Combinatorial Synthesis of Magnesium Tin Nitride Semiconductors

Nitride materials feature strong chemical bonding character that leads to unique crystal structures, but many ternary nitride chemical spaces remain experimentally unexplored. The search for previously undiscovered ternary nitrides is also an opportunity to explore unique materials properties, such as transitions between cation-ordered and -disordered structures, as well as to identify candidate materials for optoelectronic applications. Here, we present a comprehensive experimental study of MgSnN2, an emerging II–IV–N2 compound, for the first time mapping phase composition and crystal structure, and examining its optoelectronic properties computationally and experimentally. We demonstrate combinatorial cosputtering of cation-disordered, wurtzite-type MgSnN2 across a range of cation compositions and temperatures, as well as the unexpected formation of a secondary, rocksalt-type phase of MgSnN2 at Mg-rich compositions and low temperatures. A computational structure search shows that the rocksalt-type phase is substantially metastable (>70 meV/atom) compared to the wurtzite-type ground state. Spectroscopic ellipsometry reveals optical absorption onsets around 2 eV, consistent with band gap tuning via cation disorder. Finally, we demonstrate epitaxial growth of a mixed wurtzite-rocksalt MgSnN2 on GaN, highlighting an opportunity for polymorphic control via epitaxy. Collectively, these findings lay the groundwork for further exploration of MgSnN2 as a model ternary nitride, with controlled polymorphism, and for device applications, enabled by control of optoelectronic properties via cation ordering.

For cation composition, Rutherford backscattering spectroscopy (RBS), a low-throughput method, was used to characterize Mg/(Mg+Sn) for three points per sample library for four libraries to generate a calibration.Glassy carbon was used as a substrate to eliminate substrate overlap with the anion peaks (SI Figure 1A); additionally, a short growth time was used so that element signals (proportional to film thickness) would be well-resolved.X-ray fluorescence (XRF) mapping was used to quickly acquire composition spectra for full sample libraries.The XRF data was then converted to Mg/(Mg+Sn) using RBS results and integrated peak areas from XRF. Figure S1B gives a comparison of RBS and calibrated XRF data for two sample libraries.The XRF calibration somewhat underestimates Mg/(Mg+Sn) at the lowest Mg composition.The effect of changing stoichiometry on crystallinity can also be observed directly using SEM (Figure S2).At ambient temperature and below Mg/(Mg+Sn) = 0.5, the films have very small grains with no preferred orientation, while above Mg = 0.5, triangular crystallites are formed that increase in size with increasing Mg content.Such crystallites are consistent with the columnar grains often observed in thin films grown by sputtering methods.The Mg ≈ 0.5 composition can also be compared across a range of temperatures: above 300 ºC, grain size increases dramatically, with the largest grains observed at 400 ºC.The 500 ºC sample returns to columnar morphology, possibly correlated to the (002) texturing observed in that library.

Figure S1 :
Figure S1: A) Example RBS data showing well-resolved signals for C, N, O, Mg, and Sn.Use of glassy carbon rather than Si eliminates overlap with the anion signals, simplifying analysis.B) Measured RBS (circles) and RBS-calibrated XRF (square shapes) data for two libraries, shown in separate colors.

Figure S2 :
Figure S2: SEM comparisons of MgSnN2 samples with different Mg/(Mg+Sn) deposited at ambient temperature (left column) and with Mg/(Mg+Sn) ≈ 0.5 deposited at different temperatures (right column).

Figure S3 :
Figure S3: Photographs of sample libraries (with Mg-rich compositions on the left-hand side of each image) showing evolution of libraries with time.Top row: high Mg/(Mg+Sn) content library, which rapidly oxidized even in a N2 environment.Bottom row: 300 ºC library showing a much smaller degree of oxidation after nine months.Library rows are 2 inches across.

Figure S4 :
Figure S4: Calculated diffraction patterns for the three cation-disordered structure types for MgSnN2, each compared with the two lowest-energy ordered structures of that type.Both Q and 2 (Cu K) are given.

Figure S6 :
Figure S6: Additional optical properties calculated from spectroscopic ellipsometry.A)  and B)  for ambient temperature sample library presented in Figure 3A.The black arrows indicate the stoichiometric sample for this library.C)  and D)  for the Mg/(Mg+Sn) ≈ 0.5 samples presented in Figure 3B.Calculated  and  for the lowest-energy cation-ordered wurtzite (SG 33) and rocksalt (SG 13) are indicated with dashed lines in A-D.E) n (solid traces) and k (dashed traces) functions for the Mg/(Mg+Ss) ≈ 0.5 samples, using the same color scale as C) and D).

Table S1 :
Predicted structures with space group name and number; polymorph energies compared across multiple levels of computation; optoelectronic structure with band gaps, absorption thresholds ( = 103 cm−1) with indirect/forbidden transitions marked with *, electron and hole effective masses, and electronic and ionic dielectric contributions.