Wide-band-gap Semiconductors

1992

Chapter 1

General reviews

Thin films and devices of diamond, silicon carbide and gallium nitride

Robert F. Davis ,     Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC, USA

Robert F. Davis, Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC 27607, USA.

The extreme properties of diamond, SiC and GaN provide combinations of attributes for high-power, -temperature, -frequency and optoelectronic applications. The methods of deposition, the results of chemical, structural, microstructural and electrical characterization and device development are reviewed for thin films of these three materials. Problems and areas of future research are also noted.

1 Introduction

The principal driving force for the current resurgence in interest in diamond (C), SiC and GaN (as well as cubic-BN, AlN, InN and alloys of these materials) is their potential as high-power, high-temperature microelectronic and/or optoelectronic devices resistant to radiation damage.

Johnson [1] has argued that the basic limitation on the high frequency and high power capabilities of selected discrete transistor devices is set by the product of the breakdown electric field, EB, and the saturated (limiting) electron velocity, Vs, i.e. the velocity at which an electron has enough energy to emit an optical phonon. Johnson’s figures of merit (EBVS/π)² for selected semiconductors and their associated ratios to those for silicon are presented in table 1.

Table 1

Johnson’s figure of merit for a transistor’s power and frequency performance for selected semiconductor materials.

Keyes [2] has considered the switching speed of transistors in integrated circuits for computer logic applications. In this case a high limiting velocity allows high speed devices. A problem of small, closely spaced devices is the very high density of power dissipation, i.e. the production of heat. The heat flows from the device by conduction through the semiconductor material thereby encountering a thermal resistance that is inversely proportional to the size of the device. A lower limit is set on the device size by the maximum permissible thermal resistance and, thus, a high thermal conductivity, σT, is highly desirable in the device material. An additional set of figures of merit which includes Vs, σT and the dielectric constant K has been developed by Keyes [2] and is shown in table 2.

Table 2

Keyes’ figure of merit for the speed of a transistor for selected semiconductors.

Finally GaN and SiC are also of interest as light-emitting devices and for other optical applications. Important optical device applications for the nitrides include semiconductor lasers, light-emitting diodes (LEDs) and optical detectors. Although band-to-band transitions in SiC occur indirectly via intermediate energy levels in the conduction band and thus negate the possibility of laser applications, doping of the material with various donor and acceptor species allows electroluminescent devices to be produced. The creation of an abrupt p(aluminium doped)-n(nitrogen doped) junction in the α(6H)–SiC polytype (see discussion of polytypism in SiC in section 3) also allows the fabrication of blue LEDs in this form of SiC.

The following sections present an overview of the results and conclusions of recent materials research on diamond (diamond-like materials are not discussed), SiC and GaN as they pertain to microelectronics and optoelectronics, as well as a description of problems encountered and suggestions for future directions for research to overcome these problems.

2 Diamond

2.1 Background

The free energy difference between metastable diamond and stable graphite is only 453 cal/mol [3] at 1 atm. Thus the rarity of diamond is surprising when one considers that this is less than RT (≈600cal/mol) at room temperature. There are several reasons for the uncommon occurrence of diamond. Firstly, the kinetics of graphite formation are considerably faster than those of diamond, as would be predicted by the large activation energy barrier between these two materials [4]. Secondly, the entropy of diamond is less than the entropy of graphite by 0.78 cal/K mol [3]. Finally, a large activation energy barrier exists between graphite and diamond which prevents even small particles from forming at room temperature under equilibrium conditions. The density of diamond (3.51 g/cm³) is significantly greater than the density of graphite (2.25 g/cm³); thus, the former becomes the equilibrium form of C only at high pressure.

The breakthrough in the low-pressure synthesis of diamond was mainly a result of research by Soviet scientists in the late 1970s and early 1980s which indicated that gas activation techniques could greatly increase the growth rate of diamond while suppressing the graphite deposition. The results of these efforts clearly demonstrated that a concentration of atomic hydrogen, exceeding that of the equilibrium concentration associated with the thermal disassociation of the reactant hydrocarbon gas(es), was necessary to achieve diamond growth [5]. The flow of the hydrocarbon gas and hydrogen could now be simultaneous. In 1977, Derjaguin and Fedoseev [6] described three approaches to produce higher concentrations of atomic hydrogen than obtained previously: catalytic, electrical discharge and via heated tungsten filament. The significant results of this research were not published in English until 1981 [7] with additional details in 1984 [8]. This was followed by intensive efforts throughout the 1980s at the National Institute for Research in Inorganic Materials in Japan. This research quickly resulted in numerous publications [9–15] containing scanning electron micrographs of highly faceted, thick continuous films, confirming electron diffraction results and a host of spectroscopic data which removed all doubt that diamond films containing very small amounts of graphite could be grown at reasonable rates under highly metastable conditions on a variety of substrates, including Si, SiC, WC, Ta, W, Si3N4, Cu, Ni, Mo, BN, SiO2, Ti, Hf, Nb and Al2O3. This work also resulted in several new techniques for gas activation and diamond deposition. These have been grouped into five major categories [16]: (i) thermally activated chemical vapor deposition (CVD); (ii) high-frequency plasma enhanced CVD; (iii) direct-current (DC) discharge assisted CVD; (iv) combustion flame; and (v) hybrids of these and others. Only (i), (ii) and (iii) are considered likely to result in films suitable for electronic applications. Common to all techniques is the production of a high supersaturation of atomic H with a supersaturation of carbonic species and a substrate temperature in the range of 975–1275 K. Other similarities include process parameters of gas pressure (∼0.1 atm), the percentage of hydrogen in the gas phase (∼95–99.9%) and the resulting film morphology.

2.2 Nucleation and growth

2.2.1 The role of atomic hydrogen

One of the most important aspects of any discussion of the mechanisms and theory of diamond growth from the vapor phase is the role of atomic H. Firstly, within all the successful growth schemes which use H2 and a C-containing gas mixture for vapor phase deposition. The CH4 (or other hydrocarbon) and H2 are excited and dissociate into atomic H, hydrocarbon radicals and, in the case of plasma deposition, ions. The hydrocarbon species then adsorb onto the surface of the substrate and begin to grow via various hydrogen releasing reactions. These clusters may contain single, double or triple bonded carbons. However, the supersaturation of atomic H rapidly attacks all but the sp³ bonded diamond component. This may help to maintain sp³ bonding in this phase and thus reduce the amount of graphitic or amorphous C which deposits on the substrate.

Atomic H may also play a variety of other important roles in the deposition of diamond. Diamond bonds are all single bonds; graphite contains both single and double bonds. The dangling bonds on the surface of diamond are energetically unstable, and the cleaved surface will soon reconstruct to a ‘graphite-like’ surface containing a mixture of single and double bonds to reduce the surface energy. A monolayer of reacted atomic H bonded to the single bonds of diamond will prevent the reconstruction. However, because the H-C bond is stronger than the C-C bond, no diamond growth can occur, unless a bound H atom can be displaced from the surface.

The continuous exposure of the H-terminated surface to the incoming atomic H provides an escape from the no-growth situation just noted by providing vacant sites created by the reaction between the bonded and gas phase atomic H species. Because the H–H bond is stronger than the H-C bond, this is an exothermic event. In most CVD diamond growth, the ratio of H2 to C is ∼50:1 [17]. Thus the most probable next event is that another H will fill a vacant site. However, occasionally, a C-containing species will interact with the vacant site and attach to the diamond crystal, thereby generating diamond growth. Experimental data indicates that ∼10⁴ H have to be produced for each C atom that attaches to the diamond lattice. Some H are lost through collisions in the gas phase or with the wall of the growth chamber, but most are lost through H–H recombination of the growth surface. This inefficient use of atomic H is a significant economic factor in the growth of diamond via CVD.

Finally, it has been proposed by Frenklach [18] that molecular hydrogen at high temperatures suppresses the formation of polycyclic aromatic hydrocarbon radicals (PA). These species are graphite precursors and thus inhibit the formation of diamond. The H2 reacts with the PA to form a molecule (PAH). If a radical PA forms a PAH molecule, the latter can react and attach to another PAH molecule and extend the array. Eventually the array will evolve into a graphitic rather than a diamond structure. The evolution of PAH arrays is arrested by suppressing the formation of radicals with excess molecular H.

2.2.2 Growth on non-diamond substrates

One generally desires to obtain perfect films with a smooth surface morphology and a flat, defect free interface between the epitaxial partners. In thermodynamic theory, it is assumed that the mode of thin film growth is determined by the mode of nucleation [19,20]. It is also assumed that no chemical reactions, alloying, interdiffusion or other changes occur at the interface. The nucleus on a substrate is considered to be cap-shaped having a contact angle q. The surface and interfacial energies per unit area s are considered to be forces which are resolved to obtain Young’s equation:

(1)

where the subscripts s, i and o refer to the substrate, interface and overgrowth, respectively. The equation gives rise to three cases:

(2)

[21]in this case ss < so + si, and the cohesion energy between the adsorbing and condensing (i.e., nuclei are forming, as noted above) adatoms is stronger than the adhesion energy between the substrate atoms and the adatoms. Island growth occurs from the beginning of deposition if the surface energy inequality given above pertains.

(3)

[22]in this case sso + si. The adhesion energy exceeds the cohesion energy and the film tends to grow layer-by-layer.

(4)

[23]in this case sso + si but the strain energy in the overgrowth is large with respect to so. This mode of growth has been observed in most heteroepitaxial systems in which the surface energy criterion noted in (iii) is obeyed by the materials systems.

In SK growth, one or several complete monolayers are formed. The total strain in the layer increases due to lattice mismatch but is not relieved (at least sufficiently) by the formation of dislocations at the interface (misfit or other) or inside the film. These lattice energies must be expended. As a result, three-dimensional islands similar to those in the Volmer-Weber mode begin to grow, because the total energy of the substrate/film system is reduced if the adhesive forces at the interface have a smaller range than the cohesive elastic forces throughout the film.

Chemically unterminated diamond surfaces possess the highest surface energies (>4000 erg/cm²) of any known material. Therefore, for the nucleation and growth of diamond directly on any of the non-diamond substrates noted above, the most accommodating growth mode that one can expect would be that of Volmer-Weber. Indeed, observations of diamond growth via microscopy have shown that the individual crystals invariably grow from islands of nucleation which indicates that essentially all thin films of diamond grow by this growth mode.

2.3 Characterization of the deposited films

As described above, the deposition of films from C precursors may result in the formation of amorphous C, graphite and/or diamond. The primary characteristics of graphite and diamond are a result of their sp² and sp³ bonding, respectively. To determine the nature of the bond type(s) in the film, one may use Auger electron spectroscopy (AES) or Raman spectroscopy.

Lurie and Wilson [24] have shown that the AES fine structure of the C KLL peak is sensitive to the types of C bonding and is not affected significantly by charging of the sample. However, the sensitivity of this technique to the concentration of sp² bonding is not presently known.

Raman spectroscopy is the most frequently used technique for the determination of the types of C bonding in the thin films and coatings. However, there is some uncertainty due to crystallite size effects and significant scattering cross section differences between sp² and sp³ type bonding. In the case of most CVD films, the 1332 cm–1 line indicative of diamond is accompanied by a broad band centered around 1500 cm–1 and believed to be due to the sp² or graphitic bonding. A major problem is that the incorporation of low concentrations of sp² bonding may destroy any evidence of diamond due to its ∼50 times larger scattering cross section [26]. This controversy has been fully discussed in refs. [27,28].

Scanning electron microscopy is by far the most used technique for the characterization of the surface morphology of diamond films. The overall microstructure and the morphology of the individual diamond particles are dependent on the hydrocarbon concentration in the gas feed. They may also depend on the substrate temperature, the pretreatment of the substrate, the total pressure and the total time of deposition.

Transmission electron microscopy [25] is used to investigate the structural defects which are usually extensive in CVD diamond, and can strongly affect the electrical properties of the films. These include planar defects (microtwins and stacking faults), line defects (dislocations), point defects (impurities and vacancies) and graphite. Twining and stacking faults are the most predominant defects observed in diamond samples.

2.4 Applications of diamond films in electronics

Device activity regarding the production of discrete electronic devices in natural or synthetic diamond has been moderate, albeit continuous, over the past four decades, especially in the former Soviet Union and more recently in the United States and Japan. An overview of these efforts has been presented by Gildenblat [29]. The use of natural or high pressure synthetic diamond crystals has allowed considerable advancement in the development of the fabrication routes necessary for device fabrication. However, at present the use of these materials limits the potential for broad application because of high cost, low area and device integration problems. In theory, most of these problems can be alleviated by the use of the diamond thin films now being grown at low pressures and temperatures, and described above.

Several groups have electrically characterized these low pressure polycrystalline diamond films of which a selected number shall be mentioned. Pan et al. [30] have measured carrier lifetimes and found the values to be related to grain size and to vary from less than 100 ps in the smallest grain films to above 600 ps in the largest. Mobilities were 10–20 cm²/Vs. Fujimori et al. [31] and Nishimura et al. [32] have determined the electrical character of B-doped conducting diamond films deposited on monocrystalline diamond and sapphire substrates, respectively. In the former study, the resistivities were in the range of 1–10–3 Ω cm. The activation energy for conduction was 0.013 eV, which is lower than the value obtained for B-doped diamond produced by high pressure synthesis. In the latter study, the minimal resistivity was also 1 Ω cm at room temperature. In this case the activation energy ranged from 0.048 to 0.16 eV when the B2H6/CH4 ratios decreased from 200 ppm to 20 ppm. From Hall measurements at room temperature, mobilities and polarities and carrier concentrations were obtained by assuming a single carrier model. All samples were p-type and mobilities were of the order of 1 cm²/Vs, and carrier concentrations were 10¹⁹–10²⁰/cm³ at room temperature.

Grot et al. [33] have conducted extensive studies of ohmic and Schottky contacts to conducting diamond films deposited on Si and diamond. They found that the measured conductivity of the diamond film was strongly affected by the surface treatment. Exposure to H2 plasma produced a conductive surface layer from which ohmic I–V characteristics of Au/diamond contacts were determined regardless of the doping level of the diamond film. By contrast, cleaning in a selected chemical mixture of CrO3 + H2SO4 followed by a rinse in H2O2 + 30% NH4OH resulted in Au-gate Schottky diodes with excellent rectifying characteristics. The sheet resistance was in the range of 5–60 kΩ/q. The measured activation energy for the diode series resistance was ∼0.15 eV. The reverse current was also thermally activated with an activation energy of 0.45 eV. Similar rectifying characteristics were observed for AlPd contacts to homoepitaxial diamond films. Annealing of the Au contacts to 580°C did not affect the characteristics of the diodes. As a result of this research, rudimentary MESFETs have now been fabricated [34–36]. The rectification and internal photoemission in metal/CVD diamond/p-type Si structures have also been reported [37–39].

2.5 Summary

It is a time of considerable excitement in the field of diamond synthesis. At this point we have only a rudimentary understanding regarding the nucleation and growth phenomena which attend the formation of diamond although the presence of some atomic hydrogen appears to be necessary in most growth schemes. Moreover, mono-crystalline thin films have only been deposited on diamond substrates while only highly faceted, polycrystalline films, often with rough surfaces, have been produced on any other substrate. Field and gate insulators, contacts and materials with shallow donor levels are also needed to further advance this material in the electronics arena.

By contrast, characterization techniques for the verification of the presence of the diamond phase (and other carbon phases) are well established. Moreover, boron-doped, p-type films with moderate activation energies have been grown and simple electronic devices with high operating temperatures have been fabricated on conventional synthetic diamond.

3 Silicon carbide

3.1 Background

In certain close-packed structures such as SiC, there exists a special one-dimensional polymorphism called polytypism. Polytypes are alike in the two dimensions of the close-packed planes but differ in the stacking sequence in the dimension perpendicular to these planes. In SiC, the stacking sequence of the close-packed planes of covalently bonded tetrahedra (either SiC4 or CSi4) can be described by the ABC notation. If the ABC stacking is repetitive (ABCABC …), one obtains the zincblende structure. This is the only cubic SiC polytype and is referred to as 3C or β-SiC where the 3 refers to the number of planes in the periodic stacking sequence. The hexagonal (… ABAB …) sequence is also found in SiC. Furthermore, both can also occur in more complex, intermixed forms yielding a wider range of ordered, larger period, stacked hexagonal or rhombohedral structures of which 6H is the most common. All of these non-cubic structures are known collectively as α-SiC.

From an electronics viewpoint, SiC possesses a range of band gap energies at 300 K [40], e.g., 2.2 eV (3C), 3.0 eV (6H) and 3.3 eV (2H), as well as high values of saturated electron drift velocity (2 × 10⁷ cm/s for (6H)) [41] (a slightly higher value has been predicted [42] for 3C (beta) because of reduced phonon scattering), junction breakdown electric field (5 × 10⁶ V/cm) [43] and thermal conductivity (3.5W/cm°C at 300 K) [44].

The high thermal conductivity of SiC also indicates the potential for high-density integration of SiC devices. This idea has received considerable impetus with the recent development and scale-up of the seeded-sublimation growth technique for producing commercially viable single crystal α(6H)–SiC boules from which highly thermally conducting wafers are attainable. Significant advances have also been made in the growth of monocrystalline beta and alpha (6H) thin films primarily via chemical vapor deposition (CVD), in the other technologies necessary for viable devices and in the fabrication of excellent discrete devices. The following sections review the recent advances in this field, especially in the areas of thin-film growth, device fabrication and material and device characterization.

3.2 Thin-film growth and characterization

Until recently, monocrystalline Si (100) has been almost universally adopted as the substrate of choice for the growth of β-SiC thin films because of the availability of the former in well-characterized and reproducible forms of controlled purity. The mismatches in the coefficients of thermal expansion (about 8%) and the lattice parameters (about 20%) have been accommodated via the initial reaction of the Si(100) surface with a carbon-containing gas. The successful epitaxial deposition of relatively thick (up to 30μm), crack-free β-SiC films on this β-SiC converted layer using individual carbon- and silicon-containing gases has been initially reported by Nishino et al. [45,46] and subsequently employed by numerous groups in the United States, Europe and especially Japan (see refs. [47–51] for reviews and collected papers describing this research).

Transmission electron microscopy (TEM) has shown the converted layer to be microscopically rough, varying in thickness from 5 to 12 nm. It contains a high density of {111} microtwins and intrinsic stacking faults. In addition, localized regions of the converted layer exhibit disorder [53]. For a review of the results of several analyses of the converted layers and the SiC films, see refs. [52–54].

Plan-view and cross-sectional studies of the β-SiC films also reveal extensive concentrations of microtwins, intrinsic stacking faults and inversion domain boundaries (IDBs) [55–57] on the {111} planes. The density of defects is higher at the interface; it decreases over a distance of 3–4 μm from the interface and becomes approximately equal to that near the surface, even for films 20 μm thick. To eliminate the IDBs, vicinal Si substrates were used with the 〈100〉 inclined 2–4° towards 〈011〉 [55,58,59].

The research regarding the deposition of β-SiC on Si (100) has now sufficiently matured such that new deposition technologies are now being investigated to determine if the film quality can be improved. Fuyuki et al. [60] have recently reported atomic layer-by-layer control using gas source molecular beam epitaxy of the deposition of monocrystalline β-SiC within the temperature range 1250–1320 K on β-SiC(100) substrates previously prepared on Si(100) by the two-step CVD process described above.

〉 has also recently been reported by Sugii et al. [61]. In this case, the Si surface was initially subjected to a beam of C2H2 at 1173 K for 600 s to effect the epitaxial growth of a 1–2 nm layer of SiC, as is the case in CVD. Monocrystalline SiC was subsequently deposited on this layer at 1173 K using beams of C2H2 and Si simultaneously. Cross-sectional TEM revealed the layers to be nearly as good as that produced via CVD at higher temperatures. Photostimulated epitaxial deposition of β-SiC on sapphire (a-Al2O3) (0001) has been achieved by Nakamatsu and co-workers [62,63] using an ArF laser, C2H2 and Si2H6 in the temperature range of 1253–1425 K, and a pressure of 10–2 Pa. Wavelengths shorter than 310 nm were effective for epitaxial growth. Similar studies using an Ar laser and CH3SiCl3 and (CH3)2 SiCl2 carried in N2 or CO2 have been conducted by Salun et al. [64]. The 20–60 μm diameter of the beam caused the growth of SiC rods at a rate of 1–70 μm/s depending on the starting compounds and the power density of the laser. Silicon carbide SiC was formed when (CH3)2SiCl2 in N2 was used, but only graphite occurred when CH3SiCl3 was used.

The almost universal use of Si has allowed considerable progress in the understanding of thin film growth as well as the additional procedures necessary to produce devices (see below). In an attempt to substantially reduce the concentrations of all defects simultaneously in the SiC films, growth on the Si(0001) and C(0001) faces of commercial (Acheson-derived) 6H–SiC single crystals substrates has been investigated at NCSU within the temperature range of 1683–1823 K at 1 atm total pressure [65–67]. The resulting films were β-SiC(111) at all temperatures studied. Similar research has been conducted by Powell et al. [68] using 6H–SiC crystals with the (0001) surface subjected to various polishing, oxidation and etching treatments.

Low concentrations of microtwins and stacking faults were observed in these films. High resolution cross-sectional TEM showed an abrupt and coherent β-SiC/α-SiC interface. However, examination in plan view revealed the presence of double positioning boundaries (DPBs) on both the Si and C faces [0] [70]. All films produced on these off-axis substrates were the pure 6H polytype. Matsunami and co-workers [71,72] and Powell et al. [73] have recently reported similar off-axis growth studies. The high quality of these films were demonstrated in the latter study from the results of optical and electron microscopy and low-temperature photo-luminescence.

3.3 Device fabrication and characterization

Operable SiC devices have been successfully fabricated in α-SiC in previous concentrated efforts to develop this material as a semiconductor (for an overview of this earlier work see ref. [74]). However, the devices were produced in Lely crystals, the character of which was not reproducible from run to run. A brief review of selected device research and characterization in b- and α-SiC materials conducted primarily in the 5 years preceding this publication is discussed in the following two subsections.

Suzuki and co-workers [75,76] have deposited N- and B-doped β-SiC(100) films on Si(100) substrates to produce p–n junction diodes having a mesa structure. They determined their electrical character to 773 K. No significant chemical degradation was observed at any temperature. At room temperature, the device had a forward turn-on voltage of about 1.2 V and a reverse leakage current of 5 μA at 5 V. However, both parameters seriously degraded with an increase in temperature. The curves of forward current density versus voltage contained three distinct regions at all temperatures which indicated the following: (1) excess currents at applied voltages lower than about 0.6 V; (2) exponentially increasing currents at 0.6–1.1 V; (3) gradually increasing currents at voltages greater than 1.1 V.

The linear I–V characteristics of p–n junction diodes recently produced in doped, low defect density α(6H)–SiC films [77] showed linear behavior with a peak inverse voltage of 455 V and avalanche breakdown. The breakdown initially occurred at ∼420 V with a reverse IL of 4 μA, increasing to 50 μA at 450 V. Thereafter the device operated in avalanche with current increasing linearly with voltage. The value of the ideality factors in forward bias where two linear regions were observed were 1.6 and 2.0 thus indicating a compromise between diffusion current and generation-recombination dominated processes. Heating the diode to 625 K caused (1) decrease in the turn-on voltage from ∼2.4 V to ∼2.0 V, due to a reduction in band gap, and thus an increase in the intrinsic carrier concentration, and (2) a slight increase in the pre-breakdown leakage current due to new thermally generated carriers. Otherwise the graphs were identical.

Research on α(6H)–SiC MESFETs has also been conducted [77]. This device had a gate length of 10 μm, a gate width of 1 mm and a source to drain distance of 30 μm. At room temperature (293 K), the device showed very good current saturation to a drain voltage (VD) of 40 V These devices continued to operate well at temperatures to 773 K.

Shibahara et al. [78], Fuma et al. [79] and Palmour and co-workers [80] have produced working, enhancement mode (normally off) metal/oxide/semiconductor field effect transistors (MOSFETs) on β-SiC(100) films deposited on Si(100). The MOSFET reported by Shibahara et al. [78] was fabricated in boron-doped, p-type SiC layers (about 2 μm thick and with a carrier concentration of about 1 × 10¹⁵ cm–3) with the source and drain formed by ion implantation of P+ and subsequent annealing at 1350 K for 3.6 × 10³ s. The IL between the source and gate was less than 1 μA for a gate bias of ±10V. The threshold voltage and effective mobility of the FET were estimated to be 2.8 V and 100 cm²/V s. The ID did not saturate. Fuma et al. [79] produced their MOSFET device in an n–p–n structure deposited on a p-type Si substrate. The p-type dopant was obtained from B2H6. The drain current tended to saturate at room temperature. The maximum transconductance was 200 μS/mm. These characteristics degraded considerably above 473 K. The devices fabricated by Palmour et al. [80] used a p-layer (Al-doped) deposited on n-type Si into which a dual N implant (peak concentrations = 5 × 10¹⁴ and 3.4 × 10¹⁴ cm–3) was produced at 773 K. Several lengths were investigated in this work and that of Fuma et al. [79] described above. The devices [80] showed good saturation at room temperature and could be pinched off using a VG of −4V (VD = 5V). The maximum transconductance was 1.65 mS/mm for a 3.5 μm gate length device at 673 K. The devices also showed good stability to 673 K; however, the IL increased from 10 μ A (298 K) to 80 μA (673 K).

The material structure and the device I–V characteristics of a 10 mm channel length α(6H)–SiC inversion-mode MOSFET have been reported in [77]. Very good current saturation was obtained at room temperature to a VD = 30 V. A maximum transconductance of 0.27 mS/mm was observed at VG = 24 V The threshold voltage was very high, about VG = 9 V. This device continued to operate very well to 923 K. The leakage current was 2 mA at VD = 25 V and VG = 0 V. The maximum transconductance at 923 K was 0.43 mS/mm at VG = 6V.

Recent research on light emitting diodes has been conducted in Germany by Ziegler et al. [81], in Japan by Suzuki et al. [82], Matsunami et al. [83], Ikeda et al. [84], Matsunami [85] and Koga et al. [86] and in the United States by Edmond [87]. These investigators have prepared Al/N p–n junctions in 6H–SiC using either the combination of a Si solution contained in a graphite crucible, a rotating dipping method and resulting sequential double epitaxial growth, or double epitaxial growth via CVD.

As shown by Ikeda et al. [84], main mechanisms for light emission are attributed to (i) donor-acceptor (D–A) pair recombination (∼480nm), (ii) bound exciton recombination at localized centers related to Al impurities (∼455 nm) and (iii) free exciton recombination (∼425 nm).

Blue emitting LEDs having a 470 nm peak wavelength and spectral halfwidth of 70 nm at 20 mA are being produced and marketed commercially [87]. These diodes feature a 5 V maximum reverse voltage with a typical forward voltage of 3 V at 20 mA and a maximum forward operating current of 50 mA. The external quantum efficiency at 20 mA is as high as 2.5 × 10–2% with a typical output of 2.0 × 10–2%.

Dmitriev et al. [88] have produced a three-color (blue(470 nm)–green(510 nm)–red(650 nm)) SiC single-crystal display using epitaxial films grown on a(6H) substrates by the containerless method noted above at ∼1775 K. The p-regions were produced by ion implantation with subsequent annealing at 2073 for 5 min. The diameter of each diode was 300–500 μm with a distance between emitters of 300 μm.

Finally, Schottky diodes fabricated in both 6H and 4H polytopes have been used for ultraviolet photodetectors [89]. A semitransparent layer of Cr was used as a Schottky contact. The diffusion length of the generated light holes was estimated to be 0.1–0.3 μm. Glasow et al. [90] have also produced photodiodes by N implantation into p-type 6H–SiC. Maximum quantum efficiencies of ∼75% were observed around 280 nm. The diodes were sensitive in the wavelength range of 250–450 nm.

3.4 Summary

Monocrystalline thin films of high-purity β-SiC have been grown directly on Si(100) and α-SiC(0001). Problems derived primarily from interface related defects and subboundaries in the material have prevented the optimization of the quality of the films. However, deposition on these materials oriented 2–4° off-axis allowed the elimination of the APBs in the material grown on Si(100) and the DPBs in the material deposited on SiC(0001); the latter resulting in a combination of a-SiC on α-SiC appears especially viable for device fabrication. Progress in areas related to device fabrication has allowed continual improvement in several types of devices. However, extensive research is still required to reduce the defect density of the wafers and films, to improve the MOS characteristics, to achieve low resistance ohmic contacts and to further purify the source gases.

4 III–V Nitrides

4.1 Background

The commercial realization of light emitting diodes and semiconductor lasers in the previous decade has generated much interest in thin films of the wider bandgap semiconductor III–V mononitrides, especially GaN. This material, in the wurtzite form, has a bandgap of 3.45 eV (near UV region) at room temperature. It also forms a continuous range of solid solutions with AlN (6.28 eV) and InN (1.95 eV). Thus optoelectronic devices having specifically engineered band gaps with energies from the visible to the deep UV region are theoretically possible with these materials. The following subsections review the recent research on GaN to the present. The problems which are currently limiting the development of this material and the other nitrides as well as devices made from them are also noted.

4.2 Gallium nitride

Films of GaN have been grown from the vapor phase by several techniques including low and high pressure CVD, plasma-enhanced CVD, reactive ionized-cluster beam deposition, reactive and ionized molecular beam epitaxy. Reactive sputtering of gallium in NH3, N2−NH3 or N2−Ar mixtures and solution growth from Ga or Ga–Bi mixtures have been occasionally used to grow crystalline or amorphous films and small bulk crystals, respectively. Sapphire has been the most widely used substrate, although its lattice parameters and coefficients of thermal expansion are about 23% and about 25% greater, respectively, than those of GaN. Recently, films of cubic β- and wafers of hexagonal α(6H)–SiC, Si and GaAs single crystals have been employed as substrates. Two reviews of the deposition methods and luminescent properties of this material have recently appeared [91,92]. The following summarizes selected portions of the research since that time.

The recent research regarding pure GaN has been concerned with four main aspects: (1) achievement of the cubic β-GaN; (2) the use of unusual techniques to deposit the material at low (<1300K) temperatures; (3) the application of an A1N buffer layer between the substrate (usually sapphire) and the GaN film; and (4) the use of electron beams to achieve p-type material and p-n junctions.

The first reference to the apparent achievement of the cubic (zincblende structure) β-GaN is the research of Mizuta et al. [flux ratios between 0.63 and 3.9 which supports the common theory that the high donor concentrations are associated with N vacancies. Resistivities >10⁶ Ω cm were achieved. The latter group [96] reported the lattice constant to be 0.45 nm. TEM revealed a high dislocation density in the GaN due to the large lattice mismatch which extended into the GaAs. The electrical resistivity exceeded 100 Ω cm. The band gap was reported to 3.26 eV, in agreement with the results of Paisley et al. [94]. Most recently, Moustakas and colleagues [97] have achieved the growth of β-GaN on Si(100) via MBE at low temperatures. Extensive X-ray analyses were conducted.

The surface morphology of GaN films may be markedly improved if an AIN ‘buffer’ layer is first deposited on any of the substrates noted above (especially sapphire) as shown initially by Yoshida et al. [98,99]. Amano et al. [100] and Akasaki and co-workers [101] reported that the use of the AIN layer reduces microscopic fluctuation in crystallite orientation, relaxes the strain between sapphire and GaN and promotes lateral growth of the film due to the decrease in interfacial free energy between the film and the substrate. These investigators used the positive effects of this layer to grow high quality GaN films in which they observed stimulated edge emission [102] at room temperature in long (2.15 mm) GaN cavities prepared at atmospheric pressure using trimethylgallium and ammonia. They have also produced p–n junctions and light-emitting diodes in this material, as discussed below.

Khan et al. [103,104] used a metalorganic (MO) CVD process similar to that of Akasaki and Amano but at 76 Torr. The AIN buffer layer was 50 nm thick. They reported [103] electron carrier densities as low as 1 × 10¹⁷ with mobilities in excess of 300 cm²/Vs at room temperature. Photoluminescence line widths as low as 25 meV were achieved. In addition, they reported [104], for the first time, vertical cavity stimulated emission at room temperature from 1.5 μm thick GaN layers. The gain coefficient was ∼10⁴ cm–1.

It would seem more logical to use a GaN buffer layer produced at lower temperatures than the growth because of the exact lattice and expansion match with the subsequently deposited film. This approach has been successfully used to obtain high resistivity GaN films by Nakamura [107]. Following an initial heating of the (0001) Al2O3 substrate to 1323 K in H2, a 10–20 nm/GaN buffer layer was deposited between 723 and 873 using trimethylgallium and NH3 as the Ga and N sources, respectively. The temperature of this assembly was then elevated to 1273–1300 K to deposit the 4μm film. The surface was mirror-like over a 50 nm surface. The electrical properties of the film were a strong function of the thickness of the initial buffer layer. The films on the 20 nm layer possessed room temperature carrier concentration and mobility values of 4 × 10¹⁶/cm³ and 600 cm²/Vs; for those deposited on 120 nm layers, these values became 2 × 10¹⁷ and 380 cm²/V s, respectively, or essentially those room temperature values reported by Amano et al. [100] and Akasaki et al. [101,102], for GaN films deposited on an ∼50 nm AIN buffer layer.

Nearly all researchers who have grown GaN by CVD and other processes have also investigated in situ doping with Zn and Mg. Both of these dopants as well as Cd, Be, Ge and Si compensate for the n-type character of GaN, but until recently, none had been used successfully in the making of p-type GaN. However, electrical measurements following cathodoluminescence studies by Amano et al. [105] showed that low-energy electron-beam irradiation of GaN films (LEEBI) produced distinct p-type conduction. The LEEBI exposure not only produced compensation for the n-type character but increased the p-type conductivity and enhanced the photo-luminescence efficiency [106] of the material grown and doped by metal organic vapor phase epitaxy. The hole concentration and mobility and the resistivity at 293 K were ∼2 × 10⁶ cm–3, ∼8 cm²/Vs and ∼35 Ω cm, respectively. The best GaN was deposited at 1310 K on a 50 nm AlN buffer layer previously deposited on sapphire at 873 K. The Mg concentration was ∼10²⁰/cm³.

Nakamura and co-workers [108–110] have also employed and investigated Mg doping in 4 μm thick GaN films with and without LEEBI. Each of the films was deposited on a 25 nm thick GaN buffer layer deposited at 783 K in the manner described above [106]. The as-doped films exhibited p-type character even without the LEEBI treatment; however, the room temperature hole concentration was only 2 × 10¹⁵/cm³ with an average (over the entire sample) mobility of 9 cm²/ Vs. Following the exposure to the electron beam, the carrier concentration increased to within the range of (3–7) × 10¹⁸/cm³, but the average mobility remained at the same value [108].

In subsequent work, the Mg-doped films were only annealed in N2 up to 1273 K for 20 min at each temperature. The resistivity decreased dramatically from ∼10⁶ Ω cm at 773 K to a saturation point of 2 Ω cm at ∼973 K. The free-hole concentration and mobility after the high-temperature anneal were 3 × 10¹⁷/cm³ and 10 cm²/Vs, respectively. Considerable improvement in the photoluminescence spectra were also noted. In later research, these authors [110] showed that the introduction of H from annealing in an NH3 ambient caused the resistivity of the N2 annealed, Mg-doped samples to decrease to the as-grown value of 10⁶ Ω cm. A hydrogenation process whereby acceptor (Mg)-H neutral complexes are formed in tandem with hole compensation was proposed.

Koide et al. [111] have successfully doped GaN n-type using Si from a SiH4 source. A free electron concentration of 10¹⁶–10¹⁸ cm–3 and a resistivity of 1–10–2 Ω cm were obtained at room temperature. Blue m/i/n/n+ LEDs were fabricated using a Si-doped n+ layer. The peak wavelength was ∼485 nm and brightness was ∼50 mcd at a forward current of 10 mA.

As a result of the growth and doping studies described above, both Japanese research teams have reported the successful fabrication of a p–n junction LED which showed strong band edge emission [112,113]. Akasaki and Amano [112] used an AlGaN/GaN double heterostructure diode. A typical DC-EL spectrum at room temperature showed a blue emission peak at 423 nm due to the transition in the Mg center in the p-AlGaN layer. Ultra violet emission at 372 nm was believed to originate from band-to-band transitions in the GaN. The I–V characteristics of these junctions showed that large current densities (to 850 A/cm²) could be realized. The LEDs produced by Nakamura et al. [113] were fabricated from GaN layers containing Mg (p-type) and Si (n-type) dopants. The external quantum efficiency was as high as 0.18%. The electroluminescence peak wavelength and the full width half maximum of this peak were reported to be 430 nm and 55 nm, respectively. The forward voltage was 4 V at 20 mA.

4.3 Summary

The potential of GaN nitride thin films for applications in high-power and high-frequency electronic devices and optoelectronic devices is considerable. The lack of a suitable substrate is a problem of considerable magnitude. The ideal substrate would be GaN, but bulk monocrystalline GaN does not exist despite several attempts with different growth methods. Films grown on sapphire, as well as those deposited on β-SiC and GaAs, are also highly defective in terms of dislocations, twins and stacking faults. Moreover suitable ohmic and rectifying contacts for n- and p-type GaN do not exist. Insulating field and gate materials must be developed for GaN devices, as the native oxide of Ga is not stable at high temperatures. Etchants and etching procedures must also be developed for GaN. These challenges are frequently compounded by the presence of a shallow donor band which is possibly caused by N vacancies. The question of whether these vacancies occur (if they do) as a result of intrinsic or extrinsic (as a result of deposition) non-stoichiometry has not been answered. However, the recent advances of Akasaki and his coworkers and Nakamura et al. in the fabrication of p-type GaN and p–n junction light emitting diodes via the electron beam stimulation of the Mg dopant is very encouraging and may considerably advance the technology of this material. It is also encouraging that several routes to lower-temperature processes have been successful. Thus, the commercial reality of devices from these materials, individually, as solid solutions and/or as heterostructures may be slowly approaching reality.

Acknowledgements

The authors acknowledge the support of the Office of Naval Research under Contracts N00014-188-K-341 (diamond), N00014-82-K-0182 P005 (silicon carbide) and N00014-86-K-0686 (nitrides). Appreciation is also expressed to Dr. B.E. Williams, now of General Electric and Professor J.T. Glass, of NCSU for the many discussions regarding the growth and characterization of diamond films; Professor H.J. Kim, now of Seoul National University; Dr. J. Ryu, now of Northeastern University; Drs. C.H. Carter, J.W. Palmour, J.A. Edmond and H. Kong, now of Cree Research, Inc., Dr. K.L. More of Oak Ridge National Laboratory, and Professor S. Nutt of Brown University for the growth and characterization of the SiC films as well as the device fabrication and testing and helpful discussion; to Dr. Galina Keiner of the Naval Research Laboratory for discussions regarding SiC devices and Dr. Z. Sitar and M. Paisley for the research efforts and discussions regarding the nitride mtfrials.

References

1. Johnson, E.O. RCA Rev.. 1965; 26:163.

2. Keyes, R.W. Proc. IEEE. 1972; 60:225.

3. Bent, H.A. Second Law of Thermodynamics. New York: Oxford University Press, 1965; 325.

4. Anthony, T.R.Freer, R., eds. The Physics and Chemistry of Carbides, Nitrides and Borides; 185. Kluwer, Dordrecht, 1990:133. [NATO ASI Series E: Applied Sciences].

5. Derjaguin, B.V., Fedoseev, D.V. Sci. Amer.. 1975; 233:102.

6. Derjaguin, B.V., Fedoseev, D.V. Surf. Coat. Technol.. 1977; 26:133.

7. Spitsyn, B.V., Bouilov, L.L., Derjaguin, B.V. J. Cryst. Growth. 1981; 52:L183.

8. Varnin, V.P., Teremetskaya, I.G., Fedoseev, D.V., Derjaguin, B.V. Sov. Phys. Dokl.. 1984; 29:419.

9. Matsumoto, S., Sato, Y., Kamo, K., Setaka, N. Jpn. J. Appl. Phys.. 1982; 21:L183.

10. Matsumoto, S., Sato, Y., Kamo, K., Setaka, N. J. Mater. Sci.. 1982; 17:3106.

11. Kamo, K., Sato, Y., Matsumoto, S., Setaka, N. J. Cryst. Growth. 1983; 62:642.

12. Matsumoto, S., Matsui, Y. J. Mater. Sci.. 1983; 18:1785.

13. Matsumoto, S. J. Mater. Sci. Lett.. 1985; 4:600.

14. Sawabe, A., Inuzuka, T. Appl. Phys. Lett.. 1985; 46:146.

15. Sawabe, A., Inuzuka, T. Thin Solid Films. 1986; 137:89.

16. Zhu, W., Stoner, B.R., Williams, B.E., Glass, J.T. Proc. IEEE. 1991; 79:621.

17. Bachman, P.K., Messier, R. Chem. Eng. News. 1989; 67:24.

18. Frenklach, M. J. Appl. Phys.. 1989; 65:5142.

19. Markov, I., Kaischew, R. Kristall. Tech.. 1976; 11:685.

20. Bauer, E., Poppa, H. Thin Solid Films. 1976; 12:167.

21. Volmer, M., Weber, A. Z. Phys. Chem.. 1926; 119:277.

22. Frank, F.C., van der Merwe, J.H. Proc. Roy. Soc. A. 1949; 198:205.

23. Stronski, I.N., Krastanov, L. Sitzungsber. Akad., Wiss. Wien, Math. – Naturwiss., K1 IIb. 1938; 146:797.

24. Lurie, P.G., Wilson, J.M. Surf. Sci.. 1977; 65:467.

25. Williams, B.E., Glass, J.T. J. Mater. Res.. 1990; 5:801.

26. Wada, N., Gaczi, P.J., Solin, S.A. J. Non-Cryst. Solids. 1980; 35:543.

27. Nemanich, R.J., Glass, J.T., Lucovsky, G., Shroder, R.E. J. Vac. Sci. Technol. A. 1988; 6:1783.

28. Knight, D.S., White, W.B. J. Mater. Res.. 1989; 4:385.

29. Gildenblat, G. Sh., Grot, S.A., Badzian, A. Proc. IEEE. 1991; 79:647.

30. L.S. Pan, P. Pianetta, D.R. Kania and K.V. Ravi, presented at the 1989 SDIO Diamond Workshop, Washington, DC.

31. Fujimori, N., Imaj, T., Doi, A. Vacuum. 1986; 36:99.

32. K. Nishimura, K. Das, J. Glass, K. Kumagai, K. Miyata, K. Kobashi and Y. Kawate, presented at the 1989 SDIO Diamond Workshop, Washington, DC.

33. Grot, S.A., Gennady, Sh., Gildenblat, Hatfield, C.W., Wronski, C.R., Badzian, A.R., Badzian, T., Messiers, Russell. IEEE Electron Device Lett.. 1990; 11:100.

34. Gildenblat, G. Sh., Grot, S.A., Hatfield, C.W., Wronski, C.R., Badzian, A.R., Badzian, T., Messier, R.Glass, J.T., Messier, R.F., Fujimori, N., eds. Electrical Properties of Homoepitaxial Diamond Films; 162. Materials Research Society, Pittsburgh, PA, 1990:350–356. [Mater. Res. Soc. Symp. Proc.].

35. Fujimori, N., Imai, T., Nakahata, H., Shiomi, H., Nishibayashi, Y.Glass, J.T., Messier, R.F., Fujimori, N., eds. Diamond, Boron Nitride, Silicon Carbide and Related Bandgap Semiconductors; 162. Materials Research Society, Pittsburgh, PA, 1990:56–62. [Mater. Res. Soc. Symp. Proc.].

36. Shiomi, H., Nishibayashi, Y., Fujimori, N. Jpn. J. Appl. Phys.. 1989; 28:L2153.

37. Gildenblat, G. Sh., Grot, S.A., Wronski, C.R., Badzian, A.R., Badzian, T., Messier, R. Appl. Phys. Lett.. 1988; 15:586.

38. Gildenblat, G. Sh., Grot, S.A., Wronski, C.R., Hicks, M.C., Badzian, A.R., Badzian, T., Messier, R. in: IEDM. 1988; 88:626.

39. Hicks, M.C., Wronski, C.R., Grot, S.A., Gildenblat, G. Sh., Badzian, A.R., Badzian, T., Messier, R. J. Appl. Phys.. 1989; 65:2139.

40. Jepps, N.W., Page, T.F.Krishna P., ed. Progress in Crystal Growth and Characterization, 7 – Crystal Growth and Characterization of Polytope Structures. Pergamon Press: New York, 1983:259–307.

41. von Muench, W., Pettenpaul, E. J. Appl. Phys.. 1977; 48:4823.

42. Das, P., Ferry, D.K. Solid-State Electron. 1976; 19:851.

43. von Muench, W., Pfaffender, I. J. Appl. Phys.. 1977; 48:4831.

44. Bergemeister, E.A., von Muench, W., Pettenpaul, E. J. Appl. Phys.. 1979; 50:5790.

45. Nishino, S., Hazuki, Y., Matsunami, H., Tanaka, T. J. Electrochem. soc.. 1980; 127:2674.

46. Nishino, S., Powell, J.A., Will, H.A. Appl. Phys. Lett.. 1983; 42:460.

47. Davis, R.F., Sitar, Z., Williams, B.E., Kong, H.S., Kim, H.J., Palmour, J.W., Edmond, J.A., Ryu, J., Glass, J.T., Carter, C.H., Jr. Mater. Sci. Eng. B. 1988; 1:77.

48. G.L. Harris and C.Y-W. Yang (eds.), Amorphous and Crystalline Silicon Carbide, Springer Proceedings in Physics, Vol. 34 (Springer-Verlag, New York, 1988).

49. Aselage, T., Emin, E., Woods, C., eds. Novel Refractory Semiconductors, 97. Pittsburgh, PA: Materials Research Society, 1987. [Mater. Res. Soc. Symp. Proc.].

50. Freer, R.F., eds. The Physics and Chemistry of Carbides, Nitrides and Borides, 185. Dordrecht: Kluwer, 1990. [NATO ASI Series E; Applied Sciences].

51. Glass, J.T., Messier, R.F., Fujimori, N., eds. Diamond, Boron Nitride, Silicon Carbide and Related Wide Bandgap Semiconductors, 162. Pittsburgh, PA: Materials Research Society, 1990. [Mater. Res. Soc. Sym. Proc.].

52. Kim, H.J., Davis, R.F., Cox, X.B., Linton, R. J. Electrochem soc.. 1987; 134:2269.

53. Nutt, S.R., Smith, D.J., Kim, H.J., Davis, R.F. Appl. Phys. Lett.. 1987; 50:203.

54. Glass, J.T., Wang, Y.C., Kong, H.S., Davis, R.F.Choi, H.K., Hall, R., Ishiwara, H., Nemanich, R.J., eds. Heteroepitaxy on Silicon: Fundamentals, Structures, and Devices; 116. Materials Research Society, Pittsburgh, PA, 1988:337–349. [Mater. Res. Soc. Proc.].

55. Shibahara, K., Nishino, S., Matsunami, H. J. Cryst. Growth. 1986; 78:538.

56. Pirouz, P., Chorey, C.M., Powell, J.A. Appl. Phys. Lett.. 1987; 50:221.

57. Chorey, C.M., Pirouz, P., Powell, J.A., Mitchell, T.E.Green M.L., Baglin J.E.E., Chin G.Y., Deckman H.W., Mayo W., Narainham D., eds. Semiconductor-Based Heterostructures: Interfacial Structure and Stability. The Metallurgical Society: Philadelphia, PA, 1987:115–126.

58. Powell, J.A., Matus, L.G., Kucmarski, M.A., Chorey, C.M., Cheng, T.T., Pirouz, P. Appl. Phys. Lett.. 1987; 51:823.

59. Kong, H.S., Wang, Y.C., Glass, J.T., Davis, R.F. J. Mater. Res.. 1988; 3:521.

60. Fuyuki, T., Nakayama, M., Yoshinobu, T., Shiomi, H., Matsunami, H. J. Cryst. Growth. 1989; 95:461.

61. Sugii, T., Aoyama, T., Ito, T. J. Electrochem. Soc.. 1990; 137:989.

62. Nakamatsu, H., Kawai, S.Aselage, T., Emin, D., Woods, C., eds. Novel Refractory Semiconductors; 97. Materials Research Society, Pittsburgh, PA, 1987:189–194. [Mater. Res. Soc. Symp. Proc.].

63. Nakamatsu, H., Hirata, Ki., Kawai, S.Ehrlich, D.J., Higashi, G.S., eds. Laser and Particle-beam Chemical Processing for Microelectronics; 101. Materials Research Society, Pittsburgh, PA, 1988:397–402. [Mater. Res. Soc. Symp. Proc.].

64. Salun, V.S., Knaev, I.F., Serbinov, L.A., Reshetnikov, I.E., Ormont, A.B., Byvalin, D.A., Ryabova, L.A., Malinovskii, V.K. Sov. Tech. Phys. Lett.. 1987; 13:344.

65. Kong, H.S., Glass, J.T., Davis, R.F. Appl. Phys. Lett.. 1986; 49:1074.

66. Kong, H.S., Glass, J.T., Davis, R.F. J. Mater. Res.. 1989; 4:204.

67. Kong, H.S., Jiang, B.L., Glass, J.T., Rozgonyi, G.A., Moore, K.L. J. Appl. Phys.. 1988; 63:2645.

68. Powell, J.A., Larkin, D.J., Matus, L.G., Choyke, W.J., Bradshaw, J.L., Henderson, L., Yoganathan, M., Yang, J., Pirouz, P. Appl. Phys. Lett.. 1990; 56:1353.

69. Kong, H.S., Kim, H.J., Edmond, J.A., Palmour, J.W., Ryu, J., Carter, C.H., Jr., Glass, J.T., Davis, R.F.Aselage, T., Emin, D., Woods, C., eds. Novel Refractory Semiconductors; 97. Materials Research Society, Pittsburgh, PA, 1987:233–245. Mater. Res. Soc. Symp. Proc.

70. Kong, H.S., Glass, J.T., Davis, R.F. J. Appl. Phys.. 1988; 64:2672.

71. Shibahara, K., Kuroda, N., Nishino, S., Matsunami, H. Jpn. J. Appl. Phys.. 1987; 26:L1815.

72. N. Kuroda, K. Shibahara, W. Yoo, S. Nishino and H. Matsunami, Extended abstracts of the 19th Conf. on Solid State Devices and Materials, Tokyo, Japan (1989) p. 227.

73. Powell, J.A., Larkin, D.J., Matus, L.G., Choyke, W.J., Bradshaw, J.L., Henderson, L., Yoganathan, M., Yang, J., Pirouz, P. Appl. Phys. Lett.. 1990; 56:1442.

74. Campbell, R.B., Chang, H.C.Willardson, R.K., Beer, A.C., eds. Semiconductors and Semimetals; 7B. Academic Press, San Diego, CA, 1978:625–683.

75. Furukawa, K., Uemoto, A., Shigota, M., Suzuki, A., Nakajima, S. Appl. Phys. Lett.. 1986; 48:1536.

76. A. Suzuki, A. Uemoto, M. Shigota, K. Furukawa and S. Nakajima, Extended abstracts of the 18th Int. Conf. on Solid State Devices and Materials (1986) pp. 101–104.

77. Davis, R.F., Kelner, G., Shur, M., Palmour, J.W., Edmond, J.A. Proc. IEEE. 1991; 79:677.

78. Shibahara, K., Saito, T., Nishino, S., Matsunami, H. IEEE Electron Device Lett.. 1986; 7:692.

79. Fuma, H., Misura, A., Tadano, H., Sugiyama, S., Takigawa, M. Jpn. J. Appl. Phys. Lett.. 1988; 27:2143.

80. Palmour, J.W., Kong, H.S., Davis, R.F. J. Appl. Phys.. 1988; 64:2168.

81. Ziegler, G., Lanig, P., Theis, D., Weyrich, C. IEEE Trans. Electron Devices. 1983; 30:30.

82. Suzuki, A., Ikeda, M., Nagao, N., Matsunami, H., Tanaka, T. J. Appl. Phys.. 1976; 47:4546.

83. Matsunami, H., Ikeda, M., Suzuki, A., Tanaka, T. IEEE Trans. Electron Devices. 1977; 24:958.

84. Ikeda, M., Hayakawa, T., Yamagiwa, S., Matsunami, H., Tanaka, T. J. Appl. Phys.. 1979; 50:8215.

85. Matsunami, H. Optoelectron. Devices Technol.. 1987; 2:29.

86. Dmitriev, V.A., Ivanov, P.A., Morozenko, Ya.V., Popov, I.V., Chelnokov, V.E. Sov. Tech. Phys. Lett.. 1985; 11:101.

87. Edmond, J.A., Kong, H.-Sh., Carter, C.H., Jr. Physica B. 1993; 185:453.

88. Dmitriev, V.A., Morozenko, Ya.V., Popov, I.V., Suvorov, A.V., Syrkin, A.L., Chelnokov, V.E. Sov. Tech. Phys. Lett.. 1986; 12:221.

89. Verenchikova, R.G., Sankin, V.I. Sov. Tech. Phys. Lett.. 1988; 14:756.

90. Glasow, P., Ziegler, G., Suttrop, W., Pensl, G., Helbig, R.Kukkonen, M., eds. Optoelectronic Technologies for Remote Sensing from Space; 868. SPIE, Bellingham, WA, 1987:40–45. [SPIE].

91. Davis, R.F., Sitar, Z., Williams, B.E., Kong, H.S., Kim, H.J., Palmour, J.W., Edmond, J.A., Ryu, J., Glass, J.T., Carter, C.H., Jr. Mater. Sci. Eng. B. 1988; 1:77.

92. Pancove, J.I.Glass, J.T., Messier, R.F., Fujimori, N., eds. Diamond, Boron Nitride, Silicon Carbide and Related Bandgap Semiconductors; 162. Materials Research Society, Pittsburgh, PA, 1990:515–524. [Materials Research Soc. Symp. Proc.].

93. Mizuta, M., Fujieda, S., Matsumoto, Y., Kawamara, T. Jpn. J. Appl. Phys.. 1986; 25:L945.

94. Paisley, M.J., Sitar, Z., Posthill, J.B., Davis, R.F. J. Vac. Sci. Technol. A. 1989; 7:701.

95. R.C. Powell, G.A. Tomasch, Y.-W. Kim, J.A. Thornton and J.E. Greene, in: J.T. Glass, R.F. Messier and N. Fujimori (eds.), Mater. Res. Soc. Symp. Proc., Vol. 162 (Materials Research Society, Pittsburgh, PA, 1990) pp. 525–530.

96. H. Morkoc, private communication.

97. Lei, T., Moustakas, T.D., Graham, R.J., He, Y., Berkowitz, S.J. J. Appl. Phys.. 1992; 71:1745.

98. Yoshida, S., Misawa, S., Gonda, S. Appl. Phys. Lett.. 1983; 42:427.

99. Yoshida, S., Misawa, S., Gonda, S. J. Vac. Sci. Technol. B. 1983; 1:250.

100. Amano, H., Asaki, T., Akasaki, I., Toyoda, Y. Appl. Phys. Lett.. 1986; 48:353.

101. Akasaki, I., Amano, H., Koide, Y., Hiramatsu, K., Sawaki, N. J. Cryst. Growth. 1989; 98:209.

102. Amano, H., Asaki, T., Akasaki, I. Jpn. J. Appl. Phys. Lett.. 1990; 29:205.

103. Khan, M.A., Kuznia, J.N., Van Hove, J.M., Olsons, D.T., Krishnankutty, S., Kolbas, R.M. Appl. Phys. Lett.. 1991; 58:985.

104. Khan, M.A., Olson, D.T., van Hove, J.M., Kuznia, J.N. Appl. Phys. Lett.. 1991; 58:1515.

105. Amano, H., Akasaki, I., Kozawa, T., Hiramatsu, K., Sawaki, N., Ikeda, K., Ishii, Y. J. Lumin.. 1988; 40&41:121.

106. Amano, H., Kito, M., Hiramatsu, K., Akasaki, I. Jpn. J. Appl. Phys. Lett.. 1989; 28:L2112.

107. Nakamura, S. Jpn. J. Appl. Phys. Lett.. 1991; 30:L1705.

108. Nakamura, S., Senoh, M., Mukai, T. Jpn. J. Appl. Phys.. 1991; 30:L1708.

109. Nakamura, S., Mukai, T., Senoh, M., Iwasa, N. Jpn. J. Appl. Phys.. 1992; 31:L139.

110. Nakamura, S., Iwasa, N., Senoh, M., Mukai, T. Jpn. J. Appl. Phys.. 1992; 31:436.

111. Koide, N., Kato, H., Sassa, M., Yamasaki, S., Manabe, K., Hashimoto, M., Amano, H., Hiramatsu, K., Akasaki, I. J. Cryst. Growth. 1991; 115:639.

112. I. Akasaki and H. Amano, in: T.D. Moustakas (ed.), Mater. Res. Soc. Symp. Proc., Vol. 242 (Materials Research Society, Pittsburgh, PA, 1992) pp. 383–389.

113. Nakamura, S., Mukai, T., Senoh, M. Jpn. J. Appl. Phys.. 1991; 30:L1998.

Optical physics and laser devices in II–VI quantum confined heterostructures

Arto V. Nurmikko ,     Division of Engineering and Department of Physics, Brown University, Providence, RI, USA

A.V Nurmikko, Division of Engineering and Department of Physics, Brown University, Providence, RI 02912, USA.

Robert L. Gunshor,     School of Electrical Engineering, Purdue University, West Lafayette, IN, USA

We review some of the recent advances in wide-gap II–VI heterostructures, with emphasis on device physics and engineering of quantum confinement in ZnSe-based multilayer structures for blue-green diode lasers and LEDs. An important element in these quantum well configurations are quasi-two-dimensional exciton effects which dominate absorption but are also of relevance in both\spontaneous and stimulated emission processes.

1 Introduction

There is considerable contemporary excitement in the field of wide-gap II–VI compound semiconductors. A rapid rate of progress has occurred in heteroepitaxial techniques, leading to profound improvement in material quality and purity, as well as understanding of the physics of electronic states in the new II–VI quantum well (QW) and superlattice systems [1]. In another key development, useful p-type doping of ZnSe has recently been realized [2] and, while still not fully optimized, has nonetheless led to p–n heterojunction proof-of-concept devices for light emitters in the blue and the green. The culmination of this work was the achievement in 1991 of diode laser operation in the (Zn, Cd)Se/ZnSe [3,4] and the (Zn, Cd)Se/Zn(S, Se) QWs [5]. Details of the pioneering work at 3M Laboratories can be found in the article by C.T. Walker in this volume; we focus here on the work by the Brown-Purdue team.

It is well known that the large ionic component in the chemical bonding of, e.g., ZnSe enters fundamentally into defining its optical and other physical properties. One positive consequence of this is the strongly excitonic character which literally ‘colors’ the photoresponse of QW structures so that both absorptive and emissive processes contain a substantial excitonic component up to device temperatures in these quasi-two-dimensional (2D) systems. On the other hand, difficulties of doping and fabrication contacts represent the reverse side of the coin. Furthermore, the mechanical and thermal properties are other aspects that need to be carefully examined in device design, together with the question of material stability, e.g. in pursuing room temperature diode lasers.

2 Electronic confinement and two-dimensional electron-hole systems in wide-gap II–VI QWs: quasi-two-dimensional exciton effects in (Zn, Cd)Se QWs

The study of exciton effects in wide-gap II–VIs is one of the most direct ways to verify good type I quantum well confinement in these semiconductors. For example in bulk ZnSe, excitons are only observed at low temperatures (near liquid helium). This is chiefly due to the fact that the exciton–LO phonon (Fröhlich) interaction is very strong and leads to rapid