1 eV Ga(NAsSb) grown by MOVPE using di-tertiary-butyl-arsano-amine (DTBAA)

Dilute nitrides (N), like (GaIn)(NAs) or Ga(NAsSb), are promising materials for use in various applications. The small covalent radius of N in comparison to As has a large influence on the lattice constant in a GaAs matrix. Furthermore, the strong band gap reduction of 120 - 180 meV/%N is explained by the band anti crossing model.^1–5 This strong influence on lattice constant and band gap with rather small amounts of N makes the dilute nitrides a perfect candidate for band gap and lattice constant engineering. In combination with a fourth element, one can tune the band gap and lattice constant independently. For the growth of solar cells lattice matched to GaAs or Ge, one has to reduce the strain in Ga(NAs). This can be realized by incorporating an atom with a larger covalent radius than Ga or As. Additionally, the addition of these elements have to further reduce the band gap in order to achieve the desired 1 eV material. Suitable elements would be either In (substituting Ga) or Sb/Bi (substituting As). Accordingly, it is possible to achieve the 1 eV solar subcell with (GaIn)(NAs), Ga(NAsSb), Ga(NAsBi) or a combination of these.^6–15 

In literature, Ga(NAsSb) and (GaIn)(NAsSb) materials have been reported as being lattice matched to GaAs and showing a 1 eV band gap.^16–23 All of these samples were obtained with molecular beam epitaxy (MBE). The best multi junction solar cell efficiencies containing a 1 eV subcell were achieved with a (GaIn)(NAsSb) solar subcell grown with MBE.²³ In metalorganic vapor phase epitaxy (MOVPE) dilute nitrides are grown conventionally with 1,1-dimethylhydrazine (UDMHy) in combination with tri(m)ethylgallium (TMGa, TEGa), tertiarybutylarsine (TBAs) (or Arsine), and either trimethylindium (TMIn) or tri(m)ethylantimony (TMSb, TESb). Applying TEGa, TBAs, and UDMHy, the N incorporation is strongly reduced when alloying with TMIn.^24,25 This effect is even stronger when alloying with TMSb or TESb instead of TMIn.^26,27

So far, MOVPE growth of Sb in combination with N is very complex^26–28 and no high quality 1 eV cell lattice matched to GaAs have been realized as of yet. This is related to the dramatic N incorporation reduction when Ga(NAs) is alloyed with Sb. We recently presented the novel precursor for As and N namely di-tertiary-butyl-arsano-amine (DTBAA).²⁹ It not only provides N, which incorporates 10 - 20 times more efficiently in GaAs in comparison to UDMHy, but also shows no N incorporation decrease in (GaIn)(NAs) when grown with TEGa, TMIn, and DTBAA. This leads to an increase in the N incorporation efficiency by up to 80 times in comparison to UDMHy.³⁰ This result has motivated us to study Ga(NAsSb) growth in order to realize a 1 eV solar subcell material with MOVPE using DTBAA, TEGa, and TESb.

All samples in this study were grown using an Aixtron AIX 200 GFR MOVPE machine with a horizontal reactor heated by six infrared lamps. Palladium purified H₂ was used as a carrier gas for the conventional precursors like TEGa, TESb, and TBAs, as well as the novel As/N precursor DTBAA. The reactor pressure was held constant at 50 mbar by a total gas flow of 6800 sccm. The graphite block embedded in the quartz liner is designed for a 2” substrate, which can be rotated with the gas foil rotation technique for homogeneous growth. To ensure a high quality GaAs surface, a 250 nm thick GaAs buffer was grown at 625 °C prior to each experiment. The layers of interest were grown at either 475 °C, 500 °C, or 550 °C. The temperature steps from buffer growth (625 °C) to layer growth were carried out in an ambient environment of TBAs to prevent As desorption. Roughly 10 nm thick 5QW Ga(NAsSb) structures with a 20 nm thick GaAs barrier were grown, and finished with the quaternary layer for a realistic surface roughness feedback. The Ga(NAsSb) layer was grown with TEGa, TESb, and DTBAA. No TBAs was applied, as the DTBAA provides N and As simultaneously. For the GaAs barrier growth however, TEGa and TBAs were used. Several techniques were applied for detailed material investigation. High resolution X-ray diffraction (HRXRD) using a Panalytical X’Pert Pro system with the Cu Kα1 wavelength (1.5405 Å) was utilized for strain as well as growth rate determination. In order to do this, patterns around the (004) reflex were scanned and simulated with the X’Pert Epitaxy software. A rapid thermal annealer (RTA), Jipelec JetFirst 200C, was used for post growth heat treatment to decrease the defect density and increase the optical signal of the Ga(NAsSb) material. Accordingly, samples were annealed for 10 seconds in the temperature range of 650 – 700 °C. Room temperature photoluminescence spectroscopy (RT-PL) of annealed samples was performed to determine the PL-peak position as well as PL-peak intensity. For the sample excitation a frequency doubled Nd:YAG laser with a 532 nm line was used. Additionally, a liquid N₂ cooled Ge detector in combination with a standard lock-in technique was used to obtain the PL signal, which was dispersed by a 1 m grating monochromator (THR 1000, Jobin-Yvon). With an atomic force microscope (AFM), Nanoscope IIIa, surface morphology of the Ga(NAsSb) layer was investigated. Lattice matched Ga(NAsSb) bulk layer of 140 nm thickness was investigated with secondary ion mass spectrometry (SIMS) by RTG Mikroanalyse GmbH in Berlin, detecting oxygen (O), carbon (C), nitrogen (N) and antimony (Sb). For scanning transmission electron microscopy (STEM) investigations cross-sectional samples were prepared in the [010] crystallographic direction by first grinding and polishing with a Multiprep System (Allied High Tech Systems) up to a thickness of approximately 40 μm for investigation. The final preparation step was argon ion milling utilizing a Gatan precision polishing system. Plasma cleaning (Plasma Cleaner Model 1020 Fischione Instruments) was applied for 2 min to remove parasitic hydrocarbons. The STEM investigations were performed in a double-Cs corrected JEOL JEM 2200 FS field emission STEM operating at 200 kV using the high angle annular dark field (HAADF) technique.

For a quaternary material, HRXRD analysis gives an infinite combination of possible results. Therefore, additional experimental investigations are needed to determine the composition. In our case we correlated RT-PL of annealed samples with the strain obtained from HRXRD and compared to the composition of a selected sample by SIMS. For the estimation of the PL peak energy as a function of the nitrogen and antimony contents, we apply the double-band anti crossing model suggested by Lin et al.³¹ when taking into account the quantization effects and the shift of the conduction and valence band edges due to the strain inherent in the studied QW structures.^32,33

First we will present the results at 500 °C, in particular the variation of the individual partial pressures of TESb, DTBAA and TEGa as well as V/III variation and the impact on Sb and N incorporation. We will then discuss the surface roughness obtained by AFM for crystal quality analysis. Furthermore, Ga(NAsSb) samples will be discussed grown in the temperature range of 475 – 550 °C. Finally, we will present the 475 °C results including the near 1 eV Ga(NAsSb) structure. The trends in each graph are indicated with guides to the eye.

The N and Sb incorporation versus TESb supply for two different DTBAA partial pressures is depicted in Figure 1a). One can clearly see how the N incorporation decreases nonlinearly for a given DTBAA supply when increasing TESb. The Sb incorporation itself, however, seems to increase linearly for the lower DTBAA partial pressure. The Sb incorporation slope reduces drastically with higher DTBAA supply. It is thus obvious that the Sb incorporation is strongly affected by the supplied DTBAA, and therefore either the V/V or the V/III ratio. Figure 1b) shows the surface roughness on the right and the growth rate on the left hand side. We observe a higher surface roughness for higher TESb supply. This effect is more severe for higher DTBAA partial pressure. As the DTBAA provides As and N at the same time, we have to clarify which effects lead to this Sb/N/As interaction. It is known from pure Ga(AsSb) growth that the Sb incorporation is very sensitive to the As/Sb ratio,³⁴ which we would increase in our case with higher DTBAA. Furthermore, the growth rate is decreasing due to either higher TESb gas phase supply, higher Sb incorporation or Sb surfactant effect. This will be discussed in the results of the V/V experiments depicted in figure 3a).

The DTBAA variation results are plotted in figure 2. As expected, increasing the DTBAA supply leads to a higher N incorporation. However, as mentioned previously, Sb incorporation decreases with additional DTBAA. The surface roughness increases drastically with higher DTBAA supply. Higher N concentration itself leads to a slightly worse surface quality; nevertheless, roughness values of 4 nm (RMS) cannot be explained by an N incorporation of around 2 %. As the Sb incorporation decreases, Sb is either desorbing or segregating to the surface and causing defects during the growth, thus, increasing the surface roughness. At low growth temperatures and V/III ratios, group V elements also have an impact on growth rate. Therefore, a linear increase in growth rate is observed.

To understand whether the N/Sb incorporation processes or the gas phase ratio (either V/V or V/III) is accountable for the N/Sb interaction, V/III ratio experiments are performed. Figure 3a) shows results of an experiment with a constant DTBAA/TESb ratio of 9 and constant TEGa partial pressure of 8.15E-3 mbar. The V/III ratio was varied by changing the DTBAA/TESb gas phase supply at a fixed V/V ratio, thus, increasing the DTBAA and TESb supply by the same factor. This experiment reveals that either the V/III ratio or the N incorporation process itself is responsible for the Sb incorporation decrease and not the V/V ratio. With increasing V/III ratio, the N incorporation increases linearly. Simultaneously, we observe an exponential decrease in the Sb incorporation together with a higher surface roughness. For a fixed V/V ratio we observe competing behavior in N and Sb incorporation by increasing all group V elements. Surprisingly, the growth rate increases despite a fixed TEGa supply. As the Sb incorporation is simultaneously decreasing, interactions on the surface (e.g. different surface reconstructions) may lead to a higher growth rate. At the same time the surface roughness is strongly increased, leading to the assumption of a 3D growth.

If low V/III ratios allow for sufficient Sb incorporation, while high DTBAA partial pressures are needed for N incorporation, varying TEGa may lead to an additional insight of N and Sb incorporation. Furthermore, one can distinguish whether it is the N incorporation or the V/III ratio which has a huge impact on the Sb incorporation. Therefore, the DTBAA and TESb supply were fixed while the TEGa supply was varied. Increasing TEGa supply (thus, lowering the V/III ratio) leads to a higher growth rate and slightly lower N incorporation as depicted in Figure 3b). The most important change is seen in simultaneously higher Sb incorporation. Additionally, the surface roughness decreases with lower V/III ratio (higher TEGa). The growth rate, however, does not double as we would expect by doubling the TEGa supply.

In the end, the variation of TESb, DTBAA, and TEGa makes it possible to control the Sb and N content. The challenge lies in finding the parameters as well as in understanding exactly which parameters interact with each other. All of the data in this publication shows that low V/III ratios are key for both a high quality surface as well as sufficient Sb incorporation for a given DTBAA partial pressure.

Temperature dependent investigations can be seen in figure 4. Therefore, samples are grown between 475 °C and 550 °C with a rather low V/III ratio of 2.5. One clearly sees how the N incorporation decreases drastically with higher temperature. At the same time, the Sb incorporation increases. In ternary Ga(AsSb), however, Sb incorporation decreases with higher temperature. This behavior shows that not only the gas phase ratios, but possibly both the N incorporation and temperature, have a large influence on the Sb incorporation. One certainly cannot distinguish between temperature induced precursor decomposition or temperature effect itself, but this gives us a further technique for N and Sb incorporation control. Furthermore, higher temperatures increase the growth rate by 35 %. This indicates that the effective V/III ratio changes with higher temperature, leading to a lower V/III ratio (higher TEGa decomposition with higher temperature) and therefore to a higher probability for Sb incorporation. As expected, almost no N incorporation is observed at high temperatures, underlining the need for low temperature growth for dilute nitrides.

For a 1 eV lattice matched Ga(NAsSb) on GaAs material, concentrations of around 2.4 % N and 6 % Sb are needed. To take into account both decreasing N incorporation with increasing Sb supply and the sensitivity of Sb incorporation to the V/III ratio, an initial N incorporation of 3.2 % in Ga(NAs) was chosen. These layers were realized at 475 °C with V/III ratio around 1.3 or below. Afterwards, TESb was successively added, thus, increasing the V/III ratio up to 1.5. HRXRD pattern as well as AFM micrograph of the lattice matched 1 eV Ga(NAsSb) bulk material grown with MOVPE are shown in Figure 5a). The annealed sample was investigated with RT-PL. Figure 5b) shows the PL spectrum with the PL peak position of 1.0 eV which corresponds to the Ga(NAsSb) band gap. To verify the composition and to investigate the O and C incorporation, SIMS measurements were obtained. SIMS results of the same lattice matched 1 eV structure are depicted in Figure 6. One can clearly see the Sb concentration, which is higher than the N content. The calibrated SIMS data reveal an Sb incorporation of around 5.5 % Sb and 2.2 % N with an relative error of 10 %. Unfortunately, we also incorporated C in the 10¹⁸ atoms/cm³ range. Furthermore, unintentional O incorporation due to impurities contained in the novel As/N precursor DTBAA were observed in the 10¹⁷ atoms/cm³ range. This precursor was produced on laboratory scale, no sophisticated purification steps have been applied to the synthesized precursor up to now. Furthermore, ethyl groups from the Ga and Sb source might also lead to unintentional C incorporation. The highlighted peaks occur due to regrowth of the Ga(NAsSb) sample with GaAs. Figure 7a) shows an overview image of the sample investigated by STEM. Images have been acquired at comparably low detector angles (41 - 164 mrad) of the Annular Dark Field (ADF) detector. Since N has a lower atomic number than the GaAs matrix and Sb has a higher one, their influences partially compensate each other, therefore the Ga(NAsSb) layer does not exhibit contrast in conventional high angle annular dark field images. The utilized low detector angles are sensitive to the static atomic displacements introduced by N and Sb.³⁵ Consequently, Ga(NAsSb) appears brighter than GaAs. The intensity profile which is shown as inset in Figure 7a) suggests that the lower interface between GaAs and Ga(NAsSb) is abrupt, whereas the upper interface is not quite abrupt which indicates a segregation of Sb and which will be investigated in more detail in the future. Figure 7b) shows the intensity map of the region selected in 7a) which has then been evaluated using the method developed by T. Wegele et al.³⁶ The evaluation shows the intensity fluctuation within Ga(NAsSb) layer is not significantly higher than that in GaAs layer, indicating that Ga(NAsSb) layer is also quite homogenous.

In this work we presented the incorporation behavior of N and Sb in a Ga(NAsSb) material as a function of V/III, V/V ratios, DTBAA, TESb partial pressures and growth temperature. HRXRD, AFM, and RT-PL of annealed samples were used for the detailed characterization of the N and Sb content, as well as for the growth rate and surface morphology. The obtained data shows that low V/III values are favorable for simultaneously high N and Sb incorporation. Using this knowledge a 1 eV Ga(NAsSb) material lattice matched to GaAs was grown at 475 °C, which exhibited RT-PL after annealing at 650 °C for 10 s. The high C incorporation originating from low growth temperatures and possible hydro carbon groups can be decreased with higher growth temperatures. O incorporation is believed to come from the impurities in the DTBAA, which can be solved by high quality purification in the future. For the next stage in this research, a Ga(NAsSb) 1 eV solar cell device will be developed to compare the results to existing MBE grown cells, as well as to MOVPE grown (GaIn)(NAs) cells.

This work was supported by the German Research Foundation (DFG) in the frame of the Research Training Group (GRK) 1782: “Functionalization of Semiconductors”. This project has also received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No 641899.