{"id":"6c01634b-f40b-4c4a-afbb-354fb068ce3f","arxiv_id":"2506.21068","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Rh gem-dicarbonyls on TiO2(110) adopt a square-planar geometry aligned along the [001] direction, as confirmed by STM/nc-AFM, IRAS and DFT, while XPS reveals coexisting IR-invisible Rh clusters.","lead":"Rhodium gem-dicarbonyls on a titanium dioxide surface were created and imaged atom by atom for the first time, showing the two carbon monoxide molecules aligned along the surface rows. The study shows that infrared spectroscopy alone misses other rhodium species present at the same time on the catalyst.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The double-lobed SPM features are not spectroscopically tagged as the IRAS gem-dicarbonyl; IRAS at the exact SPM preparation would close this gap, making the concern conditional rather than fatal.","rationale":"The paper's core structure is coherent: IRAS provides a clean fingerprint, DFT reproduces the mode splitting to within a few cm−1 and excludes the monocarbonyl, and XPS links the 310.1 eV Rh component and 288.7 eV CO/Rh component to the same thermal evolution as the IRAS bands. The weakest point is the species identity of the SPM features, exactly as the reader stated. The preparation mismatch (0.005 ML/270 K vs 0.05 ML/200–250 K) plus the unassigned 2113 cm−1 species makes it possible, though not likely, that the imaged double-lobes are a different Rh carbonyl or a CO-decorated cluster. This does not invalidate the DFT-supported square-planar model, but it does limit the strength of the 'direct visualization' claim. A single IRAS measurement at the SPM preparation would close the gap. I found no internal contradiction severe enough to reject or to lower the verdict below CONDITIONAL; the occasional wording tension around why clusters are IR-invisible is not load-bearing. The reader's CONDITIONAL verdict is therefore the right one, and no change is needed.","tokens_in":17327,"tokens_out":9905,"duration_ms":125739,"concrete_test":"Run IRAS on a sample prepared exactly as the SPM sample (0.005 ML Rh deposited at 100 K, followed by 1 L CO at 100 K and annealing to ≈270 K) and look for the 2104/2043 cm−1 pair. Detection of the pair would show that the SPM-observed species is spectroscopically active under identical preparation conditions, whereas absence of the pair would mean the SPM double-lobes cannot be assigned to the IRAS gem-dicarbonyl without further evidence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing step is the identification of the STM/nc-AFM double-lobed features (Figs. 2a,b and 3) with the Rh gem-dicarbonyl that produces the 2104/2043 cm−1 IRAS doublet (Fig. 1). The SPM sample was prepared at 0.005 ML Rh, 1 L CO, annealed to ≈270 K, whereas the IRAS sample used 0.05 ML Rh, 0.5 ML CO, annealed to 200–250 K. The assignment of the SPM lobes rests on the structural match to the DFT+U minimum (Fig. 2c) and on the disappearance/reappearance behavior upon annealing, but no measurement spectroscopically tags the imaged lobes. Because the paper also detects an unassigned 2113 cm−1 carbonyl species whose fate on annealing is explicitly left open, the SPM lobes could in principle be a different Rh carbonyl or a small CO-decorated Rh cluster. If so, the claimed first direct visualization and the [001] orientation conclusion would lose their direct observational support, even though DFT and IRAS would still favor the square-planar model. This is an addressable experimental gap, not an internal inconsistency; the multi-technique consistency is otherwise credible.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a multi-technique surface-science study of rhodium gem-dicarbonyls, Rh(CO)2, on rutile TiO2(110). The authors synthesize the species by depositing Rh atoms at low temperature, exposing them to CO, and annealing to 200–250 K. Infrared reflection absorption spectroscopy (IRAS) shows peaks at 2104 and 2043 cm−1, which they assign to the symmetric and asymmetric C–O stretches of Rh+(CO)2. DFT+U and HSE06 calculations reproduce these frequencies and predict a square-planar geometry in which the two CO molecules are aligned along the [001] direction, parallel to the bridging O rows. Low-temperature STM and nc-AFM images show bright double-lobed features oriented along [001] on top of bridging O rows, which the authors identify as the gem-dicarbonyl complexes. XPS shows multiple Rh species, including a component at 310.1 eV that correlates in thermal behavior with the IRAS gem-dicarbonyl bands and a lower-binding-energy component assigned to Rh clusters that are not detected by IR. The central claims are that this is the first direct visualization of the gem-dicarbonyl on this surface, that the geometry is square-planar with CO along [001], and that IR-invisible Rh clusters coexist with the gem-dicarbonyl species.","tokens_in":17530,"tokens_out":4150,"duration_ms":47613,"significance":"If the identification of the scanning-probe features with the IRAS-detected gem-dicarbonyl is correct, the paper provides the first direct structural visualization of a Rh gem-dicarbonyl on a well-defined oxide surface, resolving a previous conflict between IR-based alignment assignments (Hayden et al. versus Tang et al.). The multi-technique approach is well suited to the single-atom catalysis problem, where IR spectroscopy alone can miss or misassign species. Strengths include the use of two independent DFT functionals with a clearly stated gas-phase CO scaling procedure, the combination of XPS and IRAS thermal-evolution data, and the explicit discussion of IR-invisible species. The paper is generally careful and does not overstate the theoretical agreement; the main gap is the lack of a direct spectroscopic tag connecting the imaged double-lobed features to the 2104/2043 cm−1 IRAS species.","major_comments":[{"comment":"The central claim that the STM/nc-AFM double-lobed features are Rh gem-dicarbonyls is not spectroscopically verified. The SPM sample was prepared at 0.005 ML Rh with 1 L CO and annealed to about 270 K, whereas the IRAS sample used 0.05 ML Rh with 0.5 ML CO and annealing at 200–250 K. The manuscript offers no measurement that tags the imaged lobes as the species producing the 2104 and 2043 cm−1 IRAS bands; the assignment rests on the geometric match to the DFT minimum and on the thermal behavior of separate samples. Because the paper also leaves the 2113 cm−1 carbonyl species unassigned and explicitly notes the presence of clusters in the SPM images, the possibility remains that the double lobes belong to a different Rh carbonyl or a CO-decorated cluster. To support the \"first direct visualization\" claim, the authors should either acquire IRAS under the SPM preparation conditions, perform a CO-removal/re-imaging experiment, or clearly temper the claim to reflect the indirect nature of the assignment.","section":"Section 3.2, Figs. 2 and 3"},{"comment":"There is an internal inconsistency about the CO coverage on the IR-invisible Rh clusters. In the Discussion, the authors state that the IR-invisible species likely bind \"to very few or no CO molecules, or the dipole moment of those CO species being too low,\" which leaves open the possibility that the clusters are bare or nearly bare. The Conclusions, however, assert that the clusters are \"almost certainly covered by CO at these temperatures.\" These statements should be reconciled. If the clusters are indeed CO-covered, the absence of their IR signal requires a quantitative explanation, for example a strong orientation/selection-rule effect or very low intrinsic intensity; if they are not CO-covered, the concluding sentence should be revised to avoid an unsupported assertion.","section":"Section 4 vs. Conclusions"},{"comment":"The 2113 cm−1 peak is observed alongside the gem-dicarbonyl doublet and disappears upon annealing to 250 K, yet the manuscript explicitly leaves it unassigned and states that it is unclear whether it transforms into the \"normal\" gem-dicarbonyl or into Rh clusters. This is a load-bearing ambiguity because the SPM images are assigned partly on the basis of the IRAS sample preparation, and the 2113 cm−1 species may represent a second Rh carbonyl that could appear in the SPM images. The authors should either provide a candidate assignment (with supporting DFT calculations or literature comparison) or state more explicitly how the presence of this species affects the interpretation of the SPM and XPS data.","section":"Section 3.1.1, Fig. 1 and Section 4"}],"minor_comments":[{"comment":"Typo: \"located hat higher binding energies\" should read \"located at higher binding energies.\"","section":"Section 4"},{"comment":"The phrase \"The p signal is more pronounced in our design compared to other setups\" would benefit from a brief explanation of what \"p signal\" refers to (presumably the p-polarized reflectivity difference or the signal-to-noise of the p-polarized measurement).","section":"Section 2"},{"comment":"The sentence \"The opposite orientations of these inequivalent CO molecules\" is ambiguous; the authors appear to mean opposite asymmetries or opposite tilts of the two lobes, not opposite orientations in a crystallographic sense. Please rephrase.","section":"Section 3.2, Fig. 3"},{"comment":"Reference 10 is formatted inconsistently (\"Yang C., G. C. W.\"); please correct the author list and title formatting.","section":"References"},{"comment":"The sentence \"We have observed these vibrational frequencies of the carbonyl species most frequently in our experiments\" is awkward; consider rephrasing to \"In most experiments, the gem-dicarbonyl bands appeared at 2104 and 2043 cm−1, with occasional 1–4 cm−1 shifts...\"","section":"Section 3.1.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is a strong multi-technique study and the core chemical interpretation is likely correct, but the direct-visualization claim needs either additional experimental support (e.g., IRAS at the SPM preparation conditions) or a more cautious wording. The internal inconsistency about CO coverage on the IR-invisible clusters should be fixed. These are addressable within the scope of a revision; I do not see a fundamental flaw that would require rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real news is the imaging. This paper shows the first low-temperature STM and nc-AFM pictures of isolated Rh gem-dicarbonyls on TiO2(110), and they look exactly like the DFT square-planar structure with both CO ligands along [001]. That directly contradicts Hayden et al.'s perpendicular alignment and reinforces the Tang/Sautet prediction. The IRAS data, including the polarization-dependent peak signs, independently point the same way. That concordance is the paper's strength: IR, SPM, XPS, and theory are mutually consistent and the orientation conclusion rests on direct visualization, not just a frequency match. The DFT frequencies are scaled by gas-phase CO, which is a standard but real free parameter; the authors are transparent about it. The XPS story—that IR-invisible Rh clusters coexist with the gem-dicarbonyls—is well argued and quantitatively supported by the C 1s intensity accounting. This is a solid, useful benchmark paper for single-atom catalysis on oxides.\n\nThe main soft spot is exactly what the stress-test note flags: the SPM features are not spectroscopically tagged as the 2104/2043 cm−1 species. The SPM prep used 0.005 ML Rh and ~270 K anneal; IRAS used 0.05 ML and 250 K. That is a real gap, but a narrow one. The double-lobed geometry and the DFT match give strong circumstantial identification, and the 2113 cm−1 mystery band is honestly left unassigned. Could the SPM lobes be a CO-decorated cluster or a different carbonyl? In principle yes, but the lobes sit on bridge-oxygen rows with the exact spacing expected for the gem-dicarbonyl, and they disappear/reappear with annealing in a way that tracks the IR bands. I think the concern is conditional, not fatal: a single IRAS run at the exact SPM preparation would close it. The paper would be stronger if the authors did that or explicitly acknowledged why they could not.\n\nMinor issues: the 2113 cm−1 species deserves more than a footnote, and the 'inequivalent lobes' explanation via OH interaction is plausible but speculative—though they say so.\n\nWho is this for? Surface scientists and anyone using IR to fingerprint single-atom catalysts. It deserves serious peer review; the imaging alone justifies it. I would send it to review with a request to address the prep mismatch, and I would cite it for the orientation result.","headline":"Direct STM/nc-AFM images settle the 25-year orientation debate for Rh gem-dicarbonyls on TiO2(110), with a modest but addressable gap between the imaged and infrared-detected species.","tokens_in":18181,"tokens_out":851,"would_cite":true,"duration_ms":11966,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Rh gem-dicarbonyls on TiO2(110) are square-planar, aligned along [001], and infrared-invisible Rh clusters coexist with them.","keywords":["rhodium gem-dicarbonyl","TiO2(110)","single-atom catalysis","scanning tunneling microscopy","non-contact atomic force microscopy","infrared reflection absorption spectroscopy","X-ray photoelectron spectroscopy","density functional theory"],"falsifier":"A direct test would be to run IRAS and STM on one sample at the same Rh coverage and annealing temperature, so the number of [001]-oriented double-lobes per area can be compared quantitatively with the 2104/2043 cm−1 band intensities as CO is dosed and annealed; if the two do not track together, the image-to-spectrum assignment is wrong. A second test is to heat an SPM-prepared surface past the dicarbonyl decomposition temperature and then re-dose CO: if the double-lobe features reappear without the dicarbonyl IR bands, or persist after the bands are gone, the imaged species is not the gem-dicarbonyl. Measuring XPS and IR on the same preparation would also settle whether the 310.1 eV component and the two CO stretch bands grow and decay together on the same sample.","tokens_in":17092,"feed_emoji":"🔬","tokens_out":10224,"duration_ms":95574,"temperature":0.7,"pith_summary":"The paper sets out to establish what rhodium gem-dicarbonyls—individual Rh atoms each holding two CO molecules—look like and where they sit on the model surface rutile TiO2(110), and to show that infrared spectroscopy alone can misrepresent such systems. Combining IRAS, STM, nc-AFM, XPS, and DFT, it argues that the dominant Rh(CO)2 species is square-planar with both CO molecules aligned along the [001] direction on top of the bridging oxygen rows, a geometry that matches theory and contradicts an earlier claim of perpendicular alignment. It further reports that XPS detects more Rh species than the IR spectra reveal, specifically small Rh clusters that carry little or no CO and are therefore invisible to infrared. If correct, the work provides the first direct visualization of this catalytic intermediate on a single-crystal oxide and a concrete warning that IR-only studies of single-atom catalysts can overstate sample uniformity. A sympathetic reader would take the paper's central contribution to be the convergence of four independent techniques on one structural assignment, plus the demonstration that missing species can hide in plain sight.","feed_headline":"Direct STM images show Rh gem-dicarbonyls lie along titania's [001] rows","feed_subtitle":"IR bands, atomic-scale images, and theory agree on square-planar Rh(CO)2, while XPS finds hidden Rh clusters.","key_machinery":"The central object is the gem-dicarbonyl complex Rh+(CO)2 on TiO2(110): a single Rh adatom coordinated by two surface bridging-oxygen anions and two CO molecules in a square-planar arrangement. The argument is carried by the coincidence of four independent measurements on this object: the symmetric/asymmetric IRAS pair at 2104/2043 cm−1; the azimuthal dependence of p-polarized IRAS on a dielectric single crystal, which lets the orientation of the CO dipoles be read from the spectra; the double-lobed STM and nc-AFM contrast with [001] orientation and a position centered between bridging O atoms; and DFT+U with HSE06 frequency calculations that reproduce the 59 cm−1 mode splitting. A supporting mechanism is the initial-state approximation for Rh 3d core-level shifts combined with Bader charge analysis, which the paper uses to argue that the high-binding-energy XPS component is a Rh+ carbonyl rather than a Rh3+ species.","core_discovery":"On rutile TiO2(110), rhodium gem-dicarbonyls prepared by depositing Rh at low temperature, dosing CO, and annealing to 200–250 K are square-planar Rh+(CO)2 units whose two CO ligands lie along the [001] crystallographic direction, parallel to the Ti and bridging-O rows. The paper reports direct STM and nc-AFM images of these complexes as double-lobed features centered between two bridging oxygen anions, with the two lobes oriented along [001], and it assigns the IRAS bands at 2104 and 2043 cm−1 to their symmetric and asymmetric CO stretches on the basis of HSE06 frequencies of 2098.9 and 2042.6 cm−1. The XPS Rh 3d5/2 feature at 310.1 eV that appears and disappears with the dicarbonyl bands is assigned to this complex; despite the high binding energy, Bader analysis (+0.68 e) and the absence of multiplet splitting place the Rh in the +1 state, not +3. The paper also establishes that other Rh species coexist with the dicarbonyls: a substantial cluster population is visible in STM, and the C 1s signal after oxide CO desorption is only about half of what it would be if every Rh atom carried two CO molecules.","pith_inferences":["If the image-to-spectrum assignment holds, the double-lobe contrast gives a site-by-site way to count gem-dicarbonyls and watch them form, distort, and decompose on TiO2(110), something ensemble IR measurements cannot do; a natural next step would be a temperature-programmed STM series on one and the same sample.","The IR-invisible cluster population suggests an explicit test for operando single-atom catalysis studies: whenever IR shows only gem-dicarbonyl bands, an independent counting technique such as XPS, STEM, or TPD should be used to check whether undetected clusters could be carrying the catalytic turnover.","The OH-induced asymmetry seen here implies that hydroxyl coverage could be used deliberately to tune CO stretching frequencies and binding strengths of surface carbonyls; DFT could map which OH arrangements produce which frequency shifts and predict whether the distorted species is more or less reactive.","The same dielectric-IRAS azimuthal analysis could be ported to other metal-oxide-supported gem-dicarbonyls such as Ir, Pd, Pt, and Ni to determine their orientation even on surfaces where scanning probe imaging is difficult."],"forward_implications":["On TiO2(110), the square-planar, [001]-aligned Rh(CO)2 structure, not the perpendicular one proposed by earlier work, is the starting geometry for interpreting Rh carbonyl chemistry on this surface.","Infrared-only characterization of supported Rh can miss a substantial population of clustered species, so conclusions about single-atom uniformity drawn from IR band intensities alone should be treated with caution.","Rh gem-dicarbonyls on TiO2(110) are stable only up to roughly 250 K; above that they decompose and CO-covered Rh clusters form, which delimits the temperature window in which this intermediate can participate in catalysis.","Surface hydroxyl groups can break the symmetry of the two CO ligands in the imaged complexes, a distortion the paper links to a plausible mechanism for tuning the geometry and reactivity of the species.","The low-temperature synthesis route—Rh deposition at 80 K followed by CO exposure and mild annealing—avoids the high CO pressures and chloride contamination of earlier preparations and is proposed to transfer to other carbonyl-forming metals."],"supporting_citations":[{"why":"Supplies the DFT prediction that the Rh gem-dicarbonyl aligns along the bridging O rows and is stable under reducing CO conditions; this is the configuration the SPM and IRAS results confirm.","marker":"16"},{"why":"Describes the IRAS setup with incidence-angle selection that lets the paper infer CO dipole orientation from p-polarized spectra on a dielectric surface.","marker":"24"},{"why":"Earlier preparation and IR assignment of Rh geminal dicarbonyl on TiO2(110), giving the 2104/2043 cm−1 bands the paper builds on.","marker":"44"},{"why":"Shows that Rh adatoms can be deposited and immobilized on TiO2(110) at 80 K and provides Rh 3d binding energies used for the XPS assignment.","marker":"46"},{"why":"Characterizes CO adsorption on TiO2(110), including its desorption temperature and IR bands, allowing the paper to separate CO on the oxide from CO on Rh.","marker":"47"},{"why":"Earlier FT-RAIRS claim of perpendicular alignment of Rh gem-dicarbonyl on TiO2(110), which the present data contradict.","marker":"50"},{"why":"Reports room-temperature agglomeration of Rh gem-dicarbonyls on TiO2(110) and attributes it to hydrogen/hydroxyls, supporting the stability window and OH interaction.","marker":"58"},{"why":"Provides the XPS Rh 3d binding-energy reference for Rh gem-dicarbonyl on alumina used to assign the 310.1 eV component.","marker":"59"}],"fun_headline_variants":["STM images reveal Rh(CO)2 sit along titania's [001] rows","Square-planar Rh(CO)2 on TiO2(110) imaged directly","IR, STM, AFM, and theory pin Rh gem-dicarbonyls on TiO2","Hidden Rh clusters lurk beneath clean IR spectra of dicarbonyls","Rh gem-dicarbonyls on TiO2(110): direct images settle geometry"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the double-lobed features imaged by STM and nc-AFM are the same chemical species that produces the 2104 and 2043 cm−1 infrared bands, even though the microscopy and spectroscopy samples were prepared at different Rh coverages and slightly different annealing temperatures and the imaged complexes were not spectroscopically tagged.","fun_headline_variants_meta":{"raw":{"variants":["STM images reveal Rh(CO)2 sit along titania's [001] rows","Square-planar Rh(CO)2 on TiO2(110) imaged directly","IR, STM, AFM, and theory pin Rh gem-dicarbonyls on TiO2","Hidden Rh clusters lurk beneath clean IR spectra of dicarbonyls","Rh gem-dicarbonyls on TiO2(110): direct images settle geometry"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000305,"raw_usage":{"total_tokens":1823,"prompt_tokens":1088,"completion_tokens":735,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":704,"completion_tokens_details":{"reasoning_tokens":623}},"tokens_in":704,"tokens_out":735,"duration_ms":7320,"temperature":1.0,"reasoning_tokens":623,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:33:49.224516+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test would be to run IRAS and STM on one sample at the same Rh coverage and annealing temperature, so the number of [001]-oriented double-lobes per area can be compared quantitatively with the 2104/2043 cm−1 band intensities as CO is dosed and annealed; if the two do not track together, the image-to-spectrum assignment is wrong. A second test is to heat an SPM-prepared surface past the dicarbonyl decomposition temperature and then re-dose CO: if the double-lobe features reappear without the dicarbonyl IR bands, or persist after the bands are gone, the imaged species is not the gem-dicarbonyl. Measuring XPS and IR on the same preparation would also settle whether the 310.1 eV component and the two CO stretch bands grow and decay together on the same sample.","supporting_citations":[],"review_version":1}