Valence and conduction band offsets for sputtered AZO and ITO on (010) (Al_0.14Ga_0.86)₂O₃

β-Ga₂O₃ is under development for power switching and control electronics as well as solar blind UV detection [1–11]. The β polymorph of Ga₂O₃ has a bandgap of ∼4.8 eV and is commercially available in large diameter bulk and epitaxial form [1–3, 8]. The absence of p-type doping capability [12, 13] has led to a focus on vertical Schottky rectifiers and gate all-around FinFET-like devices which operate in accumulation mode in the on-state [1, 3–5, 7–10]. Additionally for lateral devices, the (Al_xGa_1−x)₂O₃/Ga₂O₃ heterostructure has been recently demonstrated using modulation doping of the barrier layer [14–21], as β-Ga₂O₃ is nonpolar. There is strong interest in these heterostructures involving β-(Al, Ga)₂O₃ monoclinic phase alloys, in which the bandgap can be varied from 4.8 to 6 eV [14–21]. Typically, the x value is between 0.12 and 0.18 in the alloys. One of the drawbacks of β-(Al_xGa_1−x)₂O₃/Ga₂O₃ heterostructures is the high contact resistances encountered when fabricating Ohmic contacts [14–20].

One potential method for reducing the contact resistance is to include a lower gap transparent conducting oxide as an interlayer between the metal and the wide bandgap semiconductor [22–27]. The most commonly used TCOs are indium tin oxide (ITO) and aluminum zinc oxide (AZO). Sputtered films of both of these materials have been reported as intermediary layers for forming Ohmic contacts on Ga₂O₃ [22, 25, 27]. The use of Pt/ITO bi-layers to β-Ga₂O₃ improved the Ohmic contact properties compared to Pt/Ti through formation of an interfacial layer with lower bandgap and higher doping concentration than the Ga₂O₃ [22]. AZO interlayers are also proven effective in lowering contact resistance of Ti/Au contacts on n-type, Si implanted β-Ga₂O₃. A specific contact resistance of 2.82 × 10⁻⁵ Ω cm² was reported for annealing at 400 °C, while Ti/Au contacts without the AZO interlayer did not lead to Ohmic behavior [25]. The band alignment at the heterointerfaces is obviously critically important in determining the favorability of carrier transport [28–39]. Currently, there have been no reports for the band alignments of these TCOs on (Al_xGa_01−x)₂O₃.

In this paper, we report on the determination of the band alignment in the AZO/(Al_0.14Ga_0.86)₂O₃ and ITO/(Al_0.14Ga_0.86)₂O₃ heterostructures, in which the TCOs were deposited by sputtering onto (Al_0.14Ga_0.86)₂O₃ grown by molecular beam epitaxy (MBE). This alloy composition is typical of that used in heterostructure transistors. The valence band offset was obtained from XPS measurements [40] and by measuring the respective bandgaps of the TCOs, we obtained the conduction band offset in the heterostructures and determined the band alignment type.

1.5 nm AZO or ITO was deposited by RF magnetron sputtering on the (Al_0.14Ga_0.86)₂O₃/Ga₂O₃ structures and thicker layers (150 nm) on quartz substrates at room temperature using 3 in. diameter composite ITO (In₂O₃/SnO₂ 90/10) or AZO (Al₂O₃/ZnO₂ 2/98 wt%) targets. The RF power was 125 W and the working pressure was 5 × 10⁻⁶ Torr in a pure Ar ambient. The DC bias on the electrode under these conditions is in the range of 30–40 V. Both thick (150 nm) and thin (1.5 nm) layers of the dielectrics were deposited to be able to measure both bandgaps and core levels on the β-(Al_0.14Ga_0.86)₂O₃. Hall measurements on the thick AZO films showed resisitivity of 7.4 Ω cm⁻¹, carrier concentration of 1.78 × 10¹⁸ cm⁻³ and Hall mobility of 26 cm² V⁻¹ s⁻¹. X-ray diffraction 2θ scans showed a (002) peak at 34.3°, indicative of a [001] preferred orientation with the c-axis perpendicular to the substrate, similar to previous reports [41]. The thick ITO, XRD showed peaks due to (400) and (222), indicating the coexistence of [100] and [111] textures [42, 43]. The resistivity was 2.6 × 10⁻¹ Ω cm, with Hall mobility of 19 cm² V⁻¹ s⁻¹ and carrier concentration of 9 × 10¹⁶ cm⁻³. For substrate cleaning pre-deposition, the following rinse sequence was employed: acetone, IPA, N₂ dry, and finally Ozone exposure for 15 min. These films were deposited onto epi (Al_0.14Ga_0.86)₂O₃ that was grown by MBE. The sample was grown by oxygen plasma-assisted MBE using a Ga flux of 6 × 10⁻⁸ Torr, Al flux of 2 × 10⁻⁸ Torr, an oxygen plasma power of 280 W and a chamber pressure of 1.2 × 10⁻⁵ Torr. Under these conditions it is possible to maintain the phase purity of the alloy [44]. These AGO (Al_0.14Ga_0.86)₂O₃ layers were doped with Si to produce an n-type carrier density of 10¹⁷ cm⁻³, with mobility of 180 cm² V⁻¹ s⁻¹ at room temperature and were 55 nm thick. The donor concentration was determined by electrochemical capacitance–voltage profiling at a frequency of 740 Hz on calibration samples and the composition was determined by x-ray diffraction on these same samples [44]. These epitaxial layers were grown on top of Sn-doped (6.3 × 10¹⁸ cm⁻³) bulk β-phase Ga₂O₃ single crystal substrates (500 μm thick) with (010) surface orientation grown by the edge-defined film-fed growth method. The heterostructure samples are shown schematically in figure 1.

Zoom In Zoom Out Reset image size

The chemical state of the ITO, AZO, and β-(Al_0.14Ga_0.86)₂O₃ and identification of peaks for high resolution analysis were obtained from XPS survey scans in a ULVAC PHI XPS system. This employs a monochromatic, Al x-ray source (energy 1486.6 eV) at a source power of 300 W. The analysis area was 10 μm in diameter with a take-off angle of 50° and an acceptance angle of ±7°. The electron pass energy was 23.5 eV for the high-resolution scans and 93.5 eV for the survey scans.

Charge compensation was achieved with a dual beam charge neutralization system with simultaneous low-energy electron and ion beams [45]. Using the position of the adventitious carbon (C–C) line in the C 1s spectra at 284.8 eV, charge correction was performed. During the measurements, the samples and electron analyzers were electrically grounded to provide a common reference Fermi level. Differential charging is a concern for semiconductor band offset measurements [45] and while use of an electron flood gun does not guarantee that differential charging is not present, our experience with oxides on conducting substrates has been that the differential charging is minimized with the use of an electron gun. Calibrations without the guns verified that was the case [36].

Reflection electron energy loss spectroscopy (REELS) was employed to measure the bandgap of the AZO and ITO [46]. By taking a linear fit to the leading plasmon peak and finding its zero energy with the background, a direct measurement of valence to conduction band energy is made. REELS spectra were obtained using a 1 kV electron beam and the hemispherical electron analyzer.

Figure 2 shows the stacked XPS survey scans of thick (150 nm) AZO and ITO, 1.5 nm AZO and ITO on β-(Al_0.14Ga_0.86)₂O₃, and an (Al_0.14Ga_0.86)₂O₃ reference sample. There is no evidence of metallic contamination from the sputtering process, which if present in sufficient quantities can form oxides that reduce the overall bandgap of the dielectrics and affect the resulting band alignment [47–50].

Zoom In Zoom Out Reset image size

We obtained the valence band maximum (VBM) from linear fitting of the leading edge of the valence band and the flat energy distribution from the XPS measurements, and finding the intersection of these two lines [40, 49]. This is shown in figure 3 for the thick AZO and ITO, and the (Al_0.14Ga_0.86)₂O₃. The VBMs were measured to be 3.0 ±0.2 eV for β-(Al_0.14Ga_0.86)₂O₃, 2.53 ± 0.4 eV for the AZO and 2.74 eV ± 0.4 eV for the ITO.

Zoom In Zoom Out Reset image size

The measured bandgaps for the AZO and ITO were 3.2 ± 0.30 eV and 3.5 ± 0.30 eV, respectively, from the REELS data of figure 4. The bandgap of the β-(Al_0.14Ga_0.86)₂O₃ was determined to be 5.0 ± 0.3 eV, from XPS O1s based electron energy loss measurements [40, 41]. This is consistent with past work on powdered samples of (Al_xGa_1−x)₂O₃ over the composition range x = 0–0.4 [51]. If we use the theoretical relationship derived by Peelaers et al [29], we would expect a bandgap of 5.14 eV at our composition of x = 0.14, close to the measured result. The difference in bandgaps between AZO and ITO and β-(Al_0.14Ga_0.86)₂O₃ are therefore 1.8 and 1.5 eV, respectively, and then it is necessary to determine how these are partitioned between valence and conduction bands.

Zoom In Zoom Out Reset image size

To determine the band alignment, we used the standard core level spectra approach due to Kraut et al [40]. This method relies on precise measurement of a core level and the valence band edge for each material investigated and the shift of the core levels when the two materials have formed the heterojunction. The equation used to calculate the offset is:

$$\begin{eqnarray*}&&{\rm{\Delta }}{E}_{{V}}=\left({E}_{{\rm{c}}{\rm{o}}{\rm{r}}{\rm{e}}}^{1}-{E}_{{\rm{V}}{\rm{B}}{\rm{M}}}^{1}\right)-\left({E}_{{\rm{c}}{\rm{o}}{\rm{r}}{\rm{e}}}^{2}-{E}_{{\rm{V}}{\rm{B}}{\rm{M}}}^{2}\right)-\left({E}_{{\rm{c}}{\rm{o}}{\rm{r}}{\rm{e}}}^{1}-{E}_{{\rm{c}}{\rm{o}}{\rm{r}}{\rm{e}}}^{2}\right).\end{eqnarray*}$$

It is important to use a well-defined core level since the offsets are small compared to the core level energy and more deviation is expected at higher core level energies.

High resolution XPS spectra of the VBM-core delta region are shown in figure 5 for the β-(Al_0.14Ga_0.86)₂O₃ and thick AZO and ITO samples. These were used to determine the selected core level peak positions. Figure 6 shows the XPS spectra for the β-(Al_0.14Ga_0.86)₂O₃ to AZO and ITO core delta regions of the heterostructure samples. These values are summarized in table 1 and were then used to calculate ΔE_v.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 7 shows the band alignment of the AZO/β-(Al_0.14Ga_0.86)₂O₃ and ITO/β-(Al_0.14Ga_0.86)₂O₃ heterostructures. Both are nested, type I systems. The valence band offset is −0.59 ± 0.10 eV and conduction band offset is −1.21 ± 0.25 eV for the AZO/β-(Al_0.14Ga_0.86)₂O₃. For the ITO/β-(Al_0.14Ga_0.86)₂O₃ heterostructure, the values are −1.18 ±0.20 eV for valence band offset and −0.32 ± 0.05 eV for the conduction band offset. These were obtained using the differences in bandgaps and the directly measured valence band offset, i.e.: ${\rm{\Delta }}{E}_{C}={E}_{g}^{{\rm{A}}{\rm{Z}}{\rm{O}}\,{\rm{o}}{\rm{r}}\,{\rm{I}}{\rm{T}}{\rm{O}}}-{E}_{g}^{{\rm{A}}{\rm{l}}{\rm{G}}{\rm{a}}{\rm{O}}}-{\rm{\Delta }}{E}_{V}.$ The band offsets are negative for both oxides on β-(Al_0.14Ga_0.86)₂O₃, ensuring good electron and hole transport and hence the ability to reduce contact resistance when used as an interlayer in metal stacks on this wide bandgap material.

Zoom In Zoom Out Reset image size

To place the work in context, figure 8 shows the reported values for band offsets of dielectrics on (Al_xGa_1−x)₂O₃. All of the reported values to date have been for oxides intended as gate dielectrics [30, 52, 53], which require a larger bandgap than the (Al_xGa_1−x)₂O₃ and typical need conduction band offsets of at least 1 eV to obtain good electron confinement. The work reported in this paper for AZO and ITO shows the negative offsets required for enhancing carrier transport across the heterointerface when used as an interlayer between the metal contact and the (Al_xGa_1−x)₂O₃.

Zoom In Zoom Out Reset image size

The band alignment at both AZO/β-(Al_0.14Ga_0.86)₂O₃ and ITO/β-(Al_0.14Ga_0.86)₂O₃ heterojunctions is a nested gap (type I) band offset. The valence and conduction band offsets are negative in both cases, and can be used to enhance carrier transport across the heterointerface. The use of the AZO and ITO interlayers may be a convenient approach for improving Ohmic contact resistance on n-type β-(Al_xGa_1−x)₂O₃.

The project or effort depicted was partially sponsored by the Department of the Defense, Defense Threat Reduction Agency, HDTRA1-17-1-011, monitored by Jacob Calkins. The content of the information does not necessarily reflect the position or the policy of the federal government, and no official endorsement should be inferred.