{"id":"6cd61610-e696-4096-92ea-11a869fcf57c","arxiv_id":"1908.03376","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Cleaning steps set the work function and band bending of air-exposed GaN surfaces over a range larger than 1 eV, and UV light reduces the work function of HCl-treated GaN by altering the surface dipole.","lead":"This paper measures how three standard cleaning steps (acid etch, heating, oxygen plasma) change the electric fields and voltages at the surface of two types of gallium nitride crystals. It finds that ultraviolet light lowers the voltage of acid-cleaned samples by changing surface dipoles rather than screening the internal field, which may change how past experiments are interpreted.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The no-screening claim rests on an unquantified null result: 'no shift of the core levels' is reported without a detection limit, so a 0.2 eV WF drop on HCl surfaces could in part be band-bending screening.","rationale":"The reader's weakest assumption identifies the same load-bearing gap: the paper asserts 'no shift of the core levels' under UV illumination without stating the sensitivity of that measurement. The central claim in Section III.D is not that the WF changes—that is well measured by both XPS and KPFM—but that the change is entirely a surface-dipole effect and not a surface photovoltage from band-bending screening. The only direct evidence against screening is the unquantified core-level null result. The supporting Au Fermi-edge control demonstrates the absence of charging but does not calibrate the core-level response to a known band-bending change. This concern is most acute for the GaN(0001) HCl surface, where the reported dark BB is 0.0 eV and a slightly upward true BB would allow a 0.1–0.2 eV photovoltage to masquerade as a dipole change. For GaN(1-100) after HCl the reported negative BB makes the screening explanation less plausible because screening would increase the WF, but the paper's general conclusion about Kelvin-probe SPV applies to all surfaces and therefore depends on the null result for the O2-plasma surfaces as well. Because the missing quantification is a stated condition rather than a demonstrated error, the existing CONDITIONAL verdict stands, with the condition being that the authors report the detection limit and confidence intervals for the core-level shifts under illumination.","tokens_in":11019,"tokens_out":19913,"duration_ms":229379,"concrete_test":"On the same HCl-treated GaN(0001) and O2-plasma GaN(0001) samples used in Section III, acquire at least 10 Ga2p3/2 and N1s spectra alternating dark and 365 nm LED illumination at fixed X-ray flux. Fit each spectrum with the same Voigt/Shirley procedure, compute the dark-vs-LED centroid shift and its 95% confidence interval (e.g., from the fit covariance or a bootstrap). Also record the SEC cutoff in the same runs. If the upper confidence bound for the core-level shift is <0.05 eV while the SEC cutoff shifts by ≥0.2 eV, the no-screening/dipole attribution is supported; if the confidence interval reaches ~0.1 eV, the conclusion must be weakened to 'screening below ~0.1 eV cannot be excluded.'","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central inference in Section III.D is that the work-function decrease observed under 365 nm illumination on HCl-exposed GaN is a surface-dipole effect, because 'no shift of the core levels' indicates that band bending is not screened. The paper never quantifies this null result: there is no stated detection limit, no confidence interval, and no repeated-measurement statistics for the Ga2p3/2 or N1s centroids in the dark/illuminated comparison. The Section II control using the Au Fermi edge at 0.00±0.03 eV checks for charging, not for the minimum band-bending change that would be visible in the GaN core levels. This matters most for the GaN(0001) HCl surface, whose reported dark BB is 0.0 eV: if the true BB were only 0.1–0.2 eV upward (within the uncertainty of the 20±10 meV bulk Fermi-level reference used in Eq. 1), a surface photovoltage of that size would lower the WF by exactly the observed amount and would be masked by an unquantified 'no shift.' The (1-100) result is less vulnerable because the reported negative BB makes screening increase the WF, but the paper's broad conclusion that Kelvin-probe SPV alone cannot characterize GaN band bending rests on the null result for all surfaces, including O2-plasma surfaces with upward BB.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a combined XPS and Kelvin-probe study of the electronic properties of air-exposed GaN(1-100) and GaN(0001) surfaces subjected to three device-processing-compatible cleaning steps: HCl etching, annealing in N2, and O2 plasma exposure. The authors quantify adsorbate coverages, band bending, work function, and surface dipoles in the dark and under UV illumination. The central claim is that on HCl-treated surfaces, UV illumination reduces the work function by at least 0.2 eV while the XPS core levels do not shift, implying that the band bending is not screened and that the work-function change is a photo-induced modification of the surface dipole. On this basis the authors argue that surface photovoltage measurements by Kelvin probe alone cannot be used to characterize GaN surface band bending.","tokens_in":11360,"tokens_out":5805,"duration_ms":63442,"significance":"If the central claim holds, the paper provides practically useful cleaning protocols for resetting air-exposed GaN surfaces into defined electronic states and delivers a cautionary result for the interpretation of Kelvin-probe surface photovoltage measurements on GaN. The strength of the paper is that the relative trends across cleaning steps are directly measured with two complementary techniques, the work-function span is large and internally consistent, and the adsorbate analysis supports the qualitative picture of cleaning-dependent surface dipoles. However, the load-bearing conclusion that the observed photovoltage is entirely a surface-dipole effect rests on an unquantified null result: the reported absence of core-level shifts is given without a detection limit. The absolute band-bending values also depend on a self-calibrated reference, which adds systematic uncertainty to the absolute claims. These issues are addressable but require revision.","major_comments":[{"comment":"The central claim that the work-function decrease of HCl-treated surfaces under 365 nm illumination occurs 'without screening of the band bending' is supported only by the statement that 'no shift of the core levels' is observed. No detection limit, confidence interval, or repeated-measurement statistics are reported for the Ga 2p3/2 or N 1s centroid positions in the dark/illuminated comparison. The Au Fermi-edge control described in Section II checks for charging, not for the minimum band-bending change that would be visible in the GaN core levels. On the GaN(0001) HCl surface, whose reported dark band bending is 0.0 eV, a surface photovoltage of 0.1-0.2 eV would lower the work function by exactly the observed amount and would be masked by an unquantified 'no shift.' Please provide the centroid shift, its uncertainty, and an explicit statement of the minimum detectable surface photovoltage, or restrict the conclusion accordingly.","section":"Section III.D and Table I"},{"comment":"The absolute band-bending scale is set by the material constant [E_VBM - E_Ga2p3/2] = 1115.4 +/- 0.1 eV, which the authors obtain by averaging their own 25 XPS spectra of GaN samples with different orientations and adsorbates. If those spectra were not acquired at flat band, the zero of the absolute band-bending scale is systematically offset, and the stated uncertainty of 0.1 eV reflects only the uncertainty in the VBM linear fit, not this systematic effect. This matters for the absolute claims of 'flat band' after HCl etching on GaN(0001) and a downward band bending of -0.2 eV on GaN(1-100), as well as for the derived surface-dipole amplitudes. Please provide a calibration of the bulk reference against an independently known flat-band condition, or explicitly state the systematic uncertainty and temper the absolute values.","section":"Section III.B, Eq. (1)"},{"comment":"The XPS and KPFM illumination conditions are not equivalent: the XPS measurements use a 365 nm LED with an excitation density of 10-100 mW/cm2, whereas the KPFM measurements use a 405 nm LED with an excitation density of about 0.1 mW/cm2. The no-screening evidence comes exclusively from the XPS measurements at 365 nm, while the KPFM data at 405 nm are presented as complementary values of the same surface-dipole change. Since the magnitude of a surface photovoltage depends strongly on photon energy and excitation density, the two data sets are not directly comparable. Please either perform the KPFM measurements under the same above-band-gap illumination conditions, or explicitly restrict the claim to the 365 nm excitation used for the XPS core-level null result.","section":"Section III.D and Section II"}],"minor_comments":[{"comment":"The text states that the work function is modulated over a range of 0.6 eV in N2 atmosphere, but the KPFM values in Table I span from 4.25 to 5.10 eV, i.e., about 0.85 eV. Please clarify whether the 0.6 eV value refers to a single orientation or correct the statement.","section":"Section III.C and Table I"},{"comment":"The electron affinity of 4.06 eV used for the dipole estimates is taken from a scanning tunneling spectroscopy study of the GaN(1-100) surface and is applied to the GaN(0001) surface as well. Please justify this transfer or add a caveat that the absolute dipole values for GaN(0001) are approximate.","section":"Section III.C"},{"comment":"The statement that 'similar trends are observed for the cleaned GaN(0001) surface' would be better supported by showing the corresponding dark/illumination core-level spectra for the GaN(0001) surface, particularly for the O2-plasma-treated case where a large upward band bending is claimed.","section":"Section III.D and Fig. 3"},{"comment":"The comparison between XPS (vacuum, 365 nm) and KPFM (N2 atmosphere, 405 nm) results would benefit from an explicit discussion of how the different measurement environments and photon energies affect the observed work-function changes, especially the role of physisorbed water.","section":"Section II"}],"recommendation":"major_revision","confidential_remarks":"The detection-limit issue is likely addressable with existing data by reporting centroid shifts and repeat measurements. If the authors can quantify the minimum detectable core-level shift and either calibrate or appropriately caveat the absolute band-bending scale, the paper's central claim would be substantially strengthened. The fit to the journal's scope is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know: the paper is a solid experimental surface-science letter that gives the first quantitative band-bending and work-function values for GaN(1-100) after common cleanings, and it reports a UV-induced work-function decrease on HCl-treated surfaces that it attributes to a surface dipole change rather than band-bending screening. That claim holds up better than the stress-test suggests, because the KPFM control uses 405 nm (sub-band gap) light, where conventional surface photovoltage is not expected, and the (1-100) HCl surface has a negative band bending so screening would raise, not lower, the work function.\n\nWhat's genuinely new: the m-plane data, the systematic comparison of three cleanings, and the photo-dipole effect on HCl surfaces, which prior SPV studies missed. The paper also does well in using complementary XPS and KPFM, and the relative trends across cleaning steps are internally consistent. The practical point—that these cleanings can re-initialize air-exposed GaN surfaces into defined states—is useful.\n\nThe soft spots are real but not fatal. The authors report \"no shift of the core levels\" under 365 nm illumination without stating a detection limit, so the claim of zero band-bending screening is not fully quantified. A referee should ask for the peak-position uncertainty. The absolute band-bending scale rests on a self-calibrated constant [E_VBM - E_Ga2p3/2] = 1115.4 ± 0.1 eV averaged from their own spectra; if those spectra weren't all flat-band, the absolute values shift, though the relative trends and the dipole argument survive.\n\nOne nuance the stress-test misses: the KPFM data, taken with sub-band-gap 405 nm light, shows a similar (larger) work-function decrease on HCl surfaces. That's strong evidence for a light-induced dipole change because sub-band-gap photons don't generate the usual SPV. The paper could have foregrounded this more.\n\nWho's this for: anyone doing GaN surface prep, device processing, or Kelvin-probe SPV on nitride surfaces. It deserves a serious referee. I'd send it out, and ask for the core-level shift uncertainty.","headline":"New quantitative GaN(1-100) band-bending and work-function data after common cleanings, with a UV-induced dipole effect on HCl surfaces that is supported by sub-band-gap KPFM.","tokens_in":11925,"tokens_out":6380,"would_cite":true,"duration_ms":58548,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Under ultraviolet illumination, the work function of HCl-cleaned GaN surfaces drops by at least 0.2 eV while the Ga 2p3/2 core level does not move, showing the photovoltage is a light-induced surface-dipole change, not band-bending…","keywords":["GaN","surface band bending","work function","surface photovoltage","surface dipole","adsorbates","Kelvin probe force microscopy","X-ray photoelectron spectroscopy"],"falsifier":"Measure the Ga $2p_{3/2}$ core-level position under the same 365 nm illumination with a detection limit below roughly 50 meV (for example, by longer acquisition or a calibrated reference line). If any core-level shift appears on the HCl-treated surfaces while the work function drops by 0.2 eV, part of the photovoltage is band-bending screening, and the dipole-only attribution would be falsified.","tokens_in":10851,"feed_emoji":"💡","tokens_out":13578,"duration_ms":113774,"temperature":0.7,"pith_summary":"This paper asks whether three standard, device-compatible cleaning steps—HCl etching, annealing in nitrogen, and oxygen-plasma exposure—leave air-exposed GaN surfaces in predictable electronic states. Using X-ray photoelectron spectroscopy and Kelvin probe force microscopy in the dark and under ultraviolet light, the authors find that the cleaning steps change both the surface band bending (over several hundred millielectronvolts) and the work function (by more than 1 eV in vacuum), and that the changes track the adsorbates left by each step. The central finding is that ultraviolet illumination lowers the work function of HCl-treated surfaces by at least 0.2 eV while leaving the XPS core levels unshifted, meaning the photovoltage comes from a light-induced change in the surface dipole, not from screening of the band bending. If this is right, past Kelvin-probe surface-photovoltage studies of HCl-treated GaN that interpreted their signal as band-bending screening need to be reconsidered, and the three cleaning steps become a practical tool for resetting GaN surfaces to defined electronic states.","feed_headline":"UV light drops the work function of HCl-cleaned GaN by 0.2 eV","feed_subtitle":"The drop happens without shifting the core levels, so Kelvin-probe photovoltage alone cannot measure GaN band bending.","key_machinery":"The analysis rests on two complementary probes and a bookkeeping identity. Band bending is extracted from the Ga $2p_{3/2}$ core-level binding energy through $$\\mathrm{BB} = [E_{\\mathrm{VBM}} - E_{\\mathrm{Ga}\\,2p_{3/2}}] + E_g - BE_{\\mathrm{Ga}\\,2p_{3/2}} - [E_{\\mathrm{CBM}} - E_F]_{\\mathrm{bulk}},$$ using the measured constant $[E_{\\mathrm{VBM}} - E_{\\mathrm{Ga}\\,2p_{3/2}}] = 1115.4\\pm0.1$ eV; the work function is read from the secondary-electron cutoff, and the Kelvin probe gives the contact potential difference. Subtracting the band bending from the work function, with the electron affinity taken as 4.06 eV, isolates the surface dipole $\\delta$. The decisive move is the illumination comparison: when the core levels do not shift under UV light, any measured work-function change must equal a change in the surface dipole, $\\Delta\\delta_{\\mathrm{LED}} = WF_{\\mathrm{LED}} - WF_{\\mathrm{dark}}$. The three cleaning steps are themselves part of the machinery, since they serve as the controllable way to populate or strip specific adsorbate terminations ($\\cdot\\mathrm{ClO}_n\\mathrm{H}_m$, $\\cdot\\mathrm{CO}_n\\mathrm{H}_m$, $\\cdot\\mathrm{OH}_n$, $\\cdot\\mathrm{H}$, and $\\mathrm{GaO}_x + \\cdot\\mathrm{OH}_n$).","core_discovery":"On n-type GaN(1-100) and GaN(0001) surfaces exposed to air, the three cleaning steps reproducibly produce very different electronic landscapes. HCl etching leaves Cl- and H-containing adsorbates that act as surface donors, giving flat bands on (0001) and downward band bending with an electron accumulation layer on (1-100). Annealing at 400–500 °C in N2 removes the Cl and yields a moderate 0.1–0.2 eV upward band bending, while O2 plasma forms a thin GaOx capping layer with OH terminations and drives a 0.5–0.6 eV upward band bending together with a large inward surface dipole. Under 365 nm illumination, no Ga $2p_{3/2}$ core-level shift is observed on any surface, ruling out photo-induced screening of the band bending. Yet on the HCl-treated surfaces the work function drops by at least 0.2 eV, so the authors attribute the surface photovoltage to a photo-induced change of the surface dipole built by the Cl/OH adsorbates. The conclusion is that Kelvin-probe surface photovoltage alone cannot separate dipole changes from band-bending screening, so earlier interpretations of such measurements on HCl-treated GaN need revision.","pith_inferences":["If the 0.2 eV photo-induced dipole drop is specific to Cl/H termination, dosing a clean GaN surface with HCl in vacuum and cycling the UV light should reproduce the effect; if it appears only after air exposure, co-adsorbed water or oxygen is part of the mechanism.","The result implies contact-barrier engineering on GaN may be achievable purely through surface chemistry: because the cleaning steps change the work function by more than the band bending, Schottky barrier heights on air-exposed GaN should shift by roughly the dipole change, a prediction the paper does not test.","The same two-channel picture (band-bending screening plus dipole change) likely applies to other III-nitrides and to oxide-covered semiconductors; comparative Kelvin-probe/XPS measurements under sub-band-gap versus above-band-gap illumination would show whether the dipole channel is a general adsorbate phenomenon.","Because the paper relies on a single electron-affinity value (4.06 eV) for absolute dipole amplitudes, a direct measurement of the affinity on atomically clean surfaces of both orientations would sharpen all quantitative dipole estimates; the relative trends between cleaning steps would presumably survive."],"forward_implications":["Kelvin-probe surface photovoltage measurements on HCl-treated GaN cannot be read directly as band-bending screening; a UV-induced work-function drop of at least 0.2 eV can be a pure surface-dipole effect.","The three cleaning steps give a practical recipe to set air-exposed GaN(1-100) and GaN(0001) surfaces into defined electronic states, with band bending tunable over several hundred millielectronvolts and the work function by more than 1 eV in vacuum.","HCl etching produces an electron accumulation layer on GaN(1-100), so Cl/H adsorbates act as surface donors rather than merely removing surface states.","O2 plasma exposure, by creating a thin GaOx layer with OH terminations, drives a large upward band bending and inward dipole, a state that must be accounted for when later processing contacts on the surface.","Earlier reports that interpret UV-induced photovoltage on HCl-treated GaN only in terms of band-bending screening (references [3,7,59-61]) should be revisited with a core-level probe."],"supporting_citations":[{"why":"Supplies the Ga 2p3/2 method for extracting surface band bending and reviews how the cleaning steps affect GaN surfaces.","marker":"[18]"},{"why":"Establishes that HCl etching reduces C and O contaminants and suppresses reoxidation, a premise for the adsorbate assignment.","marker":"[21]"},{"why":"Documents that annealing removes residual Cl left by HCl etching, linking the N2-annealing step to the disappearance of the Cl-related surface dipole.","marker":"[22]"},{"why":"One of the earlier surface-photovoltage interpretations on HCl-treated GaN that the paper argues must be revised.","marker":"[3]"},{"why":"Another earlier Kelvin-probe surface-photovoltage study on HCl-treated GaN that attributes UV-induced work-function changes to band-bending screening, the interpretation the paper challenges.","marker":"[7]"},{"why":"Provides the 4.06 eV electron affinity for GaN(1-100) used to convert work functions into surface-dipole amplitudes.","marker":"[30]"},{"why":"Reports a Ga 2p3/2 core-level shift under illumination, the contrast case that highlights the absence of band-bending screening here.","marker":"[55]"},{"why":"Supplies the surface-photovoltage framework linking excitation density and surface-state density to surface-photovoltage amplitude, used to explain why no screening is seen.","marker":"[57]"},{"why":"Provides the charge-transfer and redox-reaction picture that grounds the assignment of adsorbate-induced surface dipoles.","marker":"[12]"},{"why":"Supplies the InN electron-accumulation analogue used to interpret the downward band bending after HCl etching.","marker":"[31]"}],"fun_headline_variants":["UV alters surface dipole, not band bending, on HCl-cleaned GaN","HCl-cleaned GaN shows UV-induced dipole change, no band bending shift","GaN SPV mystery: UV changes dipole, not band bending on HCl surfaces","Photo-induced dipole drop on HCl-GaN misleads Kelvin probe band bending","UV exposure shifts work function on HCl-GaN without core-level change"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that the 0.2 eV UV-induced work-function drop is entirely a surface-dipole effect rests on the assumption that the XPS core-level measurement would have detected any band-bending screening of a few tens of millielectronvolts, since the paper reports no core-level shift without stating the detection limit.","fun_headline_variants_meta":{"raw":{"variants":["UV alters surface dipole, not band bending, on HCl-cleaned GaN","HCl-cleaned GaN shows UV-induced dipole change, no band bending shift","GaN SPV mystery: UV changes dipole, not band bending on HCl surfaces","Photo-induced dipole drop on HCl-GaN misleads Kelvin probe band bending","UV exposure shifts work function on HCl-GaN without core-level change"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000263,"raw_usage":{"total_tokens":1656,"prompt_tokens":1057,"completion_tokens":599,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":673,"completion_tokens_details":{"reasoning_tokens":497}},"tokens_in":673,"tokens_out":599,"duration_ms":6572,"temperature":1.0,"reasoning_tokens":497,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:14:21.242974+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the Ga $2p_{3/2}$ core-level position under the same 365 nm illumination with a detection limit below roughly 50 meV (for example, by longer acquisition or a calibrated reference line). If any core-level shift appears on the HCl-treated surfaces while the work function drops by 0.2 eV, part of the photovoltage is band-bending screening, and the dipole-only attribution would be falsified.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Ga 2p3/2 method for extracting surface band bending and reviews how the cleaning steps affect GaN surfaces."},{"cited_title":"Bermudez, Surf","cited_arxiv_id":null,"evidence_quote":"Establishes that HCl etching reduces C and O contaminants and suppresses reoxidation, a premise for the adsorbate assignment."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents that annealing removes residual Cl left by HCl etching, linking the N2-annealing step to the disappearance of the Cl-related surface dipole."},{"cited_title":"As largely reported for the GaN(0001) surface, HCl etch- ingdecreases to low levels the areal concentration of typical surface contaminants (C and O) [Ref","cited_arxiv_id":null,"evidence_quote":"One of the earlier surface-photovoltage interpretations on HCl-treated GaN that the paper argues must be revised."},{"cited_title":"Tripathy, S","cited_arxiv_id":null,"evidence_quote":"Another earlier Kelvin-probe surface-photovoltage study on HCl-treated GaN that attributes UV-induced work-function changes to band-bending screening, the interpretation the paper challenges."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the 4.06 eV electron affinity for GaN(1-100) used to convert work functions into surface-dipole amplitudes."},{"cited_title":"Kahn, Mater","cited_arxiv_id":null,"evidence_quote":"Reports a Ga 2p3/2 core-level shift under illumination, the contrast case that highlights the absence of band-bending screening here."},{"cited_title":"Grandjean, J","cited_arxiv_id":null,"evidence_quote":"Supplies the surface-photovoltage framework linking excitation density and surface-state density to surface-photovoltage amplitude, used to explain why no screening is seen."},{"cited_title":"González-Posada, R","cited_arxiv_id":null,"evidence_quote":"Provides the charge-transfer and redox-reaction picture that grounds the assignment of adsorbate-induced surface dipoles."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the InN electron-accumulation analogue used to interpret the downward band bending after HCl etching."}],"review_version":1}