Electronic and chemical structure of the H2O/GaN(0001) interface under ambient conditions

We employed ambient pressure X-ray photoelectron spectroscopy to investigate the electronic and chemical properties of the H2O/GaN(0001) interface under elevated pressures and/or temperatures. A pristine GaN(0001) surface exhibited upward band bending, which was partially flattened when exposed to H2O at room temperature. However, the GaN surface work function was slightly reduced due to the adsorption of molecular H2O and its dissociation products. At elevated temperatures, a negative charge generated on the surface by a vigorous H2O/GaN interfacial chemistry induced an increase in both the surface work function and upward band bending. We tracked the dissociative adsorption of H2O onto the GaN(0001) surface by recording the core-level photoemission spectra and obtained the electronic and chemical properties at the H2O/GaN interface under operando conditions. Our results suggest a strong correlation between the electronic and chemical properties of the material surface, and we expect that their evolutions lead to significantly different properties at the electrolyte/electrode interface in a photoelectrochemical solar cell.

. Photoemission spectra of Ga 3d obtained under several experimental conditions: as received, pristine (UHV), and 0.1 mbar of H2O at room temperature (RT), 373 K, and 773 K. Note: The binding energy scale was calibrated to the N 1s of N-Ga bond at the BE of 397.8 eV.
Photoemission spectra in the isothermal conditions were stabilized within ~10 min after introducing H2O vapor at particular pressures. The data acquisition time typically was about 0.5 h, and for higher H2O pressures (~5 mbar), up to 1 h. No noticeable time-dependent changes were observed in the photoemission spectra at room temperature. In isobaric conditions, the spectra were recorded after annealing the sample at a rate of 5 K/min, followed by a 20-min period for stabilization. We observed minor time-dependent variations in the intensities of spectral peaks (within 10%) in the isobaric experiments. However, such variations influenced the relative quantity of different chemical species formed at the H2O/GaN interface, not the types of these species. The photoemission spectra of Ga 3d are shown in Figure S1.

II.
Surface pretreatment An undoped GaN(0001) wafer was purchased from MTI, USA (Ga-face, Resistivity<0.5 Ohmcm). The GaN(0001) wafer was cleaned in a preparation chamber of the AP-XPS system using a procedure reported by Burmudez et al., 1-3 involving cycles of 0.5 keV N2 + bombardment at an angle of 45º with respect to the surface normal for contaminant removal, followed by annealing to 1173 K in a N2 pressure of 3×10 -7 mbar for surface restoration. 4,5 After such pretreatments, we observed a sharp low energy electron diffraction (LEED) pattern that suggested the formation of a GaN(0001)-(1×1) reconstruction. Immediately after cleaning and recording LEED patterns in the preparation chamber, the GaN crystal was transferred into an adjacent XPS analysis chamber to confirm the absence of any surface contaminants. We monitored the crystal temperature using a K-type (chromel-alumel) thermocouple sandwiched between the GaN(0001) wafer and a molybdenum sample holder. Prior to measurement, the thermocouple was calibrated using a Lumasense Pyrometer. The photoemission spectra of C 1s were recorded frequently, and no carbon contamination was detected throughout the experimental procedures. Deionized water (Milli-Q, Millipore) was freeze-pump-thawed several times to remove gaseous contaminants prior to the introduction of vapor into the reaction cell.
Single crystals of Cu(111), Ag(111), Pt(111), and Au(111) were purchased from Goodfellow, USA, and they were cleaned using standard procedures before being introduced into the reaction cell as reference samples to create a standard working equation for the work function measurements. In order to clean these single metal crystals, cycles of Ar + bombardment were followed by annealing. The following parameters were used: 1 keV and 800 K (Cu(111)), 1 keV and 800 K (Ag(111)), 1.25 keV and 1100 K (Pt(111)), and 0.5 keV and 1100 K (Au(111)). Surface cleanliness also was monitored by XPS. No contaminants were observed within the detection limit in the C 1s, N 1s, or O 1s regions in the survey spectra. Argon gas (99.999% pure) was used to probe work function, and was used as received. A residual gas analyzer (RGA) connected to a differential pumping stage monitored the purity of the Ar and the H2O vapor in the reaction cell. No contaminants were measured above the detection limit of the RGA.

III. Surface analysis
Spectral analysis: The photoemission energy scales were referenced to the Fermi edge of clean Au(111) and were recalibrated on a weekly basis. Samples were measured in the dark to minimize the surface photovoltage (SPV) effect. However, in order to exclude any band bending effect, we used the Ga-N component with a binding energy (BE) of 397.8 eV, 6,7 for BE scale calibration. Therefore it is important to note that we used two methods for BE scale calibration, in which the BE scale was referred either to the N 1s of the N-Ga bond at 397.8 eV or to the Fermi edge of pristine Au(111). In order to identify the chemical shifts the calibration method using N 1s was performed, since chemical species were assigned based on known BE values. 6,7 However this method excludes the band bending effect, which is different from the chemical shifts and causes a shift of all peaks in the spectra. The analyses of the photoemission spectra were performed using CasaXPS software by fitting a Voigt function, which was obtained by convoluting Gaussian and Lorentzian functions, with a defined BE, full width at half maximum (FWHM), and a Gaussian:Lorentzian mixing ratio of 70:30. A flexibility of 0.1 -0.2 eV for the BE and FWHM was used for peak fitting. The FWHM of the Au 4f7/2 peak was 0.5 eV with a pass energy of 20 eV and the inherent lifetime broadening of this peak has been reported to be ~0.3 eV. 8 The BEs of the Ga-N component in Figure 2c were obtained by fitting the Ga 3d spectra with multiple components, including spin-orbital splitting and formation of oxides and hydroxyls as shown in Figure S2. Figure S2. Peak fitting of the Ga 3d photoemission spectra under two representative conditions: a) pristine (UHV) and b) 0.1 mbar of H2O at 773 K. A and B represent Ga 3d5/2 and Ga 3d3/2 due to spin-orbital splitting for the Ga-N component, respectively, and C represents Ga oxidation/hydroxylation without identifying specific surface species. Note: The binding energy scale was calibrated to the N 1s of N-Ga bond at the BE of 397.8 eV. Estimates of work function: Changes in surface work functions were determined from the photoemission spectra of gas-phase Ar at the near-surface region of the GaN(0001) wafer. This method has been used previously for liquid surfaces 9,10 and nanoparticles. 11,12 In this study, we estimated the work functions of GaN(0001) from BE shifts in the Ar 2p3/2 spectra recorded in the reaction cell at an Ar pressure of 0.5 mbar. The changes in the BE of the near-surface Ar atoms were correlated to the work function of the material surface. 11,12 By using reference samples for which the work functions are well-known, the BE of Ar can be converted directly into the work function of the GaN(0001) surface. Single crystals of Cu, Ag, Pt, and Au with (111) orientations have been reported to have work functions of 4.85 ± 0.10, [13][14][15] 4.60 ± 0.16, 15-18 5.90 ± 0.20, 11,19,20 and 5.27 ± 0.12 eV, 11,15,18,21-26 respectively. By plotting these values as a function of the BE of the Ar 2p3/2 peak ( Figure S3), we obtained the following linear dependence: where WF is the work function of a particular sample, and BE is the BE of the Ar 2p3/2 peak. Therefore, the work function of the GaN(0001) wafer under various experimental conditions could be estimated by the introduction of Ar at a pressure of 0.5 mbar, while H2O was still present on the GaN surface. The Ar 2p spectra for samples treated under several experimental conditions were recorded, after which the BEs of Ar 2p3/2 were extracted. The changes in the work function were determined using Equation S1, and they are listed in Table S1. Table S1. Measured and estimated values of band bending (BB), work function (WF), electron affinity (EA), ionization energy (IE), and their relative changes under different experimental conditions with respect to pristine GaN. Note: The values were obtained from peak fittings of photoemission spectra and do not necessarily indicate the accuracy of the spectrometer. Electronic properties: Using the method described in the main text, we were able to determine the changes in band bending, work function, and electron affinity under different experimental conditions (summarized in Table S1). Figure S4 shows the photoemission spectra of O 1s and N 1s, involving band bending and SPV effects under a few chosen experimental conditions that were used to estimate the electronic properties. The evolution of band bending, work function, and electron affinity are plotted and shown in Figure S5.