|Home | About | Journals | Submit | Contact Us | Français|
P-type doping in high Al-content AlGaN alloys is a main challenge for realizing AlGaN-based deep ultraviolet optoelectronics devices. According to the first-principles calculations, Mg activation energy may be reduced so that a high hole concentration can be obtained by introducing nanoscale (AlN)5/(GaN)1 superlattice (SL) in Al0.83Ga0.17N disorder alloy. In this work, experimental evidences were achieved by analyzing Mg doped high Al-content AlGaN alloys and Mg doped AlGaN SLs as well as MgGa δ doped AlGaN SLs. Mg acceptor activation energy was significantly reduced from 0.378 to 0.331eV by using MgGa δ doping in SLs instead of traditional doping in alloys. This new process was confirmed to be able to realize high p-type doping in high Al-content AlGaN.
High Al-content AlGaN alloys are ideal materials for deep ultraviolet (DUV)1 optoelectronic devices due to their large direct band gaps with operating wavelengths from 364nm down to 200 nm2,3,4,5. The external quantum efficiency (EQE) of AlGaN-based DUV light-emitting diodes (LEDs), however, is as extremely low as 0.1%, which is still a formidable obstacle6,7. N-type AlGaN can be produced relatively easily8,9. The very low p-type doping efficiency in AlGaN hinders the further improvement of AlGaN-based DUV LEDs. The difficulty to realize p-type doping is related to the high acceptor activation energy (EA), the compensation by nitrogen vacancies, the increased hole scattering, and the limited acceptor solubility10,11. For the most widely used p-type dopant of Mg, its EA in AlxGa1−xN increases monotonically with increasing Al-content from 0.17eV in GaN to 0.51eV in AlN11,12. This behavior indicates that only a very tiny fraction (~10−9) of Mg dopants can be activated in AlN at room temperature 11. Therefore, decreasing Mg acceptor activation energy is one of the most challenges in AlGaN-based DUV optoelectronic devices.
Great efforts have been devoted to improve p-type conduction in group-III nitrides13,14,15,16,17,18,19,20. Different from suppressing the charge separation effect in InGaN-based devices, polarization doping has been applied to increase the hole concentration in AlGaN alloys by ionizing Mg acceptor in the polarization field13,14. Alternative acceptor-donor co-doping and non-equilibrium growth with Mg pulse doping and Mg δ-doping have also been developed to reduce the acceptor activation energy, and thus, increase the hole concentration and enhance the p-type conductivity of AlGaN alloys15,16,17,18,19,20. So far, most experiments were focused on the p-type conductions of GaN and low Al-content AlGaN alloys, but the bottlenecks of the p-type doping in high Al-content AlGaN still remain20,21,22.
Many works were concentrated on the doping in the superlattices (SLs)23,24,25,26,27,28,29, where a periodic oscillation of the valence band edge was created by the valence band discontinuity and innate polarization fields in Mg-doped AlGaN/GaN SLs, resulting in the accumulation of holes near the valance band edge close to the Fermi energy forming the so-called two-dimensional (2D) hole gases. In our previous work, we have used SiGa δ doping in Al0.6Ga0.4N alloys to increase the n-type carrier density30. Recent theoretical works predicted that the nanoscale (AlN)m/(GaN)n (m>n) SL could convert the valence-band maximum (VBM) from the crystal-field split-off hole to heavy hole band, leading to the increase of the transverse electric (TE) polarized light emission efficiency31,32. The influence of the nearest and next-nearest (NN) atoms on Mg electronic structures in nanoscale (AlN)5/(GaN)1 SL substitution for Al0.83Ga0.17N disorder alloy was theoretically investigated by Zhong et al.1. The results showed that the EA decreases if the NN Ga atom number increased and the Mg-centered tetrahedron volume decreased. In this way, the Mg acceptor activation energy can significantly be reduced to 0.26eV, very close to that of GaN, in (AlN)5/(GaN)1 SL by MgGa δ-doping1. Recently, improved p-type conductivity was achieved in multidimensional Al0.63Ga0.37N/Al0.51Ga0.49N SLs33. In this work, we use MgGa δ doping in (AlN)m/(GaN)n SLs to study Mg acceptor activation energy, aiming to find a proper way to minimize it in high Al-content AlGaN.
In this work, traditional Mg doped AlGaN alloys (in which Al, Ga, Mg and N atoms arrived at the substrate at the same time), Mg doped AlGaN SLs and MgGa δ doped AlGaN SLs were grown on c-plane sapphire substrates by using the radio-frequency plasma-assisted molecular beam epitaxy system (rf-MBE, SVTA 35-V-2). The growth details are shown in Table 1 Sample D1 and D2 were Mg doped AlGaN alloys with traditional doping method. Mg doping was continuously carried out for 30min. Sample D3 and D4 were Mg doped AlGaN SLs. Mg doping was continuously performed for 10s at each cycle and the total deposition lasted for 180 cycles. Hence the real Mg doping time was also 30min. Sample D5 and D6 were MgGa δ doped AlGaN SLs with a cycle period of 15s for 180 cycles. Although the cycle period of sample D5 and D6 was 15s, the effective growth time in a cycle was still kept as 10s and the total effective growth time was also 30min. For sample D5, the growth process consisted of two loops as shown in Fig. 1. During the growth of AlGaN thin films, the nitrogen flow rate was set at 2.65 sccm under 375W rf-plasma power. Prior to the growth, nitridation was performed at 810°C for 10min under 500W rf-plasma power with a nitrogen flow rate of 2.65 sccm. AlGaN films were examined by high-resolution x-ray diffraction (HRXRD, Bede D1) and high-resolution transmission electron microscopy (HRTEM, JEOL JEM 2010 FEF UHR). Ni/Au electrodes (15nm Ni and 50nm Au) were made by thermal evaporation with templates of 150×150μm2 in area. Current versus voltage (I-V) characteristics were measured by using a semiconductor device analyzer (Keithley 4200, Keithley Instruments).
Figure 2 shows XRD patterns of AlGaN films grown on sapphire substrates. AlGaN (0002) peaks were found between the GaN (0002) peak at 34.543° and AlN (0002) peak at 36.033°. The Al2O3 (0006) peaks were normalized at 41.700°. The full widths at half maximum (FWHM) of the Mg-doped AlGaN (0002) peaks were around 800arcsec in alloys and 1000arcsec in SLs. AlGaN peaks were fitted with Gauss model to get more accurate peak information. According to the alloy crystal parameter formula and Bragg’s law 2dhklsinθ=nλ as well as hexagonal interplanar distance formula , the compositions of AlxGa1−xN thin films were determined by using standard crystal parameter cGaN of 0.5189nm and cAIN of 0.4981nm. In this way, the compositions of all samples from D1 to D6 were Al0.97Ga0.03N, Al0.75Ga0.25N, Al0.79Ga0.21N, Al0.53Ga0.47N, Al0.76Ga0.24N and Al0.47Ga0.53N, respectively. The compositions of (AlN)m/(GaN)n in D3 and D5 nearly matched to the designed value of 4:1, while those in D4 and D6 nearly approached to the designed 1:1.
As XRD results reveal only the macro compositions of the AlGaN films, the (AlN)m/(GaN)n SLs were confirmed by HRTEM results. Figure 3(a) shows the cross-sectional HRTEM image of AlGaN SLs grown on sapphire in sample D5. The total thickness of AlGaN SLs were about 210nm. The magnified HRTEM image of AlGaN SLs and the corresponding FFT image are shown in Fig. 3(b) and (c). The growth direction of AlGaN SLs on sapphire was , in agreement with the XRD results. As shown in Fig. 3(c), five extra diffraction spots were obtained along  axis in one unit. The four quinquesection spots were attributed to the (AlN)4/(GaN)1 SLs, indicating that the monolayer SL structure was achieved and NN Ga atom number increased in sample D5. The one bisection spot might be caused by the dislocations. Hence sample D5 was measured to be (AlN)4/(GaN)1 superlattices, achieving the designed (AlN)m/(GaN)n SLs structure with increased NN Ga atom number.
Figure 4 shows the I-V characteristics of sample D4, D5 and D6 at different temperatures. The linear I-V behavior indicated the Ohmic contacts between Ni/Au electrodes and AlGaN films. The good ohmic behavior could be attributed to the formation of p-type NiO34. As for sample D5, the resistances were deduced to be 13.67GΩ, 5.776GΩ, 2.090GΩ, 1.042GΩ, 547.8MΩ and 340.3MΩ at temperatures of 50, 80, 110, 140, 170 and 200°C, respectively. The insets show the corresponding Arrhenius plot of the resistivity (ρ) versus temperature (T). Since , the ionization energy (EA) of Mg in AlxGa1−xN in sample D5 was fitted to be 0.331eV as shown in Fig. 4(b). The same procedures were also performed for other samples. As for the asymmetry I-V curves as shown in Fig. 4(c), EA was calculated separately under positive and negative voltages and then made of an average. The EA for samples D1-D6 were 0.386, 0.378, 0.358, 0.344, 0.331, and 0.311eV, respectively. The highest EA as 0.386eV was in sample D1 and the lowest EA as 0.311eV lied in sample D6, in match with calculated results that high EA in high Al component alloys, and low EA in low Al component SLs.
Figure 5 shows the dependence of EA on Al composition in AlGaN with three doping methods. Obviously, the EA reaches the lower values in MgGa δ doped AlGaN SLs, the medium in Mg doped SLs and the higher in traditional Mg doped AlGaN alloys. As for Al content around 0.8, the EA decreases from 0.378 to 0.358, then to 0.331eV by the three methods in turn as Mg doped AlGaN alloys, Mg doped SLs and MgGa δ doped SLs. Hence, Mg acceptor activation energy can be significantly reduced from 0.378 to 0.331eV by using MgGa δ doping in (AlN)4/(GaN)1 SLs instead of traditional Mg doping in Al0.8Ga0.2N alloys. The difference between the theoretical value (0.26eV) and the experimental one (0.331eV) is attributed to two reasons: firstly, the macro function of Mg activation energy in Ga and Al was measured in this experiment while only the Mg activation energy in Ga was calculated to be 0.26eV; secondly, the as-grown (AlN)4/(GaN)1 SLs were not perfect single crystal with dislocations which might affect the Mg activation energy. Therefore, we have experimentally proved the theoretical prediction that Mg acceptor activation energy can be significantly decreased in (AlN)m/(GaN)n SL1.
In conclusion, we have systematically studied Mg doping in high Al-content AlGaN by using different doping methods. For high Al-content AlGaN, Mg acceptor activation energy can be significantly reduced from 0.378 to 0.331eV by using MgGa δ doping in (AlN)4/(GaN)1 SLs instead of traditional Mg doping in Al0.8Ga0.2N alloys. Our experimental study verifies the prediction of the first-principles calculations1, and provides potential applications in AlGaN-based DUV optoelectronic devices.
How to cite this article: Wang, X. et al. Experimental evidences for reducing Mg activation energy in high Al-content AlGaN alloy by MgGa δ doping in (AlN)m/(GaN)n superlattice. Sci. Rep. 7, 44223; doi: 10.1038/srep44223 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work is supported by the NSFC under Grant Nos 11574235, 11175135, J1210061 and the MOST China under Grant Nos 2014GB109004. The authors would like to thank Dr. He Zheng, Fan Cao and Ligong Zhao for the technical support.
The authors declare no competing financial interests.
Author Contributions X.W. carried out the experiments and drafted the manuscript. W.W. participated in the design of the study and performed the analysis. J.L.W. participated in the measurements. H.W. conceived the study and participated in its design. C.L. supervised the overall study and polished the manuscript. All authors read and approved the final manuscript.