domingo, 30 de mayo de 2010

Growth and applications of Group III- nitrides

Growth and applications of
Group III-nitrides


Introduction
Group III-nitrides have been considered a promising system for semiconductor devices applications since 1970,especially for the development of blue- and UV-lightemitting diodes. The III–V nitrides, aluminium nitride (AlN), gallium nitride (GaN) and indium nitride (InN), are candidate materials for optoelectrical applications at such photon energies, because they form a continuous alloy system (InGaN, InAlN, and AlGaN) whose direct optical bandgaps for the hexagonal wurtzite phase range from 1.9 eV for _-InN and 3.4 eV for _-GaN to 6.2 eV for _-AlN. The cubic modifications have bandgaps in the range from 1.7 eV for _-InN and 3.2 eV for _-GaN to 4.9 eV for _-AlN (figures 1 and 2) [1–6]. Other advantageous properties include high mechanical and thermal stability, large piezoelectric constants and the possibility of passivation by forming thin layers of Ga2O3 or Al2O3 with bandgaps of approximately 4.2 eV and 9 eV. The spontaneous and piezoelectric polarization in the wurtzite materials) and the high electron drift velocities (2 _ 105 m s−1 [7]) of GaN can be used to fabricate highpower transistors based on AlGaN/GaN heterostructures. In addition, AlN is an important material with a variety of applications such as passive barrier layers, high-frequency acoustic wave devices, high-temperature windows, and dielectric optical enhancement layers in magneto-optic multilayer structures Very informative reviews of the growth techniques and structural, optical and electrical properties of Group IIInitrides and their alloys have been presented by Strite et al .A good overview of applications of Group III-nitride based heterostructures for UV emitters and high-temperature, high-power electronic devices is provided in [12] and [13]. This review focuses on the development of the different growth techniques successfully applied to the deposition of Group III-nitride epitaxial films and heterostructures, such as chemical transport and metalorganic chemical vapour deposition (MOCVD), sputtering and molecular beam epitaxy (MBE). The quality of state-of-the-art material and its application for optical and electronic devices are discussed in detail in order to point out possible limitations, promising developments and future trends. The first systematic effort to grow InN, GaN and AlN by chemical vapour deposition or sputtering processes took place in the 1970s in order to characterize the optical and structural properties of thin films. At that time, neither metalorganic precursors containing In or Al with electronic grade purity, plasma sources for nitrogen radicals compatible with MBE systems, nor substrate material with reasonably good thermal and lattice matches to the nitrides were available. The InN and GaN material had large concentrations of free electrons, presumed to result from oxygen impurities and intrinsic defects, and the structural quality of the AlN films was not good enough for optical or electronic applications. Primarily, the development of MOCVD and plasma-induced molecular beam epitaxy (PIMBE) over the last eight years has led to a number of recent advances and important improvements in structural properties.
Crystal structure, polarity and polarization of InN, GaN and AlN
In contrast to cubic III–V semiconductors like GaAs and InP with the zincblende structure, the thermodynamically stable phase of InN, GaN and AlN, is the hexagonal wurtzite structure (_-phase). Beside the _-phase, a metastable _-phase with zincblende structure exists and a cubic high-pressure modification with NiAs structure was observed for pressures above 25 kbar in the case of AlN. Because the _- and _-phases of Group III-nitrides only differ in the stacking sequence of nitrogen and metal atoms (polytypes), the coexistence of hexagonal and cubic phases is possible in epitaxial layers, for example due to stacking faults. The hexagonal crystal structure of Group III-nitrides can be described by the edge length a0 of the basal hexagon, the height c0 of the hexagonal prism and an internal parameter u defined as the anion–cation bond length along the (0001) axis. Because of the different cations and ionic radii (Al3C: 0.39 A° , Ga3C: 0.47 A° In3C: 0.79 °A[15]), InN, GaN and AlN have different lattice constants, bandgaps and binding energies as shown in table Both wurtzite and zincblende structures have polar axes (lack of inversion symmetry). In particular, the bonds in the h0001i direction for wurtzite and h111i direction for zincblende are all faced by nitrogen in the same direction and by the cation in the opposite direction. Both bulk and surface properties can depend significantly on whether the surface is faced by nitrogen or metal atoms. The most common growth direction of hexagonal GaN is normal to the f0001g basal plane, where the atoms are arranged in bilayers consisting of two closely spaced hexagonal layers, one with cations and the other with anions, so that the bilayers have polar faces. Thus, in the case of GaN a basal surface should be either Ga- or N-faced. By Gafaced we mean Ga on the top position of the f0001g bilayer, corresponding to polarity. Ga-faced does not mean Ga-terminated; termination should only be used.

Growth of Group III-nitride films and crystals
The history of the production of Group III-nitrides covers more than a century. AlN powder was first made in 1862 from liquid Al and N2 gas. The major difficulty with this direct reaction method is that the surface film of AlN on Al is very adherent and impedes further reaction. In another technique, AlN powder can be formed from Al electrodes in a direct current arc. This produces only small amounts of AlN per day and the powder usually has several per cent excess aluminium because even here the AlN skin is protective. A more useful method for making AlN is to react AlF3 powder with NH3 gas at high temperatures. The overall chemical reaction at 1000 _C is: AlF3.s/ C NH3.g/ ! AlN.s/ C 3HF.g/: (12) In order to promote the formation of AlN, it is necessary to keep the NH3 partial pressure above 1 bar and the HF gas must be continually removed. At 1 atm pressure, a minimum of 25 molecules of NH3 are needed for each AlN molecule produced. The earliest investigations of GaN powder were reported by Johnson and co-workers, who described the conversion of metallic Ga in a NH3 stream by the chemical reaction: 2Ga.l/ C 2NH3.g/ ! 2GaN.s/ C 3H2.g/: (13) They obtained a black powder by flowing ammonia over metallic gallium at 1000 _C. Ejder contained Ga in an open sintered alumina boat. The boat was placed in a silica glass tube in a tube furnace, and a mixture of nitrogen and ammonia was passed over the boat. The reaction between Ga and NH3 started below 1000 _C and the thin crust ofGaN which formed on the surface of the molten Ga then decreased the rate of further GaN formation by lowering the evaporation rate of Ga from the metal surface. By this method whiskers, needles and prisms (sizes up to 500 _m) were formed on the edges of the boat and on the walls of the surrounding silica glass tube. These early growth processes and investigations resulted in AlN and GaN powders and very small crystals which were used to determine basic physical properties like crystal structure, lattice constants and optical properties. These results enabled the identification of substrate materials suitable for the heteroepitaxy of Group III-nitrides.

Substrates for heteroepitaxy
One particular difficulty in the growth of thin films is the unavailability of sufficiently large (>1 cm) single crystals for use as substrates for homoepitaxial growth. Thus up to now, heteroepitaxial growth is a practical necessity and the choice of substrate is critical. This problem is well recognized and there have been a number of studies on the effects of the substrate on the structural, electrical and morphological properties of thin films of these compound semiconductors [76–88]. Possible substrate materials with low thermal expansion and lattice mismatch for vapour phase epitaxy (VPE) and MOCVD are limited to those unaffected by high concentrations of ammonia and hydrogen at temperatures in excess of 1000 _C. This limits the use of Si, GaAs and GaP, if no low-temperature buffer layer can be grown as a first step of device fabrication. Even for PIMBE, in which the growth temperatures are about 250 _C lower than for VPE and MOCVD, the substrate surfaces have to be stable under the influence of nitrogen radicals at 800 _C. For device production processes, the substrate of choice has to be available in a minimum size of two inches, with atomically flat surfaces, and in large quantities at acceptable prices. Under the presuppositions mentioned above, sapphire and silicon carbide are the most popular substrate materials. Although sapphire (_-Al2O3) has a rhombohedral structure, it can be described by a hexagonal cell that is larger than the basic rhombohedral unit cell. To date, four orientations of sapphire have been used as substrates: (10N10), (0001), (2N1 N10) and (11N20). The lattice mismatch between GaN and (0001) sapphire is 13.9%. This table also includes the thermal expansion mismatch which is just as important for epitaxial growth. For various commercially available sapphire orientations, the relationship of orientation, lattice mismatch and crystallographic symmetry with a GaN epitaxial film are given. From the viewpoints of lattice mismatch and crystal symmetry, (10N10) sapphire (m-plane) seems the most suitable for GaN growth. However, the c-axis of a GaN film grown on a (10N10) sapphire substrate has a nonzero inclination with respect to the c-axis of the sapphire substrate. This means that twins may be generated. This is a major disadvantage of a (10N10) plane compared to a (0001) plane. Because of the high in-plane lattice mismatch (up to
−29%) between the (0001) oriented films of InN, GaN and AlN and the (0001) sapphire, it seems surprising that epitaxial growth is possible. Transmission electron microscopy (TEM) was used to investigate the epitaxial growth of hexagonal GaN by MBE on the (0001) basal plane of Al2O3. The in-plane orientation obtained between layer and substrate gives a high lattice misfit of −13:9%, which induces a stress relaxation process directly in the interface region. High resolution TEM (HRTEM) images reveal f11N20gSapphire lattice fringes terminating at the interface between GaN and Al2O3(0001) [79]. Because the Burgers vectors of misfit dislocations are parallel to the (0001) interface plane, gliding is limited to the (0001) planes and therefore these misfit dislocations are confined at the interface. This interfacial relaxation process is very effective and the extension of dislocations into the GaN layer is suppressed. In the case of pseudomorphic growth the number of f10N10gGaN lattice fringes is expected to be the same as the number of f11N20gSapphire lattice fringes. The plastic relaxation of GaN by the formation of misfit dislocations leads to a termination of f11N20gSapphire lattice fringes at the interface. The quantitative evaluation of Fourier-filtered HRTEM images reveals an average of 8:3 _ 0:7 Al2O3 lattice spacings between two terminating f11N20gSapphire fringes. By forming a coincidence lattice between GaN and Al2O3, a large portion (−11:8%) of the high mismatch is compensated by dislocations confined at the interface. The relaxation process allows an epitaxial growth of GaN on (0001)Al2O3 with a density of non-confined dislocations of about 1010 cm−2 in the GaN. The non-confined dislocations are caused by the residual strain reduction from −2:1% measured near the interface to −0:2% obtained at the surface for a 1 _m thick GaN film. The lattice constant a.GaN/ D 3:1892 °A in the surface region reveals a residual strain of −0:2% in the GaN film after growth. With thermal expansion coefficients perpendicular to the c-axis of 5:6_10−6 K−1 for GaN and 7:3_10−6 K−1 for Al2O3 a lattice misfit of −13:75% can be calculated for a growth temperature of 800 _C. Cooling of the sample from growth temperature to room temperature causes a film strain of −0:12%, which is similar to the measured surface strain. Consequently it can be concluded that at the growth temperature, the film is almost completely relaxed and the residual strain close to the surface is only a thermal effect.




Bárbara Scarlett Betancourt Morales
CAF

No hay comentarios:

Publicar un comentario