Investigation of a novel AlZnN semiconductor alloy

The formation of an AlZnN alloy was investigated as a route to tune the bandgap of Zn 3 N 2 . A signiﬁcant shift of the bandgap was observed in the deposited ﬁlms, increasing from 1.4 eV for Zn 3 N 2 to 2.8 eV for AlZnN alloys with an AlN fraction of (cid:1) = 0.19. The refractive index followed a similar trend, approaching that of AlN. The charge carrier density of AlZnN samples was signiﬁcantly reduced reaching values in the order of 10 16 cm (cid:3) 3 . (cid:1) 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). were studied by Energy Dispersive X-ray Spectroscopy (EDS), Grazing Incidence X-ray Diffraction (GIXRD) at 1 (cid:3) incidence, Spectroscopic Ellipsometry (SE) and Hall effect measurements, using the equipment detailed in previous papers For electrical measurements, ohmic contacts were obtained using high purity Indium annealed at 350 (cid:3) C for 30 s.


Introduction
Zinc Nitride (Zn 3 N 2 ) is a II-V semiconductor with potential for application in photonics devices, such as solar cells, because of its narrow direct bandgap and earth-abundant elemental composition [1][2][3][4][5]. The basic semiconductor properties of Zn 3 N 2 have been demonstrated in simple devices such as thin-film transistors [6][7][8][9]. However, to meet the demands of advanced semiconductor applications such as heterojunction solar cells, a Zn 3 N 2 -compatible materials system must be developed that will allow bandgap engineering through semiconductor alloying, and the design of semiconductor heterostructures. The first investigation of such a materials system was reported by Wu et al., who demonstrated a Mg 3x Zn 3(1-x) N 2 semiconductor alloy with a tuneable bandgap in the range of 1.2-2.1 eV for Mg content up to Â = 0.59 [10].
In this letter, we investigate the formation of an alternative AlZnN alloy and report the optical and electrical properties of such films. AlN is an interesting alloying candidate because it is already well established in the highly successful LED, laser and electronic technologies based on (Al)InGaN semiconductors. Furthermore, because of the wide bandgap of AlN, we anticipate that the bandgap of AlZnN will be tuneable over a wider range than that of MgZnN.

Film deposition
AlZnN films were deposited on unheated glass substrates by sputtering from an AlZn target (80 at. % Zn, 20 at. % Al) at a constant current of 400 mA in N 2 plasma. The base pressure in the deposition chamber was 7 Â 10 -7 Torr and the process pressure was in the order of 10 -3 Torr. Sputtering in low N 2 flow rates (15 sccm) resulted in a low Al content in the deposited films, whilst higher Al concentration was obtained by sputtering in nitrogen-rich conditions (45 sccm). Under nitrogen-rich conditions, the different sputtering yields of Al and Zn species caused a gradual depletion of Zn from the surface of the target, resulting in the formation of films with higher Al content. Similar process parameters were used to deposit Zn 3 N 2 using a Zn target in the same sputter chamber. The thickness of films studied here is in the range of 80 to 150 nm.

Characterisation
The physical, optical and electrical properties of the deposited films were studied by Energy Dispersive X-ray Spectroscopy (EDS), Grazing Incidence X-ray Diffraction (GIXRD) at 1°incidence, Spectroscopic Ellipsometry (SE) and Hall effect measurements, using the equipment detailed in previous papers [5,11]. For electrical measurements, ohmic contacts were obtained using high purity Indium annealed at 350°C for 30 s.

Results and discussion
The composition of the deposited films was evaluated with quantitative EDS analysis. The Zn and Al peaks of the EDS spectra are shown in Fig. 1a for 5 samples: ZN, a reference Zn 3 N 2 sample, and AZN-1 to AZN-4, four AlZnN samples with increasing Al content. The samples were composed of Zn, Al, and N with the results listed in Table 1. Carbon and oxygen were also detected, however, in line with our previous experience of these materials, carbon is readily found as a surface contaminant (2-5 at. %), and oxygen is found in large amounts in the substrate, therefore they were deemed experimental artefacts and not a part of the film composition [5]. A binary alloy of the following form was considered: Al 2x -Zn 3(1-x) N 2 . The AlN and Zn 3 N 2 fractions were calculated from experimental Al, Zn, and N atomic fractions as: starting from the Zn/Al and Zn/N atomic ratios in Al 2x Zn 3(1-x) N 2 . The calculated fractions are listed in Table 1 showing a maximum AlN fraction of Â = 0.19 was obtained. The AlN and Zn 3 N 2 fractions are plotted in Fig. 1b, demonstrating that the composition of AlZnN samples follows the expected trend of stoichiometric Al 2x Zn 3(1-x) N 2 (dashed line). It is likely that the deposited films have a N-rich composition because the average values for the two fractions are offset from the stoichiometric ratio and are in a N-rich region. Given this result, it is more accurate to refer to the composition as Al 2x Zn 3(1-x) N 2+<delta> .
The structure of the AlZnN films was investigated by GIXRD, shown in Fig. 1c. Diffraction peaks corresponding to crystal planes (4 0 0), (4 4 0), and (3 2 1) of the Zn 3 N 2 structure were identified in the pattern of ZN [5,12]. These features shifted to higher angles in AZN samples, indicating that the size of the crystal lattice decreased as a result of the incorporation of Al into the structure. The preferred (4 0 0) orientation was strongly suppressed in samples AZN-2 and AZN-3 (x > 0.10), and the additional observed diffraction peaks became broader as the films transitioned to an amorphous structure. We assign the broad features observed in AZN-2 and AZN-3 to the (3 2 1), (5 2 1), and (2 2 2) diffraction peaks, which are appropriately shifted from their expected positions in the Zn 3 N 2 crystal structure because of Al incorporation. For instance, the diffraction peak (3 2 1) in AZN-2 is traced to its original position in sample ZN, shown by the dashed line in Fig. 1c. We attribute the transition to amorphous films to a non-optimal growth surface and growth conditions. However, the growth of single crystal AlZnN layers with high Al content may prove challenging even under optimised growth conditions because of the different crystal structures of Zn 3 N 2 and cubic AlN.
The most direct evidence of alloying is the effect on the optical properties which we probe through SE measurements and by using a two-layer optical model consisting of: (a) the AlZnN film and (b) the glass substrate. The refractive index and extinction coefficient of the AlZnN samples obtained from the models are shown in Fig. 2a,b. For samples AZN-1 to AZN-4, there is a very clear shift of the absorption onset to higher energies, and the refractive index decreased accordingly. A Cody absorption plot was used to obtain the bandgap of each sample, shown in Fig. 2c [5,13]. The bandgap and refractive index obtained from this analysis are shown as a function of Al content in Fig. 3a,b. A shift of the optical properties of the AlZnN samples in the direction of AlN was observed with increasing Al content (E g = 5.34 eV and n = 1.7 for cubic AlN) [14]. Note we refer to values for cubic AlN based on the principle that the structure of Zn 3 N 2 is a cubic lattice and more likely to be compatible with cubic rather than the wurzite AlN structure. We report a maximum bandgap of 2.8 eV measured for Â = 0.19. From Hall effect measurements, we obtained the resistivity and carrier density of the AlZnN samples, shown as a function of Al content in Fig. 3c, d. All measured samples displayed n-type conductivity as seen in Zn 3 N 2 and also the recent demonstration of MgZnN. However, our AlZnN samples were found to have a significantly higher resistivity and lower carrier density than Zn 3 N 2 . The samples with the highest Al content could not be measured as they were effectively insulating, attributed to carrier depletion at the interfaces of the films which is observed in low-doped semiconductor thin films. The observation of a composition-dependent carrier density, which varies from 2 Â 10 19 cm À3 in Zn 3 N 2 to less than 10 17 cm À3 in AZN-3 explains the non-linear dependence of the bandgap on composition seen in Fig. 3a. In the case of highlydoped Zn 3 N 2 we expect a significant Burstein-Moss shift to increase the optical bandgap [3]. For the carrier densities measured in the AlZnN samples, the Burstein-Moss shift has little effect on the optical bandgap, and we can consider the measured bandgaps as intrinsic. A comprehensive first principles study by Kumagai et al. obtained a value of 0.84 eV for the intrinsic bandgap of Zn 3 N 2 [3]. When this value is plotted in Fig. 3a, the bandgap dependence with Al content shows an approximately linear behaviour.
In Zn 3 N 2 films, the unintentional n-type doping is often attributed to nitrogen vacancies and oxygen contamination [3,15].W e propose that the composition of the AlZnN samples studied here is introducing N-interstitials and Zn vacancies as compensating acceptor states [15]. Furthermore, we suggest that the binding energy of oxygen donor states may increase in AlZnN, as seen in oxygen-doped AlGaN [16]. The reduced carrier density of these samples is a promising result and suggests that further studies of AlZnN samples could provide insight on donor defects in Zn 3 N 2 . Furthermore, with a significant reduction in background n-type doping, these results suggest that high-quality AlZnN samples are candidates to realise p-type doping in this materials system.

Conclusions
In summary, we have investigated the optical and electrical properties of AlZnN alloys with a maximum AlN fraction of Â = 0.19. Analysis of optical measurements showed that the deposited films were semiconductor alloys with properties between Zn 3 N 2 and AlN. A maximum bandgap of 2.8 eV was measured for Â = 0.19. The electrical properties of the AlZnN films changed drastically, reaching carrier concentrations as low as 7x10 16 cm À3 , which was attributed to partial compensation and potentially an increase in the donor binding energy with increasing Al content. This low intrinsic doping suggests that high-quality AlZnN is a good candidate for the development of p-type doping in Zn 3 N 2 -related materials, which is a critical step to demonstrating devices such as lasers, LEDs and solar cells. These results all point to AlZnN being an attractive alloy system for use in heterostructures based on Zn 3 N 2 .