Beryllium oxide (BeO) is a large bandgap material with extreme properties that make it an ideal gate dielectric for pairing with other wide bandgap semiconductors such as silicon carbide (SiC) and gallium nitride (GaN). In this regard, the authors have utilized x-ray photoemission spectroscopy to determine the valence band offset (VBO) between atomic layer deposited (ALD) BeO and epilayers of the cubic form of silicon carbide (3C-SiC) grown on crystalline silicon (Si) substrates. The BeO VBO with 3C-SiC epilayers grown on both Si (111) and (001) substrates was determined to be 1.6 ± 0.1 and 1.5 ± 0.1 eV, respectively. Applying the band alignment rules of transitivity and commutativity, the authors additionally determine the VBO for BeO with GaN, aluminum nitride, and hexagonal boron nitride to be 0.9 ± 0.2, 0.7 ± 0.3, and 1.0 ± 0.2 eV, respectively. Utilizing the reported bandgap for ALD BeO (8.0 ± 0.1 eV) and literature values for SiC and the group III-nitrides (III-N), the authors find a type I band alignment with conduction band offsets >1 eV in all cases. These results indicate that BeO is a promising dielectric for wide bandgap SiC and III-N high-power, high-temperature, and high-frequency device applications.
I. INTRODUCTION
Beryllium oxide (BeO) is a large bandgap dielectric (Eg > 8 eV) with many extreme properties including high dielectric constant (k = 6.7), high breakdown field (Ebd > 6 MV/cm), high thermal conductivity (κ > 15 W/m K), and high hardness (H > 30 GPa).1–6 These extreme properties make BeO an excellent choice for various high-temperature, high-power, and high-frequency device applications when paired as a gate dielectric with wide bandgap semiconductors with similar extreme properties such as silicon carbide (SiC) and gallium nitride (GaN).7–10 Such device applications, however, require large valence and conduction band offsets (CBOs) of at least 1 eV at the interface between the dielectric and semiconductor.11 To date, there have been relatively few investigations of the band alignment between BeO and narrow and wide bandgap semiconductors.8,9,12 In this regard, we have utilized x-ray photoemission spectroscopy (XPS) to measure the valence band offset (VBO) between atomic layer deposited (ALD) BeO and cubic silicon carbide (3C-SiC) epilayers grown on crystalline silicon (Si) substrates. From the band alignment rules of transitivity and commutativity, we additionally determine the VBO for BeO with III-V nitride (III-N) semiconductors GaN, aluminum nitride (AlN), hexagonal boron nitride (h-BN), and indium nitride (InN). Using previously reported values for the bandgaps of ALD BeO, 3C-SiC, and the III-Ns,1,13,14 we were able to calculate the CBO at the interface between BeO and these materials and find that the measured VBOs and calculated CBOs meet the desired >1 eV device requirement in all cases.
II. EXPERIMENT
The details of the 3C-SiC epilayer growth and the BeO ALD process have been described in detail previously.1,15 Briefly, the growth of 500 nm thick 3C-SiC epilayers on Si (001) and (111) oriented two inch diameter substrates was achieved by chemical vapor deposition using propane (C3H8) and silane (SiH4) at 1135 °C.15 Prior investigations of 3C-SiC growth on Si substrates have shown 3C-SiC (111)/Si (111) and 3C-SiC (001)/Si (001) heteroepitaxial relationships.16 Thus, the 3C-SiC epilayers grown on Si (001) and (111) substrates are assumed to have (111) and (001) orientation as well.
Prior to BeO ALD, the 3C-SiC/Si epilayers were cleaned in dilute HF to remove the native surface oxide.17 While dilute HF cleaning removes any surface SiO2, it leaves a hydrophilic OH terminated SiC surface ideal for the ALD growth of an oxide material.17–19 This is in contrast to the HF cleaning of Si which leaves a hydrophobic H terminated surface that typically exhibits delayed ALD oxide growth.20,21 The ALD BeO growth on the HF cleaned 3C-SiC/Si (001) and (111) substrates occurred at 250 °C and 0.3 Torr using Be(C2H5)2 (DEBe) and H2O as precursors and N2 as a purge gas.1 The ALD growth sequence consisted of DEBe pulse and purge times of 0.4 and 5 s, respectively, followed by H2O pulse and purge times of 0.4 and 5 s, respectively. The ALD BeO growth rate for these conditions was ∼0.8–0.9 Å/cycle. After deposition, the ALD BeO/3C-SiC/Si samples were annealed at 600 °C in an N2 environment. Such annealing has been shown to induce crystallization of the BeO film.2 We also note that Lee et al. have recently demonstrated the crystalline growth of BeO on 4H-SiC (0001) using a similar ALD process without a postdeposition anneal.22
To establish an initial estimate of the VBO at the BeO/3C-SiC interface, the method of Kraut was utilized.23 This method has been previously described in detail.24,25 Briefly, the method relies on referencing distinct core levels (CLs) in BeO and SiC to their respective valence band maxima (VBM) and then measuring the relative position of these core levels with respect to one another at their interface, as follows:
where ΔEv is the valence band offset between the two materials, (CL − VBM) is the relative position of the core level to the valence band maximum of the bulk material, and ΔCLint is the relative position of the core levels in the two materials at the interface [i.e., ΔCLint = (CLBeO – CLSiC)int].
(CL − VBM)bulk is typically measured using an uncoated substrate or a thin film substantially thicker than the photoelectron mean free path (>20 nm) to minimize contributions to the photoemission spectra from the underlying substrate/interface. In our case, we utilized the Be 1s core level from a 128 nm thick ALD BeO film grown directly on an Si (001) substrate to determine (CL − VBM)BeO.1 For (CL − VBM)SiC, we instead utilized initially the previously reported values by Seyller for the Si 2p and C 1s core levels of 98.7 ± 0.07 and 281.0 ± 0.07 eV for n-type 6H-SiC (0001).26 Additional discussion and subsequent refinement of these values will be presented later.
To determine ΔCLint, 2 nm of BeO was grown on the 3C-SiC/Si (111) and (001) substrates by ALD, and the BeO thicknesses were confirmed by spectroscopic ellipsometer measurements.1 After growth, the BeO/SiC/Si and BeO/Si samples were transferred ex situ to a Phi VersaProbe II scanning XPS system equipped with a hemispherical electron energy analyzer and a monochromated Al anode x-ray source (1486.6 eV) with a full width half maximum of <0.5 eV for the Ag 3d5/2 core level.25 All samples were floated on insulating tape using standard dual-beam charge compensation (10 eV Ar+ ion and low energy electron flood exposure). As the method of Kraut relies purely on measuring relative peak positions, no attempts to perform absolute charge corrections were made. However, the binding energy scale was calibrated using the Ag 3d5/2 and Au 4f7/2 core levels from clean samples. The peak positions for all XPS core levels were determined via curve fitting using a mixed Gaussian-Lorentzian lineshape and Shirley background with casa xps software.27 The VBM was determined via linear regression analysis of the steepest slope of the turn-on for the photoemission valence band spectra.28
III. RESULTS AND DISCUSSION
Figure 1 shows XPS spectra of the Be 1s core level and the valence band obtained from the 128 nm BeO film. The Be 1s core level was well fitted using a single mixed Gaussian-Lorentzian line shape centered at 112.5 ± 0.03 eV [see Fig. 1(a)]. The valence band spectrum shown in Fig. 1(b) is dominated by the O 2s core level located at ∼22.6 eV. Linear regression of the leading edge of the valence band photoemission turn-on [see the inset of Fig. 1(b)] places the valence band maximum at 2.3 ± 0.1 eV. Accordingly, (Be 1s – VBM)BeO was determined to be 110.2 ± 0.1 eV.
(a) XPS spectra of the Be 1s core level and (b) XPS valence band spectra for a 100 nm ALD BeO thin film grown on the Si (001) substrate. The inset of (b) focusses on the upper portion of the BeO valence band illustrating the VBM position.
(a) XPS spectra of the Be 1s core level and (b) XPS valence band spectra for a 100 nm ALD BeO thin film grown on the Si (001) substrate. The inset of (b) focusses on the upper portion of the BeO valence band illustrating the VBM position.
Figure 2 shows XPS spectra of the Be 1s, Si 2p, and C 1s core levels from the 2 nm ALD BeO on the 3C-SiC/Si (001) interface. For reasons to be clarified later, the XPS spectra in Fig. 2 were all hard shifted to align the Be 1s core level with that shown in Fig. 1. Note that this does not affect the valence band alignment calculation as only relative positions between the various core levels are measured. As mentioned previously, the Be 1s core level was well fitted by a single peak [see Fig. 2(a)]. The Si 2p core level, however, was best fitted using two peaks [see Fig. 2(b)]. One at 99.4 ± 0.03 eV attributed to Si–C bonding in SiC and another smaller one at 100.5 eV attributed to Si–O bonding at the BeO/SiC interface.29 Fitting of the C 1s core level required up to four peaks due to carbon in SiC and the presence of adventitious surface carbon with various different chemical states deposited during ex situ transfer [see Fig. 2(c)].17–19 However, the C 1s signal originating from the SiC underlayer was well separated and resolved at 281.5 ± 0.03 eV relative to the adventitious surface carbon at >283 eV. Although not shown, similar Be 1s, Si 2p, and C 1s spectra were obtained from the 2 nm ALD BeO on 3C-SiC/Si (111) interface.
XPS spectra from 2 nm of ALD BeO on 3C-SiC/Si (001). (a) Be 1s, (b) Si 2p, (c) C 1s core levels, and (d) BeO/SiC valence band.
XPS spectra from 2 nm of ALD BeO on 3C-SiC/Si (001). (a) Be 1s, (b) Si 2p, (c) C 1s core levels, and (d) BeO/SiC valence band.
Included in Fig. 2 is also the valence band spectrum obtained from the 2 nm BeO/3C-SiC (001) interface. In some cases, it is possible to distinguish the valence band edge for both materials in a valence band spectrum from their interface and thus directly observe/measure the VBO.24,25 As shown in Fig. 2(d), the valence band spectrum for the 2 nm BeO/3C-SiC (001) interface does appear to show the valence band edge for both SiC and BeO at 0.5–2.5 and 2.7–4.3 eV, respectively. Performing linear regression analysis on these portions of the valence band spectrum, we located the 3C-SiC and BeO VBM at 0.4 and 1.9 eV, respectively, suggesting a BeO/3C-SiC (001) VBO of 1.5 eV. For the BeO/3C-SiC (111) interface, a similar analysis indicated a slightly higher VBO of 1.6 eV.
To confirm the VBO values derived from the valence band spectra, the BeO/SiC VBO was next calculated using Eq. (1), the Be 1s and Si 2p peak positions from Fig. 2 and the previously determined values for (Be 1s – VBM)BeO and (Si 2p – VBM)SiC. This calculation yielded a BeO/SiC VBO value of 1.6 ± 0.1 eV for both the BeO/SiC (001) and (111) interfaces. This value is slightly higher than that determined from the VB spectra for the BeO/3C-SiC (001) interface but is in excellent agreement with the value determined for the BeO/3C-SiC (111) interface. However, a significantly higher VBO of 1.8 ± 0.1 eV was instead calculated for both interfaces using the Be 1s and C 1s core levels.
To resolve the small discrepancies between the VBO obtained by the different methods and further reduce possible errors in the measurements, we next attempted to utilize the valence band spectra acquired from both the 2 nm BeO/3C-SiC interface and the 128 nm BeO thin film to obtain the 3C-SiC valence band spectrum and thus independently determine (Si 2p – VBM)SiC and (C 1s – VBM)SiC. This was accomplished by subtracting the BeO only valence band spectrum shown in Fig. 1(a) from the BeO/SiC valence band spectrum shown in Fig. 2(d). A similar approach has been previously utilized by Miyazaki and others to obtain the valence band offset between Si and gate dielectric materials such as SiO2, SiN, and ZrO2.30–33 In our case, a rigid binding energy shift of the BeO/SiC valence band spectra was performed to align the Be 1s and O 2s core levels to that from the 128 nm BeO film on Si (001) before performing the subtraction. The BeO valence band spectrum was then scaled such that the O 2s peak intensity from the thick BeO film matched that of the 2 nm BeO/3C-SiC sample. The difference between the two spectra is shown in Fig. 3. The spectra closely resemble those previously published for 6H-SiC (0001) and 3C-SiC (001) surfaces with both the C 2s-Si 3s and C 2s-Si 3p states clearly resolved.34,35 Linear regression analysis of the leading edge of the VB spectra indicated that the VBM was located at 0.8 ± 0.1 eV for 3C-SiC (001) and 0.7 ± 0.1 eV for the 3C-SiC (111) orientation. Taking the BeO VBM to be 2.3 ± 0.1 eV indicates VBO of 1.5 ± 0.1 and 1.6 ± 0.1 eV for the BeO/3C-SiC (001) and (111) interfaces, respectively, in agreement with the values obtained directly from the interfacial valence band spectra.
XPS spectra from 3C-SiC (001) valence band obtained via subtracting the valence band spectrum obtained from a 128 nm thick BeO film from the valence band spectrum acquired from a 2 nm BeO/3C-SiC (001) interface.
XPS spectra from 3C-SiC (001) valence band obtained via subtracting the valence band spectrum obtained from a 128 nm thick BeO film from the valence band spectrum acquired from a 2 nm BeO/3C-SiC (001) interface.
Utilizing the SiC VBM position determined from Fig. 3 for 3C-SiC (111), we also arrive at values for (Si 2p – VBM)SiC and (C 1s – VBM)SiC of 98.7 ± 0.1 and 280.8 ± 0.1 eV, respectively. The former is in excellent agreement with the value initially utilized based on the results of Seyller for the 6H-SiC (0001) surface,26 while the latter value is 0.2 eV lower than reported by Seyller. However, utilizing the lower (C 1s – VBM)SiC value to calculate the BeO/SiC VBO using Eq. (1) results in a VBO of 1.6 ± 0.1 eV in excellent agreement with the value calculated using the Si 2p core level and resolves the slight disagreement previously observed. For the 3C-SiC (001) orientation, we similarly arrive at values for (Si 2p – VBM)SiC and (C 1s – VBM)SiC of 98.6 ± 0.1 and 280.7 ± 0.1 eV, respectively. While slightly lower than that determined for the (111) orientation, we arrive at a BeO/3C-SiC (001) VBO of 1.5 ± 0.1 eV when using either Si 2p or C 1s core level in Eq. (1). Based on this, we conclude that the BeO/3C-SiC (001) and (111) VBOs are 1.5 ± 0.1 and 1.6 ± 0.1 eV, respectively.
We do note that there has been a significant range in values (1–2 eV) reported for (Si 2p – VBM)SiC and (C 1s – VBM)SiC in the literature for various polytypes and orientations of SiC (see Tables I and II).29–55 Some of this variability may be attributed to differences in the line width for the x-ray source, peak fitting of the unresolved or partially resolved Si 2p3/2,1/2 spin doublet, the method utilized to locate the valence band maximum,28 surface charging, surface cleaning/preparation methods,56,57 and differences in SiC polytypes and orientations. For the latter, we note that the valence band alignment between the various polytypes of SiC has been found to be ±0.1 eV with most of the differences in bandgap being accommodated by the CBO.58–60 Therefore, the differences in polytype should be a small contribution to the variability observed in Tables I and II. The more likely source is the presence of various surface states that exist in the SiC bandgap and can confuse the location of the VBM in photoemission spectra.36 As shown by Seyller and others, such surface states are only fully passivated when SiC is annealed in H2 at high temperatures (>1000 °C) or exposed to atomic H sources to produce fully hydrogen terminated surfaces.26,61,62 In this regard, it is interesting to note that the lowest reported values for (Si 2p – VBM)SiC are generally those where the SiC surface preparation included etching or annealing in H2. The low value here in turn suggests that the ALD BeO film has similarly passivated any existing SiC surface states and may have a correspondingly low density of interfacial defect or trap states.
Summary of Si 2p – VBM values for 4H/6H-SiC (0001) and 3C-SiC (111) and (001) surfaces. Mono = monochromatic, SI = semi-insulating.
Material . | X-ray source . | Si 2p – VBM (eV) . | Reference . |
---|---|---|---|
(0001) 4H-SiC, n- and p-type | Mg Kα | 99.34 | 39 |
(0001) 6H-SiC, n-type | Mg Kα | 98.9 | 40 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.98 ± 0.13 | 29 |
(0001) 4H-SiC, n-type | Mono Al Kα | 99.06 ± 0.28 | 41 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.85 ± 0.28 | 42 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.48 ± 0.28 | 43 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.56 ± 0.28 | 44 |
(0001) 6H-SiC, n-type | Mono Al Kα | 98.8 ± 0.13 | 45 |
(0001) 6H-SiC, SI | Mono Al Kα | 100.33 | 46 |
(0001) 6H-SiC, n-type | Synchrotron | 98.7 | 26 |
(0001) 6H-SiC, n-type | Al Kα | 99.1 | 34 |
(0001) 6H-SiC, n-type | Al Kα | 99.3 ± 0.1 | 36 |
(0001) 6H-SiC, n-type | Al Kα | 99.7 | 47 |
(0001) 4H-SiC, n-type | Al Kα | 98.98 ± 0.13 | 48 |
(0001) 4H-SiC, n-type | Al Kα | 100.98 | 49 |
(111) 3C-SiC, n-type | Al Kα | 99.3 ± 0.1 | 38 |
(001) 3C-SiC, n-type | Zr Mα | 99.1 ± 0.2 | 50 |
Material . | X-ray source . | Si 2p – VBM (eV) . | Reference . |
---|---|---|---|
(0001) 4H-SiC, n- and p-type | Mg Kα | 99.34 | 39 |
(0001) 6H-SiC, n-type | Mg Kα | 98.9 | 40 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.98 ± 0.13 | 29 |
(0001) 4H-SiC, n-type | Mono Al Kα | 99.06 ± 0.28 | 41 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.85 ± 0.28 | 42 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.48 ± 0.28 | 43 |
(0001) 4H-SiC, n-type | Mono Al Kα | 98.56 ± 0.28 | 44 |
(0001) 6H-SiC, n-type | Mono Al Kα | 98.8 ± 0.13 | 45 |
(0001) 6H-SiC, SI | Mono Al Kα | 100.33 | 46 |
(0001) 6H-SiC, n-type | Synchrotron | 98.7 | 26 |
(0001) 6H-SiC, n-type | Al Kα | 99.1 | 34 |
(0001) 6H-SiC, n-type | Al Kα | 99.3 ± 0.1 | 36 |
(0001) 6H-SiC, n-type | Al Kα | 99.7 | 47 |
(0001) 4H-SiC, n-type | Al Kα | 98.98 ± 0.13 | 48 |
(0001) 4H-SiC, n-type | Al Kα | 100.98 | 49 |
(111) 3C-SiC, n-type | Al Kα | 99.3 ± 0.1 | 38 |
(001) 3C-SiC, n-type | Zr Mα | 99.1 ± 0.2 | 50 |
Summary of C 1s – VBM values for 4H/6H-SiC (0001) and 3C-SiC (111) and (001). Mono = monochromatic, SI = semi-insulating.
Material . | X-ray source . | C 1s – VBM (eV) . | Reference . |
---|---|---|---|
(0001) 6H-SiC, n-type | Mono Al Kα | 281.26 ± 0.1 | 37 |
(0001) 4H-SiC, n-type | Mono Al Kα | 281.3 ± 0.1 | 51 |
(0001) 6H-SiC, n-type | Al Kα | 281.3 | 52 |
(0001) 6H-SiC, n-type | Al Kα | 281.3 ± 0.2 | 34 and 36 |
(0001) 6H-SiC, n-type | Synchrotron | 281.0 ± 0.1 | 26 |
(000−1) 6H-SiC, n-type | Mg Kα | 281.55 ± 0.1 | 53 |
(000−1) 6H-SiC, n-type | Synchrotron | 280.08 ± 0.02 | 54 |
(111) 3C-SiC, n-type | Al Kα | 281.3 ± 0.1 | 38 |
(001) 3C-SiC, n-type | Mono Al Kα | 281.45 ± 0.1 | 55 |
Material . | X-ray source . | C 1s – VBM (eV) . | Reference . |
---|---|---|---|
(0001) 6H-SiC, n-type | Mono Al Kα | 281.26 ± 0.1 | 37 |
(0001) 4H-SiC, n-type | Mono Al Kα | 281.3 ± 0.1 | 51 |
(0001) 6H-SiC, n-type | Al Kα | 281.3 | 52 |
(0001) 6H-SiC, n-type | Al Kα | 281.3 ± 0.2 | 34 and 36 |
(0001) 6H-SiC, n-type | Synchrotron | 281.0 ± 0.1 | 26 |
(000−1) 6H-SiC, n-type | Mg Kα | 281.55 ± 0.1 | 53 |
(000−1) 6H-SiC, n-type | Synchrotron | 280.08 ± 0.02 | 54 |
(111) 3C-SiC, n-type | Al Kα | 281.3 ± 0.1 | 38 |
(001) 3C-SiC, n-type | Mono Al Kα | 281.45 ± 0.1 | 55 |
Having determined the VBO, we can next calculate the corresponding CBO for the BeO/3C-SiC interface using the value of 8.0 ± 0.1 eV previously determined for the bandgap of ALD BeO (Ref. 1) and 2.4 eV for the bandgap of 3C-SiC.13 Doing so, CBOs of 4.1 ± 0.2 and 4.0 ± 0.2 eV were calculated for both the BeO/3C-SiC (001) and (111) interfaces, respectively. As shown in Fig. 4, this results in a type I band alignment with >1 eV VBO and CBO that is ideal for high-power, high-temperature, and high-frequency device applications.
Schematic diagram for the band alignment between ALD BeO and 3C-SiC (001).
In order to determine the BeO VBO with other polytypes of SiC and the group III-nitrides, we next apply the band alignment rules of transitivity and commutativity to prior VBO measurements for 4H/6H-SiC/3C-SiC,63,64 BN/SiC,65,66 AlN/SiC,34,36 GaN/SiC,38,67 and InN/SiC (Ref. 29) interfaces. The transitivity and commutativity rules for VBOs, respectively, state that68
where 1/2 signifies the BeO/SiC interface, 2/3 signifies the SiC/X interface, and 3/1 signifies the X/BeO interface in question. Utilizing the previously reported X/SiC VBO values summarized in Table III, we calculate the BeO VBO with hexagonal BN (h-BN), AlN, GaN, and InN to be 1 ± 0.2, 0.7 ± 0.3, 0.9 ± 0.2, and 2.1 ± 0.2 eV, respectively. To determine the VBO between cubic BN (c-BN) and BeO, we first determined the VBO between 4H-SiC and c-BN to be 1.2 ± 0.2 eV utilizing the reports by Shammas et al.66 for a c-BN/nanocrystalline diamond (nc-D) interface (VBO = 0.8 ± 0.1 eV) and Goto et al.69 for a nc-D/4H-SiC interface (VBO = 0.43 eV). These results allowed us to determine that the value of BeO/c-BN VBO is 0.5 ± 0.2 eV.
Utilizing the reported bandgaps for 4H/6H-SiC and the III-Ns,13,14 we can also calculate the BeO CBO with these materials. The results are summarized in Table IV. As can be seen, a type I band alignment is present in all cases. Excluding the BeO/c-BN interface, the interfacial VBOs and CBOs also approach or exceed the >1 eV requirement for gate dielectric applications. Thus, BeO appears to also be an excellent dielectric choice for III-N based high-power, high-temperature, and high-frequency device applications. These results are particularly motivating given the recent demonstrations of epitaxial growth of BeO on GaN substrates.7,70
Summary of reported interfacial valence band offsets between SiC and III-N materials. The error bars for each entry reflect either the experimental variability reported for a single measurement or the range of values reported in which multiple investigations of the same interface have been made. h = hexagonal, w = wurtzite, c = cubic/zinc-blende, and nc-D = nanocrystalline diamond.
Interface . | VBO (eV) . | Reference . |
---|---|---|
3C-SiC/6H-SiC | 0.1 ± 0.1 | 63 |
3C-SiC/4H-SiC | 0.06 ± 0.02 | 64 |
h-BN/6H-SiC (0001) | 0.7 ± 0.2 | 65 |
w-AlN/6H-SiC (0001) | 1.1 ± 0.3 | 34 and 36 |
w-GaN/3C-SiC (111) | 0.75 ± 0.25 | 38 and 66 |
w-InN/4H-SiC (0001) | 0.55 ± 0.23 | 29 |
nc-D/4H-SiC (0001) | 0.43 | 69 |
c-BN/nc-D | 0.8 ± 0.1 | 66 |
Summary of BeO interfacial valence and conduction band offsets with SiC and III-N materials. h = hexagonal, w = wurtzite, c = cubic/zinc-blende.
Wide bandgap semiconductor . | Eg (eV) . | BeO VBO (eV) . | BeO CBO (eV) . |
---|---|---|---|
3C-SiC (001) | 2.4 | 1.5 ± 0.1 | 4.1 ± 0.2 |
3C-SiC (111) | 2.4 | 1.6 ± 0.1 | 4.0 ± 0.2 |
6H-SiC (0001) | 3.0 | 1.7 ± 0.1 | 3.3 ± 0.2 |
4H-SiC (0001) | 3.3 | 1.6 ± 0.1 | 3.2 ± 0.1 |
h-BN (0001) | 6.0 | 1.0 ± 0.2 | 1.0 ± 0.3 |
c-BN (111) | 6.4 | 0.5 ± 0.2 | 1.1 ± 0.3 |
w-GaN (0001) | 3.4 | 0.9 ± 0.3 | 3.6 ± 0.3 |
w-AlN (0001) | 6.2 | 0.7 ± 0.3 | 1.3 ± 0.3 |
w-InN (0001) | 0.7 | 2.1 ± 0.3 | 5.2 ± 0.3 |
Wide bandgap semiconductor . | Eg (eV) . | BeO VBO (eV) . | BeO CBO (eV) . |
---|---|---|---|
3C-SiC (001) | 2.4 | 1.5 ± 0.1 | 4.1 ± 0.2 |
3C-SiC (111) | 2.4 | 1.6 ± 0.1 | 4.0 ± 0.2 |
6H-SiC (0001) | 3.0 | 1.7 ± 0.1 | 3.3 ± 0.2 |
4H-SiC (0001) | 3.3 | 1.6 ± 0.1 | 3.2 ± 0.1 |
h-BN (0001) | 6.0 | 1.0 ± 0.2 | 1.0 ± 0.3 |
c-BN (111) | 6.4 | 0.5 ± 0.2 | 1.1 ± 0.3 |
w-GaN (0001) | 3.4 | 0.9 ± 0.3 | 3.6 ± 0.3 |
w-AlN (0001) | 6.2 | 0.7 ± 0.3 | 1.3 ± 0.3 |
w-InN (0001) | 0.7 | 2.1 ± 0.3 | 5.2 ± 0.3 |
IV. SUMMARY AND CONCLUSIONS
In summary, we have utilized XPS to determine the interfacial band alignment between BeO and 3C-SiC. Utilizing the band alignment rules of transitivity and commutativity, we have additionally determined the band alignment between BeO and III-N semiconductors. In all cases, we find a type I alignment with the determined interfacial VBO and CBO approaching or exceeding 1 eV. This indicates that BeO is a promising dielectric for SiC and III-N based high-temperature, high-power, and high-frequency device applications.
ACKNOWLEDGMENTS
The authors would like to acknowledge Bruce Tufts of Intel for supporting and encouraging the measurements performed at Intel and Joseph Shammas of Intel for thoughtful reading and insight during the manuscript preparation. The work was partly done at the Texas Nanofabrication Facility supported by National Science Foundation (NSF) (Grant No. NNCI-1542159).