Photopotential Measurements on Undoped n-InP at Open Circuit Potential According to the Aqueous pH

In this work we studied some fundamental concepts of n-InP interfacial properties based on acid-base equilibrium in the dark and also after the surface illumination at open circuit voltage (Voc). The variation of the flatband potential on undoped n-InP semiconductor was investigated according to aqueous pH. This study was complemented by the effect of the illumination at Voc conditions through photopotential measurements. The impact of the surface lighting on acid-base equilibrium was analyzed by in-situ interfacial capacitance measurements in the dark and by ex-situ X-ray photoelectron spectrometry. © The Author(s) 2018. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0201804jes]

The fundamental properties of many semiconductor/electrolyte systems were monitored by the flatband potential position (E FB ). 1,2ts position governed the interfacial characteristics and controlled the electrochemical properties.A classical way of measuring the variation of the flatband potential position was to change the pH of the medium.][5][6][7][8][9][10][11][12] For III-V semiconductors such as binary (GaAs 3,5 or GaP 6,7 ) or ternary 8 compounds, the acid-base equilibrium was complicated by the specific adsorption sites which induced localized charges on the surface. 13Previous works have shown a quasi Nernstian variation of the flatband potential on n-InP for (100) 14 and (111) faces 13 over the whole pH range in aqueous medium and also in non aqueous medium such as acetonitrile 15 and liquid ammonia. 16cid-base equilibrium was studied in the dark from interfacial capacitance measurements in the depletion region avoiding faradaic transfer at the interface semiconductor/electrolyte. This included changing the voltage of the semiconductor artificially through the use of a potentiostat.The band bending owing to electron depletion in the semiconductor (SC) changed depending on the voltage.In the depletion region the Boltzmann relation described the distribution of electrons in the space charge region and the electric field was determined by Gauss'law.Poisson's equation could be solved within that region to give the Mott-Schottky equation that links the reverse square of the capacitance (C) to the interfacial polarization (E). 17Under depletion conditions the flatband potential was actually deduced from C −2 = f(E) straight line which extrapolation gave E FB for a zero value of There are other ways to evaluate the flatband potential such as measuring the photopotential at the open circuit voltage (Voc).As a function of radiation intensity the photopotential was the change in the Fermi level due to the promotion of electrons to the conduction band, and it reached theoretically a maximum at the flatband potential. 13,1815][16]18 This was particularly true in terms of surface electrical field, charge, capacitance, layer of the space charge region. 1,2,19As a consequence, undoped n-InP exhibited a wider range of interfacial polarization avoiding faradaic transfer at the interface SC/electrolyte.These improved conditions were required for interfacial capacitance measurements compared to high doped SC.The impact of the illumination at Voc conditions was also studied using photopotential measurements on undoped n-InP whose larger space charge layer was well suited for this study.This z E-mail: anne-marie.goncalves@uvsq.froriginal procedure provided an opportunity to "revisit" some of fundamental concepts of interfacial properties by considering acid-base equilibrium in the dark and after the illumination on undoped n-InP at Voc conditions.

Experimental
InP semiconductor (n-types, undoped) wafers with a <100> orientation were purchased from MCP Electronic Materials in which no impurities have been voluntarily added.However, the intrinsic carrier concentration given by the supplier was estimated to be about 5 × 10 15 atoms.cm−3 .The wafers were cut into small squares with an area around 0.12 cm 2 .Oxide free InP surfaces were obtained using an initial chemical-mechanical polishing with methanol bromide solution (2%) followed by a thoroughly methanol rinsing.Just before experiment the surface is dipped in 2 M HCl solution then washed in pure water and dried under argon. 14n-situ capacitance experiments were performed in aqueous media using a classical three-electrodes configuration.A platinum electrode was used as counter electrode.All potentials were measured vs. a mercury sulfate electrode (MSE).The ohmic contact was made on the back side of n-InP by a thin conductive layer of gold which was deposited by electron beam evaporator technique.In-situ capacitance measurements were performed in the dark under a constant argon flux and using 2273 parstat at a frequency of 1.03 kHz.
Photopotential measurements required the semiconductor illumination which was provided by a led light engine using Schott KL 2500 Led.The brightness was equivalent to a 250-Watt halogen cold light source at an irradiance about 36 mW/cm 2 .The electrochemical cell was composed of an optical window that ensured the illumination access.The light source was connected to one end of the optical fiber.This work required an appropriate experimental design to provide the better reproducibility of photopotential measurements.A constant distance was kept between the optical window and the optical fiber.In the electrochemical cell, a constant distance was also kept between the optical window and the semiconductor surface.To modify the acid-base equilibrium at the junction SC/electrolyte, we used several buffered solutions (pH = 0, 2, 4, 8, 9, 10, 12) from Alfa Aesar.Buffered solution from pH 8 to 10 were borate-based chemical solutions while buffered solutions of pH 2 and pH 4 were glycine-based chemical solutions.For each photopotential measurement at specific pH, the electrochemical cell was previously rinsed several times with deionized water.Between each buffered pH, the solution was emptied out without removing the semiconductor.Photopotential experiments were performed from acidic to alkaline pH under constant argon flux.A light chopper was used during photopotential measurements to control the stability of the Voc position at the interface semiconductor/electrolyte in the dark and under illumination.
The chemical composition of InP surface was analyzed by Xray photoelectron spectroscopy (XPS) using a Thermoelectron VG-ESCALB 220i XL system.High-resolution XPS conditions have been used, in a constant analyzer energy mode with a pass energy of 8 or 20 eV.The monochromatic Al K R line or Al K R line of a dual anode was used as X-ray excitation.Large X-ray spots (3 mm) were usually used to explore the surface.The homogeneity of the surface modification was also controlled on lines by using small X-ray spots (100 μm).

Results and Discussion
The experimental process based on the chemical-mechanical polishing of n-InP samples avoided the formation of InP oxides which had been clearly demonstrated by X-ray photoelectron spectrometry. 20nP electrodes were successively immersed in several buffered aqueous solution.A local variation of the pH induces an evolution of the potential drop through the Helmholtz layer which was considered as capacitive component (C H ). 21 The excess of charge was compensated inside the interface which could involve a modification of the semiconductor (SC) interface.The equivalent capacitance of the interface could be described according to the relation (1) where Csc was the only bias-dependent capacitance. 22C = 1/C H + 1/Csc [1]   A contact between n-InP and water acted as an electrochemical junction.A dynamic equilibrium was reached after equalization of both Fermi levels from the junction.At Voc conditions, the equilibrium junction of InP/aqueous electrolyte made the semiconductor in depletion region.1/C H became negligible, and the capacitance was directly related to the space charge layer of InP: Mott-Shottky plots provided a linear relation between the reverse square of the capacitance and the interfacial polarization (E).
Where "q" was the charge, "N D " was the doping level, " ε • " was the electric permittivity of empty space and "ε" was the electric per- mittivity of InP, "k b " was the Boltzmann constant and "T" was the temperature.We noted that the ratio k b T/q became negligible.
For several buffered solutions Mott-shottky straight lines were reported on the Figure 1.According to the pH, the flatband was deduced from the intercept of the regression line.Whatever the pH, the slopes were kept constant.The intrinsic carrier was calculated from the slope.An average value of 6 × 10 15 atoms/cm 3 was deduced.This result remained close to the supplier value.In aqueous media, the high chemical stability of a bare InP surface provided reproducible interfacial capacitance. 14,23As was shown in the inset graph of Figure 1, the average change of the flatband potential was reported against the pH.A quasi Nernstian variation of the flatband potential on undoped n-InP was also observed in the whole range of pH (Fig. 1, inset).The relation 4 described the Nernstian variation of the flatband potential with the pH.
Where "E • FB " was the flatband potential at a pH of zero.Solid states studies fully agree with the prediction of Lewis acidbase chemistry at the interface semiconductor/water.This result was consistent with previous studies on higher doped n-InP where precautions have been taken to avoid the formation on the surface of oxide/hydroxide mixture at neutral pH. 14,23he acid-base equilibrium was explored after illumination at Voc conditions.These experiments gave access to the photopotential which corresponds to the potential variation at Voc in the dark and under illumination. 18As we expected, the photopotential increased until a limit value as soon as the light intensity increased.The limit value was due to the position saturation of the open circuit voltage under illumination.It gradually moved closer to the flatband potential. 13For the same light intensity, the photopotential measurements were performed at several pH.Two photopotential variations were reported on the Figure 2 at acidic and alkaline pH.The photopotential positions were shifted according the pH as it was predicted by the flatband potential variations with the pH (Fig. 1).A photopotential enlargement was observed as soon as the pH increased (Fig. 2).In the dark a higher band bending was indeed observed in alkaline pH as compared to the acidic pH.In the dark, the more the Voc position was near the valence band, the more the photopotential could increase.That was currently being observed at pH 12 (Fig. 2).For both pH, a high stability of the Voc in the dark was shown.However after illumination, a contrasted behavior was detected according to the pH.While the photopotential remained stable at acidic pH, a gradual decrease of Voc under illumination was revealed over time.As a consequence at acidic pH, the photopotential measurement was constant over time while a significant decline of the photopotential was shown at alkaline pH (Fig. 3).The instability of the photopotential could result from modifications of n-InP interface which seemed to occur only under illumination because a high stability of the Voc position was observed in the dark whatever the pH (Fig. 2).The question that was raised was how a SC interface could be modified without applying an external voltage or current at the n-InP/alkaline interface.Even if the photopotential occurred at Voc conditions, the equilibrium of both Fermi Levels from the junction appeared to be disturbed by the SC illumination.Interfacial capacitance measurement in the dark was a very sensitive and reliable in-situ method to study the variation of charge distribution at the SC/electrolyte interface.Mott-Schottky plot was indeed a good indicator of in-situ chemical modification progress of the interface n-InP/electrolyte.In the Figure 4, Mott-Schottky plots were recorded in the dark at a pH of 0 (Fig. 4A) and a pH of 9 (Fig. 4B).For both pH the effect of n-InP illumination at Voc conditions on Mott-Schottky plots were analyzed (Fig. 4A(2) and Fig. 4B(2)).At acidic pH a perfect overlapping of the C −2 -V straight lines was observed (Fig. 4A).This behavior proved that at pH 0, the chemical composition of the n-InP surface is unchanged after an illumination of the SC surface at Voc condition (Fig. 4A(2)).In contrast, at alkaline pH, the light effects on the interfacial capacitance measurements have been clearly evidenced.After illumination, a positive shift of the straight line (recorded in the dark) with a slight decrease of the Mott-Schottky slope was indeed revealed.Such positive shifts of C −2 = f(E) plots have already been observed at pH 9, on doped (10 18 atoms.cm−3 ) n-InP semiconductor after an external positive polarization under illumination. 24,25,20In these previous works, the illumination under positive polarization generated an anodic photocurrent transient which was associated to the anodic oxidation of InP surface and the growth of a thin "InPO 4 -like" oxide layer. 24,25The positive shift of the Mott-Schottky plot was clearly attributed to a chemical modification of the SC surface. 24,25In the present work, the positive shift of Mott-Schottky plots could suggest a similar chemical modification of the undoped n-InP surface at Voc conditions under illumination.
To check this assumption, under illumination at the undoped n-InP interface, a constant positive polarization was applied at V = 0.8 V/Voc.A photo-anodic charge of 2.5 mC.cm −2 was measured from this external polarization.In these conditions, an anodic photocurrent transient very similar to that recorded on doped n-InP 25,26 was observed suggesting the growth of a very similar oxide at alkaline pH.After the positive polarization under illumination, the Mott-Schottky plot was also recorded (Figure 4B(3)).The same interfacial capacitance features was found as those observed after the illumination at Voc conditions (Fig. 4B(2)), namely a positive shift of the flatband potential with a slight decrease of the initial slope.From these results, it could be suggested, that the deviations of Mott-Schottky plots recorded after an illumination at Voc conditions of n-InP (Fig. 4B(2)) were consistent with the growth of an oxide at the SC surface at alkaline pH.
X-ray photoelectron spectroscopy (XPS) was a surface-sensitive quantitative spectroscopic technique that measures the chemical environment of the surface.XPS analyses were performed to understand the nature of the chemical modification after the photopotential experiments.Reproducible indium and phosphorus photoelectron peaks were recorded for different surface treatments at pH 9 (Fig. 5).At Voc conditions, after soaking the sample 3600 s in the dark at pH 9, no significant changes have been observed for P 2p and In 3d peaks as compared with a sample freshly deoxidized.In spite of the long soaking time in the dark at pH 9, these contributions overlapped each other with the same shape (full width at half maximum value) and Counts/s Binding energy (eV) In 3d the same energy position (Fig. 5(1)).However, as soon as the illumination at Voc conditions of the n-InP surface has been performed, both peaks were modified (Fig. 5(2)).A new contribution positively shifted in binding energy was revealed for P 2p peak and a positive shift with enlargement of In3d peak were also observed (Fig. 5(2)).These modifications increased drastically when the sample was still illuminated but kept under positive polarization (Fig. 5(3)).7][28][29][30][31] Furthermore, referring to the literature and to the experimental data showing a positive shift of the flatband potential by capacitance measurements and new contributions of P 2p and In 3d peaks positively shifted by XPS analyses, the interface was classically characterized by the formation of a thin oxide with a composition close to InPO 4 .For undoped n-InP, under positive external polarization and under illumination, the formation of InPO 4 -like compound could be also suggested from capacitance measurements and XPS analyses (Fig. 5(3)).For undoped n-InP, after the illumination at Voc conditions, the same experimental trends were shown as those observed after positive polarization under illumination.At Voc conditions under illumination, the formation of a thin oxide of InPO 4 -like might be also formed on the undoped n-InP interface.At Voc conditions, it could be supposed that photogenerated holes were able to migrate to the SC surface with a sufficient band bending to provide electronic transfers leading to a stable thin oxide of InPO 4 -like at alkaline pH.

Conclusions
On undoped n-InP surface freshly deoxidized, we have shown a quasi Nernstian variation of the flatband potential with the pH.Photopotential measurements have been used to "revisit" some of fundamental aspects of acid-base equilibrium at the interface n-InP/electrolyte after illumination of the semiconductor.A decrease of the photopotential measurements was only revealed at alkaline pH.This instability resulted from a decrease of the Voc position under illumination.In-situ interfacial capacitance measurements and ex-situ XPS analyses were performed to understand the effect of the illumination on undoped n-InP at Voc conditions.From capacitance measurements a positive shift of Mott-Schottkky plots with a slight decrease of the initial slope was evidenced.This modification of the interface was supported by XPS analyses showing a partial corrosion of n-InP at open circuit voltage under illumination.Indeed, the exchange current was sufficient to provide the SC photo-corrosion.By combining the benefits of both techniques, the formation of a thin oxide with a composition close to InPO 4 at the interface was suggested at alkaline pH after illumination on undoped n-InP at Voc conditions.The presence of a thin oxide at Voc conditions (under illumination) could explain the difficulties to obtain a perfect Nernstian variation of the flatband potential at alkaline pH.The question remains open for the other III-Vs and small gap semiconductors.

Figure 2 .
Figure 2. Variations of the open circuit voltage (Voc) under chopped light according to the experiment duration onto undoped n-InP at buffered aqueous pH (2 and 12).

Figure 4 .
Figure 4. In-situ capacitance at a frequency of 1.03 kHz in the dark on freshly deoxidized undoped n-InP surface.Mott-shottky straight lines obtained (A) at pH = 0 and (B) at pH = 9. (1): without semiconductor illumination (2): after photopotential measurements.(3): under illumination after the formation of a thin oxide using a constant polarization at V = 0.8 V/Voc and involving an anodic charge of 2.5 mC.cm −2 .

Figure 5 .
Figure 5.Comparison at pH = 9 of XPS spectra for the In 3d (left) and P 2p (right) contributions observed for different undoped n-InP surface treatments.(1): After 3600 s in the dark at Voc conditions or after a surface freshly deoxidized.(b): After 1000 s under illumination at Voc conditions (photopotential).(3): Under illumination after the formation of a thin oxide using a constant polarization at V = 0.8 V/Voc and involving an anodic charge of 2.5 mC.cm −2 .