Ultrathin oxides (UOs) and ultrathin nitrides (UNs) play a crucial role in forming lattice-mismatched semiconductor heterostructures that are fabricated by using semiconducting grafting approach. The grafting approach has shown its great potential to realize GaN-based heterojunction bipolar transistors by fulfilling the missing high-performance p-type nitrides with other p-type semiconductors. A handful of UO and UN dielectrics readily available by atomic layer deposition (ALD) satisfy the requirements of double-sided surface passivation and quantum tunneling for semiconductor grafting. Due to the states existing between the UO or UN conduction band and that of the GaN, the ALD deposited UO or UN layer can generate significant effects on the surface band-bending of GaN. Understanding the band parameters of the interface between UO or UN and c-plane Ga-face GaN can guide the selection of interfacial dielectrics for grafted GaN-based devices. In this study, we performed x-ray photoelectron spectroscopy measurements to obtain the band-bending properties on c-plane, Ga-face GaN samples coated by different ALD cycles of ultrathin-HfO2 or ultrathin AlN. The valence band spectra of GaN coated with ultrathin-ALD–Al2O3, ALD–HfO2, or PEALD–AlN/ALD–Al2O3 were further analyzed to calculate the valence and conduction band offsets between the ALD dielectrics and the Ga-face GaN under different thicknesses and post-deposition annealing conditions of the dielectrics.

Gallium nitride (GaN) has shown its strong competency for power devices and radio frequency (RF) power device applications, benefiting from its high breakdown electric field and high electron saturation velocity.1 Unipolar GaN devices have been well developed over the last few decades.2 However, the limited electrical properties of p-type GaN, e.g., the low hole mobility, the difficulty in achieving high hole concentration and good ohmic contacts, have restricted the development of GaN-based bipolar devices.3 The recently demonstrated semiconductor grafting approach makes it possible to heterogeneously integrate high-quality p-type non-nitride single crystalline semiconductors with n-type single crystal GaN to create interfaces with an interface trap density comparable to lattice-matched epitaxy.4 The enabling interface material in the grafted heterostructures is an ultrathin dielectric, e.g., an ultrathin-Al2O3 layer, which serves as a double-sided passivation and quantum tunneling layer, and can be formed by using atomic layer deposition (ALD).4 The grafted GaN-based devices with Al2O3 as the interfacial layer have been recently reported with outstanding performance,5–7 and the effects of the Al2O3 on the surface band-bending on Ga-face GaN have revealed further details on the interface properties between the ALD–Al2O3 and Ga-face GaN.8 

Aside from ALD–Al2O3,9–12 there are several other ultrathin oxide (UO) and ultrathin nitride (UN) materials that can also potentially provide double-sided passivation and quantum tunneling functions and can be deposited by ALD to fulfill the requirements of grafting. ALD–HfO213 and plasma-enhanced-ALD–AlN (PEALD–AlN)14–19 are the alternative choices of passivation dielectrics of GaN surface among others. However, the effects of these UO and UN on the surface band-bending of GaN are unknown. Measuring the interfacial electrical parameters between the UO/UN and the GaN is of great importance to understand and optimize GaN-based grafted devices for the following reasons. First, while all UO/UN could provide sufficient passivation to the GaN surface, it is reasonable to expect that the charge carrier tunneling probabilities through the UO/UN interfacial layer may be different. Second, due to the charge trapped inside the UO/UN layer, the effects on the surface band-bending of GaN may be different. Third, the reliability of the UO/UN interface under high current flow may be different. Lastly, the UO/UN band alignment with GaN under different UO/UN treatment conditions may also be different.

A few studies on the band-bending of GaN with ALD–HfO2 or PEALD–HfO2 were reported in the past.20–22 Nevertheless, these studies lacked the variation of dielectric thickness and, in particular, involved too much thermal budget, which is not applicable and even potentially harmful to the grafted heterostructure interface.4 Notably, a recent study showed that when the ALD material thickness is below 2 nm, variations in the thickness can significantly alter the material band structure.23 Therefore, a comprehensive study of the band-bending of c-plane Ga-face GaN under the influence of other UO/UN becomes necessary for the future development of grafted GaN-based heterostructure devices.

In addition to band-bending properties of GaN, the valence band XPS spectra can also provide information about the valence band offset (VBO or ΔEV) between the Ga-face GaN and the deposited dielectric materials.24 Therefore, together with the existing data reported previously,8 the VBO between the GaN and ALD–Al2O3, ALD–HfO2, or PEALD–AlN/ALD–Al2O3 can be calculated, respectively. And, the conduction band offset (CBO or ΔEC) can be determined with the reported bandgap values of the ALD dielectrics accordingly.

In this paper, we present the XPS measurements of GaN valence band spectra on c-plane, Ga-face GaN coated with ultrathin-ALD–HfO2 or PEALD–AlN/ALD–Al2O3 with the thickness range suitable for grafting applications [Fig. 1(a)]. The post-deposition annealing (PDA) parameters are also selected with a thermal budget that is compatible with that needed for the grafting process. The surface band-bending values of GaN with various ultrathin-ALD-dielectric coatings, as well as the VBOs and CBOs between various ultrathin-ALD dielectrics and the Ga-face GaN, are determined.

FIG. 1.

(a) A schematic diagram of the XPS measurement on the ALD-dielectric-coated Ga-face GaN epi. (b) A schematic band diagram of Ga-face n-GaN surface coated with UO or UN with surface band-bending qψS illustrated. The energy difference of 0.005 eV between EF and ECBM was calculated based on the secondary ion mass spectrometry (SIMS) result.8 (c) Plot of the measured surface band-bending values qψS of the n-GaN epi as a function of the number of ALD cycles of HfO2, AlN/Al2O3, and Al2O3, respectively. For the AlN/Al2O3 points, the number of cycles of AlN is referred to. The lines between the data points are drawn to guide the view of the data. The Al2O3 data were replotted from previous work for comparison and the expected GPC of the ALD–Al2O3 was ∼0.1 nm/cycle.8 

FIG. 1.

(a) A schematic diagram of the XPS measurement on the ALD-dielectric-coated Ga-face GaN epi. (b) A schematic band diagram of Ga-face n-GaN surface coated with UO or UN with surface band-bending qψS illustrated. The energy difference of 0.005 eV between EF and ECBM was calculated based on the secondary ion mass spectrometry (SIMS) result.8 (c) Plot of the measured surface band-bending values qψS of the n-GaN epi as a function of the number of ALD cycles of HfO2, AlN/Al2O3, and Al2O3, respectively. For the AlN/Al2O3 points, the number of cycles of AlN is referred to. The lines between the data points are drawn to guide the view of the data. The Al2O3 data were replotted from previous work for comparison and the expected GPC of the ALD–Al2O3 was ∼0.1 nm/cycle.8 

Close modal

The growth conditions and material characterizations of the n-type doped GaN (n-GaN) samples are described in detail in our previous publication.8 It is noted that the n-GaN layer from the previous study and this study is the same and has a doping concentration of 1 × 1018 cm−3 (Fig. S1 in the supplementary material, Ref. 8). Before each ALD process, the GaN samples were cleaned thoroughly using the identical procedures that were described previously.8 The ALD–HfO2 was deposited on the n-GaN epi layer using tetrakis(dimethylamido)hafnium (TDMAHf) precursor and H2O vapor at the reactor temperature of 200 °C in a Savannah thermal ALD system. The ALD–HfO2 was deposited between 5 and 20 cycles, depending on the recipes. The expected growth per cycle (GPC) of HfO2 at 200 °C is ∼0.1 nm/cycle from the calibration by the ALD instrument manufacturer, though the initial cycles can be slower.25,26 The ultrathin-HfO2-coated sample was immediately subject to PDA processes and XPS measurements. The same n-GaN epi sample was reused for all the HfO2 recipes with a thorough oxide removal process after each experiment, which is the same as our previous work involving ALD–Al2O3.8 Atomic force microscopy (AFM) analyses were conducted on the sample that was treated with recipe Nos. Ref, H5, and H5a350 (Fig. S1 in the supplementary material). The root mean square roughness (Rq) of a 5 × 5 μm2 area increased slightly from 0.41 to 0.49 nm after the deposition of five cycles ALD–HfO2.

The PEALD–AlN/ALD–Al2O3 process was performed in an Oxford OpAL ALD system with the chuck temperature kept at 300 °C. First, the n-GaN samples (from the same n-GaN epi wafer as the above experiment on HfO2) were subjected to a remote-plasma pretreatment (RPP) process to remove any residual native oxide at the surface and terminate exposed Ga dangling bonds with nitrogen atoms (N).27 Subsequently, the PEALD–AlN deposition process (with 3/6/9 cycles in the respective recipes) was performed with trimethylaluminum (TMA) as the Al precursor and N2–H2 plasma as the N precursor.14 The effective GPC was pre-calibrated to be ∼0.05 nm/cycle. Immediately following the PEALD–AlN deposition and without removing the sample from the ALD chamber, ten cycles of thermal ALD–Al2O3 were deposited using TMA and H2O vapor as the precursors of Al and O, respectively, without changing the chuck temperature. The expected ALD–Al2O3 GPC is ∼0.1 nm/cycle. The ALD–Al2O3 deposition was to cover the as-grown AlN surface and prevent AlN from being further oxidized in atmosphere.14 AFM images were taken on recipe Nos. Ref and A3 (Fig. S2 in the supplementary material). The Rq of a 5 × 5 μm2 area increased from 0.33 to 0.40 nm after deposition of three cycles PEALD–AlN and ten cycles ALD–Al2O3. Transmission electron microscope (TEM) image of the PEALD–AlN/ALD–Al2O3 stack has been shown in the previous work and a reduced interface state density has been observed from capacitance–voltage (C–V) characteristics.17 

The operating parameters for XPS measurement have been described in detail in the previous work8 [Fig. 1(a)]. The spot size on the HfO2-coated sample is changed to 400 μm for better signal-to-noise ratio in this experiment. The representative survey spectra are shown in Fig. S3 in the supplementary material. Detailed error analysis of the XPS measurement, including the estimation of the error bars in Fig. 1(c), was also demonstrated in our previous work.8 

The energy difference between the valence band maximum (EVBM or EV) and the Fermi level (EF) of GaN, qϕS,GaN [Fig. 1(b)], can be obtained from the XPS spectra of GaN valence band via the method described in our previous work,8 and the corresponding band-bending, qψS, as shown in Fig. 1(b), can be determined. Figure 1(c) (also see Table I) shows the calculated qψS values from the measured results as a function of the ALD cycles for both HfO2 and AlN/Al2O3 under different PDA conditions. The band-bending value of recipe No. Ref sample, i.e., bare GaN, was remeasured and is in good agreement with our previous report.8 For the purposes of comparison, the band-bending values of GaN coated with ultrathin Al2O38 were plotted in Fig. 1(c).

TABLE I.

Summary of the c-plane, Ga-face GaN sample treatment conditions, measured S,GaN, upward band-bending energy S.

Recipe No.MaterialALD cycles (CY)Estimated thickness (nm)PDAqϕS,GaN (eV)qψS (eV)
Ref … … … 2.95 +0.45 
H5 HfO2 0.5 … 2.92 +0.47 
H5a350 HfO2 0.5 350 °C 5 min 2.77 +0.63 
H5a650 HfO2 0.5 650 °C 5 min 2.85 +0.55 
H10 HfO2 10 1.0 … 2.88 +0.52 
H15 HfO2 15 1.5 … 2.72 +0.67 
H20 HfO2 20 2.0 … 2.82 +0.58 
A3 AlN/10 CY Al2O3 1.2 … 2.76 +0.64 
A3a350 AlN/10 CY Al2O3 1.2 350 °C 5 min 2.59 +0.81 
A6 AlN/10 CY Al2O3 1.3 … 2.82 +0.58 
A6a350 AlN/10 CY Al2O3 1.3 350 °C 5 min 2.64 +0.75 
A9 AlN/10 CY Al2O3 1.5 … 2.83 +0.56 
A9a350 AlN/10 CY Al2O3 1.5 350 °C 5 min 2.61 +0.78 
Recipe No.MaterialALD cycles (CY)Estimated thickness (nm)PDAqϕS,GaN (eV)qψS (eV)
Ref … … … 2.95 +0.45 
H5 HfO2 0.5 … 2.92 +0.47 
H5a350 HfO2 0.5 350 °C 5 min 2.77 +0.63 
H5a650 HfO2 0.5 650 °C 5 min 2.85 +0.55 
H10 HfO2 10 1.0 … 2.88 +0.52 
H15 HfO2 15 1.5 … 2.72 +0.67 
H20 HfO2 20 2.0 … 2.82 +0.58 
A3 AlN/10 CY Al2O3 1.2 … 2.76 +0.64 
A3a350 AlN/10 CY Al2O3 1.2 350 °C 5 min 2.59 +0.81 
A6 AlN/10 CY Al2O3 1.3 … 2.82 +0.58 
A6a350 AlN/10 CY Al2O3 1.3 350 °C 5 min 2.64 +0.75 
A9 AlN/10 CY Al2O3 1.5 … 2.83 +0.56 
A9a350 AlN/10 CY Al2O3 1.5 350 °C 5 min 2.61 +0.78 

For the ultrathin-ALD–HfO2-coated GaN, the upward band-bending of the Ga-face GaN first increased from 0.45 to 0.67 eV with increasing number of ALD cycles from 0 to 15 and then decreased to 0.58 eV for the 20-cycle case. In comparison with the ALD–Al2O3,8 the band-bending of ultrathin-HfO2-coated GaN shows an opposite trend as a function of the ALD cycles. Especially, the upward band-bending of GaN was strengthened with the coating of HfO2. After performing a thermal annealing for the five-cycle-HfO2-coated GaN sample at 350 °C for 5 min, which is the optimal annealing conditions for the grafting process,4 the upward band-bending increased from 0.47 to 0.63 eV. The band-bending values obtained from this study indicate that employing HfO2 as the interfacial layer in grafted GaN-based heterostructures is not suitable for bipolar junction transistor applications due to a large energy barrier height formed at the base-collector junction.28Figure 1(c) also showed that at a higher annealing temperature, i.e., 650 °C for 5 min, 0.55 eV upward band-bending was observed. The larger band-bending of ultrathin-HfO2-coated n-GaN than that of bare n-GaN or ultrathin-Al2O3-coated n-GaN is not only observed in this study but also in some other publications.20–22 From the analysis in our previous study,8 the energy states between the conduction band minimum (CBM or EC) of the Al2O3 and that of GaN,29,30 which are ionized and show positive charge, can suppress the band-bending of ultrathin-Al2O3-coated n-GaN. However, the CBO between ultrathin-HfO2 and GaN (0.17–0.67 eV) is measured to be much smaller than that between ultrathin-Al2O3 and GaN (1.31–1.96 eV), which is demonstrated in details in the second part of the discussion. This smaller CBO between ultrathin-HfO2 and GaN leads to less charge compensation effect from the states between CBMs, which can explain the observed difference in n-GaN band-bending between ultrathin-Al2O3-coated samples and ultrathin-HfO2-coated ones.

Figure 1(c) also plots the band-bending values of ultrathin AlN/Al2O3-coated Ga-face GaN as a function of the AlN deposition cycles, where even larger upward band-bending was observed than that of the ultrathin-HfO2-coated GaN. For the three, six, and nine cycles of AlN, the upward band-bending values are 0.64, 0.58, and 0.56 eV, respectively. The larger band-bending than that of a bare Ga-face GaN indicates that more negative fixed charge exists at the surface after the PEALD–AlN coating. Furthermore, the band-bending increases to 0.81, 0.75, and 0.78 eV after PDA at 350 °C for 5 min for the three, six, and nine cycles, respectively. The PEALD–AlN is up to 0.45 nm after nine cycles of deposition according to the GPC of ∼0.05 nm/cycle. It will become evident from the later discussion that the contribution from this ultrathin AlN to the GaN valence band XPS spectra is indistinguishable [Fig. 2(c) and 5], indicating that the measured band-bending values in Fig. 2(b) are results of a combined effect of GaN epi and the ultrathin PEALD–AlN layer. It is, therefore, reasonable to speculate that the epitaxial relationship between the ultrathin AlN with GaN14 has strengthened the overall polarization of the sample. Moreover, the PDA has further enhanced the already strengthened polarization. With the stronger spontaneous polarization from AlN, the measured upward band-bending becomes larger. Although the AlN/Al2O3 (UN/UO) combination is also not favorable for bipolar transistor applications,28 the improved passivation effect from this dielectric stack has proved to be critical to the GaN-based unipolar devices.16 

FIG. 2.

Normalized XPS spectra of GaN valence band with (a) ALD–Al2O3 (processed from the raw data reported in earlier work); (b) ALD–HfO2; and (c) PEALD–AlN/ALD–Al2O3. The “W/PDA” means a PDA process at 350 °C for 5 min. The arrows indicate the direction of changes when increasing the number of ALD cycles.

FIG. 2.

Normalized XPS spectra of GaN valence band with (a) ALD–Al2O3 (processed from the raw data reported in earlier work); (b) ALD–HfO2; and (c) PEALD–AlN/ALD–Al2O3. The “W/PDA” means a PDA process at 350 °C for 5 min. The arrows indicate the direction of changes when increasing the number of ALD cycles.

Close modal

Inspired by the method reported by McDonnell et al.,24 we normalized the measured valence band XPS spectra of the GaN coated with various ultrathin-ALD UO/UN layers, after subtracting the Shirley backgrounds31 and correcting for the binding energy to eliminate the shift from different band-bending values. From the normalized XPS spectra shown in Fig. 2, one can clearly observe the changes among the valence band spectra due to the introduction of ultrathin-ALD UO/UN layers. The major changes occurred at high binding energies of the valence band, i.e., above 5 eV. As shown in Fig. 2, the valence band spectrum of the bare GaN has a “valley” region in the middle of the spectra, and the valley was gradually “filled” with the increase of the ALD cycles for ultrathin-ALD–Al2O3 [Fig. 2(a)] and ALD–HfO2 [Fig. 2(b)] cases. As for the ultrathin PEALD–AlN/ALD–Al2O3 case [Fig. 2(c)], the valley was partially filled after introducing the ALD layer. However, no significant difference or clear trend of valley filling was observed among the spectra with different ALD cycles of AlN. It is noted that the shapes of the “filled” spectra roughly follow that of the bare GaN, indicating that the extra electron intensities shown in these spectra come from the ALD UO/UN layers. Between the binding energy 9 and 12 eV of the spectra, it is observed that the coating of either 20-cycle Al2O3 [Fig. 2(a)] or AlN/10-cycle Al2O3 [Fig. 2(c)] has extended the spectra to higher binding energy while the 20-cycle-HfO2 [Fig. 2(b)] has narrowed the valence band spectra slightly toward the lower binding energy, both relative to that of the bare GaN case. The extension of the valence band spectra with Al2O3 is due to the spectrum contribution from the Al2O3 valence band signal. The binding energy of Al2O3 valence band XPS spectrum ranges up to ∼12 eV.32 Therefore, as the ALD–Al2O3 layer becomes thicker, the signal contribution from the valence band of Al2O3 becomes significant. The above phenomenon is related to the material property of the ALD dielectric and could be useful for future density-functional theory (DFT) study of the ultrathin-ALD-dielectric/GaN structure.

The high signal-to-noise ratio obtained from the long data acquisition time enables us to observe the valence band spectrum differences by subtracting the valence band spectrum of the bare GaN from those of the samples with ALD coatings.24 The remaining spectra after the subtraction are the valence band contributions of the ALD dielectrics. With the valence band spectrum differences around the VBM of GaN, the VBO between ALD dielectrics and the Ga-face GaN can be calculated, respectively.

Figure 3 shows the calculated valence band spectrum differences between the different ultrathin-ALD–Al2O3-coated GaNs and the bare GaN. As an examination of the binding energy correction and the normalization process, the difference between two bare GaN spectra obtained from two separate XPS measurements8 is calculated. As shown in Fig. 3(a), a white noise like spectrum is obtained as expected, which indicates the reliability of the data processing. For ALD–Al2O3 up to ten cycles, the valence band spectrum contribution from Al2O3 is nearly indistinguishable from that of the bare GaN, as shown in Figs. 3(b)3(e), which indicates that the valence band spectrum of the GaN with initial ALD deposition is still quasi-GaN. Though some distortions are observed in Figs. 3(c)3(e), which could be the indication of the effects of ALD–Al2O3, they are not significant enough to obtain the VBO under the data collection conditions. Our results are consistent with the report by Werner et al., which shows that for initial cycles of ALD–Al2O3, the oxygen/aluminum atomic ratio is significantly higher than 3/2, the stoichiometric ratio of bulk Al2O3.33 Therefore, for a small number of ALD cycles, the ALD–Al2O3 material properties could be different from the thicker Al2O3. As the number of ALD cycles increases beyond 15, the spectrum contribution from ALD–Al2O3 can be clearly extracted as shown in Figs. 3(f) and 3(g). Therefore, the energy difference between the Fermi level and VBM of ALD–Al2O3, qϕAl2O3, can, thus, be extracted from a linear fitting. The corresponding VBO between ALD–Al2O3 and Ga-face GaN is then calculated and shown in Fig. 6(b) and listed in Table II. The VBO is 1.99 eV for the 15-cycle-ALD–Al2O3 and 1.94 eV for the 20-cycle case. Both lie within the error bar from each other [Fig. 6(b)]. Although the VBO of the 15-cycle case is ∼0.5 ± 0.3 eV larger than the result obtained using a different method,22 the difference may be caused by different surface cleaning and annealing conditions.21 Considering the bandgap of ALD–Al2O3 to be 7.0 ± 0.1 eV,34 the CBO between ultrathin-ALD–Al2O3 and Ga-face GaN is about 1.61 ± 0.3 and 1.66 ± 0.3 eV for the 15-cycle case and the 20-cycle case, respectively (see Fig. S4 in the supplementary material).

FIG. 3.

GaN valence band spectrum differences: (a) between two bare GaN spectra and (b)–(e) between spectra of various ultrathin-ALD–Al2O3-coated GaN and the bare GaN spectrum, calculated from the data in Fig. 2(a).

FIG. 3.

GaN valence band spectrum differences: (a) between two bare GaN spectra and (b)–(e) between spectra of various ultrathin-ALD–Al2O3-coated GaN and the bare GaN spectrum, calculated from the data in Fig. 2(a).

Close modal
TABLE II.

Summary of the sample treatment conditions and measured VBO.

Recipe No.MaterialALD cycles (CY)Estimated thickness (nm)PDAVBO (eV)
6a Al2O3 15 1.5 … 1.99 
7a Al2O3 20 2.0 … 1.94 
H5 HfO2 0.5 … 1.04 
H5A350 HfO2 0.5 350 °C 5 min 1.47 
H5A650 HfO2 0.5 650 °C 5 min 1.39 
H10 HfO2 10 1.0 … 1.73 
H15 HfO2 15 1.5 … 1.72 
H20 HfO2 20 2.0 … 1.74 
A3 AlN/10 CY Al2O3 1.2 … 2.01 
A3A350 AlN/10 CY Al2O3 1.2 350 °C 5 min 1.77 
A6 AlN/10 CY Al2O3 1.3 … 1.83 
A6A350 AlN/10 CY Al2O3 1.3 350 °C 5 min 1.91 
A9 AlN/10 CY Al2O3 1.5 … 1.85 
A9A350 AlN/10 CY Al2O3 1.5 350 °C 5 min 2.14 
Recipe No.MaterialALD cycles (CY)Estimated thickness (nm)PDAVBO (eV)
6a Al2O3 15 1.5 … 1.99 
7a Al2O3 20 2.0 … 1.94 
H5 HfO2 0.5 … 1.04 
H5A350 HfO2 0.5 350 °C 5 min 1.47 
H5A650 HfO2 0.5 650 °C 5 min 1.39 
H10 HfO2 10 1.0 … 1.73 
H15 HfO2 15 1.5 … 1.72 
H20 HfO2 20 2.0 … 1.74 
A3 AlN/10 CY Al2O3 1.2 … 2.01 
A3A350 AlN/10 CY Al2O3 1.2 350 °C 5 min 1.77 
A6 AlN/10 CY Al2O3 1.3 … 1.83 
A6A350 AlN/10 CY Al2O3 1.3 350 °C 5 min 1.91 
A9 AlN/10 CY Al2O3 1.5 … 1.85 
A9A350 AlN/10 CY Al2O3 1.5 350 °C 5 min 2.14 
a

Recipe No. used in our previous publication.8 

For ALD–HfO2, the valence band spectrum contributions from HfO2 can be distinctly separated from the raw valence band spectra, as shown in Fig. 4. The contribution also increases with intensity when the number of ALD cycles increases from 5 to 20, as expected. The VBOs between ultrathin-ALD–HfO2 and GaN are demonstrated in Fig. 6 and listed in Table II. For the case of five cycles without PDA, the VBO between HfO2 and Ga-face GaN is 1.04 eV, which is significantly lower than the VBO values of cases with a greater number of ALD cycles. The results are consistent with a previous report, i.e., the ultrathin-ALD dielectric can change significantly in band parameters under different thicknesses.23 A PDA of 350 °C 5 min and 650 °C 5 min can increase the VBO of five-cycle-HfO2-coated GaN to 1.39 and 1.47 eV, respectively. For ALD cycle numbers larger than 5, the VBO remains at ∼1.73 eV. If considering the bandgap of ALD–HfO2 to be about 5.3–5.8 eV,35,36 the CBO between ultrathin-ALD–HfO2 and Ga-face GaN is about 0.17–0.67 eV for the cases in which the ALD cycle numbers are larger than 5. The CBO values are significantly smaller than those between ultrathin-ALD–Al2O3 and Ga-face GaN (see Fig. S4 in the supplementary material).

FIG. 4.

GaN valence band spectrum differences: (a) and (d)–(f) between spectra of various ultrathin-ALD–HfO2-coated GaN and the bare GaN spectrum and (b) and (c) between spectra of five-cycle-ALD–HfO2-coated GaN with different PDA conditions and the bare GaN spectrum, calculated from the data in Fig. 2(b).

FIG. 4.

GaN valence band spectrum differences: (a) and (d)–(f) between spectra of various ultrathin-ALD–HfO2-coated GaN and the bare GaN spectrum and (b) and (c) between spectra of five-cycle-ALD–HfO2-coated GaN with different PDA conditions and the bare GaN spectrum, calculated from the data in Fig. 2(b).

Close modal
FIG. 5.

GaN valence band spectrum differences: (a), (c), and (e) between spectra of various ultrathin PEALD–AlN/10 CY ALD–Al2O3-coated GaN and the bare GaN spectrum and (b), (d), and (f) between spectra of various ultrathin PEALD–AlN/10 CY ALD–Al2O3-coated GaN with PDA and the bare GaN spectrum, calculated from data in Fig. 2(c).

FIG. 5.

GaN valence band spectrum differences: (a), (c), and (e) between spectra of various ultrathin PEALD–AlN/10 CY ALD–Al2O3-coated GaN and the bare GaN spectrum and (b), (d), and (f) between spectra of various ultrathin PEALD–AlN/10 CY ALD–Al2O3-coated GaN with PDA and the bare GaN spectrum, calculated from data in Fig. 2(c).

Close modal
FIG. 6.

VBOs between ultrathin-ALD UO/UNs and Ga-face GaN.

FIG. 6.

VBOs between ultrathin-ALD UO/UNs and Ga-face GaN.

Close modal

For the case of PEALD–AlN/ALD–Al2O3, the valence band spectrum contributions due to the dielectric coating can also be distinctly separated from the raw valence band spectra (see Fig. 5). Nevertheless, unlike the ALD–HfO2 spectra shown in Fig. 4, the PEALD–AlN/ALD–Al2O3 spectrum intensities stay at nearly the same level among all the six treatment recipes and do not show any increase with more cycles of PEALD–AlN for up to nine cycles. This phenomenon indicates that the separated spectra shown in Fig. 5 are mostly from ALD–Al2O3, and the spectrum contribution from the ultrathin PEALD–AlN is indistinguishable. The indistinguishable contribution from the ultrathin AlN layer is speculated to be a quasi-GaN valence band spectrum due to the epitaxy AlN growth.14 The VBOs extracted from the spectra shown in Fig. 6(b) further confirms the above statement. As shown in Fig. 6(b), the VBOs of all six cases of PEALD–AlN/ALD–Al2O3 are at the same level with that of ALD–Al2O3 cases. Since only ten cycles of ALD–Al2O3 were deposited for all the PEALD–AlN/ALD–Al2O3 recipes, compared with the case of ten cycles of ALD–Al2O3 directly deposited on Ga-face GaN [Fig. 3(d)], one can see that the ultrathin PEALD–AlN on the Ga-face GaN surface can enhance the deposition of ALD–Al2O3 at the initial cycles. Since the spectrum contribution from the ultrathin PEALD–AlN is indistinguishable, the CBOs can, thus, be estimated with regards to the bandgap of ALD–Al2O334 (Fig. S4 in the supplementary material) through Eq. (1).

The measured VBOs between all tested ALD dielectrics and Ga-face GaN are summarized in Fig. 6. Because the VBMs of ALD dielectrics are extracted through linear fitting, which has not considered the uncertainty caused by the instrumental Gaussian resolution, the error in Fig. 6 is estimated to be the total Gaussian resolution σ, i.e., 0.2 eV, which was measured with the same XPS instrument.8 

Considering that n-type GaN is more prevalent than the p-type, the CBO values are more relevant to many GaN-based device applications. As discussed before, according to the VBO values shown in Fig. 6 and in Table II and based on the bandgap values of the ALD–Al2O334 and ALD–HfO2,35,36 CBO values between the ultrathin dielectrics and the GaN can be estimated by

CBO=EG,UO/UNEG,GaNVBO,
(1)

where EG,UO/UN is the bandgap of the ALD UO/UN and EG,GaN is the bandgap of the GaN, i.e., 3.4 eV.37 The calculated CBO results are plotted in Fig. S4 in the supplementary material.

In this paper, the surface band-bending properties of c-plane, Ga-face GaN coated with ultrathin-ALD–HfO2 or PEALD–AlN/ALD–Al2O3 were characterized. For ALD–HfO2 treatment recipes, the band-bending first increases from 0.45 eV of the bare Ga-face GaN to 0.67 eV of 15-cycle-ALD–HfO2-coated and starts to decrease with further ALD cycles. For PEALD–AlN/ALD–Al2O3 treatment recipes, the band-bending is enhanced due to the enhanced polarization from the ultrathin PEALD–AlN, making a ∼0.59 eV band-bending after deposition and ∼0.78 eV after PDA at 350 °C for 5 min. The VBOs and CBOs between the ALD dielectrics and Ga-face GaN were further extracted. The band parameters measured in this work could guide the selection of appropriate interfacial ALD dielectric for GaN-based grafted heterojunction devices.

See the supplementary material for AFM images, XPS survey spectra, and the summary of CBO data.

This work was supported by Air Force Office of Scientific Research under Grant No. FA9550-21-1-0081.

The authors have no conflicts to disclose.

Jiarui Gong: Conceptualization (equal); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Writing – original draft (lead). Zheyang Zheng: Methodology (supporting); Writing – review & editing (supporting). Daniel Vincent: Methodology (supporting). Jie Zhou: Methodology (supporting). Jisoo Kim: Methodology (supporting). Donghyeok Kim: Methodology (supporting). Tien Khee Ng: Methodology (supporting); Resources (supporting); Writing – review & editing (supporting). Boon S. Ooi: Methodology (supporting); Resources (supporting); Supervision (supporting). Kevin J. Chen: Methodology (supporting); Supervision (supporting); Validation (supporting). Zhenqiang Ma: Conceptualization (equal); Funding acquisition (lead); Project administration (lead); Resources (lead); Supervision (lead); Validation (lead); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
J.
Tsao
 et al, “
Ultrawide-bandgap semiconductors: Research opportunities and challenges
,”
Adv. Electron. Mater.
4
(
1
),
1600501
(
2018
).
2.
M.
Asif Khan
,
J.
Kuznia
,
D.
Olson
,
W.
Schaff
,
J.
Burm
, and
M.
Shur
, “
Microwave performance of a 0.25 μm gate AlGaN/GaN heterostructure field effect transistor
,”
Appl. Phys. Lett.
65
(
9
),
1121
1123
(
1994
).
3.
L. S.
McCarthy
 et al, “
GaN HBT: Toward an RF device
,”
IEEE Trans. Electron Devices
48
(
3
),
543
551
(
2001
).
4.
D.
Liu
 et al, arXiv:1812.10225 (
2018
).
5.
D.
Liu
 et al, “
229 nm UV LEDs on aluminum nitride single crystal substrates using p-type silicon for increased hole injection
,”
Appl. Phys. Lett.
112
(
8
),
081101
(
2018
).
6.
D.
Liu
 et al, “
226 nm AlGaN/AlN UV LEDs using p-type Si for hole injection and UV reflection
,”
Appl. Phys. Lett.
113
(
1
),
011111
(
2018
).
7.
S. J.
Cho
 et al, “
P-type silicon as hole supplier for nitride-based UVC LEDs
,”
New J. Phys.
21
(
2
),
023011
(
2019
).
8.
J.
Gong
 et al, “
Influences of ALD Al2O3 on the surface band-bending of c-plane, Ga-face GaN
,”
Jpn. J. Appl. Phys.
61
(
1
),
011003
(
2021
).
9.
H.
Wang
,
J. W.
Chung
,
X.
Gao
,
S.
Guo
, and
T.
Palacios
, “
Al2O3 passivated InAlN/GaN HEMTs on SiC substrate with record current density and transconductance
,”
Phys. Status Solidi C
7
(
10
),
2440
2444
(
2010
).
10.
Z.
Liu
,
G.
Ng
,
H.
Zhou
,
S.
Arulkumaran
, and
Y.
Maung
, “
Reduced surface leakage current and trapping effects in AlGaN/GaN high electron mobility transistors on silicon with SiN/Al2O3 passivation
,”
Appl. Phys. Lett.
98
(
11
),
113506
(
2011
).
11.
Y.
Yuan-Zheng
,
H.
Yue
,
F.
Qian
,
Z.
Jin-Cheng
,
M.
Xiao-Hua
, and
N.
Jin-Yu
, “
GaN MOS-HEMT using ultra-thin Al2O3 dielectric grown by atomic layer deposition
,”
Chin. Phys. Lett.
24
(
8
),
2419
(
2007
).
12.
Y.
Yuan-Zheng
,
H.
Yue
,
Z.
Jin-Cheng
,
F.
Qian
,
N.
Jin-Yu
, and
M.
Xiao-Hua
, “
A study on Al2O3 passivation in GaN MOS-HEMT by pulsed stress
,”
Chin. Phys. B
17
(
4
),
1405
(
2008
).
13.
Y.
Chang
 et al, “
Structural and electrical characteristics of atomic layer deposited high κ HfO2 on GaN
,”
Appl. Phys. Lett.
90
(
23
),
232904
(
2007
).
14.
S.
Huang
,
Q.
Jiang
,
S.
Yang
,
C.
Zhou
, and
K. J.
Chen
, “
Effective passivation of AlGaN/GaN HEMTs by ALD-grown AlN thin film
,”
IEEE Electron Device Lett.
33
(
4
),
516
518
(
2012
).
15.
S.
Huang
,
Q.
Jiang
,
S.
Yang
,
Z.
Tang
, and
K. J.
Chen
, “
Mechanism of PEALD-grown AlN passivation for AlGaN/GaN HEMTs: Compensation of interface traps by polarization charges
,”
IEEE Electron Device Lett.
34
(
2
),
193
195
(
2013
).
16.
C.
Liu
,
S.
Liu
,
S.
Huang
, and
K. J.
Chen
, “
Plasma-enhanced atomic layer deposition of AlN epitaxial thin film for AlN/GaN heterostructure TFTs
,”
IEEE Electron Device Lett.
34
(
9
),
1106
1108
(
2013
).
17.
S.
Liu
 et al, “
Al2O3/AlN/GaN MOS-channel-HEMTs with an AlN interfacial layer
,”
IEEE Electron Device Lett.
35
(
7
),
723
725
(
2014
).
18.
T.-E.
Hsieh
 et al, “
Gate recessed quasi-normally OFF Al2O3/AlGaN/GaN MIS-HEMT with low threshold voltage hysteresis using PEALD AlN interfacial passivation layer
,”
IEEE Electron Device Lett.
35
(
7
),
732
734
(
2014
).
19.
S.
Yang
 et al, “
Investigation of SiN x and AlN passivation for AlGaN/GaN high-electron-mobility transistors: Role of interface traps and polarization charges
,”
IEEE J. Electron Devices Soc.
8
,
358
364
(
2020
).
20.
T.
Cook
, Jr.
,
C.
Fulton
,
W.
Mecouch
,
R.
Davis
,
G.
Lucovsky
, and
R.
Nemanich
, “
Band offset measurements of the GaN (0001)/HfO2 interface
,”
J. Appl. Phys.
94
(
11
),
7155
7158
(
2003
).
21.
J.
Yang
,
B. S.
Eller
,
C.
Zhu
,
C.
England
, and
R. J.
Nemanich
, “
Comparative band alignment of plasma-enhanced atomic layer deposited high-k dielectrics on gallium nitride
,”
J. Appl. Phys.
112
(
5
),
053710
(
2012
).
22.
J.
Yang
,
B. S.
Eller
, and
R. J.
Nemanich
, “
Surface band bending and band alignment of plasma enhanced atomic layer deposited dielectrics on Ga-and N-face gallium nitride
,”
J. Appl. Phys.
116
(
12
),
123702
(
2014
).
23.
M.
Si
 et al, “
Why In2O3 can make 0.7 nm atomic layer thin transistors
,”
Nano Lett.
21
(
1
),
500
506
(
2020
).
24.
S.
McDonnell
 et al, “
Hfo2 on MoS2 by atomic layer deposition: Adsorption mechanisms and thickness scalability
,”
ACS Nano
7
(
11
),
10354
10361
(
2013
).
25.
R. L.
Puurunen
, “
Growth per cycle in atomic layer deposition: A theoretical model
,”
Chem. Vap. Deposition
9
(
5
),
249
257
(
2003
).
26.
R. L.
Puurunen
, “
Analysis of hydroxyl group controlled atomic layer deposition of hafnium dioxide from hafnium tetrachloride and water
,”
J. Appl. Phys.
95
(
9
),
4777
4786
(
2004
).
27.
S.
Yang
 et al, “
High-quality interface in Al2O3/GaN/GaN/AlGaN/GaN MIS structures with in situ pre-gate plasma nitridation
,”
IEEE Electron Device Lett.
34
(
12
),
1497
1499
(
2013
).
28.
J.
Gong
 et al, arXiv:2109.04597 (
2021
).
29.
M.
Esposto
,
S.
Krishnamoorthy
,
D. N.
Nath
,
S.
Bajaj
,
T.-H.
Hung
, and
S.
Rajan
, “
Electrical properties of atomic layer deposited aluminum oxide on gallium nitride
,”
Appl. Phys. Lett.
99
(
13
),
133503
(
2011
).
30.
M.
Matys
,
R.
Stoklas
,
M.
Blaho
, and
B.
Adamowicz
, “
Origin of positive fixed charge at insulator/AlGaN interfaces and its control by AlGaN composition
,”
Appl. Phys. Lett.
110
(
24
),
243505
(
2017
).
31.
D. A.
Shirley
, “
High-resolution x-ray photoemission spectrum of the valence bands of gold
,”
Phys. Rev. B
5
(
12
),
4709
(
1972
).
32.
J.
Tao
,
J.
Chai
,
Z.
Zhang
,
J.
Pan
, and
S.
Wang
, “
The energy-band alignment at molybdenum disulphide and high-k dielectrics interfaces
,”
Appl. Phys. Lett.
104
(
23
),
232110
(
2014
).
33.
F.
Werner
 et al, “
Electronic and chemical properties of the c-Si/Al2O3 interface
,”
J. Appl. Phys.
109
(
11
),
113701
(
2011
).
34.
E. O.
Filatova
and
A. S.
Konashuk
, “
Interpretation of the changing the band gap of Al2O3 depending on its crystalline form: Connection with different local symmetries
,”
J. Phys. Chem. C
119
(
35
),
20755
20761
(
2015
).
35.
H.
Yu
 et al, “
Energy gap and band alignment for (HfO2)x (Al2O3)1−x on (100) Si
,”
Appl. Phys. Lett.
81
(
2
),
376
378
(
2002
).
36.
M. C.
Cheynet
,
S.
Pokrant
,
F. D.
Tichelaar
, and
J.-L.
Rouvière
, “
Crystal structure and band gap determination of HfO2 thin films
,”
J. Appl. Phys.
101
(
5
),
054101
(
2007
).
37.
S.
Pearton
,
J.
Zolper
,
R.
Shul
, and
F.
Ren
, “
GaN: Processing, defects, and devices
,”
J. Appl. Phys.
86
(
1
),
1
78
(
1999
).

Supplementary Material