Lead Salt Thin Film Semiconductors for Microelectronic Applications

Lead Salt Thin Film Semiconductors for Microelectronic Applications 2010 Authors

Shaibal Mukherjee Electrical Engineering and Computer Science, Northwestern University Evanston, IL - 60208, USA

Donghui Li Electrical and Computer Engineering, University of Oklahoma Norman, OK - 73019, USA

Anurag Gautam Department of Chemistry, University of Victoria, Victoria British Columbia V8W 3V6, Canada

Jyoti P. Kar Department of Materials Science and Engineering, Yonsei University Seoul 120-749, South Korea

Zhisheng Shi Electrical and Computer Engineering, University of Oklahoma Norman, OK - 73019, USA

Transworld Research Network, T.C. 37/661 (2), Fort P.O., Trivandrum-695 023 Kerala, India

Published by Transworld Research Network 2010; Rights Reserved Transworld Research Network T.C. 37/661(2), Fort P.O., Trivandrum-695 023, Kerala, India Authors Shaibal Mukherjee Donghui Li Anurag Gautam Jyoti P. Kar Zhisheng Shi Managing Editor S.G. Pandalai Publication Manager A. Gayathri Transworld Research Network assumes no responsibility for the opinions and statements advanced by the Authors ISBN: 978-81-7895-501-8

Contents Chapter 1 Background 1.1. Introduction 1.1.1. Lead salt energy band structure 1.1.2. Band gap engineering 1.1.3. Dielectric property and refractive index 1.2. IV-VI semiconductor growth 1.3. Applications of lead salt materials 1.3.1. Detectors 1.3.2. Lasers 1.3.3. Quantum dots

1 1 3 7 9 9 12 12 14 15

Chapter 2 Growth pits of lead salt epitaxial layers grown on silicon substrates 2.1. Defects in semiconductors 2.1.1. Threading dislocation in IV-VI materials 2.1.2. Growth pits in II-VI semiconductors 2.2. Investigation of growth pits in IV-VI semiconductor 2.2.1. Samples preparation 2.2.2. Composition of micro-cuboids 2.3. Multiple growth pits 2.4. Origin of growth pits 2.5. Conclusions

16 16 17 20 23 23 25 26 27 28

Chapter 3 A novel in-situ surface treatment during MBE growth of PBSE on CAF2/SI(111) heterostructure 3.1. Introduction of surface treatment for II-VI epitaxy 3.1.1. N-plasma Si surface treatment 3.1.2. Improvement of growth quality by surface treatment 3.1.3. Growth model 3.2. In-situ treatment for IV-VI epitaxy 3.2.1. Motivation 3.2.2. Traditional growth of lead salt on Si 3.2.3. Design of experimental procedure 3.2.4. In-situ surface treatment procedure 3.2.5. Results and Discussion 3.2.6. Conclusions

29 29 29 31 32 34 34 34 36 36 39 41

Chapter 4 Recent developments of IV-VI lead salt semiconductor quantum well structures 4.1. Background 4.2. Emission power from QW structures 4.3. MBE growth on [110] oriented BaF2 substrate 4.4. [110] Oriented electrically pumped IV-VI edge-emitting laser 4.4.1. Consistency in infrared laser emission 4.5. Mid-IR vertical cavity surface emitting laser (VCSEL) 4.5.1. Above-room-temperature pulsed operation 4.5.2. Optically pumped CW operation 4.6. Room temperature CW photoluminescence 4.7. Fabrication of free-standing multiple quantum well microstructures 4.7.1. Background 4.7.2. Pulsed mode PL emission from MQW micropillars 4.7.3. Fabrication of IV-VI microtubes/rods 4.7.4. Optical characterization of microtubes/rods 4.8. Conclusions

42 42 43 45 46 51 52 54 55 56 57 57 57 59 59 63

Chapter 5 Photonic bandgap defect structure: Cavity without cleaving 5.1. Background 5.2. Introduction 5.3. Finite difference time domain (FDTD) method 5.4. Two dimensional FDTD equations 5.4.1. TE Waves 5.4.2. TM waves 5.5. Plane wave expansion (PWE) method 5.6. Finite difference method (FDM) 5.7. Lead chalcogenide defect cavity PBG structure 5.7.1. Mid infrared photonic bandgap formation 5.7.2. Modal analysis by FDM scheme 5.7.3. Modal analysis by FDTD scheme 5.8. Experimental steps for air hole formation 5.9. Conclusions

63 63 65 65 66 66 69 70 70 73 74 77 77 79 81

Chapter 6 Summary References

81 83

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Lead Salt Thin Film Semiconductors for Microelectronic Applications, 2010: 1-88 ISBN: 978-81-7895-501-8 Authors: Shaibal Mukherjee, Donghui Li, Anurag Gautam Jyoti P. Kar and Zhisheng Shi

Lead salt thin film semiconductors for microelectronic applications 1

Shaibal Mukherjee1, Donghui Li2 Anurag Gautam3, Jyoti P. Kar4 and Zhisheng Shi2

Electrical Engineering and Computer Science, Northwestern University, Evanston, IL - 60208, USA 2 Electrical and Computer Engineering, University of Oklahoma, Norman, OK - 73019, USA 3 Department of Chemistry, University of Victoria, Victoria, British Columbia V8W 3V6, Canada 4 Department of Materials Science and Engineering, Yonsei University, Seoul 120-749, South Korea

CHAPTER 1: BACKGROUND 1.1. Introduction IV–VI lead salt semiconductor devices can be traced back to the 1874 when the first rectification was observed with natural galena (PbS) by F. Brann et al [1]. From that point forward, narrow band gap IV–VI lead salt semiconductors like PbTe, PbSe, and PbS and their ternary and quaternary alloys such as PbSnSe, PbSnTe, PbSrSe, and PbSnSeTe have been widely used in various solid state midinfrared devices, such as light emitting devices (LEDs), laser diodes (LDs), detectors, and thermoelectric generators [2, 3].

Due to their two orders of magnitude lower non-radiative Auger recombination rates as compared to those of III-V and II-VI narrow band gap semiconductors, IV-VI semiconductors optoelectronic devices such as detectors and lasers are considered to be the best candidates for roomtemperature operation. Moreover, the high permittivity of lead salt materials Correspondence/Reprint request: Dr. Shaibal Mukherjee, School of Electrical Engineering and Computer Science, Northwestern University, Evanston, IL – 60208, USA. E-mail: mukherjee.shaibal@yahoo.com

2

Shaibal Mukherjee et al.

will effectively shield electric charges from defects, which in turn yields superior and fault-tolerant devices [4]. IV-VI semiconductors present one of the rare naturally occurring semiconductors and the name “lead salts” for the IV-VI semiconductors is in common use in infrared optoelectronics [5-7].

The binaries PbS, PbTe, and PbSe (ordered by their band gap equivalent wavelengths), and typical mixed crystals of this material family like the ternaries PbSnTe, PbSnSe, PbCdSe, PbSSe cover a band gap range from 50meV to 500meV, with a corresponding wavelength range of roughly 3µm to 30µm. Fig. 1.1 gives an overview by showing the relation of the band gap of some mixed crystals on the lattice periodicity length. These crystals have rock-salt structure.

Ternaries with Eu, Mn, and Gd are of further interest, since they result in semimagnetic materials [8]. Quaternary material is grown for strain and stress engineering. While these materials are the most widely investigated ones due to their possible use in infrared devices, others like SnSe have also been used for early quantum structures [9].

In contrast to the tetrahedrally-coordinated group IV semiconductors like diamond, silicon, germanium, and the III–V (GaAs, AlAs, InAs, etc.) or II– VI semiconductors (ZnTe, CdTe, etc.), in the IV–VI compounds, there are 10

Figure 1.1. The band gap of typical IV-VI materials as a function of the lattice periodicity length (upper abscissa) or of the relative lattice misfit to Si (lower abscissa). For comparison also data for Si, Ge and some III-V and II-VI compounds are shown. (Source: Zogg et al. (1989)).

Lead salt thin film semiconductors for microelectronic applications

3

instead of 8 bonding electrons per atom pair. The six electrons per atom pair promote hybridization with the bonds orbitals nearly at 90° angles, resulting in a 6-fold coordination of the atoms in the crystal lattice. As a consequence, the IV–VI compounds predominantly crystallize in the rock-salt structure. As a result, many physical and electronic properties of the IV–VI semiconductors differ strongly from that of the tetrahedrally bonded semiconductors. In addition, the lead salt compounds are mechanically much softer than their tetrahedrally bonded counterparts. This easy plastic deformability has significant implications for device fabrication processes.

1.1.1. Lead salt energy band structure The direct band gap minimum of IV-VI semiconductors is fourfold and located at the L-points (on the (111) axes and on the Brillouin zone surface), in contrast to the Γ-point (in the center of the Brillouin zone) minima of the technically more common semiconductors. For comparison, parts of the band structures of PbSe and GaAs are shown in Fig. 1.2. The conduction and valence bands of PbSe are quite similar, so that electrons and holes have comparable masses. These masses, however, differ along the (111) axes and perpendicular to them, giving ellipsoidal energy surfaces for technically relevant carrier concentrations. The order of magnitude of the corresponding mass anisotropies ranges from 10 for PbTe to 1 for PbS. As a typical example, the constant energy surfaces of PbSe are shown in Fig. 1.3 and band structure in Fig. 1.4.

Figure 1.2. Comparison of typical IV-VI and III-V band structures. (Source: Tacke et al).

4

Shaibal Mukherjee et al.

PbSe

Figure 1.3. Constant energy surfaces of PbSe. Aside of the band edge minima at the L-points (open half-ellipsoids), the ÎŁ-pockets are also shown, since they must be considered in connection with hot carriers. (Source: Tacke et al).

Figure 1.4. Band structure of PbSe calculated by non-local empirical pseudopotential methods.

Lead salt thin film semiconductors for microelectronic applications

5

In contrast to the majority of technically important semiconductors, the lead salt band gap widens with temperature. This is explained qualitatively by a structural transition close to 0K, which “softens” the relevant energies. This fact must be considered when interpreting experimental optoelectronic data.

It leads for instance to absorption of radiation originating from electrically or optically pumped regions embedded in non-pumped material. This is due to the relative blue shift of the active region band gap by localized heating via pumping losses, which pushes the emission energy into band-to-band absorption of the cladding.

The wavelength dependence of the refractive index is of typical semiconductor behavior. The refractive index increases with decreasing wavelength within the band gap spectral range with a maximum close to the optical band edge. The temperature behavior corresponds to that of typical semiconductors, but the temperature-induced change is negative relative to the usual, in accordance with the different energy shift of the gap with temperature. The refractive index n is quite high and around n = 5, the static dielectric constant is of the order of 1000 giving rise to large depletion lengths. For PbTe and carrier concentrations of the order of 1017cm−3, the depletion length exceeds 300nm. These large depletion lengths make the carrier concentrations quite constant across quantum structures. This is an important feature, since it allows comparatively straightforward interpretation of experimental data that depend on carrier concentrations.

Due to the large dielectric response, excitons should have very large Bohr radii and very low binding energy; in accordance with this expectation, distinct exciton transitions are not observed directly in this narrow band gap material. There are, however, interpretations of experimental data that are consistent with the existence of such states with binding energies of one to two inverse centimeters, i.e. around 0.1 meV [10].

The effective masses of electrons and holes are similar. Due to the four valley structure and electron mass anisotropy, electron effective masses are effective averages of the masses for motion parallel to the minimum rotation ellipsoid axes (ml) and perpendicular (mt). While the masses of electrons and holes of the IV-VI compounds are smaller than those of heavy holes of III-V semiconductors, the density of states masses md* are comparable, since they are mixed of the transverse and longitudinal masses mt and ml by md* = 42/3(ml mt2)1/3 [11].

Since the IV-VI semiconductors are long wavelength infrared materials, free electron absorption in general is strong at wavelengths of interest, with absorption coefficients beyond 102cm−1 for a material with a carrier concentration of 1×1018cm−3. Free carrier absorption depends on wavelength, the absorption coefficient α being proportional to the carrier concentration n,

6

Shaibal Mukherjee et al.

the wavelength λ squared, and inversely proportional to the free carrier effective mobility mass meff and mobility µ or, using the relation between mobility, mass, and scattering time τ with quantum energy ћω of the photons.

From liquid nitrogen to room temperature, the mobility of good grade material is typically not limited by impurities, but rather by optical and acoustical phonon scattering. The reason is qualitatively found in the very high dielectric constant which shields impurities. In this range of temperature T, the mobility empirically follows a power law like T−5/2 for pure crystals [12]. Room temperature mobilities are of the order of 1000cm2V−1s−1. Due to the similar band extrema of electrons and holes, the mobilities of p-type and n-type materials are comparable, the n-type mobility being usually a little larger. The mobility of ternary material is lowered by alloy scattering as usual. However, this effect is sometimes outbalanced by a reduction of the effective mass resulting from the shrinking of the gap. This effect is found notably in the narrow band gap materials Pb1−xSnxSe and Pb1−xSnxTe, that even go through zero band gap due to band inversion. Due to the small mass, mobility increases [13].

The minority carrier lifetime in narrow gap semiconductors is typically limited by the Auger effect. There is a strong difference between IV-VI materials on one side, and on the other side, II-VI and III-V material, in that the Auger recombination rate is two orders of magnitude smaller in the IV-VI material with comparable band gap energy, which is the reason for the IV-VI dominance in mid-infrared diode lasers. This effect was found both theoretically [14-16] and experimentally [17, 18]. The Auger effect is due to one carrier being excited within a band while two others recombine, one being an electron from the same band, and the other a hole from the valence band. This effect depends on the detailed band structure via energy and momentum conservation. The multi-valley lead salt band structure leads to a special mechanism, intervalley-scattering, where a large momentum is taken up by the intraband transition of the Auger carrier [14, 16]. A pump-probe optical experiment by Klann et al [17] on PbSe with carrier concentration 3x1018cm−3 evaluated an Auger coefficient close to 10−28cm6s−1, and thus an effective lifetime of 10−9s, as predicted theoretically [15]. In comparison, Auger recombination coefficients were found to be (at 300 K) 60, 210 and 90 in units of 10−28 cm6s−1, for InAs, InAs0.91Sb0.09 and an InAlAsSb multiple quantum well, respectively [19].

Current/voltage characteristics of p-n junctions are usually found to be dominated by carrier generation/recombination characteristics that probably are due to imperfections and/or material impurities. The corresponding nonradiative recombination times of PbSe were 0.7 ns for bulk material and 0.45

Lead salt thin film semiconductors for microelectronic applications

7

ns for MBE grown material. The defect dominated lifetime is an indicator of a substantial material growth optimization potential [20].

1.1.2. Band gap engineering As stated in the previous section, one of the important parameters of the electronic band structure of the lead salt compounds such as PbSe, PbTe, PbS, is their small and direct energy gap of less than 410 meV at the L-valley point of the Brillouin zone (as can be seen in Fig. 1.1 and Fig. 1.4). As a result, photons can be directly absorbed or emitted at the band edges. The conduction and valence bands have nearly mirror symmetry with almost equal effective masses for the electrons and holes. Because of the narrow energy gaps, the energy bands are strongly non-parabolic and the effective masses of electrons as well as holes are rather small (0.02m0–0.08m0 for the transverse masses). In addition, the bands are anisotropic, i.e., the Fermi surfaces are elongated ellipsoids of revolution around the <111> axes, characterized by a longitudinal and transverse effective mass ml and mt parallel and perpendicular to the 8-fold <111> directions. While for most semiconductors the energy band gap increases with decreasing temperature, in the lead salt compounds it decreases upon cooling to from room temperature to 4 K by about 150 meV. This represents a large relative change with respect to the absolute value of the band gap, and thus, the emission of lead salt based diode lasers can be tuned of over a broad wavelength range just by changing of operation temperature. This effect is used for spectroscopic applications of these lasers.

For adjustment of the fundamental absorption edge of infrared detectors as well as the adjustment of the emission wavelength of lead salt diode lasers, alloying of the lead salt compounds with other chalcogenide compounds is used [21-23]. This is shown in Fig. 1.1, where the band gap energies of several IV–VI compounds and their alloys are plotted versus lattice constant and compared with various III–V, II–VI and group IV semiconductors. Band gap engineering is also a crucial prerequisite for fabrication of quantum confined low dimensional heterostructures. This is based on the alloying of different semiconductor materials by which the energy band gap of the material can be varied over a wide range. For the IV–VI semiconductors, a number of different ternary and quaternary alloys have been used for this purpose. Of the classical alloys within the IV–VI compounds, the ternary PbSnTe and PbSnSe single phase pseudo binary alloys have been widely used for far infrared applications up to wavelengths of 30 µm (see Fig. 1.1).

Since the conduction and valence bands in the tin chalcogenides are exchanged compared to the lead chalcogenides, the band gap of the ternary

8

Shaibal Mukherjee et al.

alloy decreases linearly with increasing Sn content reaching zero for a certain alloy composition. Other pseudo binary IV–VI alloys with exchanged group IV (e.g. Pb/Ge) or group VI element (e.g. Se/Te) are less useful for band gap engineering since the achievable variations of the band gap energies are comparatively small, while the lattice constants change rather strongly with varying alloy composition. For these alloys, the largest tuning range can be achieved in the PbSe1–y Sy system with a band gap variation from 172–310 meV at 77 K. Since other IV–VI compounds such as SnS, GeTe or GeSe do not crystallize in the rock-salt crystal structure, only a limited miscibility regime exists for their alloys with the lead salt compounds. The same applies also for alloys with MnTe or CdTe. Only Pb1–xCdxS exhibits a relatively wide miscibility region and shorter wavelength IV–VI diode lasers have been fabricated in this alloy system.

In order to obtain higher band gap materials, alloys of the lead salts with rare earth mono-chalcogenides (EuTe, EuSe, EuS, YbTe, SmTe, etc.) or alkaline earth chalcogenides (SrTe, SrSe, SrS, BaTe, BaS, CaTe, CaS) have been used. These compounds share the same rock-salt crystal structure as the lead salt compounds. Of the rare earth mono-chalcogenides most are metallic, with the exception of those with the divalent rare earth elements Eu, Yb, Sm and Tm. The corresponding highly ionic mono-chalcogenides are semiconductors with energy band gaps in the 1-2 eV range. The Eu chalcogenides are the ones with most stable rare earth ion in the 2+ state due to the half filled 4f shell. Thus, a complete miscibility exists for the Pb1– xEuxX systems (X stands for Te, Se, S), resulting in a large band gap tunability of more than 1.5 eV. The lattice constants as well as energy band gaps of the Pb1–xEuxX alloys depend nonlinearly on Eu content. For Eu contents below about 10%, the change in energy band gap is very large, with dEg/dx = 3.5, 3.0 and 5 eV for Pb1–xEuxTe, Pb1–xEuxSe and Pb1–xEuxS, respectively, with little changes in the lattice constants. Consequently, Pb1– xEuxX alloys have been the most widely used high energy band gap alloys for lead salt based heterostructures and devices.

The alternative group of materials for high energy band gap alloys are the alkaline-earth chalcogenides. These compounds such as CaX and SrX have energy band gaps above 4 eV and lattice constants comparable to that of the lead salt compounds. For the alloys Pb1–xCaxS and Pb1–xSrxS, a complete miscibility exists, and for Pb1–xCaxTe, Pb1–xSrx Te and Pb1–xSrx Se solid solutions have been obtained for Sr or Ca mole fractions at least up to 15%25.

For other alloys with alkaline-earth chalcogenides such as BaTe or MgTe, only a very small solubility region exists. In the lead/alkaline earth chalcogenide alloys again the energy band gap increases very rapidly with alkaline earth content with little change in the lattice constants. This makes

Lead salt thin film semiconductors for microelectronic applications

9

these alloys equally well suited for fabrication of lead salt based quantum well structures and MIR diode lasers [13].

One general problem associated with the high energy band gap lead salt alloys is the very rapid deterioration of the electrical properties of the epitaxial layers with increasing rare-earth or alkaline-earth content. In fact, for all of these alloys a decrease of the low temperature electron mobility by more than two orders of magnitude has been observed already for alloy compositions of only 10%. This has been explained by very strong alloy scattering of the charged carriers [24], which can even lead to a disorderinduced metal to insulator transition.

1.1.3. Dielectric property and refractive index An important property of the lead salt compounds is their huge static dielectric constant. This results from the fact that these materials, in particular PbTe, are close to a structural phase transition from the cubic to a rhombohedral phase that exhibits obvious ferroelectric properties. Thus, although PbTe remains cubic at all temperatures, the temperature dependence of the static dielectric constant ε0 ∝ C/(T–θ) can be associated with an extrapolated negative Curie temperature θ of –75 K. As a consequence, the static dielectric constant increases strongly with decreasing temperature, reaching a value of ε0 = 1350 at 4 K. The dielectric constants of the lead compounds in the mid-infrared spectral region are also exceptionally large, causing a refractive index above 5 for PbSe and PbTe (Fig. 1.5). By combining the lead salt compounds with wide band gap and low refractive index materials such as EuTe, EuSe, SrSe or BaF2 (as in Fig. 1.5), very high reflectivity epitaxial Bragg interference mirrors, as required in vertical cavity surface emitting lasers (VCSELs), can be obtained [25]. The refractive index contrast achieved by combining the lead salt compounds with these materials is up to five times larger as compared to those of III–V materials. As a result, the optical quality of lead salt based Bragg mirrors is significantly better than those of other materials [26].

1.2. IV-VI semiconductor growth In the early 1960s, the first semiconductor laser was successfully demonstrated on a GaAs substrate, later other group III-V compounds were used to fabricate a laser. The primitive lasers were all homojunctions and were functional only at low temperatures, and exhibited single mode emission around ~ 0.8 µm. Soon after, heterojunction lasers consisting of an array of mono-crystalline layers (e.g., GaAs, AlGaAs), were demonstrated on

10

Shaibal Mukherjee et al.

Figure 1.5. Refractive index of various III–V, II–VI, IV–VI, group IV semiconductors and selected fluorides and oxides plotted versus lattice constants.

(Source: Schwarzl et al).

GaAs substrate. In these multiple epitaxial layer structures, one layer generated stimulated emission, with the other layers providing both optical as well as electrical confinement of photons and charge carriers, respectively.

This design facilitated lasing at room temperature with an emission power of several milliwatts [27].

Various fabrication methodologies have been adopted to make these types of layered epitaxial structures. Liquid phase epitaxy (LPE) method was initially more widely used, followed by vacuum deposition techniques including vapor phase epitaxy (VPE), metal organic chemical vapor deposition (MOCVD), and molecular beam epitaxy (MBE). Among these methods, MBE has the unique advantage of epi-layer growth with precise and accurate control of layer composition and thickness. As a result of this precise control, MBE offered new opportunities for successful fabrication of micro- as well as nano-scale active layers, especially important in the development of quantum well (QW) lasers.

In 1964, the first mid-infrared p–n junction laser was made by Butler et al [28]. using Pb1–xSnxTe, and since then, efficient mid- and far-infrared IV–VI compound diode lasers have been fabricated, finding their main applications for remote sensing of gaseous pollutants in trace gas sensing devices, toxic gas analysis systems, human breath analysis in medical diagnostics and industrial process control. Recent progress in epitaxial growth techniques has led to the fabrication of lead salt midinfrared diode lasers operating at

Lead salt thin film semiconductors for microelectronic applications

11

temperatures up to 225 K in continuous (cw) mode [29] and up to 60 °C in pulsed mode [30]. In fact, lead salt IV–VI compound diode lasers can cover the whole 3-30 µm wavelength region and feature the highest operation temperature of conventional infrared band gap lasers. Apart from this, IV–VI MBE has also been used for the fabrication of detector arrays on Si substrates and of novel thermoelectric devices [31]. In addition, MBE has been extensively used for the growth of low dimensional electronic and magnetic systems like quantum wells [32], superlattices [33] and self-organized quantum dots [34].

In contrast to other compound semiconductors, the main constituents in IV–VI MBE are supplied from compound effusion sources loaded with PbSe, PbTe, PbS, SnTe, SnSe or GeTe, etc. This is due to the fact that the IV–VI compounds evaporate predominantly in the form of binary molecules, which means that nearly congruent evaporation occurs. While the degree of dissociation is only few percent for lead salt compounds, it notably increases for the tin and germanium chalcogenides. Therefore, additional group VI flux is usually used to assure the right stoichiometry of the epilayers, but only a very small excess group VI is required. By changing the total group IV to group VI flux ratio, the background carrier concentration and type of carriers can be controlled flexibly.

Excess group IV flux leads to n-type and excess group VI flux to p-type conductivity in the epilayers. For the high energy band gap alloys required for band gap engineering, the group IV elements are substituted by rare earth (Eu, Yb) or alkaline earth elements (Sr, Ca). Due to the very low vapor pressure of their chalcogenide compounds, these elements are supplied from elemental sources, which must be complemented by the corresponding amount of excess group VI flux. Dopants such as Bi, Tl or Ag can be supplied either from elemental or compound (e.g., Bi2Se3, Bi2Te3 or Tl2Te) sources. Generally, effusion temperatures for the IV–VI compounds are in the range of 400 to 600 °C, for the group VI elements (Te, Se) in the range of 150–300 °C and for the rare earth and alkaline earths around 450 to 900 °C. The substrate temperatures are typically in the range of 250 °C to 400 °C, where a unity sticking coefficient of all components except for Te, Se or S exists. Above 400 °C a significant reevaporation from the lead salt layers occurs and the growth rates and layer composition become increasingly difficult to control. 2D layer-by-layer growth can be achieved down to temperatures as low as 150 °C, as proven by in situ RHEED studies in which pronounced RHEED intensity oscillations were found over a large temperature range [22, 23].

Concerning the substrate materials for lead salt epitaxial growth, not only the lattice mismatch to common semiconductor substrates such as Si or GaAs

12

Shaibal Mukherjee et al.

is rather large (10% and more, see Fig. 1.1), but also the crystal structure and even more the thermal expansion coefficient β of the IV–VI compounds of around 20 × 10–6/K differ strongly from that of Si as well as that of the zincblende type III–V or II–VI compounds (β typically less than 6 × 10–6/K). As a consequence, large thermal strains are induced in the epitaxial layers during cooling of the samples to room temperature and below after sample growth.

For thicker epilayers, this leads to a significant structural degradation of epilayers or even to the formation of cracks and peeling during thermal cycling. A best compromise in these respects is achieved for BaF2 substrates, in spite of its different crystal structure (calcium fluoride structure). As shown in Fig. 1.1, BaF2 shows only a moderate lattice-mismatch to PbSe or PbTe (-1.2% and +4.2%, respectively) and moreover, the thermal expansion coefficient is almost exactly matched to that of the lead salt compounds.

Therefore, the epilayers are stable against thermal cycling. Furthermore BaF2 is highly insulating and optically transparent in the mid-infrared region. Thus, BaF2 has been the most widely used substrate material for the IV–VI epitaxial growth. Due to its high ionic character, good BaF2 surfaces can be obtained easily only for the (111) surface orientation. Therefore, in most cases, IV–VI heterostructures have been grown in this surface orientation. However the problems still exist regarding to the BaF2 substrates, such as low thermal conductivity, hard to cleave along (110) or (100) orientation etc.

1.3. Applications of lead salt materials 1.3.1. Detectors Narrow-gap lead chalcogenides (IV–VI compounds) such as Pb1-xSnxTe or Pb1-xSnxSe have been known for a long time and were employed as early as the 1970s to fabricate infrared detector arrays for thermal imaging [35].

They can cover the infrared (IR) range from <3 µm to >30 µm by choosing appropriate chemical compositions. The properties of IV–VI materials are remarkably different from the well-known II–VI or III–V compounds (see Table I). The ultimate sensitivities obtainable with IV–VI compounds are similar to those of II–VI narrow bandgap semiconductors such as Cd1-xHgxTe (MCT) under similar conditions (i.e., bandgap and operating temperature). They are limited by Auger recombination in both cases (see Table-1). This leads to maximal RoA (differential resistance at zero bias times area) products over a certain doping concentration: e.g., optimal concentrations for 0.1 eV bandgap devices at 80 K are ~2 ×1014 cm-3 to ~5 ×1015 cm-3 for MCT and ~1 ×1017 cm-3 to ~3 ×1017 cm-3 for Pb1-xSnxTe or Pb1-xSnxSe. The much higher optimal concentrations in Pb1-xSnxTe and Pb1-xSnxSe are due to the larger homogeneity region and high

Lead salt thin film semiconductors for microelectronic applications

13

Table 1.1. Some Properties of IV–VI Materials versus MCT.

permittivities in these materials. These high permittivities, as well as the high thermal expansion coefficients (7 times higher than Si), were one reason why development for IR focal-plane array (FPA) thermal imaging was nearly stopped in the early 1980s in favor of MCT. However, IV–VI materials remained the only choice to fabricate mid-wave infrared (MWIR) laser diodes before the invention of the quantum cascade laser [36], and continue to be of importance today.

Renewed interest started in the mid-1980s with the growth of IV–VI epitaxial layers on non-latticematched substrates such as Si. High-quality layers were obtained by molecular-beam epitaxy (MBE) on Si(111) substrates with a very thin CaF2 buffer layer for compatibility. Of most importance is the relief of the thermal mismatch strain by glide of dislocations: the dislocations glide on {100} planes, which are inclined with respect to the (111) surface plane, i.e., the Schmid factors and therefore the shear strains are not zero. On each temperature change, the threading ends of the misfit dislocations glide to remove the mechanical thermal strain built up. They never block (as might occur in zincblende type II–VI and III–V semiconductors) and in addition may even react to decrease the density of threading dislocations and therefore increase the structural layer quality [37]. Linear IV–VI IR sensor arrays on Si(111) were demonstrated with cut-off wavelength ranging from ~3 µm to ~15 µm [4]. A two dimensional (2D) FPA with the IV–VI layer grown epitaxially on an active Si chip containing the readout multiplexing electronics was also demonstrated, proving the concept that whole IR FPAs might be fabricated by post-processing of complete and tested Si-complementary metal oxide semiconductor (Si-CMOS) circuits [38].

14

Shaibal Mukherjee et al.

1.3.2. Lasers Lead salt semiconductor diode lasers have been widely used for highresolution gas spectroscopy [39] due to the numerous absorption lines of almost all molecular gases in the MIR spectral region. Due to their large wavelength tunability and their narrow line widths, lead salt lasers have been an excellent tool for gas sensing applications. In mid-infrared vertical cavity surface emitting lasers (VCSELs), the laser resonator is created by epitaxial growth of planar Bragg interference mirrors. Therefore, the laser light is coupled out perpendicular to the epilayer surface. VCSELs offer a variety of advantages such as a small beam divergence, large mode spacing and single mode operation and the possibility of monolithic integration, and they have a great potential for reducing threshold currents and increasing the operation temperatures.

Lead salt mid-infrared VCSELs were first demonstrated by Springholz’s research group [40] for wavelengths between 4.5 and 6 µm. This development was based on the realization of novel high-reflectivity epitaxial Bragg mirrors and ultra-high finesse microcavity structures for the infrared region. These mirrors consist of multilayer stacks of optically transparent dielectric layers with alternating low and high refractive index and an optical thickness equal to one quarter of the target wavelength λT for which the reflectivity is maximized. The high reflectivity is purely caused by multipleinterference effects and can be arbitrarily tuned from 0 to 100% just by changing the layer sequence and/or number of layer pairs. Due to the quite short length of the active region of around 1 µm of VCSELs compared to several hundreds µm for edge emitters, the reflectivity of the cavity mirrors must be very high, i.e., typically above 98%. This can be only achieved either by using a large number of λ/4 layer pairs or by combining materials with very large refractive index contrast. As shown in Fig. 1.1, the combination of lead salt materials (with high refractive index) (n ~ 5) with wide band gap materials such as EuTe, EuSe or BaF2 with n ~ 2.5 to 1.5 clearly fulfills this requirement. Due to the resulting very high refractive index contrast ∆n/n of up to 100%, ultra-high reflectivity Bragg mirrors with reflectivity above 99.5% can be achieved already with 3 layer pairs [27], in great contrast to the case of III–V Bragg mirrors, which require typically around 20 periods because the achievable refractive index contrasts are only around 15%. In addition, these materials are epitaxially compatible with the lead salt compounds, featuring lattice constants and thermal expansion coefficients that fit reasonable well to that of the lead salt compounds. Due to the very high refractive index contrast, lead salt Bragg mirrors exhibit ultra-broad high-reflectivity stop band regions with a spectral width of up to 80 % of the

Lead salt thin film semiconductors for microelectronic applications

15

central mirror wavelength. This makes them highly suitable also for passive optical elements such as low loss omni-directional mirrors and interference filters [41].

1.3.3. Quantum dots Strained-layer heteroepitaxy has become a powerful tool for fabrication of self-assembled semiconductor nanostructures [42, 43]. It is based on the natural tendency of highly strained layers to spontaneously form coherent, i.e., dislocation free three-dimensional (3D) nanoislands on the surface of a thin 2D wetting layer [44]. This islanding is driven by the highly efficient strain relaxation possible within the islands due to lateral elastic expansion or compression in the directions of their free side faces [45]. For islands larger than a certain critical size, such relaxed elastic energy outweighs the corresponding increase in free surface energy, leading to an effective lowering of the total free energy of the system. Overgrowth of these islands with a higher band gap barrier material results in a confinement of the free carriers in all three directions of space and thus, self-assembled quantum dots are formed. Due to the statistical nature of growth, however, these dots exhibit considerable variations in size and shape. This results in a large inhomogeneous broadening of the quantum confined energy levels and the corresponding optical transitions [46].

In addition, there is little control over the lateral arrangement and position of the nanoislands. Both factors pose considerable limitations for device applications. Three-dimensional stacking of self-assembled quantum dots in multilayer or superlattice structures provides an effective tool for controlling the vertical and lateral arrangement of the dots. PbSe/PbEuTe quantum dot superlattices represent a particular interesting system for investigation of such interlayer correlations. On the one hand, the elastic anisotropy is particularly large and therefore, an exceedingly efficient vertical and lateral ordering takes place. On the other hand, different dot stacking types occur for different spacer thicknesses as well as growth conditions.

Therefore, this system is very well suited for testing of various theoretical predictions of superlattice growth models. In addition, comprehensive systematic studies on interlayer correlation mechanisms have been carried out in this material system [47, 48].

Quantum dots can be made by self-organized growth from quantum-well structures, and also by photolithographic techniques. Due to the possibility to grow adjacent layers of quite different lattice constant in the IV-VI material family, strain can well be incorporated into superlattice structures. The inherent strain can be designed and predetermined by selection of the material

16

Shaibal Mukherjee et al.

composition and layer thickness. The first reported self-organized quantum-dot structure was made by MBE from Pb1â&#x2C6;&#x2019;xEuxTe (x = 0.05 to 0.1) spacers and PbSe wells, with some 5.5% lattice mismatch [35]. Growth resulted in PbSe islands that were embedded by consecutive spacers. While these islands appeared at irregular positions in the first PbSe layer, they got organized in subsequent periods, resulting in a face centered cubic (FCC) superstructure of the single dots localized in the wells. The lattice constant of the resulting dot crystal was shown to be variable and determined by changing the superlattice period. The comparative ease of making these quantum-dot crystals bears considerable potential for optoelectronic device applications.

CHAPTER 2: GROWTH PITS OF LEAD SALT EPITAXIAL LAYERS GROWN ON SILICON SUBSTRATES 2.1. Defects in semiconductors It is extremely crucial to understand the behavior of defects in semiconductors in order to obtain improved performance from semiconductor devices having wide applications in civil and military technology. The focus of this research is the study of one kind of defects (growth pits) in IV-VI Pb-salt semiconductors. The goal is to understand how this kind of defect originate, their shapes, and their effects on the basic properties of the materials.

The physical characteristics of semiconductors are determined both by the properties of the host crystal and by the presence impurities and crystalline defects. Dopant impurities, which typically substitute for a host crystal atom, introduce electronic states in the bandgap close to the valence and conduction band edges and thus determine the type and conductivity of the material. These so-called shallow level defects enable the wide range of semiconductor devices available today. However, crystal lattice defects or other impurities, which introduce electronic states deeper in the bandgap and are referred to as deep level defects, also modify the properties of the semiconductor and thus may make a semiconductor unsuitable for its intended applications.

When a semiconductor film is grown epitaxially on a substrate that has a slightly different lattice constant, the lattice mismatch strain in the film may be relaxed by the introduction of misfit dislocations or other lattice defects.

The electronic and optoelectronic devices for many important semiconductor applications are complex engineered structures in which both the lattice mismatch strain and the defects must be tightly controlled.

As is well known, some of the defects in the semiconductors play a dramatic role on the performance of the devices. For instant, detectivity (D*) of photovoltaic (PV) detectors is restrained by threading dislocations and

Lead salt thin film semiconductors for microelectronic applications

17

operability of detectors is limited by growth pits density in the semiconductors. While the main defects in Pb-salt semiconductors are exactly threading dislocations and growth pits.

2.1.1. Threading dislocation in IV-VI materials The threading dislocations in IV-VI semiconductors have already been studied [49, 50]. Researchers adopted various methodologies to reduce the density of threading dislocation in lattice-mismatched epitaxial layers. For Pb-salt semiconductors grown on the Si (111) substrates, the lattice and thermal expansion mismatch strain can be relaxed by the dislocation glide mechanism in the main {100} <110> glide system for epilayers which have the NaCl crystal structure. Fig. 2.1 demonstrates a schematic diagram of the glide and crystallographic geometry. Nearly complete strain relaxation is obtained in layers with a thickness of several micrometer, and on each temperature cycle ranging from 80 K to 300 K. Even after more than 1400 such cycles, plastic strain relaxation still occurs on each temperature change, and it was estimated that the layers had undergone a cumulative plastic deformation exceeding 400% in report [51]. The structural quality has changed only slightly after these procedures. Even higher quality of epilayers is obtainable if the samples are annealed at 200 â&#x20AC;&#x201C; 400 0 C in vacuum or Se vapor environment. An estimate based on geometrical arguments and the experimental findings was performed which predicted movements of the threading ends of misfit dislocations in the centimeter range in layers with dislocation densities of 108 cm-2 [52].

Figure 2.1. Schematic drawing of the arrangement of the {100} <110> glide system for the NaCl type PbSe (111) layers on Si(111). (a) Perspective drawing; (b) arrangement of possible a/2<110>-type Burgers vectors; dashed lines: Burgers vectors which are inclined to the interface and therefore belong to glissile dislocations.

(Source Muller et al).

18

Shaibal Mukherjee et al.

Figure 2.2. Distribution of etch pits on rectangular PbSe (111) islands on Si(111) before (a) and after (b) a thermal cycle to 300 0C. The crystallographic orientation is plotted pointing into the same directions as in Fig. 2.1 (Source Zogg et al).

Etch-pit (threading dislocation) densities of typical PbSe layers after growth were 1 Ă&#x2014; 107 to 3 Ă&#x2014; 107 cm-2. To demonstrate the easy mobility of threading ends, rectangular islands were etched into the layers. Fig. 2.2(a) shows the etch-pit density of an as-grown island. A homogeneous distribution of etch pits is observed. Similar islands of the same sample (but where no etch-pit treatment was applied) were heated to 300 0C, and etched after cool down to RT for etch-pit analysis. Fig. 2.2(b) shows a micrograph of such an island. The most striking feature is the extremely reduced etchpit density over most of its interior. On all such rectangular islands investigated, similar patterns were reproducibly observed. All showed the same features, a very low dislocation density in the interior, and an increased density along a U-shaped band located a few micrometers away from the edges. This band runs along three of the four edges only, while no increased density is observed along the fourth edge (Fig. 2.2(b) right side). This latter edge runs parallel to the [101] direction formed by the intersection with the surface of the (010)-glide planes inclined away from the edge (i.e., the (010)-glide planes plotted in Fig. 2.1 with the island on the left). On the other three edges with a different crystallographic orientation, a band with increased etch-pit density was observed. The asymmetry is correlated with the mechanical strain, which decreases to zero at the edges. For most orientations, it becomes too small near the edge to move the threading segment of the misfit dislocation completely out of the island. Only for edges oriented along [101] and with the (010)-glide planes inclined away as described, the strain field is changing in a manner that the threading ends can escape completely.

Lead salt thin film semiconductors for microelectronic applications

19

Figure 2.3. Etch-pit density in PbSe (111) on Si(111) vs thickness d of the layers (points). A 1/d2 dependence is indicated by the dashed line. (Source Fach et al).

The decrease of dislocation densities (etch-pit densities) with increasing layer thickness is plotted in Fig. 2.3. A similar sample was etched down to different thicknesses and annealed, and the revealed etch pits in the interior of the samples were counted. With increasing thickness, the threading dislocation density at the surface of the layer decreases with approximately the square of the thickness d, which is different from III-V or II-IV semiconductors, where a 1/d scaling is observed, or even a leveling off of the density for thicker layers [53]. The 1/d dependence can be understood as the probability that two threading dislocations meet during growth in order to annihilate is proportional to dislocation density Ď 2 [54]. The near 1/d2 dependence found in the study for annealed PbSe can be explained as follows: For a certain grown layer with thickness d (and before performing anneals or temperature changes), the threading dislocation density existed during growth scales with 1/d as described. When changing the temperature (after growth is completed), one has to take into account that each threading end of the misfit dislocations is able to glide along the whole layer dimension. If a threading dislocation does not encounter and react with any other threading dislocations on its glide across the whole sample, no change in the 1/d dependence results. However, assuming that this moving threading end encounters another threading end and reacts (annihilates or fuses) with it, a further dislocation reduction takes place. Remembering that all threading dislocations behave in this manner, the result must be that this reduction

20

Shaibal Mukherjee et al.

again scales with 1/d in order to finally achieve the observed 1/d2 dependence.

An explanation of this dependence has to take into account the inhomogeneous distribution of dislocations across the layer thickness. The junctions where two threading ends have reacted in the interior of the layer may be able to move further towards the interface as long as the dislocation density is low. Near the interface, the dislocation density increases and the junctions encounter more obstacles to move down. Thus, a higher decrease of the dislocation density is possible closer to the surface. Clearly, a 1/d2 dependence as compared to a 1/d dependence leads to a faster decrease of dislocation density with increasing layer thickness, which is of value for growth of thick layers with very low dislocation density. The exact nature of the dislocation interactions for these NaCl-type semiconductors is not known and deserved further studies.

2.1.2. Growth pits in II-VI semiconductors As compared to IV-VI materials, growth defects in II-VI semiconductors have been investigated extensively [55-57]. For example, in the prevailing IIVI material, Hg1-xCdxTe (MCT), different growth defects have been identified as dislocations, stacking faults and twins, voids and hillocks, as well as precipitates. Fig. 2.4 shows typical defect complexes (one kind of voids defects) in MCT. From the images, one can see the voids exist inside hillocks, or voidhillock complexes. These defect complexes appear to be associated with a nest of ‘decorating’ dislocations. This becomes apparent upon defect etching small sections of films such that the location of each individual defect complex is relatively precisely correlated during the etching process.

The large voids, with sizes usually greater than 5 µm, nucleated at the substrate-epilayer interface. The smaller voids, however, appeared to nucleate away from the substrate-epilayer interface. Direct evidence of these voids on cleaved cross-sections of MBE-grown films has been obtained earlier [58] and reproduced in Fig. 2.5. As displayed, usually the voids, once nucleated either at the substrate-epilayer interface or away from it, continued to replicate through to the top surface of the films.

Fig. 2.6 displays a fraction of the voids appears to close before reaching the film surface. Here the voids nucleated slightly away from the filmsubstrate interface, continued to replicate for a while as the growth progressed, but then relatively rapidly closed off at significant depths from the film surface. The whole sequence was completed before two-thirds of the film growth was completed. Examination of the top surface of these films does

Lead salt thin film semiconductors for microelectronic applications

21

Figure 2.4. (a) Defects which appear to be voids inside hillocks. (b) defect complexes where the hillocks appear to be associated with void edges. (Source: Chandra et al).

Figure 2.5. Cross-sectional and three dimensional SEM microphotographs of void defects nucleated at various stages of MBE growth: at the growth interface, (a) in the middle of the growth run, (b) near the end of the growth run. (Source: Aoariedn et al).

22

Shaibal Mukherjee et al.

Figure 2.6. Cross-sectional SEM microphotograph of void defects which “closed” before the end of the growth. These voids will be completely hidden and not apparent from the top surface. (Source: Frazier et al).

Figure 2.7. Dislocation etching of a MBE epilayer containing “hidden” voids. (a) Dislocation nests are not apparent on the top surface, (b) but become apparent upon removal of half of the film. (Source: Shih et al).

not reveal any indication of voids within the depth of the films. These only become apparent upon examination of cleaved cross-section of these films.

Voids, which close during growth, can nucleate either at the interface or away from the interface. Furthermore, these voids may not always be leaving any “fingerprints” on the top surface. Defect etching these films did not indicate a presence of nesting dislocations. However, void density increased upon

Lead salt thin film semiconductors for microelectronic applications

23

selective and progressive removal of the epilayers. When defect etching was performed following a partial removal of the film, dislocation nests were observed decorating these voids in some cases. An example is displayed in Fig. 2.7. No nests or decorative pattern of dislocations were observed upon defect etching the top surface of a selected small section of a MBE film (Fig. 2.7 (a)). Upon removal of approximately half of the film thickness by chemical etching, this section was then defect etched. A number of dislocation nests springs into view (Fig. 2.7 (b)).

As stated above, void defects during MBE growth of HgCdTe can nucleate at the substrate-epilayer interface as well as away from it. The latter kind of void defects is smaller in size compared to voids nucleating at the interface and appears to be usually present as a defect complex, which leads to the formation of an associated decorative dislocation nest. Both kinds of voids usually penetrate through to the surface of the films. However, examples where voids have closed before reaching the top surface of the films have been found. No indications of these hidden voids can be observed by examining the top surface of the films. Elimination of fluctuations in growth conditions, usually likely to be associated with multi-layer and multicomposition films, eliminated the formation of these complex defects.

2.2. Investigation of growth pits in IV-VI semiconductor Although IV-VI semiconductors like Pb1-xSnxSe show a comparably high capability in comparison to II-VI and III-V materials, such as Hg1-xCdxTe and InSb [59, 60], not many efforts have been devoted to exploring the full potential of IV-VI materials. Recently, an investigation of defects in the IV-VI Pb-salt materials was carried out by our research group and a significant conclusion in nature of growth pits was made according to our experimental results.

2.2.1. Samples preparation In this phase, a monocrystalline PbSe layer was epitaxially grown on silicon substrate in a customer-designed MBE apparatus. Throughout the growth process, the reflection high energy electron diffraction (RHEED) patterns were recorded for confirmation of the quality of epilayer. Further details regarding to the growth have been described in elsewhere [61]. The surface morphology of epilayer was then characterized by Nomarski microscopy and scanning electron microscopy (SEM). Energy dispersive xray analysis (EDXA) was performed on the special area of the epilayer surface to investigate the composition of the growth defects.

24

Shaibal Mukherjee et al.

Figure 2.8. Nomarski micrograph of an as-grown PbSe epilayer on silicon substrate.

Figure 2.9. (a) SEM image of PbSe epilayer grown on a Si (111) substrate, (b) plan view, (c) 60o oblique view, (d) three dimensional view of the circled part in Fig. 2.9 (a).

Lead salt thin film semiconductors for microelectronic applications

25

Fig. 2.8 shows a Nomarski micrograph of an as-grown PbSe epilayer surface on a Si (111) substrate. A number of bright spots with various sizes are conventionally called “growth pits” or surface defects, which are also prevalent in epitaxial semiconductor materials such as Hg1-xCdxTe and SiC [62, 63]. The detailed information about the growth pits in PbSe epilayer is presented through SEM observation. Fig. 2.9 (a) shows a typical SEM image of a PbSe epilayer grown on a Si (111) substrate. Higher resolution images of a small growth pit (marked by a circle) are displayed in Fig. 2.9 (b) and (c) in plan view and in 60° oblique view respectively. It is very interesting to note that the growth pit, as in Fig. 2.9 (a), is composed of two micro-sized cuboids, both of which have rectangular facets approximately parallel to the (111) crystal plane of Si substrate. The 3-dimensional (3D) display of the micro-cuboids in Fig. 2.9 (c) shows that the upper part of the larger microcuboid extrudes outwards from the epilayer, and the facets of the microcuboid are vertically intersecting each other. Fig. 2.9 (d) describes the three dimensional image image of another micro-cuboidal growth pit. While the orientations of the micro-cuboids differ from one structure to another, similar micro-cuboids are readily observed in other MBE-grown PbSe samples.

2.2.2. Composition of micro-cuboids EDXA is conducted in order to investigate the chemical composition of the micro-cuboids. From the EDXA spectrum of the micro-cuboidal growth pit, as shown in Fig. 2.10, the two most intense peaks are identified to be from “lead” and “selenium” elements. Elemental analysis of EDXA shows that there is a very small compositional difference between micro-cuboids and the surrounding epilayer. The atomic percentages of Pb and Se are 51% and 46%, respectively, which indicates that both the micro-cuboids and the epilayer are made of PbSe. Considering that IV-VI semiconductors like PbSe and PbTe possess a cubic NaCl-type structure, it is reasonable to identify the micro-cuboids as PbSe micro-crystals since both facets and regular shapes of micro-cuboids are of single crystalline character. The facets of the microcuboids in Fig.2.9 are most likely attributed to the {100} plane of lead salt materials. The RHEED patterns, however, indicate a three-fold rotational symmetry during the growth, which means that the crystalline orientation of the PbSe epilayers is along the <111> direction of Si substrate. The discrepancy of crystal orientations between PbSe epilayer and micro-crystal suggests that the PbSe micro-crystals may originate from a self-assembly growth mode which is completely different from the growth mechanism of the surrounding PbSe epilayers.

26

Shaibal Mukherjee et al.

Figure 2.10. A typical EDXA spectrum of micro-cuboids.

2.3. Multiple growth pits In addition to the smaller growth pits (~1µm), several larger growth pits were also observed as shown in Fig. 2.11 with a 60° oblique view. The size of those pits is more than 5µm. Obviously, this larger growth pit is a direct result of a stack of micro-cuboids with the size of each cuboid around 0.5 to 1 µm. However, every cuboid has a specific crystal orientation, and the size of single micro-cuboids is not uniform. Some of the smaller cuboids are so close to their neighbors, that it is difficult to identify individual cuboid because they overlap or merge with each other, appearing in irregular geometrical shapes.

Figure 2.11. SEM image of a cluster of single micro-cuboids.

Lead salt thin film semiconductors for microelectronic applications

27

2.4. Origin of growth pits The space between a micro-cuboid and surrounding epilayer is always quite narrow, which indicates that the micro-cuboids are in fact embedded in PbSe epilayer rather than simply lying loosely on the surface. This fact is confirmed in Fig. 2.13, where a cross-sectional SEM image of a cleaved PbSe sample clearly displays that the “roots” of the micro-cuboid are located near the middle of the epilayer. The depth of craters increases with the increase of size of cuboids. The nucleation of PbSe micro-crystals/micro-cubiods occur at various stages of the epitaxial growth, while the geometrical dimensions of micro-crystals increase along specific crystal orientations as the epitaxial growth proceeds.

Figure 2.12. SEM image of crater after the PbSe surface was blown by high purity nitrogen gas.

Figure 2.13. Cross-sectional SEM image of a cleaved sample which clearly shows that the “roots” of the micro-cuboid are located near the middle of the epilayer.

28

Shaibal Mukherjee et al.

2.5. Conclusions The growth mechanism of lead salt defects in forms of micro-cuboid has been depicted in this chapter. Each individual micro-cuboid comprises of the criteria of a high quality PbSe micro-crystal. But, in a collective fashion, the random distribution of crystalline orientation of micro-cuboids is a strong indication of polycrystalline growth. This leads to a tentative assumption that PbSe polycrystalline seeds provide the initial stage of micro-cuboidal growth.

As growth continues, single crystalline PbSe micro-crystals start growing randomly along various crystal orientations of the polycrystalline seeds.

Apparently, the growth along the <100> orientation is faster than the growth in other directions since the {100} facets have a lower surface energy than that of the higher index planes. The validity of this reasoning is reinforced by the growth of other PbSe nano-crystals whose geometrical shapes vary from spherical to highly faceted truncated octahedrons to cubic with sizes ranging from several nanometers to tens of nanometers [64]. It is considered that once the lead salt polycrystalline seeds are formed at a certain stage during the crystal growth, the high growth rate of micro-cuboids promotes the out-ofplane growth and the self-assembling of micro-cuboids.

It is interesting to note that, dislocation density of IV-VI materials is not increased near growth pits unlike its II-VI MBE-grown counterparts. The dislocation density is always higher near growth defects like voids and void/hillock for II-VI materials [65]. Whereas for IV-VI materials, the gaps between micro-cuboidal defects and the epilayer are considered to play an important role in the reduction of growth dislocation. These gaps prevent stems of micro-cuboids from coming in contact with the epilayer surface, and thus the micro-cuboids could not generate extra strain zones on the epilayer area surrounding them. Once passivated by post-growth treatments, these larger defects, which are actually highly-crystalline lead salt materials, are not expected to negatively impact the performance and reliability of fabricated devices. Moreover, it should be noted that the similar kind of lead salt micro-crystals are also observed in PbSe and PbSnSe epilayers grown on a BaF2 (111) substrate.

Possible explanations behind the polycrystalline seed nucleation include the large lattice mismatch between lead salt materials and silicon substrates and/or surface contaminations. In this chapter, only preliminary results and discussions about the structure and morphology of the growth pits in lead salt semiconductor on Si substrate are represented. In order to further investigate the growth mechanism of defects, IV-VI epilayers grown on the different substrates should be considered.

Lead salt thin film semiconductors for microelectronic applications

29

CHAPTER 3: A NOVEL IN-SITU SURFACE TREATMENT DURING MBE GROWTH OF PBSE ON CAF2/SI(111) HETEROSTRUCTURE 3.1. Introduction of surface treatment for II-VI epitaxy In-situ surface treatment is extensively used in the interfaces during the growth of semiconductors on mismatched substrates. For example, a nitrogen-plasma (one kind of in-situ surface treatment) was applied to the interfaces between substrates and epilayers during MBE growth of ZnSe on Si (111) for improving the quality of the epilayers [66].

It is well-known that a high quality ZnSe epilayers are difficult to grow due to the chemical and mechanical mismatch between ZnSe and Si. A ~ 4.3% lattice mismatch and dissimilar thermal expansion coefficients responsible for the creation of crystal defects like dislocations and stacking faults [67, 68]. On the other hand, it is known that an imbalance in the interface charge is one of the serious problems associated with the growth of a polar semiconductor on a nonpolar one [69]. For example, at the (100) interface between Si and the Se plane of ZnSe, the interface bonds will each receive a total of 5/2 electrons per bond instead of the two electrons required. This excess charge would generate a huge internal electric field. Harrison et al [69] suggested that in order to eliminate the excess charge the growth process itself produces modifications in the ideal atomic arrangement resulting in a distorted interfacial geometry. Moreover, if the bonding problems at the interface are such that in the absence of any flux of Se, the sticking coefficient of Zn on the Si surface is very small [70]. Furthermore, chemical reactions between Si and Se at the interface during the very initial stage of growth cause the formation of a thin SiSex amorphous interlayer [71, 72]. This layer hinders the smooth growth of ZnSe, resulting in a high density of crystal defects in the films. From the above statements, it is clear that the control of the initial stages of growth is essential in order to reduce the defect density.

The researchers describe the effect of Si substrate surface irradiation with plasma of nitrogen (N-plasma) prior to the MBE growth of ZnSe, and found that a substantial improvement on the crystal quality of ZnSe epilayers on Si(111) can be achieved by this novel substrate surface treatment with Nplasma. The implementation of such surface treatment helps in obtaining two dimensional growth of ZnSe on Si, as revealed by means of reflection highenergy electron diffraction (RHEED) oscillations at the initial stages of growth.

3.1.1. N-plasma Si surface treatment N2 gas is introduced inside the MBE chamber when the oxide is fully desorbed from the Shiraki-cleaned Si (111) surface. In order to increase the

30

Shaibal Mukherjee et al.

reactivity of nitrogen with the Si surface, an RF-plasma discharge source is used to produce highly reactive nitrogen species [73]. After the N-plasma treatment the substrate surface shows a RHEED pattern with bulk 1×1 streaks and a blurred 7×7 reconstruction, as seen in Fig. 3.1.

Fig. 3.2 shows the RHEED pattern obtained after the growth of 8 monolayers (ML) of ZnSe by conventional MBE directly on Si(111) without the N-plasma treatment. The intense bulk spots and the diffused background are indications of a three-dimensional growth mode. With further growth, crystal defects become prominent by the appearance of extra-spots indicating the formation of twinned regions, as observed by the RHEED pattern of Fig. 3.3 that corresponds to the growth of 30 ML of ZnSe. It is difficult to achieve a smooth ZnSe growth by this technique. The three-dimensional islands resulting from the direct MBE growth on untreated Si substrates are illustrated in Fig. 3.4.

Figure 3.1. RHEED patterns of N-plasma treated Si(111) substrate. (Source: MéndezGarcía et al).

Figure 3.2. RHEED patterns of Growth of 8 ML ZnSe on an untreated Si(111) substrate.

(Source: López-Lópezet et al).

Lead salt thin film semiconductors for microelectronic applications

31

Figure 3.3. RHEED patterns of 30 ML of ZnSe grown by conventional MBE directly on an untreated Si(111) substrate. (Source: Méndez-García et al).

Figure 3.4. Atomic force microscopy (AFM) image of 30 ML of ZnSe grown by conventional MBE directly on a treated Si substrate. (Source: Hernández-Calderón et al).

3.1.2. Improvement of growth quality by surface treatment A massive improvement in the epitaxial growth quality is achieved during the growth on a Si surface pre-treated with N-plasma. Fig. 3.5 shows the RHEED pattern obtained after the growth of 8 ML of ZnSe by pulsed MBE on an N-plasma treated substrate. The RHEED pattern looks very streaky; moreover a 2-fold reconstruction can be observed. This indicates a two-dimensional growth mode and an atomically ordered epilayer surface.

The improvement in epitaxial growth quality after surface treatment is established by comparing the RHEED patterns with the corresponding patterns obtained from direct MBE growth on untreated substrates (Fig. 3.2).

The RHEED pattern, as shown in Figs. 3.5, have a low background and are much streakier, indicating a flat surface.

32

Shaibal Mukherjee et al.

Figure 3.5. RHEED patterns of 8 ML of ZnSe grown by pulsed MBE on an N-plasma treated Si(111). (Source: L贸pez-L贸pezet et al).

Figure 3.6. AFM image of a 30 ML of ZnSe grown by pulsed MBE on an N-plasma treated Si substrate. (Source: Hern谩ndez-Calder贸n et al).

Auger measurements reveal a better ZnSe-coverage for the N-plasma treated Si-substrate surface, supporting the idea of a layer by layer growth promotion by this technique. The AFM image in Fig. 3.6 confirms the flatter ZnSe surface obtained on the N-plasma treated Si surface.

3.1.3. Growth model The two dimensional growth could be induced by the formation of a nitrogen interlayer that avoid the formation of the SiSex compound, and could balance the electric charge at the ZnSe/Si interface, thus allowing the growth of a smooth and uniform ZnSe epilayer.

Lead salt thin film semiconductors for microelectronic applications

33

Figure 3.7. Schematic picture of a possible atomic arrangement in the early stages of the ZnSe growth. (Source: Uhrberg et al).

Figure 3.8. Schematic picture of a possible atomic arrangement of the Si surface after the N-plasma treatment. (Source: Uhrberg et al).

For the (111) interface a sub-monolayer amount of nitrogen is enough.

For example, charge balance occurs when the N atoms resplaces Si atoms and covers roughly ~ 50% of the surface sites. These N atoms share three chemical bonds with the substrate, as illustrated in Fig. 3.7. In this way the epitaxial growth continues smoothly continue with a complete monolayer of Zn atoms bonded to the surface formed by Si and N, as shown in Fig. 3.8. A different sub-monolayer thick of N could be possible depending on the particular atomic geometry at the substrate (Si) surface. The partial coverage of the Si(111) surface with nitrogen could be reflected by the blurred 7Ă&#x2014;7 reconstruction after the N-plasma treatment, as depicted in Fig. 3.2.

34

Shaibal Mukherjee et al.

3.2. In-situ treatment for IV-VI epitaxy Although various methods have been applied to reduce the defects in IVVI epilayers grown by MBE, such as low and high temperature cycling, defects are still a major challenge for device applications especially for the structure grown on Si substrate. In particular, the growth pits and etch pits in the Pb-salt semiconductors are relatively higher than those in III-V and II-VI materials. For instance, the threading dislocations density in Pb-salt epilayers is usually from high 106 to low 108 cm-3, compared to MCT with dislocation density of only 104 ~ 105 cm-3. Hence, reduction of dislocations in IV-VI semiconductors is critical for high performance device applications. Here a novel method of surface treatment is detailed in order to reduce dislocation density. This method, as described in the following section, will reduces the thermal mismatch as well as the lattice mismatch between the substrate and epilayer interfaces.

3.2.1. Motivation The continuous improvement of IV-VI materials grown on silicon substrates by MBE system is a key step in the development of IV-VI infrared semiconductor devices. In this chapter, a novel surface treatment method during MBE growth of monocrystalline PbSe on Si(111)-oriented substrates is described. Details of the experimental procedures, supported by reflection high-energy electron diffraction (RHEED) patterns, are reported. The positive effect of the in-situ surface treatment method is realized in terms of improved electrical and morphological properties of PbSe epitaxial layers. To be more specific, the carrier mobility increases almost three-fold at 77 K and nearly two-fold at 300 K compared to their respective values from epilayers without surface treatment. The density of the growth pits undergoes almost a threefold reduction, whereas the density of the threading dislocations reduces ~ four times after surface treatment.

3.2.2. Traditional growth of lead salt on Si One of the most important applications for IV-VI lead salt semiconductors is the detection of light in the mid- and far- infrared region.

Growth of homogeneous, high quality materials on Si substrates makes IV-VI semiconductors a promising candidate for the large format focal plane array (FPA) detectors [74]. It has already been shown that IV-VI detectors especially those based on Pb1â&#x2C6;&#x2019;xSnxSe materials grown on Si substrate offer high sensitivity similar to that of MCT detectors [75, 76]. Nonetheless, the

Lead salt thin film semiconductors for microelectronic applications

35

full potential of IV-VI semiconductor detectors is not yet entirely realized, mainly due to the lattice and thermal expansion mismatch of IV-VI materials with Si substrate. These mismatches introduce defects such as dislocations. As is well known, defects in semiconductors play a critical role on the performance of the devices. There are primarily two kinds of defects in epitaxial IV-VI semiconductor materials â&#x20AC;&#x201C; threading dislocations and growth pits.

The thermal expansion and lattice parameter mismatches between PbSe and silicon substrates are about 12% and 700% at 300 K respectively (as shown in Table-3.1). Despite these large mismatches, high quality IV-VI layers can still be grown on (111)-oriented silicon substrates successfully and hence reported [77]. Initial work in this area focused on using a stacked BaF2/CaF2 buffer layer on Si substrate, but later MBE-growth results have shown that just a thin CaF2 layer produces good PbSe epitaxy [78]. This fluoride buffer layer, even when as thin as ~ 20 Ă&#x2026;, appears to be necessary for good PbSe epitaxy since attempts to grow PbSe directly on Si(111) substrate have resulted in poor layer quality [79]. The layer-by-layer growth modes for both CaF2 and PbSe are observed by RHEED intensity oscillations and reported in literature. After growth, the epilayers are reported to be entirely free of cracks and their crystalline quality is reasonably good as evidenced by high resolution X-ray diffraction (HRXRD) with full-width half-maximum (FWHM) values of typically less than 200 arcsec [80].

However, the dislocation density in all these reports has still been in the range of high 107 cm-2, which needs to be further reduced for better device performance.

Table 3.1. Properties of semiconductor materials related to surface treatment at 300 K.

36

Shaibal Mukherjee et al.

Until date, post growth annealing seems to be the most effective way implemented to reduce the dislocation density in IV-VI epilayers after the growth on mismatched substrates. However, the lowest dislocation density achieved by annealing seems to be limited to a mid- 106 cm-2 even after 1400 cycles [81]. So it is highly desirable to develop other ways, such as surface treatment, to reduce defect density either before or after growth. Based on this idea, a MBE-growth procedure is meticulously designed to obtain a high quality growth of IV-VI lead salt epilayers with much reduced defect density. Several buffer layers are applied between CaF2/Si substrate and PbSe epilayer in relation to their crystal structure, lattice constant and thermal expansion coefficient.

3.2.3. Design of experimental procedure Based on the properties of IV-VI semiconductors and other materials encountered during an MBE growth of PbSe on CaF2/Si(111) heterostructure, an experimental procedure is adopted for further reduction of growth defects and thread dislocations. Considering material properties such as the crystal structure, lattice parameters and thermal expansion coefficients, MBE growth sources, and growth chamber conditions; a detailed growth approach has been designed. This design takes essential growth parameters; such as substrate temperature, source flux, growth time; into consideration. The detailed growth mechanism is pictorially depicted as shown in Fig. 3.9. In the traditional growth method, the PbSe film is directly grown on CaF2/Si(111) substrate without any additional surface treatment. However, due to the modified growth procedure, several layers are grown between a substrate and the final epitaxial layer as demonstrated in Fig. 3.9. In order to carry out the novel growth procedure, strontium MBE source is kept open for a while right after the CaF2 growth. Therefore, it is expected that a single atomic layer of SrF2 is formed on top of the CaF2 surface at high surface temperature. It is important to mention here that no extra strain is generated during this growth step because of the same crystal structure of CaF2 and SrF2. In the next step, selenium source is opened for growth of SrSe on SrF2. It is presumed that the Se atoms are only reacted by Sr atom on the outer surface and thus the surface is covered by Se atoms after this step of growth. Finally, PbSe is grown on SrSe with reduced lattice mismatch as compared to direct PbSe growth on CaF2.

3.2.4. In-situ surface treatment procedure As stated above, Pb-salt semiconductors are generally grown either on cleaved bulk BaF2 or on silicon with fluoride buffer layers. In this improved in-situ growth procedure, PbSe thin films are grown on a (111)-oriented Si

Lead salt thin film semiconductors for microelectronic applications

37

Bulky PbSe

Lead atoms Selenium Atoms Strontium atoms Fluorine atoms Calcium atoms

Si wafer Figure 3.9. Schematic diagram of layer by layer growth of the in-situ surface treatment.

Figure 3.10. RHEED pattern (20 kV) from a clean Si (111) surface at 770 °C with the streaky (7x7) surface reconstruction along [112] incidence.

substrate covered by several buffer layers using a compound PbSe source and an elemental Se source. A single-sided polished p-type Si substrate (resistivity ~ 100 ohm-cm) was prepared by the modified-Shiraki cleaning method for epitaxial growth. When the substrate temperature reached 770 ºC, a clear 7×7 reconstruction RHEED pattern (as shown in Fig. 3.10) is observed. This indicates the top oxidation layer is thoroughly removed and the substrate surface is ready for fluoride buffer layer growth. A 20 Å thick CaF2 layer is then grown as a buffer layer whose crystalline quality, as measured by RHEED pattern, is given in Fig. 3.11 (a).

38

Shaibal Mukherjee et al.

(a)

(b)

Figure 3.11. RHEED image in the [112] azimuth (a) during growth of the CaF2 layer at a growth temperature of 770 0C and (b) during growth of the SrF2 layer at a temperature of 530 0C.

Figure 3.12. RHEED image in the [112] azimuth during exposure of the SrF2 /CaF2 layer in selenium flux at a temperature of 530 0C.

The substrate with buffer layers is then transferred to the main MBE chamber for PbSe growth. At this time, an optimized surface treatment method is applied to CaF2/Si heterostructure. When the substrate is heated up to 530 ºC, the strontium source associated with the main chamber is opened for 200 seconds. Thus the substrate surface is covered with SrF2 as evidenced by RHEED in Fig.3.11 (b). Subsequently, the selenium source attached to the main chamber is opened for 30 seconds and at this point, the surface is covered with SrSe as evidenced by RHEED image shown in Fig. 3.12. Finally, the substrate temperature is set at 375 ºC and a PbSe single layer is grown on the treated substrate. The RHEED pattern in Fig.3.13 confirms the epitaxial growth of the lead salt layer. During the growth of PbSe, a 12% Se-to-PbSe flux ratio is maintained while the growth rate is kept constant at 2.0 µm/hr.

Lead salt thin film semiconductors for microelectronic applications

39

Figure 3.13. RHEED image in the [112] azimuth during growth of the PbSe layer at a temperature of 375 0C.

3.2.5. Results and Discussion The buffer layers grown by the optimized in-situ surface treatment at the interfaces between CaF2/Si and IV-VI epilayer are to reduce lattice and thermal mismatches between adjacent layers in a gradual manner [82]. At first, a thin CaF2 layer is grown on the Si (111) substrate and the lattice mismatch is calculated to be 0.6% at 300 K (as seen from Table-3.1). The thickness of CaF2 layer is controlled within the critical thickness of CaF2 on Si substrate at the same temperature. In the next step, CaF2 layer is exposed to strontium flux so as to form a single atomic layer of SrF2, whose thermal expansion coefficient nearly matches with that of CaF2. There is no visible change on the RHEED patterns during the growth of SrF2 as observed in Fig.3.11 (a) and Fig.3.11 (b). This is due the similar crystal structures of CaF2 and SrF2. It is interesting to note that their lattice mismatch is only 5 % at 300 K (as shown in Table-3.1). However, there is an indirect evidence which supports the formation of SrF2 on top of CaF2. If a CaF2 layer, without the strontium treatment, is directly exposed to the selenium flux at 530 ยบC, RHEED patterns would not produce any visible change even after long time of exposure. But, there is an obvious change in the RHEED patterns observed due to the strontium exposure followed by selenium treatments as depicted in Fig. 3.12. In the next step, SrF2 layer is exposed to selenium flux in order to deposit a single atomic layer of SrSe on SrF2 surface. As previously mentioned, there is a visible change on the RHEED patterns during growth of the SrSe layer because the crystal structures of SrF2 (Cubic Closest Packing) and SrSe (Rock salt) are totally deferent. Actually, SrF2 and SrSe layers are

40

Shaibal Mukherjee et al.

elastic-strain layers, which help in releasing strain in epilayers by reducing lattice and thermal mismatch between the CaF2/Si heterostructure and the PbSe epilayer [83, 84] Finally, a monocrystalline PbSe layer is grown on top of SrSe surface (as seen from RHEED patterns in Fig. 3.13) because their crystal structures are same and their lattice mismatch is only 2% at 300 K. After the MBE growth, the samples are taken out and the growth pit density is measured under the Nomarski microscope. The improvement by the in-situ surface treatment on the reduction of growth pits density is almost three-fold. The growth pits density decreases from 1735 cm-2 by traditional MBE-growth to 621 cm-2 as shown in Fig. 3.14. This, on the other hand, signifies that the theoretical defect limited operability of opto-electronic devices such as infrared photo-detectors fabricated on such improved IV-VI materials increases from 99.60% to 99.85%, where the pixel operability is predicted for 30 µm × 30 µm pixels with 30 µm pixel spacing. The etch pits density, as characterized by SEM in Fig. 3.15, reduces from a level of 2×108 cm-2 to 5×107 cm-2 due to the surface treatment. Moreover, the in-situ surface treatment helps in increasing the mobility of majority carrier from 9,875 cm2/V.s (carrier concentration: 8.0 × 1016 cm-3) to 29,912 cm2/V.s (carrier concentration: 7.2 × 1016 cm-3) at 77 K and from 647 cm2/V.s (carrier concentration: 2.0 × 1017 cm-3) to 1,073 cm2/V.s (carrier concentration: 1.7 × 1017 cm-3) at 300 K. It is well-known that the performance of detectors is limited by carrier mobility, which is dependent on mobility governed by threading dislocation. It is, therefore, highly expected that the IV-VI device performance would significantly improve by using the aforementioned highquality MBE-grown epilayers.

(a)

(b)

500µm

500µm

Figure 3.14. Nomarski images of growth pits (a) in traditional growth: pit density is 1352 cm-2 (size <5µm) and 383 cm-2 (size >5µm) and (b) in optimized growth: pit density is 454 cm-2 (size <5µm) and 167 cm-2 (size >5µm).

Lead salt thin film semiconductors for microelectronic applications

41

(a)

(b)

Figure 3.15. SEM images of etch pits (a) in traditional growth (PbSe on CaF2/Si) and (b) in optimized growth (after in-situ surface treatment), both in etchant for 5 minuets at 75 0C.

3.2.6. Conclusions In summary, a novel surface treatment method on CaF2/Si is described, by applying which the quality of IV-VI epilayer improves considerably. The reduction of growth pits and etch pit densities in the epilayer by the optimized surface treatment method are almost three-fold and four-fold respectively. Moreover, improvements in carrier mobility is more than threefold at 77 K and two-fold at room temperature respectively. This achievement is extremely important for the IV-VI midinfrared devices especially for detector

42

Shaibal Mukherjee et al.

applications. As is detailed in this chapter, the preliminary results are very encouraging and could lead to a significant progress in the performance of future devices based on IV-VI materials.

CHAPTER 4: RECENT DEVELOPMENTS OF IV-VI LEAD SALT SEMICONDUCTOR QUANTUM WELL STRUCTURES 4.1. Background The strain relaxation mechanism in IV-VI materials takes place by dislocation glide [85]. This process is pictorially depicted in Fig. 4.1. The Burgers vectors are of type a/2 <110>, as indicated in the figure. Unlike other zinc-blend type semiconductors, the primary gliding planes of lead salts are the {100} planes. This {100} planes remained inclined with respect to the surface for (111) oriented substrates [86]. A mechanical strain is generated either as a result of molecular beam epitaxy (MBE) growth or due to thermal, and this strain is relaxed only by dislocation glide on the {100} type planes.

This relaxation mechanism takes place because the Schmid factors are higher for the (111) growth orientation. For the (100) growth orientation, the Schmid factors for dislocation glide in the primary {100} planes are zero, leading to an absence of this glide mechanism along the main {100} planes.

Figure 4.1. Schematic of the arrangement of the {100}<110> glide system for PbSe (111) layers.(Source:H. Zogg, et al).

Lead salt thin film semiconductors for microelectronic applications

43

The misfit dislocations, which are arranged with a three-fold symmetry, lie along the <110> directions i.e., at the intersection of the {100} type glide planes with the (111) oriented interface. Once the dislocations have crossed the whole material thickness, no defect remains in the interior of the material, instead, a glide step remains on the surface parallel to a <110> direction [86].

If more than one dislocation glides within the same slip plane, the surface step is higher. [111] and [110] growth, therefore, have the advantage of lower dislocation density.

4.2. Emission power from QW structures Several non-sensing applications such as infrared (IR) countermeasures require output powers higher than 1 W at thermo-electrically (TE) cooled temperatures. The natural cleavage plane for BaF2, which is popularly used to grow IV-VI epitaxial layers, is (111). MBE growth orientations along {100} as well as {110} necessitate a complex, time-consuming chemo-mechanical polishing step.

Figure 4.2. Peak output power vs. peak pumping power at temperatures 250 K and 300 K.

44

Shaibal Mukherjee et al.

The results from [111]-oriented growths are very encouraging and could have significant implications for IR applications that need high output power.

Our group has observed a peak output power up to 4.9 W at room temperature (300 K) and 61.5 W at 180 K from photoluminescence (PL) of PbSe/PbSrSe multiple quantum well (MQW) structures grown by MBE. Because single-mode output powers of conventional IV-VI [100] edgeemitting lasers are typically â&#x2030;¤ 1 mW, even at 77 K, questions as to whether IV-VI lead salt materials could generate higher power were raised. Fig. 4.2 shows the calibrated output powers of PL emissions at room temperature and at 250 K with different pumping intensities with a spot-size of 1.2 mm in diameter on the sample [87]. The maximum peak output power observed is 4.9 W at room temperature and 17.2 W at 250 K with a peak pumping power of 122 kW and 73.5 kW, respectively. Fig. 4.3 demonstrates the emission at different heat-sink temperatures from 180 K to 310 K with a constant pumping intensity of 75 kW with the same pumping source and spot size [87]. The inset in the Fig. 4.3 demonstrates the shift in QW spontaneous emission peaks with temperature (dÎť/dT ~ 8 nm/K), nearly linear and shifting towards higher wavelengths (red shift) as heat-sink temperature decreases. The maximum PL power observed was 61.5 W at 180 K.

Figure 4.3. Peak output power at different temperatures. Inset shows the peak emission wavelength from the QW as a function of temperature.

Lead salt thin film semiconductors for microelectronic applications

45

4.3. MBE growth on [110] oriented BaF2 substrate MBE growth along {110}-orientated BaF2 substrates, which remained unexplored for a long time, was another promising phase for IV-VI semiconductor technology. The huge challenge for achieving this goal was to develop an optimized polishing recipe which would lead to form a scratchless substrate for epitaxial growth. A chemo-mechanical polishing recipe for [110] BaF2 substrate for MBE growth was developed. This new development is extremely exciting and is undoubtedly a breakthrough for mid-IR lead salt semiconductor lasers. The growth of PbSe/PbSrSe QW structures on polished [110] BaF2 substrates has successfully been accomplished. In situ reflection high energy electron diffraction (RHEED) during MBE growth demonstrated a high-quality two dimensional (2D) growth mode of the lead salt material on the BaF2 substrates.

Fig. 4.4 shows room temperature PL emission [88] from a seven pair PbSe (20 nm)/PbSrSe (30 nm) QW structure grown on [110] BaF2. The results are compared with the PL emission of the same structure on polished [111] cleaved substrate grown in the same MBE run. As can be seen that the PL intensity of [110] sample is about 1.5 times higher than that of [111] sample. This result indicates high material quality and is very encouraging for making infrared opto-electronic devices.

Figure 4.4. Comparison of room temperature photoluminescence from samples on [110] and [111] BaF2 substrates.

46

Shaibal Mukherjee et al.

Figure 4.5. High resolution x-ray diffraction spectrum of the (220) reflection from a PbSe thin film on a (110) BaF2 substrate.

The (110) BaF2 substrate and 3.9 Âľm (110) PbSe thin film were investigated with HRXRD of the symmetrical (220) reflections [89]. The measured full width at half maximum (FWHM) of (110) BaF2 substrate was 23 arcsec, which is similar to the theoretical value of the FWHM. The (220) rocking curve of the PbSe thin film is shown in Fig. 4.5 and has a FWHM of 60 arcsec. This indicates that a high crystalline quality PbSe epitaxial layer was obtained. Typical linewidths of (222) rocking curves for PbSe thin films grown on (111) BaF2 substrates are between 90 and 150 arcsec [89]. The slight broadening of FWHM for PbSe single layer compared with the (110) BaF2 substrates might be due to crystal size effects, non-uniform strain, and dislocations. The dislocation density of the PbSe thin film has been estimated from the rocking curve measurements [87] to be 1Ă&#x2014;107 cm-2. The higher quality of (110) PbSe thin films should improve the performance of lead salt optoelectronic devices.

4.4. [110] Oriented electrically pumped IV-VI edge-emitting laser While cost effective, the fabrication of MIR lasers on BaF2 substrates is tremendously challenging because these materials do not cleave along the same planes. BaF2 has a natural cleavage plane of (111) while Pb-salt epi-layers cleave along (100). The resultant fabrication has proven quite

Lead salt thin film semiconductors for microelectronic applications

47

challenging, therefore, before reaching out to achieve our ultimate goal of demonstrating lead-salt lasers on inexpensive BaF2 substrates, we have first elected to fabricate edge-emitters successfully on [110]-oriented lead-salt substrates, as the orientation is supposed to be the vital component for the lifting of degeneracy.

A thoroughly polished, smooth and uniform substrate is essential for the MBE growth of laser structures. As shown in Fig. 4.6 (a), the unpolished PbSnSe wafer had numerous deep grooves and saw marks on the surface.

After applying the optimized chemo-mechanical polishing, [110]-oriented lead salt substrate looked like Fig. 4.6 (b) under 100X Nomarski lens. As illustrated, the optimized polishing recipe resulted in a defect free, smooth, and uniform substrate. The polished substrate was deemed epi-ready after scrupulous cleaning with chemical reagents such as acetone, propanol, and methanol.

Figure 4.6. Nomarski images of a [110]-oriented PbSnSe substrate (a) before polishing, (b) after chemo-mechanical polishing, and the RHEED pattern from (c) the PbSnSe polished substrate and (d) after MBE growth.

48

Shaibal Mukherjee et al.

The two-dimensional uniformity of the substrate surface was confirmed by the streakiness of the reflected high energy electron diffraction (RHEED) pattern (as shown in Fig. 4.6 (c)). As depicted in Fig. 4.7, the resultant laser structure consisted of a 150 nm thick bottom p-PbSe cladding layer, followed by a 2.05 µm thick graded p-PbSrSe confinement layer, an active layer (thickness ~ 450 nm) consisting of 9 pairs of PbSe/Pb0.97Sr0.03Se multiple quantum wells (MQW), a 150 nm thick n-PbSe top confinement layer, terminated by a 150 nm thick n-PbSe top cladding layer. During the MBE growth, the PbSe growth rate of 1.30 µm/h. with a substrate temperature of 375 ºC was maintained. The streaky RHEED pattern observed from the laser structure during the MBE process shown in Figure 1.3.6 (d) confirmed the two-dimensional nature of the semiconductor multilayer growth on the PbSnSe substrate. Other relevant information about the MBE growth is detailed in [90].

The 20 µm stripe laser patterns were then realized by lithographically patterning as well as cleaving along the natural cleavage (100) plane along the MBE-grown structure. The photoresist separating the adjacent laser stripes was then completely removed and individual laser pieces with lengths of approximately 300 µm were cleaved and mounted on the metallic housing for optical characterization.

As shown in Fig. 4.8, the individual laser piece was mounted in an episide down fashion, and indium was electro-deposited on the laser head to enhance the adhesion of the laser piece with metal housing. The bottom contact of the laser piece with housing was accomplished by indium-plated copper wire. Pulsed laser pumping was performed at a pulse repetition frequency (PRF) of 100 kHz, and an IFS 66/S Fourier Transform Infra-Red

Figure 4.7. Schematic structure of an electrically pumped MQW IV-VI laser.

Lead salt thin film semiconductors for microelectronic applications

49

Figure 4.8. Picture of an edge emitting laser mounted on metallic housing.

spectrometer (FTIR), equipped with a liquid nitrogen cooled MercuryCadmium-Telluride (MCT) photo-detector in conjunction with a preamplifier, was used to obtain time-resolved lasing spectra in step scan mode.

The threshold current density was found to be 1.98 kA/cm2, and it increased with increasing heat sink temperature. The device was excited with 500 ns width current pulses and the laser emission obtained from that above threshold was multimodal. At 77 K, the peak lasing emission wavelength was 5.8 µm at 350 mA injection current. A standard blackbody calibration was carried out to calibrate the emission power from the lasing device, and the collected data were plotted vs. injection current as shown in Fig. 4.9.

The inset provides the I-V characteristics of the device. The series resistance was measured to be 0.56 ohm. The lasing spectrum at λ = 5.2 µm at the highest achievable temperature of 158K is shown in Fig. 4.10, from which the emission linewidth was determined to be 0.45 cm-1 when the device was excited by 500 mA current pulses. The larger value of linewidth can be attributed to the limitation of achieving a maximum FTIR optical resolution.

The peak laser emission shifted as the injection current increased. From the 350 to 800 mA current range, the tunability of peak emission was almost 0.14 cm-1/mA, while above 800 mA, the tunability was 0.32 cm-1/mA. The pulse-width was increased from 500 ns to 4.0 µs at a constant PRF of 100 kHz to observe the effect of it on laser emission. Emission power decreased with increasing pulse-width, and no emission was obtained above 4.0 µs i.e.

40% duty cycle of current pumping.

50

Shaibal Mukherjee et al.

Figure 4.9. Laser light output per facet vs. injection current (L-I) at several heat-sink temperatures. The inset shows I-V characteristics of the laser diode at 77 K.

Figure 4.10. Laser emission spectrum at 158 K of the laser structure of Fig. 4.7.

Lead salt thin film semiconductors for microelectronic applications

51

4.4.1. Consistency in infrared laser emission In order to examine the consistency and reliability of lasing emission over time, emissions were recorded from the device at different time intervals, keeping other parameters like heat-sink temperature and current injection constant. As shown in Fig. 4.12, the initial emission (black curve) was multi-modal. Even after the measurement was completed, the device was left on continuous current excitation so that the device had sufficient time for joule heating effects to take place. After 2 and 4 hrs time intervals, data collection was re-started, and at those times, multi-modal lasing emission (blue curve) with a lesser number of peaks took place. Moreover, the maximum peak value increased from the initially measured emission peak.

When the time interval was increased to 14 and 16 hrs, no significant change (red curve) was observed in either the number of modes or the peak values.

The total lasing power per facet was computed to be constant at ~ 2.43 mW, despite the change in the number of emission modes. As a plausible explanation, internal heat generation must have occurred in the sample which otherwise helped in shifting the modal emission from the IV-VI semiconductor structure. When the heat sink temperature was increased, the higher order lasing peaks underwent more loss compared to the lower order peaks. That should account for the smaller number of peaks obtained

Figure 4.11. Laser light output per facet vs. injection current (L-I) at T = 77 K and 100 K for a pump pulse-width of 4 Âľs.

52

Shaibal Mukherjee et al.

Figure 4.12. Stimulated emission from the lasing device at different time intervals for a current injection of 500 mA at a 5% duty cycle and T = 100 K.

from the lasing sample. Because the total emission power calculation (based on incorporating the integral area of all lasing modes) was found to be same, it can be inferred from the experiment that the emitted lasing power from IVVI laser was indeed consistent with time.

4.5. Mid-IR vertical cavity surface emitting laser (VCSEL) MBE growth of high quality IV-VI materials both on BaF2 as well as Sisubstrates has been well developed and optimized. High quality Pb1-xSrxSe with Sr composition from zero to one can be routinely grown in our lab with MBE [91, 92] and their refractive indices, energy bandgap and lattice constant have been intensively studied [90]. Epitaxial thickness of MBEgrown samples can be routinely controlled within 3% error.

As an example, Fig. 4.13 illustrates [93] the designed and measured reflectivity spectra of BaF2/Pb1-xSrxSe 位/4 distributed Bragg reflector (DBR).

The high quality BaF2/PbSrSe broadband high-reflectivity DBR is one of the key elements that contribute to the success of our infrared VCSEL development.

The first VCSEL with BaF2/PbSrSe DBR was demonstrated in 1999. Fig. 4.14 demonstrates [95] the achievement of 289 K operation of the VCSEL, with a PbSe 位-cavity, by optical excitation with Ho:YAG laser. Shortly afterwards, another VCSEL with PbSe/PbSrSe 位/2-cavity QW active region

Lead salt thin film semiconductors for microelectronic applications

53

Figure 4.13. Three-pair BaF2/Pb0.97 Sr 0.03Se λ/4 DBR with reflectivity up to 99.4%.

Figure 4.14. The first IV-VI VCSEL with PbSe λ−cavity.

operated at a threshold of 10.5 kW/cm2 and produced a peak output power of 300 mW. The group at Naval Research Lab that expertise in GaSbbased type-II VCSELs measured these two VCSELs and compared the Pbsalt VCSEL with the regular GaSb VCSEL as shown in Fig. 4.15. The threshold of QW IV-VI VCSEL is the lowest in all regular optically pumped VCSELs.

54

Shaibal Mukherjee et al.

Figure 4.15. The IV-VI QW VCSEL with λ/2−cavity.

4.5.1. Above-room-temperature pulsed operation Since high power diode lasers are available around 1 µm, we shifted to Nd:YAG laser (1.064 µm) for pumping. Fig. 4.16 and 4.17 show optically pumped lead salt vertical-cavity surface-emitting lasers (VCSELs) with a 9period PbSe/PbSrSe quantum well active region operating above-roomtemperature in pulsed mode [94]. A power output of 40 mW is obtained at room temperature and does not show saturation. The room-temperature threshold pump density is 200 kW/cm2.

Figure 4.16. VCSEL peak output power as a function of pump power density.

Lead salt thin film semiconductors for microelectronic applications

55

Figure 4.17. VCSEL emission at different temperatures.

4.5.2. Optically pumped CW operation In order to establish continuous wave (CW) pumping, a VCSEL structure optimized at lower temperature is grown in MBE. Without further processing, epi-side-up on-wafer testing with CW Nd:YAG was performed.

The main purpose is to investigate the heating effect for future diode pumped VCSEL. The sample has an 11 period PbSe/PbSrSe QW active region.

As shown in Fig. 4.18, the CW emissions are observed up to 230 K with emission around 4.03 Âľm [95]. The lowest threshold pump density is 2.6 kW/cm2 at 190 K. The low CW threshold shows the feasibility for a diodepumped mid-IR VCSEL system.

Figure 4.18. VCSEL output power as a function of incident CW pump power density with a spot size of 200 Âľm in diameter.

56

Shaibal Mukherjee et al.

4.6. Room temperature CW photoluminescence Strong CW photoluminescence from mid-infrared lead-salt PbSrSe/PbSe multiple quantum-wells grown on (111) BaF2 substrates are observed at temperatures from 100 K to 300 K [96]. The maximum output power was 2.9 and 0.6 mW at room temperature with epi-side down and epi-side up mounting, respectively. At 225 K, the temperature difference between the active region and the heat sink was about 32 ºC for epi-side down mounting and 75 ºC for epi-side up mounting (as demonstrated in Fig. 4.19). The peak energies shifted toward high energy with increasing temperature.

Figure 4.19. Temperature-dependent CW PL spectra under a constant excitation power of 2 W with episode down (upper) and epi-side up (lower) mounting, respectively. The exciting source is 1.064 µm CW YAG laser with TEM00 mode.

Lead salt thin film semiconductors for microelectronic applications

4.7. Fabrication microstructures

of

free-standing

multiple

57

quantum

well

4.7.1. Background In this section, the recent progress in fabricating novel micro-sized IV-VI QW structures which have vast applications in the field of microelectromechanical systems (MEMS) that demands free standing micro-sized objects in various shapes and sizes, has been described.

All these applications necessitate the requirement of precisely controlled thickness of epi-layers with smooth and uniform surfaces. This can be achieved by already well-established thin film deposition schemes such as molecular beam epitaxy (MBE) and metal-organic chemical vapor deposition (MOCVD). A controlled release of the thin film, grown by these deposition processes, from the substrate creates miniaturized objects for various device applications. This yielding of thin film from the substrate depends on the mismatch between the lattice parameters as well as temperature coefficients of expansion of the constituent epitaxial layers in the multilayer semiconductor structures and thereby an inherent biaxial strain [97] is generated. This strain has a significant effect on the electronic band structure [98] of the superlattice semiconductor. The geometry of these novel objects can be engineered by suitably defining the discrepancy among internal stresses generated in multilayered thin films with the help of some fundamental processing parameters. The internal stress itself depends on the thickness of individual layer and etching environment which the sample has undergone. Micro and nano structured devices based on III-V semiconductors are already quite well realized in both industry as well as research fields. The fundamental theory for fabrication of such miniaturized structures is based on very popular â&#x20AC;&#x153;generalâ&#x20AC;? method [99]. Recently, by adapting this technological aspect, we could fabricate micro-sized rods and tubes on IV-VI semiconductor materials [100, 101]. The large surface area to volume ratio assists in heat dissipation, which is of utmost importance for IV-VI semiconductor because of their low thermal conductivity. The large refractive index of these materials can be useful in obtaining more optical confinement when used as waveguide. Recently, it was shown that the microstructure dimensions were a strong function of fabrication environment such as etching time, temperature etc [102].

4.7.2. Pulsed mode PL emission from MQW micropillars Another important phase of those microfeatures included is the wavelength-sized micropillars [103-106]. Compared to other geometries of

58

Shaibal Mukherjee et al.

such micro-sized devices, micropillars have the advantages of larger photon collection efficiencies and their wavelength-size microcavities. The free standing micropillar arrays based on the lead salt materials were manufactured with the purpose of better heat dissipation during their high temperature operations as efficient mid-infrared light emitters. The micropillar sample was grown by molecular beam epitaxy (MBE) on BaF2(111) substrate [107]. The as-grown sample consists of a 20 pair PbSe/Pb0.97Sr0.03Se multiple quantum well (MQW) structure. The epi-layer consisted of a multiple quantum well structure with a 22 nm thick PbSe quantum well and 25 nm thick Pb0.97Sr0.03Se barrier layer. After the photolithography and consequent processing steps, pillars were formed on the epi-layer. The spacing between individual pillars (with an etching height of ~ 940 nm as shown in Fig. 4.20) was 8 Âľm, where the diameter of each pillar was 5 Âľm.

The pulsed photoluminescence measurements on such pillar array were carried out at different heat sink temperature and are plotted in Fig. 4.21. The figure [108] exhibits a consistent blue-shifting of photoluminescence peak with the increase in heat sink temperature. The temperature tunability at the emission wavelength obtained from the pillar sample was ~ 3.62 cm-1/K. The trend of the reduction of spontaneous emission coming from the pillar sample with the corresponding reduction in temperature was noted. The linewidth was 49.16 cm-1 at 77 K, and this reduced to 34.27 cm-1 at 300 K. The strong PL emission emanating from the sample together with the regular PL peak shift with temperature indicated the MBE growth quality of the epi-layer was retained in the micropillar sample.

Figure 4.20. SEM image of the pillar arrays illustrating undercut in wet etching leading to inclined pillar sidewalls.

Lead salt thin film semiconductors for microelectronic applications

59

Figure 4.21. Pulseominescence spectra of micropillar array (pillar diameter = 5 µm and inter-pillar distance = 8 µm) at several heat-sink temperatures (Intensity at 200 K, 250 K and at 300 K were multiplied by 2, 7 and 20 times respectively).

4.7.3. Fabrication of IV-VI microtubes/rods Lead salt epi-layers of various thicknesses are grown on a polished (111) BaF2 substrate. The entire multilayer structure generally consists of a thin PbSe buffer layer (~ 0.7 µm) as well as a thin BaF2 (~ 0.4 µm) sacrificial layer. The sacrificial layer of BaF2, lattice-matched with the PbSe layer, could be etched with de-ionized (DI) water, and as a result, the top singlecrystalline PbSe layer will roll-up taking the shape of microtubes/rods as illustrated in [109]. At low temperature, BaF2 substrate etched away more from the edges of the sample and thereby the PbSe buffer layer started rolling into micro-cylinders but due to the slower etching rate of BaF2 sacrificial layer top PbSe layer only grew cracks on it. But upon increasing the DI water temperature, the etching rate of the sacrificial layer enhances resulting in the folding back of top PbSe layer into rods. Wet etching of BaF2 layer guiding to the microtube/rod building is pictorially depicted in Fig. 4.22. The change in such microcylinder dimension can be achieved by suitable alteration of initial epitaxial layer thickness and etching time.

4.7.4. Optical characterization of microtubes/rods In order to carry out the optical characterization, a single free-standing microrod of 342.96 µm long with a diameter of 17.28 µm is glued by silver paste on copper ribbon (Fig. 4.23 (a)) in order to have means for heat dissipation during the measurement at room temperature.

60

Shaibal Mukherjee et al.

Figure 4.22. Schematic of microtube/rod formation.

Figure 4.23. (a) SEM micrograph of the microrod mounted on copper ribbon with silver paste during its optical characterization, (b) Photoluminescence spectra from mounted MQW microrod and its reference sample.

Lead salt thin film semiconductors for microelectronic applications

61

Pulsed photoluminescence (PL) excitation is then carried out by a 1.064 µm Nd:YAG laser (τpulse = 30 ns, pulse repetition rate = 10 Hz) which generates TEM00 mode incident normally on the sample surface.

The lasing power density is 44.31 kW/cm2 and the spot-size of the laser was focused to 5.0 mm, with the help of a 2˝-diameter CaF2 lens, which is well enough to cover the entire area of the sample. The PL emission is collected at a resolution of 16 cm-1 onto InSb detector, cooled by liquid nitrogen through a series of optical mirrors, perpendicular to the rod axis and parallel to the direction of excitation. Measurements are accomplished on various places on both the microrod as well as its reference sample (before processing of rods) in order to judge the repeatability of the results.

There has been 4.01 times reduction in the PL intensity from the rod sample compared to its reference as seen from Fig. 4.23(b). The excited surface area of the reference sample (19.63 × 106 µm2) is calculated to be 2563 times larger than that of the microrod sample (7659.86 µm2).

In case of cylindrical morphologies compared to its planar counterpart, there is always an enhancement factor of ten [110] in light extraction efficiency due to the internal reflection occurring inside the structure itself from various facets formed on their own. Thus by taking all these things into considerations and also assuming that the total emitted light from the sample is collected at the detector, there is a confirmed enhancement by 64 factor [111] in pulsed PL power density from microrod. Although there have not been adequate theories to account for this huge intensity gain factor, this is undoubtedly promising for future high power opto-electronic devices.

One of the major applications of mid-infrared opto-electronic devices is in the field of chemical and industrial gas sensing [112-114]. A large number of fundamental molecular vibrations as well as most of the first overtones take place in the mid-infrared region (2.5-25 µm) of electromagnetic spectrum. The nature of the spectral bands in this frequency region is likely to be sharp, distinct and have very high values of absorption coefficients; thereby leaving specific “footprint” for individual small gas molecule in the mixture. Therefore this necessitates manufacturing novel devices that show narrow and sharp PL peak at room temperature. This can be achieved by utilizing the well-established fiber-optics technology which not only restrict the light by the virtue of its waveguide nature but also improves signal power by boosting the signal to noise ratio. This is illustrated by comparing the pulsed PL spectra from a single microtube both before and after inserting in a hollow quartz fiber.

62

Shaibal Mukherjee et al.

Figure 4.24. SEM images of MQW microtube (a) cross-section, (b) surface (some portions damaged due to high-power laser excitation), (c) crosssectional view of microtube inserted inside silica fiber.

Figure 4.25. Comparative analysis of photoluminescence spectra at room temperature from MQW microtube inserted inside silica glass fiber as well as outside fiber.

SEM images of the morphological views of microtube (diameter = 480 Âľm) both outside and inside of hollow quartz fiber (diameter = 1051.11 Âľm) are shown in Fig. 4.24. The pulsed lasing power used for optical characterization is 397.33 kW/cm2. The emitted light from the microtube samples is collected parallel to the rod axis and perpendicular to the direction of excitation. Fig. 4.25 describes an increase of the light emission by a factor

Lead salt thin film semiconductors for microelectronic applications

63

of 1.19 together with a narrower linewidth of PL peak due to the optical confinement nature of glass fiber [111]. These all signify that the microstructures are very robust in nature and they can be easily handled while implementing into any devices. The increment in light emission power coupled with the reduction in optical linewidth is obviously encouraging for mid-infrared industrial and research applications.

4.8. Conclusions The modern epitaxial technology that allows high quality alternating material growth and advanced laser concepts is resurrecting IV-VI mid-IR lasers. The [110] and [111] orientated QW laser structure combined with much improved substrate thermal conductivity as well as the low Auger recombination promise the success of the lead-salt mid-IR lasers. The noteworthy success in mid-IR opto-electronic research based on lead salt QW structures holds the promise of a compact, highly efficient laser/detector system in the not too distant future. These light sources will eventually lead to many applications such as a high efficiency, field-portable, mid-IR laser spectroscopy systems.

CHAPTER 5: PHOTONIC BANDGAP DEFECT STRUCTURE: CAVITY WITHOUT CLEAVING 5.1. Background For a few decades, lead chalcogenide diodes have been one of the most popular and commercially available semiconductor MIR lasers. However, due to the low operating temperatures and low external efficiencies, the performances of lead salt lasers could not reach the desired level. Even at low cryogenic heat-sink temperature, lasing output powers for single mode operation are less than few mWs. Moreover, an affinity towards multimodal operation and mode hopping is quite familiar. The threshold level of laser operation for IV-VI materials is enhanced due to the existence of four-fold Lvalley degeneracy near band extrema. The degeneracy does not get lifted off for [100] orientation which is the most common growth orientation for QW laser structures. This, in turn, limits exploiting the supreme advantages of lead chalcogenide materials for high temperature and long wavelength operation i.e., reduced threshold level originating from a low nonradiative recombination rate. Also [100] growth orientation leads to inferior epitaxial material quality because it does not allow dislocation gliding phenomenon.

It is well-known that the QW gain threshold is minimum for [111] lead salt crystal orientation compared to its other two counterparts. Therefore, this orientation suits for higher temperature laser operation of lead salt materials.

64

Shaibal Mukherjee et al.

Also experimentally we already have exhibited superior performances from lead salt vertical cavity surface emitting laser (VCSEL) grown on (111) BaF2 substrate [115]. Those VCSELs have demonstrated pulsed-mode operation near room temperature. Lasing output power as high as 300 mW and threshold density as low as 10.5 kW/cm2 are achieved [116]. Pulsed laser emission at elevated temperature from VCSELs on (111) substrate is obtained [117, 118]. The continuous wave (CW) lasing at 4.03 ¾m is observed up to 230 K. The minimum value of threshold excitation density of 2.6 kW/cm2 has been measured at 190 K, 65 °C lower than that of the pulsed operation.

However the design complexity of distributed Bragg reflector (DBR) during MBE fabrication and low power operation severely prevent the popularity of VCSELs while realizing long wavelength and high power lasing devices. The best growth substrate for realizing high temperature devices is silicon (Si). Si (thermal conductivity = 1.56 W/cm K) has almost 78 times higher thermal conductivity than lead salt materials (thermal conductivity = 0.02 W/cm K) at 300 K. Moreover, by using Si one can reduce manufacturing cost of device fabrication as well as avail of the established device fabrication technologies on this substrate. Although unlike lead salt materials, the natural cleavage plane for Si is (111). Therefore in order to fabricate lasing device, there is a necessity to adopt some valid technology which does not undergo a cleaving process while making a Fabry-Perot resonating cavity.

The best solution in this regard is achieved by implementing the idea of well-known photonic bandgap (PBG) methodology. The physics of PBG does not depend on crystal growth orientation. PBG scheme demonstrates a perfect optical analogy of semiconductor bandgap where the electric potential distribution is governed by a periodicity of regularly distributed lattice dielectric medium (mainly air) in another highly contrasting dielectric medium. This helps in the confinement of light modes by creating a specific frequency bandgap which in other way opens a new horizon of tailoring the light-matter interaction in the crystal lattice. Due to the existence of bandgap effect in photonic crystals, optical waves corresponding to the bandgap frequency are forbidden [119] to propagate whereas waves outside this range are allowed to propagate. This basically provides a provision for suitably engineering the crystal design and localizing light modes within a semiconductor slab. The size of the bandgap is determined by two factors: the dielectric contrast of the constituent materials that build up the photonic crystal and the higher dielectric material filling fraction [120]. Several research studies [121-125] regarding photonic crystal fabrication and characterizations in the semiconductor structure are cited in the literature.

Lead salt thin film semiconductors for microelectronic applications

65

5.2. Introduction A periodic arrangement of atoms and molecules is known as crystal. If a tiny basic building block of atoms or molecules is repeated in space, a crystal lattice structure would be formed. Quite similar to the crystal structure, if the dielectric constant of a material system is periodically altered in space t is known as photonic crystal (PC). It is well known that all semiconductor materials have a specific energy bandgap in between conduction and valence band. Electrons and holes are “forbidden” to occupy any energy level within the bandgap. In a similar way, photonic crystal possesses a “forbidden” frequency band in which no propagation of electromagnetic waves is possible. Based on the number of directions the dielectric periodicity is exhibited, one-, two-, or three- dimensional photonic crystal or photonic bandgap structure is generated. Light scattering occurring inside a PBG structure can produce many photonic phenomena analogous to the electronic behavior inside crystals.

PBG crystal structure is analogous to normal crystal lattice in terms of atomic periodicity, except the fact that the periodicity is of the order of a micrometer rather than a fraction of a nanometer. The implementation of PBG effect to achieve a semiconductor laser by controlling spontaneous emission was demonstrated by Yablonovitch in 1987 [126]. The fundamental idea behind the concept was to design a photonic structure in order to originate a frequency bandgap that coincides with the semiconductor spontaneous emission. Since that time, the concept of PBG phenomenon is well adopted and practiced all over the world. Various significant applications [127-129] of PBG phenomenon are noticed to obtain millimeter and microwave frequencies in order to achieve desired performances for antennas and waveguides.

5.3. Finite difference time domain (FDTD) method The FDTD method is a well established mathematical analysis tool which is extensively implemented for integrated and diffractive optical device simulations. Through rigorous and accurate numerical solution of electromagnetic Helmholtz’s equations, this methodology helps in modeling light propagation, diffraction and scattering phenomena, and polarization and reflection effects inside a crystal. It also assists in modeling material anisotropy and dispersion without the requirement of any presumed field behavior. FDTD method is a very powerful simulation tool that provides effective analysis of sub-micron devices with very finite structural details.

The miniaturization of device dimension helps in light confinement and

66

Shaibal Mukherjee et al.

consequently, the large dielectric constant difference of the semiconductor materials to be implemented in a typical device design.

5.4. Two dimensional FDTD equations Opti-FDTD software is implemented to perform the FDTD simulation for the lead salt PBG structure. The FDTD methodology is fundamentally based on numerical solutions to electromagnetic Helmholtz’s equations. As shown in Fig. 5.1, the photonic structure is arranged in the X-Z plane [130].

The light propagation inside the device is considered to be along Z-direction.

The Y-direction is considered to be infinite. This fundamental approximation removes all the ∂ ∂y partial derivatives from electromagnetic Maxwell’s equations and divides them into two independent sets of numerical equations.

These are known as transverse electric (TE) and transverse magnetic (TM) equations.

The spatial steps in the X- and Z-directions are designated as ∆x and ∆z respectively. Each mesh point in the crystal lattice structure is linked with a specific type of dielectric material (i.e., lead salt substrate or air) and contains information about its basic properties such as dielectric constant and dispersion parameters.

Figure 5.1. Numerical representation of the 2D computational domain.

5.4.1. TE waves For two-dimensional TE case in X-Z plane, electric and magnetic field components like E y , H x and H z would exist. The electromagnetic equations in lossless media are given as:

Lead salt thin film semiconductors for microelectronic applications

∂E y ∂t

=

67

1 ⎛ ∂H x ∂H z ⎞ − ⎜ ⎟ ε ⎝ ∂z ∂x ⎠ 1 ∂E y

∂H x = ∂t µ 0 ∂z ∂H z 1 ∂E y =− ∂t µ 0 ∂x

(5.1)

where ε = ε0εr is the dielectric permittivity of the crystal lattice medium and µ0 is the vacuum magnetic permeability. ε0 and εr are the vacuum permittivity and the relative permittivity of the crystal medium respectively. The refractive index of the medium is given by n = ε r .

Each component of the electromagnetic field inside the crystal lattice structure is represented by a two-dimensional (2D) array - E y (i, k ) , H x (i, k )

and H z (i, k ) - equivalent to the 2D mesh grid demonstrated in Fig. 5.1. The indices i and k designate step size in space along the X- and Z- directions, respectively. Fig. 4.2 illustrates the distribution of the TE field components in the mesh structure in account. The distribution model of the TE field pattern is explained as follows. The Ey field, as shown in Fig. 5.2, coincides in space with the mesh node location in Fig. 5.1. The solid lines in Fig. 5.2 symbolize the mesh in Fig. 5.1. The Ey field is designated to be the center of the FDTD cell in space. Each FDTD shell is indicated by dashed lines in Fig. 5.2. The magnetic field components, Hx and Hz, form the cell edges.

The positions of the Ey field is associated with the integer values of indices i and k. The locations of Hx field is associated with integer values of i and (k + 0.5) indices. Similarly, the Hz field is associated with integer (i + 0.5) and k indices. Implementing numerical discrimination of the field components and applying them in the first derivative of equation (4.1), we would have the following relation: E yn (i, k ) − E yn −1 (i, k ) ∆t

(

)

(

)

1 ⎡ n − 12 n− 1 Hx i, k + 1 − H x 2 i, k − 1 ⎤ 2 2 ⎥⎦ ε∆z ⎢⎣ 1 ⎡ n − 12 n− 1 Hz i + 1 ,k − Hz 2 i − 1 ,k ⎤ − 2 ⎥⎦ 2 ε∆x ⎢⎣ =

(

)

(

)

(5.2)

The total set of numerical equation (4.1) takes the following form in Cartesian co-ordinate system:

68

Shaibal Mukherjee et al.

Figure 5.2. Position of the TE fields in the computational domain.

E yn (i, k ) = E yn −1 (i, k ) +

(

(

)

(

)

(

)

∆t ⎡ n − 12 n− 1 Hx i, k + 1 − H x 2 i, k − 1 ⎤ 2 2 ⎥⎦ ε∆z ⎢⎣

)

∆t ⎡ n − 12 n− 1 Hz i + 1 ,k − Hz 2 i − 1 ,k ⎤ 2 2 ⎦⎥ ε∆x ⎢⎣ ∆t n+ 1 n− 1 H x 2 i, k + 1 = H x 2 i, k + 1 + E yn (i, k + 1) − E yn (i, k ) 2 2 µ ∆z 0

−

(

n+ 1

Hz

2

)

(

)

[

(i + 1 2 , k ) = H (i + 1 2 , k ) − µ∆∆t x [E n− 1 2 x

]

n y

(i + 1, k ) − E yn (i, k )]

(5.3)

0

The scheme is actually based on implementing Yee’s [131] numerical approach to establish central difference approximations for both temporal and spatial derivatives. The superscript n denotes the time steps used in the numerical algorithm. The space sampling for the calculation is done on a submicron scale. To ensure numerical stability of the mathematical calculation in the absorbing medium, we have utilized Courant-Friedrichs-Levy (CFL) condition [132] as follows: 2 2 ∆t ≤ 1 ⎛⎜ c 1 / (∆x ) + 1 / (∆z ) ⎞⎟ ⎝ ⎠

(5.4)

Lead salt thin film semiconductors for microelectronic applications

69

5.4.2. TM waves Similarly, for the 2D TM case in X-Z plane, electric and magnetic field components like Ex, Hy and Ez would exist. The Helmholtz’s equations in lossless media are given as:

∂H y ∂t

1 ⎛ ∂E x ∂E z ⎞ − ⎟ ⎜ µ 0 ⎝ ∂z ∂x ⎠ 1 ∂H y

=−

∂E x =− ∂t ε ∂z ∂E z 1 ∂H y = ε ∂x ∂t

(5.5)

The distribution of the TM field components in the computational domain is shown in Fig. 5.3. In this case, the electric field components Ex and Ez form the FDTD cell edges. The magnetic field component Hy is associated with the center of the FDTD cell. The numerical equations in the TM algorithm can be demonstrated in a similar way to equation (5.3).

Figure 5.3. Position of the TM fields in the computational domain.

70

Shaibal Mukherjee et al.

5.5. Plane wave expansion (PWE) method The band structure of the lead salt PBG lattice for both TE and TM modes are plotted with the help of very popular PWE method. By this methodology, the Maxwell’s equation in a transparent, time invariant, sourceless, and non-magnetic medium can be described as:

G G ω2 G G 1 (5.6) ∇ × G ∇ × H (r ) = 2 H (r ) ε (r ) c G where ε (r ) is the dielectric permittivity which is a function of space. c is the G G speed of light in vacuum, and H (r ) is the optical magnetic field vector G G (r ) is H having a definite angular frequency ω. The time dependence of designated by a term e iωt . Equation (4.6) is known as the “master” equation [133]. This equation represents a Hermitian eigen value problem.

According to the Bloch theorem, the magnetic field can be represented in a medium with infinite periodicity as: G G G G G H (r ) = e ikr hk (r ) (5.7) G G GG G where h (r ) = h r + R for all possible arrangements of crystal lattice vectors G R . Thus, combining equations (5.6) and (5.7), we can obtain:

(

G ⎡1 G ∇ + i k × ⎢ ∇ + ik ⎣ ε rG

(

)

(

)

⎤ G ω2 G ⎥ × hk = 2 hk c ⎦

)

(5.8)

Equation (5.8) is the fundamental equation by solving which one can get the dispersion relation inside a particular crystal lattice structure. This equation is generally transformed into a finite domain by simply expanding the magnetic field in a finite basis of simple plane waves. Several methodologies [134, 135] have already been adopted in solving the final discretized problem. The final solution of the discretized problem would G result in a dispersion relation between modal frequencies and wave vector k , generally plotted in the form of a band diagram.

5.6. Finite difference method (FDM) There are a number of numerical approaches available for the electromagnetic analysis of structures having complicated geometries and designs. Numerical schemes such as finite element, finite difference, and

Lead salt thin film semiconductors for microelectronic applications

71

boundary element methods are all well known and well established.

However, most of these popular mathematical and analytical tools involve complex algebra and therefore in order to find a numerical solution they all prove to be time killer. Therefore, a fast and precise method for the modal analysis of a complicated structure such as the PBG crystal lattice has been implemented in this study.

The method fundamentally incorporates finite difference (FD) scheme. It is based on the perturbation correction technique [136, 137], devised with a field convergence algorithm. This method removes the limitation of FDTD scheme in obtaining individual modal field distribution. This limitation of FDTD arises because the source used in the method is an impulse function in the time domain covering an infinite spectrum, thus field solutions are superposition of all possible modes. But FDM, through a small correction in field and mode index evaluates individual mode field solutions in the crystal lattice structure.

The concerned lead salt PBG structure is constructed with the usual 2 2 discretization notation and index profile, n ( x, y ) = ni j where ni j corresponds to the index profile along the cross-section of the lattice. Since the field at the boundaries must vanish, assuming an initial value of the field, e next to one boundary, approximate field distribution and mode index are calculated.

After this primary calculation, taking these values as the initial conditions, the verification of the discretized Helmoltz’s equation at all grid points by the basic mode convergence equation is carried out. The field value is represented by ei,j and can be given as:

ei , j =

ei , j +1 + ei , j −1 + ei +1, j + ei −1, j

(

4 − (∆x ) k n 2

2 0

2 i, j

−n

2 eff

(5.9)

)

In equation (5.9) k0 is free space wave vector, ni,j is the refractive index at the computational point and neff is the effective refractive index of the semiconductor material. In this case, we assume ∆x = ∆y . Through a series of convergence scans, the resulting field distribution yields a more accurate value of mode index [138], neffm through the equation

2 = neffm

2ε i , j −1 2ε i , j +1 ⎡ ⎤ ei , j −1 + ei , j +1 ⎢ei +1, j + ei −1, j + ⎥ ε i , j −1 + ε i , j ε i , j +1 + ε i , j +∞ +∞ ⎢ ⎥ ∫−∞ ∫−∞ ⎢ ⎧⎪ ε i, j − ε i, j −1 ε i, j − ε i, j +1 ⎥ ei , j dx dy ⎫ ⎪ 2 ⎢ − ⎨4 + + − k 02 (∆y ) ε i , j ⎬ ei , j ⎥ ⎢ ⎥ ⎪⎩ ε i , j + ε i , j −1 ε i , j + ε i , j +1 ⎪⎭ ⎣ ⎦ k

2 0

+∞ +∞

(∆y )2 ∫−∞ ∫−∞

2 i, j

e dx dy

(5.10)

72

Shaibal Mukherjee et al.

The aforesaid process of recurring convergence scan and consequent assessment of the mode effective index is continued till a desired precision in the value of neffm is achieved. For very fast convergence, the algorithm is altered with a relaxation factor by which the change in the field before and after convergence scans is added in the next step of convergence. The method described above results in the scalar fields of the waveguide modes. In order to get the solution for the polarized modes, the approximate modes can be calculated through the TE and TM modal solution of the constituent slab waveguides. For mode field correction, the FD discretization can simply be adapted by incorporating the semivectorial Helmoltz’s equation. In case of the TE modes, the continuity condition leaves the scalar Helmoltz’s equation unchanged to represent the semi-vectorial form. For the TM modes, the corresponding equation [138] is given as:

⎛ ∂2 ⎞ ⎞ ∂ ⎛ ∂ ⎜⎜ 2 + k 02 n 2 − β 2 ⎟⎟ E y + ⎜⎜ E y log n 2 ⎟⎟ = 0 ∂y ⎝ ∂y ⎠ ⎝ ∂y ⎠

(5.11)

Considering a three-point centered difference approximation for the operator ∂ 2 ∂y 2 and implementing a uniform sampling grid, the field in case of semi-vectorial form, is converged to an x-polarized mode as given by the following convergence scan equation.

ei +1, j + ei −1, j + ei , j =

(

2ε i , j −1

ε i , j −1 + ε i , j

4 − k (∆y ) ε i , j 2 0

2

ei , j −1 +

ε i, j − n )+ ε i, j 2 eff

2ε i , j +1

ei , j +1

ε i , j +1 + ε i , j − ε i , j −1 ε i , j − ε i , j +1 + + ε i , j −1 ε i , j + ε i , j +1

(5.12)

From the field distribution, the neffm is then calculated using the integral as in equation (5.13).

2 = neffm

2ε i , j −1 2ε i , j +1 ⎡ ⎤ ei , j −1 + ei , j +1 ⎢ei +1, j + ei −1, j + ⎥ ε i , j −1 + ε i , j ε i , j +1 + ε i , j +∞ +∞ ⎢ ⎥ ∫−∞ ∫−∞ ⎢ ⎧⎪ ε i, j − ε i, j −1 ε i , j − ε i , j +1 ⎥ ei , j dx dy ⎫ ⎪ 2 ⎢ − ⎨4 + + − k 02 (∆y ) ε i , j ⎬ ei , j ⎥ ⎢ ⎥ ⎪⎭ ⎪⎩ ε i , j + ε i , j −1 ε i , j + ε i , j +1 ⎣ ⎦ k 02 (∆y )

2

+∞ +∞

∫ ∫

−∞ −∞

ei2, j dx dy

(5.13)

Lead salt thin film semiconductors for microelectronic applications

73

In case of y-polarized mode, equations (5.12) and (5.13) are exactly followed with the indices i, j interchanged.

5.7. Lead chalcogenide defect cavity PBG structure The photonic semiconductor structure consists of PbSe-Pb0.98Sr0.02Se multiple quantum well structure, grown on BaF2 substrate, which has arrays of dielectric air columns periodically distributed in hexagonal fashion on the surface (as shown in Fig. 5.4(a)). Fig. 5.4(b) demonstrates the symmetry points of the first Brillouin zone in the periodic hexagonal crystal. The structure has the following constituent material parameters: refractive indices are nPbSe = 5.0 (bulk PbSe), nPbSrSe = 4.6 (Pb0.98Sr0.02Se) and na = 1 (air). The active layer is considered to have a λ/2 thickness, where λ is the emission wavelength. In order to choose an optimized design along two-dimensional crystal structure, a scanning has been done to calculate modal band structure and the corresponding bandgaps in the crystal lattice for varying radius of air-hole.

The cross section of the designed hexagonal crystal lattice structure constructed during FDM simulation is illustrated in Fig. 5.5. The diagram illustrates the formation of a single defect inside the PC structure by omitting the central air-hole within a computational window of a 600 × 600 mesh points.

Figure 5.4. (a) Top view of triangular photonic crystal lattice structure, (b) Schematic of first Brillouin zone for the hexagonal lattice pattern.

74

Shaibal Mukherjee et al.

Figure 5.5. Top View of refractive index variation over the cross-section of hexagonal crystal lattice.

5.7.1. Mid infrared photonic bandgap formation The bandgap structure is illustrated in Fig. 5.6 where all the frequency eigenvalues from the dispersion curves are combined along the single vertical line for each specific scanned value of radius. The crystal is optimally designed to have a diameter of dielectric air-hole (d) of 0.63 µm with a periodic lattice constant (∧) of 0.96 µm. The air-fraction (d/∧) in the photonic crystal plays an important role in creating bandgap.

The band frequency tends to rise with the increase of air-fraction for a fixed lattice constant (∧) in the crystal as can be seen in Fig. 5.6(a).

In Fig. 5.7, both transverse electric (TE) as well as transverse magnetic (TM) band structures are plotted with the help of PWE method. However, it is mention-worthy that the resonating optical modes in such a periodically patterned crystal are not distinctively TE or TM but they can be thought as TE-like and TM-like [140]. The band diagrams in Fig. 5.7(b) demonstrate one narrow and another quite broad TM-like photonic bandgaps exist in the structure, but no TE-like bandgaps. The bandgap values are populated in Table-5.1. Moreover, the broad TM bandgap covers the mid-infrared spectral region where room temperature photoluminescence from lead salts takes place. Therefore we focus to engineer a defect cavity mode in this frequency bandgap region.

Lead salt thin film semiconductors for microelectronic applications

75

(a)

(b) Figure 5.6. (a) Reduced band map for TE and TM mode, (b) band map for TE, TM, and joint mode.

76

Shaibal Mukherjee et al.

(a)

(b)

Figure 5.7. Band structure of the photonic crystal for (a) TE and (b) TM polarized light.

Table 5.1. TM bandgaps in the hexagonal periodic photonic lattice structure on IV-VI lead salt semiconductor.

Serial Number TM1 TM2

Bandgap 1.916 µm – 2.038 µm 3.959 µm – 5.872 µm

Lead salt thin film semiconductors for microelectronic applications

77

The defect cavity is designed (as illustrated in Fig. 5.4(a)) by omitting the central air column, four vertical (two on top and two at bottom) lattice atoms along X-direction surrounding the central defect are made to have a diameter of 0.8 µm, two adjacent horizontal atoms along Zdirection surrounding the central defect are made to have a diameter of 0.46 µm.

5.7.2. Modal analysis by FDM scheme To appreciate the confinement of mode to the central region, mode field distributions are calculated using the FDM technique. Different values of emission wavelength and crystal lattice parameters have been selected to optimize the variation in mode field distribution on changing these parameters. The final goal is to achieve a single modal laser emission from the lead chalcogenide defect cavity PC structure. It is noticed that the field distribution appears to be largely Gaussian at single modal condition. However, when the emission wavelength becomes multimodal, the field assumes a distribution different from the usual gaussian shape. This can be seen in Fig. 5.8, which depicts the distribution of electric field for the fundamental and multi mode at emission wavelengths of 4.17 µm [141].

The effective modal refractive index at the resonating condition is formulated to be ~ 4.04. FDM perturbation technique determines the field distribution in the crystal at a fixed wavelength and therefore it is quite challenging while determining the resonating wavelength for a specific optical cavity. This necessitates the deployment of FDTD approach which produces a wideband response for the field distribution in the cavity by exciting the crystal with a gaussian pulse. Moreover, by applying FDTD approach we could verify the results of FDM approach.

5.7.3. Modal analysis by FDTD scheme The uniaxial perfectly matched layer (UPML) absorbing boundary condition [142-144], which is a very efficient method in dealing with scattering by particles in vacuum, is implemented as the boundary layers outside the FDTD computational window. In order to excite the photonic crystal structure, a Gaussian modulated continuous wave (GMCW) point source is considered which can be expressed as:

[

]

Einc = A. exp − 0.5.(t − t0 ) / T 2 sin (ωt ) 2

(5.14)

78

Shaibal Mukherjee et al.

(a)

(b) Figure 5.8. Resonating electric field distribution at λ = 4.17 µm (a) when multimodal, (b) when single modal.

Intensity (arb. unit)

1.0 0.8 0.6 0.4 0.2 0.0 3.0

3.5

4.0

4.5

5.0

5.5

6.0

Emission Wavelength (µm)

Figure 5.9. Spectral response from DFT calculations after FDTD analysis of the resonating defect cavity.

Lead salt thin film semiconductors for microelectronic applications

79

where A is the input wave amplitude, T is the half width, ω is the angular frequency, t0 is the time offset. The rigorous numerical analysis produces precise and efficient time domain response for the field distribution in the defect cavity.

To generate the spectral response corresponding to the time domain behavior, discrete Fourier transform (DFT) is applied. This calculates spectral response for a specific wavelength as shown by: T

N

0

n =0

F (ω ) = ∫ F (t ). exp(− jωt )dt = ∑ F (n ). exp(− jωn∆t ).∆t

(5.15)

where F(n) is the time domain response, N is the number of time steps. The spectral field obtained from the defect cavity (as in Fig. 5.4(a)) in the photonic crystal is seen to resonate at 4.17 µm and is plotted in Fig. 5.9. The result seems to be in exact correlation with the numerical results obtained from the FD perturbation correction analysis. The optimized quality factor for the resonating mode is calculated to be 5200.

5.8. Experimental steps for air hole formation In order to realize PC lasing cavity, one has to undergo through contact lithography and wet etching method. One of the most important parameters of PC air-hole is the inclination of its side wall with the base substrate.

During the theoretical simulation, it is always considered that the side wall is perfect in terms of the surface roughness and inclination. Even a minute alteration in these parameters affects the modal behavior of the defect cavity emission inside the crystal lattice structure. The surface roughness is not a problem for wet etching method. The only challenge is the vertical inclination. In order to verify our presumption, we have selected a photomask which would lead to an air-hole with a diameter of 50 µm as shown in Fig. 5.10. It is to be noted that the dimension of this mask has nothing to do with our original theoretical PC laser dimension. We considered a bigger airhole mask just to study the perfection of wet etching procedure.

The air-hole is etched on a PbSe single layer grown on (111) BaF2 substrate. From Fig. 5.11, it can be seen that the air-hole is not perfectly vertical to the base substrate. Rather it poses an acute angle with the base.

Therefore the wet etching method is proved to be unsuitable for PC laser structure fabrication. Thus dry etching or plasma etching seems to be the only feasible option remaining to fabricate PC lasing cavity on lead salt substrate.

Regarding this we are doing a joint collaborative study with Penn State University and the preliminary results of their dry etching procedure on our MBE-grown lead salt material is illustrated in Fig. 5.12.

80

Shaibal Mukherjee et al.

Figure 5.10. Realization of air-hole by photolithography and wet etching method.

Figure 5.11. Air-hole side wall inclination to base not exactly vertical after wet etching method.

Lead salt thin film semiconductors for microelectronic applications

81

Figure 5.12. Plasma etching of lead salt material (Courtesy: Penn State University).

5.9. Conclusions In conclusion, a theoretical investigation of spontaneous mid-infrared emission from IV-VI semiconductor defect cavity in the hexagonal photonic crystal is elaborated in this chapter. The design is aimed to solve out challenges of the formation of resonating cavity for lead salt materials fabricated on Si(111) or BaF2(111) substrates. The band structure calculations of the periodic crystal are performed using PWE method. Two approaches have been implemented to analyze and understand modal field distribution in the defect cavity. FD perturbation correction method and FDTD algorithms are very popular and well-established mathematical tools for optical waveguide analysis. It has been demonstrated that the FDTD results reasonably agree with that of FD perturbation method. A single TM-like mode working at 4.17 Âľm, having an optimized Q-factor of 5200, resonates in the designed defect cavity. The prospective practical applications of the single mid-infrared emission are mainly in industrial trace-gas sensing systems and emission monitoring.

CHAPTER 6: SUMMARY The research described in this book has focused on the IV-VI lead salt semiconductor growth and characterization for mid-infrared microelectronic

82

Shaibal Mukherjee et al.

device applications. Several novel approaches, that have revolutionized the field of thin film semiconductors, have been reported here. The ultimate goal is to develop a high-performance mid-infrared detector and lasers for various civil and military application. Processing techniques and procedures, as well as theoretical design and calculations have been successfully developed, which could lead to high-temperature CW laser and high-sensitivity detector operation.

In this book, an introduction of primary properties of IV-VI lead salt semiconductors and their applications are illustrated. This follows a thorough investigation of the growth defects in lead salt epilayers. Here, the growth pits in IV-VI semiconductors are studied meticulously.

Morphological and structural images are provided to verify the properties of growth pits in order to understand the origin of growth defects in Pb-salt epilayers. In the next chapter, a description of the in-situ surface treatment methodology, which is carried out during MBE-grown PbSe on CaF2/Si (111) heterostructure, is provided. Detailed experimental procedures are provided and supported by in-situ RHEED images of growth steps. The promising results are listed in forms of an improved electrical and optical performances of the epilayers. The modern epitaxial technology that allows high quality alternating material growth and advanced laser concepts is resurrecting IV-VI mid-IR lasers. The [110] and [111] orientated QW laser structure combined with much improved substrate thermal conductivity as well as the low Auger recombination promise the success of the lead-salt mid-IR lasers. The third chapter described some work in the domain of recent developments of IV-VI lead salt light emitting devices. An electrically excited QW laser on [110] oriented lead salt substrate is described for the first time in the literature. Fabrications of novel microstructures in the form of rod, tube and pillar, having enormous applications in MEMS and NEMS, are detailed. A theoretical investigation of spontaneous mid-infrared emission from IV-VI semiconductor defect cavity in the hexagonal photonic crystal is elaborated in this chapter. The design is aimed to solve out challenges of the formation of resonating cavity for lead salt materials fabricated on Si(111) or BaF2(111) substrates. The band structure calculations of the periodic crystal are performed using PWE method. FD perturbation correction method and FDTD algorithms have been implemented to analyze and understand modal field distribution in the defect cavity. These are very popular and well-established mathematical tools for optical waveguide analysis. FDTD results reasonably tally with the simulation results by FD perturbation method. By thorough and accurate calculation, a single TM-like mode is achieved at 4.17 Âľm, having an optimized Q-factor of 5200, resonates in the designed defect cavity.

Lead salt thin film semiconductors for microelectronic applications

83

References 1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Braun, F. 1874, Annalen der Physik und Chemie, 153, 556.

Zogg, H., Fach, A., Maissen, C., Masek, J., Blunier, S. 1995, Optical Eng., 34, 1964.

Hicks, L.D., Harman, T.C., Sun, X., Dresselhaus, M.S. 1996, Phys. Rev. B, 53, R10493.

Zogg, H., Blunier, S., Hoshino, T., Maissen, C., Masek, J., Tiwari, A.N. 1991, IEEE Transac. Electron Dev., 38, 1110.

Katzir, A., Rosman, R., Shani, Y., Bachem, K.H., Böttner, H., Preier, H.M. 1989, Handbook of solid state lasers, Peter K. Cheo (ed.), Marcel Dekker, New York, Basel.

Bauer, G., Kriechbaum, M., Shi, Z., Tacke, M. 1995, J. Nonlin. Opt. Phys.

Mater., 4, 283.

Tacke, M. 2000, Helm, M. (ed.), Optoelectronic properties of semiconductors and superlattices, Gordon & Breach Science Publishers SA, Amsterdam.

Bauer, G., Pascher, H. 1991, Semimagnetic semiconductors and diluted magnetic semiconductors, M. Averous, M. Balkanski (ed.)., Plenum Press, New York, 209.

Bauer, G., Pascher, H., Kriechbaum, M. 1987, Phys. Scrip. T, 19, 147.

Ishida, A., Fujiyasu, H. 1985, Jpn J. Appl. Phys., 24, L956.

Landolt-Börnstein. 1983, New Series, Group III, Vol 17: Semiconductors, O. Madelung, M. Schulz, H. Weiss (ed.), Springer, Berlin, Heidelberg, New York.

Nimtz, G., Schlicht, B. 1983, Springer Tracts Mod. Phys., 98, Springer, Berlin.

Tacke, M. 1995, Infrared Phys. Technol., 36, 447.

Ziep, O., Genzow, D., Mocker, M., Herrmann, K. H. 1980, Phys. Stat. Sol. B, 99, 129.

Mocker, M., Ziep, O. 1983, Phys. Stat. Sol. B, 115, 415.

Rosman, R., Katzir, A. 1982, IEEE J. Quant. Electron., 18, 814.

Klann, R., Buhleier, R., Elsaesser, T., Lambrecht, A., 1991, Appl. Phys. Lett., 59, 885.

Findlay, P. , Pidgeon, C.R., Murdin, B.N., van der Meer, A.F.G., Langerak, C.J.G.M., Ciesla, C.M., Oswald, J., Springholz, G., Bauer, G. 1998, Phys. Rev. B, 58, 12908.

Lindle, J.R., Meyer, J.R., Hoffman, C.A., Bartoli, F.J., Turner, G.W., Choi, H.K. 1995, Appl. Phys. Lett., 67, 3153.

Xu, J., Tacke, M. 1992, Infrared Phys., 33, 151.

Partin, D.L., Heremans J. 1994, Handbook on Semiconductors, T.S. Moss and S. Mahajan (ed.), North-Holland, Amsterdam, 369–450.

Springholz, G. 2003, Lead chalcogenides – Physics and Applications, D. Khoklov (ed.), Taylor and Francis Inc., 123.

Springholz, G., Zogg, H., Shi, Z. 1999, Thin Films: Heteroepitaxial Systems, W.K. Liu, M.B. Santos (ed.), World Scientific, Singapore, 621–688.

Prinz, A., Brunthaler, G., Ueta, Y., Springholz, G., Bauer, G., Grabecki, G., Dietl, T. 1999, Phys. Rev. B, 59, 12983.

84

Shaibal Mukherjee et al.

25. Springholz, G., Schwarzl, T., Heiss, W. 2006, Mid-infrared Semiconductor Optoelectronics, A. Krier (ed.), Springer Verlag.

26. Schwarzl, T., Heiss, W., Springholz, G. 1999, Appl. Phys. Lett., 75, 1246.

27. Kapon, E., Figueroa, L., Katzir, A., Rosman, R., Shani, Y., Bachem, K.H., Bottner, H., Preier, H.M., Sam, R.C., German, K.R., Jones, W.B., Jr. 1989, Handbook of Solid State Lasers; P.K. Cheo (ed.), Marcel Dekker Inc.: Rochester, NY, 1-611.

28. Butler, J.F., Calawa, A.R., Phelan, R.J., Jr., Harman, T.C., Strauss, A.J., Rediker, R.H. 1964, Appl. Phys. Lett. 5, 75.

29. Feit, Z., Kostyk, D., Woods, R.J., Mak, P. 1991, Appl. Phys. Lett., 58, 343.

30. Schiebel, U., Rohr, J. 1999, J. Infrared Phys. Technol., 40, 325.

31. Hicks, L.D., Harman, T.C., Sun, X., Dresselhaus, M.S. 1996, Phys. Rev. B, 53, R10493.

32. Springholz, G., Ihninger, G., Bauer, G., Olver, M.M., Pastalan, J.Z., Romaine, S., Goldberg, B.B. 1993, Appl. Phys. Lett., 63, 2908.

33. Yuan, S., Springholz, G., Bauer, G., Kriechbaum, M. 1994, Phys. Rev. B, 49, 5476.

34. Springholz, G., Pinczolits, M., Bauer, G. 1998, Science, 282, 734.

35. Rogalski, A., Adamiec, K., Rutkowski, J. 2000, Narrow Gap Semiconductor Photodiodes, Bellingham, WA: SPIE. 36. Faist, J., Capasso, F., Sivco, D.L., Sirtori, C., Hutchinson, A.L., Cho, A.Y. 1994, Science, 264, 553.

37. Muller, P., Zogg, H., Fach, A., John, J., Paglino, C., Tiwari, A.N., Krejci, M., Kostorz, G. 1997, Phys. Rev. Lett., 78, 3007.

38. Zogg, H., Alchalabi, K., Zimin, D., Kellermann, K. 2003, IEEE Transac.

Electron. Dev., 50, 209.

39. Grisar, R., Böttner, H., Tacke, M. 1992, Monitoring of Gaseous Pollutants by Tunable Diode Lasers, G. Restelli (ed.), Kluwer Academic Publishers, Dordrecht.

40. Schwarzl, T., Heiss, W., Springholz, G., Aigle, M., Pascher, H. 2000, Electron.

Lett., 36, 322.

41. Baumgartner, E.W., Schwarzl, T., Springholz, G., Heiss, W. 2006, Appl. Phys.

Lett., 89, 051110.

42. Bimberg, D., Grundmann, M., Ledentsov, N.N. 1998, Quantum Dot Heterostructures, Wiley, Chichester.

43. Stangl, J., Holý, V., Bauer, G. 2004, Rev. Mod. Phys., 76, 725.

44. Schchukin, V.A., Ledentsov, N.N., Bimberg, D. 2004, Epitaxy of Nanostructures, Springer Verlag, Berlin.

45. Tersoff J., LeGoues, F.K. 1994, Phys. Rev. Lett., 72, 3570.

46. Springholz G., Bauer, G. 2007, Phys. Stat. Sol. B, 8, 2752.

47. Raab, A., Lechner, R.T., Springholz, G. 2002, Appl. Phys. Lett., 80, 1273.

48. Lechner, R.T., Schülli, T., Holy, V., Springholz, G., Stangl, J., Raab, A., Metzger, T.H., Bauer, G. 2004, Appl. Phys. Lett., 84, 885.

49. Fach, A., John, J., Muller, P., Paglino, C., Zogg, H. 1997, J. Electron. Mater., 26, 873.

50. Müller, P., Zogg, H., Fach, A., John, J., Paglino, C., Tiwari, A.N., Krejci, M. 1997, Phys. Rev. Lett., 78, 3007.

Lead salt thin film semiconductors for microelectronic applications

85

51. Zogg, H., Blunier, S., Fach, A., Maissen, C., Müller, P., Teodoropol, S., Meyer, V., Kostorz, G., Dommann, A., Richmond, T. 1994, Phys. Rev. B, 50, 10801.

52. Zogg, H., Müller, P., Fach, A. 1995, Mater. Res. Soc. Symp. Proc., 379, 27.

53. Speck, J.S., Brewer, M.A., Beltz, G., Romanov, A.E., Pompe, W. 1996, J. Appl.

Phys., 80, 3808.

54. Kroemer, H., Liu, T.-Y., Petroff, P.M. 1989, J. Cryst. Growth, 95, 96.

55. Chandra, D., Aqariden, F., Frazier, J., Gutzler, S., Orent, T., Shih, W.D. 2000, J. Electron. Mater., 29, 887.

56. Chang, Y., Badano, G., Zhao, J., Grein, C.H., Sivananthan, S., Aoki, T., Smith, D.J. 2003, Appl. Phys. Lett., 83, 4785.

57. Sabinina, I.V., Gutakovsky, A.K., Sidorov, Y.G., Latyshev, A.V. 2005, J. Cryst.

Growth, 274, 339.

58. Chandra, D., Shih, H.D., Aqariden, F., Dat, R., Gutzler, S., Bevan, M.J., Orent, T. 1998, J. Electron. Mater., 27, 640.

59. Rogalski, A. 1988, Infrared Phys., 28, 139.

60. Zogg, H., Fach, A., Maissen, C., Masek, J., Blunier, S. 1994, Opt. Eng., 33, 1440.

61. Shi, Z., Lv, X., Zhao, F., Majumdar, A., Ray, D., Singh, R., Yan, X.J. 2004, Appl. Phys. Lett., 85, 2999.

62. Ma, X., Chang, H., Zhang, Q., Sudarshan, T. 2005, J. Cryst. Growth, 279, 425.

63. Chen, L., Wu, Y., Yu, M.F., Wang, S.L., Qiao, Y.M., He, L. 2001, J. Infrared Millim. Wave, 20, 406.

64. Houtepen, A.J., Koole, R., Vanmaekelbergh, D., Meeldijk, J., Hickey, S.G. 2006, J. Am. Chem. Soc., 128, 6792.

65. Wijewarnasuriya, P.S., Zandian, M., Young, D.B., Waldrop, J., Edwall, D.D., Mclevige, W.V., Lee, D., Arias, J., D’Souza, A.I. 1999, J. Electron. Mater., 28, 649.

66. Mendez-Garcia, V.H., Lopez-Lopez, M. 1999, J. of Crys. Growth, 210/202, 518.

67. Romano, L.T., Knall, J., Bringans, R.D., Biegelsen, D.K. 1994, Appl. Phys. Lett., 65, 869.

68. Bringans, R.D., Biegelsen, D.K., Swarts, L.E., Ponce, F.A. Tramontana, J.C. 1992, Phys. Rev. B, 45, 13400.

69. Harrison, W.A., Kraut, E.A., Waldrop, J.R., Grant, R.W. 1978, Phys. Rev. B, 18, 4402.

70. Park, R.M., Mar, H.A. 1986, Appl. Phys. Lett., 48, 529.

71. Bringans, R.D., Olmstead, M.A. 1989, Phys. Rev. B, 39, 12985.

72. Romano, L.T., Bringans, R.D., Knall, J., Biegelsen, D.K., García, A., Northrup, J.E. 1994, Phys. Rev. B, 50, 4416.

73. Park, R.M. 1992, J. Vac. Sci. Technol. A, 10, 701.

74. Elizondo, S.L., Zhao, F., Kar, J., Ma, J., Smart, J., Li, D., Mukherjee, S., Shi, Z. 2008, J. Electron. Mater., 37, 1411.

75. Zogg, H., Blunier, S., Hoshino, T., Maissen, C., Masek, J., Tiwari, A.N. 1991, IEEE Transac. Elect. Dev., 38, 1110.

76. Zogg, H. 1986, Appl. Phys. Lett., 49, 933.

77. Zogg, H., Fach, A., John, J., Masek, J., Müller, P., Paglino, C. 1995, Narrow Gap Semiconductors, Institute of Physics Publishing Ltd., London, 160.

86

Shaibal Mukherjee et al.

78. McCann, P.J., Fang, X.M., Liu, W.K., Strecker, B.N., Santos, M.B. 1997, J. of Crys. Growth, 175/176, 1057.

79. Müller, P., Fach, A., John, J., Tiwari, A.N., Zogg, H., Kostorz, G. 1996, J. of Appl. Phys,. 79, 1911.

80. Wu, H.Z., Fang, X.M., McAlister, D., Salas, R., Jr., McCann, P.J. 1999, J. of Vac. Sc. and Technol. B, 17, 1263.

81. Zogg, H. 1999, SPIE, 3629, 52.

82. Springholz, G., Zogg, H., Shi, Z. 1999, Thin Films: Heteroepitaxial Systems, W.K. Liu and M.B. Santos (ed.), World Scientific, Singapore, 621–688.

83. Gautier, C., Breton, G., Nouaoura, M., Cambon, M., Charar, S., Averous, M. 1998, Thin Sol. Film., 315, 118.

84. Tacke, M. 1995, Infrared Phys. Technol., 36, 447.

85. Zogg, H., Blunier, S., Fach, A., Maissen, C., Muller, P., Teodoropol, S., Meyer, V., Kostorz, G., Dommann, A., Richmond, T. 1994, Phys Rev B, 50, 10801.

86. Elliott, C.T., Gordon, N.T., Baker, I.M., Zogg, H., Ishida, A., Kimata, M., Whatmore, R.W., Watton, R., Wood, R.A., Micklethwaite, W.F.H., Johnson, A.J., Biefeld, R.M., Kurtz, S.R., Asahi, H., Capper, P., Reine, M.B., Rogalski, A., Meyer, J.R., Vufgaftman, I., Kane, M.,J. 2001, Handbook of Solid State Lasers, P. Capper, C.T. Elliott (ed.), Electronic Materials, Kluwer academic publisher: Boston, MA, 43-96.

87. Majumdar, A., Xu, H.Z., Khosravani, S., Zhao, F., Jayasinghe, L., Shi, Z. 2003, Appl. Phys. Lett., 82, 493.

88. Zhao, F., Lu, X., Keay, J.C., Ray, D., Singh, R., Majumdar, A., Yan, X., Shi, Z. 2005, J. Crys. Growth, 285, 54.

89. Ayers, J.E., 1994, J. Crys. Growth, 135, 71.

90. Majumdar, A., Xu, H.Z., Zhao, F., Keay, J.C., Jayasinghe, L., Khosravani, S., Lu, X., Kelkar, V., Shi, Z. 2004, J. Appl. Phys., 95, 939.

91. Xu, G., Fang, X.M., McCann, P.J., Shi, Z. 2000, J. Crys. Growth, 209, 763.

92. Wu, H., Zhao, F., Jayasinghe, L., Shi, Z. 2002, J. Vac. Sci. Technol B, 20, 1356.

93. Bewley, W.W., Felix, C.L., Vurgaftman, I., Meyer, J.R., Xu, G., Shi, Z. 2000, Electron Lett., 36, 539.

94. Xu, H.Z., Zhao, F., Majumdar, A., Shi, Z. 2003, Electron Lett., 39, 659.

95. Zhao, F., Wu, H., Majumdar, A., Shi, Z. 2003, Appl. Phys. Lett., 83, 5133.

96. Zhao, F., Lv, X., Majumdar, A., Shi, Z. 2004, Appl Phys Lett., 84, 1251.

97. Kubota, K., Vaccaro, P.O., Ohtani, N., Hirose, Y., Hosoda, M., Aida, T. 2002, Phys. E, 13, 313.

98. Grundmann, M. 2003, Appl. Phys. Lett., 83, 2444.

99. Schmidt, O.G., Jin-Phillipp, N.Y. 2001, Appl. Phys. Lett., 78, 3310.

100. Majumdar, A., Guan, Z.P., Zhao, F., Li, D., Ray, D., Jain, S., Mukherjee, S., Shi, Z. 2006, Appl. Phys. Lett., 88, 171111-1.

101. Jain, S., Mukherjee, S., Guan, Z.P., Ray, D., Zhao, F., Li, D., Shi, Z. 2007, Phys.

E, 39, 120.

102. Mukherjee, S., Jain, S., Zhao, F., Kar, J.P., Li, D., Shi, Z. 2008, J. Mat. Sc. Mat. Electron., 19, 237.

103. Gayral, B., Gerard, J.M.,Lemaitre, A., Dupuis, C., Manin, L., Pelouard, J.L. 1999, Appl. Phys. Lett., 75, 1908.

Lead salt thin film semiconductors for microelectronic applications

87

104. Reithmaier, J.P., Sek, G., Loffler, A., Hofmann, C., Kuhn, S., Reitzenstein, S., Keldysh, L.V., Kulakovskii, V.D., Reinecke, T.L., Forchel, A. 2004, Nature, 432, 197.

105. Graham, L.A., Huffaker, D.L., Deppe, D.G. 1999, Appl. Phys. Lett., 74, 2408.

106. Deppe, D.G., Graham, L.A., Huffaker, D.L. 1999, IEEE J. Quant. Electron., 35, 1502.

107. Mukherjee, S., Jain, S., Zhao, F., Kar, J.P., Shi, Z. 2007, J. Microelec., 38, 1181.

108. Mukherjee, S., Jain, S., Kar, J.P., Shi, Z. 2007, International Workshop on Phys. of Sem. Dev., 413.

109. Schmidt, O.G., Eberl, K. 2001, Nature, 412, 42.

110. Nurnus, J., Vetter, U., Koenig, J., Glatthaar, R., Lambrecht, A., Weik, F., Tomm, J.W. 2005, Proceedings of the SPIE,5840, 212.

111. Mukherjee, S., Jain, S., Zhao, F., Kar, J.P., Li, D., Shi, Z. 2008, Microelec. Eng., 85, 665.

112. Weidmann, D., Tittel, F.K., Aellen, T., Beck, M., Hofstetter, D., Faist, J., Blaser, S. 2004, Appl. Phys. B, 79, 907.

113. Richter, D., Lancaster, D.G., Curl, R.F., Neu, W., Tittel, F.K. 1998, Appl. Phys.

B, B67, 347.

114. McCabe, S., MacCraith, B.D. 1993, Elect. Lett., 29, 1719.

115. Shi, Z., Xu, G., McCann, P.J., Fang, X.M., Dai, N., Felix, C.L., Bewley, W.W., Vurgaftman, I., Meyer, J. R. 2000, Appl. Phys. Lett., 76, 3688.

116. Felix, C.L., Bewley, W.W., Vurgaftman, I., Lindle, J.R., Meyer, J.R., Wu, H.Z., Xu, G., Khosravani, S., Shi, Z. 2001, Appl. Phys. Lett., 78, 3770.

117. Zhao, F., Wu, H., Jayasinghe, L., Shi, Z. 2002, Appl. Phys. Lett., 80, 1129.

118. Xu, H., Zhao, F., Majumdar, A., Jayasinghe, L., Shi, Z. 2003, Electron. Lett., 39, 659.

119. Joannopoulos, J.D., Meade, R.D., Winn, J.N. 1995, Princeton University Press, Princeton, N.J. 120. Joannopoulos, J.D., Villeneuve, P.R., Fan, S. 1997, Nat., 386, 143.

121. Krauss, T., Song, Y.P., Thoms, S., Wilkinson, C.D.W., DelaRue, R.M. 1994, Electron. Lett., 30, 1444.

122. Krauss, T.F., Rue, R.M.D.L., Brand, S. 1996, Nat., 383, 699.

123. Oâ&#x20AC;&#x2122;Brien, J., Painter, O., Cheng, C.C., Lee, R., Scherer, A., Yariv, A. 1996, Electron. Lett., 32, 2243.

124. Baba T., Matsuzaki, T. 1996, Jpn. J. Appl. Phys., 35, 1348.

125. Hamano, T., Hirayama, H., Aoyyagi, Y. 1997, Conference on Lasers and ElectroOptics, vol. 11 of 1997 OSA Technical Digest Series (Optical Society of America, Washington, D.C.), 528.

126. Yablonovitch, E. 1987, Phys. Rev. Lett., 58, 2059.

127. Sigalas, M.M., Biswas, R., Li, Q., Crouch, D., Leung, W., Jacobs-Woodbury, R., Laogh, B., Nielsen, S., McCalmont, S., Tuttle, G., Ho, K.M. 1997, Microwave Opt. Tech. Lett., 15, 153.

128. Gauthier, G.P., Courtay, A., Rebeiz, G.M. 1997, IEEE Trans. Antennas Propag., 45, 1310.

129. Maloney, J.G., Kesler, M.P., Shirley, B.L., Smith, G.S. 1997, Microwave Opt. Tech. Lett., 14, 261.

130. Opti-FDTD user manual, Optiwave.

88

Shaibal Mukherjee et al.

131. Yee, K.S. 1966, IEEE Trans. Antennas. Prop., AP-14, 302.

132. Taflove, A., Brodwin, M.E. 1975, IEEE Trans. Microwave Theory Tech., MTT23, 623.

133. Joannopoulos, J.D., Meade, R.D., Winn, J.N. 1995, Princeton University Press.

134. Johnson, S.G., Joannopoulos, J.D. 2000, Opt. Exp., 8, 173.

135. Guo, S., Albin, S. 2003, Opt. Exp., 11, 167.

136. Chaudhuri, P.R., Paulose, V., Zhao, C., Lu, C. 2004, IEEE Photon. Technol.

Lett., 16, 1301.

137. Chaudhuri, P.R., Ghatak, A.K., Pal, B.P., Lu, C. 2005, Opt. Las. Technol., 37, 61.

138. Snyder, A.W., Love, J.D. 1983, London:Chapman & Hall, 487.

139. Kawano, K., Kitoh, T. 2001, 2001, Chapter 4, New York:Wiley.

140. Painter, O., Vuckovic, J., Scherer, A. 1999, J. Opt. Soc. Am. B, 16, 275.

141. Mukherjee, S., Bi, G., Kar, J.P., Shi, Z. 2009, Opt. Appl., 2009, XXXIX, 499.

142. Berenger, J.P. 1994, J. Comput. Phys., 114, 185.

143. Katz, D.S., Thiele, E.T., Taflove, A. 1994, IEEE Microwave Guid. Wave Lett., 4, 268.

144. Berenger, J.P. 1996, J. Comput. Phys., 127, 363.