Effects of cation stoichiometry on surface morphology and crystallinity of ZnGeN₂ films grown on GaN by metalorganic chemical vapor deposition

ZnGeN₂, a ternary heterovalent II-IV-N₂ compound, has recently attracted growing interest for applications in light emitters^1–3 and photovoltaics,^4,5 thanks to its complementary properties to GaN.^6–9 Hypothetically, the ideal ZnGeN₂ structure can be derived from a GaN crystal by replacing every two Ga atoms by one Zn atom and one Ge atom. An ideal octet-rule-preserving unit cell of ZnGeN₂, in which every N atom is bonded with two Zn and two Ge atoms, will have the orthorhombic symmetry of either space group Pna2₁ (sometimes referred to instead as Pbn2₁, depending on the particular choice of principal axes) or Pmc2₁, with the first structure being the lower energy configuration.⁴ Deviation from the ideal coordination of the cations around N will result in a locally octet-rule-violating wurtzite structure (space group P6₃mc).⁴ Although the octet-rule-violating, fully cation-disordered wurtzite structure has significantly higher predicted formation energy per formula unit than the ordered Pna2₁ or Pmc2₁ structures,⁵ this situation may be achieved, for example, in a kinetically limited growth regime.¹⁰ The lattice parameters for the ordered and disordered structures have been measured to be similar.¹⁰ The lattice mismatch of the Pbn2₁ structure with GaN is less than 1%.¹¹ The orthorhombic Pna2₁ (Pbn2₁) ZnGeN₂ structure has a predicted bandgap very close to that of GaN (3.4 eV).¹¹ From the theoretical prediction based on first-principles calculations, the valence band of ZnGeN₂ can be as high as 1.1 eV above that of GaN at the heterointerface.^6,7 This large band offset has inspired novel designs for high quantum efficiency optoelectronic devices including green light-emitting diodes,¹ ultraviolet laser diodes,² and quantum cascade lasers.³ On the other hand, the presence of cation disorder has been predicted to shrink the bandgap significantly, by up to almost 3 eV.^5,12 Such a cation-disorder-dependent bandgap can be useful for designing broadband solar absorbers,⁵ although such experimental verification for ZnGeN₂ is still lacking.

Prior efforts on the synthesis of ZnGeN₂ include reports on (i) powder samples synthesized from the reaction of NH₃ and Zn₂GeO₄,^13,14 (ii) polycrystalline films grown by the reaction of a Cl₂–N₂–HCl mixture with Zn and Ge vapors,¹⁵ (iii) powder samples synthesized under high pressure–high temperature conditions from a mixture of Zn₃N₂ and Ge₃N₄,¹⁶ (iv) radio frequency sputtered polycrystalline films,¹⁷ (v) vapor–liquid–solid (VLS) synthesis of polycrystalline ZnGeN₂^10,18,19 and single crystal rods/plates,^10,19 and (vi) metalorganic chemical vapor deposition (MOCVD) growth of single crystalline films on sapphire substrates with c-plane¹⁹ and r-plane orientations.^20,21 Recently, we have reported a study on the MOCVD growth of single crystalline ZnGeN₂ films on c-, r-, and a-plane sapphire substrates.²² 

In this work, we have systematically investigated the MOCVD growth of ZnGeN₂ films grown on GaN templates with (0001) orientation. Key growth parameters such as growth temperature, total reactor pressure, and cationic precursor flow rates were mapped to determine their impacts on the cation stoichiometry of ZnGeN₂ films grown on closely lattice-matched GaN. The surface morphology and crystalline quality of the films showed strong correlation with the cation stoichiometry. Comprehensive material characterization indicated that the films are substantially disordered on the cation sublattice.

The ZnGeN₂ films were grown in a custom designed dual-chamber MOCVD system. GaN-on-sapphire templates with (0001) out-of-plane orientation were used as substrates. The substrates were ∼3.5 μm to 4.5 μm thick and unintentionally doped with free electron concentrations in the range of low 10¹⁷ cm⁻³. Diethylzinc (DEZn), germane (GeH₄), and ammonia (NH₃) were used as the precursors of Zn, Ge, and N, respectively. N₂ was used as the carrier gas. Samples were grown at temperatures between 600 °C and 775 °C with reactor pressures between 200 Torr and 500 Torr. The DEZn-to-GeH₄ molar flow rate ratio (II/IV ratio) ranged between 15 and 75. The NH₃ molar flow rate was kept at 178 mmol/min. The substrates were cleaned ex situ using acetone and isopropanol, rinsed with deionized water, and blown dry using nitrogen. Prior to epi-growth, the substrates were in situ annealed for 3 min at 900 °C under a combination of N₂ and NH₃ flow. Table I summarizes the growth conditions for a series of samples in this study.

The atomic percentages of Zn, Ge, and N in the ZnGeN₂ films were measured by energy dispersive x-ray spectroscopy (EDS). Surface morphologies of the films were investigated by field emission scanning electron microscopy (FESEM). The thicknesses of the films were estimated from cross-sectional SEM imaging. FEI Apreo LoVac analytical SEM was used for both the EDS measurements and the FESEM imaging. Atom probe tomography (APT) was used to investigate the compositional homogeneity of the cations using a CAMECA Local Electrode Atom Probe (LEAP) 5000 XR system. APT samples were analyzed at a base temperature of 50 K, a laser pulse energy of 15 pJ, and an average evaporation rate of 0.005 ions per laser pulse. APT needle-shaped specimens were prepared in an FEI Nova 200 Focused Ion Beam (FIB) using standard lift-out and annular milling techniques. A Bruker Icon 3 atomic force microscope (AFM) was used to measure the roughness of the films. X-ray diffraction measurements were performed to determine the crystallinity and growth direction of the films using Bruker D8 Discover XRD with the Cu Kα source. Crystalline quality was investigated by scanning tunneling electron microscope (STEM) imaging using a Thermofisher probe-corrected Titan STEM operated at 300 kV. Raman spectra were captured at room temperature using a Renishaw–Smiths detection combined Raman–IR microprobe and 785 nm excitation wavelength. Cathodoluminescence (CL) measurements were performed using a Thermo Fisher Quattro Environmental Scanning Electron Microscope (ESEM) equipped with a Horiba H-Clue CL detector. A 325 μm wavelength cw He–Cd laser was used as the excitation source for the photoluminescence (PL) measurements. The excitation laser power was ∼1.4 mW. The excitation spot size was ∼100 μm in diameter. The spectra were resolved using a 0.8 m double monochromator with 2 nm resolution and detected with a Burle C31034A UV-enhanced photon counting cooled photomultiplier tube.

To investigate the effects of MOCVD growth parameters, i.e., growth temperature (T_G), total reactor pressure (P), and group II/IV molar ratio (R_II/IV), on the stoichiometry and crystallinity of the ZnGeN₂ films, three sets of samples were grown while varying only one of these parameters at a time within a set. The total gas flow by volume was kept constant for all growth experiments. The effects of T_G, P, and R_II/IV on Zn/(Zn + Ge) composition and film thickness are plotted in Figs. 1(a)–1(c). The cation-to-anion ratios were measured to be stoichiometric, within the measurement error. The incorporation rate of Zn into the crystal depends on the concentration of Zn adatoms on the growth surface. The adatom concentration is directly proportional to the pressure of Zn in the vapor phase (⁠$PZnv$⁠) and the flow rate of Zn (J_Zn) but inversely proportional to the equilibrium vapor pressure of liquid Zn (P_Zn).²³ For the samples C, B, and A, as shown in Fig. 1(a), the Zn/(Zn + Ge) composition monotonically decreased from 0.54 to 0.45 with an increase in growth temperature from 600 °C to 700 °C. This decrease is attributed to the effect of the increase in the equilibrium vapor pressure of Zn with an increase in temperature. For example, the equilibrium vapor pressure of Zn increases to ∼75 Torr at 700 °C from ∼15 Torr at 600 °C.²⁴ The vapor pressure of Ge is ∼11 orders of magnitude less than that of Zn over this temperature range. For the samples E, D, and C, as shown in Fig. 1(b), the Zn/(Zn + Ge) composition increased from 0.5 to 0.54 with an increase in total reactor pressure from 200 Torr to 500 Torr. This effect can be ascribed to the increase in the pressure of Zn in the vapor phase (⁠$PZnv$⁠) with an increase in the total reactor pressure (P). Finally, the increase in the II/IV ratio from 15 to 25 (samples G, F, and B, respectively) led to the increase in the Zn/(Zn + Ge) composition from 0.45 to 0.5, as expected [Fig. 1(c)].

Atom probe tomography (APT) measurements were performed on a ZnGeN₂ film grown at T_G = 660 °C with P = 500 Torr and R_II/IV = 25 (sample J). The Zn/(Zn + Ge) composition measured by APT was found to be 0.47. APT provided insight on the compositional homogeneity of Zn and Ge atoms across the cross section of the ZnGeN₂ films. 3D atom maps of Zn and Ge from APT data are shown in Fig. 2(a). Zn and Ge atoms are shown in violet and green dots, respectively, in the atom maps. Regions of interest are shown in Fig. 2(a) as blue cylindrical slices, labeled i–iii. These measure 20 nm in diameter in the x–y plane and 3 nm in thickness along the z axis. These regions were selected to generate two-dimensional contour plots (2DCPs) of the Zn-to-Ge ratio, as shown in Fig. 2(b). The assessment of the 2DCPs revealed relatively homogeneous distributions with a range of 0.8–1.0 Zn/Ge. Statistical analyses were performed within the regions of interest, boxed in black in Fig. 2(a), measuring 20 × 15 nm in the xz plane and 2 nm thick, to define the likelihood of segregation or clustering of Zn and Ge. Frequency distribution analyses (FDA) allowed for the comparison of the experimental distribution and a calculated binomial distribution for each species. The resulting distributions are shown in Fig. 2(c) along with the statistical measurements of the Pearson coefficient (μ) and the probability-value (p-value). The binomial and experimental distributions are close, indicating a nearly random distribution. The μ reported here is independent of sample size and is analogous to a goodness of fit for which a value close to one indicates a complete association in the occurrence of the solute atoms and a value close to 0 indicates a random distribution. The p-value is a measure of confidence. A p-value of 0.01 indicates a confidence level of 99%. The p-value is not independent of sample size, and therefore, μ is the more reliable indicator for segregation. Both Zn and Ge demonstrate high p-values (close to 1) and low μ values, which indicate a lack of segregation or inhomogeneity. The combined analyses indicate a homogeneous distribution of Zn and Ge. Thus, there is no indication of the presence of a secondary phase, such as Zn₃N₂ or Ge₃N₄.

The growth rates of the ZnGeN₂ films were determined from cross-sectional SEM imaging. The measured values were in agreement with the estimated growth rates from the in situ reflectivity monitor. The error bars arise from the uncertainty due to the surface roughness plus the small variation in the thickness from the edge toward the center of the 2 in. wafer. With 500 Torr total pressure and a II/IV ratio of 25 (samples A–C), the growth rates of the films were close to 1.1 μm/hr. These growth rates did not vary significantly with the growth temperature [Fig. 1(a)]. At 600 °C, the growth rate remained almost constant (∼1.1 μm/hr) as the total pressure decreased from 500 Torr to 350 Torr (sample D) but dropped to ∼0.45 μm/hr at 200 Torr (sample E) [Fig. 1(b)]. On the other hand, for T_G = 650 °C and P = 500 Torr, the growth rate increased monotonically, from ∼0.9 µm/hr to 1.1 μm/hr with an increase in the II/IV ratio from 15 to 25 (samples G, F, and B).

The plan-view FESEM images of the films shown in Fig. 1 are shown in Fig. 3. The surface morphologies of the films grown under otherwise identical conditions (P = 500 Torr and R_II/IV = 25) but at T_G = 700 °C (sample A), 650 °C (sample B), and 600 °C (sample C) are plotted in Figs. 3(a)–3(c), respectively. Figures 3(d) and 3(f) show the films grown at T_G = 650 °C and P = 500 Torr [the same as the sample B in Fig. 3(b)] but with R_II/IV = 20 (sample F) and 15 (sample G), respectively. The films shown in Figs. 3(e) and 3(g) were grown with chamber pressures of 350 Torr (sample D) and 200 Torr (sample E), respectively, but otherwise under the same conditions as the sample C shown in Fig. 3(c), with T_G = 600 °C and R_II/IV = 25. From Fig. 1, it is understood that, depending on the combination of T_G, P, and R_II/IV, the grown film can be zinc-poor, stoichiometric, or zinc-rich. As seen here, the surfaces of the zinc-poor films [Figs. 3(a) and 3(f)] consist of facets aligned along certain preferred directions and the aspect ratios of the facets vary with growth conditions. In the case of zinc-rich films [Figs. 3(c) and 3(e)], the surfaces consist of crystallites of arbitrary shapes and sizes, without dominant orientations. Finally, the stoichiometric or near-stoichiometric films [Figs. 3(b), 3(d), and 3(g)] show planar surfaces, which evolved with growth temperature. Figures 3(h) and 3(i) are plan view FESEM images of two additional stoichiometric films grown at T_G = 730 °C (sample H: P = 500 Torr and R_II/IV = 45) and T_G = 775 (sample I: P = 500 Torr and R_II/IV = 75), respectively, showing larger features on the surface of the film grown at relatively higher temperatures. Figures 4(a)–4(c) show the 3D images obtained from the AFM scans of three stoichiometric ZnGeN₂ films grown at T_G = 600 °C (sample E), 650 °C (sample B), and 730 °C (sample H), respectively. The estimated film thicknesses and rms roughness values of these films were ∼0.45 μm and 2.7 nm, ∼1.1 μm and 5.4 nm, and ∼1.4 μm and 7.2 nm, respectively. Figure 4(d) is an AFM image of a stoichiometric film grown at ∼735 °C (sample K) but for only 20 min resulting in an estimated thickness of ∼0.4 μm. The rms roughness value is ∼2.6 nm, which is roughly 1/3 of that of the film grown at 730 °C but for 60 min [Fig. 4(c)]. In general, it was observed that for a similar growth temperature, the rms roughness increases as the film grows thicker. In addition, AFM images in Figs. 4(a)–4(c) indicate a gradual change in the dimensions of the surface features with the growth temperature. For 600 °C, the film surface [Fig. 4(a)] consists of needle-like features with vertical and lateral dimensions in the order of ∼10 nm and ∼200 nm, respectively. As the growth temperature increases, the feature sizes increase in both vertical and lateral dimensions. For the film grown at 730 °C, the surface is comprised of hillocks with average heights and widths of ∼30 nm and 500 nm, respectively [Fig. 4(c)]. The evolution of the needle-like features to the hillocks is likely caused by the increased adatom mobility at elevated temperatures, facilitating the coalescence of adjacent structures. It is worthwhile noting that such evolution of surface features was observed regardless of the film thickness.

STEM imaging was carried out to better understand the mechanism causing the intriguing dependence of the surface morphology of the ZnGeN₂ films on Zn/(Zn + Ge) composition. The low angle annular dark field STEM (-LAADF) images of a Zn-rich (sample C) and a Zn-poor (sample A) film are shown in Figs. 5(a) and 5(b), respectively. The cross section of the Zn-rich film in Fig. 5(a) shows a columnar morphology. The columnar growth starts right from the interface between the substrate and the epi-film. The columnar cross-sectional morphology was also observed in the case of polycrystalline ZnSnN₂ films grown by combinatorial sputtering for a Zn/(Zn + Sn) composition between 0.45 and 0.7.²⁵ On the other hand, the dominant features of the cross section of a Zn-deficient film as shown in Fig. 5(b) are tilted filament-like structures, which give rise to the surface morphology shown in Fig. 3(a). Under highly Ge-rich growth conditions, the growth of rod-like structures is observed. This phenomenon may be caused by the presence of excess Ge on the growth surface. Figures 5(c)–5(e) represent the cross-sectional STEM-LAADF images of three near-stoichiometric films grown at T_G = 600 °C (sample E), 650 °C (sample B), and 775 °C (sample I), respectively. Either the total reactor pressure (P) or the II/IV ratio (R_II/IV) was varied to obtain near-stoichiometric cation compositions in these films. The near-stoichiometric films have more uniform cross sections than do the off-stoichiometric films. As the growth temperature increases, the density of extended defects decreases. This effect is likely due to the higher mobility of the adatoms on the growth surface at elevated temperatures.

Figures 6(a)–6(c) show the XRD 2θ–ω scan profiles obtained from a stoichiometric (sample B), a Zn-rich (sample C), and a Zn-poor (sample A) ZnGeN₂ film, respectively. In all cases, the highest intensity peak corresponds to the GaN (002) plane. For the stoichiometric film [Fig. 6(a)], the XRD 2θ–ω scan profile shows the strong ZnGeN₂ (002) peak at 2θ = 34.64°,^20,22 which has comparable intensity to the GaN (002) peak. No other ZnGeN₂ peak was observed between 2θ = 20° and 90° except the ZnGeN₂ (004) peak at 2θ = 73.07°.^20,22 However, for the Zn-rich film [Fig. 6(b)], two 2θ–ω peaks at 2θ = 32.42° and 36.72° positions are obvious, whereas the one at 2θ = 36.72° position is noticeable for Zn-poor films [Fig. 6(c)]. These two peaks can be assigned to ZnGeN₂ (100) and (101) peaks, respectively, assuming a wurtzite structure.²⁶ It is likely that the intensity of the ZnGeN₂ (002) peak is low so that it overlaps with the GaN (002) shoulder in Figs. 6(b) and 6(c). The XRD 2θ–ω scan profiles in Fig. 6 indicate that stoichiometric films are single crystalline with (001) out-of-plane orientation, whereas Zn-rich or Zn-poor films are polycrystalline. Moreover, the positions of (002) and (004) peaks indicate that ZnGeN₂ has a slightly smaller lattice constant than GaN along the c-axis.

It is challenging to distinguish between the ordered (Pna2₁) and disordered (P6₃mc) ZnGeN₂ structures grown along the c-direction from the XRD 2θ–ω scan profiles because of the similar Bragg conditions of the primary planes of these polymorphs.¹⁰ The TEM nano-diffraction patterns were used to investigate the crystallinity and crystal structures of the MOCVD grown films. A 15 × 15 array of nano-diffraction patterns was captured from ∼830 nm × 830 nm area from each investigated sample. In general, for the Zn-rich or Zn-poor films, substantial variations were observed among individual diffraction patterns in the array, whereas all 225 patterns were identical for nominally stoichiometric films. Moreover, the stoichiometric films grown under different growth conditions show very similar nano-diffraction patterns. Figures 7(a)–7(c) show the average of the 225 nano-diffraction patterns captured from a Zn-rich (sample C), a Zn-poor (sample A), and a stoichiometric ZnGeN₂ (sample I) film, respectively. These position-averaged diffraction patterns indicate that the near-stoichiometric films are single crystalline, whereas the off-stoichiometric films are polycrystalline in structure. In addition, the structural variations in the Zn-rich and Zn-poor films were markedly different, an observation associated with the different cross-sectional morphologies of these films [see Figs. 5(a) and 5(b)]. Figures 7(d) and 7(e) show two simulated electron diffraction patterns calculated assuming completely disordered P6₃mc ([110] view direction) and ordered Pna2₁ ([100] view direction) structures, respectively, of ZnGeN₂. Hollow squares in Fig. 7(d) mark the space-group absences in the simulated pattern. Neither of these patterns adequately matches the experimental pattern in Fig. 7(c) to comment about the degree of disorder in the cation sublattice of the grown films. The experimental pattern is consistent with both the simulated diffraction patterns, except for the presence of the forbidden peaks, indicated by the open squares in Fig. 7(d). These forbidden peaks are also observed in the diffraction pattern of the underlying GaN template (not shown). Therefore, they most likely appear as a result of the inhomogeneous strain caused by the bending of the TEM lamella. Inhomogeneous strain can also be introduced by local variations in the distribution or ordering of the cations or by inhomogeneous stress in the lamella due to the lattice mismatch between the film and the substrate. Other important factors that can result in the appearance of additional diffraction spots are a thicker-than-ideal TEM lamella and amorphous layers on both sides of the lamella, as these conditions can potentially cause dynamical diffraction.

Figure 7(f) is a high magnification STEM image showing the alternating cation layers in a stoichiometric ZnGeN₂ film. For this lattice orientation, there should be equal distributions of Zn and Ge atoms in each cation column for both the fully ordered Pna2₁ phase and wurtzite-like random distributions of the cations. Therefore, we do not expect to obtain information on the degree of cation ordering from such an image. The contrast variation observed here could be a result of the presence of strain, crystalline defects, phase segregation, or non-uniform sample thickness.

Figure 8 shows the room temperature unpolarized Raman spectrum of a single crystalline stoichiometric ZnGeN₂ film (sample I, red curve), a Zn-rich film (sample C, blue curve), and a Zn-poor film (sample A, green curve) along with that of the GaN-on-sapphire template (black curve). A ZnGeN₂ unit cell has 78 optical phonon modes, all of which are Raman active.²⁷ None of these modes are seen here. The three narrow peaks observed in the spectrum are from phonon modes in the GaN template. In the case of the stoichiometric film, two broad features, one between 600 cm⁻¹ and 700 cm⁻¹ and the other around 830 cm⁻¹, are identified with peaks in the phonon density of states (DOSs) in ZnGeN₂.^19,28,29 These two features can also be seen, although they are less pronounced, in the spectrum from the Zn-rich film (blue curve). They are absent in the spectrum from the Zn-poor film (green curve). Cation disorder in ZnGeN₂, evident in x-ray diffraction spectra, has been shown to lead to the relaxation of the momentum conservation rule and to the corresponding appearance of DOS features in the Raman spectra.¹⁰ The absence of ZnGeN₂ Raman peaks in the spectrum in Fig. 8 is perhaps not surprising. In Ref. 10, the relative intensities of the Raman vs DOS peaks have been shown to increase for vapor–liquid–solid (VLS)-grown ZnGeN₂ on going from a growth temperature of 758–850 °C, with a corresponding change in the x-ray powder diffraction scan from disordered (P6₃mc) to ordered (Pna2₁). For the film shown in Fig. 8, the growth temperature is 775 °C, but the MOCVD growth process may be more kinetically limited than the very different, near-equilibrium VLS process, even for roughly the same growth temperatures.

Figure 9(a) shows the CL spectra of a stoichiometric ZnGeN₂ film grown at 730 °C (sample H) measured at three different temperatures (77 K, 123 K, and 298 K). The electron beam energy and beam current were set as 5 keV and 23 nA, respectively. The sampling area was the same in all cases. No CL peak was observed near the 3.4 eV predicted bandgap of orthorhombic ZnGeN₂, even at 77 K. At room temperature, a broad peak was observed centered at ∼615 nm (2.02 eV). At lower temperatures, this broad peak shows clear splitting into two peaks, with one centered at ∼560 nm (2.22 eV) and the other at ∼640 nm (1.94 eV). The intensities of both peaks increase as the temperature decreases. We attribute these peaks to yellow-band luminescence (YBL)³⁰ from the underlying GaN template and from the ZnGeN₂ film. Similar peaks were observed in the room temperature PL spectra of ZnGeN₂ films grown on sapphire substrates.²² Figure 9(b) shows the PL spectrum of the ZnGeN₂ film along with the PL spectrum of a GaN-on-sapphire template as a reference. Both spectra were measured at 80 K. From Fig. 9(b), it is evident that the peak centered at ∼630 nm in the spectrum of the film (red) is primarily associated with the defect PL from the ZnGeN₂ layer, with possibly some contribution from the defect PL from the GaN substrate, whereas the peaks at wavelengths below 400 nm correspond to the near-band edge and impurity PL from the GaN substrate. Such “yellow-band-like” defect PL has been observed previously in ZnGeN₂ grown by VLS¹⁸ and by MOCVD.^21,22 Figure 9(c) plots room temperature PL spectra collected from a Zn-poor (sample A), a stoichiometric (sample I), and a Zn-rich (sample C) film. The spectra from the Zn-poor and stoichiometric films show broad defect peaks at similar wavelengths near 625 nm, but no luminescence from the Zn-rich film was observed. This “yellow-band-like” luminescence may be related to Ge-at-Zn anti-site or Zn-vacancy-type defects.^31,32 However, further investigation such as defect spectroscopy is still required to confirm the origin of these defect levels.

In conclusion, ZnGeN₂ films were grown by MOCVD on the closely lattice-matched GaN-on-sapphire templates. The stoichiometry of the films was found to be highly dependent on the growth parameters, including growth temperature, total reactor pressure, and the group II/IV molar ratio. This work has demonstrated that stoichiometric ZnGeN₂ films on GaN can be achieved in a wide growth window by tuning combinations of these key parameters. The crystallinity and surface morphologies of the films were found to be highly dependent on the stoichiometry. APT measurements indicated that the local distribution of cations at the sub-nanometer scale is homogeneous. Near-stoichiometric films were found to be single crystalline, with the planar surface morphology, whereas the Zn-rich or Zn-poor films were polycrystalline, with surfaces consisting of crystallites and facets. Room temperature Raman spectra support the presence of significant cation disorder in the crystal. The measured CL and PL peaks show PL from the ZnGeN₂ layer reminiscent of the “yellow band” PL commonly observed in GaN.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

The authors acknowledge partial funding support from the National Science Foundation (Grant No. DMREF-SusChEM-1533957) and the U.S. Department of Energy (Grant No. DE-EE0008718). Karim, Zhao, Zhu, and Hwang also acknowledge support from the Seed Grant from the Institute for Materials Research at the Ohio State University and the Center for Emergent Materials, an NSF-funded MRSEC under Award No. DMR-1420451.