14

Table 2-5. Carrier concentration and Hall mobility for the different growth methods.

The typical range of carrier concentrations and mobilities for the different growth methods including MOVPE, PA-MOVPE, HVPE, MBE, MEE, and sputtering was discussed in detail and summarized in Table 2-5.

2.1.4 Optical Properties of InN

Until 2001, the measured bandgap of 1.89 eV has been commonly accepted for InN [Tan86a]. However, a few groups recently showed by PL measurements that the band gap energy of InN is in between 0.65 and 0.90 eV, [Dav02a, Dav02b, Dav02c, Wu02, Tat02, Hor02, Sai02, Miy02] which is much smaller than 1.89 eV.

Evidence of a narrower band gap for InN was reported in 2001. Inushima et al. insisted that the fundamental absorption edge of MBE grown InN layer lies around 1.1 eV, which is much lower than the previously reported values [Inu01]. Davydov et al. reported a band gap value of 0.9 eV for high quality MBE grown InN, studied by means of optical absorption, PL, photoluminescence excitation (PLE) spectroscopy, as well as by ab initio calculation [Dav02a]. Figure 2-4 shows photoluminescence spectra for MBE grown InN sample which showed that the band gap of InN was much less than the previously reported value (around 1.9 eV) [Dav02a]. They further studied in detail with different high quality hexagonal InN films grown by different epitaxy methods. Analysis of optical absorption, PL, PLE, and photoreflectivity data obtained on single crystalline hexagonal

15

InN film leads to the conclusion that the true band gap of InN is Eg ~ 0.7 eV [Dav02b, Dav02c].

The larger band gap (~1.89 eV) cited in the literature may be due to the formation of oxynitrides, which have much larger band gaps than that of InN. As can be seen in Fig. 2.5, the energy gap data less than 1 eV were obtained for single crystalline InN film with a relatively low carrier concentration, while the larger values were mostly for polycrystalline InN film [Bhu03a]. It should also be pointed out that the band gap obtained from epitaxial films shows a remarkable dependence of carrier concentration, which is different from the larger one obtained from polycrystalline films. Polycrystalline films show a similar band gap (~ 2 eV) in spite of the wide range variation of carrier concentration 1016-1021 cm-3.

Figure 2-4. Photoluminescence spectra for MBE grown InN.

As Motlan et al. [Mol02] reported, oxygen incorporation is one of the causes for the large band gap energy. Therefore, the larger values may be related to oxygen incorporation into grown InN because polycrystalline films can contain a high density of oxygen atoms at their grain boundaries.

16

Figure 2-5. Band gap energy for InN films as a function of carrier concentration. Davydov et al. [Dav02c] showed that the sample with band gap in the region of 1.8-

2.1 eV contained up to 20 % of oxygen, much higher than for samples with narrow band gap. It can be assumed that oxygen is responsible for a high concentration of defects. Therefore, this increase of the band gap energy can be caused by formation of oxynitrides, which have a much larger band gap than that of InN.

2.1.5. Indium Nitride (InN) andIindium Gallium Nitride (InxGa1-xN) Applications

The latest progress in improving the InN film quality indicates that the InN film almost meets the requirements for application to practical devices. Nowadays, the bandgap energy of InN is known as 0.7 eV and thus InxGa1-xN layers can be used as absorber layers in tandem solar cells where the mole fraction of indium (x) is varied from 0 to 1 which tunes the bandgap from 0.7 to 3.4 eV. This energy range covers the majority of the solar spectrum, therefore improving efficiency.

In addition to the tandem solar cells, InN can also be applied to LED and LD similar to other III-V nitride compounds. Because the reported band gap value of InN is about 0.7 eV, which is compatible with the wavelength of the optical fiber, another very

17

important potential application of InN, fabrication of high-speed LD and PD in the optical communication system, is expected.

It is expected to be a highly promising material for the fabrication of high performance high electron mobility transistor (HEMT). InN as a HEMT channel requires a larger band gap barrier to induce and confine electrons. The significant lattice mismatch between InN and GaN or AIN can result in a large piezo-electric charge, which is very advantageous for HEMT applications. The strained InxGa1-xN or InxAl1-xN is also a good choice as a barrier layer.

InxGa1-xN is a very important compound semiconductor among III-V nitride compounds because the InxGa1-xN active layer emits light by the recombination of the injected electrons and holes into this active layer. The addition of a small amount of indium into the GaN was very important in obtaining a strong band-to-band emission because GaN without the indium could not emit a strong band-to-band emission at RT. This reason is considered to be related to deep localized energy states.

Currently, InxGa1-xN is usually applied for the active layer in LEDs and LDs for this characteristic of the deep localized energy states, which can facilitate the efficiency of the band-to-band emission. For InxGa1-xN-based LDs, however, the TDs (threading dislocations) density had to be decreased to lengthen the lifetime by using the ELOG (Epitaxial Lateral Overgrowth). For InxGa1-xN-based LEDs, the lifetime of the LEDs is more than 100,000 hours in spite of the large number of dislocations. This difference in lifetime-behavior between LDs and LEDs is probably caused by the difference in the operating current density in the two devices. The operating current density of LDs is about one order higher than that of LEDs. Numerous studies have investigated the origin

18

of these defects, and their effects on the structural, optical, electronic, and morphological properties of heteroepitaxial InxGa1-xN layers [Chen06, Cho04, Jin06, Lil06].

2.2 Thermodynamic Analysis and Phase Separation in the InxGa1-xN System

Thermodynamics give the guideline for the epitaxial growth process for all techniques, including MOVPE, since epitaxial growth is simply a highly controlled phase transition. A thermodynamic understanding of epitaxy allows the determination of alloy composition as well as the solid stoichiometry.

The thermodynamics of mixing of semiconductor alloys (III/V, II/VI, and IV/IV) determines many characteristics of the growth process as well as the properties of the resultant materials. For example, Thermodynamic factors may limit the mutual solubility of the two (or more) components of an alloy. When the sizes of the constituent atoms are sufficiently different, miscibility gap exist. In addition to solid-phase immiscibility in important alloys systems such as GaInAsP and InxGa1-xN, this size difference also leads to microscopic structures far different than the random, totally disordered state normally expected for alloys. Both miscibility gaps and deviations from a random distribution of the atoms constituting the lattice affect the electrical and optical properties of semiconductor alloys in ways that are extremely important for many types of devices.

The thermodynamics of the surface must also be considered in any effort to understand the growth processes as well as the characteristics of the materials produced epitaxially.

The basic goal of thermodynamics, as applied to epitaxy, is to define the relationship between the compositions of the various phases in an equilibrium system at constant temperature and pressure.