual nature are left unchallenged. Such the use of analogies in teaching must address the teaching method rather than the individual thought process of the student (Harrison & Treagust, 1993). Effective teaching using analogy Research is limited as educational research on analogy only became a significant field in the late 1980s (Aubusson et al., 2009). There has been little experimental research on the way students draw on analogy in the absence of a teacher (Aubusson et al., 2009; Duit, 1991; Thiele & Treagust, 1992), hence, examination style questions will be used as a post-test in this present study. However, the question remains on whether analogies are both interpreted and applied in the way in which they are intended by educators. The currently supported view by academics is that “when learners construct their own knowledge, it is both transferable to and usable in later learning situations” (Thiele & Treagust, 1992, p.3; Sutula & Krajcik, 1988). The relational reasoning of analogy is considered by Richland and Simms (2015) to be the “cognitive underpinning of higher order thinking”, promoting advanced reasoning (p. 177). According to Fredricks et al. (2004), it is thought that in constructing one’s own knowledge, their cognitive engagement will promote a deeper understanding as well as a willingness to further the complexity. Engagement is imperative in instilling the “willingness to exert the effort necessary to comprehend complex ideas and master difficult skills"

128 • Science Extension Journal

(Fredricks et. Al, 2004, p.60). Such the increased engagement when using analogy in classroom instruction can have benefits on the motivation and dedication of students in education (Fredricks et. Al, 2004; Richland & Simms, 2015). Analogies and Chemistry Education Analogies can be particularly helpful in science due to the highly abstract nature of theories and models (Harrison & Treagust, 1993; Nagel, 1961; Muniz & Oliver-Hoyo, 2014; Richland & Simms, 2015). They can serve as “initial models, or simple representations, of scientific concepts” which drive as both inspiration and explanations for scientific discovery (Aberšek, 2016, p. 4; Kaiser, 1989). Analogies engage students in scientific thinking in classroom settings through learning and teaching as well as textbooks (Aberšek, 2016; Richland & Simms, 2015; Metsala & Glynn, 1996; S˛endur et. al, 2011). Analogy is generally considered beneficial in chemistry education as it can be “utilized as motivation for the connection of concepts such that students create and maintain a rich, holistic viewpoint of science overall" (Muniz & Oliver-Hovo, 2014, p. 25). In chemistry education, analogies are particularly helpful in providing a bridge between unfamiliar and abstract concepts and preconceived knowledge (Thiele & Treagust, 1992). Chemistry often deals with scientific phenomena such as particles, energy and matter which are invisible to the naked eye and unfamiliar (Aubusson et al., 2009; Dilber & Bahattin, 2008; Duit, 199; Muniz & Oliver-Hoyo, 2014; Richland & Simms, 2015). Often visualisations are used to help students with these concepts, however, without the “aligning and mapping between this representation and the natural phenomenon,” can fail to be as effective as analogical reasoning (Richland & Simms, 2015, p. 185). In a chemistry education context, the desired model of bridging analogy gives students the opportunity to develop inferences, prompts conceptual change and moves from original ideas about a target phenomenon to reformulate them based on comparison with the analogy itself (Richland & Simms, 2015). The concept of limiting and excess reagents was chosen as it is a topic that is typically taught through the simple analogy of a recipe. Analogical reasoning of a recipe appeals to all as it is not a niche analogy and such it is easily understood and brief. Limiting and excess reagents is also a preliminary topic, and such, the Year 10s did not have any prior knowledge on the topic that would compel the sophistication of their answers.

Scientific Research in School Volume 3 Issue 1 2021

Scientific research question When learning high school chemistry content with analogy, it is more beneficial than learning chemistry traditionally?

Scientific hypothesis That in incorporating analogy into instruction on limiting and excess reagents, in high school chemistry, will lead to a positive difference in how they answer exam-style questions.

Methodology Two different video lessons introducing the Chemistry content of limiting and excess reagents were presented to two randomly selected groups of Year 10 students at a co-educational independent high school in Sydney, Australia – one video involved analogy (Video A: Analogy), whilst the other was traditionally scientific (Video B: Traditional). Instructional Material Both videos explained the concept of limiting and excess reagents, sharing the same explanations and chemical examples, whilst Video B (Traditional) used solely chemical examples, Video A (Analogy) used an analogy of constructing hot dogs and burgers from their constituent parts. To ensure that the same scientific terms were used, parts of the videos were identical. The videos were both between 6:30 and 7:30 minutes in length to approximately mitigate the variable of time and were presented by the same Chemistry teacher. A detailed summary of the identical and corresponding content can be found in Figure 2 and full videos are available on YouTube. 1 Implementation 34 Year 10 students were chosen as they would not yet have encountered limiting reagents in science. During a Science class, students followed a link which randomly assigned them to either the analogy or traditional video which they watched on their own device. In order to investigate the realities of analogy use in exam-question situation after the video, all students then participated in an identical post-test with a combination of both quantitative and qualitative questions that challenged their knowledge and understanding of the concept. Two key questions are included in Figure 3.

Analysis methodology The responses were then collected, randomized and declassified from treatment and control, in order to avoid confirmation bias when the results were reviewed. The quantitative answers and simpler qualitative responses were marked either correctly or incorrectly. The main qualitative response, where students were asked to “Explain how to find the limiting and excess reagents in chemistry”, was assessed on sophistication through the criteria of the accuracy, length and depth of the responses and given a mark out of 5. Finally, the responses were re-identified as being completed by students who had watched the analogy and traditional chemistry videos, and strong differences or similarities were sought, particularly associated with the desired and undesired effect from Thiele & Treagust’s 1992 model. The quantitative data was used to make graphs and averages in order to identify immediate trends.

Results Answers to exam-style questions When given a question of similar difficulty to the examples in the videos, 15 students in the analogy treatment were able to find the limiting reagent, however, only 13 could both name the excess reagent and calculate how many atoms of the reagent would be left over. The traditional treatment saw similar numbers where again 15 students could find the limiting reagent, whilst only 12 could find the excess reagent. Students were asked in part B of the extension question “Which is the excess reactant, and how many atoms will be remaining once the reaction has completed as much as possible”. Only 2 students from the traditional treatment were able to answer both the excess reagent and the number of atoms remaining correctly. It is not enough evidence to say that Video B was more beneficial as there were also two respondents from the traditional treatment that failed to answer the first qualitative question (explaining limiting and excess reagents) and instead wrote “I don’t really know just guessing” and “I was really confused by this video and don’t really understand”.

Video A: http://bit.ly/BCChemistryA Video B: http://bit.ly/BCChemistryB 1

Science Extension Journal • 129

Scientific Research in School Volume 3 Issue 1 2021

Video A (analogy) Explanation of how a lack of hot dog buns limits the amount of hot dogs assembled.

Video B (traditional) Explanation of how a lack of Sodium ions limits the amount of Sodium Chloride formed.

Defining Limiting and Excess Reagents Chemistry Question #1: “In the sodium chloride reaction, if there are 4 sodiums and 1 chlorine available. How many sodium chlorides can be produced?” Worked Solution (using analogy):

Worked Solution (using visualisation of atoms):

Explanation of how a lack of burger meat limits the amount of burgers produced.

Explanation of how a lack of magnesium ions limits the amount of magnesium chloride produced.

Chemistry Question #2: “In the magnesium chloride reaction, if there are 3 magnesiums and 4 chlorine available. How many magnesium chlorides can be produced?” Worked solution (using analogy):

Worked Solution (using visualisation of atoms):

Revision of definitions of limiting and excess reagents Figure 2: A flowchart of the analogy and traditional videos with similarities and differences

130 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Figure 3: Two key questions from the post-test worksheet that allowed for both quantitative and qualitative responses.

Student explanations In the post test, students were asked to qualitatively “Explain how to find the limiting and excess reagents in chemistry”. In the analogy treatment, 7 people mentioned ‘chemistry’ in their response which, while somewhat trivial, it does suggest an ability to separate the analogy from the target chemistry knowledge. In the traditional treatment only 4 people mentioned ‘chemistry’ in their response. Table 1 shows the quotes that encapsulate the emergent themes that arose when looking over the qualitative responses. In the analogy group, there was less of a tendency to use examples compared to the traditional group, and a focus on the misconception that the limiting reagent was “the first ingredient that is used up”. The traditional treatment saw a misconception that the element with the “larger number is excess and smaller number is limiting” which could have led to the inability to correctly answer the later extension question.

Table 1: Exemplar quotes from student responses when asked to explain limiting and excess reagents. Analogy “For example, in a hot dog if there were 5 buns and 3 sausages”

Traditional “I was really confused by this video and don’t really understand”

“The limiting reagent is the first ingredient that is used up in a chemical reaction”

“Larger number is excess and smaller number is limiting”

“i.e. magnesium chloride (MgCl2)”

“Limiting: by how much there is less than the other element” “e.g. 1Na + Cl -> 1NaCl” “For example: if there are 3 magnesium and chloride”

The written answers, in response to the question ‘Explain how to find the limiting and excess reagents in Chemistry’, were assessed on their level of sophistication and given a rating out of five. These ratings amongst the treatment groups are graphed in Figure 5. An example of a 5-rated response which came out of the analogy treatment was, “In Chemistry, if you

Science Extension Journal • 131

Scientific Research in School Volume 3 Issue 1 2021

Traditional

Analogy

Traditional

6 5 4 3 2 1 0

4 5 6 7 Score of Interest

Figure 6: Graph of interest in videos reported by students

Analogy

need or want to find the limiting and/or excess reagent you need first the chemical equation and find out how many compounds can be made out of the elements you’re given. The limiting reagent will be the element you run out of first therefore being unable to make any more compounds, and the excess will be the left over element”. A 2-rated response, also from the analogy treatment, would be “Limiting = used all the atoms to create ions, therefore the reaction stops. Excess = the leftover atoms after the reaction stops”.

RATING OF RESPONSE (5 = MOST SOPHISTICATED) Figure 4: Graph comparing the sophistication of responses of the analogy and traditional group to the question “Explain how to find the limiting and excess reagents in chemistry”

Assessing treatments for helpfulness and interest After watching the video, students were asked about how helpful and interesting they found the instruction. Students were asked to choose a number on a scale from 1 to 9. The results for each treatment group are graphed in Figures 5 and 6. 10

Analogy

Traditional

Figure 6 shows a large differentiation in the scores of interest for both treatment groups. However, the average score of interest for the analogy group was 6.59 and 6.24 for the traditional group. This small increase for the analogy group was also evident in the scores of helpfulness where the analogy average score was 7.94 and the traditional group score of 7.41.

Results There is not significant evidence to suggest that one group answered questions more correctly however there is a tendency for students who watched the analogy video to offer more sophisticated answers (as seen in Figure 5 where 4 students in the analogy group had 5rated responses compared to 1 in the traditional group). Despite this trend, this data, with limited sample size, is not sufficient to draw the conclusion that teaching using the analogy would result in more sophisticated understanding.

Discussion

The literature led to the hypothesis that analogies are more beneficial than traditional instruction when used in science education (Duit, 1991; Richland & Simms, 2015; Sutula & Krajcik, 1988; Thiele & Treagust, 1992), there was very limited clear evidence for this in the results.

2 0

Score of Helpfulness

Figure 5: Graph of helpfulness of videos reported by students

In Figure 5, it can be seen that while most students in both treatment groups selected a high level of helpfulness (6-9 out of 9) there were three students in the traditional who only selected a score mid-score of helpfulness (4-6 out of 9) for the traditional treatment showing a larger differentiation in responses.

132 • Science Extension Journal

Interestingly, there was some evidence to infer that the analogy had the undesired effect rather than the desired where one participant referred to the element as an “ingredient” (Thiele & Treagust, 1992). However, this undesired effect could be explained by the misconception that was created in both videos where we said that the limiting reagent was the element that was “all used up”. This is not an ideal definition; although, it is commonly used in early stages of learning limiting

Scientific Research in School Volume 3 Issue 1 2021

and excess reagents in chemistry. It provides the basis of knowledge necessary for simpler questions, however, as the questions become harder it does not apply. This is a common problem in science where students are taught concepts with simple claims in order to form a basis of knowledge that can then be increased in difficulty (Thiele & Treagust, 1992; Zeitoun, 1984). Whilst the definition of “all used up” was not unreasonable to use for the experiment, with the extension question it became ineffective as the question did not see one of the elements “all used up” and therefore made it hard for students to conclude which the limiting reagent was. The majority of the students could not solve the extension question due to this misconception, however, 2 students from the traditional treatment were able to answer it correctly. Whilst this isn’t a significant difference, the analogy of a recipe could have reaffirmed this notion of “all used up” where the ingredients were used up. In order to address this misconception, future investigation should use the definition of “first to be used up” instead. More difficult ratios would be more beneficial as an easier question may not allow for differentiation in responses which is what was found in the first test group and why more difficult extension questions were introduced. In the extension question majority of students were able to pick that oxygen was the excess reagent, however, could not get the number of oxygens left over correct, indicating that they had either misinterpreted the gas or were dependent on guessing oxygen as it was the largest number given. This variable had been predicted and it was attempted to be controlled including an example in the videos where there was more chlorine than magnesium, and yet magnesium was the limiting reagent. In the extension question, due to the higher level of difficulty, students may have been prone to pick the largest number rather than attempting to solve the equation themselves. A harder question was deliberately chosen to assess whether students were truly using higher order thinking as hypothesised (Duit, 1991; Richland & Simms, 2015; Sutula & Krajcik, 1988). The results indicate that students were rather guessing and dismiss that analogies were more beneficial, with no students in the analogy treatment obtaining a correct answer. Future research As there was minimal difference in short term assessment, future research could test the theory that learning analogy in classrooms approach has been effective in supporting creation of long-term memory easier to access on a later occasion (Richland & Simms, 2015). This could be done through a second test,

delivered a few weeks after the first test to see if analogy has any impact on memory retention of scientific concepts. The video should be changed to avoid the phrase “all used up”, even though it’s the commonly accepted phrase in most chemistry textbooks, it created a misconception that harboured the students from delving into the extension question that challenged their understanding. Student interviews could be more beneficial than surveys, through probing on comparison on the method on analogical teaching to their traditional methods of classroom teaching.

Conclusion In my research project I investigated limiting and excess reagents in chemistry and how when students learn this abstract concept using the analogy of the components of a hot-dog they may be able to better understand chemistry exam questions. From my literature review, I found that analogy is a particularly helpful model in explaining chemistry concepts such as limiting and excess reagents, which are otherwise difficult to understand without attachments to real-world understandings, however they do not always have the desired effect and can occasionally confuse students. Using scientific principles, I designed a controlled experiment where 34 Year 10 science students were randomly allocated to watch one of two instructional videos on limiting reagents that I created in collaboration with my research supervisor. 17 watched a video which used the analogy of hot dogs and burgers, whilst the other 17 watched a traditional educational video. Quantitative and qualitative data was collected and assessed through a post-test. Despite my hypothesis that using analogy would improve understanding and responses on chemistry exam-style questions, there was minimal observable difference between the two groups. Future research might involve a delayed test to see whether students who watched the analogy video found the learning more memorable and applicable in the long term.

Acknowledgements I would like to thank Dr Matthew Hill for his involvement and dedication to my vision throughout the supervision of my project. I would also like to thank Dr Katie Terrett for her feedback and motivation in morning classes. Thank you also to the Year 10 classes that participated in this study for their time and consideration.

Science Extension Journal • 133

Scientific Research in School Volume 3 Issue 1 2021

References Aberšek, B. 2016, ‘Teaching With Analogies: Examples Of A Self-Healing Porous Material’, Problems of Education in the 21st Century, vol. 70, pp. 4-7. Aubusson, P.J., Treagust, D.F. & Harrison, A. 2009, ‘Learning and teaching science with analogies and metaphors’, The world of Science education Handbook of Research in Australasia, vol. 1, Sense Publishers, pp. 199–216. Champagne, A.B., Gunstone, R.F., & Klopfer, L.E. 1985, Instructional consequences of students’ knowledge about physical phenomena in L.H.T. West & A.L. Pines (Eds.). Cognitive structure and conceptual change, Orlando, FL: Academic Press, pp. 259-266. Dilber, R. & Bahattin, D. 2008, ‘Effectiveness of Analogy on Students’ Success and Elimination of Misconceptions’, LatinAmerican Journal of Physics Education, pp. 174–183. Duit, R. 1991, ‘The role of analogies and metaphors in learning science’, Science Education, vol. 75, pp. 649–72. Duit, R., Roth, W. M., Komorek M., & Wilbers J. 2001, Fostering conceptual change by analogies – between Scylla and Carybdis. Learning and Instruction, 11 (4), pp. 283-303. Gentner, D., & Gentner, D.R. 1983, Flowing waters or teeming crowd: Mental models of electricity in D. Gentner & A.L. Stevens (Eds.), Mental models, Hillsdale, NJ: Erlbaum, pp. 99129. Harrison, A.G. & Treagust, D.F. 1993, ‘Teaching with analogies: A case study in grade-10 optics’, Journal of research in science teaching, vol. 30, no. 10, pp. 1291–307. Harrison, A.G. & Treagust, D.F. 2006, ‘Teaching and Learning With Analogies: Friend or Foe?’, Metaphor and Analogy in Science Education: 1. Ed, Science & technology education library, vol. 30, pp. 11–25. Kaiser, W. 1989, Analogien in Physik und Technik im 19. und 20. Jahrhundert [Analogies in physics and technology in the

134 • Science Extension Journal

19th and 20th centuries]. Berichte Wissenschaftsgeschichte, 12 (1), pp. 19-34.

zur

Metsala, J.L. & Glynn, S. 1996, ‘Teaching with analogies: Building on the science textbook’, The Reading Teacher, vol. 49, no. 6, pp. 490–2. Muniz, M.N. & Oliver-Hoyo, M.T. 2014, ‘On the use of analogy to connect core physical and chemical concepts to those at the nanoscale’, Chem. Educ. Res. Pract, vol. 15, no. 4, pp. 87–823. Nagel, E. 1961, The structure of science: Problems in the logic of scientific explanation, London: Routledge & Kegan Paul. Richland, L.E. & Simms, N. 2015, ‘Analogy, higher order thinking, and education’, Wiley interdisciplinary reviews. Cognitive science, vol. 6, no. 2, pp. 177–92. S˛endur, G., Toprak, M. & Pekmez, E.S. 2011, ‘An analysis of analogies used in secondary chemistry textbooks’, Procedia Computer Science, World Conference on Information Technology, vol. 3, pp. 307–11. Sutula, V., & Krajcik, J.S. 1988, The effective use of analogies for solving mole problems in high school Chemistry, National Association of Research in Science Teaching, Lake Ozark, MO. Thiele, R.B. & Treagust, D.F. 1992, Analogies in Senior High School Chemistry Textbooks: A Critical Analysis, Curtin University of Technology, Science and Mathematics Education Centre. Zeitoun, H.H. 1984, ‘Teaching Scientific Analogies: a proposed model’, Research in science & technological education, vol. 2, no. 2, pp. 107–25.

Scientific Research in School Volume 3 Issue 1 2021

Synthesis of Pyrimethamine Jess Samuelson Barker College Pyrimethamine, introduced in 1952, is an essential medicine due to its anti-malarial, lifesaving properties. However, this essential medicine is now harder to buy due to Turing Pharmaceuticals price increase of the essential medicine, raising the price by 5000% overnight, raising concerns surrounding the ethics of big pharma companies such as Turing. In 2016, Sydney Grammar School was able to successfully synthesise pyrimethamine, cheaply, accessibly and within a high school laboratory, bringing to light the unethical price for pyrimethamine. This report will outline Sydney Grammar’s method to successfully replicate and produce a comparable yield to Sydney Grammar to prove the reliability, accessibility, and the inexpensive nature of their method. The compound itself was analysed at the University of Sydney using nuclear magnetic resonance spectroscopy (NMR) to determine the purity and structure of the compound made. After completing the first two parts of the Sydney Grammar method, a greater yield and similar purity was found to that of Sydney Grammar, further emphasising the reliability of their method. Literature Review Malaria is a protozoan parasitic disease transmitted by the female Anopheles mosquito and is found in tropical and sub-tropical regions of the world (Bloland & WHO, 2001) (Figure 1). According to the World Health Organisation (WHO), infection rate is going down, from

238 million new infections in 2000 and 736 000 deaths (WHO, 2020) to 229 million new infections and 409 000 deaths in 2019 (WHO, 2020). There are five different strains of malaria, Plasmodium falciparum, P. vivax, P. ovale, P. malariae and P. knowlesi, with P.falciparum being responsible for 90% of infections (Ashley and Phyo, 2018).

Figure 1: Global map of the geographical distribution of Malaria (Source: WHO, 2020, pg19)

Science Extension Journal • 135

Scientific Research in School Volume 3 Issue 1 2021

According to the WHO malaria website, symptoms will appear in a non-immune individual 10-15 days after being bitten by an infected mosquito (WHO Global, 2021). Early symptoms include fever, headache, and chills and may be difficult to recognise as malaria (WHO Global, 2021). If not treated over a span of 24 hours, infection can progress to severe illness, with an infected individual experiencing nausea, vomiting and diarrhea (CDCP, 2020) and can be fatal, potentially leading to multi-organ failure (WHO Global, 2021). Children with severe malaria usually develop severe anaemia, respiratory distress, or cerebral malaria (WHO Global, 2021). Pyrimethamine Pyrimethamine, (Figure 2), is an antimalarial drug and 2,4-daminopyrimide derivative. 2,4-diamniopyrimides were found to be powerful antagonists of pteroylglutamic acid, further suggesting that pyrimethamine had anti-malarial activity (Falco et al., 1951). It was first synthesised by Gertrude Elion in 1952 and came into medical use in 1953, sold under the name of Daraprim. WHO declared pyrimethamine as an essential medicine (satisfies the priority health care need of the population determined by WHO) as it was an effective drug for treating P.falciparum strains of malaria. It is commonly used in combinations with other compounds, such as sulfoxine, when treating malaria and is combined with sulfoxine and amodiaquine for chemoprevention (WHO, 2019). The drug has also proved effective when treating pregnant women and children with malaria and for treating other diseases such as Toxoplasmosis, caused by a parasite in undercooked, contaminated meat.

Figure 2: Chemical structure of pyrimethamine.

However, in the late 20th century, new strains of P.falciparum were found to have developed in Southeast Asia and Africa, leading to an increase in child mortality (Ashley and Phyo, 2018). In 1990, WHO stopped recommending the triple-drug combination of mefloquine-sulfadioxine-pyrimethamine as resistance to the combination was proven (Oaks et al., 1991). Issues with accessibility Accessibility to pyrimethamine is extremely important, owing to its natural anti-malarial properties and its use

136 • Science Extension Journal

in treating fatal illnesses and diseases. In 2015, Turing pharmaceuticals acquired market rights for Pyrimethamine (Daraprim) from CorePharma and raised prices by 5000% in the United States overnight (Pollard, 2015). To put this into perspective, Daraprim originally cost $13.50 per tablet, which was raised to $750 per tablet (USYD, 2016). For comparison, in Australia, Daraprim costs between 1-2 dollars per tablet. Martin Shkreli, founder of Turing claimed, ‘This isn’t the greedy drug company trying to gouge patients, it is us trying to stay in business.” (Pollard, 2015). Unfortunately, this is not an isolated issue, with drugs used to treat cancer, hepatitis C and high cholesterol also having unfair prices to begin with, while drugs such as cycloserine, used to treat tuberculosis, had a $343.33 price rise per pill. (Pollard, 2015). Even in 2021, Daraprim remains at $750 USD per tablet, affecting those in developing nations, who are more likely to be at risk of developing these illnesses and may be unable to afford the unnecessarily expensive price. As a result of the high price, research around the development of analogues has been stunted, although they could have the potential to be highly effective against fatal diseases. Breaking Good Breaking Good is a project led by Associate Professor Alice Motion at the University of Sydney. The project provides an opportunity for high school students to contribute to drug development to shed light on ethical issues such as raising the price of Daraprim. In 2016, students at Sydney Grammar School were able to cheaply synthesise the compound in a three-part method, using accessible materials that are safe to use by high school students. They were able to successfully produce 3.7g worth of Daraprim in a high school lab (USYD, 2016). This summates to approximately $111000 which is equivalent to 148 tablets of Daraprim. Sydney Grammar School is the only school that has been attempted the synthesis of Daraprim. My project will aim to test the reliability of the Sydney Grammar Method by repeating the three-part method in the Barker College school laboratory, measuring the yield and purity of the compound after each step. This will allow us to compare yields to Sydney Grammar and compare purity, meaning we can analyse the reliability of their method effectively. Reliability within a School Laboratory It is harder to achieve a successful synthesis within a school laboratory due to the limited technology at our disposal and the difficulty in obtaining chemicals.

Scientific Research in School Volume 3 Issue 1 2021

However, based off Sydney Grammar’s successful synthesis, it can be expected that with the same equipment at our disposal that a similar yield and purity can be gained within a Barker laboratory. Planned Synthesis The planned three-step synthesis is the exact procedure Sydney Grammar carried out, testing the reliability by replicating the method within a different school laboratory (Figure 3).

layer and was separated from the aqueous phase. The organic layer was washed with brine (100mL) and dried over anhydrous sodium sulfate, after which the solvent was removed in vacuo to yield 2-(4-chlorophenyl)-3oxopentanentrile (BBG-2020-10, 11.23g, 0.054mol, 81.98%) as a red oil (Figure 4).

Figure 4: Chemical structure of 2-(4-chlorophenyl)-3oxopentantrile.

Figure 3: Synthetic pathway of pyrimethamine.

Scientific research question How reliable is the Sydney Grammar School Method to synthesise pyrimethamine?

Scientific hypothesis That pyrimethamine can be synthesised with a comparable yield at Barker College following the Sydney Grammar School method.

Methodology Part 1 4-chlorophenylacetonitrile (10g, 0.066mol, 1 equiv.), ethyl propionate (6.87g, 0.067mol, 1.05 equiv.) and potassium tert-butoxide (16.31g, 0.15mol, 2.3 equiv.) were combined in Tetrahydrofuran (THF) (100mL) with stirring. The colourless solution turned red, and the temperature of the solution rose rapidly. The reaction flask was covered in aluminium foil to retain heat. All potassium tert-butoxide dissolved after an hour, leaving a homogeneous solution. The reaction mixture was worked up by pouring it into 250mL of 1M Hydrochloric acid (HCl) solution and dichloromethane (100mL) was added. An emulsion was noticed, so a saturated solution of brine was added to further separate the layers. However, a precipitate formed, so the precipitate was filtered out of the organic

Part 2 2-(4-chlorophenyl)-3-oxopentanenitrile (BBG-2020-10, 11.23g, 0.0541mol, 1 equiv.) was dissolved in a mixture of toluene (100ml) and 2-methylpropan-1-ol (12mL). Concentrated H2SO4 (2mL) was added, and the mixture was refluxed for 4 hours using a Dean Stark apparatus. A thin-layer chromatography (TLC) (dichloromethane) of the reaction mixture indicated that some starting material was still present. The main products were two close spots corresponding to the E and Z isomers of the enol ether. The reaction mixture was worked up by pouring onto a solution of saturated sodium hydrogen carbonate and dried over anhydrous sodium sulfate (150mL). The organic layer was separated and dried over anhydrous sodium sulfate. Dichloromethane (100mL) was added to the organic layer. Triethylamine(10mL) and chromatography silica gel (50g) was added to the reaction mixture to convert the unreacted starting material to its very polar triethylammonium enolate salt. TLC indicated that the polar baseline material had disappeared from solution and absorbed on to the silica. The reaction mixture was filtered to remove the silica gel and washed with 1M HCl (50 mL) and deionised water (50mL) to remove all traces of triethylamine. The solvent was dried over anhydrous sodium sulfate and the solvent was removed in vacuo to yield 2-(4-chlorophenyl)-3-(2methylpropoxy)-pent-2-enenitrile as an orange oil (10.85g, 0.044mol, 80.61%) (Figure 5).

Figure 5: Chemical structure of 2-(4-chlorophenyl)-3-(2methylpropoxy)-pent-2-enetrile.

Science Extension Journal • 137

Scientific Research in School Volume 3 Issue 1 2021

Step 1: Synthesis of 2-(4-chlorophenyl)-3-oxopentanentrile.

Figure 6: 1H NMR spectra after step 1.

H NMR (300 MHz, chloroform-d): δ 1.06 (t, J = 7.2 Hz, 3H), 2.66 (m, 2H, H4), 4.65 (s, 1 H, H3), 7.33 (d, J = 8.4 Hz, 2H, H1), 7.41 (d, J= 8.2 Hz, 2H, H2). 1

Figure 7: TLC after step 1.

138 • Science Extension Journal

Figure 9: TLC after substitution reaction

Scientific Research in School Volume 3 Issue 1 2021

Step 2: Synthesis of 2-(4-chlorophenyl)-3-(2-methylpropoxy)-pent-2-enenitrile

Figure 2: TLC after step 1.

Figure 8: 1H NMR spectra after step 2.

H NMR (300 MHz, chloroform-d): δ 0.96 (d, J = 76.7 Hz, 6H, H3), 1.30 (t, J = 7.6 Hz, 3H), 2.12-1.84 (m, 2H), 2.78 (q, J = 7.5 Hz, 2H, H4), 3.83 (d, J = 6.4 Hz, 2H), 7.30 (d, J = 8.8 Hz, 2H, H2), 7.58 (d, J = 8.8 Hz, 2H, H1).

Discussion Step 1: Synthesis of 2-(4-chlorophenyl)-3oxopentanentrile. O N Cl

OEt

KOtBu, THF RT, 82%

O N

Figure 10: Reaction of 4-chlorophenylacetonitrile (2) to form 2-(4-chloroophenyl)-3-oxopentanenitrile (3)

The first step of the synthesis was performed to form Compound 3 from a condensation reaction between Compound 2 and ethyl propionate. The potassium tertbutoxide added is a strong base, allowing deprotonation of the CH2 group within Compound 3. The reaction between Compound 3 and ethyl propionate removed the ethanol from the compound. The yield produced for this

step was 82%, which is slightly smaller than Sydney Grammar’s yield of 90% (SGS, 2016). This may be the result of the precipitate that formed during the work-up phase, and while in the process of removing the precipitate, some of the compound may have been lost. However, this result is still positive, as it is within the same range of Sydney Grammar, taking the complications into account. The polar baseline material that appeared on the TLC (Figure 7) matched the expected increased polarity of compound 3 compared to compound 2. The 1H NMR spectrum revealed the high purity of the substance. The doublet at 7.33 ppm was assigned to the aromatic H1 protons and the doublet at 7.41 ppm was assigned the aromatic H2 protons. This was the expected result, as hydrogen atoms on a benzene ring tend to appear between 7ppm to 8ppm due to the deshielding effect of the benzene ring. The singlet signal at 4.65ppm was assigned to the H3 hydrogen adjacent to the carbonyl

Science Extension Journal • 139

Scientific Research in School Volume 3 Issue 1 2021

and nitril groups. The singlet is further upfield than the two aromatic doublets as the nitrile group and carbonoxygen double bond withdraws electrons, causing the Hydrogen atom to become deshielded in comparison to the upfield signals. The signal at 2.66ppm from the CH2 of the ethyl group, which was expected to be a quartet, but may have been a multiplet due to impurities or restricted rotation at this part of the molecule. The triplet signal at 1.06 ppm indicated that attached to the ethyl group CH2 was a CH3 group (Figure 6). Finally, there is an impurity to be noted between 5.5 and 5.0 ppm, which is most likely residue from the THF solvent used. There is comparable yield between Sydney Grammar’s and Barker College’s compound 3, Sydney Grammar obtained a 90% yield and Barker college obtaining an 82% yield. The NMR shows a mostly pure substance which is comparable to Sydney Grammar. All peaks align from both spectrums, presenting that compound three was formed (Figure 11), meaning that the hydrogens of compound 3 are in similar environments to that of Sydney Grammar. This further emphasises the reliability of Sydney Grammar’s method, as we were able to produce similar results with no significant differences.

Step 2: Synthesis of 2-(4-chlorophenyl)-3-(2methylpropoxy)-pent-2-enenitrile. Et

O N

HO cat. H2SO4 toluene, reflux 81%

O N

Figure 12: 2-(4-chloroophenyl)-3-oxopentanenitrile (3) to form 2-(4-chlorophenyl)-3-(2-methylpropoxy)-pent-2enenitrile (4).

The second step of the synthesis was performed to form Compound 4 under reflux, where compound 3 underwent a substitution reaction. A Dean Stark apparatus (Figure 13) removed water, shifting the equilibrium action to the right. TLC indicated that not all starting material had been removed (Figure 9). Sydney Grammar had undergone a similar outcome, so this was to be expected. To remove this residual starting material, triethylamine was added and reacted with the starting material to form triethylammonium salt. Silica gel was added to remove the salt from the reaction mixture. The NMR revealed that some starting material remained (Figure 8). The yield produced for this step was 81%, which is significantly greater than Sydney Grammar’s yield of 60% (SGS, 2016). This may be caused by the amount of starting material remaining after the Dean Stark Apparatus.

Figure 13: Dean Stark Apparatus Source: (Torsaeter and Abtahi, 2003). Figure 11: Comparison between Barker College (Top) and Sydney Grammar School (Bottom) 1H NMR after step 1. (After: SGS, 2016)

140 • Science Extension Journal

The 1H NMR spectrum revealed to be less pure than compound 3. The H1 and H2 signals were assigned two doublets at 7.58 ppm and 7.30 ppm and was assigned to the aromatic H1 and H2. Similar to compound 3, this was the expected result, as these signals come from the benzene ring which tend to appear between 7ppm to 8ppm due to the deshielding effect of the benzene ring. The doublet signal at 3.83 ppm was assigned to the

Scientific Research in School Volume 3 Issue 1 2021

hydrogens adjacent to the carbonyl and CH3 group (H3). The signal is further upfield than the two doublets as the carbon-oxygen double bond withdraws electrons, causing the Hydrogen atom to become deshielded in comparison to the upfield signals. The signal at 2.78 ppm from the hydrogens assigned is a quartet, and it indicates that it is adjacent to the same carbonyl group, meaning Hydrogen atoms become deshielded. The multiplet signal between 2.12 and 1.84 ppm indicates that the Hydrogen is adjacent to a CH3 group. The signal at 1.30 ppm is a triplet and is adjacent to a CH2 group. The doublet at 0.96 ppm is the CH3 group of the isopropyl compound (Figure 8). There is a significant difference in yields between Sydney Grammar’s and Barker College’s compound 3, Sydney Grammar obtained a 60% yield and Barker College obtaining an 81% yield. Unfortunately, the NMR for Sydney Grammar’s step two synthesis could not be found, so the purities of the two substances could not be compared.

Future Research Unfortunately, due to other commitments and time constraints, the third part of the synthesis was unable to be completed. If further research is to be commenced, step three would need to be successfully synthesised to further solidify Sydney Grammar’s method as reliable. The method is straightforward due to the accessible materials. Comparing yields and NMR would give a more accurate result of the reliability and potential improvements. Proving this method as reliable is important, as it will be able to shed light on the unethical practice of Turing Pharmaceuticals, while also putting a more reliable, cheaper, and more accessible method into place.

Conclusion My research project explored the reliability of Sydney Grammar School’s method of synthesizing pyrimethamine. I synthesised pyrimethamine as a threepart process and observed the yield and purity of the compound. Data was collected by NMR from Sydney University to test the purity of each compound and the mass was of each compound was used to calculate the yield after each part. Data analysis involved interpreting the data from the NMR to determine the chemical structure of the compounds. The results of my data showed a comparable yield and purity to that if Sydney Grammar, leading me to accept my hypothesis that pyrimethamine can be synthesised with a comparable yield at Barker College following the Sydney Grammar School method.

The first two steps of Sydney Grammar’s method of producing pyrimethamine proved successful in testing reliability, with both steps produced the desired compound for the synthesis. The yields produced by Barker College were either comparable or greater than that of Sydney Grammar, and the 1H NMR of step one revealed a similar purity level in both compounds. This provides evidence that the Sydney Grammar method is reliable, as high yields and purity were able to be obtained in two school laboratories, both produced by high school students, further emphasisng the unneeded expenses and inaccessibility of the drug.

Acknowledgements I wish to thank Dr Katie Terret for helping me throughout all aspects of my report and my experiment. I would like to thank Kymberley Scroggie for analysing my compounds and providing me with NMR.

References Ashley, E. and Phyo, A.P. (2018). Drugs in Development for Malaria. pp.861–871. Bloland, P. (2001). Drug resistance in malaria, pp.1–10, 12– 19. Centres for Disease Control and Prevention (2020). CDC Malaria - FAQs. [online] www.cdc.gov. Available at: https://www.cdc.gov/malaria/about/faqs.html#:~:text=Sympt oms%20of%20malaria%20include%20fever [Accessed 9 Jun. 2021]. Falco,E.A., Goodwin, L.G., Hitchings, G.H.,Rollo, I.M and Russell, P.B. (1951). 2:3-Diaminopyrimidines-A new series of antimalarials. British Journal of Pharmacology and Chemotherapy, [online] 6(2), pp.185–200. Available at:https://bpspubs.onlinelibrary.wiley.com/doi/abs/10.1111/j. 1476-5381.1951.tb00634.x [Accessed 9 Jun. 2021]. Oaks, S., Mitchell, V., Pearson, G. and Carpenter, C. (1991). Malaria: Obstacles and Opportunities. 1st ed. Pollack, A. (2015). Drug Goes From $13.50 a Tablet to $750, Overnight. The New York Times. [online] 20 Sep. Available at: https://www.nytimes.com/2015/09/21/business/a-hugeovernight-increase-in-a-drugs-price-raises-protests.html [Accessed 9 Jun. 2021]. Sydney Grammar School (2016). Consult Terrett. [online] Our Experiment. Available at: https://malaria.ourexperiment.org/daraprim_synthesis/15412/ Third_synthesis_of_24chlorophenyl3oxopentanenitrile_SGS _104.html [Accessed 13 Jun. 2021]. The University of Sydney (2016). Students make $750 drug cheaply with Open Source Malaria team. [online] The University of Sydney. Available at: https://www.sydney.edu.au/news-

Science Extension Journal • 141

Scientific Research in School Volume 3 Issue 1 2021

opinion/news/2016/11/30/students-make--750-drug-cheaplywith-open-source-malaria-team-.html [Accessed 9 Jun. 2021].

World Health Organisation (2020). World Malaria Report 2020.

Torsaeter, O. and Abtahi, M. (2003). Experimental Reservoir Engineering Laboratory. Norwegian University of Science and Technology.

World Health Organisation (2021). Malaria. [online] www.who.int. Available at: https://www.who.int/healthtopics/malaria#tab=tab_2 [Accessed 9 Jun. 2021].

World Health Organisation (2019). World Health Organization Model List of Essential Medicines. [online] pp.23–25. Available at: https://apps.who.int/iris/bitstream/handle/10665/325771/WH O-MVP-EMP-IAU-2019.06-eng.pdf [Accessed 9 Jun. 2021].

142 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Antibacterial activity and degradation pathways of methallyl isothiocyanate Ollie Bacon Barker College Methallyl isothiocyanate is an analogue of the well-researched antibacterial agent allyl isothiocyanate, found in plants belonging to the Brassicaceae family. Candidacy of methallyl isothiocyanate as an alternative antibacterial agent was investigated. The antibacterial activity of methallyl isothiocyanate was determined by testing multiple concentrations of the compound against K-12 E. coli achieved through serial dilutions. The compound was found to not display activity at the experimental concentrations. Multiple explanations for this are provided including an insufficient range of concentrations used for testing and structural differences between methallyl isothiocyanate and allyl isothiocyanate. Chemical degradation pathways of methallyl isothiocyanate are proposed based on literature inquiring into the decomposition of allyl isothiocyanate in water. Literature review Between 2001 and 2013, 107 bacterial toxin-mediated outbreaks were confirmed in Australia, affecting 2,219 people with 47 hospitalisations and 13 deaths occurring (May, Polkinghorne & Fearnley, 2016). The true number of outbreaks is likely higher due to underreporting which is common for bacterial infections. From this incident, 48% of outbreaks resulted from commercial food services inadequately controlling the temperature of environments where food was stored and handled (May, Polkinghorne & Fearnley, 2016), making conditions favourable for bacterial growth. Bacterial infections which are contracted by humans from eating or handling contaminated food are treated with the use of antibiotics. Over time, bacterial strains can develop favourable DNA mutations which render currently used antibiotic drugs ineffective, demonstrating the natural process of antibiotic resistance. The misuse of antibiotics through unnecessary or incorrect consumption is causing the rapid development of antibiotic resistance, thus reducing the ability to treat common bacterial infections (World Health Organisation, 2020). This will likely lead to a significant increase in global deaths if new antibacterial agents are not developed. Antibiotic resistance is therefore one of the largest threats to global health and as such, more effective methods of treating bacterial infections must be explored (World Health Organisation, 2020). One cause of bacterial infection is Escherichia coli (E. coli) which presents a significant danger to public health despite certain strains being used in laboratories. This is

due to the wealth of knowledge surrounding this organism, the ability to be easily genetically modified and being considered as biologically safe vehicles for propagation of various genes and vectors (Kuhnert, Nicolet & Frey, 1995). In 2011, Shiga toxin-producing E. coli strains (STEC) were the cause of an outbreak in Germany in which 3,842 cases of human infections were recorded (Beutin & Martin, 2012) with person to person and foodborne transmission being the most effective mode of pathogen spread. Other unrelated STEC outbreaks highlight the ability of a wide variety of wild, domestic and captive animals to act as reservoirs for zoonotic pathogens (Kim, Lee & Kim, 2020).

Figure 1:General chemical structure of isothiocyanate functional group. R represents the glucosinolate side chain which varies between each ITCs.

As technological advances and the trend of mass production for maximised profit continue, human activities which result in the contamination of water sources and animal products will require new methods to mitigate the effects of these pathogens (Kim, Lee & Kim, 2020). One such method will include the development of novel antibacterial agents (Munita & Arias, 2016) in order to combat a future with increased antibiotic resistance.

Science Extension Journal • 143

Scientific Research in School Volume 3 Issue 1 2021

There are many chemical compounds that display antibacterial properties however the commercial potential for each compound’s application depends on a variety of factors such as toxicity to humans, cost, effectiveness and environmental impact (Nowicki et al., 2016). Plausible candidates may be found in the isothiocyanate group (Figure 1). An isothiocyanate (ITC) is a natural plant product found in vegetables belonging to the Brassicaceae family such as horseradish, cauliflower and cabbage (Dufour, Stahl & Baysse, 2015) and are known to protect plants against microbial infection (Bending & Lincoln, 2000). ITCs are produced from the enzyme, myrosinase, acting on plant glucosinolates when the plant tissues are disrupted (Dufour, Stahl & Baysse, 2015) (Figure 2). The result is a family of volatile, electrophilic substances, some of which have been shown to display an inhibitory effect on various pathogenic bacteria (Chacon, Buffo & Holley, 2006). ITCs have also been found to have an additive effect which allows them to be used in conjunction with other antibiotic agents to reduce dosages but maintain the same potency (Dufour, Stahl & Baysse, 2015). This may have the potential to slow down the rate at which a microbial strain develops antibiotic resistance (Dufour, Stahl & Baysse, 2015).

ITCs can be divided into two classes – aliphatic and benzenic – with studies showing that short-chained benzenic ITCs generally display more activity against gram negative and gram positive bacteria than shortchained aliphatic ITCs (Dufour et al., 2012; Nowicki et al., 2016). Despite this, the aliphatic compound allyl isothiocyanate (AITC) (Figure 3) is of particular interest in the scientific community due to its specific side chain structure. This unique structure gives AITC its antibacterial and additive properties (Dufour, Stahl & Baysse, 2015). AITC is effective as an antibacterial agent in both liquid and vapour form (Lin, Preston & Wei, 2000), supporting its current application as a food preservative in Japan provided it comes from a natural source (Nadarajah, Han & Holley, 2005). This is due to the compound being highly volatile, toxic and carcinogenic in its pure extracted form (GHS Classification Guidance by the Japanese Government 2013). It should be noted that unlike its application as a treatment for bacterial infections mentioned previously, the aim of using AITC in food preservation is to kill pathogenic bacteria to prevent the pathogen from entering the body. Furthermore, AITC has a limited shelf life due to the compound’s decomposition at room temperature however, despite these limitations, no superior alternatives have been identified as low concentrations are effective on various pathogenic

Figure 2: Degradation of glucosinolates by myrosinase. Myrosinases catalyse the hydrolysis of glucosinolates to give unstable aglucones and glucose (1). The products that are formed vary depending on the glucosinolate side chain (represented by R) and on the reaction conditions. The products are thiohydroxamate-O-sulfonate (2), epithionitrile (3), thiocyanate (4), nitrile (5), oxazolidine-2-thione (6), and ITC.

144 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

microorganisms and do not detract from the sensory quality of the food (Nadarajah, Han & Holley, 2005). Consequently, this report has selected to investigate the antibacterial potential of the synthetic compound methallyl isothiocyanate (MAITC) as an alternative. MAITC (Figure 4) is an analogue of AITC – differing structurally only by a CH₃ group (Fizer 2013) – and as such may possess similar antibacterial properties to AITC with potentially more favourable chemical properties. It is well known that an allyl side chain, as in AITC, is a reactive structure. Adding a methyl group as in MAITC may mitigate the reactivity of the side chain in comparison, leading to a more stable compound which may possess more favourable properties which make it easier to handle and use as a food preservative. MAITC has also been thought to possess anticarcinogenic properties and shown to have an antiinflammatory response to mast cell-mediated inflammatory reactions, similarly to other isothiocyanates (Han et al., 2012). Although numerous biological properties are associated with this compound, the antibacterial potential of MAITC remains unknown due to the lack of research in this area (Han et al., 2012). A similar investigation was conducted in 2019 which tested the antibacterial potential of another compound, propyl isothiocyanate (Cheng, 2019). The results found this compound – which differed from AITC by a single carbon bond in place of the carbon-carbon double bond – to display no signs of antibacterial activity in the diluted concentrations however the pure form displayed complete inhibition (Cheng, 2019). Degradation pathways were discussed to explain this phenomenon although further investigation is required. Similarly, this report will investigate whether various concentrations of MAITC display antibacterial activity against K-12 E. coli.

Figure 4: Chemical structure of methallyl isothiocyanate

Scientific research question Does methallyl isothiocyanate have comparable antibacterial activity to the activity reported for allyl isothiocyanate against K-12 E. coli.?

Scientific hypothesis That methallyl isothiocyanate will display antibacterial activity against K-12 E. coli, measured by the zone of inhibition.

Methodology Chemicals Methallyl isothiocyanate (MAITC) was purchased from Sigma Aldrich (Sydney, Australia) Bacteria K-12 E. coli and chloramphenicol antibiotic was purchased from Southern Biological (Knoxfield, Victoria, Australia). 9 plates of Mueller Hinton agar were prepared, according to the manufacturer's instructions. Sterilisation A 70% alcohol solution was used to wipe the bench. Metal forceps and glass pipettes were dipped in the 70% alcohol solution before burning the alcohol off in a Bunsen burner. Different sterile glass pipettes were used for transferring each substance and concentration. Three 6mm diameter wells were created pushing the large end of a glass pipette into the agar and removing the remaining circle.

Figure 3: Chemical structure of allyl isothiocyanate

Serial Dilutions A standard solution of MAITC was created by dissolving 0.20 g of MAITC into 40 mL of DMSO and 60 mL of water in a 100 mL volumetric flask. This created a stock solution of concentration 2000 mg/L. Serial dilutions were carried out into 20 mL volumetric flasks by pipetting 1 mL of solution and 19 mL of distilled water into 20 mL volumetric flasks to create solutions with the following concentrations:

Science Extension Journal • 145

Scientific Research in School Volume 3 Issue 1 2021

Table 1: Serial dilutions of MAITC

Solution

Concentration (mg/L)

1 2 3

2000 100 5

Total volume (mL) 100 20 20

Inoculation and plating Each agar plate was divided into 4 equal sections using a marker and labelled with the solution it would contain (Figure 5). The agar plates were relocated to a fume hood containing an ignited Bunsen burner to create an updraft of air, removing contaminants from the work site. The lid of a plate was removed. A well was created in three of the four sections of agar using the large end of a sterile glass pipette and the agar discs were removed to create the well. Using sterile disposable glass pipettes, 2 drops of K-12 E. coli were placed on each plate and spread evenly around the plate using a glass spreader. 1 drop of the 2000mg/L MAITC solution was put into the well labelled MAITC 1 and 1 drop of DMSO (acting as a negative control) was put into their respective wells. Sterile metal forceps were used to transfer the antibiotic tablet (acting as a positive control) onto the plate and the remaining well (acting as a negative control) was left blank. The lid was reapplied. This process was repeated for the remaining two plates with the label MAITC 1.

Figure 5: Diagram of agar plate setup (divided into 4 sections).

The same process was used to prepare three plates containing the 100 mg/L solution of MAITC (labelled MAITC 2), and similarly to prepare the 5 mg/L solution labelled MAITC 3. Each plate was sealed using tape and placed in an incubator at 37℃ for 36 hours.

146 • Science Extension Journal

Determination of inhibitory concentration The annular radius was measured using vernier callipers. If a zone of inhibition was detected, the substance was deemed to be antibacterial (Figure 6). If no zone of inhibition observed, the substance specific concentration of the substance was deemed to be ineffective at inhibiting bacterial growth.

Figure 6: Measurement of annular radius (Source: Bell et al. 2018, pp. 19).

Results A test for each concentration was conducted 3 times (Figure 7). In every case, the chloramphenicol positive control displayed antibacterial activity measured by the zone of inhibition (Table 2). The blank section of agar and DMSO (acting as negative controls) showed no inhibition (Figure 8) on every plate. All plates for each concentration of MAITC tested displayed no antibacterial activity (Table 2) determined by the absence of a zone of inhibition.

Figure 7: All 9 plates containing 3 repetitions of each concentration

Scientific Research in School Volume 3 Issue 1 2021

Table 2: Zone of inhibition of chloramphenicol and different concentrations of MAITC against K-12 E. coli

Concentration of MAITC (mg/L) 2000 100 5

Average zone of inhibition (mm) MAITC Chloramphenicol No 8.7 inhibition No 9.0 inhibition No 8.3 inhibition

Figure 8: Agar plate of each concentration of MAITC

Discussion A lawn of K-12 E. coli was successfully grown on each agar plate. The inhibitory activity demonstrated by the positive control (chloramphenicol) and lack of activity in both the negative controls (blank well and well filled with DMSO) show the results of the different concentrations of MAITC to be valid and reliable. Despite the structural similarities to AITC, the experimental concentrations of MAITC were found to not possess antibacterial activity. It is possible that the concentrations selected did not provide a wide enough range for MAITC to display activity. A highly relevant research paper investigating the potential candidacy for the analogue propyl isothiocyanate as an antibacterial agent displayed identical results in a similar range of concentrations however when tested in its pure, undiluted form, the compound displayed antibacterial properties (Cheng 2019). It could therefore be speculated that using undiluted MAITC would yield a similar outcome. Further research would need to be conducted to determine whether MAITC is antibacterial at higher concentrations, which would be valuable to the scientific literature, however the application of an efficacious dose of MAITC as an antibacterial agent would require a clinically unacceptable quantity compared to existing, more potent antibacterial agents. If a zone of inhibition was observed, an ANOVA test with post-hoc tukey would have been conducted to analyse the difference in the means between concentrations of the compound against K-12 E. coli however, as no zone of inhibition was detected this was not conducted. Previous studies have suggested AITC’s antibacterial activity results from either its ability to alter the structures of essential proteins, attack the active site of enzymes or interfere with bacterial enzymic activity, respiration, metabolism and the transcription of genes (Kawakishi & Kaneko, 1985; 1987; Dufour, Stahl & Baysse, 2015, p.20; Luciano & Holley, 2009). As conflicting proposals for the mechanism of AITC’s antibacterial activity exist, the true mode of action remains unknown. Despite this, one potential explanation for MAITC’s inactivity at the tested concentrations may be the presence of an additional CH₃ group (Fizer, 2013). As previously mentioned, ITCs consist of a functional group (Figure 1) and a side chain which differs between each ITC, giving it specific properties. Chemical reactions can be easily initiated at either the C or N centre of the –N=C=S group (Figure 1) without losing reactive possibilities of the other centre (Fizer, 2013). As this functional group is common to all

Science Extension Journal • 147

Scientific Research in School Volume 3 Issue 1 2021

ITCs, it would suggest that different side chains (i.e. between AITC and MAITC) are likely the cause of difference between the activity of the two compounds. In other words, the allyl side chain, as in AITC, holds partial responsibility for the antibacterial activity of the compound compared to the additional methyl group present in MAITC side chain which may alter the properties of the compound, making this activity is no longer possible. A phenomenon which may provide another explanation for MAITC’s inactivity is the degradation of AITC in water. Research conducted by Olaimat and Holley in 2013 proposed that the bactericidal activity of AITC is strongest at room temperatures as the compound is more stable than at higher temperatures with a 2009 study finding that AITC decomposes rapidly in water at 37℃ (Luciano & Holley 2009). Similarly, degradation pathways may also occur with MAITC when dissolved in water, although further exploration of the decomposition products has yet to be carried out. MAITC is insoluble in water (as is AITC) and was therefore mixed first with DMSO prior to carrying out the serial dilutions. After introducing water to the MAITC-DMSO solution both heat and gas were produced, which may indicate the occurrence of chemical degradation when mixing MAITC with either water (as supported by the literature) or even possibly the solvent DMSO. It is unlikely that the reaction resulted from the mixing of DMSO and water as there is no evidence in the literature to support this. Multiple papers propose different decomposition products of AITC in water (Figure 9) however the majority of them agree the products are not responsible for antibacterial activity. If similar products occur from the degradation of MAITC in water then it could be speculated that the pure, undiluted form of the compound may display antibacterial activity and the serial dilutions carried out in this investigation inhibited this result.

Figure 10: Rearrangement pathway from AITC to allyl

148 • Science Extension Journal

Figure 9: Proposed decomposition products of AITC (Source: Cheng, 2019)

A proposed degradation pathway of AITC to allyl thiocyanate (Figure 10) includes an allylic shift via a circular six membered ring transition state to achieve degradation (Pecháček, Velíšek & Hrabcová, 1997). The importance of the carbon-carbon double bond on the allyl side chain for this could potentially allow MAITC to degrade by following this pathway as this bond is present in both compounds. An alternative degradation pathway can be seen in Figure 11 which involves the formation of dimethallylthiourea from MAITC. This pathway does not rely on the importance of the carbon-carbon double bond and instead involves the addition of an ion to the functional group of the compound. There is a strong case for this pathway as the hydroxide ions would have been present during the serial dilations in which water was combined with the compound.

Scientific Research in School Volume 3 Issue 1 2021

Figure 9: Proposed degradation pathway for formation of dimethallylthiourea from MAITC

Conclusion MAITC was found to not display antibacterial activity at the tested concentrations against K-12 E. coli as determined by a lack of an observable zone of inhibition. These results are supported by all positive and negative controls working as intended on each plate for all concentrations. The findings may be a result of the experimental concentrations being too low and as such further investigation is advised. Another explanation may be the additional CH₃ group to the AITC side chain to form MAITC, which could potentially inhibit the mechanism for AITC’s antibacterial activity. It may also be due to chemical degradation through similar pathways to those proposed for AITC or the proposed pathway for the degradation of MAITC with hydroxide ions to produce dimethallylthiourea. This potential pathway suggests that MAITC may display antibacterial when in its undiluted, pure form however subsequent research is required.

Acknowledgements I wish to show my appreciation to Dr Katie Terrett for her supervisory assistance in generating this research report and Dr Alison Gates for her expertise and assistance in carrying out the practical experiment.

References Bending, G.D. & Lincoln, S.D. 2000, ‘Inhibition of soil nitrifying bacteria communities and their activities by

glucosinolate hydrolysis products’, Soil Biochemistry, vol. 32, no. 8–9, pp. 1261–9.

Biology

and

Beutin, L. & Martin, A. 2012, ‘Outbreak of Shiga Toxin– Producing Escherichia coli (STEC) O104:H4 Infection in Germany Causes a Paradigm Shift with Regard to Human Pathogenicity of STEC Strains’, Journal of Food Protection, vol. 75, no. 2, pp. 408–18. Chacon, P., Buffo, R. & Holley, R. 2006, ‘Inhibitory effects of microencapsulated allyl isothiocyanate (AIT) against Escherichia coli O157:H7 in refrigerated, nitrogen packed, finely chopped beef’, International Journal of Food Microbiology, vol. 107, no. 3, pp. 231–7. Cheng, S. (2019). Antibacterial activity and chemical degradation pathways of propyl isothiocyanate. Barker Science Extension Journal, 1(1). Dufour, V., Alazzam, B., Thepaut, M., Ermel, G. & Baysse, C. 2012, ‘Antimicrobial Activities of Isothiocyanates Against Campylobacter jejuni Isolates’, Frontiers in Cellular and Infection Microbiology, vol. 2, viewed 14 June 2021, <https://www.frontiersin.org/articles/10.3389/fcimb.2012.00 053/full>. Dufour, V., Stahl, M. & Baysse, C. 2015, ‘The antibacterial properties of isothiocyanates’, Microbiology, vol. 161, no. 2, pp. 229–43. Fizer, M. 2013, ‘Methallyl Isothiocyanate’, Synlett, vol. 24, no. 15, pp. 2019–20. Han, N.-R., Kim, I.-K., Kim, H.-M. & Jeong, H.-J. 2012, ‘Methallyl isothiocyanate inhibits the caspase-1 activity through the inhibition of intracellular calcium levels’, Biochimie, vol. 94, no. 3, pp. 816–22. Kawakishi, S., and T. Kaneko. 1985. ‘Interaction of oxidized gluta- thione with allyl isothiocyanate’. Phytochemistry vol. 24, pp. 715–718.

Science Extension Journal • 149

Scientific Research in School Volume 3 Issue 1 2021

Kawakishi, S., and T. Kaneko. 1987. ‘Interaction of proteins with allyl isothiocyanate. J. Agric’. Food Chem. vol. 35, pp. 85–88. Kim, J.-S., Lee, M.-S. & Kim, J.H. 2020, ‘Recent Updates on Outbreaks of Shiga Toxin-Producing Escherichia coli and Its Potential Reservoirs’, Frontiers in Cellular and Infection Microbiology, vol. 10, viewed 14 June 2021, <https://www.frontiersin.org/articles/10.3389/fcimb.2020.00 273/full>. Kuhnert, P., Nicolet, J. & Frey, J. 1995, ‘Rapid and accurate identification of Escherichia coli K-12 strains.’, Applied and environmental microbiology, vol. 61, no. 11, pp. 4135–9. Lin, C.-M., Preston, J.F., III & Wei, C.-I. 2000, ‘Antibacterial Mechanism of Allyl Isothiocyanate†’, Journal of Food Protection, vol. 63, no. 6, pp. 727–34. Luciano, F.B. & Holley, R.A. 2009, ‘Enzymatic inhibition by allyl isothiocyanate and factors affecting its antimicrobial action against Escherichia coli O157:H7’, International Journal of Food Microbiology, vol. 131, no. 2, pp. 240–5. May, F.J., Polkinghorne, B.G. & Fearnley, E.J. 2016, Epidemiology of bacterial toxin-mediated foodborne gastroenteritis outbreaks in Australia, 2001 to 2013, vol. 40, no. 4, p. 10. Ministry of Health, Labour and Welfare, Ministry of the Environment 2013, ‘GHS Classification Result Table’ GHS Classification Guidance Japanese Government, Japan. Munita, J.M. and Arias, C.A., 2016. ‘Mechanisms of antibiotic resistance’. Microbiology spectrum. vol 4. no. 2 Nadarajah, D., Han, J.H. & Holley, R.A. 2005, ‘Use of mustard flour to inactivate Escherichia coli O157:H7 in ground beef under nitrogen flushed packaging’, International Journal of Food Microbiology, vol. 99, no. 3, pp. 257–67. Nowicki, D., Rodzik, O., Herman-Antosiewicz, A. & Szalewska-Pałasz, A. 2016, ‘Isothiocyanates as effective agents against enterohemorrhagic Escherichia coli : insight to the mode of action’, Scientific Reports, vol. 6, no. 1, p. 22263. Olaimat, A. and Holley, R. 2013, ‘Effects of changes in pH and temperature on the inhibition of Salmonella and Listeria monocytogenes by Allyl isothiocyanate’. Food Control, vol. 34, pp. 414-419. Pecháček, R., Velíšek, J. & Hrabcová, H. 1997, ‘Decomposition Products of Allyl Isothiocyanate in Aqueous Solutions’, Journal of Agricultural and Food Chemistry, vol. 45, no. 12, pp. 4584–8. World Health Organization (2020). Antibiotic resistance. [online] World Health Organization. Available at: https://www.who.int/news-room/fact-sheets/detail/antibioticresistance.

150 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

Optimising the synthesis of the 4-iodo analogue of pyrimethamine Maxine Wu Barker College The emergence and rapid spread of drug resistant P. falciparum, the most prevalent malaria pathogen, has threatened to curtail the efficacy and therapeutic lifetime of antifolate drug treatments like pyrimethamine, due to point mutations in the dihydrofolate reductase (DHFR) enzyme of the parasite. Currently, the burden of malaria is highest in developing counties, and the deteriorating efficacy of existing anti-malarial drugs have increased the cost and complexity of routine prophylaxis and treatment. The present work proposes to overcome such resistance, using a simple rational drug design strategy which focuses on changing substituents on the pyrimethamine ring to synthesise new pyrimethamine analogues as potential drug leads. This report discusses methods used to successfully optimise the synthesis of the 4-iodopyrimethamine analogue and its possible future applications in cross-coupling reactions. Literature Review Malaria is a protozoan disease that continues to present major public health concerns for most endemic areas of the world (Weiss et al., 2019). Once infected, malaria attacks the body’s red blood cells, and if not treated within 24 hours, severe complications often lead to death (WHO, 2020). The burden of malaria is highest in developing countries, affecting vulnerable and marginalised populations – particularly children, pregnant women, migrants and refugees (Murphy, 2006). The optimisation of access to malaria intervention is essential for achieving universal health coverage and promoting well-being of all ages, having the potential to alleviate poverty, improve equity, hence contributing to overall socio-economic development in affected areas.

Globally, there were 229 million cases of malaria in 2019, with an estimated death toll of 409,000 (WHO, 2020). Of the five Plasmodium parasite species, Plasmodium falciparum is the most virulent causative agent, and in 2018, accounted for 99.7% of cases in the African region, 71% in the Eastern Mediterranean, 65% in the Western Pacific and 50% in the South-East Asian region (Figure 1). Since 2000, the scale-up of malaria control interventions has fuelled bold aims for disease eradication, having significantly reduced global morbidity and mortality (Weiss et al., 2019). This was driven by several factors, in particular, increased funding, effective vector control, and improved case reporting and surveillance (Cotter, Sturrock & Hsiang, 2013). Although remarkable

Figure 1: Spatial distribution of P. falciparum incidence (Source: Weiss et al., 2017) Science Extension Journal • 151

Scientific Research in School Volume 3 Issue 1 2021

Figure 2: Graphs displaying the trends in a) malaria case incidence rate between 2000 and 2019 and b) mortality rate between 2000 and 2019 (Source: WHO, 2020)

progress has been made, the rate of reduction in both mortality and morbidity has slowed dramatically (Figure 2) (WHO, 2020). Currently, malaria parasites have developed resistance to almost every anti-malarial drug available (Dhiman, 2019). Initially, these agents were highly effective, but due to drug pressure, the selection of resistant parasites has allowed for the continued proliferation of malaria (White, 2004). The subsequent propagation of resistant strains by local transmission and migration of parasite reservoirs are key factors contributing to the increased cost and complexity of achieving parasitological cure (Wernsdorfer, 1994). Accordingly, since present malaria control interventions are unlikely to address these epidemiological changes, the systematic exploration of novel strategies and methods are urgently needed. The antimalarial drug pyrimethamine (1, Figure 3) is a specific derivative of 2,4- diaminopyrimidine and is currently sold under the trade name Daraprim

(Tse, Korsik & Todd 2019). Since its development in 1952, pyrimethamine has been most widely used as an anti-malarial drug. However, the antiprotozoal agent has also been used in the treatment of other protozoan diseases such as toxoplasmosis and trypanosomiasis caused by Toxoplasma gondii and Trypanosoma brucei, respectively (Adane & Bharatam, 2008). In addition, pyrimethamine has also been used to treat Pneumocystis carinii pneumonia that often affects immune-compromised patients (Cirioni et al., 2000).

Figure 3: Structure of pyrimethamine (1)

Drugs used in malaria treatment target specific biochemical processes vital to the growth of the disease causing parasite (Warhurst, 2002). The

Figure 4: Bio-synthetic pathway of DHFR catalysed hydrogenation of dihydrofolate (After: Sardarian et al., 2003) 152 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

mechanistic function of pyrimethamine owes its effectiveness to its structural similarities with one of the domains of the bifunctional dihydrofolate reductase-thymidylate synthase (DHFR-TS) of P. falciparum, namely, the dihydrofolate reductase (DHFR) enzyme (Sirawaraporn et al., 1997). Due to its ability to inhibit the DHFR enzyme domain in the folate bio-synthetic pathway (Figure 4), pyrimethamine prevents the catalysation of the NADPH-dependent reduction of dihydrofolate to tetrahydrofolate. Consequently, this affects the biosynthesis of purines and several amino acids as well as cofactors imperative for DNA synthesis and cell multiplication, and hence, pyrimethamine prevents further proliferation of the plasmodia (Peterson, Walliker & Wellems, 1988). Unfortunately, the efficacy of pyrimethamine over the years has been blunted by the rapid spread of resistance that has enervated scientific efforts (Bloland 2001). Knowledge of the molecular structure of the DHFR enzyme and its interactions with inhibitors and substrates, as well as how mutations can affect these functions are imperative for rational drug design. Recently, solved crystal structures of the wild-type and mutant P. falciparum DHFR complexed with pyrimethamine have yielded insights into the mechanisms of resistance resulting from mutations (Figure 5A) (Yuthavong et al., 2005). Although the connectivity of the P. falciparum DHFR-TS domains (Figure 5A, denoted in red and blue respectively) are still relatively unknown, the structure of this enzyme-inhibitor complex confirms that several amino acid residues at the active site are engaged in hydrogen bonding with dihydrofolate, and highlights other interactions which holds the substrate molecule in an orientation that allows for inhibition of the enzyme function (Figure 5B). However, mutations in the side chains as well as in the main chain configuration at the active site have created a steric constraint which is detrimental to the binding of the rigid parachlorophenyl group in pyrimethamine (Yuthavong et al., 2005). This has caused the subsequent dislocation of the phenyl ring, and significantly reduced the binding affinity for pyrimethamine (Lozovsky et al., 2009).

Figure 5: A) Crystal structure for P. falciparum DHFR-TS (Different shades denote individual sub-units. Grey dashed curves represent possible linkages based on intermolecular space in crystal packing. Junction between DHFR (blue) and TS (red) are drawn as dark green. Ligands bound in active sites are coloured as follows: dUMP in magenta and NADPH in cyan). B) Enzyme-inhibitor interactions at the active site of the wild-type DHFR. Four mutated amino acid residues are denoted with red labels. C) Diagram of interactions, four residues responsible for resistance are bolded and underlined. (Source: Yuthavong et al., 2005)

Science Extension Journal • 153

Scientific Research in School Volume 3 Issue 1 2021

Furthermore, the Serine residue 108 (S108N) has also been found to be a crucial point mutation found in all pyrimethamine resistant strains (Sirawaraporn et al., 1997). Cumulative effects on inhibitor binding are seen with ancillary mutations at residue Asparagine 51 (N51L), Cysteine 59 (C59R) and Isoleucine 164 (1164L) which have caused a significant reduction in the binding affinity for pyrimethamine (Figure 5C) (Yuthavong et al., 2012). Such mutation-based resistance poses questions about research into designing new and potent antifolate drugs with improved affinity against wild-type and mutant DHFR as a means to minimise the burden of P. falciparum malaria (Adane & Bharatam, 2008). One approach to overcoming drug resistance involves the development of new analogues where small structural changes to the chemical structure of the compound can significantly improve drug efficacy (Yuthavong, et al., 2005). Although crossresistance may be expected for inhibitors with similar structures, given that the DHFR enzyme needs to remain functional, it is likely that there are limitations to the number of mutational combinations that can occur in the DHFR domain (Tarnchompoo et al., 2018). Thus, this approach assumes that suitable structural modifications can be exploited. This research aims to optimise the synthesis of a novel 4-iodopyrimethamine analogue (2) that was completed by a Barker College student in 2020 (Wong, 2020) (Figure 6B). The iodine analogue (2) is of interest due to the larger atomic radius and lower electronegativity compared to the chlorine atom in pyrimethamine (1). By replacing the chlorine atom with an iodine atom at the R4 position (Figure 6A), it is hypothesised that this could potentially enhance binding interactions at the DHFR enzyme site, or at least provide interesting structure-activity information. Additionally, given the significant effort into the synthesis of other various pyrimethamine analogues, the collection of inhibition constants (KI and IC50 values) acquired via biological testing against both mutant and wildtype P. falciparum from previous studies can be used quantitively to compare and assess the drug efficacy and inhibition potency of the novel 4iodopyrimethamine analogue (Nattee et al., 2017).

154 • Science Extension Journal

Figure 6: A) General structure of pyrimethamine analogues. B) Proposed 4-iodopyrimethamine analogue (2)

Furthermore, the carbon-iodine bond can be used as a “synthetic handle” for metal catalysed crosscoupling reactions, creating a pathway for the synthesis of more complex analogues with increased binding affinity (Mao et al., 2009). In the current literature, efforts have been made to exploit the Suzuki Miyaura cross-coupling procedure, and halogenated pyrimidines with boronic acids have been used as a common approach for the preparation of a diverse set of substituted pyrimidines (Richardson & Stevens, 2002). Since the carbon-iodine bond at the R4 position can be easily functionalised, the optimisation of the synthesis of the 4-iodopyrimethamine analogue can create a starting point for a library of potential antimalarial drugs in the future. Previously, the successful synthesis of compound 2 was conducted following the 2016 synthetic pathway developed by Sydney Grammar School (SGS, 2016) for the synthesis of pyrimethamine (1) (Figure 7) (Wong, 2020). However, due to issues with synthetic yields and product purity, the final mass of 4-iodopyrimethamine was insufficient to conduct biological testing against P. falciparum. The focus of this report aims to resolve these issues and optimise yields such that biological data on the efficacy of 4-iodopyrimethamine in combating malarial pathogens can be conducted.

Scientific Research in School Volume 3 Issue 1 2021

Step 1: Synthesis of 2-(4-iodophenyl)-3oxopentanenitrile

Figure 7: Pathway for the synthesis of 4-iodopyrimethamine analogue (2) with the yields from the previous synthesis

Scientific research question How can the synthesis of 4-iodopyrimethamine (2) be optimized to improve yields and purity to allow for biological testing?

Scientific hypothesis That an optimized synthesis of 4iodopyrimethamine (2) can be successfully undertaken to produce enough material for biological testing.

Methodology General experiment details 1

H spectra were recorded at 300 K using a Bruker Avance DRX400 NMR spectrometer in deuterated solvents. Residual acetone (δ 2.05) and chloroform (δ 7.26) were used as internal reference for 1H NMR spectra. The data is reported as chemical shift (δH ppm), relative integral, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet) and assignment. Atom labels on structures are to illustrate 1H NMR spectral assignments and do not necessarily correspond to the IUPAC names given. Analytical thin layer chromatography was performed with Merck Kieselgel 60 F254 (0.2 mm) pre-coated aluminium sheets, and visualisation was achieved by inspection under UV light. Throughout the reaction process Thin Layer Chromatography (TLC) was conducted to gauge the progress of the reaction and determine the point of completion. TLC analysis was conducted with either 50:50 Dichloromethane (DCM) : Hexane, or pure DCM.

Figure 8: 2-(4-iodophenyl)-3-oxopentanenitrile (4)

4-iodophenylacetonitrile (10.00 g, 0.046 mol, 1 equiv.), ethyl propionate (5.00 g, 0.049 mol, 1.05 equiv.) and potassium tert-butoxide (10.47 g, 0.093 mol, 2 equiv.) were combined in THF (100 mL) at room temperature, with stirring in a round bottom flask. The reaction mixture turned to a dark red and heated up rapidly. The reaction was sealed and stirred for 2 hours. The reaction mixture was worked up by the addition of 1.0 M HCl (100 mL) to the reaction vessel. The acidified reaction mixture was transferred to a separating funnel and the aqueous layer was extracted with DCM (3 x 65 mL). The combined organic layer was washed with brine (100mL), dried with anhydrous sodium sulfate, filtered, and concentrated in vacuo to afford 2-(4-iodophenyl)-3oxopentanenitrile (4) (12.699 g, 0.042 mol, 91%) as a reddish oil. TLC was conducted with 100% DCM as the eluent. The crude 2-(4-iodophenyl)-3oxopentanenitrile (4) was used without purification in the second step of the synthesis. Step 2: Synthesis of 2-(4-iodophenyl)-3-(s2methylpropoxy)-pent-2-enenitrile

Figure 9: 2-(4-iodophenyl)-3-(s2-methylpropoxy)-pent-2enenitrile (5)

Science Extension Journal • 155

Scientific Research in School Volume 3 Issue 1 2021

2-(4-iodophenyl)-3-oxopentanenitrile (12.699 g, 0.042 mol, 1 equiv.) was dissolved in a mixture of toluene (65.0 mL) and 2-methylpropan-1-ol (6.50 mL, 0.21 mol). 18M H2SO4 (1.00 mL) was added, and the mixture was refluxed for 10 hours in a Dean Stark apparatus. The reaction mixture was poured onto a saturated sodium hydrogen carbonate in a separating funnel and the aqueous phase was extracted with DCM (2 x 50 mL). The combined organic extracts were dried over anhydrous sodium sulfate. Addition of 5.0 mL of triethylamine to the reaction mixture converted the unreacted starting material to its very polar triethylammonium enolate salt. Chromatography silica (50.0 g) was added to the organic phase, which was made up to 200mL with dichloromethane and stirred for 1.5 hours. The organic phase was filtered using vacuum filtration and rinsed with 1M HCL (2 x 50 mL) and deionised water (50 mL) in a separating funnel to remove all traces of triethylamine. The combined organic extracts were dried over anhydrous sodium sulfate and filtered. The solvent was removed in vacuo to yield 2-(4-iodophenyl)-3-(2-methylpropoxy)-pent2-enenitrile (5) (11.36 g, 0.032 mol, 75%) as a red oil. This product was used in the next step of synthesis.

Figure 10: 4-iodopyrimethamine analogue (2)

Step 3: Synthesis of 4-iodopyrimethamine analogue

2-(iodophenyl)-3-(2-methylpropoxy)-pent-2enenitrile (11.36 g, 0.0320 mol, 1 equiv.) was dissolved in DMSO (90.0 mL). Guanidine hydrochloride (6.40 g, 0.0640 mol, 2 equiv.) was stirred into the solution followed by sodium methoxide powder (4.00 g, 0.0704 mol, 2.2 equiv.). The solution became dark red in colour on addition of the sodium methoxide, which dissolved into the solution within an hour. No precipitation of sodium chloride was observed. The solution was allowed to stand at room temperature for 48 hours. Crystals appeared in the reaction mixture. These were filtered and isolated for analysis. The remaining 156 • Science Extension Journal

reaction mixture was poured onto water and extracted with DCM. 100% ethanol was added to the organic layer and left for 48 hours. No additional crystallisation occurred.

Scientific Research in School Volume 3 Issue 1 2021

Results Step 1: Synthesis oxopentanetitrile (4)

2-(4-iodophenyl-3-

Figure 11: 1H NMR spectra of crude product 4 1H

NMR (500 MHz, chloroform-d): δ 7.77-7.73 (2H, d, H1), 7.12-7.07 (2H, d, H2), 4.61 (1H, s, H3), 2.69-2.64 (2H, m, CH2), 1.06 (3H, t, CH3)

Figure 12: TLC of step 1 after reaction was stirred for 2hrs

Figure 13: Previous report’s TLC of step 1

Science Extension Journal • 157

Scientific Research in School Volume 3 Issue 1 2021

Figure 14: 1H NMR of crude product 5. 1

H NMR (500 MHz, chloroform-d): δ 7.70 (2H, d, H1), 7.21 (2H, d, H2), 3.85 (2H, d, CH2), 2.78 (2H, q, CH2), 2.02-1.91 (1H, m, CH), 1.30 (3H, t, CH3), 0.93 (6H, d, CH3) Step 2: Synthesis of 2-(4-iodophenyl)-3-(2-methylpropoxy)-pent-2-enenitrile (5)

Figure 13: Previous report’s TLC of step 1

Figure 15: Previous report’s 1H NMR experiment after step 2. Hydrogen environments cannot be assigned due to complexity

158 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

yield obtained for this step was 91% which is significant increase from the previous report’s 65% yield and is likely a result of optimising the reaction (Wong, 2020). By performing a scaled-up synthesis with 11.33g of starting material, a greater amount than the 7g used previously, loss of the target compound 4 was minimised during extraction and separation from aqueous impurities.

Figure 16: TLC after work up in step 2

Figure 19: Mechanism of the reaction between iodophenylacetonitrile and ethyl propionate

Figure 17: Previous synthesis’ TLC after work up in step 2

Discussion Step 1: Synthesis of 2-(4-iodophenyl)-3oxopentanenitrile (4)

Figure 18: Reaction of 4-iodophenylacetonitrile (3) to form 2(4-iodophenyl)-3-oxopentanenitrile (4)

Step 1 of the synthesis was performed to afford Compound 4 from a condensation reaction between 4-iodoacetonitrile (3) and ethyl propionate. More specifically, addition of the strong base potassium tert-butoxide caused deprotonation at the CH2 group of Compound 3, and reaction with ethyl propionate led to the elimination of ethanol (Figure 19). The

Additionally, filtration of compound 4 from sodium sulfate after drying was optimised by using a sintered funnel rather than filter paper, as was used previously. These considerations attributed to a yield percentage increase of 40%. The appearance of polar baseline material on the TLC (Figure 12) corresponded to the formation of the more polar Compound 4. In comparison to the TLC plate from the previous report, this TLC is cleaner (Figure 12), indicating that the compound has a higher purity than the synthesis prior (Figure 13). The 1HNMR spectrum is highly complex, likely due to the combination of enol and keto tautomers in equilibrium (Figure 20). The doublet signal between 7.77-7.73 ppm and 7.12-7.07 ppm were assigned to the H1 and H2 aromatic protons, respectively. This was expected since hydrogen atoms on a benzene ring characteristically appear in the 7-8 ppm region. The H3 singlet was assigned to the signal further upfield at 4.61 ppm, due to the electron withdrawing nature of the nitrile group and carbon-oxygen double bond causing a deshielding of the hydrogen atom in this environment. The quartet signal between 2.69-2.64 ppm was assigned to the CH2 on the ethyl group. This is characteristically a quartet but due to possible leftover impurities or the appearance of enol tautomer, the spectra has become more complicated. The distinctive triplet splitting pattern

Science Extension Journal • 159

Scientific Research in School Volume 3 Issue 1 2021

at 1.06 ppm indicated the presence of the CH3 on the ethyl group.

Figure 20: Mechanism of reaction between enol and keto tautomers of compound 4 in equilibrium

The information provided by the 1HNMR was able to confirm the formation of Compound 4. Although comparisons cannot be made with the 1HNMR analysis of the previous compound since it was conducted with a different solvent (acetone-d6, (CD3)2CO), there was enough evidence to conclude that the optimisation of step 1 was achieved. Future trials can focus on obtaining a cleaner sample by undertaking further purification of the crude product, such that complete characterisation of this compound can be undertaken. Step 2: Synthesis of 2-(4-iodophenyl)-3-(2methylpropoxy)-pent-2-enenitrile (5)

Figure 21: Reaction of 2-(4-iodophenyl)-3-oxopentanenitrile (4) to form 2-(4-iodophenyl)-3-(2-methylpropoxy)-pent-2enenitrile (5)

Step 2 of the synthesis was performed under reflux whereby a substitution reaction led to the formation of Compound 5 from Compound 4. A Dean Stark apparatus was used to remove water and drive the equilibrium reaction in the forward direction. To remove the remaining starting material, triethylamine was added to convert it into triethylammonium enolate salt. Similar to the previous synthesis, the addition of silica gel at this stage facilitated the removal of the salt from the reaction mixture. However, after the work up, the previous TLC (Figure 17) indicated that there were still remnants of the starting compound leftover, most likely due to inefficient separation of the silica 160 • Science Extension Journal

gel via decanting, which reappeared in the separating funnel. The re-protonation of the triethylammonium enolate salt stuck to the silica gel occurred after addition of HCl, reforming the enol tautomer of Compound 4 (Figure 22). To optimise this step, the sintered funnel was again incorporated as an alternative to separate the silica gel from the crude product mixture, and the resultant purity was evident in the TLC which showed that the starting material reaction spot was less apparent than the previous attempt. (Figure 16). Furthermore, the 1 HNMR spectra of this compound appeared to be of a higher purity than the previous synthesis. Thus, by optimising the reaction, a yield of 75% was obtained from this step, which is a 32% increase from the previous yield of 57%.

Figure 22: Mechanism of the reaction between iodophenylacetonitrile and ethyl propionate

In the previous synthesis, impurities in the 1H NMR made the spectrum difficult to analyse and as a result, 13C NMR was used to verify formation of product. However, the 1H NMR for this optimised procedure was able to confirm the successful synthesis of Compound 5. The doublet signal at 7.70-7.21 ppm were assigned to the aromatic H1 and H2 hydrogen environments which were previously seen in step 1. The presence of two new doublet signals further downfield at 3.85 ppm were assigned to the H4 hydrogen protons due to their position next to the highly electronegative oxygen atom which can cause a deshielding of the hydrogen in that environment. Compound 5 can also be confirmed by the appearance of the new multiplet signal from 2.02-1.91 ppm which was assigned to the H5 hydrogen environment. This is because it has many neighbouring protons which can cause the signal to split multiple times. The characteristic quartet splitting pattern at 2.78 ppm corresponded to CH2 group from the ethyl group. Lastly, the distinctive triplet splitting pattern at 1.30 ppm and the doublet splitting pattern at 0.92 indicate the presence of three CH3 groups, thus confirming the successful synthesis of Compound 5.

Scientific Research in School Volume 3 Issue 1 2021

Step 3: Synthesis of 4-iodopyrimethamine analogue (2)

Figure 23: Reaction of 2-(4-iodophenyl)-3-(2-methylpropoxy)pent-2-enenitrile (5) to form 4-iodopyrimethamine (2)

Step 3 of synthesis was initiated by the deprotonation of the guanidine ion by sodium methoxide. Following this, guanidine reacted at the nitrile carbon of Compound 5, and subsequent electron rearrangement and the elimination of 2methylpropanol afforded the pyrimidine ring seen in Compound 2. Unfortunately, the optimisation of step 3 was not effective due to time constraints which only permitted one experimental attempt at isolation and purification. Although there was a higher amount of solid material obtained, 1H NMR results were not conclusive due to the problems encountered with dissolving the compound in chloroform-d. Given that the previous synthesis performed the 1H NMR experiment in this solvent, the compound crystallised in this step may not have been the desired 4-iodopyrimethamine analogue. Another possibility could be that the compound contained too many impurities, making it insoluble. Instead, MeOD solvent was used to run the experiment, but no signals were observed from the 1H NMR spectra. Thus, the identification of the solid is still awaiting structural confirmation. Although different solvents were used in combination to crystallise the product, like the previous synthesis, the same difficulties were encountered during the third stage of the synthesis which could potentially be a result of the 4-iodopyrimethamine analogue (2) having a higher degree of solubility compared to the original pyrimethamine compound (1). Due to these issues, the pyrimethamine analogue will not be able to undergo biological testing, but future efforts could

focus on finding a suitable solvent to precipitate out the desired compound such that a sufficient amount can be produced. Overall, the optimisation of the first two steps of the synthetic pathway for 4-iodopyrimethamine (2) was successful. Future considerations such as using column chromatography for further purification at the end of each step could be carried out such that a complete characterisation and analysis of 4iodopyrimethamine can be conducted. Future Research

Future possibilities of the 4-iodopyrimethamine analogue include applications in the preparation of a diverse set of other analogues via substitution at the R4 position. Previously, Suzuki coupling reactions have been able to successfully effect the cross-coupling of pyrimidines, allowing the construction of carbon-carbon bonds between functionalised substrates that were previously inaccessible or required multi-step procedures (Biajoli et al., 2014). In one pilot experiment, a successful reaction scheme utilising a Sandmeyer reaction afforded an iodo-pyrimethamine analogue that was then cross-coupled with 4methoxybenzeneboronic acid to furnish a substituted biphenyl at the R3 position of the analogue (Richardson & Stevens 2002) (Figure 24). For the 4-iodopyrimethamine analogue, although the iodine is in the R4 rather than R3 position (Figure 24), the highly reactive electrophile can still act as an effective additive for accelerating these couplings with satisfactory results (Figure 25) (Dobrounig, Trobe & Breinbauer 2017). Thus, even if the potency and inhibitory activity of the 4iodopyrimethamine analogue isn’t particularly effective, given that the optimised synthetic pathway is much simpler than current analogues reported in the literature, it can function as a precursor for the future synthesis of more complex structures and molecules that can effectively inhibit the DHFR enzyme substrate.

Figure 24: A successful Suzuki-cross coupling application (After: Richardson & Stevens 2002) Science Extension Journal • 161

Scientific Research in School Volume 3 Issue 1 2021

Figure 25: Proposed possible future applications of 4-iodopyrimethamine analogue via. Suzuki-Miyaura pathway

Conclusion

References

This report details the successful optimisation of the first two steps of a synthetic pathway developed previously for the synthesis of the 4iodopyrimethamine analogue (2). The results of the optimisation of step three cannot be confirmed due to difficulties encountered with dissolving the solid obtained at the conclusion of step 3, hence the product is still waiting for 1H NMR analysis so that structural confirmation can be made.

Adane, L. & Bharatam, P.V. 2008, ‘Modelling and informatics in the analysis of P. falciparum DHFR enzyme inhibitors.’, Current medicinal chemistry.

In this experiment, thin layer chromatography (TLC) was carried out to gauge the progress of each reaction, and structural conformation was undertaken by using 1H NMR data to confirm that the desired product was produced. Further comparisons with the previous synthesis’ TLC and 1 H NMR spectra allowed for the efficacy of the optimisation to be confirmed. Significant quantities have been made, and thus, future research can focus on the purification of each reaction product to obtain full characterisation. Additionally, after successful optimisation of the final step of the synthesis, cross-coupling reactions can be investigated for their potential as novel synthetic pathways towards the synthesis of new pyrimethamine analogues in the future.

Acknowledgements I wish to thank Dr Katie Terrett for her guidance throughout the entire project. In particular, her extensive knowledge and expert advice, supervision of the synthesis and for overseeing the completion of this report proved to be invaluable. I would also like to acknowledge the collaborators of Breaking Good who assisted with providing the 1H NMR experimental data which was greatly appreciated.

Biajoli, A.F.P., Schwalm, C.S., Limberger, J., Claudino, T.S. & Monteiro, A.L. 2014, ‘Recent Progress in the Use of PdCatalyzed C-C Cross-Coupling Reactions in the Synthesis of Pharmaceutical Compounds’, Journal of the Brazilian Chemical Society, viewed 20 February 2021, <http://www.gnresearch.org/doi/10.5935/01035053.20140255>. Bloland, P.B. 2001, ‘Drug resistance in malaria’, World Health Organisation, p. 32. Cirioni, O., Giacometti, A., Barchiesi, F. & Scalise, G. 2000, ‘Inhibition of growth of Pneumocystis carinii by lactoferrins alone and in combination with pyrimethamine, clarithromycin and minocycline’, Journal of Antimicrobial Chemotherapy, vol. 46, no. 4, pp. 577–82. Cotter, C., Sturrock, H. & Hsiang, M. 2013, ‘The changing epidemiology of malaria elimination: new strategies for new challenges (vol 382, pg 900, 2013)’, The Lancet, vol. 382, pp. 858–858. Dhiman, S. 2019, ‘Are malaria elimination efforts on right track? An analysis of gains achieved and challenges ahead’, Infectious Diseases of Poverty, vol. 8, no. 1, p. 14. Dobrounig, P., Trobe, M. & Breinbauer, R. 2017, ‘Sequential and iterative Pd-catalyzed cross-coupling reactions in organic synthesis’, Monatshefte Fur Chemie, vol. 148, no. 1, pp. 3–35. Lozovsky, E.R., Chookajorn, T., Brown, K.M., Imwong, M., Shaw, P.J., Kamchonwongpaisan, S., Neafsey, D.E., Weinreich, D.M. & Hartl, D.L. 2009, ‘Stepwise acquisition of pyrimethamine resistance in the malaria parasite’, Proceedings of the National Academy of Sciences, vol. 106, no. 29, pp. 12025–30. Mao, J., Hua, Q., Xie, G., Guo, J., Yao, Z., Shi, D. & Ji, S. 2009, ‘Iodine-Catalyzed Suzuki-Miyaura Coupling Performed in Air’, Advanced Synthesis & Catalysis, vol. 351, pp. 635–41. Murphy, S.C. 2006, ‘Malaria and Global Infectious Diseases: Why Should We Care?’, AMA Journal of Ethics, vol. 8, no. 4, pp. 245–50. Nattee, C., Khamsemanan, N., Lawtrakul, L., Toochinda, P. & Hannongbua, S. 2017, ‘A novel prediction approach for antimalarial activities of Trimethoprim, Pyrimethamine, and Cycloguanil analogues using extremely randomized trees’, Journal of Molecular Graphics & Modelling, vol. 71, pp. 13– 27. Peterson, D.S., Walliker, D. & Wellems, T.E. 1988, ‘Evidence that a point mutation in dihydrofolate reductase-thymidylate

162 • Science Extension Journal

Scientific Research in School Volume 3 Issue 1 2021

synthase confers resistance to pyrimethamine in falciparum malaria’, Proceedings of the National Academy of Sciences, vol. 85, no. 23, pp. 9114–8.

Wong, K. 2020, ‘Synthesis of a pyrimethamine analogue’, Barker College Science Extension Journal, vol. 2, no. Number 1, pp. 47–60.

Richardson, M.L. & Stevens, M.F.G. 2002, ‘Structural Studies on Bioactive Compounds. Part 37.1 Suzuki Coupling of Diaminopyrimidines: A New Synthesis of the Antimalarial Drug Pyrimethamine’, Journal of Chemical Research, vol. 2002, no. 10, pp. 482–4.

Yuthavong, Y., Tarnchompoo, B., Vilaivan, T., Chitnumsub, P., Kamchonwongpaisan, S., Charman, S.A., McLennan, D.N., White, K.L., Vivas, L., Bongard, E., Thongphanchang, C., Taweechai, S., Vanichtanankul, J., Rattanajak, R., Arwon, U., Fantauzzi, P., Yuvaniyama, J., Charman, W.N. & Matthews, D. 2012, ‘Malarial dihydrofolate reductase as a paradigm for drug development against a resistance-compromised target’, Proceedings of the National Academy of Sciences, vol. 109, no. 42, pp. 16823–8.

Sardarian, A., Douglas, K.T., Read, M., Sims, P.F.G., Hyde, J.E., Chitnumsub, P., Sirawaraporn, R. & Sirawaraporn, W. 2003, ‘Pyrimethamine analogs as strong inhibitors of double and quadruple mutants of dihydrofolate reductase in human malaria parasites’, Organic & Biomolecular Chemistry, vol. 1, no. 6, pp. 960–4. Sirawaraporn, W., Sathitkul, T., Sirawaraporn, R., Yuthavong, Y. & Santi, D.V. 1997, ‘Antifolate-resistant mutants of Plasmodium falciparum dihydrofolate reductase’, Proceedings of the National Academy of Sciences, vol. 94, no. 4, pp. 1124– 9.

Yuthavong, Y., Yuvaniyama, J., Chitnumsub, P., Vanichtanankul, J., Chusacultanachai, S., Tarnchompoo, B., Vilaivan, T. & Kamchonwongpaisan, S. 2005, ‘Malarial (Plasmodium falciparum) dihydrofolate reductase-thymidylate synthase: structural basis for antifolate resistance and development of effective inhibitors’, Parasitology, vol. 130, no. Pt 3, pp. 249–59.

Sydney Grammar School Synthesis 2016, ‘Daraprim Synthesis’, Pyrimethamine Synthesis: Status at the end of 2016, viewed 20 February 2021, <https://malaria.ourexperiment.org/daraprim_synthesis>. Tarnchompoo, B., Chitnumsub, P., Jaruwat, A., Shaw, P.J., Vanichtanankul, J., Poen, S., Rattanajak, R., Wongsombat, C., Tonsomboon, A., Decharuangsilp, S., Anukunwithaya, T., Arwon, U., Kamchonwongpaisan, S. & Yuthavong, Y. 2018, ‘Hybrid Inhibitors of Malarial Dihydrofolate Reductase with Dual Binding Modes That Can Forestall Resistance’, ACS Medicinal Chemistry Letters, vol. 9, no. 12, pp. 1235–40. Tse, E.G., Korsik, M. & Todd, M.H. 2019, ‘The past, present and future of anti-malarial medicines’, Malaria Journal, vol. 18, no. 1, p. 93. Warhurst, D.C. 2002, ‘Resistance to Antifolates in Plasmodium Falciparum, the Causative Agent of Tropical Malaria’, Science Progress, vol. 85, no. 1, pp. 89–111. Weiss, D.J., Lucas, T.C.D., Nguyen, M., Nandi, A.K., Bisanzio, D., Battle, K.E., Cameron, E., Twohig, K.A., Pfeffer, D.A., Rozier, J.A., Gibson, H.S., Rao, P.C., Casey, D., Bertozzi-Villa, A., Collins, E.L., Dalrymple, U., Gray, N., Harris, J.R., Howes, R.E., Kang, S.Y., Keddie, S.H., May, D., Rumisha, S., Thorn, M.P., Barber, R., Fullman, N., Huynh, C.K., Kulikoff, X., Kutz, M.J., Lopez, A.D., Mokdad, A.H., Naghavi, M., Nguyen, G., Shackelford, K.A., Vos, T., Wang, H., Smith, D.L., Lim, S.S., Murray, C.J.L., Bhatt, S., Hay, S.I. & Gething, P.W. 2019, ‘Mapping the global prevalence, incidence, and mortality of Plasmodium falciparum, 2000–17: a spatial and temporal modelling study’, The Lancet, vol. 394, no. 10195, pp. 322–31. Wernsdorfer, W.H. 1994, ‘Epidemiology of drug resistance in malaria’, Acta Tropica, vol. 56, no. 2, pp. 143–56. White, N.J. 2004, ‘Antimalarial drug resistance’, Journal of Clinical Investigation, vol. 113, no. 8, pp. 1084–92. WHO 2020, World Malaria Report 2020, viewed 20 February 2021, <https://www.who.int/publications/i/item/9789240015791>.

Science Extension Journal • 163

91 Paciﬁc Highway Hornsby NSW 2077 Australia

t w

+61 2 8438 7999 barker.college

2021 Scientific Research in School

Articles inside

Optimising the synthesis of the 4-iodo analogue of pyrimethamine

As simple as making hot dogs: Using analogies to teach limiting and excess reagents in high school chemistry

Synthesis of the 3,4-methylenedioxy analogue of Pyrimethamine

Antibacterial activity and degradation pathways of methallyl isothiocyanate

Concentration of Lycopene in Different Varieties of Tomatoes

Synthesis of Pyrimethamine

Concentration of Allicin in Garlic

Influence of ultraviolet light on the stability of allicin in aqueous garlic extract

Effect of time after harvest on chlorophyll concentration in spinach leaves