Impact of various cleaning procedures on p‐GaN surfaces

This work discusses the influence of different cleaning procedures on p‐gallium nitride (GaN) grown on sapphire by metal–organic chemical vapor deposition. The cleaned p‐GaN surface was transferred into an ultrahigh vacuum chamber and studied by an X‐ray photoelectron spectrometer, revealing that a cleaning with a so‐called “piranha” procedure results in low carbon and oxygen concentrations on the p‐GaN surface. Contrary, a cleaning that solely uses ethanol represents a simple cleaning but leads to an increase of carbon and oxygen contaminations on the surface. Afterward, the cleaned p‐GaN samples underwent a subsequential vacuum thermal cleaning at various temperatures to achieve an atomically clean surface. X‐ray photoelectron spectroscopy (XPS) measurements revealed residual oxygen and carbon on the p‐GaN surface. Thus, a thermal treatment under a vacuum did not entirely remove these organic contaminations, although the thermal cleaning reduced their peak intensities. The complete removal of carbon and oxygen contaminants was only achieved by argon ion sputtering, which is accompanied by a strong depletion of the nitrogen on the p‐GaN surface. The treatments cause a large number of surface defects preventing the formation of a negative electron surface when the p‐GaN is activated with a thin layer of cesium.


| INTRODUCTION
Gallium nitride (GaN) has garnered increasing interest, especially since the Nobel Prize was awarded for developing GaN lightemitting diodes (LEDs) in 2014. 1,2 The band gap is manipulated by adding dopant atoms, such as indium (In), magnesium (Mg), and aluminum (Al) into the crystal lattice, 3 offering new opportunities for the use of III-V semiconductors in photonic devices, data-storage applications, UV detectors, and high-power electronics. A new application of p-GaN arose in the last years emerged from the ability to provide high-current electron beams for accelerator-based light sources, such as free-electron laser (FEL) and terahertz (THz) radiation. 4,5 P-type GaN with Mg doping can form a negative electron affinity (NEA) surface when activated with an alkali metal, such as cesium (Cs). When used as a photocathode, the cesiated p-GaN can reach high quantum efficiency (QE) values of up to 70%. [6][7][8] Achieving an atomically clean surface is an important prerequisite for the formation of the NEA surface. Strong acids, such as sulfuric acid (H 2 SO 4 ), hydrogen peroxide (H 2 O 2 ), hydrofluoric acid (HF), or hydrochloric acid (HCl), are usually used for cleaning III-V semiconductor surfaces. [9][10][11] A subsequent thermal cleaning under vacuum represents the most common way to clean semiconductor surfaces in order to receive an almost atomically clean surface, but at temperatures above 600 C, semiconductor surfaces might suffer from a depletion of their original composition. 12 Another alternative method represents an ion bombardment by inert gases such as argon (Ar), hydrogen (H 2 ), or nitrogen (N 2 ).
However, for many materials, such ion sputtering causes damages and thus undesired electronic property changes. 13 Consequently, NEA formation in the band structure can no longer occur.
In this work, X-ray photoelectron spectroscopy (XPS) was used to monitor the composition of p-GaN surfaces. Different surface conditions of p-GaN cleaned by 99% ethanol (EtOH) and "piranha" solution, followed by a vacuum thermal cleaning and bombarded with Ar + , were compared. We demonstrated that the surface carbon (C) and oxygen (O) peak intensities vary to the wet chemical cleaning and remained on the surface after a subsequent thermal vacuum treatment. Furthermore, an excessive thermal treatment and Ar + ion irradiation causes surface damages in the form of a deterioration of the initial surface; thus, an NEA formation is not achievable.

| MATERIALS AND METHODS
Commercially available 5 μm thick p-GaN layers grown on sapphire (Al 2 O 3 ) with metal organic chemical vapor deposition (MOCVD) were used. The samples were Mg doped and had a carrier concentration in the range of 6 Â 10 16 -1 Â 10 17 cm À3 , provided by the supplier. 14 One p-GaN sample was cleaned in 99% pure EtOH in an ultrasonic bath for 15 min, while another sample was treated with a socalled "piranha" solution (H 2 SO 4 : H 2 O 2 ) at 140 C for 15 min. In the last step, the sample was dipped into hydrochloric acid (HCl) (36%) and rinsed with deionized water (H 2 O). The "piranha" solution is a strong oxidizing agent and can remove most organic contaminants, whereas HCl removes metal ions from the surface.
The wet chemically cleaned samples were mounted on a molybdenum flag in inert gas atmosphere and were transported under a dry N 2 atmosphere into an ultrahigh vacuum (UHV) chamber. In the UHV chamber, a thermal cleaning was carried out using a thermoelement.
The temperature on the p-GaN surface was measured with an IR sensor, calibrated by the supplier.
The UHV preparation chamber was directly connected to an XPS, and the sample surface was studied after each treatment step at room temperature. The sample was transported under UHV conditions from the preparation chamber into the XPS chamber via a manipulator. The  The fitting of the experimental data was obtained using Casa XPS and a Lorentzian asymmetric (LA) line shape (1.53, 243), and a Shirley background was applied to all spectra.
After the wet chemical cleaning, several thermal cleanings, Ar + irradiation, and Cs activation, the surface morphology was studied ex situ by a scanning electron microscope (SEM), combined with energy dispersive X-ray spectroscopy (EDX). The SEM images were carried out on a Zeiss NVision 40 FIB/SEM microscope with an electron beam energy of 20 kV and a secondary electron (SE) detector. The EDX measurements were done with a Bruker ® QUANTAX EDS spectrometer.

| Wet chemical cleaning
The XPS spectra of O 1s, C 1s, and Ga 2p 3/2 are shown in Figure 1  The main peak was located at a binding energy (BE) of approximately 532 eV and accompanied by a small shoulder at 530 eV, which was derived from a less intense photoemission peak of gallium oxide (Ga x O y ). The third subpeak was located at a BE of 533.5 eV, which was caused by adsorbed H 2 O or a carbonyl (C=O) compound. A C peak at 284.9 eV was found in the C 1s spectra, which was derived from adventitious C. At a higher BE, a small shoulder at 286.4 eV was observed, which corresponded to a hydroxyl (C-OH) compound.
The Ga 2p 3/2 peak showed a fitting into two components. The cleaning. Additionally, the O concentration was also reduced because the O 1s peak intensity decreased after the cleaning with the "piranha" solution. Therefore, the "piranha" cleaning was able to reduce the C and O contamination layer on the p-GaN surface, and consequently, the Ga 2p 3/2 intensity increased.
Furthermore, the atomic concentrations taken from the survey spectra (shown in Figure 2) support the statement that the "piranha" cleaning reduced especially the C contamination on the p-GaN surface, while using EtOH led to a significantly increase of O and C.
After the "piranha" treatment, a shift toward lower BEs in all XPS photoemission peaks was observed, which was an effect of the reduction of C and O compounds on the p-GaN surface.

| Thermal cleaning in vacuum
The p-GaN layer was furthermore studied by a vacuum thermal cleaning at different temperatures through the sapphire substrate by using a thermoelement from the backside. The changes in the O and C photoemission peak intensities, which depended on the applied temperature, are shown in lines 1 to 6 in Figure 3. Several temperatures between 320 C and 650 C were applied to the p-GaN sample to remove surface O and C. The XPS measurements were carried out when the sample was cooled down to room temperature, but after the thermal cleaning, the presence of O and C was always observed on the p-GaN surface, and only a partial reduction of the C and O concentration was achieved. Even at a temperature of 650 C, it was not possible to remove the C and O completely as a residual amount of O and C remained on the surface, resulting in a small, broad peak that was still detectable. It is well-known that C, O, and other impurities were incorporated into the crystal lattice as unwanted impurities and that C remained on the surface when a semiconductor is manufactured by MOCVD. [15][16][17] The crystal quality and carrier mobility can suffer due to these unwanted impurities that also have an undesirable effect on the photocathode performance.
The atomic concentrations taken from example survey spectra (shown in Figure 4) showed a Ga:N ratio of 1.05 for the p-GaN surface cleaned at 320 C. This GaN ratio began to change and turned into 1.39 when the surface was thermal cleaned at 570 C. Furthermore, the Ga:N ratio changed to 1.61 when cleaned with 650 C, and thus, the surface was enriched with Ga atoms, thereby indicating that a depletion in the N atoms occurred.

| Ar + sputtering
C and O remained on the p-GaN surface and were not removed completely by a thermal cleaning. The p-GaN surface was therefore irradiated by Ar + with 1.5 kV energy for 10 min. Figure 5 shows a comparison of detailed XPS spectra of O 1s, C 1s, and Ga 2p 3/2 of the thermally cleaned p-GaN surface at 650 C and the Ar + irradiated surface.  Table 1 summarizes the quantitative values that were obtained from the survey spectra, which are shown in Figure 6. A Ga:N ratio of F I G U R E 2 Comparative survey spectra of original p-GaN surface (red), cleaned with EtOH only (green), and cleaned with a "piranha" procedure (blue).

F I G U R E 3 O 1s and C 1s
photoelectron spectra for the p-GaN surface cleaned at several temperatures (lines 1-6) under vacuum.

F I G U R E 4
Comparative survey spectra of p-GaN cleaned at 320 C (red), at 570 C (green), and at 650 C (blue).
F I G U R E 5 O 1s, C 1s, and Ga 2p 3/2 photoemission spectra for the p-GaN surface after thermal cleaning at 650 C (red), and after Ar + irradiation at an energy of 1.5 kV for 10 min (green).
T A B L E 1 Atomic concentrations derived from the survey spectra for a p-GaN surface after a cleaning at 650 C and after Ar + sputtering.

Treatment
Ga 2p   Afterwards, a thin layer of Cs was deposited on the cleaned p-GaN surface, but no NEA was achieved, and thus, no p-GaN photocathode was received.
Other researchers should note the phenomenon of the N depletion when working on p-GaN photocathodes due to thermal heating and ion sputtering as it seems one of the critical factors for determining the QE.
High quantum efficiencies were reported for p-GaN samples treated at moderate temperatures without using additionally Ar + ion sputtering in a previous publication. 18

| Surface morphology
SEM measurements were carried out to study the p-GaN surface morphology after the p-GaN underwent a wet chemical cleaning, several thermal cleanings, Ar + sputtering, and a thin layer deposition by Cs. Additionally, the surface composition was characterized by EDX spectroscopy whereby the p-GaN sample was taken out of the UHV F I G U R E 6 Comparative survey spectra for the original p-GaN surface (red), cleaned at 650 C (green), and after Ar + sputtering (blue).
F I G U R E 7 (A) Scanning electron microscope (SEM) image of a p-GaN surface that underwent thermal cleaning at 650 C, Ar + sputtering, and a Cs activation. The energy dispersive X-ray spectroscopy (EDX) measurements showed a depletion in N atoms at positions 1 and 2 at the surface and (B) resulting surface crates and destroyed p-GaN surface from Ar + sputtering.
chamber. Hence, the resulting SEM and EDX spectra might not show the original surface conditions because the sample was exposed to air.
The extensively treated p-GaN surface showed irregularities in the surface morphology and large black holes were found. The excessive thermal cleaning has caused the appearance of these holes, where the original p-GaN layer was destroyed. EDX measurements confirmed that the composition of N and Ga atoms changed in this area. The N atomic concentration inside the black holes was half of that of the Ga atoms when EDX was measured at position 2 in Figure 7A. Furthermore, the p-GaN surface showed an unequal GaN ratio of 1:0.84 at position 1 in Figure 7A in which depletion of N atoms must also have occurred.
Additionally, round crates were found at the surface which are shown in Figure 7B. The appearance of these round crates was associated with a damaging effect due to Ar + irradiation. The crates were homogenous and had almost the same diameter as sown in Figure 7B.
Therefore, it is assumed that the Ar + irradiation caused an unwanted change in the surface morphology.
These results showed that a depletion in N atoms occurred for an excessively treated p-GaN surface. Furthermore, the examined p-GaN surface showed a preferential sputtering of N atoms when irradiated with Ar + . After the wet chemical cleaning, the p-GaN sample underwent a subsequential vacuum thermal cleaning at temperatures between 320 C and 650 C. In situ XPS measurements showed that the vacuum thermal treatment was not able to remove C and O entirely, but the thermal cleaning reduced the peak intensities and thus their relative concentrations on the surface. The remaining O and C contaminations were assumed to be residuals from the MOCVD process. The complete removal of these contaminants can only be achieved by Ar + irradiation on the p-GaN surface, which was accompanied by a strong depletion of the N atoms. An excessive thermal cleaning above 600 C and an Ar + irradiation consequently caused a large number of defects on the semiconductor surface, which could be validated by SEM and EDX. After these excessive treatments, Cs was deposited onto the p-GaN surface, but no NEA surface was achieved. Thus, no p-GaN photocathode was received.

| CONCLUSION
Other researchers should note the phenomenon of preferential depletion when working with p-GaN, as it may be one of the critical factors for the obstruction of an NEA surface and thus a p-GaN photocathode.
Commercially available p-GaN with higher purity and ideally no impurities grown by other methods represent another alternative compared to p-GaN grown by MOCVD. However, the remaining surface impurities, especially C and O, could potentially be removable by an ion bombardment of helium (He), H + , or argon clusters at low energies. To date, no sufficient ion bombardment was reported in which a p-GaN surface was irradiated with He + or H + ions and examined on its surface with XPS. A suitable sputtering method might yield a C-free p-GaN surface without damaging it which consequently would lead to a better photocathode performance.