During experimentation, devices were constructed from both sputter-deposited and MBE ZnSnxGe1-xN2 alloy films. Shing, NC Coronel, NS Lewis, HA Atwater, “Fabrication and Characterization of ZnSnxGe1-xN2 Alloys for Light Absorbers” IEEE 42nd PVSC Manuscript, 2015. MBE46G. The image shows the film with a thickness of ~30 nm, with pure heteroepitaxy on the substrate and with single crystal quality.
Motivation and Background
ZnSnxGe1-xN2 alloys also have the advantage of being composed of more abundant elements than InxGa1−xN or AlxGayIn1-x-yP alloys, which can become either too difficult or expensive to produce as consumption increases. It was hypothesized that nanocrystallinity would be an effect of deposition rather than the initial lattice mismatch of the ZnSnN2 and the substrates (Table 1.1). Moreover, sputtered ZnSnxGe1-xN2 thin films of 0 ≤ x ≤ 1 also exhibit tunable band gaps, including the band gaps covering the Green Gap region.
Scope of thesis
It was also observed that at constant 175C deposition temperatures but different growth times, the surface properties became rougher as the deposition time increased. Unshaded regions grow faster than shaded regions, causing topological differences and increasing RMS roughness as deposition time increases. Higher base pressure sputtering gave less preferential orientation with (002) planes, with reflections likely from and (421) planes present in the diffraction pattern.
Degenerate semiconducting properties
To see if carriers would freeze their donor energy levels, resistance experiments were conducted at lower temperatures. However, the resistivity does not change on an order of magnitude over the measured temperature range. Looking at the general resistance equation (Equation 2.1), for mobility with tens of cm2V-1s-1 the exponential value of a degenerate carrier concentration (1020 cm-3) should omit the exponential value of the charge term q, and resistances are found in the milliohm -cm regime.
Photoconductivity
According to this mechanism, an electron from the conduction band interacts in the dark with an oxygen molecule that adsorbs to the surface. In this mechanism, a shallow trap continuously exchanges electrons with the conduction band, disrupting the formation and recombination of charge carriers in a deep trap. In addition to the shallow traps, adsorbed oxygen can also play a role in the conduction band.
Improvements in reactive RF sputtered ZnSnN 2 films
As noted in Chapter 2, one of the challenges with ZnSnN2 is the degenerate carrier concentration. ZnSnxGe1-xN2 alloys fabricated by initial reactive RF two-target Zn0.75Sn0.25 and Ge co-sputtering, although they showed higher resistances, their resistances did not exceed 500 milliohm cm - still a reflection of degenerate carrier concentrations. The sputtering plasma in the chamber was maintained with an argon:nitrogen gas flow ratio of 1:3 at a pressure of 3 mTorr.
The stoichiometry of the samples was measured using a scanning electron microscope (SEM) equipped with an energy-dispersive spectrometer (EDS). The carrier type was determined using the hot probe technique, where the samples were electrically connected to a multimeter while a heated (~150C) probe was touching the surface of the sample near one of the electrodes. Vegard's law indicates that the corresponding peak for ZnSnxGe1-xN2 alloys must lie between these two values.
The linear shift with Sn content shown in 3.1 indicates that phase segregation also does not occur for alloys with Sn atomic concentrations as low as 2%, thus extending the trend for the ZnSnxGe1-xN2 alloy range. Using the full-width-half-max of the (002) peak in a Debye-Scherrer approximation, it was deduced that the film consisted of nanocrystalline grains that were tens of nm in size, which corresponds to the average grain size observed in SEM- images of the surface of the films (Fig. 2.1). ZnGeN2 [7], in which the density of states in the conduction band contributed by Sn extends further into the band gap than the states contributed by Ge.
Variations in the measured band gaps for a given composition can be attributed to the dispersion of samples, variations in voltage, or to variations in substrates.
Non-degenerate low-tin alloys
The resistivity increased exponentially with decreasing Sn concentration to ~ 105 �-cm for ZnSnxGe1-xN2 alloys with <5% at. Thus, decrease in the produced Sn content increases the resistance of the ZnSnxGe1-xN2 alloys. The exponential increase in resistivity with reduced Sn concentration can be attributed to an increase in the band gap and to a reduction in the number of thermally activated carriers in the conduction band of the alloys.
The Sn concentration can also reflect the location of the Fermi level (EF) relative to the energy of the conduction band (Ecb) and relative to the excited carrier energies. If EF were to remain the same while Ecb was increasing, we would expect the resistivity to increase as Sn is replaced by additional Ge atoms. a) b). In addition, the resistivity of the sample with low Sn concentration varied by two orders of magnitude in the temperature range 160-300 K, while the resistivity of the sample with higher Sn concentration varied only by 100 milliohm-centimeters in the same temperature range.
These results are characteristic of semiconductors, for which the resistance increases with decreasing temperature, compared to metals, for which the resistance decreases with decreasing temperature. Furthermore, the exponential trend of the temperature resistance observed for the low Sn sample indicates that the low Sn samples behave in a non-degenerate manner, differing from the rest of the ZnSnxGe1-xN2 alloys that have a higher Sn content . Linearly fitted slopes of the data gave activation energies of 70 meV and 10 meV for carriers at 5% at.
A higher slope of 5% in the Sn trendline indicates a higher activation energy for thermalized carriers.
Spectra of ZnSnN2 could be distinguished from the quaternary ZnSnxGe1-xN2 spectra in peak shape and positioning. Raman shifts for the quaternary alloys consisted of four major peaks extending from the lowest detectable wavenumber to 1000 cm-1. By plotting peak position to Sn content (Figure 3.6), Peak 1 and Peak 2 appear to change directly with Sn content, while Peak 3 and Peak 4 do not appear to correlate as linearly.
Calculations and experimental results for Raman shifts and phonon density of states have been performed for the ternary alloys ZnSnN2 and ZnGeN2. Thus, for the defective sputtered materials, k-wave vectors are not well defined, and Raman spectra observed for ZnSnN2 are found to reflect the calculated average phonon density of states over all k-wave vectors. The Raman spectra of ZnSnxGe1-xN2 quaternary alloys also consist of broad peaks that shift with Sn content.
Previous Raman studies have only been performed on ZnSnN2 and ZnGeN2, but provide limits on the Raman shifts of the quaternary alloy. Low-tin ZnSnxGe1-xN2 alloys appear at lower wavenumbers than the peaks reported for ZnGeN2 by Viennois et al. Although vibrations are still not detectable in polycrystalline quaternary alloys, Peak 2 may include wurtzite-E2 and A1 analogs for the II-IV-V system, where nitrogen atoms primarily vibrate and cations are relatively quiescent [36].
Ultimately, the peak shifts in the Raman spectra are important because they demonstrate that in these Zn-IV-Nitride systems the decrease in Sn content and increase in Ge content gradually translates the lattice phonon behavior.
Work function can be measured using photoelectron spectroscopy (PES) or Kelvin Probe Force Microscopy (KPFM). Through PES, the work function of a material can be determined from the photoelectric effect. The material work function can be calculated relative to the work function of the probe tip.
The shift of the SEC to a lower binding energy after exposure to air represents an increasing work function. This agreement is shown for different samples in Figure 4.4, where the work functions of fresh vacuum transferred samples are compared with their values after exposure to air. Therefore, the lower work function values of fresh vacuum transferred samples best represent those of the ZnSnN2 surface.
The work functions of surfaces sputtered with argon ions vary depending on the samples and sputtering time, but do not represent the ZnSnN2 work function. Looking at the UPS spectra from the beginning of the valence band in Figure 4.7, electron counts exist from the beginning of the valence band beyond 0 eV binding energy. Although there is minor surface contamination (<4% O,C), the fresh vacuum transferred values most accurately represent the ZnSnN2 surface work function.
The work function and valence band onset, combined with band gap measurements, show the band positions for the ZnSnN2 surface versus vacuum.
XPS band alignment measurements
In this work, a series of ZnSnxGe1-xN2 alloys were grown on hexagonal-based substrates: sapphire and GaN templates on sapphire. Electron transmission diffraction simulations of wurtzite GaN showed that RHEED patterns were similar to spots from the ZnSnxGe1-xN2 films grown on sapphire (Figure 5.5). The ZnSnxGe1-xN2 RHEED reflections have similar symmetry to the simulated hexagonal GaN pattern for transmission diffraction, as expected for the ZnSnxGe1-xN2 crystal structure.
The single presence of the (002) ZnSnxGe1-xN2 peak and only its higher reflections indicate that the films are oriented. Similarly, Figure 5.9 displays the diffractograms of ZnSnxGe1-xN2 MBE films grown on hexagonal-based GaN substrates with strong Pendellosung oscillations, even with ZnSnN2 films. Figure 5.13a) and b) show high-resolution transmission electron microscopy [HRTEM] of GaN interfaces on ZnSnN2 and ZnSnxGe1-xN2.
The substrate reflections are clearly separated from the ZnSnN2 reflections, demonstrating the lattice mismatch, while the ZnSnxGe1-xN2 reflections are on the substrate peaks. The substrate polarity is thus a potential influential factor in the resulting electronic properties of ZnSnxGe1-xN2. Sputtered ZnSnxGe1-xN2 thin films have demonstrated chemical stability with various solvents, acids and bases.
Smaller concentrations of oxidized compounds (FeCp2+ and CoCp2+) cause a limiting current at negative potentials. Figure 6.4: Photoelectrochemical tests of a) ZnSnN2 and b) ZnSnxGe1-xN2 electrodes. If improvements in the photoactivity of ZnSnxGe1-xN2 films are achieved and suitable protective layers for ZnSnxGe1-xN2 films are found, Zn-IV-Nitride materials can be used as light absorbers in photoelectrochemical cells. However, the ZnSnN2 films were still degenerate and the ZnSnxGe1-xN2 films followed a similar resistivity trend as the sputtered films.
