Xjenza Vol. 11 - 2006 - Needs Cover

Xjenza 11 (2006) In print

1

Invited Article The legacy of Mendeleev and his periodic system Eric R. Scerri* Department of Chemistry & Biochemistry, UCLA, Los Angeles, California 90095-1569, USA http://www.chem.ucla.edu/dept/Faculty/scerri

Received & Accepted: 18 September 2006 Published online: 30 September 2006

Next year is the one hundredth anniversary of the death of one of the most famous scientists of all time, the Russian chemist Dimitri Ivanovich Mendeleev (18341907). It is therefore appropriate to consider the lasting influence of Mendeleev and of his chef d’ouvre, the periodic system of the elements.

As is well known, Mendeleev was not the first to arrive at the periodic system although his version was the one that made the greatest impact in the scientific community.

Before Mendeleev there were at least five scientists who produced some very respectable periodic systems, but were not able to capitalize too much upon their discoveries. The French geologist De Chancourtois obtained the very first periodic system, in fact a three dimensional system which he inscribed on the outer surface of a metal cylinder. Two English chemists Newlands and Odling, who coincidentally were born in the same London borough of Southwark, independently published periodic tables and both realized the need to reverse the positions of the elements tellurium and iodine.

Just to back-track a little, the basic principle which all of these pioneers had realized, was that if the elements were ordered according to increasing atomic weight there would be an approximate repetition in their properties after certain well defined intervals. But there are a few stubborn exceptions to ordering the elements strictly according to atomic weight. Although the atomic weight of iodine is lower than that of tellurium, it was clear to Newlands, Odling and later Mendeleev, that the chemical properties of these elements required an exception to the ordering principle.

in his system. But this is an issue that even contemporary scholars of the periodic system continue to debate.

According to one school of thought, the successful accommodation of already known scientific facts is regarded as being equally important to successful predictions, in the acceptance of a new scientific development. This is despite the fact that predictions make a more dramatic impact especially on laypersons.

But even if one were to discount Mendeleev’s highly successful predictions of new elements it is clear that he also went a good deal further than Lothar Meyer in correcting the then known atomic weights of a number of elements such as uranium, titanium and beryllium.

And what about the lasting influence of Mendeleev’s system? Needless to say the periodic system served to unify and organize the chemistry of the elements. No longer were students of chemistry obliged to memorize the properties of all the known elements. Henceforth they could learn the properties of at least one element from each column and could in principle make good predictions concerning the properties of other group members. More fundamentally perhaps, the eightcolumn table paved the way to G.N. Lewis’ octet rule of chemical bonding and the notion of electron shells, some of which contain eight electrons. But I am getting ahead of the story.

Another early system was due to the enigmatic, Danish born, Gustav Hinrichs whose path to discovery included some fanciful analogies with planetary astronomy and the newly discovered spectroscopic frequencies obtained from various elements.

The closest precursor in chronological terms was the German chemist Lothar Meyer who arrived at a fully mature periodic system almost at the same time as Mendeleev. But a number of factors conspired to thwart Lothar Meyer’s efforts including an untimely delay in the publication of his most elaborate periodic table. The usual account of the rivalry between Lothar Meyer and Mendeleev has it that only Mendeleev possessed the courage to make predictions on the properties of the elements that occupied empty spaces Dimitri Mendeleev, Portrait by Ilya Repin

* Corresponding Author: e-mail: scerri@chem.ucla.edu, Tel: +1 (310) 206 7443.

Scerri ER, The legacy of Mendeleev and his periodic system

Mendeleev, unlike some of his contemporaries, rejected any suggestion that the periodic system implied the existence of any form of primary matter of which all the elements were composed. He maintained that all the elements were strictly individual, indestructible and irreducible. But the evidence began to point in the opposite direction as Mendeleev was nearing the end of his life. Several revolutionary discoveries in physics began to show that the atoms of the various elements were reducible and that there was a deep sense in which all elements are indeed made of primary matter, namely electrons, protons and neutrons.

In 1879 J.J. Thomson in Cambridge discovered the electron, a particle that seemed to occur in the atoms of all elements. Shortly afterwards in Paris, the work of Becquerel, and especially the Curies, also implied that atoms of elements consisted of components that were coming apart in the course of radioactive decay.

Thomson then attempted to explain the form of the periodic system by postulating the existence of rings of electrons embedded in the positive charge that comprised his plum pudding model of the atom. This was the origin of today’s electronic configurations, which have become the explanatory paradigm in much of chemistry. The key to an atom’s properties lies in the number of outer-shell electrons and that in turn is obtained by taking into account the configuration of all the electrons in the atom.

2

not assume the experimentally observed order of subshell filling?

In addition there are some continuing debates concerning the best way in which to represent Mendeleev’s periodic system. Should it be the original eight column short-form table, or the more contemporary eighteen column medium-long form or perhaps even a 32 column longform table which more naturally accommodates the rare earth elements into the main body of the table?

Alternatively some favor pyramidal tables while others prefer the left-step form originally proposed by Charles Janet in the 1920s. And very recently Philip Stewart of Oxford University has resuscitated a continuous spiral form of table.

Many chemists argue that the form of the table is of no importance but surely this is not so when rival forms position elements such as hydrogen and helium in quite different groups. Some philosophers of chemistry have argued that there may be an objective ‘fact of the matter’ regarding the kinship of helium for example. If this were so it would enable chemists to settle whether helium should be aligned with the noble gases or the alkaline earths, as it sometimes is on electronic grounds. And this is not a matter of convention as might be the choice between a pyramidal or a rectangular form of table.

The origin of electronic configurations is most frequently and incorrectly attributed to Niels Bohr who introduced the quantum theory to the study of the atom. But Bohr was essentially tidying up Thomson’s pre-quantum configurations and taking advantage of a more accurate knowledge of how many electrons each of the atoms actually possessed. Further developments in quantum theory, including Pauli’s Exclusion Principle and Schrödinger’s Equation, led to a more rigorous theoretical explanation of the form of the periodic system.

Now it became clear why the first two periods contain two and eight elements respectively. The exact solution of the Schrödinger equation for the hydrogen atom reveals a set of characteristic quantum numbers. The manner in which these numbers are related to each other is rigorously constrained by the theory and the outcome is that the first two shells contain a maximum of two and eight electrons respectively. The build-up of successive elements, by the addition of a proton and an electron (and variable numbers of neutrons), produces period lengths of two and eight elements respectively. So far so good, but the third shell contains 18 electrons according to quantum mechanics and yet the third period in the modern periodic table contains eight and not 18 electrons.

If one takes into account the precise order in which electron shells and especially their sub-shells are filled this goes some way towards explaining the length of the third period, but the explanation is no longer neat and rigorous but is strictly semi-empirical. Here then is one essential aspect of the periodic system that continues to challenge the ingenuity of theorists and physicists to this day. Can a more fundamental solution be found that does

One form of Mendeleev's periodic table, from the 1st English edition of his textbook (1891, based on the Russian 5th edition).

[Image taken from www.wikipedia.org]

I would like to now turn to an area where Mendeleev’s views have not been refuted but are indeed being reexamined in an attempt to clarify the philosophical foundations of chemistry. This topic concerns the distinction between elements regarded as basic substances as opposed to elements regarded as simple

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Scerri ER, The legacy of Mendeleev and his periodic system

substances. The latter notion is by far the better known especially among chemists. Since Lavoisier all attention has been directed towards ‘elements’ in the sense of substances that can be isolated following the decomposition of compounds.

But there is a more abstract sense of the term ‘element’, namely a bearer of properties that cannot be isolated. It is carbon the basic substance that exists in all the isotopes of carbon or indeed in any of the allotropic forms of carbon. The fact that the element as a basic substance cannot be isolated has meant that its importance was relegated and perhaps even forgotten altogether especially since the rise of positivism in science. But Mendeleev did not ignore the distinction between simple substances and basic substances. More than most authors in modern chemistry, Mendeleev devoted considerable attention to elaborating the distinction.

In many publications as well as subsequent editions of his Principles of Chemistry, he repeatedly emphasized that the periodic system was primarily a classification of elements as basic substances.

The distinction became rather important following the discovery of isotopes in the early years of the 20th century. Within a short period of time there suddenly seemed to be a profusion of ‘atoms’ or simplest possible substances to which all matter could be reduced. The periodic table, which was supposed to classify the simplest possible substances, was confronted with a major challenge. Some chemists including Fajans even called for an abandonment of Mendeleev’s periodic system in favor of a more complicated table of isotopes.

But other chemists, most notably Paneth, appealed to Mendeleev’s distinction to argue that that the periodic table should remain as the focus of attention for chemistry, since what mattered more in chemical terms was the classification of elements as basic substances.

All that the discovery of isotopes implied was an increase in the number of possible simple substances, or the less fundamental notion of elementhood.

Of course when really pushed even a contemporary chemist might admit the need for the concept of element as basic substance. For example one can ask a chemist about the substance that occupies the sixth place in the periodic system. What does it mean to refer to just an atom of carbon? Which of the particular isotopes would it be? The answer is that the sixth place in the periodic table does not refer to any particular isotope of carbon but the abstract notion of a carbon atom. Or in macroscopic terms what is the substance that belongs to the sixth place in the periodic table? Is it diamond, graphite or buckminsterfullerene? The answer is that it is none of these forms but whatever substance underlies these allotropes. It is carbon existing as the basic substance.

3

if at all, the elements survive following compound formation. Some of these philosophers hold that the notion of elements as simple substances is perfectly consistent with that of elements as basic substances and deny the lack of properties that is sometimes associated with them. Others believe that there is an irresolvable complementarity between the two notions and maintain, as Paneth did, that elements in the form of basic substances lack all properties with the possible exception of a few microscopic attributes such as atomic number.

To conclude, Mendeleev provided chemistry with its most profound organizing principle that even anticipates the discovery of quantum mechanics.

The periodic system has also provided chemistry with its most potent icon, which is recognized by anyone with even a passing knowledge of chemistry.

Even if students forget everything they ever learned, they tend to remember the existence of the periodic system.

Moreover, the distinction between elements as basic substances and as simple substances, as stressed by Mendeleev, continues to exercise the minds of the current generation of historians and philosophers of science. For the discovery of the mature periodic system and for the clearest elaboration of the nature of elements we are indebted to Dimitri Ivanovich Mendeleev (1834 - 1907).

Dr. Eric R. Scerri is a lecturer in the Department of Chemistry & Biochemistry at UCLA (USA) and is the editor-in-chief of the journal ‘Foundations of Chemistry’ published

by Springer. He has just written a book on the history and philosophy of the periodic system.

The Periodic Table: Its Story and Its Significance, Oxford University Press, 2006.

He is the author of over 100 articles in chemistry, history and philosophy of science and chemical education.

But whereas contemporary chemists adhere to an implicit notion of basic substances, if any, the growing community of philosophers of chemistry have recently devoted a good deal of attention towards clarifying the notion further since it is implicated in a number of perennial chemical questions such as the question of how,

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

4

Discussion Article Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications Owen Casha*, Joseph Micallef, Ivan Grech and Edward Gatt Department of Electronic Systems Engineering, Faculty of Engineering, University of Malta, Msida MSD 06, Malta.

Summary. The market trends of the rapidly growing communication systems require new product architectures and services that are only realisable by utilising technologies beyond that of planar integrated circuits. Microsystems technology (MST) is one such technology which can revolutionise radio frequency (RF) and microwave applications. This article discusses the enabling potential of the MST to meet the stringent requirements of modern communication systems. RF MST fabrication technologies and actuation mechanisms empower conventional processes by alleviating the substrate effects on passive devices and provide product designers with high quality versatile microscale components which can facilitate system integration and lead to novel architectures with enhanced robustness and reduced power consumption. An insight on the variety of components that can be fabricated using the MST is given, emphasizing their excellent electrical performance and versatility. Research issues that need to be addressed are also discussed. Finally, this article discusses the main approaches for integrating MST devices in RF and microwave applications together with the difficulties that need to be overcome in order to make such devices readily available for volume-production.

Keywords: MST, Micromachining, MEMS, RF Integrated Circuits Received: leave Blank, Accepted: leave blank Published online: leave blank

Introduction The rapidly growing communication systems entail demands for a wide range of radio frequency (RF) and microwave transceivers, and as a consequence, the research and development of such systems has dramatically increased in these last years. Amongst the common demands of present day and future RF and microwave applications, one finds the need for lower weight, low power consumption and low cost with increased functionality, frequency of operation and component integration. Consumers also demand for highly personalised, ever-present access to information together with growing volumes of data traffic and the need of higher quality products.

Such market trends require new product architectures that are enabled by technologies beyond that of planar integrated circuits (ICs). Microsystems technology (MST) is one such technology and is on the verge of revolutionising RF and microwave applications (Hilbert & Morris 2002). RF MST provides product designers with high quality versatile threedimensional (3-D), microscale design components that can be manufactured using traditional IC batch fabrication techniques. These components have the potential for enhancing performance and robustness together with the reduction of power consumption and size of modern wireless architectures (Nguyen 2005).

RF MST components are generally classified into two categories: micromachined components and micro-electro-mechanical systems (MEMS). The difference between them is that in an RF MEMS component, in addition to the micromachined RF element, there is an electromechanical actuator which is able to convert a voltage or current signal into a

mechanical movement to provide re-configurability to the device. RF MEMS are generally actuated using electrostatic, piezoelectric, magnetic or electro-thermal mechanisms. In contrast to European terminology, in the US and Asia, the term MEMS loosely represents microsystems technology (Lucyszyn 2004).

Very high quality factor (Q) passive devices, such as tunable capacitors with Q values up to 300 at 1 GHz (Young & Boser 1996, Yoon & Nguyen 2000) and inductors with Q values up to 85 at 1 GHz (Van Shuylenbergh et al. 2002), transmission line structures (Daneshmand et al. 2004, Kintis & Berenz 1997), mechanical resonators (Wang et al. 2004a, 2004b) and switches with insertion loss as low as 0.1 dB (Yao et. al 1999) are achievable using MST. Such devices can be implemented as discrete components or, for highly-integrated portable applications, merged with ICs through post-processing manufacturing techniques. MST components have been a very interesting area for research and development showing their excellent performance at microwave and even at mm-wave frequencies (Milosavljevic 2004). Such excellent performance comes from the possibility of the MST to limit the substrate coupling effects and the availability of metal structures with low parasitic capacitance and contact resistance.

RF MST structures are already being developed for switching applications such as transmit/receive duplexers (TDD) and band/mode selection which is fundamental for modern wireless multi-mode multi-standard applications.

Also micromachined film bulk acoustic resonators (Otis & Rabaey 2003) are already carried in many mobile handset platforms for two-way voice and data traffic (Coventor Inc. 2005). New applications are emerging as

* Corresponding Author: e-mail: oecash@eng.um.edu.mt Telephone: (+356) 2340 2388 / (+356) 7905 0635 Fax: (+356) 2134 3577

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

the existing technology is applied to the miniaturisation and integration of conventional devices. Also its flexibility is starting to be exploited to overcome the limitations exhibited by standard planar integrated RF devices. The ultimate goal in the utilisation of MST structures in RF and microwave applications is to propagate the device-level benefits all the way up to the system level to attain new levels of performance not achievable otherwise.

5

lines and multi-chip modules are possible structures which can be realised through this technology. With substrate bonding techniques, high purity dielectric substrates are bonded together, using flip-chip technology in order to create both electrical and thermal contacts between the substrates (Lucyszyn 2004). Note that in this technique there is no etching of the substrate other than for via holes.

Micromachining Technologies

Having said that, to date communication system designers are still finding some difficulty in utilising RF MSTs since they usually rely on non-standard design methods that require custom processes together with specialist knowledge resulting in long development cycles with relatively high risk and cost (Hilbert & Morris 2002). From the industry’s perspective, the overall implementation cost is the most important factor determining the utilisation of such RF MST structures (Lucyszyn 2004). Packaging and reliability determines the cost of such structures and in fact these issues are currently the subjects of an intense research effort, since together with process and material characterisation, will eventually determine how and when MST structures are readily available for volume-production (Milosavljevic 2004, Coventor Inc. 2005).

Fabrication Technologies RF microsystems have evolved as a result of continual advances in a number of different manufacturing technologies that have merged together. These can be broadly categorised into multilayer and micromachining technologies (Lucyszyn 2004). The three characteristic features of MST fabrication are miniaturisation, multiplicity and microelectronics. Miniaturisation enables the production of compact, fast-responding devices.

Multiplicity or batch fabrication, allows millions of structures to be easily and concurrently fabricated.

Thirdly, microelectronics provides the intelligence to the MST structures. In particular for MEMS structures it allows the monolithic merger of sensors, actuators and electronic circuitry to build a unified component or system. MST fabrication techniques empower conventional IC fabrication processes to produce 3-D mechanical structures not achievable by conventional machining techniques.

Multilayer Technologies Multilayer technologies are used to increase package densities, by implementing traditional planar 2-D components with 3-D geometries (Robertson & Lucyszyn 2001). Two main approaches are used in multilayer fabrication: microfabrication and substrate bonding. In microfabrication technologies, a dielectric or metal layer is first deposited and then a photolithographic process is used to pattern the layer. This is then repeated for any other required layer. Since sub-micron feature sizes can be obtained, this technology is normally associated with monolithic circuits (Lucyszyn 2004).

Lumped-element components, 2-D and 3-D transmission

There are three main approaches used in micromachining fabrication, namely bulk micromachining, surface micromachining and micromoulding.

Figure 1: A bulk-micromachined inductor (Source: Sun Y et al. 1996)

Bulk micromachining is an extension of IC technology for the fabrication of 3-D structures, in which they are sculpted within the confines of a wafer by exploiting the anisotropic etching rates of the different atomic crystallographic planes in the wafer (Moore & Syms 1999). In addition to this, the Technical University of Darmstadt has been working on micromachined structures developed from III-V semiconductor materials (Miao et al. 1995, Beilenhoff et al. 1999). In this case crystallographic etching techniques cannot be employed as in the case of silicon. Chemical etching is a possible alternative but poorer precision and profile definition is achieved (Lucyszyn 2004). Another approach is to form structures by means of fusion bonding, i.e., building a structure by atomatically bonding various wafers. The work in McGrath et. al 1993 demonstrated a W-band micromachined air-filled metalpipe rectangular waveguide, realised by using a twowafer sandwich approach. A measured level of insertion loss of only 0.04 dB/λg at 100 GHz was achieved. Wafer bonding continues to advance, with vertically integrated micromachined filters being demonstrated at frequencies in the 10 GHz range (Harle & Katehi 2002). Figure 1 shows an example of a bulk-micromachined inductor in which the substrate has been eliminated from underneath the spiral trace to reduce substrate associated parasitics (Sun et al. 1996).

3-D structures in surface micromachining are built up by the controlled addition and removal of a sequence of thin film layers to/from the wafer surface called structural and sacrificial layers, respectively. The substrate is used primarily as a mechanical support upon which the

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

micromechanical elements are fabricated. For instance, in Treen & Cronin 1993 a 600 GHz air-filled metal-pipe rectangular waveguide structure, realised using a single wafer is presented. A very thick layer of SU-8 photoresist was used to define the waveguide. Gold was deposited onto an SU-8 former, the sacrificial layer of SU-8 was removed to leave a 3-D metal structure. The success of this approach usually depends on the ability to release or dissolve the sacrificial layers while preserving the integrity of the structural layers. Figure 2 depicts an RF switch implemented as a surface-micromachined structure (Richards and De Los Santos 2001). Other possible structures are variable capacitors, V-antennas, rectangular waveguides and microstrip filters.

6

RF MST components and applications The MST fabrication possibilities provide solutions to the designer to develop new components that do not carry the limitations of conventional ones. In this section, an insight on the variety of both true-RF MEMS and micromachined components that can be fabricated using the MST is given, together with the applications in which they can utilised, emphasising their excellent electrical performance and versatility. These include switches, variable capacitors, variable and fixed inductors, transmission lines, resonators and antennas.

RF Micromachined Components Micromachined High-Q Fixed Inductors

Figure 2: A surface-micromachined RF switch (Source: Richards and De Los Santos 2001)

In the micromoulding process, microstructures are fabricated using moulds to define the deposition of the structural layer. The structural material is deposited only in those areas constituting the micro-device structure, in contrast to the previous approaches. After structural layer deposition, the mould is dissolved using a chemical that obviously does not attack the structural material.

Figure 3: Junction of a CPW 6-dB coupler fabricated in a LIGA process (Source: Willke et al. 1998)

A common type is the LIGA process, where LIGA is a German acronym consisting of letters LI (RoentgenLIthography, mean X-ray lithography), G (Galvanik, meaning electrodeposition) and A (Abformung, meaning moulding). In the LIGA process thick photoresists are exposed to X-rays to produce the moulds. Figure 3 shows a junction of a coplanar waveguide (CPW) 6-dB coupler fabricated in a LIGA process (Willke et al. 1998).

Inductors are key passive components which determine the noise and power consumption performance of tuned circuits, such as voltage-controlled oscillators (VCOs), low noise amplifiers (LNA) and impedance matching networks. Conventional IC planar spiral inductors lie on their host low resistivity substrate and as a result suffer from unwanted effects such as low self-resonant frequency, low Q (< 10) and limited operating bandwidth.

As a consequence, circuits requiring high performance inductors usually use off-chip components. Using micromachining it is possible to reduce the parasitics affecting conventional on-chip planar inductors, thus providing high performance inductors which can be integrated.

Figure 1 shows an example of a bulk-micromachined inductor in which the substrate has been eliminated from underneath the spiral trace (Sun et al. 1996). Removing the substrate reduces parasitic capacitances, thus increasing Q-factor and the self-resonating frequency and removes the associated capacitive dielectric losses.

Induced eddy currents in the substrate and associated energy lost due to Jouleâ&#x20AC;&#x2122;s heating are also minimized (Lucyszyn 2004). The measured Q values range from 6 to 28 at frequencies from 6 to 18 GHz, with typical inductor values of around 1 nH. Questions have been raised regarding their robustness to withstand subsequent wafer processing, lack of good RF ground and susceptibility of their characteristics to electromagnetic coupling (De Los Santos 2002). These issues were addressed in the structure proposed in Jiang et al. 2000, where the structure consists of an elevated inductor suspended over a 30 Âľm deep copper-lined cavity etched in a silicon bulk.

While substrate removal and shielding together with spiral elevation do improve inductor performance, the ultimate limitation on improvement is dictated by the remaining parasitic capacitance, between metal traces and the substrate. This approach yielded a Q of 30 at 8 GHz on a 10.4 nH inductor with a self-resonating frequency of 10.1 GHz. Solenoid-like inductors above the substrate, as that shown in Figure 4 have been developed using surface micromachining (Yoon et al. 1999) yielding a Q of 16.7 at 2.4 GHz on 2.67 nH inductors and exhibiting a linear relationship between the inductance and the number of

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

turns having the shape of a solenoid.

In (Ribas et al. 2000) the inductor is located on a platform which is then raised above the substrate using scratch-drive actuators.

Figure 4: Solenoid-like inductor (Source: Yoon et al. 1999)

Another approach uses the self-assembly principle, in which planar inductor structures are brought perpendicular to the substrate, aimed at decoupling the inductor properties from those of the substrate (Chen et al. 2003, Dahlmann et al. 2001a, 2001b, 2002, Zou et al. 2001a). As a consequence of this the designer has a greater possibility to increase the number of turns (maximize inductance) or conductor track width (minimize series resistance). A typical example is shown in Figure 5.

Figure 5: 4.5-turn meander inductor after self-assembly (Source: Dahlmann et al. 2001)

There are three major methods of actuation that are used during the manufacturing process to create nearly vertical inductors: plastic deformation magnetic assembly (Zou et al. 2001a), tensile stress (Lubecke et al. 2001a, Van Shuylenbergh et al. 2002, Chua et al. 2003) and surface tension (Dahlmann et al. 2001a, 2001b, 2002).

Very high Q-factors were demonstrated as those in Chua et al. 2003 and Van Shuylenbergh et al. 2002 in which Q values of 70 and 85 at 1 GHz were demonstrated.

7

means of the available micromachining techniques it is possible to approach a similar level of performance in the context of a microscopic planar IC level.

The work in Papapolymerou et al. 1997 proposed a micromachined cavity resonator for X-band applications, in which an unloaded Q of 506 for a cavity of 16x32x0.465 mm was obtained. It can be seen that such resonators are very advantageous in applications such as emerging millimetre-wave commercial applications, whose performance levels and frequency cannot be met otherwise. For instance, a 33.2 GHz monolithic microwave IC (MMIC) oscillator was stabilised by a similar micromachined cavity, in which the obtained phase noise is â&#x2C6;&#x2019;113 dBc/Hz at a 1 MHz offset from the carrier (Kwon et al. 1999). An 18 dB improvement over its MMIC free-running counter was obtained.

At lower frequencies, cavity resonators become impractical due to their excessively large dimensions.

Micromechanical resonators, in turn, become very attractive especially in the 1 kHz-1 GHz range, where they can be used to design ultra stable oscillators (Lin et al. 2004) and low loss circuit functions, such as bandpass filters (Bannon et al. 2000) and mixers (Wong & Nguyen 2004), for a wide range of transceiver types. There are two main design approaches to design such resonators: the vertical displacement resonator, in which a cantilever beam is set into a diving board-like vertical vibration in response to an electrostatic excitation and the lateral displacement resonator, in which the motion is obtained by exciting a comb-like structure. The typical maximum resonance frequency of such resonators is about 200 MHz with a Q value of almost 9400 under vacuum conditions. In Nguyen 2005 a short review of high-Q vibrating mechanical resonators is presented.

Micromechanical resonators usually require some form of transduction mechanisms so that they can be interfaced to an electrical circuit. Capacitive transduction is commonly employed, but resonators using piezoelectric (Piazza et al. 2005) and magnetomotive methods (Roukes 2000) also exist. The choice of transduction method greatly effects the achievable Q, impedance of the devices and also the amenability of their implementation.

Further improvement of the Q-factor of micromachined inductors in order to achieve values as high as 300, for applications such as multi-band filters is an important challenge for future research (Nguyen 2005).

Micromachined Resonators The performance of macroscopic waveguide (cavity) resonators and bulky mechanical resonators are well known, in particular the fact that they are capable of exhibiting Q values in the 10,000-to-25,000 range. By

Figure 6: SEM photograph of a membrane-type FBAR (Source: Lubecke et al. 2001b).

Applications requiring higher frequencies up to a few gigahertz appear to fall in the domain of film bulk acoustic wave resonator (FBAR) technology. An FBAR

8

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

device basically consists of a layer of piezoelectric material sandwiched between metal electrodes. Such technology is one of most promising for high-Q integrated wireless filters (e.g. wireless PCS duplex filters (Bradley et al. 2000), since it can provide sharp cut-off characteristics, compact size and a practical fabrication. It can also be employed in very stable oscillators such as that presented in Otis and Rabaey 2003, where a 1.9 GHz CMOS oscillator was designed using such a micromachined resonator. In Lubecke et al. 2001b a review of such resonators is presented describing resonator design and fabrication together with performance and circuit considerations. Two types of such resonators are the membrane based resonator (see Figure 6) and the acoustic mirror type resonator.

Micromachined Transmission Lines Transmission lines are flexible components in RF and microwave electronics and are utilised very often in many circuits and systems. Unfortunately conventional transmission lines, have some limitations such as frequency dispersion and to a certain extent insertion loss, originating from the properties of the substrate or the environment in which they are constructed. Similarly to the design of the other micromachined passive devices, MST can be utilised to design transmission lines in which the influence of the substrate is reduced.

There are three main approaches how a micromachined transmission line can be designed: membrane supported microstrip, coplanar microshield and the top-side-etched coplanar waveguide. In the membrane supported microstrip the transmission line is defined on a thin membrane, in which the substrate underneath the trace is bulk-etched via backside processing, to achieve a dielectric constant close to unity. The coplanar microshield was proposed to overcome the limitation exhibited by the membrane supported microstrip which has no intrinsic ground plane, by including the ground planes defining a ground-signal-ground structure. The top-side-etched coplanar waveguide relies on opening etch windows through the top passivation layer to create a pit underneath the line, thus simplifying the substrate etching process when compared to the first approach.

Micromachined waveguide uses micromachining techniques which are aimed at overcoming the lower-dimension bound of conventional machining techniques (Richards and De Los Santos 2001).

Typical uses of micromachined transmission lines are in the design of impedance tuners (Lubecke et al. 1998, Chiao et al. 1999b) and filters (Kim HK et al. 1999, 2002). Such applications usually require some form of tuning arrangement: the most common form is to make use of MEMS surface micromachined switches or variable capacitors. Another approach was used in the design of an impedance tuner in a coplanar stripline technology, in which a sliding planar backshort on top of a coplanar transmission lines forms a moveable short circuit and allows for variations in the length of short circuit transmission line stubs using a scratch-drive actuator (Chiao et al. 1999b).

RF MEMS Components Actuation Mechanisms The appropriate choice of an electromechanical actuator to be used in an RF MEMS component depends a lot on the available fabrication technologies and the requirements of the application being considered. There are mainly four types of actuation mechanisms: electrostatic, piezoelectric, magnetic and electrothermal.

The scratch-drive actuator (SDA) is a direct drive actuator that can be based on one of these actuation mechanisms. It has become a common means of achieving mechanical movement in RF MEMS components (Li et al. 2002, Minotti et al. 1998).

Electrostatic actuation is the most commonly employed mechanism because it gives the facility to produce small components that are robust, simple to fabricate, relatively fast and which consume almost no control power. Table 1 provides a comparison between these four actuation mechanisms. More details on these mechanisms can be found in Lucyszyn 2004 and Richards & De Los Santos 2001.

Actuation Mechanism

Actuation Voltage

Power Demand

Electrostatic

Can be quite high in particular for the design of good isolation switches

Piezoelectric

Lower than that required in Electrostatic Actuation

Low

Magnetic

Low

High

Electrothermal

Low

High

Structure

Actuation Speed

Simple Low

Robust to changes in the environment Can suffer from parasitic thermal expansion of the layers being used Bulky Difficult to fabricate Provide High Contact Force

Fast

Fast

Slow

Slow

Table 1: Comparison of RF MEMS actuation mechanisms

Switches The switch is considered as the most important RF MEMS component (Lucyszyn 2004). The excellent performance of prototype MEMS switches demonstrated in Yao 2000 (insertion loss of 0.1 dB, isolation of 50 dB from DC to 4 GHz and high linearity) shows their great potential for replacing lossy and power hungry semiconductor switches (e.g. Pin-diode switches or GaAs-based FET), such as in numerous applications such as switchable filters, tunable antennas, high efficiency RF power amplifiers and other reconfigurable RF circuits (DeNatale 2004). System architectures can be greatly enhanced, in terms of greater performance and functionality and reduced complexity and cost, if switch performance can be improved even further (Lucyszyn 2004).

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

They achieve such impressive performance characteristics because of their micromechanical construction, which not only provides a mechanical operation mechanism much like that of macroscopic mechanical switches, but allows the use of metal materials with substantially lower resistivity than semiconductors (Nguyen 2005). Current research is studying new structures with fast switching and low actuation voltage together with improving reliability, packaging and cost issues that can be tolerated in many telecommunication applications (Milosavljevic 2004, Nguyen 2005).

Nearly all RF MEMS switches are based on out-of-plane electrostatically actuated suspension bridge or cantilevertype as shown in Figure 2. The condition of pull-down occurs when the electrode separation decreases below two-thirds of the fully open condition (Lucyszyn 2004).

In practice, the actuation voltage for such designs is too high for many applications and so meandering can be introduced to lower the effective spring constant (Pacheco et al. 2000, Peroulis et. al 2003) since the actuation voltage is proportional to the square root of the spring constant. Other ways of lowering the actuation voltage are described in (Milosavljevic 2004); where a review on RF MEMS switches is also presented.

There are two generic types of RF MEMS switch: (i) the ohmic contact (metal-air-metal, MAM) or series switch (De Los Santos 1999) and (ii) the capacitive membrane (metal-insulator-metal, MIM) or shunt switch (Muldavin & Rebeiz 1999, 2000, Park et al. 2000). The advantages of the series switch are that a very low ON state insertion loss and very high OFF state isolation can be achieved.

On the other hand, such switches are very susceptible to stiction, corrosion and microscopic bonding of the contact electrodesâ&#x20AC;&#x2122; metal surfaces and usually require a considerable force to create a good metal-to-metal contact. The advantages of the capacitive membrane switch are that it has a longer lifetime and that the insertion loss is independent of the contact force relaxing the requirements of the actuation mechanism. With such a switch a trade-off between insertion loss and isolation exists (Lucyszyn 2004).

Although electrostatic actuation is the most commonly used type since such switches are usually fabricated using surface micromachining which is most compatible process technology with IC fabrication processes, other actuation mechanism are sometimes employed. RF MST has been used to implement magnetically actuated and thermally actuated switches. With the former, a micromachined magnetic latching switch has been shown in (Ruan et al. 2002) operating from DC to 20 GHz and with a worst-case insertion loss of 1.25 dB and an isolation of 46 dB. In (Blondy et al. 2001) an electrothermally actuated mm-wave switch is demonstrated, constructed using a stress-controlled dielectric bridge which buckles when heated. The insertion loss has been estimated to be 0.2 dB at 35 GHz. Lower speeds are reached which such actuation mechanisms when compared to the electrostatic form.

9

Variable Capacitors Variable capacitors or varactors are essential components wherever circuit tunability is required, like for instance, in variable frequency matching circuits, VCOs and phase shifters. For such applications, having a high Q is very important for maximising noise performance and minimising loss. Varactors have traditionally resisted monolithic integration due to their process incompatibility which results in devices with sub-optimal properties such as low Q, low self-resonating frequency, high sensitivity to even medium RF power levels and generally do not exhibit linear frequency tuning characteristics.

A number of publications demonstrate the performance improvement achieved by MEMS varactors (Young et al. 1998, Chen et al. 2003, Kassem and Mansour 2004, Liu 2002, Mehmet 2004, Yoon & Nguyen 2000) over conventional integrated diode or MOS varactors. For instance, the work in Yoon & Nguyen 2000 demonstrated a 1-4 pF varactor with a Q â&#x2030;&#x2C6; 300 at 1 GHz. MEMS based varactors usually take two forms: parallel plate and interdigitated. In the parallel plate approach the top plate is suspended a certain distance from the bottom plate by suspension springs, and this distance is made to vary in response to the electrostatic force between the plates induced by an applied voltage (Kim et al. 2002) or by using two SDAs located at the opposite sides of the capacitor (Chiao et al. 1999b). Another possible version is a structure in which the dielectrics can be electrostatically positioned between the two metal plates (Yoon & Nguyen 2000). Although not so common, development of electrothermally actuated RF MEMS varactors is possible as demonstrated in Hirano et al.

1995 and Feng et al. 1999.

Figure 7: A variable MEMS capacitor in which the actuation electrodes (E1-E3) are spaced differently from the capacitor plates (E1-E2).

(Source: Zou J. et al. 2001b)

Although such structures are very convenient to build due to simplicity of their fabrication, such types of MEMS capacitors have a maximum theoretical tuning range of 50% due to the collapse of the capacitor structure as the voltage in increased beyond a certain value (pull-in voltage). This problem can be minimised with the use of a passive series feedback capacitance to remove this instability (Seeger & Crary 1997). A MEMS parallelplate capacitor with a wider tuning range was proposed in Chen et al. 2003, Liu 2002 and Zou J. et al. 2001b by spacing the actuation electrodes differently from the capacitorâ&#x20AC;&#x2122;s plates, yielding a theoretical 100% tuning range as shown in Figure 7. However in practice, the obtained tuning range was of about 70%. In Kassem & Mansour 2004 a two movable-plate nitride loaded MEMS

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

varactor gave a tuning range of about 250%-280% whilst in Cai Y. et al. 2005 , a multi-step actuator is proposed to increase the tuning range of the standard parallel plate MEMS varactor.

Generally traditional MEMS varactors require lengthy cantilever beams to attain low valued stiffness constants (and so low activation voltages) such that in these designs there is a trade-off between Q and actuation voltage. The work in Sani et al. 2007, breaks this trade-off by proposing two novel structures for tunable MEMS capacitors which do not require lengthy beams. When linearity of tuning is more important than dynamic range, circular parallel-plate variable capacitors similar to that proposed in Chiao et al. 1999b are preferred. In this case the gap between the two plates is fixed and the overlapping area can be varied by means of two circular SDA.

Figure 8: Interdigitated MEMS-based varactor (Source: Borwick et al. 2003)

In the interdigitated approach, shown in Figure 8, the effective area of the capacitor is varied by changing the degree of engagement of the fingers of the comb like plates, similar to the macro-scale variable capacitors used in radio receivers (Borwick et al. 2003). While interdigitated MEMS capacitors do not suffer from the pull-in voltage limitation, have a high tuning linearity and are generally more compact, they exhibit low-Q values and a low selfâ&#x20AC;&#x201C;resonance frequency in comparison with the parallel-plate capacitors.

The most important research issues related to RF MEMS varactors are improvement in tuning range, lowering of actuation voltages and improvement in actuation speed together with reduction of microphonics and providing reliable packaging (Nguyen 2005).

Variable Inductors Very few attempts to build tunable true-RF MEMS inductors have been reported in literature to date. In Sun et al. 2001, the inductor is divided into four segments and a relay network made of MEMS switches change the overall inductance depending on the switching configuration. Although MEMS devices are used in such a structure to provide tuning, one may argue that these are not true-MEMS tunable inductors (Lucyszyn 2004). An approach to design a true-MEMS variable inductor is presented in Lubecke et al. 2001a. The inductor consists of two loops, that are self-assembled above the substrate

10

using the residual stress between the polysilicon and gold layers, with a relative angle between them, as shown in Figure 9.

Figure 9: Self-assembling variable inductor: mutually coupled loops (about 1200 Âľm long) bend at different rates when heated to allow controlled variation of inductance (Source: Lubecke et al. 2001b)

The two loops can be thermally controlled. The differential motion results from a cross-member corrugation structure in the inner loop that causes it to bend with temperature at a different rate than the outer loop. The motion affects the mutual component of the total inductance of the structure, thus a variation in inductance is achieved when a DC current flows through the structure. This technique offers a continuous tuning range of 18 %. The main problem with such approach is that the original angle separating the two inductors depends a lot on the deposition temperature of gold during the fabrication and is very hard to control (Abidine et al. 2003).

To alleviate this problem, in Abidine et al. 2003 the tunability of the inductance mutual component is achieved with the use of thermal actuators that control the spacing between the main and secondary inductor. In this case the inner inductor is again off the substrate because of the residual stress between the metal and the polysilicon layer whilst the outer inductor is attached to a beam that is connected to an array of thermal actuators.

When the array is actuated, the beam buckles and lifts up the outer inductor, thus varying the angle separating the two inductors. In this case the tuning range achieved was of 13 %.

Steerable Antennas In terms of true RF MEMS components, steerable antennas are the most difficult to implement and reach the desired performance. Antennas are very sensitive to the presence of adjacent structures: actuation mechanism in the antenna structure can easily interact with the antenna to distort the desired radiation pattern. Also if the radiating elements are detached from the supporting substrate, the size reducing property of the dielectric cannot be fully exploited (Lucyszyn 2004). Steerable antennas give the possibility of far-field radiation beamsteering and also beam-shaping. In Chiao et. al 1999a a 17.5 GHz reconfigurable Vee antenna has been reported, in which the arms of the Vee antenna used to form the radiating aperture are moved, using linear scratch drive

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

actuators. In Baek et al. 2003 2-D beam steering is achieved using magnetic actuation.

RF MST Structures: Design and Implementation In the early days of RF MST structures development, designers used to rely on expensive and time consuming prototyping cycles. Today, a lower risk approach is possible because of the commercially available RF MST design tools, which are user-friendly and enable shorter time development and thus lower design costs.

In particular, the nature of RF MEMS structures necessitates such design tools that can solve true-coupled physical analysis such as electrostatics, electromagnetic, mechanical and thermal. Whilst developing RF MEMS, designers must also take into consideration the device layout, construction and packaging together with their effect on the system performance, in particular if the RF MEMS structures will be used within an IC framework. This is again facilitated through the use of available design tools. As a matter of fact, RF MEMS design tools usually provide modelling and analysis on both a behavioural, physical and structural level.

Successful performance of RF and microwave integrated systems depends a lot on how good is the packaging of the integrated circuits. In RF MEMS development, apart from ensuring that unwanted electromagnetic interference and coupling together with unwanted resonances are not present, designers should aim at developing packaging techniques that prevent moisture and particulates, which may impair the movement of freestanding MEMS structures. Reduction of energy losses such as acoustic and thermal must also be considered. For example the reliability of an RF switch, which is very important for long-term applications (Rebeiz & Muldavin 2001, Campbell 2001, Cass 2001), is limited by organic deposits and contamination around the contact area.

These can be eliminated by a clean hermetic packaging environment, but unfortunately this is generally the most expensive step in the production chain and determines the cost of the switch, challenging volume-production. It is likely that the RF switch will continue to be most important RF MEMS component (Lucyszyn 2004) and as a matter of fact there is currently a large effort to develop low cost reliable wafer-scale packaging techniques which are compatible with MEMS switches (Rebeiz & Muldavin 2001). Two commonly used RF MEMS packaging approaches are the flip-chip assembly technique (Miller et al. 2000) and the self-packaged technique (Robertson et al. 1995).

11

components, switches, filters, VCOs, mixers and diplexers can all be replaced by their MST counterparts.

Indeed, even just direct replacement of components via MST-based ones can lead to significant performance increases. For example, analyses before and after replacement of off-chip high-Q passives by higher Q MST versions in a super-heterodyne architecture often show dramatic improvements in receiver noise figure, e.g. from 8.8 dB to 2.8 dB (Nguyen 2005).

Figure 10: RF MEMS based receiver (Source: Larson 1999)

In the other approach, the designer begins the development by actually devising a new system architecture that is not biased by the usual limitations imposed by conventional RF components. Such architecture would take the advantages provided by MST realisations, such as their microscale size and zero dc power consumption, and use them in massive quantities to enhance robustness and trade Q for power consumption (Nguyen 2005). For example shown in Figure 10, is a receiver architecture proposed in Larson 1999 which utilises an acoustic resonant intermediate frequency (IF) filter bank aimed at simplify the implementation by eliminating the need for a tunable first local oscillator (LO), by replacing it with a fixed LO which can be designed to have a better phase noise response and exploiting the switchable filter bank to affect IF band selection. From the industryâ&#x20AC;&#x2122;s perspective, the overall implementation cost is the most important factor determining the utilisation of such RF MST structures. Although using the bottom-up approach can provide superior RF performance to a conventional transceiver architecture, there is no overall economic gain in replacing conventional devices with RF MST counterparts. On the other hand, utilising the top-down approach there may be cost benefits from new architectures that are enabled with RF MST structures (Lucyszyn 2004).

RF MST Challenges

RF MST Utilisation There exist two main approaches for integrating RF MST devices into RF and microwave applications: the bottomup approach and the top-down approach. In the bottomup approach the designer uses an already established transceiver architecture and replaces the main components by RF MST structures. For example in a conventional transceiver architecture, whether it be superheterodyne or homodyne, the off-chip passive

Although MST has already demonstrated its superior RF performance over conventional approaches, the difficulty in matching the future requirements of the RF designer with the limitations of commercial foundry processes should be taken in consideration (Lucyszyn 2004).

Due to the fact that MST offers many degrees of design freedom than conventional ICs and that most of the available MST products are fabricated in confined

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

foundries, it will be difficult to have a MST “standard process” in the IC sense. As a matter of fact, to date communication system designers are still finding some difficulty in making use of the benefits of RF MST structures, also because the development of such devices requires specialist knowledge. To alleviate this issue MST development platforms (MDP) are being introduced, where an MDP is a reusable MST design component that has been designed, manufactured and validated. Making use of already proven design elements on an IC-compatible process technology platform provides broad access to MST even for non-specialists (Hilbert and Morris 2002).

However, the rapid growth of the MST industry has been impeded by a general lack of reliable material properties, understanding of processing effects on materials and process variables. Standardisation needs to be applied to the methods of characterising a process and its material properties to guarantee volume-production feasibility (Coventor Inc. 2005). Moreover special packaging techniques are required for reliable operation of such devices, especially in the case of RF MEMS structures.

Packaging determines the cost of the product and since prices are still very high, it is proving to be one of the major challenges to volume-production feasibility.

There are also inherent problems associated in particular with RF MEMS: for instance at low microwave frequencies, resonant structures are relatively bulky and can be difficult to move under electromechanical actuation (Lucyszyn 2004). Another drawback of RF MEMS is that the actuation voltage required is still relatively high when compared to the always decreasing supply voltage of ICs and the speed of the actuation is not always adequate to every application. Another issue that may present difficulty to chip manufactures is the fact that to take advantage of the benefits of micromachining in RF design, incurs extra processing steps. Nevertheless, since these fabrication techniques, may be implemented as post processing steps, chip manufacturers are not required to modify their high-volume fabrication processes, thus making RF MST structures extremely appealing (De Los Santos 2002).

Conclusion The recurring demand for wireless systems, such as wireless data links and Internet services, to be more flexible and sophisticated, yet consume very little power and occupy less space, has generated the need for a technology that can effectively reduce manufacturing size, weight and cost and improve performance together with increasing the battery life.

RF MST is widely believed to be one of the technologies that can cater for all these needs by eliminating off-chip passive components, enabling wide operational bandwidths, reducing interconnect losses and producing almost ideal switches and resonators. These provide designers with the necessary elements and features to create novel reconfigurable systems with new levels of performance not achievable otherwise.

12

The availability of reliable material properties, processing effects and variables during design and manufacturing, together with accurate, easy to use, commercially available MST design tools enables shorter time-tomarket and lower design costs. Also, the use of MDP enables a semi-custom design methodology built on the reuse of already proven design elements on an ICcompatible process technology platform, thus providing broad access to the MST allowing non-specialists to take the benefits of integrating MST components into their designs. Availability of low cost hermetic packaging techniques will surely facilitate the volume-production of MST components, in particular for RF MEMS components.

References Abidine IZE, Okoniewski M and McRory JG (2003), A new class of tunable RF MEMS inductors, Proc. of the International Conference on MEMS, NANO and Smart System, 114- 115.

Baek CW, Song S, Park JH, Lee S, Kim JM, Choi W, Cheon C, Kim YK and Kwon Y (2003), A V-band micromachined 2-D beam-steering antenna driven by magnetic force with polymer based hinges, IEEE Trans. Microwave Theory Tech., 51 (1), 325-331.

Bannon FD III, Clark JR and Nguyen CTC (2000), High frequency micromechanical filters, IEEE Journal of Solid-State Circuits, 35 (4), 512-526.

Beilenhoff K, Mutamba C, Pfeiffer J, Peerings J, Simon A, Lin C and Hartnagel HL (1999), III-V semiconductor structuring for mm and sub-mm waves, Workshop M-FrW3 European Microwave Week.

Blondy P, Mercier D, Cros D, Guillon P, Rey P, Charvet P, Diem B, Zanchi C, Lapierre L, Sombrin J and Quoirin JB (2001), Packaged millimetre wave thermal MEMS switch, Proc. 31st European Microwave Conf., London, 283–286.

Borwick RL, Stupar PA, DeNatale JF, Anderson R and Erlandson R (2003), Variable MEMS capacitors implemented into RF filter systems, IEEE Trans.

Microw. Theory Tech., 51 (1), 315–319.

Bradley P, Ruby R, Larson J, Oshmyansky Y and Figuerdo D (2000), A film bulk acoustic resonator (FBAR) Duplexer for USPCS Handset Applications, IEEE MTT-S Int. Microwave Symposium, 1, 367 – 370.

Cai Y, Wang J, Ativanichayaphong T and Chiao JC (2005), Multi-step Actuators for RF MEMS varactors and capacitive switches, TexMEMS, VII International Conference on MEMS, Mexico.

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

13

Campbell T (2001), MEMS Switch Technology Approaches the “Ideal Switch”, Applied Microwave & Wireless, 13 (5), 100-107.

Proceedings of the 52nd IEEE Electronic Components and Technology Conference.

San Diego, CA¸1027-1031.

Cass S (2001), Large Jobs for Little Devices, IEEE Spectrum, 38 (1), 72-73.

DeNatale J (2004), Reconfigurable RF Circuits Based on Integrated MEMS switches, ISSCC 2004, 1, 310-311.

Chen J, Zou J, Schutt-Aine JE and Kang SM (2003), Design and Modeling of a Micromachined High-Q Tunable Capacitor with Large Tuning Range and a Vertical Planar Spiral Inductor, IEEE Transactions on Electron Devices, 50 (3), 730-739.

Chiao JC, Fu Y, Chio IM, Delisio M and Lin LY (1999a), MEMS reconfigurable Vee antenna, IEEE MTT-S Int. Microwave Symposium, Anahein, 1515-1518.

Chiao JC., Fu Y, Choudhury D, and Lin LY (1999b), MEMS millimeterwave components, Proc. IEEE MTT-S Int. Microwave Symposium, Anaheim, 463-466.

Chua CL, Fork DK, Van Schuylenbergh K, Lu JP (2003), Out-of-plane high Q inductors on low resistance silicon, Journal of Microelectromechanical Systems, 12 (6), 989-995.

Coventor Inc. (2005), Micro-electro-mechanical systems move toward standardization, Internet address: http://www.coventor.com/media/papers/sta ndardization.pdf (Accessed: 20th December 2006).

Dahlmann GW, Yeatman EM, Young PR, Robertson ID, and Lucyszyn S (2001a), MEMS High Q Microwave Inductors Using Solder Surface Tension Self-Assembly, IEEE IMS Digest of Papers, 1, 329-332.

Dahlmann GW, Yeatman EM, Young PR, Robertson ID and Lucyszyn S (2001b) ‘High Q achieved in microwave inductors fabricated by parallel selfassembly’. Proc. 11th Int. Conf. on Solid-State Sensors and Actuators (Transducers Eurosensors XV), Munich, Germany, 1098–1101.

Dahlmann GW, Yeatman EM, Young and Lucyszyn S (2002), characteristics and mechanical assembled 3D microwave Actuators, 215–220.

PR, Robertson ID Fabrication, RF stability of selfinductors’, Sen.

Daneshmand M, Mansour RR and Sarkar N (2004), RF MEMS Waveguide Switch, IEEE Transactions on microwave theory and techniques, 52 (12), 2651-2657.

De

Los Santos HJ (1999), Introduction to Microelectromechanical (MEM) Microwave Systems, Artech House, Boston – London.

De Los Santos HJ (2002), On the Ultimate Limits of IC Inductors - An RF MEMS Perspective,

Feng Z, Zhang W ,Su B , Harsh KF, Gupta KC, Bright V and Lee YC (1999), Design and modelling of RF MEMS tunable capacitors using electro-thermal actuators, Proc. IEEE MTT-S Int. Microwave Symposium, Anaheim, 1507–1510.

Harle L and Katehi LPB (2002), A vertically integrated micromachined filter, IEEE Trans. Microw. Theory Tech., 50 (9), 2063–2068.

Hilbert J and Morris A, (2002), RF MEMS Switch Platforms Expedite MEMS Integration and Commercialization, COMMS 2002, Ypsilanti, Michigan, USA. Hirano M, Nishikawa K, Toyoda I, Aoyama S, Sugitani S and Yamasaki K (1995), Three-dimensional passive circuit technology for ultra compact MMICs, IEEE Trans. Microwave Theory Techniques, 43 (12), 2845–2850.

Jiang H, Wang Y, Yeh JLA and Tien NC (2000), Fabrication of High-Performance On-Chip Suspended Spiral Inductors by Micromachining and Electroless Copper Plating, 2000 IEEE IMS Digest of Papers, Boston, MA, 1, 279-282.

Kassem MB and Mansour RR (2004), Two MovablePlate Nitride-Loaded MEMS Variable Capacitor, IEEE Transactions on Microwave Theory and Techniques, 52 (3), 831-837.

Kim HK, Park JH, Kim YK and Kwon Y (1999), Millimeterwave micromachined tunable filters, Proc. IEEE MTT-S Int. Microwave Symposium, Anahein, 1235–1238.

Kim HK, Park JH, Kim YK, and Kwon Y (2002), Lowloss and compact V-band MEMS-based analog tunable bandpass filters, IEEE Microwaves Wireless Components Letters, 12 (11), 432 –434.

Kim Y, Lee SG and Park S, Design of the two-movable plate type MEMS voltage tunable capacitor, NSTI 2002 Nanotechnology Conference, Santa Clara.

Kintis M and Berenz JJ (1997), MEMS switched resonators for VCO applications, United States Patent No.: 5,994,982.

Kwon Y, Cheon C, Kim N, Kim C, Song I and Song C (1999), A Ka-band MMIC Oscillator Stabilized

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

with a Micromachined Cavity, IEEE Microwave and Guided Wave Letters, 9 (9), pp. 360–362.

Larson LE (1999), Microwave MEMS Technology for Next-generation Wireless Communications – Invited Paper, IEEE MTT-S Digest.

Li L, Brown JG and Uttamchandani D (2002), Study of scratch drive actuator force characteristics, Journal of Micromechanics and Microengineering, 12, 736-741.

Lin YW, Lee S, Li SS, Xie Y, Ren ZY and Nguyen CTC (2004), Series resonant VHF micromechanical resonator reference oscillators, IEEE Journal of Solid-State Circuits, 39 (12), 2477-2491.

Liu C (2002), Wide Tuning Range Variable MEMS capacitor, United States Patent No.: 6,418,006.

Lubecke VM, Barber B, Chan E, Lopez D, Cross ME and Gammel P (2001a), Self assembling MEMS variable and fixed RF inductors, IEEE trans. on Microwave Theory and Techniques, 49 (11), 2093-2098.

Lubecke VM, Barber BP and Fetter LA (2001b), High-Q MEMS for Wireless Integrated Circuits, 5th International Conference on Telecommunications in Modern Satellite, Cable and Broadcasting Service, 1, 203-209.

Lubecke VM, McGrath WR, Stimson PA and Rutledge DB (1998), Micromechanical tuning elements in a 620-GHz monolithic integrated circuit, IEEE Trans. Microwave Theory Tech., 46 (12), 2098–2103.

Lucyszyn S (2004), Review of radio frequency micromechanical systems technology, IEE Proc.Sci. Meas. Technol., 151 (2), 93-103.

McGrath WR, Walker C, Yap M, and Tai YC (1993), Silicon micromachined waveguides for millimeterwave and submillimeterwave frequencies, IEEE Microw. Guid. Wave Lett., 3 (3), 61–63.

Mehmet G (2004), Method of making an integrated electromechanical switch and tunable capacitor, United States Patent No.: 6,969,630.

Miao J, Hartnagel HL, Wilson RJ and Weiss BL (1995), Improved free standing semi-insulating GaAs membranes for sensor applications, Electronic Letters, 31 (13), 1047-1049.

Miller DC, Zhang W and Bright VM (2000), Microrelay Packaging Technology Using Flip-chip Assembly, The 13th Annual International Conference on Micro Electro Mechanical Systems, 265–270.

Milosavljevic ZD (2004), RF MEMS Microwave Review, 10 (1), 2-8.

14

Switches,

Minotti P, Bourbon G, Langlet P and Masuzawa T (1998), Arrayed electrostatic scratch drive actuators: Toward visible effects up to the human scale, Journal of Intelligent material systems and structures, 9¸837-846.

Moore DF and Syms RRA (1999), Recent developments in micromachined silicon, Electron. Commun. Eng. J., 11 (6), 261–270.

Muldavin JB and Rebeiz GM (1999), 30 GHz Tuned MEMS Switches, IEEE MTT-S Int. Microwave Symp. Dig.,Anaheim, CA, USA,1511-1514.

Muldavin JB and Rebeiz GM (2000), High-Isolation Inductively-Tuned X-Band MEMS Shunt Switches, IEEE MTT-S Int. Microwave Symp. Dig., Boston, Massachusets, USA, 169-172.

Nguyen CTC (2005), RF MEMS in Wireless Architectures, DAC 2005, Anaheim, California, USA, 416-420.

Otis B and Rabaey J, (2003), A 300-µW 1.9-GHz CMOS Oscillator Utilizing Micromachined Resonators, IEEE Journal of Solid-State Circuits, 38 (7), 12711274.

Pacheco SP, Katehi LPB and Nguyen CTC. (2000), Design of low actuation voltage RF MEMS switch, Proc. IEEE MTT-S Int. Microwave Symposium., Boston, MA, 165–168.

Papapolymerou J, Cheng JC, East J and Katehi LPB (1997), A Micromachined High Xband Resonator, IEEE Microwave and Guided Wave Letters, 7 (6), 168-170.

Park JY, Kim GH, Chung KW, Bu JU (2000), Fully Integrated Micromachined Capacitive Switches for RF Applications, IEEE MTT-S Int. Microwave Symp. Dig., Boston, Massachusetts, USA, 283-286.

Peroulis D, Pacheco SP, Sarabandi K and Katehi LPB (2003), Electromechanical Considerations in Developing Low-Voltage RF MEMS Switches, IEEE Trans. on MTT, 51 (1), 259-270.

Piazza G, Stephanou PJ, Porter JM, Wijesundara MJ, and Pisano AP, Low motional resistance ring-shaped contour-mode aluminum nitride piezoelectric micromechanical resonators for UHF applications, 18th IEEE International Conference on Micro Electro Mechanical Systems, MEMS’ 05, 20-23.

Casha O. et al.: Utilisation of Microsystems Technology in Radio Frequency and Microwave Applications

Rebeiz GM and Muldavin JB (2001), RF MEMS Switches and Switch Circuits, IEEE Microwave Magazine, 2 (4), 59-71.

Ribas RP, Lescot J, Leclercq JL, Karam JM and Ndagijimana F. (2000), Micromachined microwave planar spiral inductors and transformers, IEEE Trans. Microw. Theory Tech., 48 (8), 1326-1335.

Richards RJ and De Los Santos HJ (2001), MEMS for RF Microwave Wireless Applications - The Next Wave, Microwave Journal.

Robertson ID and Lucyszyn S (2001), RFIC and MMIC Design and Technology. IEE, London.

Robertson SV, Katehi LPB and Rebeiz GM (1995), Micromachined Self-packaged Wband Bandpass Filters, IEEE MTT-S International Microwave Symposium Digest, 3, 1543–1546.

Roukes ML, Nanoelectromechanical systems (2000), Solid State Sensor and Actuator Workshop, Hilton.

Head, SC, 367-376.

Ruan M, Tam G, Vaitkus R, Wheeler C and Shen J (2002), Micromagnetic latching RF switches, Proc.

Wireless Design Conf., London, 59–66.

Sani EA, Nasirzadeh N and Dadshzadeh G (2007), Two novel structures for tunable MEMS capacitor with RF applications, Progress in Electromagnetic Research (PIER 68), 169-183.

Seeger JI and Crary SB (1997), Stabilization of electrostatically actuated mechanical devices, Proc.

IEEE Int. Conf. on Solid-State Sensors and Actuators (Transducers), Chicago, IL, 1130–1136.

Sun XQ, Zhou S and Carr WN (2001), A micro variable inductor chip using MEMS relay, IEEE International Conference on solid-state sensors and actuators, 1137-1140.

Sun Y, Zeijl H, Tauritz JL and Baets R (1996), Suspended Membrane Inductors and Capacitors for Application in Silicon MMICs, IEEE Microwave and Millimeter-wave Monolithic Circuits Symposium Digest of Papers, 99-102.

Treen SA and Cronin, NJ (1993), Terahertz metal pipe waveguides, Proc.18th Int. Conf. on Infrared and Millimetre Waves, London, 470–471.

Van Shuylenbergh K, Chua CL, Fork DK, Lu JP and Grifths B (2002), On-chip out-of-place high-Q inductors, Proc. IEEE Conf. on High Performance Devices, Palo Alto CA, 364–374.

Wang J, Butler JE, Feygelson T and Nguyen CTC (2004b), 1.51 GHz polydiamond µmechanical disk resonator with impedance-mismatched isolating

15

support,” Proceedings, IEEE Int. MEMS Conf., Maastricht, The Netherlands, 641-644.

Wang J., Ren Z. and Nguyen CTC (2004a), 1.156-GHz self-aligned vibrating micromechanical disk resonator, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., 51 (12), 1607-1628.

Willke TL, Onggosanusi E and Gearhart SS (1998), Micromachined Thick-metal Coplanar Coupled-line Filters and Couplers, IEEE MTT-S International Microwave Symposium Digest, 1, 115–118.

Wong AC and Nguyen CTC, Micromechanical Mixer-Filters Mixlers,” IEEE/ASME J. Microelectromechanical Systems, 13 (1), 100-112.

Yao JJ (2000), Topical Review: RF MEMS from a Device Perspective, Journal of Micromechanical Microengineering, 10, 9-38.

Yao ZJ, Chen S, Eshelman S, Denniston D, Goldsmith C (1999), Micromachined Low-Loss Microwave Switches, IEEE/ASME J. Microelectromech. Syst., 8 (2), 129-134.

Yoon JB, Kim BK, Han CH, Yoon E and Kim C.K. (1999), Surface Micromachined Solenoid OnSi and On-Glass Inductors for RF Applications, IEEE Electron Device Letters, 20, 487.

Yoon JB and Nguyen CTC (2000), micromechanical capacitor with for RF applications, Technical Electron Devices Meeting, California, 489-492.

A high-Q tunable movable dielectric Digest, IEEE Int. San Francisco,

Young D and Boser BE (1996), A Micromachined Variable Capacitor for Monolithic High-Q VCOs, IEEE Solid-State and Actuator Workshop Proceedings, Hilton Head Island, SC, 86-89.

Young DJ, Malba V, Ou JJ, Bernhardt AF and Bernhard EB (1998), A Low-Noise RF Voltage-Controlled Oscillator using on-chip High-Q Three Dimensional Coil Inductor and Micromachined Variable Capacitor, Solid-state Sensor and Actuator Workshop Hilton Head Island, South Carolina, 128-131.

Zou J, Chen J and Liu C (2001a), Plastic deformation magnetic assembly (PDMA) of 3D microstructures: technology development and application. Proc. 11th Int. Conf. on Solid-State Sensors and Actuators, Munich, Germany, 1582–1585.

Zou J, Liu C and Schutt-Aine JE (2001b), Development of a wide-tuning-range two-parallel tunable capacitor for integrated wireless communication systems, 322-329.

Xjenza 11 (2006)

17

Review Article Negative Thermal Expansion Joseph N. Grima*, Victor Zammit and Ruben Gatt Department of Chemistry, University of Malta, Msida MSD 06, Malta. www: http://www.auxetic.info Summary. Materials with a negative thermal expansion coefficient contract when heated and expand when cooled.

This paper reviews mechanisms of how this unusual property can be achieved at the molecular and macroscopic level.

Some applications of this unusual property are also discussed.

Keywords: Negative thermal expansion, NTE, thermal contraction, zeolites Received: 29 September 2006 Accepted: 26 December 2006 Published online: 26 December 2006

Introduction Many materials which we encounter in everyday life expand on heating. This fact of life is easily highlighted with a simple experiment that we are all familiar with and involves a metallic sphere passing through a metal ring.

At room temperature, the sphere is of such a size that it can easily pass through the ring, but after the sphere is heated, it will have expanded with the result that it can no longer pass through the ring.

The phenomenon of thermal expansion may be explained by looking at interatomic distances. If we were to consider a crystal lattice close to absolute zero with individual atoms located next to each other in a regular pattern, the atoms would be vibrating to a relatively small extent, taking up a particular volume of space. As the temperature is increased, the atoms vibrate more and this corresponds to the material expanding as more volume is taken up by the vibrations.

Expansion in a material may be one of two forms: isotropic and anisotropic. In isotropic expansion, the material expands by the same extent in any direction (isotropically) upon heating. Polycrystalline aggregates and cubic materials undergo this form of expansion. In such cases, the extent of thermal expansion may be measured in terms of the volumetric thermal expansion coefficient αV which at constant pressure p is defined as follows: 1 ⎛ ∂V ⎞ αV = ⎜ (1) ⎟ V ⎝ ∂T ⎠ p where V is the volume of the sample and T is the temperature. Alternatively, it may be expressed in terms of the density, ρ, through: 1 ⎛ ∂ρ ⎞ αV = − ⎜ ⎟ (2) ρ ⎝ ∂T ⎠ p However, not all materials expand in an isotropic fashion and instead, the extent of expansion will be dependent on the particular direction where the measurement is taken (i.e. anisotropic expansion). To quantify anisotropic

thermal expansion (e.g. in a generalised single crystal), it is more useful to define the thermal expansion in some particular crystallographic direction and in such cases, the linear expansion coefficient, α L is used. This is defined as: 1 ⎛ ∂L ⎞ αL = ⎜ ⎟ (3a) L ⎝ ∂T ⎠ p where L is the length of the unit cell in the direction of interest. This equation may be re-written in terms of εL, the strain the direction of measurement as follows: ε L = α L dT (3b) or, more generally, for any direction, one may use the tensorial notation: ε ij = α ij dT i, j = 1,2,3 (3c) where εij are the elements of the strain tensor (i.e. ε11, ε22 and ε33 are the strains in the Ox1, Ox2 and Ox3 directions respectively and ε12 = ε 21 , ε13 = ε 31 and ε 23 = ε 32 are half the shear strains in the Ox1–Ox2, Ox1–Ox3 and Ox2–Ox3 planes respectively.

It should be noted that for isotropic systems which do not α12 = α13 = α 23 = 0 whilst experience any shear, α11 = α 22 = α 33 = α L . Also, in such cases, αV and α L are related through:

αL =

αV

(4) 3 In most cases, materials expand when heated, i.e., V increases (ρ decreases ) or L increases as T increases with the result that thermal expansion coefficient αV or αL are positive (see equations (1) – (3)). However, it should be noted that materials which defy common expectation and contract when heated do exist, and in such cases, the thermal expansion coefficients assume a negative value.

It should also be noted that for anisotropic systems, different values for α L may exist, depending on the direction of measurement. In fact, such materials may

* Corresponding Author: e-mail: joseph.grima@um.edu.mt Telephone: +356 23402274 / +356 79049865 Fax: +356 21123456.

Grima JN et al.: Negative Thermal Expansion

18

possess positive α L ’s in one direction and zero or negative α L ’s in other directions.

Alternative forms for defining the thermal expansion coefficient It is frequently the case that αV is expressed in other forms apart from the one in equation (1). For example, from the Maxwell relationship: ⎛ ∂S ⎞ ⎛ ∂V ⎞ (5) ⎜ ⎟ = −⎜ ⎟ ⎝ ∂T ⎠ p ⎝ ∂p ⎠T where S is the entropy of the system, from equations (1) and (5), the volumetric thermal expansion coefficient αV may also be written as: 1 ⎛ ∂S ⎞ (6) αV = − ⎜ ⎟ V ⎝ ∂P ⎠T Also, defining the isothermal compressibility coefficient, χT by: 1 1 ⎛ ∂V ⎞ =− ⎜ (7) ⎟ KT V ⎝ ∂P ⎠T where KT is the isothermal bulk modulus, then from equations (6) and (7): 1 ⎛ ∂S ⎞ ⎛ ∂S ⎞ (8) αV = χ T ⎜ ⎟ =− ⎜ ⎟ KT ⎝ ∂V ⎠T ⎝ ∂V ⎠T

χT = −

Alternatively, using the Maxwell relationship: ⎛ ∂p ⎞ ⎛ ∂S ⎞ ⎜ ⎟ =⎜ ⎟ ∂ T ⎝ ⎠V ⎝ ∂V ⎠T then, for constant volume conditions, αV expressed as: 1 ⎛ ∂p ⎞ ⎛ ∂p ⎞ αV = χ T ⎜ ⎟ = − ⎜ ⎟ T K ∂ ⎝ ⎠V T ⎝ ∂T ⎠V

(9) may be (10)

An important parameter when studying thermal expansion is the Grüneisen parameter γ (Grüneisen 1926).

This parameter represents the thermal pressure from a collection of vibrating atoms and is defined as: V ⎛ ∂p ⎞ γ= (11) ⎜ ⎟ CV ⎝ ∂T ⎠V where CV V is the heat capacity per unit volume at constant volume. Thus, from equation (10), the volumetric thermal expansion coefficient αV may be written in terms of the Grüneisen parameter γ as follows: χ C C αV = T V γ = V γ (12) V VKT where χT is the isothermal compressibility coefficient, KT is the isothermal bulk modulus and CV is heat capacity at constant volume. The Grüneisen parameter is dimensionless and its sign determines whether the expansivity is positive or negative.

Alternatively, for an adiabatic compression, αV may be written as:

αV =

χS Cp V

γ=

Cp VK S

γ

(13)

where Cp is heat capacity at constant pressure, χ S is the adiabatic compressibility coefficient and K S is the adiabatic bulk modulus since these terms relate to the equivalent terms at constant temperature through: K S χT C p = = (14) KT χ S CV and are defined through: 1 ⎛ dp ⎞ =V ⎜ (15) χS = − ⎟ KS ⎝ dV ⎠ S

Negative thermal expansion (NTE) Although most materials exhibit positive thermal expansion, there are, in fact, some materials which exhibit negative thermal expansion (NTE) in some particular temperature range. When such materials experience an increase in temperature in their NTE range, they contract (i.e. shrink in size) with the result that their density increases.

NTE, is generally considered to be rare, unusual and limited to certain types of structures. Nevertheless, it is important to note that NTE has been shown to take place in a very common material: water. In fact, liquid water exhibits an increase in density between when heated between 273K and 277K whilst the hexagonal form of ice has a NTE coefficient at 45K (Rottger et al. 1994).

This review will provide information on some ‘families’ of materials which appear to possess NTE and discuss the main mechanisms which are responsible for this anomalous behaviour. (). Finally, the review will consider several applications which NTE materials have.

Families of materials which undergo NTE There are a number of classes of materials, in particular the ZrW2O8 family of materials, the AM2O7 family of materials and zeolites / zeolite-like materials. These systems have attracted considerable attention particularly due to the fact that their NTE window is unusually large (i.e. exhibit NTE over a large temperature range). These classes of materials are also of interest as members within the same class exhibit NTE windows with very similar or identical mechanisms occurring.

(i) The ZrW2O8 family: Cubic zirconium tungstate (ZrW2O8) (Evans et al. 1996; Evans et al. 1999a; Evans et al. 1999b) is the best known example of large isotropic negative expansion persisting over a wide temperature range. Although it is thermodynamically stable with respect to ZrO2 and WO3 only at high temperatures (≈1380–1500 K), it can be quenched and is then metastable from the lowest temperatures up to about 1050 K. Over all this range αV is negative. The crystal is cubic, with a rather complex structure (see Fig. 1) containing WO4 tetrahedra and ZrO6 octahedra which are linked in such a way that each ZrO6 unit shares its corners with six

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

different WO4 units, while each WO4 unit shares only three of its corners with ZrO6 units and the remaining oxygen in each WO4 tetrahedron is formally singly coordinated. Other materials in the ZrW2O8 family which share the expansion properties include cubic HfW2O8 (Evans et al. 1996; Mary et al. 1996), ZrMo2O8 (Wilkinson et al. 1999; Allen & Evans 2003), Zr1−xHfxW2O8 and ZrW2−xMoxO8 (Evans 1999); and Zr1−xMxW2O8−y (M = Sc, In, Y) (Nakajima et al. 2002).

All these systems have the same crystal structure as ZrW2O8 and exhibit NTE over a wide temperature range.

NTE in this family is well explained by invoking RUM (Rigid Unit Mode) models. In such cases, certain substructures within the unit remain at a fixed distance to each other and rotate as a whole. The mechanism is explained later on.

Figure 1. The structure of ZrW2O8 at room temperature. The spheres represent oxygen atoms whilst the dark coloured tetrahedra are WO4, whilst the light coloured octahedral represent ZrO6. The rigid polyhedra are very clearly visible. The oxygen atoms act as hinges for these rigid polyhedral units. (Adapted from Barrera et al.

2005). The structure of HfW2O8 is essentially similar, with Hf replacing Zr. The ZrWMoO8 structure is also similar to that of ZrW2O8 but it contains MoO4 units as well as WO4 units.

(ii) The AM2O7 family: The AM2O7 (e.g. A = U, Th, Zr, Hf, Sn; M = P, V) (Taylor 1984) phases are structurally closely related to ZrW2O8 and form a second family of cubic materials which display isotropic negative thermal expansion under certain conditions (Korthuis et al. 1995; Koshrovani et al. 1996). This family is characterised by its distinctive structure which allows for a particular mechanism of NTE which is slightly different than in the case of the ZrW2O8 family. In the AM2O7 family, NTE arises from a deformation mechanism involving coupled 3D rotation of slightly distorted polyhedral units and nonlinear M–O–M bridges. This is essentially a slight modification of the previously mentioned RUM mechanism.

19

(iii) Zeolites and zeolite-like materials: A class of materials where NTE is quite frequent is that of zeolites and similar materials. The list of such materials includes siliceous FAU (Attfield and Sleight 1998), siliceous MWW, ITE and STT (Woodcock et al. 1999a), ISV, STF and MAPO-17 (Lightfoot et al. 2001), RHO (Bieniok & Hammonds 1998, Reisner et al. 2000, Lee et al. 2001), ERI (Attfield & Sleight 1998b, Tao and Sleight 2003b), siliceous CHA (Woodcock et al. 1999b), siliceous IFR (Villaescusa et al. 2001; Woodcock et al. 1999b) and calcined siliceous FER (Bull et al. 2003). A considerable number of these NTE cases are explained through RUMs (see below) but some studies have suggested reservations about these models (Tao and Sleight 2003a), particularly in the case of ERI (Tao and Sleight 2003b) where it has been suggested that the simpler ‘bridged transverse vibration’ mechanism (see below) is sufficient to explain some instances and also the fact that the RUM models must be modified to fit completely. A selection of these zeolites is shown in Fig 3.

(a)

(b)

(c)

(d)

Figure 3: The polyhedral structures of (a) CHA and (b) FAU, (c) ERI and (d) MWW. In all cases the visualisation is in 2X2 cell format showing the (001) plane.

Figure 2: The structure of an AM2O7 member. The large light gray polyhedra represent M2O7 units while the smaller dark gray polyhedra represent AO6 units.

Mechanisms resulting in negative thermal expansion A number of mechanisms have been proposed to explain NTE in different materials. We now review the four principal mechanisms which all operate at the atomic scale and give rise to NTE in a wide range of materials.

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

Some less frequently occurring mechanisms will also be mentioned.

(a) Mechanism 1: Shortening of bond lengths and phase changes Normally, as heat is applied to a structure, the bonds tend to lengthen and subsequently become weaker. However, there exist circumstances where solids which are about to undergo a phase change (to another solid state) exhibit certain changes in their atomic co-ordinates to the effect that certain bonds become stronger (shorter). This effect is usually the result of the formation of a less distorted crystalline system upon heating or the result of a rearrangement of the system as a consequence of a phase change.

NTE behaviour as a result of the formation of a more ordered system upon heating has been found to occur in BaTiO3, PbTiO3 where it is even more pronounced and similar perovskitic materials (Agrawal et al. 1987; Agrawal et al. 1988). In these systems, taking BaTiO3 as an example, one may observe that at room temperature, the TiO6 polyhedra are distorted (and thus the bonds are weaker and longer) but start to become more regular as heat is applied. This will result in a thermal contraction which occurs along the c crystal axis. This extent of thermal contraction is so pronounced that despite the fact that a and b vectors exhibit expansion, the net result is still a unit cell volumetric thermal contraction (i.e. negative αV ).

Figure 4: The idealised ‘crystal perovskite structure’ of BaTiO3.

The TiO6 tetrahedral units (i.e. the units with the O at the vertices and the Ti at the centre) illustrated in this figure are distorted at room temperature and become more regular as the temperature is increased, a process which results in a reduction in volume (i.e.

NTE).

Once the phase shifts from the starting tetragonal state to the cubic phase, conventional thermal expansion once again starts to occur. Thus, the NTE mechanism occurs during the shift from the tetragonal to the cubic phase.

The phenomena of these bond length changes and phase changes are in fact closely related and it is often the case that the bond length changes may be the result of the solid in question undergoing a phase change. Sometimes, the NTE window is also associated with such a phase change. In such a case, as one heats the material, the

20

structure deforms and it may become conducive to contraction. For example, in some cases, reordering of species occurs within the lattice and this reordering is responsible for the NTE behaviour. Phase changes and NTE were correlated and studied in the cases of ZrV2xPxO7 (Korthuis et al. 1995, see Fig. 2), lithium borates (Mathews et al. 1998), ZrW2O8 (Yamamura et al. 2001, see Fig. 1) and related systems such as HfW2O8 (Nakajima et al. 2002) and ZrWMoO8 (Evans et al.

2000). Other studies involving NTE and phase changes were carried out by Ishii and co-workers (Ishii et al.

1999) who studied an ammonated alkali fulleride compound which was reported to exhibit a phase change close to the NTE window whilst Tyagi et al. studied A2(MoO4)3 systems (A = Fe3+ , Cr3+ and Al3+ ) where the transition from the monoclinic to ortho-rhombic modification is accompanied by a large positive expansion followed by NTE. (Tyagi et al. 2002). Also, a phase change was reported to occur in AgI nanowires where the transition from the β to the α form was found to be responsible for NTE in the temperature range 293 433K (Wang et al. 2003).

On the other hand, theoretical investigations by Christensen and co-workers (Christensen et al. 2000) regarding f.c.c. cesium suggest that negative thermal expansion is present but there is no isostructural transition.

Figure 5: The structure of ZrV2-xPxO7. The large light gray polyhedra represent V2-xPxO7 units while the smaller dark gray polyhedra represent ZrO6 units.

(b) Mechanism 2: Bridging Atoms and RUMs Another important mechanism which results in NTE is that involving ‘bridging oxygens’, although it is important to note that the presence of ‘bridging oxygens’ does not automatically imply that the material exhibits NTE: the vast majority of systems containing bridging oxygens, do not, in fact possess the property of NTE. NTE occurring via the ‘bridging oxygens’ mechanism can be explained by considering an M-O-M bond, (where M is usually a metal) which has an angle of 180o, where we may observe a thermal displacement for the bridging oxygen in the axis perpendicular to that of the bond. This

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

21

sideways motion which is shown in Fig. 6, will result in a decrease of the M-M distance. It has long been argued that such transverse vibrational modes (Barron 1957; Blackman 1957) can lead to NTE (White 1993).

This mechanism has been used to explain a NTE window between 25K and 573K for faujasite (FAU) (Attfield & Sleight 1998a). There have been reports of materials where the phenomenon takes place using the O-M-O configuration instead and examples include: cuprite Cu2O (White et al. 1978; Schafer & Kirfel 2002), Ag2O (Tiano et al. 2003; Beccara 2002) and CuScO2 (Li et al. 2002a).

Li and co-workers (Li et al, 2005) describe an experiment designed to understand the phenomenon and provide additional insight.

Figure 6. (a) A higher amplitude in atomic vibrations due to heat results in longer bond lengths which translates into a larger volume.

In (b) the transverse vibratory-type motion of a species leads to reduction in the bond lengths (the dotted structures represent the extremal positions of the vibratory motions). The black and white spots represent species of different types.

Finally, the mechanism for NTE in Zn(CN)2 (Williams 1997) has been described as being very similar with the CN group taking the role of the species performing the transverse thermal motion. An explanation for the NTE in gold cyanide (AuCN) is also quite similar with species performing long wavelength motions perpendicular to the –[Au-(CN)-Au-(CN)]– chain axis (Hibble et al. 2003).

For a material, there will be several different vibration modes at a given temperature and they may not all contribute to NTE. It has been postulated that transverse vibration modes are favoured at low temperatures and in fact several materials such as rubidium halides exhibit NTE at very low temperatures.

Interestingly, correlations can be made between the degree of covalency (rigidity) and the magnitude of the negative thermal expansion. This concept has been used to develop the RUM (or Rigid Unit Modes) (Giddy et al.

1993) and QRUM (Quasi RUM) models. The idea behind these models is actually a ‘simple’ model (Pryde et al.

1996) which considers polyhedra (SiO4, WO4, ZrO6, etc.) linked at their corners by shared oxygens to be rigid. In this manner, large amplitude transverse vibrations of the oxygens can occur only through coupled oscillations of the tetrahedra and octahedra forming the structure. They involve no changes in intra-unit bond distances and angles, and so have relatively large amplitudes and low frequencies. The concept is illustrated in Fig. 7. It has also been used to explain NTE in calcined siliceous ferrierite (Bull et al. 2003), where it is suggested that tetrahedra oscillate via rigid modes of vibration.

Figure 7. The RUM model involves a depiction of the structures as rotating squares where the sides of squares represent rigid bonds and the vertices represent flexible hinges (eg: shared oxygens). The dotted square represents the original area of this 2D RUM.

Detailed analysis is needed to find out whether a given structure can support RUMs and, if so, of what type. The RUM approach has been developed extensively by Dove, Heine and others (Giddy et al. 1993; Pryde et al. 1996; Pryde et al. 1998; Welche et al. 1998; Heine et al. 1999; Dove et al. 2000) and related to the standard Grüneisen theory. When there are many vibrations involving oscillations with only small distortions of the polyhedra, these are termed quasi-rigid unit modes (QRUMs), and can also give rise to negative expansion. A number of NTE materials have been explained through the use of RUM and QRUM models and these include: ZrW2O8 (Pryde et al. 1998; David et al. 1999), ZrV2O7, AlPO4-17 (Attfield & Sleight 1998b), various zeolite frameworks (Attfield & Sleight 1998a; Lightfoot et al. 2001; Bieniok & Hammonds 1998), beta-quartz (Welche et al. 1998), and Sc2(WO4)3 (Weller et al. 2000), In recent years, questions have been raised regarding whether simple RUM models are sufficient to explain NTE in zirconium tungstate (Ravindran & Arora 2003; Cao et al. 2003). Ravindran and co-workers (Ravindran et al. 2001) claim that RUMs alone do not account for NTE in zirconium tungstate but one must consider also other phonons, including the bending modes of the WO4 group. The direct relationship between RUMs and NTE has also been challenged by Tao & Sleight (2003a) who cite several examples where NTE exists regardless of RUMs. The authors concede, however, that there is some form of correlation between the two. It has also been reported (Tao & Sleight 2003b) that the NTE that occurs in AlPO4-17 does not originate solely from RUMs but a very important contribution comes from other lowfrequency vibration modes. The RUM model has also been described as insufficient for completely explaining the NTE in Ag2O (Beccara 2002).

Figure 8. Rotations of rigid unit modes in the zeolite RHO framework type. The rotations are associated with phase transitions.

(Adapted from Bieniok and Hammonds 1998).

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

The concept of RUMs is becoming increasingly popular for explaining NTE in many materials. Indeed, the RUM model is perhaps the easiest mechanism to understand from all those presented in this review, especially to the layman. However, while RUM models do provide a partial explanation of how NTE arises in the materials they are identified in, they fail to give a detailed view of why it arises, from the physical aspect. RUM models are also being used to investigate the intrinsic flexibility in framework structures and the effects that property has in structures and materials (Goodwin 2006).

(c) Mechanism 3: Magnetostriction Anomalous thermal expansion behaviour, including NTE has also been shown to exist in various cases of alloys and magnetic systems. Usually, for such systems, the NTE window occurs at low temperatures where vibrational contributions are small. In magnetostriction, what occurs is that in the wholly ordered ferromagnetic state the electron spins are aligned in parallel and are associated with a specific volume. The application of pressure to a state of partial ferromagnetic order causes the parallel spins to become less ordered, in turn contributing negatively to the thermal expansion (refer to Equation 6). This almost cancels the normal vibrational positive contributions, giving very low expansion or even negative expansion. At very low temperatures, the electrons are organised in domains which take up a certain volume. Upon heating, different energy levels for electrons become viable and when these are reached, the electronic configuration which is adopted is associated with a smaller volume, hence negative thermal expansion occurs.

Perhaps the most famous example of systems which exhibits anomalous thermal expansion behaviour is the ferromagnetic Invar† alloy C0.65Ni0.35 and alloys of similar composition (Guillaume 1897; Schlosse et al. 1971; Chikazumi 1979; Collocott & White 1986; Manosa 1991; Manosa 1992; Saunders et al. 1993; van Schilfgaarde et al. 1999; Kainuma 2002). Other systems which exhibit NTE include the orthorhombic antiferromagnetic CuCl2•H2O (Harding et al. 1971), CeNiSn (Aliev et al.

1993) and the rare earth metal Holmium (White 1989).

Materials such as Lu2Fe17 and Y2Fe17 have also been shown to exhibit negative thermal expansion below approximately 400 K (Gignoux et al. 1979). Tino and Iguchi (1983) investigated the possibility of NTE in FePd alloys and correlated the thermal expansion behaviour with that of Invar. The structure and magnetic properties of Y2Al3Fe14-xMnx compounds (Hao et al. 2001) have been investigated by means of X-ray diffraction and NTE behaviour was found and attributed to magnetostriction.

Tb2Fe16Cr (Hao et al. 2005a), thermoclinic perovskite MnF3 (Hunter et al. 2004) have also been shown to possess the property of NTE due to magnetostriction.

Negative thermal expansion and spontaneous †

The alloy INVAR is so called because its thermal expansion coefficient is approximately zero and therefore it has an invariable volume when subjected to changes in temperature.

22

magnetostriction of Tb2Fe16.5Cr0.5 (Hao et al. 2005b), Dy2AlFe10Mn6 (Hao et al. 2005c), Dy2AlFe13Mn3 (Hao et al. 2005d) and Y2Fe16Al (Hao et al. 2005e) have been documented.

The magnetic effects on NTE exhibiting molecule-based magnets M[N(CN)2]2 (M = Co, Ni) (Kmety et al. 1999) have also been investigated. For example, lanthanum manganite (Huang et al. 1997), which is another NTE material, has been studied with respect to magnetostriction. Also, a negative thermal expansion coefficient has been measured for b.c.c. Cr-Mn alloys (Shiga et al. 1986) and partly explained by magnetostriction.

(d) Mechanism 4: Electronic effects The volume of a system depends partly on the electronic configuration and changes in this configuration may lead to changes in the volume the system occupies. The volume of a material may in certain cases vary due to transfer of electrons, a transition which occurs at particular temperatures (e.g.: excitation of electrons to the previously unfilled conduction band). This volume change as a result of electron transfer does not necessarily result in NTE of the material; indeed below superconductive transition temperature, the positive expansion is enhanced; for Nb the expansion is decreased and at still lower temperatures becomes negative; and for Ta it is negative immediately below the transition temperature (White 1962).

The electronic configuration of a material may also become altered, such that the material has a NTE window which is associated with the availability of two different electronic configurations for a particular temperature window (Arvanitidis et al. 2003; Sleight 2003). When the material is heated, it may adopt a different electronic configuration which results in a decrease of the volume (see Fig. 9). This has been observed for samarium fulleride Sm2.72C60 from 4.2 K up to 32 K, where there is an isosymmetric phase transition (Arvanitidis et al. 2003) and also in ytterbium fulleride (Margadonna et al. 2005) (see Fig. 10). This phenomenon is also thought to be responsible for negative thermal expansion below 60 K in YbCuAl (Mattens et al. 1980) and between 4.2 and 350 K in Sm0.75Y0.25S (Mook and Holtzberg 1981), and for the approximately zero thermal expansion of the metallic compound YbGaGe (Salvador et al. 2003). The effect may also occur in pure metals: a fcc phase of Pr has α ≈ −20 × 10−6 K−1 between 550 and 700 K, attributed to electronic excitation from 4f to 5d (Kuznetsov et al.

2003). In the case of lanthanum manganite (Huang et al.

1997), the existence of different valence states of Mn and its possible relation to NTE in the material was also pointed out. NTE was also observed in PrGaO3 (see Fig. 10), and it can also be attributed to an interaction between the phonon vibrations and the electronic excitations of the Pr ion (Savytskii et al. 2003). Investigations carried out on b.c.c. Cr-Mn alloys (Shiga et al. 1986) indicate that the phenomenon may also be partly responsible for NTE

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

in these materials. The phenomenon may also provide a partial explanation for the thermal contraction that occurs in gold nanoparticles at about 125K (Li et al. 2002b; Li et al. 2003).

Figure 9: (a) This is a schematic diagram for the relevant electronic configuration of ytterbium fulleride at very low temperatures. (b) Upon heating (in the region of 60K), electron transfer from the 4f band of Yb to the T1u band of the C60 occurs. This is accompanied by a change of the valence of Yb, such that the ionic radius decreases, (The diagram is adapted from Margadonna et al. 2005).

The process, thus, results in NTE.

23

negative. This expansion can also be strongly affected by the presence of magnetic fields. Such a case is illustrated in the investigations carried out on magnesium boride (Anshukova et al. 2003a; Anshukova et al. 2003b; Anshukova et al. 2003c).

‘Heavy fermionic’ materials (which are compounds where one of the constituent atoms has a partially filled 4f or 5f shell with high electron correlation) can present some of the most spectacular effects in thermal expansion in solids, especially at low temperatures (Ott et al. 1987).

For example, the measured expansion αV of two different polycrystalline samples (Andres et al. 1975; Ribault et al.

1979) of the cubic material CeAl3 below 1K have been reported to be negative and very large. As regards CeInCu2 (de Visser et al. 1993) it was confirmed that the development of antiferromagnetic order in a strongly correlated electron system is correlated to a strong reduction of the coefficient of volume expansion, in accordance with decreasing hybridisation. The work of Aliev and co-workers (Aliev et al. 1993) on CeNiSn is also important to consider in this respect.

The behaviour of most heavy fermion materials is further complicated by anisotropy, and also by magnetic and superconducting transitions and thus it is highly dependent on the composition.

Figure 10: Measurements of the cell volume and parameters as they vary with temperature for PrGaO3. (The measurements and graphs are taken from Vasylechko et al. 2005). The NTE window between circa 12K and 50K is evident.

An effect which is also involved in the mechanisms controlling thermal expansion and that can make them even more complex is that of superconductivity particularly at temperatures in the immediate vicinity of the superconducting transition. This is the case for the previous cited examples of Nb and Ta where cooling below the superconducting temperature results immediately in a negative thermal expansion coefficient for the latter while in the former case, the coefficient of thermal expansion decreases gradually until it becomes

(e) Other mechanisms There are other ‘mechanisms’ which result in NTE but do not relate to a class of systems. Quantum tunnelling is one such phenomenon. Tunnelling occurs for some substitutional impurities in alkali halides, and can give rise to either positive or negative thermal expansion. For example, replacement of the anion by 0.03% (CN)- in KBr causes the thermal expansion to become negative below 0.5 K, with a very large Grüneisen function of about −300 at 0.1 K (Dobbs et al. 1986). In the case of systems where inter-molecular forces are weak, certain molecules in solids may sometimes behave as rotors. This may occur in the case of symmetrically equivalent H atoms, such as in CH4 (with different nuclear spin species), where investigations have yielded evidence of negative thermal expansion below 9K (Heberlein and Adams 1987). If the temperature in some systems is sufficiently low, the rotations may cease and the molecules oscillate at fixed orientations. This is the mechanism postulated for NTE in fulleride at and below 3.4K (Aleksandrovskii et al. 1997).

Applications Materials exhibiting NTE are not only of interest due to the fact that this behaviour is highly anomalous, but also due to the fact that this anomalous behaviour can be exploited in many practical applications. For example, − In the composites industry, where negative thermal expansion materials are used as components of composites to adjust the overall thermal expansion of composites to some particular value. The composite

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

−

−

−

−

may be metal, polymer, or oxide-based. Predicting the thermal expansion of composites based on component properties is somewhat complex. Properties other than the thermal expansion of the components are important, in particular bulk or elastic moduli. It is interesting to note that a special system of carbon fibre composites and metals, with an extremely negative value of the thermal expansion coefficient, has been developed. The value is about three times that of steel, but with a negative sign. A further advantage of this system is its very low thermal conductivity and high compressive strength (Hartwig 1995).

In the production of symmetric laminated beams with useful thermal deformation properties: Wetherhold & Wang (1995, 1996) suggest that by combining laminae having a positive coefficient of thermal expansion (fortwith referred to as CTE) with laminae having a negative CTE, one can control or tailor the laminate CTE and/or the laminate thermal curvature.

The design of the beam can be naturally expressed in terms of property ratios and lamina thickness ratios. A three-layer beam offers one thickness ratio, and thus can be used either to control laminate CTE or eliminate thermal curvature. A five-layer beam offers two thickness ratios, and can be used to control the CTE and eliminate curvature.

In the construction of mechanically enhanced capillary columns through the deposition of TiO2doped silica (which has a negative thermal expansion coefficient) on the inside and outside of the silica tube results in the tube becoming compressed as it is heated, reducing the propagation of surface flaws (Berthou et al. 1993). Thus, NTE can be exploited in the design of surface-flaw-free tubes.

In the electronics area, there is a need for substrates and heat sinks that match the thermal expansion of Si. There are currently several active efforts in this area using ZrW2O8 to reduce thermal expansion. In the heat sink application, Cu/ZrW208 composites have been successfully made matching the thermal expansion of Si over at least a several hundred degree temperature range (Holzer & Dunand 1997).

In the production of ‘superior’ high field superconducting solenoid magnets: Such magnets sometimes quench by wire motion induced by electromagnetic force. It is suggested that the quick wire motions may be constrained by a high strength polyethylene fibre reinforced plastic (DFRP) bobbin with a negative thermal expansion coefficient and a low frictional coefficient (Yamanaka et al. 2002).

Dyneema–glass hybrid composite fibre reinforced plastic (DGFRP) has negative thermal expansion, low frictional coefficient and high thermal conductivity

24

−

−

−

−

and its use as material of a coil bobbin has also been described in literature (Takeo et al. 2003).

In the making of interfaces exhibiting good adhesion properties: It has been reported that additional reinforcing fibres with a negative thermal expansion coefficient such as Kevlar fibres are helpful to strengthen the reliability of the interface and enhance the actuating ability of SMA (shape memory alloy) hybrid composites (Zheng et al. 2005).

In medical applications: Another interesting application for NTE materials is that of adjusting the thermal expansion of the white composites used in teeth fillings. It is suggested that the thermal expansion of teeth and conventional fillings mismatch and this may result in failure (Versluis et al. 1996).

Several groups working on such filling materials have been supplied with zirconium tungstate.

In the production of materials which do not change shape when heated: Zero thermal expansion is, of course, of profound interest. One of the biggest uses of such materials is as substrate materials for mirrors in various telescope and satellite applications (Collins and Richter 1995).

In the photonics sector, the use of NTE materials has been suggested in chirped fibre gratings (see Fig. 12).

Wei and co-workers (Wei et al. 2001) theoretically analysed and experimentally demonstrated a simple method for adjusting the chirp of chirped fibre gratings by means of a temperature method, while the central wavelength is temperature insensitive. In this work, chirped fibre grating with tapered cross-section area was mounted under tension in a negative thermalexpansion coefficient material. Similar work has been carried out in this field by other workers including Mavoori et al. (1999) and Ngo et al. (2003), see (Fig. 11). Nishii et al. (2003) also report that channel waveguides with Bragg gratings have been fabricated on glass ceramic substrates with negative thermal expansion coefficient.

Figure 11: A schematic cross section representation of a Tunable Fibre Bragg Grating dispersion compensator, which involves a material with a NTEC encircling the coated fibre. (Diagram adapted from Ngo et al. 2003).

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

Conclusions In this review we gave an overview of existing work relating to materials exhibiting negative thermal expansion. The subject of NTE is by no means a new subject but there still exists much space for further work.

In particular, the contribution to negative thermal expansion from certain electronic effects, especially at very low temperatures, is still not completely understood.

Since a considerable number of NTE cases occur at very low temperatures, (and such conditions require huge efforts to bring about), it is evident that computer-based molecular modelling of such systems would be a valuable contribution to understanding how this counterintuitive behaviour arises. As computational power is constantly on the increase, quantum-based computer modelling of such systems will become more viable and as more information about this phenomenon becomes available, it is envisaged that a more complete picture of the phenomenon will emerge. Such research is also expected to yield a more extensive documentation of natural materials which possess this property. Furthermore, this useful property of NTE may in the future be engineered in structures as well as in synthetic materials, perhaps even resulting in the construction of structures and materials with tailor-made negative coefficients of thermal expansion at desired temperature ranges.

Acknowledgement: The financial support of the Malta Council of Science of Technology (MCST) through their RTDI programme is gratefully acknowledged.

References Agrawal D. K., Roy R. and McKinstry H. A., (1987), Ultra low thermal-expansion phases - substituted PMN Perovskites, Mater. Res. Bull., 22, 83-88.

Agrawal D. K., Halliyal A. and Belsick J., (1988), Thermal-expansion of ceramics in Pb(Zn1/3Nb2/3)O3-based solid-solution systems, Mater. Res. Bull., 23, 159-164.

Aleksandrovskii A. N., Eselson V. B., Manzhelii V. G., Udovidchenko B. G., Soldatov A. V. and Sundqvist B., (1997), Low Temp. Phys., 23, 943.

Aliev F. G., Villar R., Vieira S., and Lopez de la Torre M. A., Scolozdra R. V. and Maple M. B., (1993), Energy-gap of the ground-state of CeNiSn caused by local and long-range magnetic-moment interactions, Physical Review B, 47, 769-772.

Allen S. & Evans J. S. O., (2003), Negative thermal expansion and oxygen disorder in cubic ZrMo2O8, Physical Review B, 68, art:134101.

Andres K., Graebner J. E. and Ott H. R., (1975), Phys. Rev. Lett., 35, 1779-1782.

Anshukova N. V., Bulychev B. M., Golovashkin A. I., Ivanova L. I., Krynetskii I. B., Minakov A. A. and Rusakov A. P., (2003a), Anomalous LowTemperature Behavior of the Thermal

25

Characteristics of MgB2, J. Exp. Theor. Phys., 97, 70-77 Anshukova N. V., Bulychev B. M., Golovashkin A. I., Ivanova L. I., Krynetski I. B. and Rusakov A. P., (2003b), Low-Temperature Anomalies in MgB2 thermal Expansion; Physics of the Solid State, 45, 6–10.

Anshukova N. V., Bulychev B. M., Golovashkin A. I., Ivanova L. I., Minakov A. A., and Rusakov A. P., (2003c), Low-Temperature Anomalies in the Specific Heat and Thermal Conductivity of MgB2, Physics of the Solid State, 45, 1207– 1212.

Arvanitidis J., Papagelis K., Margadonna S., Prassides K. and Fitch A. N., (2003), Temperature-induced valence transition and associated lattice collapse in samarium fulleride, Nature, 425, 599-602.

Attfield M. P. and Sleight A. W., (1998a), Strong negative thermal expansion in siliceous faujasite, Chem. Commun., (5), 601-602.

Attfield M. P. and Sleight A. W.; (1998b), Exceptional negative thermal expansion in AlPO4-17, Chem.

Mater., 10, 2013-2019.

Barrera G. D., Bruno J. A. O., Barron T. H. K. and Allan N. L., (2005) Negative Thermal Expansion, J. of Phys.: Cond. Mat., 17, R217-R52.

Barron T. H. K., (1957), Grüneisen parameters for the equation of state of solids, Ann. Phys., 1, 77-90.

Beccara S. A., Dalba G., Fornasini P., Grisenti R. and Sanson A., (2002), Local thermal expansion in a cuprite structure: The case of Ag2O, Phys. Rev. Lett., 89, art:025503.

Berthou H., Neumann V., Dan J. P. Hintermann H. E., (1993), Mechanically Enhanced Capillary Columns, Surface & Coatings Technology, 61, 93-96.

Bieniok A., Hammonds K. D., (1998), Microporous and Mesoporous Materials, 25, 193–200.

Blackman M., (1957), On the thermal expansion of solids, Proc. Phys. Soc., B70, 827-832.

Bull I., Lightfoot P., Villaescusa L. A., Bull L. M., Gover R. K. B., Evans J. S. O., and Morris R. E., (2003), An X-ray Diffraction and MAS NMR Study of the Thermal Expansion Properties of Calcined Siliceous Ferrierite, J. Am. Chem. Soc., 125, 4342-4349.

Cao D., Bridges F., Kowach G. R., and Ramirez A. P., (2003), Correlated atomic motions in the negative thermal expansion material ZrW2O8: A local structure study, Physical Review B, 68, art:014303.

Chikazumi S., (1979), INVAR anomalies, J. Magn.

Magn. Mater., 10, 113-119.

Christensen N. E., Boers D. J. and van Velsen J. L., Novikov D. L., (2000), Negative thermal expansion coefficient and isostructural transition in fcc cesium; Physical Review B, 1, r3764r3767.

Collins E. G. & Richter S., (1995), Linear-QuadraticGaussian-Based Controller-Design for Hubble-

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

Space-Telescope, Journal of Guidance Control and Dynamics, 18, 208-213.

Collocott S. J. and White G. K., (1986), Thermalexpansion and heat-capacity of some stainlesssteels and feni alloys, Cryogenics, 26, 402-405.

David W. I. F., Evans J. S. O. and Sleight A. W., (1999), Direct evidence for a low-frequency phonon mode mechanism in the negative thermal expansion compound ZrW2O8, Europhys. Lett., 46, 661-666. de Visser A., Bakker K. and Pierre J., (1993), Anomalous negative thermal-expansion of CeInCu2, Physica B, 186-188, 577-579.

Dobbs J. N., Foote M. C. & Anderson A. C., (1986), Low-Temperature Thermal-Expansion and Specific-Heat of KBr Containing CN, Physical Review B, 33, 4178-4183.

Dove M. T, Hammonds K. D., Harris M. J., Heine V., Keen D. A., Pryde A. K. A., Trachenko K. and Warren M. C., (2000), Amorphous silica from the Rigid Unit Mode approach, Miner. Mag., 64, 377-388.

Evans J. S. O., (1999), Negative thermal expansion materials, J. Chem. Soc. Dalton Trans., (19), 3317-3326.

Evans J. S. O., Mary T. A., Vogt T., Subramanian M. A. and Sleight A. W., (1996), Negative Thermal Expansion in ZrW2O8 and HfW2O8, Chemistry of Materials, 8, 2809-2823/ Evans J. S. O., Jorgensen J. D., Short S., David W. I. F., Ibberson R. M. and Sleight A. W., (1999a), Thermal expansion in the orthorhombic gamma phase of ZrW2O8, Phys. Rev. B, 60, 1464314648.

Evans J. S. O., David W. I. F. and Sleight A. W., (1999b), Structural investigation of the negative thermal Acta expansion material ZrW2O8, Crystallographica B, 55, 333-340.

Evans J. S. O., Hanson P. A., Ibberson R. M., Duan N., Kameswari U., and Sleight A. W., (2000), LowTemperature Oxygen Migration and Negative Thermal Expansion in ZrW2-xMoxO8, J. Am. Chem. Soc., 122, 8694-8699.

Giddy A. P., Dove M. T., Pawley G. S. and Heine V., (1993), The determination of rigid-unit modes as potential soft modes for displacive phasetransitions in framework crystal-structures, Acta Crystallogr., A49, 697-703.

Gignoux D., Givord D., Givord F. and Lemaire R., (1979), Invar properties in the rare earth-3d transition-metal alloys, J. Magn. Magn. Mater., 10, 288-293.

Goodwin A. L., (2006), Rigid unit modes and intrinsic flexibility in linearly bridged framework structures, Physical Review B, 74, 134302.

Grüneisen E., (1926), Handb. Phys., 10, 1.

Guillaume C. E., (1897), C. R. Acad. Sci., 125, 235.

Hao Y., Gao Y., Wang B., Qu J., Li Y., Hu J. and Deng J., (2001), Negative thermal expansion and magnetic properties of Y2Al3Fe14-xMnx

26

compounds; Applied Physics Letters, 78, 32773279.

Hao Y., Zhao M., Zhou Y., Hu J., (2005a), Negative thermal expansion and spontaneous volume magnetostriction of Tb2Fe16Cr compound, Scripta Materialia, 53, 357–360.

Hao Y., Zhou Y., and Zhao M., (2005b), Negative thermal expansion and spontaneous magnetostriction of Tb2Fe16.5Cr0.5 compound, Journal Of Applied Physics, 97, art:116102.

Hao Y., Zhou Y., and Zhao M., (2005c), Negative thermal expansion and spontaneous magnetostriction of Dy2AlFe10Mn6 compound, Advanced Engineering Materials, 7, 517-520.

Hao Y., Zhou Y., and Zhao M., (2005d), Spontaneous magnetostriction of Dy2AlFe13Mn3 compound, Chinese Physics, 14, 1449-1452.

Hao Y., Zhou Y., and Zhao M., (2005e), Spontaneous magnetostriction of Y2Fe16Al compound, Chinese Physics, 14, 818-820.

Harding G. L., Lanchester P. C. and Street R., (1971), Low temperature magnetic thermal expansion of J. Phys. C: Solid State Phys., CuCl2. 2H2O, 4, 2923.

Hartwig G., (1995), Support Elements with Extremely Negative Thermal-Expansion, Cryogenics, 35, 717-718.

Heberlein D. C. and Adams E. D., (1970), J. Low Temp.

Phys., 3, 115.

Heine V., Welche P. R. L. and Dove M. T. (1999), Geometrical origin and theory of negative thermal expansion in framework structures, J. Am. Ceram. Soc., 82, 1793-1802.

Hibble S. J., Hannon A. C., and Cheyne S. M., (2003), Structure of AuCN Determined from Total Neutron Diffraction, Inorg. Chem., 42, 47244730.

Holzer H. & Dunand D., (1997), Processing, structure and thermal expansion of metal matrix composites containing zirconium tungstate, 4th International Conference on Composite Engineering, Hawaii.

Huang Q., Santoro A., Lynn J. W., Erwin R. W., Borchers J. A., Peng J. L. and Greene R. L., (1997), Structure and magnetic order in undoped lanthanum manganite, Physical Review B, 55, 14987-14999.

Hunter B. A., Kennedy B. J. and Vogt T., (2004), Magnetostriction in a simple trivalent manganese perovskite; Physical Review B, 69, art:020410.

Ishii K., Watanuki T., Fujiwara A., and Suematsu H., Iwasa Y., Shimoda H., Mitani T., Nakao H., Fujii Y., Murakami Y. and Kawada H., (1999), Structural phase transition in the ammoniated alkali C60 compound (NH3)K3C60, Physical Review B, 1, 3956-3960.

Kainuma R., Wang J. J., Omori T., Sutou Y., and Ishida K., (2002), Invar-type effect induced by coldrolling deformation in shape memory alloys, Appl. Phys. Lett., 80, 4348-4350.

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

Kmety C. R., Manson J. L., Huang Q., Lynn J. W. Erwin R. W., Miller J. S. and Epstein A. J., (1999), Collinear ferromagnetism and spin orientation in the molecule-based magnets M[N(CN)2]2 (M=Co,Ni); Physical Review B, 60, 60-63.

Korthuis V., Khosrovani N., Sleight A. W., Roberts N., Dupree R. and Warren Jr. W. W., (1995), Negative Thermal Expansion and Phase Transitions in the ZrV2-xPxO7 Series, Chem.

Mater., 7, 412-417.

Koshrovani N., Korthuis V., Sleight A. W. & Vogt T., (1996), Inorg. Chem., 35, 485-489.

Kuznetsov A. Y., Dmitriev V. P., Bandilet O. I. and Weber H. P., (2003), High-temperature fcc phase of Pr: Negative thermal expansion and intermediate valence state, Phys. Rev. B, 68, art:064109 Lee Y., Reisner B. A., Hanson J. C., Jones G. A., Parise J. B., Corbin D. R., Toby B. H., Freitag A., Larese J. Z., (2001), New Insight into Cation Relocations within the Pores of Zeolite Rho: In Situ Synchrotron X-Ray and Neutron Powder Diffraction Studies of Pb- and Cd-Exchanged Rho, J. Phys. Chem. B, 105, 7188-7199.

Li J., Yokochi A., Amos T. G. and Sleight A. W., (2002a), Strong negative thermal expansion along the O-Cu-O linkage in CuScO2, Chem.

Mater., 14, 2602-2606.

Li W., Wu S. Y., Yang C. C., Lai S. K., Lee K. C., Huang H. L. and Yang H. D., (2002b), Thermal Contraction of Au Nanoparticles, Physical Review Letters, 89, art:135504.

Li W., Wu S. Y., Yang C. C., Tsao F.C., Lai S. K., and Lee K. C., (2003), Quantum Thermal Contraction of Au Nanoparticles, Synthetic Metals, 135-136, 811-812 .

Li J., Sleight A. W., Jones C. Y. and Toby B.H., (2005), Trends in negative thermal expansion behavior for AMO2 (A = Cu or Ag; M = Al, Sc, In, or La) compounds with the delafossite structure, Journal of Solid State Chemistry, 178, 285.

Lightfoot P., Woodcock D. A., Maple M. J., Villaescusa L. A. and Wright P. A., (2001), The widespread occurrence of negative thermal expansion in zeolites, J. Mater. Chem., 11, 212-216.

Margadonna S., Arvantidis J., Papagelis K. and Prassides K., (2005), Negative Thermal Expansion in the Mixed Valence Ytterbium Fulleride, Y2.75C60, Chem. Mater., 17, 4474-4478.

Mary T. A., Evans J. S. O., Vogt T. & Sleight A. W., (1996), Science, 272, 90-92.

Manosa L., Saunders G. A., Rahdi H., Kawald U., Pelzl J., and Bach H., (1991), Longitudinal acoustic mode softening and Invar behaviour in Fe72Pt28, J. Phys.: Condens. Matter, 3, 2273-2278.

Manosa L., Saunders G. A., Rahdi H., Kawald U., Pelzl J., and Bach H., (1992), Acoustic-mode vibrational anharmonicity related to the anomalous thermal expansion of Invar iron alloys, Physical Review B, 45, 2224-2236.

27

Mathews M. D., Tyagi A. K. and Moorthy P. N., (1998), High-temperature behaviour of lithium borates Part II: High-temperature X-ray diffractometric and dilatometric studies; Thermochimica Acta, 319, 113-121.

Mattens W .C. M., Hölscher H., Tuin G. J. M., Moleman A. C. and de Boer F. R., (1980), Thermalexpansion and magneto-volume effects in the mixed-valent compound YbCuAl, J. Magn.

Magn. Mater., 15–18, 982.

Mavoori H., Jin S., Espindola R. P. & Strasser T. A., (1999), Enhanced thermal and magnetic actuations for broad-range tuning of fibre Bragg grating-based reconfigurable add-drop devices, Optics Letters, 24, 714-716.

Mook H. and Holtzberg F., (1981), Valence Fluctuations in Solids, ed: Falicov L. M., Hanke W. and Maple M. B. (Amsterdam: North-Holland) p 113.

Nakajima N., Yamamura Y. and Tsuji T., (2002), Phase transition of negative thermal expansion Zr1– xHfxW2O8 solid solutions, Journal of Thermal Analysis and Calorimetry, 70, 337–344.

Ngo N. Q., Li S. Y., Zheng R. T., Tjin S. C. & Shum P., (2003), Electrically Tunable Dispersion Compensator with Fixed Center Wavelength using Fibre Bragg Grating, Journal of Lightwave Technology, 21, 1568-1575.

Nishii J., Kintaka K., Nishiyama H. & Takahashi M., (2003), Photosensitive and athermal glasses for optical channel waveguides, Journal of NonCrystalline Solids, 326, 464-471.

Ott H. R., (1987), Prog. Low Temp. Phys., 11, 215.

Pryde A. K. A., Hammonds K. D., Dove M. T., Heine V., (1996), Gale J. D. and Warren M.C., (1996), Origin of (1996),the negative thermal expansion in ZrW2O8 and ZrV2O7 , J. Phys.: Condens.

Matter, 8, 10973-10982.

Pryde A. K. A., Dove M. T. and Heine V., (1998), Simulation studies of ZrW2O8 at high pressure, J. Phys.: Condens. Matter, 10, 8417-8428.

Ravindran T. R., Arora A. K. and Mary T. A., (2001), High-pressure Raman spectroscopic study of zirconium tungstate, J. Phys.: Condens. Matter, 13, 11573–11588.

Ravindran T. R. and Arora A. K., (2003), Anharmonicity and negative thermal expansion in zirconium tungstate, Physical Review B, 67, art:064301.

Reisner B. A., Lee Y., Hanson J. C., Jones G. A., Parise J. B., Corbin D. R., Toby B. H., Freitag A., Larese J. Z. and Kahlenbergbf V., (2000), Understanding negative thermal expansion and ‘trap door’ cation relocations in zeolite rho, Chem. Commun., 2221–2222 Ribault M., Benoit A., Flouquet J. and Palleau P., (1979), Thermal-expansion of Al3Ce at very lowtemperatures, J. Physique Lett., 40, L413-L417.

Rottger K., Endriss A., Ihringer J., Doyle S. and Kuhs W. F., (1994), , Acta Crystallogr., Sect. B, 50, 644..

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

Salvador J. R., Guo F., Hogan T. and Kanatzidis M. G., (2003), Nature, 425, 702.

Savytskii D., Vasylechko L., Senyshyn A., Matkovskii A., Sanjuan M. L., Bismayer U., and Berkowski M., (2003), Low-temperature structural and Raman studies on rare-earth gallates, Physical Review B, 68, art:024101.

Saunders G. A., Senin H. B., Sidek H. A. A. and Pelzl J., (1993), Third-order elastic constants, vibrational anharmonicity, and the Invar behavior of the Fe72Pt28 alloy, Physical Review B, 48, 15801-15806.

Schafer W. and Kirfel A., (2002), Neutron powder diffraction study of the thermal expansion of cuprite, Appl. Phys. A, 74, art:S1010.

Schlosse W. F., Graham G. M. and Meincke P. P. M., (1971), Temperature and magnetic field dependence of forced magnetostriction and thermal expansion of INVAR, J. Phys. Chem.

Solids, 32, 927.

Shiga M., Miyake M. and Nakamura Y., (1986), Giant Negative Thermal Expansion in Mn-Rich bcc Cr-Mn Alloys, J. Phys. Soc. Japan, 55, 22902295.

Sleight A., (2003), Materials science - Zero-expansion plan, Nature, 425, 674-676.

Takeo M., Sato S., Matsuo M., Kiss T., Takao T., Yamanaka A., Kashima T., Mito T. and Minamizato K. (2003), Dependence on winding tensions for stability of a superconducting coil, Cryogenics, 43, 649-658.

Tao J. Z. and Sleight A. W., (2003a), The role of rigid unit modes in negative thermal expansion, Journal of Solid State Chemistry, 173, 442– 448.

Tao J. Z. and Sleight A. W., (2003b), Free energy minimization calculations of negative thermal expansion in AlPO4-17, J. Phys. Chem. Solids, 64, 1473-1479.

Taylor D., (1984), Thermal-Expansion Data .3.

Sesquioxides, M2O3, With The Corundum And The A-, B- And C- M2O3 Structures, Transactions and Journal of the British Ceramic Society, 83, 92-98.

Tiano W., Dapiaggi M. and Artioli G., (2003), Thermal expansion in cuprite-type structures from 10 K to decomposition temperature: Cu2O and Ag2O, J. Appl. Crystallogr., 36, 1461-1463.

Tino Y. and Iguchi Y., (1983), Zero or negative thermalexpansion of Fe-Pd alloys and the cause of Invar characteristic, Journal of Magnetism and Magnetic Materials, 31-34, 117-118.

Tyagi A. K., Achary S. N. and Mathews M. D., (2002), Phase transition and negative thermal expansion in A2(MoO4)3 system (A=Fe3+ , Cr3+ and Al3+ ); Journal of Alloys and Compounds, 339, 207210. van Schilfgaarde M., Abrikosov I. A. and Johansson B., (1999), Origin of the Invar effect in iron-nickel alloys, Nature, 400, 46-49.

28

Vasylechko L., Pivak Ye., Senyshyn A., Savytskii D., Berkowski M., Borrmann H., Knapp M & Paulmann C., (2005), Crystal Structure and Thermal Expansion of PrGaO3 in the temperature range 12-1253K, J. Sol. Stat. Chem, 178, 270-278.

Versluis A., Douglas W. H. & Sakaguchi R. L., (1996), Thermal expansion coefficient of dental composites measured with strain gauges, Dental Materials, 12, 290-294.

Villaescusa L. A., Lightfoot P., Teat S. J. and Morris R. E., (2001), Variable-Temperature Microcrystal X-ray Diffraction Studies of Negative Thermal Expansion in the Pure Silica Zeolite IFR, J. Am. Chem. Soc., 123, 5453-5459.

Wang Y., Ye C., Wang G., Zhang L., Liu Y. and Zhao Z., (2003), In situ x-ray diffraction study on AgI nanowire arrays, Applied Physics Letters, 82, 4253-4255.

Wei Z. X., Yu Y. S., Xing H., Zhou Z. C., Wu Y. D., Zhang L., Zheng W. & Zhang Y. S., (2001), Fabrication of chirped fibre grating with adjustable chirp and fixed central wavelength, IEEE Photonics Technology Letters, 13, 821823.

Welche P. L. R., Heine V. and Dove M. T., (1998), Negative thermal expansion in beta-quartz, Phys. Chem. Miner., 26, 63-77.

Weller M. T., Henry P. F. and Wilson C. C., (2000), An analysis of the thermal motion in the negative thermal expansion material Sc2(WO4)3 using J. Phys. isotopes in neutron diffraction, Chem. B, 104, 12224-12229.

Wetherhold R. C. & Wang J. Z., (1995), Tailoring Thermal Deformation by Using Layered Beams, Composites Science and Technology, 53, 1-6.

Wetherhold R. C. & Wang J. Z., (1996), Controlling thermal deformation by using laminated plates, Composites Part B-Engineering, 27, 51-57.

White G. K., (1962), Thermal expansion of vanadium, niobium, and tantalum at low temperatures, Cryogenics, 2, 292-296.

White G. K., (1989), Phase-transitions and the thermalexpansion of Holmium, J. Phys.: Condens.

Matter, 1, 6987-6992.

White G. K., (1993,), Solids - thermal-expansion and contraction, Contemp. Phys., 34, 193-204.

White M. A., Lushington K. J. and Morrison J. A., (1978), Orientational ordering and tunneling states in solid CH2D2 as deduced from lowtemperature heat-capacity and neutron crosssection measurements, J. Chem. Phys., 69, 42274233.

Wilkinson A. P., Lind C. & Pattanaik S., (1999), A new polymorph of ZrW2O8 prepared using nonhydrolytic sol-gel chemistry, Chemistry of Materials, 11, 101-108.

Williams D. J., Partin D. E., Lincoln F. J., Kouvetakis J. and O’Keeffe M., (1997), The disordered crystal structures of Zn(CN)2 and Ga(CN)3, J. Solid State Chem., 134, 164–169.

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Grima JN et al.: Negative Thermal Expansion

Woodcock D. A., Lightfoot P., Wright P. A., Villaescusa L. A., Diaz-Cabanas M. and Camblorb M. A., (1999a), Strong negative thermal expansion in the siliceous zeolites ITQ-1, ITQ-3 and SSZ-23, J. Mater. Chem., 9, 349–351.

Woodcock D. A., Lightfoot P., Villaescusa L. A., DiazCabanas M., Camblorb M. A. and Engberg D., (1999b), Negative Thermal Expansion in the Siliceous Zeolites Chabazite and ITQ-4: A Neutron Powder Diffraction Study, Chem.

Mater., 11, 2508-2514.

Yamamura Y., Nakajima N., and Tsuji T., (2001), Calorimetric and x-ray diffraction studies of ato- b structural phase transitions in HfW2O8 and ZrW2O8, Physical Review B, 64, art:184109.

Yamanaka A., Kashima T., Nago S., Hosoyama K., Takao T., Sato S. & Takeo M., (2002), Coil bobbin composed of high strength polyethylene fibre reinforced plastics for a stable high field Physica C: superconducting magnet, Superconductivity and its Applications, 372, 1447-1450.

Yamanaka A., Kashima T., Mito T. & Minamizato K., (2003), Dependence on winding tensions for stability of a superconducting coil, Cryogenics, 43, 649-658.

Zheng Y. J., Cui L. S., Schrooten J., (2005), Effects of additional reinforcing fibres on the interface quality of SMA wire/epoxy composites, PRICM 5: The Fifth Pacific Rim International Conference On Advanced Materials And Processing, Pts 1-5 Materials Science Forum, 475-479, 2047-2050.

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

29

Xjenza 11 (2006)

30

Research Article Model of limestone weathering and damage in masonry: Sedimentological and geotechnical controls in the Globigerina Limestone Formation (Miocene) of Malta Peter A. Gatt* Department of Civil Engineering, University of Malta, Msida, Malta.

Summary. Five types and subtypes of stone used in construction in the Maltese Islands and three problematic stone types, mostly extracted from facies within the Oligo-Miocene Valletta Basin, are identified. Their nature and geotechnical behaviour is discussed in the context of specific use in masonry. These stone types represent end members of the variations in depositional and diagenetic environments in carbonates which control their level of physical heterogeneity, and ultimately affect the nature of damage seen in Globigerina limestone masonry. A model is presented linking the level of heterogeneity to the mode of salt weathering seen especially in ancient constructions.

Keywords: Globigerina limestone, bioturbation, facies, salt weathering, geotechnical behaviour, Malta Received: 03 August 2006 Accepted: 24 August 2006 Published online: 30 August 2006

Introduction The mid-Tertiary succession of the Maltese Islands comprises five Formations (figure 1a), including the finegrained sediments of the Globigerina Limestone Formation that outcrop extensively. Constructions from the Neolithic to the advent of concrete have mostly used certain facies of this Formation, which is characterised by high purity (>90% CaCO3), fine grain size and small pore size. Some fine-grained facies are also found in both Coralline Limestone Formations.

Fine-grained limestone is more damaged by salt crystallization and less affected by washing of salts by rain (Kozlowski et al., 1989). This problem becomes acute in the maritime and seasonal climatic conditions of the Maltese Islands, and requires further study. Fitzner et al. (1997) have modelled salt decay in Globigerina limestone based on salt load, although Cassar (2004) alleges that marginal non-carbonate geochemical parameters can be used for â&#x20AC;&#x2DC;predictingâ&#x20AC;&#x2122; severity of weathering.

This paper adopts a holistic approach and considers the overall nature of stone as the principal variable. A number of types of stones, considered as end members of a continuum of varieties, are for the first time systematically described on the basis of depositional environment, diagenetic potential and geotechnical properties. Pore structure is also an important control on mode of weathering (Rossi-Manaresi & Tucci, 1990). On the basis of all these factors, a model is presented that links the level of heterogeneity in stone to the diverse modes of salt weathering observed in different stone types, even when found within the same masonry construction where environmental conditions are similar.

* Corresponding Author: e-mail: mepcons@hotmail.com , Tel: +356 79603783.

Other forms of damage, including tensile fractures in masonry are also discussed.

Stratigraphy of the Globigerina Limestone Formation and extraction Rizzo (1932) subdivided the Globigerina Limestone Formation into 3 Members: the Lower, Middle and Upper Members. These are separated by the ubiquitous C1 and C2 phosphorite conglomerate beds (figure 1a). The thickness of the Lower Member ranges from 0m in west Malta to >100m in central south Malta (Gatt, 2005a).

Dimension stone has been won since prehistory from distinct facies within the Lower Member, which is here sub-divided into 3 palaeogeographical areas (figure 1b): (1) The Valletta Basin facies found in central south Malta, where the Lower Member may exceed 100m in thickness and comprises a basal bluecoloured facies succeeded by pale-yellow facies showing cyclic sedimentation. The foram Globigerinoides is usually preserved; (2) Cyclic sediments succeeding the basal C0 phosphorite conglomerate bed (Gatt, 2005a) over a palaeohigh in west Malta and east Gozo, where the Lower Member becomes more condensed and may thin out considerably.

Planktonic foraminifera are poorly preserved (Gianelli & Salvatorini, 1972).

(3) West Gozo palaeoslope producing some thickening of the Lower Member.

Gatt PA: Model of limestone weathering and damage in masonry

31

Figure 1. (a) Lithostratigraphy; (b) Localities mentioned in text.

MIDDLE deeper MEMBER C 1

1F 1F 1F

1S 1S

1F 1S 1Fa 1F 1S 1Fa 2N

b

a

2P L.C.L Fm

2B

Figure 2. Highly generalized log of the Lower Member (Malta) showing stratigraphical position of stone types identified in text. Curve shows relative sea level. Symbols show level of bioturbation (a) increasing cementation; (b) increasing compaction.

(Number of sea level cycles is unknown)

Figure 3. Photo records of masonry (exterior) classified by facies identified in text. Scale 6cm.

Localities: 1F: The Palace, Valletta; 1Fa: staircase, Association SMOM, Vall.; 1Fb /1S: Phoenicia hotel; 1Fa2: modern, Attard; 2P: S.Caterina d’Italia church, Valletta; 2N: Pinto stores, Floriana.

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

The main quarrying regions are found in central south Malta, coinciding with the Valletta Basin facies which has been the main source of dimension stone since prehistory, and in the west Gozo palaeoslope (figure 1b).

Extraction was by splitting the rock with metal wedges, albeit the local freestone shows poor cleavage. In the 20th century, dimension stone started to be won by using rotary cutters. Most quarries lack microscale and mesoscale jointing, but may show few faults that traverse the entire quarry. Present problems include the low cost of local dimension stone, sizeable quarry waste and lack of scientific applications to quarrying leading to inadequate prospecting for good quality limestone (Gatt, 2002).

Facies of the Lower Member Two main Facies Associations are recognised within the Lower Member, both are quarried. The stratigraphy is shown in figure 2. Specific strata within these two facies associations are used as dimension stone and here subdivided into stone types and sub-types shown in figure 3:

FACIES ASSOCIATION 1

1. Globigerinid wackestone-packstone facies – 1F This facies is extensively used for construction and shows a wackestone to packstone texture with grains dominated by globigerinid tests (figure 4). This freestone, translated to franka in Maltese, generally has a pale yellow to white colour with few blemishes. Bioturbation includes medium sized (<50mm) burrows. SEM studies show a micritic ground mass (<2µm) with calcite spar sometimes developing in larger pores such as empty foraminiferal chambers (figures 4 & 5). The matrix includes coccoliths, although diagenetic processes may have rendered them partly indistinguishable. It has a relatively homogenized nature, although anisotropic, showing a uniaxial compressive strength normal to bedding greater by a magnitude of 1.1 to 1.2 relative to parallel to bedding.

32

compaction, greater preservation of empty globigerinid chambers or an overall decrease in phyllosilicates that can clog pores. Facies further up the Lower Member show higher levels of porosity (>30%) and more well-preserved coccoliths seen by SEM. A number of subtypes within 1F stone can be distinguished in ancient and modern masonry constructions. These different facies can be classified on the basis of structures within the stone that reflect depositional environment (depth of water and seabed oxygenation) and diagenesis: a. Chondrites facies – 1Fa: Fine (<3mm) structures of Chondrites are common in franka stone and can be distinguished on the surface by brown spots in a pale yellow matrix that is usually darker than the 1F stone.

Chondrites indicate dysaerobic depositional environments (Goldring, 1991) of deeper water where the less diverse bioturbation increases physical homogenisation in rock. A variant of this subtype (here called 1Fa2) recently won from certain quarries may also show medium-sized (<10mm) bioturbation with distinct brown ferruginous staining. b. Medium to large bioturbated facies – 1Fb: On weathering, some 1F stone shows evidence of larger burrows with a width >10mm, also associated with the echinoid Schizaster parkinsoni. However, the difference between burrow material and surrounding is not so great as to cause significant differential weathering, although this subtype may also be a hybrid of 1S stone. Types 1Fb and 1S stone have been used for exterior boundary walls and foundations. c. Facies with dewatering structures – 1Fc: In a few cases, bedding is disrupted by flame-structures and other dewatering structures. These sediments were disturbed by the rapid expulsion of water during compaction of the seabed sediments. A number of these structures are preserved in rock cuts in Valletta. Such internal structures can create strong anisotropies in rock which are detrimental when this stone is used in construction.

The pelagic depositional palaeoenvironment (50 to 150m deep; Pedley, 1987) was characterised by moderate sedimentation rates and aerobic seabed conditions that allowed bioturbation. Low to moderate hydrodynamic environment produced little surface cementation, and since sediments comprise mostly calcite, there was little dissolution of metastable aragonite that could supply CaCO3 for further cementation. Siliciclastic content is very low and minor accessory minerals include glauconite, quartz and phyllosilicates.

There is an overall increase in porosity in Globigerina limestone further up from the interface with the Lower Coralline Formation. This increase can be related to a number of factors including less

Figure 4. Lapped section of Globigerina limestone (1F stone) showing intragranular (i) and some intergranular porosity. Sample has been lapped to produce a cross-section effect (SEM micrograph

by P. Gatt)

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

33

>30MPa when mottling by burrows is clearly visible (Cachia, 1985).

FACIES ASSOCIATION 2

Figure 5. Trochospirally coiled shell of planktonic Globigerinacea: Intragranular porosity and precipitation of micrite (m) and spar (s) within chambers. Intergranular porosity (p) developed along Mode II failure (f). (SEM mircograph by P. Gatt)

Medium-depth burial, mostly related to the development of the Oligo-Miocene Valletta Basin (Gatt, 2005a) especially in south central Malta has resulted in a number of diagenetic changes that characterise facies. Further down the Lower Member, intragranular cementation increases in sediments deposited in these deeper marine palaeoenvironments, although the base (<18m) of the Globigerina Limestone Formation shows poor intergranular cementation.

1. Neomorphic facies – 2N 2. Intensely bioturbated facies – 1S Specific levels in the Lower Member consist of sediments showing anomalously dense and large bioturbation. Local quarrymen and masons use the term ‘soll’ to describe a problematic stone known to weather rapidly and unevenly that may have a slightly darker yellow and mottled hue, although it is also claimed to be visually undifferentiated from type 1F stone when freshly cut. Type 1S stone seen in Globigerina limestone quarries can be identified in outcrop at these stratigraphical levels:

Neomorphism represents a more advanced form of diagenesis in Globigerina Limestone, although it is limited in extent compared to the Coralline Limestone Formations. Aggrading neomorphism can be identified in type 1F and 1Fa stone as patches having a slightly translucent and darker surface that break with a distinct conchoidal fracture. At the microscale, neomorphism sometimes results in the development of calcite crystals with curved surfaces. Figure 6 shows the front between neomorphic spar and original micrite from a 17th century dimension stone in Floriana.

(1) 0.3 to <3m-thick facies recurring every 2 to <10m, depending on locality. In one quarry, type 1S stone shows an increase in silica and decrease in carbonates (Testa, 1989). The repetitive intervals of 1F and 1S stone types seen in many quarries are here interpreted as cyclic sediments formed during high frequency (Milankovitch scale) climatically-driven eustatic changes.

(2) 3m from the base of the Globigerina Formation, found only on palaeohighs outside the Valletta Basin facies. The facies can be almost entirely dominated by bow-form burrows (Goldring et al., 2002).

This facies is interpreted to have formed when seabed conditions were markedly aerobic during episodes of lower sea level and higher hydrodynamic levels. Under these conditions, bioturbation becomes intense and diverse, comprising large burrows of Thalassinoides, Ophiomorpha and Planolites, also associated with echinoid tests.

Animal burrows improve circulation of water in the sediments and increase the area of sediment/water interaction, resulting in the precipitation of some calcite cement. Increased cementation affects the geotechnical properties of the stone. Fitzner et al. (1997) confirms that 1S stone has a slightly lower porosity compared to 1F stone and Xuereb (1991) reports a higher compressive strength in ‘soll’ compared to franka stone, increasing to

Figure 6. Growth of neomorphic spar (5-15 microns) within micrite (<4 microns) in dimension stone from 17th century building in Floriana. Dotted white line shows boundary between micrite (m) and microspar (s). (SEM micrograph by P Gatt)

Neomorphism in pelagic limestone is associated with deeper burial environments. Danish chalk in the North Sea shows the onset of pressure dissolution and the formation of microspar at depths >600m (Tucker, 1990).

However, burial diagenesis in Globigerina limestone is difficult to explain since burial depths may have not been so great. The fine-grained texture and the lack of larger bioturbation structures indicates deeper water depositional environment, which may have gone through some burial diagenesis, possibly affected by elevated temperatures.

Crusts (1mm thick) closely resembling neomorphic

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

calcite are also seen on the surface of some old masonry, although their formation is unrelated to burial.

Testing of samples from the same dimension stone consisting of 1F with 2N variants show that the development of neomorphism produced a significant drop in water absorption by a magnitude of 10. This is related to a reduction in porosity by the development of neomorphism which reduces the entry of dispersed water and reduces the uptake by capillarity of solutes from the ground. Additionally, the growth of neomorphic spar also decreases hygroscopic absorption as seen in figure 7. % weight increase 1F 0.4 0.3 Lower Coralline Lmst

0.2 0.1

2N

0 1

2

3

4

hours

Figure 7. Hygroscopic absorption over 4 hours. Neomorphic Globigerina limestone shows only slight increase in percent weight and early levelling of trend, indicating very low hygroscopic absorption of atmospheric water compared to 1F stone.

2. Facies with cemented seams – 2P Millimetre-thick seams of well-cemented sub-parallel horizons can be seen in the lower parts of some quarries and in ancient masonry (figure 5). These are interpreted as pressure solution seams, although without the development of stylolites. The amount of dissolution is small, distorting burrow structures only slightly. These seams may have formed by burial compaction.

Alternatively, they may be extensive microcracks healed with calcite infilling, formed as a result of fracturing during tectonic activity affecting partly lithified rock. On weathering, the seams stand out, showing their stronger and more cemented nature.

3. Blue Globigerina limestone – 2B The blue-grey coloured Globigerina limestone is easily distinguishable from other facies and is locally called ġebla l-kaħla or ħadra. Type 2B stone outcrops in dimension stone quarries of south central Malta (sometimes succeeded or replaced by an orange mottled facies) and less extensively in west Gozo. In Malta, it limits further downward excavation. This facies delimits the depocentre produced by local tectonic deepening in south central Malta linked to the formation of the Valletta Basin (Gatt, 2005a). In this relatively deeper and dysaerobic environment, 1 to >15m of carbonate and siliciclastic sediments accumulated. Metre-wide lenticular bodies of blue limestone also outcrop within the pale yellow Globigerina limestone, immediately west of the Basin at Msida and Sliema.

34

Non-carbonates in the Lower Member Siliciclastic content is very low in bulk rock of the Lower Member, although it may indicate external controls that affected sedimentation throughout the Globigerina Limestone Formation. Cyclic sedimentation consisting of alternate 1S and 1F facies is accompanied by an increase in silica in the 1S beds and a peak in phosphate levels (~633ppm) just above the termination of every 1S bed (Testa, 1989). The main phosphate precipitation events in the Globigerina Limestone Formation are the C1 and C2 phosphorite conglomerate beds, the latter extending to SE Sicily. Carbone et al., (1987) associate these beds with anoxic conditions over hardgrounds. Phosphate was precipitated during shallowing events, succeeded by rare cross-bedding over the C1 bed in Sliema (Gatt, 2005a).

The increase of phosphate in recurring 1S beds is interpreted as showing greater nutrient levels during episodes of shallower marine conditions accompanied by intense bioturbation (figure 2). This culminated in the peaking of phosphate just above the 1S beds, when increased organic productivity brings the onset of low oxygen conditions resulting in a decline of bioturbation.

Clay in the Globigerina Limestone Formation is detrital in origin and correlates with quartz, comprising 0 to 25% of the rock. Its variable occurrence up the sequence represents episodes of eustatic, tectonic and climatic changes that also triggered continental erosion (John et al., 2003), culminating in the deposition of the Blue Clay Formation. Superimposed on this complex mineralogical signal is clinoptilolite, which is of a volcanic origin (John et al., 2003) and independent of cyclic sedimentation.

Meanwhile, tectonically-controlled deepening in the Valletta Basin produced a unique pattern of carbonatesiliciclastic sedimentation within a geographically restricted area (south central Malta). This includes type 2B stone which shows an increase in non-carbonate content recorded by Murray (1890) and confirmed by Vella et al. (1997) to exceed that of both ‘soll’ and franka (1F) stone. However, the complex and geographically diverse geochemical signals makes their use in identifying stone types highly debatable. Fitzner et al., (1997), Gatt (2005b) do not find a causal relationship between slight non-carbonate mineral content (e.g. non-swelling kaolinite) and severe weathering forms seen in masonry, thereby eliminating the relevance of geochemical proxies used by Cassar (2004) in ‘predicting’ weathering in Globigerina limestone.

Discussion An obstacle to the scientific study of Globigerina limestone deterioration is the persistent use in literature of elusive vernacular terms utilized by masons and quarrymen, even if these terms have not been scientifically defined e.g. ‘soll’ (here categorised as an intensely bioturbated facies at defined stratigraphical levels).

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

Depositional depth is a controlling factor in the lithification process. Together with the diagenetic potential (sensu Schlanger & Douglas, 1974) it accounts for the local variations in diagenetic grade of the Lower Member. These factors have direct consequences on geotechnical properties and weathering behaviour.

Diagenesis starts with compaction and dewatering, affecting subtype 1Fc. Later, the dissolution of coccoliths and foraminifera with depth increased cementation and decreased porosity in type 1F stone. Increased overburden pressure brings partial dissolution, creating the 2P type of stone. Further burial leads to the ultimate end state of lithification, reached when all grains have a minimum surface-to-volume ratio (Byrne, 1965). The recrystallised type 2N stone may have partly approached this condition in absence of clay. Where clay was present, lithification was only by compaction, resulting in a stone of low strength (type 2B stone). Although burial in the Lower Member is not considered to be deep, compaction with some cementation produced stone that could be used in masonry, unlike the uncemented and less compacted Middle Member.

The complex nature of stone and response to environmental conditions has to be assessed in terms of: (1) geotechnical properties, (2) the agents of weathering of stone and (3) the forms of weathering which result from the interaction between the nature of the stone and complex environmental conditions.

1. Geotechnical properties

Porosity and compressive strength

% porosity/UCS in MPa

Definition of distinct Globigerina limestone types should encompass the complexity of this material, which was deposited circa <25 MA and later subjected to different levels of diagenesis. These different stone types are seen to weather diversely in masonry, although, Renaissance, Baroque to early colonial age constructions show a very selective extraction of stone mainly from the Valletta Basin area (figure 1b), on the basis of their known performance in particular environmental conditions. This has resulted in relatively well-preserved ancient structures despite adverse environmental conditions and time e.g. ~2kyr Punic house at Żurrieq (Mahoney, 1988).

35

P a

b

c bc

a 0 to 10

10 to 20

20 to 30

height in metres above LCL

Figure 8. Rock properties varying by depth and palaeoenvironment: Porosity (P) increases further up the Lower Member; Uniaxial Compressive Strength (shaded bars) increases marginally up the section. Localities: (a) Zebbug Miocene palaeohigh; (b) Luqa palaeoslope; (c) Handaq Early Miocene palaeobasin.

The dry density of subtypes of 1F varies from 1.5 to 1.76 Mg/m3 (Bonello, 1988). This is more comparable with the density of Tertiary chalk (circa 1.6 Mg/m3) rather than limestone, indicating the medium burial depth (circa 200300m) experienced by the Lower Member of the Globigerina Limestone Formation. However, type 1S may show a slightly higher density of 1.784 Mg/m3, attributed

by Xuereb (1991) to the presence of clay that permits greater compaction.

The modulus of elasticity (E) and Poisson’s ratio (ν) are influenced by the degree of cementation in the rock, which also influences compressional velocity in limestone (Schlanger & Douglas, 1974). Relatively higher qu, lower porosity and palaeoenvironment indicators point to greater cementation in 1S compared to 1F stone. Figure 9 shows data for E and qu for limestone and chalk, including that of Xuereb (1991) and Bonello (1998) who report a high E for franka stone (1F) compared to similar stone. This would also indicate an exceptionally high modulus ratio (E/qu) of ~1000, or double that proposed by Deere & Miller (1966) for strong intact rock, comparable to that of crystalline rock e.g. massive finegrained Taconic marble (Vermont, USA). The E for Globigerina limestone used in construction is here approximated to be ~4 to 6 GPa, on the reasonable assumption that this stone shows a low to average modulus ratio (figure 9).

25 Young's modulus (GPa)

FACIES ASSOCIATION 1 The Uniaxial Compressive Strength (qu) of the Lower Member of the Globigerina Limestone Formation ranges between 8 and >20 MPa, vaguely increasing with porosity further up the sequence (figure 8) in cores by Wardell Armstrong, (1996), although quarry samples show a negative correlation between qu and porosity (Bonello, 1988). Type 1S stone has slightly higher compressive strength (~21 MPa) than average 1F stone (14 to 20 MPa).

Bowden et al. (1998) also show that qu increases with a decrease in visually assessed clay content in ‘Globigerina marl’.

45 40 35 30 25 20 15 10 5 0

500:1

20 15 B

10

200:1 5 0 0

10

20

30

40

50

60

70

Uniaxial Compressive Strength (MPa)

Figure 9. E and qu for Tertiary limestone and chalk: (♦) Lower Coralline Limestone; (▲) Kent chalk (Bell, 1993); (○) Eocene chalk, Israel (Talesnick & Brafman, 1998); star symbols shows data by Xuereb (1991) and (B) Bonello (1988). Shaded area represents my approximated E for 1F stone. Dotted lines represent high and low modulus ratio limits by Deere & Miller (1966).

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

The tensile strength of stone has significant consequences on the integrity of masonry and construction. Because tensile strength of Globigerina limestone is always several orders of magnitude less than its compressive strength, tensile cracks can occur under many circumstances in local masonry. Mode I opening occurs initially at microscopic scale seen in figure 4, associated with Mode II failure and develops to the mesoscale. Some of the most serious damage to ancient structures is the result of tensile stress failure occurring in two types of masonry: (a) Structures built without mortar: Neolithic to Punic age constructions, e.g. Ħaġar Qim [497 653] and Mnajdra [492 651] temples, where point load stress by overburden or by movement of masonry (due to differential settlement or erosion of surrounding stone) has resulted in tensile fracturing in megaliths of type 1F stone. (b) Masonry with mortar: Several Medieval to Baroque buildings show erosion or deformation of mortar leading to point load stress between courses caused by masonry overburden. Depending on the stone’s E, this results in the formation of vertical tensile cracks that split the dimension stone. Tensile cracks may also form at the side of the dimension stone following the removal of lateral confining pressure by loss or softening of mortar (figure 10).

Figure 10. Auberge de Castille, Valletta, south corner: tensile fractures (circled) related to loss or differential deformation of mortar. Stone types labelled, also showing different forms of salt weathering.

FACIES ASSOCIATION 2 The geotechnical properties of types 2N and 2P stone are relatively untested. However, recrystallization and cementation reduces porosity and can increase compressive strength and E. Brittle stone have a higher E

36

and under stress fracture at higher frequency (Gross et al., 1995) e.g. rusting metal in masonry made of type 2N stone may develop tensile fractures parallel to axis of loading that lead to conchoidal fracturing.

Bioturbation in type 2B has thoroughly mixed the clay and carbonate sediments. The presence of clay reduces the potential for cementation (Matter, 1974), resulting in lithification by compaction. Since overburden pressure may not have exceeded circa 300m, this facies usually shows a relatively low uniaxial compressive strength of <10 MPa which makes it unreliable for use in masonry.

The presence of clay can affect weathering of limestone due to its expansion on wetting and its surface charge (Gauri & Bandyopadhyay, 1999).

The form of weathering seen in similar grey-coloured Globigerina limestone in the Middle Member, namely by splitting during wetting and drying cycles may be caused by the presence of clay.

2. Agents of weathering of stone The long-term macroscale erosion of limestone is due to carbonation, although the shorter-term damage on the micro and meso-scale is controlled by salt crystallization.

Cassar (2004) classifies stone by the degree of weathering and identifies ‘badly’ weathering stone as ‘soll’.

However, the degree of salt damage is also a function of environment, which is highly variable and should not be used as a basis for classifying stone types. Instead, it is the mode of weathering which reflects the intrinsic nature of the stone, independently of environment, which should partly categorize a type of stone.

The main sources of solutes in stone are capillarity rise, dispersed water and hygroscopic absorption, depending on specific environmental conditions namely, distance from the sea and water table, wind effect, diurnal temperature and humidity variations. The relationship between solute migration to stone surface and evaporation controls level of salt disruption (Rossi-Manaresi & Tucci, 1991). Fitzner et al. (1992) also record different damage categories for masonry and architectural decoration. The latter may show special forms of weathering due to the geometry of the stone. For example, I observed that subtype 1Fa and type 2P stone may show granular disintegration on the exterior of masonry, but result in sizeable spalling of architectural decoration on the interior (e.g. St John’s co-cathedral) and exterior (e.g. Mdina gate) respectively. In this paper, damage categories refer to the most common weathering form seen on the exterior masonry.

Where environments are very hostile to local stone, as in the case of buildings close to sea level and exposed to sea spray, Globigerina Limestone was considered inadequate and was supplanted by Upper Coralline Limestone as the main dimension stone e.g. Scamp’s Palace [567 720].

Type 2N stone has also been used in local constructions of the 17th and 18th century in environments that are conducive to rapid weathering of stone e.g. Forni Stores

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

[566 721]. In localities where capillarity rise of solutes is significant, Coralline Limestone has been used as masonry for the lower courses (e.g, in Mdina). However, in Valletta (where Coralline limestone is not readily available) this has been replaced (possibly deliberately) by type 2P stone, where dissolution seams act as inhibitors to capillarity rise.

3. Modelling mode of weathering in finelimestone

grained

A model for fine-grained limestone e.g. Globigerina limestone; Mtarfa Member, Upper Coralline Limestone (Pedley, 1987), is proposed based on the nature of the stone and the related mode of weathering (fig. 11). This is independent of salt load, which is a function of environment. This model (table 1) presents 3 conclusions based on extensive observation of external masonry in Malta: (1) Relative intensity of salt weathering is controlled by level of; (a) uniformity of distribution of cementation; (b) physical heterogeneity, namely the distribution of pores (random or clustered) and the size of pores. Intermediate heterogeneity and cementation in stone results in least severe weathering. Weathering tests on Globigerina limestone also confirm that heterogeneity has a direct effect on loss of weight in stone (Cachia, 1999); (2) Type of stone controls mode of salt weathering e.g. granular disintegration or alveolar weathering, as seen within the same ancient masonry (figure 11); (3) The slight non-carbonate content especially in the problematic 1S stone has no consequence on salt weathering.

37

Moisture with solutes preferentially fills up smaller pores, which later supply larger pores with solute. On drying, crystallisation first occurs in the larger pores. In Globigerina limestone these include empty foraminiferal chambers (~4 to 50 µm) which are more common in type 1F stone than in type 1S stone (in the latter these are partly or completely cemented).

Fitzner et al. (1997) confirms that pore sizes >3 µm constitute 12.2% and 5.7% in stone that weathers moderately and severely, respectively. Only after the larger pores are completely filled can crystallisation begin in the smaller pores.

Heterogeneity of the stone, including the pore sizes has the following controls on salt decay in Globigerina limestone (table 1): I. Highly heterogeneous stone: The greater size, density and diversity of bioturbation contribute to heterogenization in 1S stone and the Mtarfa Member (shallow marine back reef facies). Larger burrows tend to be filled with coarser grain size due to the binding of grains and production of faecal pellets by burrowing fauna or because they are infilled with sandsized grains. In type 1S stone, relatively less solute is supplied to the fewer larger pores mostly in burrows, so that more remains in the smaller pores, where salt crystallization can start at an earlier stage. Salt crystallisation is more destructive in the finer-grained matrix surrounding the burrowed areas. This explains the development of exichnia (sensu Martinsson, 1970) commonly seen in weathered type 1S outcrops (figure 5), also as a result of differential cementation. Exichnia protrusions break off, causing a significant loss in original volume and overall decline in strength.

II. Moderately heterogeneous stone

Figure 11. >200yr bastion, Mdina. Same salt load and environment; (a) Homogeneous stone (1Fa), poorly cemented, showing scaling. (b) Heterogeneous, cemented Mtarfa Member (UCL) stone develops alveolar (exichnia) mode of weathering.

Rossi-Manaresi & Tucci (1991) show that stone decay by salt crystallization occurs only in the case of particular pore structures and conclude that high crystallization pressure is associated with stone having a wide range of pore sizes, from <0.01 to 1 µm as well as larger pore sizes. Fitzner & Snethlage (1982) also associate this type of porosity with the significant development of salt decay in stone. Crystal growth in pores exerts large pressure leading to tension cracks in the stone. Cyclic wetting/drying leads to rapid disintegration of the stone.

Type 1F stone is free from localised anomalies, hence called freestone or franka, which together with uniformly distributed cementation gives the best weathering quality to stone. Early scaling passes to granular disintegration that may later develop into poorly defined exichnia (figure 5, 1F), independently of salt load. The more common larger pores tend to block capillarity rise of water coming from smaller pores. Salt crystallization in larger pores is relatively less damaging compared to that in smaller pores. However, when larger pores close to the surface are filled with growing salt crystals, these may contribute to exfoliation and crumbling of surface crust.

III. Homogenous stone Subtype 1Fa and type 2N stone used in masonry of the 17th century are relatively homogenous and show low diversity in burrow types. Subtype 1Fa lacks alveolar weathering by salt crystallisation. This is due to the homogenous texture of the stone that weathers uniformly by granular disintegration and scaling, although more rapidly compared to 1F stone due to its poor cementation.

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

38

Heterogeneous Stone type

Pore size

small <3 µm

1S

Intermediate

1Fb

Mostly in matrix surrounding burrows

1F

Homogeneous

2N 2P

dispersed

large

5.7% mostly in large burrows and connected

12.2% and dispersed (Globigerinid chambers)

Mode of Weather-

Alveolar, distinct exichnia formed

Some alveolar, exfoliation & granular disintegration

Cementation

moderate

1Fa dispersed

dispersed Mostly granular disintegration and scaling low

Table 1. Model of mode of weathering controlled by stone heterogeneity (pore sizes after Fitzner et al., 1997)

Conclusions Mode of salt weathering is related to the level of physical heterogeneity of fine-grained pure limestone. This factor is fundamental to conservation and intervention and can be used to address the following problems which are endemic in the Maltese Islands: - In cases where badly weathered stone needs to be replaced, a similar stone should be selected (figure 11 shows the effect of the contrary).

- Consolidants and non-carbonate coatings on masonry building should not be applied indiscriminately to different stone types, but are more appropriate for types with low cementation.

- Constructions in hostile environmental conditions require the selection of specific stone types that respond adequately to adverse conditions.

- Identification of stone types is fundamental in prospecting for quality limestone by the quarrying industry and can serve as a basis for a stone classification scheme and pricing of different stone types.

Acknowledgements: Tests and SEM studies were carried out in the laboratories of the former Malta Centre for Restoration.

References: Bell, F.G. 1993 Engineering Geology. Blackwell Scientific Publications, pp357.

Bonello , S. 1988 Engineering properties of rocks and soils of the Maltese Islands. Unpublished B.E.&A dissertation, University of Malta.

Bowden, A.J., Lamont-Black, J. & Ullyott, S. 1998 Point load testing of weak rocks with particular reference to chalk. Quart. Journ of Engin. Geol., 31(2), 95-103.

Byrne, J.G. 1965 Recovery, Recrystallisation and grain growth. Macmillan, New York, pp175.

Cachia, C. 1999 The effects of salts as a primary weathering parameter of Globigerina limestone.

Unpublished BE&A dissertation, Univ. of Malta.

Cachia, J. 1985 The mechanical and physical properties of the Globigerina limestone as used in local masonry construction. BE&A dissertation, Univ. of Malta.

Carbone, S., Grasso, M., Lentini, F., Pedley, H.M. 1987 The distribution and palaeo-environment of early Miocene phosphorites of SE Sicily and their relationships with Maltese phosphorites. Palaeogeog., Palaeoclimat., Paleoecol., 58, 35-53.

Cassar, J. 2004 Comparing visual and geochemical classification of limestone types: The Maltese Globigerina limestone. In: Proceed. of 10th Inter.

Congress on deterioration & Conserv. of stone, Stockholm, 569-573.

Deere, D.V. & Miller, R.P. 1966 Engineering classification and index properties for intact rock. Air Force Weapons Lab., New Mexico.

Fitzner, B. & Snethlage, R. 1982 Über Zusammenhange zwisschen Salzkristallisationdruck und Porenradienverteilung. GP News Letter, 3, 13-24.

Fitzner, B., Heinrichs, K. & Kownatzki, R. 1992 Classification and mapping of weathering forms. In Delgado, Fernando & Telmo (Eds.) 7th International congress on deterioration and conservation of stone.

Laboratório Nacional de Enginharia Civil, Lisbon, 957968.

Fitzner, B., Heinrichs, K. & Volker, M. 1997 Model for salt weathering in Maltese Globigerina limestone. In:

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com

Gatt PA: Model of limestone weathering and damage in masonry

Zezza (ed.) Origin, mechanisms and effects of salts on degradation of monuments in marine and continental environment. EU Workshop, Bari. C.U.M., 333-344.

Gatt, P.A. 2002 Quarrying in Malta. Quarry Management (UK), 29(2), 10-13.

Gatt, P.A. 2005a Syntectonic deposition of an OligoMiocene phosphorite conglomerate bed in Malta.

Central Medit. Naturalist, 4(2), 109-119.

Gatt, P.A. 2005b Assessing limestone quality for the construction industry in Malta: A geological perspective. The Architect, 35, 20-21.

Gauri, K., Bandyopadhyay, J.K. 1999 Carbonate stone: chemical behaviour, durability and conservation.

Wiley-Interscience publication. pp283.

Gianelli, L. & Salvatorini, G. 1972 I foraminiferi planctonici dei sedimenti terziari dell’Arcipelago maltese. Atti. Soc. tosc. Sci. Nat. Mem. Serie A, 79, 49-74.

Goldring, R. 1991 Fossils in the field. Longman scientific. pp218.

Goldring, R., Gruszczynski, M. & Gatt, P.A. 2002 A Bow-form burrow and its sedimentological and palaeoecological significance. Palaios, 17, 622-630.

39

Schlanger, S.O. & Douglas, R.G. 1974 The pelagic oozechalk-limestone transition and its implications for marine stratigraphy. Spec. Publs. Int.Ass. Sediment., 1, 117-148.

Talesnick, M.L. & Brafman, M. 1998 Small strain deformation characteristics of two chalks subjected to varying stress conditions. Quart. Journ. of Engin.

Geol., 31(3)161-174.

Testa, S.J. 1989 A down-column geochemical study of Lower Globigerina Limestone with special reference to the ‘soll’ layers. Unpublished B.Ed. dissertation, University of Malta.

Tucker, M.E. & Wright, V.P. 1990 Carbonate Sedimentology. Blackwell. pp 482.

Vella, A.J., Testa, S., Zammit, C. 1997 Geochemistry of soll facies of the Lower Globigerina Limestone Formation.

Xjenza, 2(1), 27-33.

Wardell Armstrong 1996 Mineral Resource Assessment report. Malta.

Xuereb, D. 1991 Elastic constants of Globigerina Limestone. Unpublished B.E & A dissertation, University of Malta.

Gross, M.R., Fischer, M.P., Engelder, T., Greenfield, R.J. 1995 Factors controlling joint spacing in interbedded sedimentary rock. In meen, M. (Ed.) Fractography.

Geol. Soc. Spec. Publ. 92, 215-233.

John, C., Mutti, M. & Adatte, T. 2003 Mixed carbonatesiliclastic record on the North African margin (Malta)coupling of weathering processes and Miocene climate.

GAS Bull., 115(2), 217-229.

Kozlowski, R., Magiera, J., Weber, J. & Haber, J. 1989 Decay and conservation of Pinczow porous limestone.

Studies in conservation, 35, 205-221.

Mahoney, L. 1988 A History of Maltese Architecture Veritas Press, Malta. pp360.

Martinsson, A. 1970 Toponomy of trace fossils. Geol.

Jour., Spec. Iss., 3, 323-330.

Matter, A. 1974 Burial diagenesis of pelitic and carbonate deepsea sediments from the Arabian Sea. Init. Repts.

DSDP, 23, 421-470.

Murray, 1890 The Maltese Islands with special reference to their geological structure. Scottish Geographical Magazine. Sept., 1-40.

Pedley, H.M. 1987 Controls on Cenozoic carbonate deposition in the Maltese Islands: review and reinterpretation. Mem. Soc. Geol. It., 38, 81-94.

Rizzo, C. 1932 Report on the Geology of the Maltese Islands. Govt. Printing Office, Valletta.

Rossi-Manaresi, R. & Tucci, A. 1990 Pore structure and the disruptive or cementing effect of salt crystallisation in various types of stone. Studies in conservation, 36, 53-58.

© 2006 Xjenza – Journal of the Malta Chamber of Scientists - www.xjenza.com