1. Introduction

The experimental evidence of narrow band gap of InN has great impact on III-nitrides research [1–4 ]. It greatly extends the wavelength coverage of In_xGa_1-xN alloy to infrared region and thus leads to more potential applications. In particular, the band gap energy of In_xGa_1-xN alloy with tunable In composition nearly perfectly matches to the solar spectrum and thus it is possible to fabricate high efficiency solar cell using the same III-nitride family materials [5,6 ]. The large absorption coefficient of In_xGa_1-xN with a magnitude of ~10⁵ cm⁻¹ yields strong light absorption for relatively thin layer [7]. In addition, as a member of nitride family, In_xGa_1-xN is a robust material for radiation and an excellent candidate for solar cells used in space. Unfortunately, the application of In_xGa_1-xN is limited by poor crystalline quality, in particular those with large In fraction [8–10 ]. Furthermore, study on the properties of In_xGa_1-xN with a large range of In compositions is much less. For example, electric transport characteristics under illumination of In_xGa_1-xN alloys with a large range of In fraction is not well studied yet, which definitely does not benefit for fabrication of devices such as solar cells and detectors.

In this article, we will report the photoconductivity (PC) behaviors of In_xGa_1-xN alloys, where the PC is found to be related with In fraction (x). It is noticed that there is a critical In fraction ($x_c$), where the PC converts from negative to positive one. Hall-effect measurements under illumination indicate that this transition from negative PC to positive one is most likely due to the different behaviors of electron mobility under illumination, which is probably influenced by In-related defects.

2. Experiment

All samples were grown by a plasma-assisted molecular beam epitaxy (PA-MBE). GaN layers grown on c-Al₂O₃ substrate by metal-organic vapor phase epitaxy were used as templates for In_xGa_1-xN epitaxy. The GaN template was degassed under ultra-high vacuum and then its surface was protected under N atom beam before the regrowth of GaN. The regrown GaN was about 100-nm-thick and then the In_xGa_1-xN was directly grown on the GaN layer using growth temperature controlled epitaxy method as we reported previously [11], where all the samples were grown under slightly In-rich condition and the In composition was controlled by the substrate temperature and Ga flux as well. The In_xGa_1-xN layers investigated in this article were all around 300-nm-thick. The In fraction of the In_xGa_1-xN had been carefully determined by using high resolution X-ray diffraction (HR-XRD, Bruker D8) measurement. Electron concentration (n) and mobility (µ) were measured by a Hall effect system with a magnetic field of 0.503 T in Vander Paul structure. Photoconductivity measurements were conducted using a continuous-wave laser with a wavelength of 808 nm (1.53 eV) at normal incidence with a spot diameter of 5 mm. Two electrodes in coplanar stripe with a length of 1 mm and spacing of 7 mm were fabricated by evaporating Ti/Al/Ni/Au. To gurantee that all samples are studies under interband excitation, the In composition in the In_xGa_1-xN layer investigated in this study is up to 0.58 since the band gap energy of In_0.58Ga_0.42N is ~1.35 eV according to the reported bowing parameter [12], which is smaller than the incident photon energy of 1.53 eV.

3. Result and discussion

It is known that most of semiconductors show positive PC under super-band gap excitation, which is mainly due to that the absorbed light generate a large number of electron-hole pairs. GaN is a good example showing the positive PC [12,13 ]. However, InN is different from GaN, where a negative PC was often observed [14,15 ]. Obviously, the PC of In_xGa_1-xN should be influenced by the In fraction. Thus, we prepared several In_xGa_1-xN samples with In composition x varying from 1 to 0.58, where the fraction of In was confirmed by XRD measurement. Figures 1(a)-1(e) show surface morphologies of those In_xGa_1-xN layers and the root mean square (RMS) values of these surfaces are also shown in Fig. 1(f). It is shown that the surface is quite flat for the InN and In_0.92Ga_0.08N, where step-flow features are observed. The RMS roughness is less than 0.5 nm in a scan area of 3 μm × 3 μm. With decreasing In composition, the surface becomes rougher but not so bad yet, with a typical RMS value of 2–3 nm. Meanwhile, step-flow growth mode is not kept anymore and the surface changes to grain-like morphology. The grain size reduces with decreasing In fraction, indicating deteriorated crystalline quality.

 

Fig. 1 AFM images of In_xGa_1-xN films in a scanned area of 3 μm × 3 μm with different In fraction: (a) 0.58, (b) 0.7, (c) 0.84, (d) 0.92, and (e) 1. (f) The RMS value of the In_xGa_1-xN layer as a function of In fraction. The In droplets were removed by dilute HCl solution before the AFM measurement.

Download Full Size | PDF

The residual electron concentration (n) and electron mobility (µ) of the In_xGa_1-xN as a function of In composition are shown in Fig. 2 . Generally, the n increases with decreasing In composition, where it changes from 3.2 × 10¹⁸ to 6.3 × 10¹⁸ cm⁻³ with decreasing x from 1 to 0.7, but slightly decreases to 3.5 × 10¹⁸ cm⁻³ for the In_0.58Ga_0.42N. The µ greatly decreases from 1740 to 189 cm²/Vs. This indicates that the crystal quality of In_xGa_1-xN is greatly deteriorated with decreasing In composition, which is reasonable since it is quite difficult to grow In_xGa_1-xN with middle In content [11], where the lattice disorder is serious.

 

Fig. 2 Electron mobility and residual electron concentration as a function of In composition for the In_xGa_1-xN samples at room temperature.

Download Full Size | PDF

The typical PC response of In_xGa_1-xN with In fraction x ranging from 1 to 0.58 are shown in Fig. 3 . The photocurrents of InN, In_0.92Ga_0.08N and In_0.84Ga_0.16N under laser irradiation show distinct drop till saturation with respect to the steady-state dark current, and then gradually recovers after the laser is off. Interestingly, the magnitude of the distinct drop gradually decreases with reducing In fraction x in In_xGa_1-xN layer, and finally the photocurrent starts to rise up at and below a critical value x when the irradiated laser beam is opened. The transition fraction $x_c$ is about 0.7, where the satuated photocurrent is almost the same as the steady state dark current. Below $x_c$, the transient photocurrent rises up to the saturation under illumination. The saturation photocurrent is positive compared with the steady-state dark current when the laser turns off. Therefore, a negative PC is observed at high In fraction samples while it gradually changes to a positive PC with decreasing In fraction.

 

Fig. 3 Photocurrent transient response of the In_xGa_1-xN with different In-fractions at room temperature.

Download Full Size | PDF

The negative PC of InN has been previously studied, where a recombination center at the energy level of near the valence band was proposed to act as a scattering center under illumination, leading to the negative PC at room temperature [14–18 ]. The typical processes to explain this negative PC is shown in Fig. 4(a) : under illumination, a large number of electrons in the valence band (process 1) and donor states (process 2) are excited to the conduction band (E_c) to form free carriers, and the corresponding quasi-Fermi levels for electrons (E_Fn) and holes (E_Fp) are formed. Then free carriers would diffuse (process 3 and 4), which usually leads to a normally positve PC. However in InN, a large number of free holes will be immediately trapped by the recombination centers (E_R), making the latter positively charged (process 5) and act as scattering centers for electrons. Since the trapping process is faster than the diffusion process, free photo-generated holes quickly captured by the recombination centers. Those centers scatter electrons and reduce their mobility. The reduction of electron mobility decreases the conductivity and this contribution is more serious than the increase of conductivity by photo-generated carriers, as reported previously [14,15 ]. Therefore, the photocurrent under laser irradiation shows a distinct drop till saturation with respect to the steady-state dark current as shown in Fig. 3, and then the current gradually recovers after the laser is off since the excited electrons recombine with the holes in the recombination centers (process 7) and thus the positively charged recombination centers return to neutral.

 

Fig. 4 The proposed energy-level diagrams showing electronic transitions responsible for the transient photocurrent, (a) and (b) correspond to the In_xGa_1xN samples with x > x_c and x < x_c.

Download Full Size | PDF

It should be mentioned that the abnormally negative PC are only observed in InN among III-nitride binary semiconductors. GaN and AlN show normally positive PC. Therefore, the scattering centers are usually thought to be originated from In-related point defects [14,17–19 ], and then less In fraction x in the In_xGa_1-xN alloy probably leads to lower density of the scattering centers. Consequently, under the same illumination, there will be less number of photo-generated free holes captured from the valence band by the recombination centers with decreasing In fraction. In other words, the concentration of the photo-generated scattering centers created by the laser illumination becomes lower and lower with decreasing In composition, and is even negligible for the In_xGa_1-xN with low In fraction. This indicates that the reduction of the carrier mobility caused by the photo-generated metastable scattering centers will be gradually weakened with decreasing In fraction, and hence the magnitude of the distinct drop of photocurrent gradually decreases with decreasing In fraction x from 1 to 0.84 as shown in Fig. 3 when the samples were irradiated. And finally in x = 0.7, the photocurrent starts to rise up at the beginning and then slowly decreases to saturation. The saturation photocurrent is almost the same as the steady-state dark current when the laser is turned off. The transient positive PC of x = 0.7 under illumination results from the lower concentration of scattering centers, the slower holes trapping (process 5) rate and that the process 2 is also a relatively slow process, and thus the photo-generated carriers diffuse more easily (process 3 and 4) and contribute to the positive PC before being captured and scattered by defects. Hence the PC gradually rises before descending. When laser is off, three processes happen as shown in Fig. 4(a): Process 6, holes recombine with electrons; Process 7, electrons recombine with those holes in the recombination centers, and thus the positively charged recombination centers return to neutral; Process 8, electrons return to the positively charged donor states. The process 8 is much harder and hence slower than processes 6 and 7 due to strong lattice relaxation [15]. Then the mobility of electrons quickly recovers to the steady-state at dark condition. And those nonequilibrium electrons remained in the conduction band result in a photocurrent jump when the light is off, as shown in Fig. 3 in x = 0.7.

It is noted that the crystal quality of In_xGa_1-xN actually becomes bad with decreasing In fraction x as indicated by the AFM image and Hall-effect measurement results as shown in Figs. 1 and 2 . This reduction of crystal quality is mainly ascribed to the generation of threading dislocations but not In-related point defects such as In vacancies and/or In-related complex defects. In other words, In-related point defects do not play the most important role in the AFM images. On the other hand, it is true that the crystal quality reduction with decreasing In fraction x is generally accompanied by the increase of density of the point defects. However, it is difficult to conclude what type of point defects were generated. From the dependence of electron concentration on In fraction shown in Fig. 2, it is clear that more donor-like point defects were generated since the residual electron concentration increased and the mobility reduced. However, this does not imply that the density of recombination centers which is probably due to the In-related defects is increased since the recombination centers are acceptor-like and should be neutral before trapping holes when the light is on. Further study should preformed to clarify the origin of those recombination centers.

As shown in Fig. 3, the PC changes to a normally positive one in In_xGa_1-xN with x<0.7, as the concentration of recombination centers are too low to play any important role any more. As shown in Fig. 4(b), large number of electrons in the valence band (process 1) and donor states (process 2) are excited to the conduction band (E_c), and carriers then diffuse (process 3 and 4) to produce positive PC. Other than the processes shown in Fig. 4(a), very few or neglible holes will be trapped by the recombination centers (E_R) and the free holes trapping process (process 5) in the high In-fraction In_xGa_1-xN tends to be negligible in the low In-fraction samples. Therefore, less or negligible scattering centers are formed, and thus a positive PC is observed, as shown in the Fig. 3 for x = 0.58. It should be noted that the process 1 and 2 correspond to different time constants. The interband transition is a quick process (in the time scale of ~100 ms) and a faster onset are usually observed if the detection equipment is pricise enough. Unfortunately this process is beyond our experimental instrument precision (~500 ms) and is often not detectable. On the other hand, the impurity-to-conduction band transition (process 2) is a slow process (in the time scale of seconds), which is due to the large lattice relaxation of impurity. That large lattice relaxation of impurity also results in the persistent PC, which was reported previously [15]. Therefore, only the process 2 was observed in sample of x = 0.58 and the rising process is slow as well due to the limitation of our detection equipment.

To further confirm the above model, temperature-dependent Hall-effect measurement was performed with and without laser irradiation for the In_xGa_1-xN with x = 0.92 and 0.58, to investigate the effect of light illumination on residual electron concentration and mobility. These two samples are typical ones showing negative (x = 0.92) and positive PC (x = 0.58), respectively. As shown in Fig. 5(a) and Fig. 5(b), with increasing temperature (T) from 100 to 300 K, the sheet residual electron density (n _s) in the In_0.92Ga_0.08N increases from 8.7 × 10¹³ to 9.1 × 10¹³ cm⁻² and the mobility µ decreases from 1620 to 1400 cm²/Vs. In comparison, the sheet residual electron density n _s in the In_0.58Ga_0.42N increases from 6.3 × 10¹³ to 1.22 × 10¹⁴ cm⁻² and the mobility µ decreases from 212 to 140 cm²/Vs. It should be mentioned that the larger density variation and smaller mobility change with increasing temperature in the In_0.58Ga_0.42N is due to the deteriorated crystal quality which results in larger density of donor-type defects and more serious impurity scatterings. When both samples are exposed under laser radiation, the n _s of the both samples are higher, since the absorption of light generates carriers in both kinds of samples. However, the mobility µ shows totally different behaviors. At all temperature range, it is reduced in the In_0.92Ga_0.08N under the light irradiation. The mobility reduction is due to In-related scattering centers (E_R) as reported in the InN as well [14–16 ]. This is supported by the negative PC at room temperature as shown in Fig. 3. However, the electron mobility almost does not change in the In_0.58Ga_0.42N upon light irradiation, which coincides with our above prediction that, since less or negligible recombination centers exist in the In_0.58Ga_0.42N, photo-generated holes are not captured under illumination and negligible scattering centers are formed. Hence the mobility under illumination does not reduce and positive PC are observed in the lower In-fraction In_xGa_1-xN.

 

Fig. 5 Residual electron concentration (circle) and mobility (square) in dark condition (open) and under illumination (solid) at temperatures from 100 to 300 K of (a)In_0.92Ga_0.08N and (b) In_0.58Ga_0.42N.

Download Full Size | PDF

4. Conclusion

In summary, negative PC is observed in the InN and the high In-fraction In_xGa_1-xN, whereas the PC gradually changed to be positive with decreasing In fraction. Transition from negative to positive PC occurred at a critical point of about x _c = 0.7. The negative PC is mainly due to that the recombination centers become positively charged, resulting from recombination with the mobile holes, and hence scatter the electrons, reduce their mobility and conductivity. The recombination centers are most likely In-related defects, and their concentration decreases with decreasing In fraction, leading to reduction of scattering centers under illumination. Therefore, a normally positive PC is observed in the low In-fraction In_xGa_1-xN. That is also supported by the temperature-dependent Hall-effect measurement with and without illumination.