Introduction

InGaN (IGN) has drawn a great deal of attention among the group-III ternary nitride materials due to its immense applications especially in the field of optoelectronics [1–3]. IGN can be considered as a solid solution of InN and GaN. The direct band gaps of InN and GaN are 0.7 and 3.4 eV, respectively. Hence, IGN thin films are promising materials for photovoltaics applications, as its band gap can be tuned in the entire range of solar spectrum [4]. Similar to its constituents, IGN crystallizes predominantly in the hexagonal wurtzite structure in the entire composition range. The basic properties of IGN materials are mostly evaluated on the basis of its binary constituents: InN and GaN. InN, being difficult to grow high quality epilayers, is the least understood material among the group-III nitride semiconductors. Many of its fundamental properties like band gap, electron effective mass, decomposition temperature and defect levels reported in the literature are highly debatable. Recently, the accurate band gap of MBE-grown epitaxial InN film has been shown to be 0.7 eV from photoluminescence (PL) measurements [5]. But, there are reports which still support the earlier result of 1.9 eV [6, 7]. In the literature, several reasons like Burstein–Moss (B–M) shift [8], Mie resonance [9], stoichiometry [10], oxygen impurity [11], defect levels, polycrystallinity, etc., are cited to explain the discrepancy found in the InN band gap. In contrast, almost all the fundamental properties of GaN are well established. Moreover, the structural, optical and electrical properties of IGN films depend mostly on the alloy composition. Thus, the study of the basic properties of IGN alloy system is very difficult and complex.

It is difficult to grow good crystalline quality IGN films in the whole composition range because of several problems. There exists a solid phase miscibility gap in IGN alloy system due to the difference in the interatomic spacing between its constituents, i.e. InN and GaN [12]. Low indium incorporation has also been observed in these alloy films due to the relatively high vapour pressure of InN compared to GaN [13, 14]. Moreover, due to the difference in the formation enthalpies of GaN and InN, surface segregation of indium is expected at the growth front. IGN films can be grown by several techniques like e-beam plasma technique [15], sputtering [16], pulsed laser deposition [17], metalorganic chemical vapour deposition (MOCVD) [18], molecular beam epitaxy (MBE) [19], hydride vapour phase epitaxy [20], etc. In the present scenario, MOCVD and MBE are considered as the leading growth techniques to grow high quality IGN epilayers. But, both these techniques employ high temperatures (>500 °C) which limit the growth of good quality indium-rich IGN films. MBE is not a commercially viable technique due to cost factor and low throughput. Though MOCVD is a commercially viable method, use of hazardous gases and high temperatures limit its applications. In order to have compositional tunability (preferably eliminating toxic reactive environments) and to expand the range of applications, a low temperature growth technique is highly desirable.

Recently, our laboratory has developed a room temperature growth technique known as modified activated reactive evaporation (MARE). By this technique, we have grown structurally good quality InN and GaN thin films [21, 22], InN films on flexible substrates like polyimide and polycarbonate [23] and also reported room temperature growth of IGN films using a mixed source of indium and gallium [24]. It may be pointed out that due to the difference in the vapour pressures of indium and gallium, it is always difficult to control the In/Ga ratio from a single evaporation source. A better control over In/Ga ratio may be achieved in IGN alloy films by employing separate sources [25].

In the present study, IGN films in the entire composition range have been grown by MARE technique employing separate In and Ga sources. The dependence of the structural, optical and electrical properties of these thin films on composition has been reported.

Experimental details

The details of the separate source MARE technique are discussed in Ref. [25]. IGN thin films have been prepared on cleaned borosilicate glass (BSG) substrates by MARE technique using a triangular geometry for the evaporation sources and substrate (Fig. 1). The substrate is placed on the radio-frequency (RF) magnetron cathode. The films are grown at room temperature (without any intentional substrate heating). The In/Ga ratio is controlled by capturing the optical emission spectra (OES) through the optical fibres (fitted to the view-port of the vacuum chamber) using Ocean Optics HR 2000 spectrometer. OES records the characteristic emission lines of the plasma. The emission spectra consist of the excited indium, gallium and nitrogen species (Fig. 2). The thickness of IGN films is in the range 600–800 nm as determined by the reflectance spectra of Filmetrics F20 spectrophotometer. The phase and crystallinity of the IGN films are studied using Phillips X’Pert Pro X-ray diffractometer operated at 40 kV and 30 mA using Cu Kα source (λ = 0.1542 nm). The Raman spectra are recorded using a Jobin–Yvon HR-800 Raman spectrometer equipped with an 18 mW Ar+ laser source (488 nm). The energy dispersive X-ray analysis (EDX) is performed using the EDAX detector attached to a field-emission scanning electron microscope (FE-SEM) FEI Quanta 200. The surface morphology and roughness of the films are studied using Digital Instruments Nanoscope IV atomic force microscope (AFM) in contact mode. The transmittance spectra in the wavelength range (300–2000 nm) are recorded using a Jasco V-570 UV–vis-NIR spectrophotometer. The room temperature photoluminescence (PL) spectra for indium-rich IGN films are obtained using 488 nm Ar+ laser excitation source whereas a He–Cd laser with excitation wavelength of 325 nm is used for gallium-rich films. The resistivity, carrier concentration and mobility of IGN films are measured in the van der Pauw geometry using a Lake Shore (7600 series) Hall effect system.

Fig. 1
figure 1

Schematics of IGN thin film growth by MARE using separate sources

Full size image
Fig. 2
figure 2

Optical emission spectra for the growth of x = 0.77 and 0.30 IGN films

Full size image

Results and discussion

The representative XRD patterns for IGN films are plotted in Fig. 3a. All the films are found to crystallize in the hexagonal wurtzite structure with preferential growth along the c-axis. Weak shoulder peaks corresponding to \( (10\mathop {\bar{1}} 0) \) and \( (10\mathop {\bar{1}} 1) \) reflections are observed with increase in gallium concentration and are found to be dominating for gallium-rich films (x < 0.10). For some of the films, \( (10\mathop {\bar{1}} 2) \) and \( (10\mathop {\bar{1}} 3) \) reflections are also observed. For more clarity, the slow scans of XRD patterns for all the IGN films near the (0002) Bragg reflections are shown in Fig. 3b. In general, crystallites are expected to grow in all the directions. The (0002) plane in wurtzite system possesses the lowest surface energy [26]. Grains with lower surface energy become larger as the films grow and the growth develops along one crystallographic direction. For pure GaN, at lower power, it is observed that the crystallites are oriented in all the directions. In order to obtain preferential growth along the c-axis for pure GaN films grown by MARE technique, the films are to be grown initially at higher powers (~300 W) for 1–2 min. Since for the growth of IGN films the power is kept constant at 150 W throughout the deposition for maintaining the In/Ga ratio, the gallium-rich films could not be preferentially grown along the c-axis. When the IGN films with x = 0.5 is tried out by controlling the corresponding indium and gallium evaporation rates, a phase-segregated (PS) IGN film with a broad hump in the XRD pattern is seen. Again, the XRD peak height is an estimate of the degree of crystallinity. The crystallinity is found to be degraded for the intermediate compositions, which is mainly due to the insolubility of the alloy constituents. The onset of other crystal orientations leads to a decrease in the height of (0002) XRD peak for gallium-rich films. Further, the asymmetry of the XRD peak for indium-rich IGN films signifies the compressive strain present in them. Though comparatively less asymmetry in XRD peak is observed in Ga-rich films, they show small tensile strain. The XRD peak broadening is mainly attributed to the variations in the compositions, size effect and presence of inhomogeneous strain. Since MARE is a room temperature growth technique and the films are grown on amorphous (BSG) substrates, strain is bound to be present in these films. The FWHM values for (0002) peak for different IGN films are plotted as a function of indium concentration in Fig. 3c. The FWHM reaches a maximum value for the intermediate alloy composition which is mainly due to the large amount of strain associated with the films in this region.

Fig. 3
figure 3

a Full range XRD patterns of representative IGN films, b slow scan XRD patterns of IGN films near (0002) reflection, c FWHM of (0002) XRD peak as a function of indium concentration

Full size image

The indium compositions are calculated from the (0002) XRD peak shift (and \( (10\mathop {\bar{1}} 0) \) peak shift for gallium-rich films) by Vegard’s law under full relaxation approximation. The variation of the c-axis lattice parameter as a function of excited In/Ga ratio present in the plasma is shown in Fig. 4a. The strong lines at 451.1 nm for indium and 417 nm for gallium are used to obtain the excited In/Ga ratio. The indium concentration for different IGN films obtained from XRD is found to vary from 5 to 93 %. The indium concentrations calculated from XRD peak shifts are plotted as a function of excited In/Ga ratio obtained from OES (Fig. 4b). Indium concentrations obtained from XRD peak shift by Vegard’s law is overestimated under full relaxation approximation [27]. This observation is verified from the downshift of the plot shown. This downshift is pronounced near the PS region marked as a shadow in the graph; it might be due to larger overestimation of indium concentration near PS due to larger strain. In the rest of the regions, the straight line behaviour of the plot signifies the fact that the ratio of indium to gallium concentration is maintained in the resultant IGN films. Hence, OES can be considered as a powerful tool to control the stoichiometry of the films.

Fig. 4
figure 4

a c-axis lattice parameter as a function of indium concentration obtained from OES (solid line is a guideline to the eye), b plot of indium concentration obtained from XRD as a function of excited indium species present in the plasma (solid line drawn is a guide to the eye and dashed line is a guide for ideal case)

Full size image

According to the theoretical calculations shown by Ho et al. [12], the solubility limit of InN in GaN is less than 2 % at 450 °C. But, for IGN alloy thin films, there are several reports of enhanced solubility [28] which is largely dependent on the growth technique. The phase segregation in these alloy films is mainly suppressed because of the time required for the growth of the films. Further, the presence of strain in the alloy thin films increases the solubility limit. However, XRD studies reveal that in case of MARE-grown IGN films, solubility is present in the entire alloy composition except the range 0.44 < x < 0.60. But, compositional fluctuations leading to peak broadening are observed in all the films [25].

The In/Ga ratio is also obtained from the EDX measurement. The indium compositions obtained from EDX are lower than the values calculated from the XRD peak shift (Fig. 5). This is because indium concentration is consistently overestimated while calculating the same from XRD peak shift considering the films to be fully relaxed.

Fig. 5
figure 5

Indium concentration obtained from EDX as a function of indium concentration calculated from XRD (solid line drawn is a guide for ideal case)

Full size image

Raman spectra of IGN films excited at 488 nm for the entire composition range are shown in Fig. 6. The peaks are deconvoluted by means of Voigt lineshapes after background subtraction. All the films exhibit the characteristic A1 (LO) and E2 (high) modes corresponding to the hexagonal wurtzite structure. The broadening of the Raman peaks is mainly attributed to the high carrier concentrations present in the films leading to structural defects which in turn result in disorder-activated Raman scattering. The Raman peak broadening is also due to the relaxation in momentum conservation due to decrease in crystallite size. In the present case, alloy disorder and compositional fluctuations also play a significant role in the Raman peak broadening. The positions of the A1 (LO) and E2 (high) modes as a function of the indium composition are plotted in Fig. 7. In a ternary semiconductor A x B1−x C, the optical phonons may exhibit one-mode or two-mode behaviour. One mode alloys only present one set of LO and TO phonons with frequencies which show an almost linear dependence from one end member (AC) to other (BC). Two-mode behaviour alloys exhibit two sets of LO and TO phonons, the frequencies of which vary continuously from those of one end-member binary (AC, BC) to the impurity mode in the other end-member binary (BC:A and AC:B), respectively. In the present study for IGN ternary alloy, we can observe linear behaviour for the A1 (LO) mode peak position with indium concentration. On the other hand, E2 (high) peak positions show non-linear dependence with the composition x. It implies that A1 (LO) mode exhibits one-mode behaviour, whereas E2 (high) mode exhibits two-mode behaviour. These results are in agreement with the observations made by Hernandez et al. [29] and Alexson et al. [30].

Fig. 6
figure 6

Raman spectra of IGN films prepared by MARE (λex = 488 nm)

Full size image
Fig. 7
figure 7

A1 (LO) and E2 (high) peak positions as a function of indium compositions

Full size image

The AFM images of x = 0.75, PS and 0.18 IGN films are shown in Fig. 8. All the films are found to be smooth with an RMS roughness of ~6–7 nm. Clear grain growth is observed for x = 0.75 and 0.18 films, whereas the grains are found to be distorted for the phase segregated IGN film. No indium or gallium metallic clusters are observed for any of the films. The two dimensional image for x = 0.18 IGN films shows the elongated grains which is due to the glancing angle geometry involved during the co-evaporation process. The average grain size obtained for x = 0.75 and 0.18 IGN films is found to be 88 and 96 nm, respectively.

Fig. 8
figure 8

AFM images of a x = 0.75, b x = PS, and c x = 0.18 IGN films. d 2-dimensional AFM image for x = 0.18 film showing the glancing angle deposition

Full size image

The transmittance spectra in the wavelength range 300–2000 nm for IGN films prepared by MARE technique employing separate sources are shown in Fig. 9a. The absorption edge shifts towards higher energies with increase in gallium incorporation into the lattice. This is also an indication that gallium replaces the indium atoms and vice versa in the hexagonal lattice. The NIR transmittance values at 1700 nm are plotted as a function of indium concentration in Fig. 9b. NIR transmittance is found to increase with increase in gallium incorporation which is because free carrier absorption decreases with decrease in the number of free carriers. The gallium-rich films are found to be ~70 % transparent in the visible region. From the transmittance and reflectance data, the absorption coefficients and band gap values are calculated by the following relations.

$$ \alpha = 2.303\frac{{\log_{10} ((1 - R)^{2} /T)}}{d} $$
(1)
$$ \alpha = A(h\nu - E_{g} )^{1/2} $$
(2)

Here, α is the absorption coefficient, T is the transmittance, R is the reflectance, d is the film thickness, E g is the direct band gap and A is a constant. The energy band gap values can be determined by plotting the squared absorption coefficient (α2) as a function of photon energy (hν) and linearly extrapolating it to α = 0 (Fig. 9c).

Fig. 9
figure 9

a Transmittance spectra of IGN films prepared by MARE. b IR transmittance at 1700 nm as a function of indium composition. c Squared absorption coefficient plots for IGN films of different indium compositions. d Band gap variation of IGN films with indium composition (dashed line shows the fitting by taking the bowing parameter into account)

Full size image

The band gap values obtained from the α2 ~ hv plots are plotted as a function of the indium concentration in Fig. 9d. The data are fitted with the following equation [31]:

$$ F(x) = P_{1} (1 - x) + P_{2} (x) - Bx(1 - x) $$
(3)

where, F(x) is the energy gap as a function of composition x. P 1 and P 2 represent the band gap at x = 0 and x = 1, respectively, and B is the bowing parameter. The fitted parameters are given as P 1 = 3.43 eV (E g for GaN), P 2 = 1.90 eV (E g for InN) and B = 1.52 eV. The band gap values obtained for pure InN and GaN match well with those obtained experimentally. The small bowing parameter matches well with the recent literature values [31]. The lowering of band gap values for indium-rich films can be attributed to the compensation of Burstein–Moss (B–M) shift due to gallium incorporation into the lattice.

The room temperature PL spectra for IGN films are shown in Fig. 10a. For indium-rich films, 488-nm excitation is used, whereas 325-nm excitation is used for the gallium-rich films. The PL peaks for indium-rich samples are broader as compared to those of gallium-rich ones. This is mainly due to higher density of structural defects present in indium-rich samples. The band gaps obtained from absorption spectra (E gTR) follow nearly the same trend as those obtained from PL measurements (E gPL) (Fig. 10b). The slight difference between E gTR and E gPL is due to the difference in the momentum selection rule between the absorption and luminescence. The optical absorption occurs between the states in the valance band and states above the Fermi level in the conduction band under the momentum (k) selection rule, while the luminescence represents the recombination of electrons distributed up to the Fermi level with the localized holes in the tail states without conserving k [32].

Fig. 10
figure 10

a Room temperature PL spectra for IGN films grown by MARE. b Comparison of the band gap values obtained from PL measurements and transmittance measurements

Full size image

The room temperature electrical resistivity and carrier concentration as a function of indium composition are shown in Fig. 11. The carrier concentration decreases by 2 orders of magnitude with increase in gallium concentration which is in accordance with the free carrier absorption observed in transmittance spectra. All the films exhibit n-type conductivity. The unintentional incorporation of impurities plays a significant role in the conductivity type in GaN and InN. In particular, hydrogen acts as a shallow donor in these semiconductors. Both interstitial and substitutional hydrogen have high solubility and give rise to n-type conductivity. From theoretical calculation, it has been proved that hydrogen acts as a double donor in group-III nitride thin films [33]. Nitrogen interstitials and nitrogen on In\Ga antisite defects can also be one of the major reasons for n-type conductivity of these IGN films [34]. The antisite defect is a double donor and with a shallow first ionization energy level and is therefore a candidate for the supply of the high background donor concentration. Together with this, unintentionally doped oxygen also contributes to the n-type conductivity in these IGN films. The room temperature mobility values of IGN films are given in Table 1. The low mobility values are mainly due to the low temperature involved in the MARE process. Low temperature introduces significant amount of structural defects in the lattice which in turn results in residue imperfection scattering and thereby lowers the mobility values. Owing to the high carrier concentration, electron–electron scattering also leads to a decrease in the carrier mobility.

Fig. 11
figure 11

Room temperature electrical resistivity and carrier concentration as a function of indium composition

Full size image
Table 1 Summary of structural, optical and electrical properties of IGN films grown by MARE
Full size table

The structural, optical and electrical properties of IGN films prepared by MARE technique are summarized in Table 1.

Conclusion

IGN films in the entire composition range are prepared on BSG substrates by separate source MARE technique. OES is used to control the In/Ga ratio in these alloy films. All the films are found to crystallize in the hexagonal wurtzite structure. From XRD, the phase segregation is observed in 0.44 < x < 0.60 composition range. From the Raman spectroscopy measurements, the characteristic A1 (LO) mode is found to exhibit two-mode behaviour where as E2 (high) mode is found to follow one-mode behaviour. The band gap of IGN films could be successfully tailored from 1.88 to 3.22 eV. The band gap values obtained from optical absorption data are found to follow the same trend as obtained from PL measurements. All the films exhibited n-type conductivity with carrier concentration decreasing with increase in gallium incorporation. The separate source MARE turns out to be a powerful technique to grow IGN films in the entire composition range at room temperature which has got potential applications in flexible electronics. The phase segregation window for IGN films can further be narrowed down using a suitable buffer layer-like GaN and post-deposition annealing.