{"id":"034a3293-24a8-451e-9ce0-1d60cbe9e8a5","arxiv_id":"2412.00951","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"ADAPT-VQE with a DFT-derived active space reproduces PBE-D3 binding energies for triazole inhibitors on Al(111), but adds no new chemical prediction beyond the classical reference.","lead":"Scientists tested a hybrid quantum-classical workflow that combines density functional theory with an ADAPT-VQE quantum algorithm to compute how two triazole corrosion inhibitors bind to an aluminum surface. The quantum algorithm reproduced the classical DFT binding energies, and a warm-started variant was several times faster, offering a template for applying quantum computing to practical materials problems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The (2e,5o) active space is likely chemically inert for these physisorbed inhibitors, so the near-exact AdaptVQE/DFT binding energies do not validate a quantum treatment of corrosion inhibition.","rationale":"Read in good faith, the paper is a useful engineering report: it integrates CP2K, Qiskit, and Braket, provides code, benchmarks LiH, and reports a chemically plausible ordering (thiol binds more strongly than triazole). None of those facts are in dispute. The issue is the interpretation of the near-identical binding energies as demonstrating viability of a quantum approach to corrosion inhibition. For the claim to hold, the active space must contain the orbitals that form the adsorbate–surface interaction. Three features of the manuscript make this assumption insecure. First, the selection rule is purely energetic ('around the Fermi level') on a metallic slab; a 4×4 Al(111) cell has many near-degenerate metal states, so the five orbitals closest to the Fermi level need not involve the molecule. Second, the equilibrium distances (3.54 Å, 3.21 Å) place the molecules in a physisorption regime where dispersion (DFT-D3) dominates the binding; such contributions are not represented in the second-quantized active-space Hamiltonian. Third, the agreement between AdaptVQE and classical DFT is exact to ~1e-6 eV, which is precisely what one expects if the active-space electrons are weakly coupled and contribute no significant correlation to the binding-energy difference. The paper's own limitation statement (Section III) concedes the active space may be too small. The vanilla VQE outlier is a red flag for robustness, but the more fundamental point is that the demonstration does not yet establish that any quantum component is describing the corrosion-relevant physics. The reader's weakest_assumption identified the same issue; this stress-test sharpens it into a falsifiable orbital-composition check. Because the reader already assigned CONDITIONAL, and the condition is exactly that the active space captures the bonding, the verdict remains CONDITIONAL: accept if the orbital-character test passes and the binding energy is stable under a modestly larger or CDD-based active space; otherwise the conclusions should be limited to 'embedding infrastructure demonstration' rather than 'viable quantum approach to corrosion inhibition.'","tokens_in":12116,"tokens_out":9375,"duration_ms":90369,"concrete_test":"Using the published GitHub inputs, extract the five active-space orbitals from the CP2K/Qiskit embedding step and compute the Mulliken weight of the inhibitor atoms in each orbital. If the summed squared weight on the inhibitor is below 30%, the active space is composed mainly of Al surface states and the reported AdaptVQE/DFT agreement does not demonstrate a quantum treatment of the adsorption bond; if the weight is high, the concern is alleviated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central 'viability' claim depends on the (2e,5o) active space describing the inhibitor–Al(111) bond. That assumption is unverified and probably false. The adsorbates sit at 3.54 Å (triazole) and 3.21 Å (thiol) from the surface; at these distances the binding is dominated by dispersion, which enters only through the classical DFT-D3 term and not through the active-space Hamiltonian. The active space is selected by canonical orbital energy ordering around the Fermi level of a 4×4 Al slab, which has a quasi-continuum of metallic states; without an orbital-character projection there is no reason to expect the five chosen orbitals to include the triazole frontier orbitals or the S lone pair. The paper's S4 assertion that the analysis 'revealed significant electronic contributions' from the π-system and sulfur lone pair is not supported by any orbital-population data. A weakly coupled two-electron active space contributes only a tiny correlation correction, so the 1e-6 eV agreement between AdaptVQE and classical DFT is the expected result for a chemically irrelevant active space, not a validation of ADAPT-VQE. The paper itself concedes in Section III that 'a larger active space might better capture these electronic coupling effects in full.' The unexplained vanilla VQE outlier (-2.326 eV) further shows that the quantum results are sensitive to algorithmic details, reinforcing that the reported agreement is not a robust property of the workflow.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a hybrid quantum-classical workflow for computing binding energies of corrosion inhibitors on Al(111). Classical periodic DFT (PBE-D3) is combined with an active-space embedding scheme: a (2e,5o) active space around the Fermi level is solved by VQE variants (vanilla VQE, ADAPT-VQE, StatefulAdaptVQE) using Qiskit and CP2K. For 1,2,4-Triazole and 1,2,4-Triazole-3-thiol, the reported ADAPT-VQE binding energies (-0.385508 eV and -1.279064 eV) agree with classical DFT to within a few micro-eV, while vanilla VQE yields -2.326 eV for the triazole. Benchmarks on LiH and H2 are used to support claims of algorithmic speedup and error-mitigation performance. The paper concludes that the workflow demonstrates the viability of hybrid quantum-classical approaches for corrosion inhibition and, more broadly, for materials science applications.","tokens_in":12462,"tokens_out":6702,"duration_ms":65547,"significance":"The manuscript's main strength is that it makes the implementation available as open-source code and integrates established tools (CP2K, Qiskit Nature, Braket) into a periodic active-space embedding workflow. The benchmarks on LiH and the error-mitigation analysis on H2 are useful reference points. If the corrosion results were robust, the paper would be a useful proof-of-concept for quantum-chemistry embedding in surface-adsorbate systems. However, the central demonstration is not convincing: the active space is very small, the agreement with classical DFT is essentially a self-consistency check, the single non-agreeing vanilla-VQE result is not adequately explained, and no error bars or convergence data are reported. The paper does not provide a new chemical prediction beyond the classical DFT and experimental results already in the literature.","major_comments":[{"comment":"The micro-eV agreement between ADAPT-VQE and classical DFT binding energies is not evidence that the quantum calculation captures the inhibitor-surface bond. At adsorption distances of 3.54 Å and 3.21 Å, the binding is dominated by dispersion, which enters only through the classical DFT-D3 term and is absent from the active-space Hamiltonian in Eq. (S1). Moreover, the active space is selected by canonical orbital energy ordering around the Fermi level of a metallic Al slab; no orbital-character projection is provided to show that the five orbitals contain the inhibitor's π system or the sulfur lone pair. The S4 claim that the analysis revealed significant electronic contributions from these orbitals is unsupported by any population, charge-density-difference, or occupation data. The agreement therefore largely reflects the quantum solver reproducing the DFT-derived active-space energy, not the physics of adsorption.","section":"Section III, Table II and S5, Eq. (1)"},{"comment":"The vanilla VQE result for 1,2,4-Triazole (-2.326 eV) is a serious outlier. Attributing this ~2 eV deviation to a looser embedding convergence threshold (2E-5 vs 1E-6 Ha) is not credible without convergence evidence: a threshold change of 1e-5 Ha cannot plausibly produce an energy shift of this magnitude. No error bars, repeated runs, or convergence curves are reported, so the robustness of the workflow is not established.","section":"Section III, Table II"},{"comment":"The claimed 5-6x speedup of StatefulAdaptVQE is demonstrated only for the LiH molecular benchmark (Table V), not for the Al(111) adsorption systems. The abstract's statement that the speedup is achieved 'while maintaining accuracy' is therefore not supported for the corrosion systems. The authors should either provide timings and energy comparisons for the main workflow or explicitly restrict the speedup claim to the molecular test case.","section":"Section III and S3, Tables IV-V"},{"comment":"The paper's own concession that 'a larger active space might better capture these electronic coupling effects in full' is in direct tension with the abstract's claim that the workflow establishes the viability of hybrid quantum-classical approaches for corrosion inhibition. As presented, the (2e,5o) active space is too small to describe adsorbate-substrate hybridization, so the viability claim extends beyond the evidence.","section":"Section III"}],"minor_comments":[{"comment":"The sentence 'In which, can be transferable to other applications...' is grammatically incomplete and should be rewritten.","section":"Abstract and Section II"},{"comment":"The main text states a vacuum gap of 25 Å in the z-direction, while Table III lists a vacuum gap of 40 Å; these values should be reconciled.","section":"Section II vs Table III"},{"comment":"Reference [38] is cited for ADAPT-VQE and for Grimsley et al., but the listed reference is Higgott, Wang, and Brierley on excited-state VQE; the correct ADAPT-VQE reference (Grimsley et al., Nature Communications 10, 3007 (2019)) is missing.","section":"References, [38]"},{"comment":"The H2 benchmark energies are given without specifying the basis set or active space, which makes the error-mitigation comparison difficult to reproduce.","section":"Section S3, Table VI"},{"comment":"The GitHub link is repeated twice in the same paragraph; the second occurrence should be removed or merged with the first.","section":"Section S8"}],"recommendation":"reject","confidential_remarks":"The core issue is that the reported agreement between the quantum and classical results is essentially built into the construction of the active space and the embedding scheme, and the one deviating result is left unexplained. The manuscript would need substantially new validation—such as a chemically meaningful active space, orbital projection data, convergence tests, and error bars—to support its central claim. In its current form, the claims outrun the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The punchline: the paper's real product is the workflow template, not the chemistry. The authors wired CP2K's DFT embedding to Qiskit's ADAPT-VQE, ran it on two triazole inhibitors on Al(111), and posted the code on GitHub. That is genuinely useful. The specific binding energies are new, but the chemical ranking (thiol > triazole) was already known from experiment and appears in their own classical DFT column. What the paper demonstrates is that you can get this pipeline to execute on periodic surface-adsorbate systems, not that VQE adds physical accuracy.\n\nWhat it does well: the implementation is reproducible, parameters are documented, and limitations are stated rather than hidden. The paper says explicitly that a larger active space might capture the electronic coupling, that hardware runs were only on small benchmark molecules, and that the main results are Qiskit simulations. The supplementary disclosure about needing to patch qiskit-braket-provider is honest. For someone setting up the same CP2K/Qiskit interface, the repo is a head start.\n\nThe soft spots. The (2e,5o) active space is selected by canonical orbital energy ordering around the Fermi level of a 4x4 Al slab. There is no orbital-character projection showing that those five orbitals include the triazole pi system or the sulfur lone pair. At the computed adsorption distances (3.54 Å and 3.21 Å) the binding is dominated by dispersion, which enters through the classical D3 term, not through the active-space Hamiltonian. So the micro-eV agreement between AdaptVQE and classical DFT is the expected result for an almost trivial active space, not a validation of the quantum method. The S4 claim that the analysis 'revealed significant electronic contributions' is not backed by any population or charge-density data. The vanilla VQE outlier (-2.33 eV vs -0.39 eV) is waved off with a looser convergence threshold, which is not a real explanation. The headline 5-6x speedup is from LiH, not from the corrosion systems. Minor: reference [38] is cited as Grimsley et al. but is actually Higgott et al., so the ADAPT-VQE citation needs fixing.\n\nWho this is for: researchers who want a working example of CP2K/Qiskit embedding on a periodic slab, and people judging whether quantum embedding is ready for surface chemistry. The chemical conclusions should be read as conditional. I would send it to review as a methods/application note, with major revision: either validate the active space with orbital population analysis, or reframe the claims as an infrastructure demonstration. It is not desk-reject material; it is a useful, honest demo with an overreaching interpretation.","headline":"A reproducible workflow demo for CP2K/Qiskit embedding on periodic slabs, but the quantum-classical agreement is built into the tiny active space, not a chemical validation.","tokens_in":12945,"tokens_out":2996,"would_cite":false,"duration_ms":29584,"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":"This paper claims that ADAPT-VQE, solving a two-electron five-orbital active space embedded in periodic DFT, reproduces the classical DFT binding energies of two triazole corrosion inhibitors on Al(111), and that a warm-started variant…","keywords":["corrosion inhibition","ADAPT-VQE","active space embedding","density functional theory","aluminum surface","1,2,4-triazole","surface-adsorbate interaction","hybrid quantum-classical workflow"],"falsifier":"Take the same two inhibitors on Al(111), but enlarge the active space to $4e,8o$ (four electrons in eight orbitals) and solve with ADAPT-VQE; compare the resulting binding energies to the DFT reference. If the match degrades beyond chemical accuracy, the $2e,5o$ selection was the reason for the reported agreement, not the quantum algorithm itself.","tokens_in":11951,"feed_emoji":"🧪","tokens_out":7215,"duration_ms":63388,"temperature":0.7,"pith_summary":"The paper uses 1,2,4-Triazole and 1,2,4-Triazole-3-thiol on Al(111) as a testbed for a hybrid quantum-classical workflow. It tries to establish that ADAPT-VQE, fed a two-electron five-orbital active space carved from a periodic DFT calculation, reproduces the classical DFT binding energies (-0.386 eV and -1.279 eV) almost exactly, while a fixed-ansatz VQE does not. The stronger thiol binding matches experimental corrosion-inhibition rankings, and a warm-started variant, StatefulAdaptVQE, is 5-6 times faster in benchmarks. The authors read this as evidence that quantum algorithms can be embedded in realistic periodic surface-adsorbate simulations and transferred to other materials problems such as carbon capture and battery materials.","feed_headline":"Quantum solver matches DFT for corrosion inhibitor binding","feed_subtitle":"ADAPT-VQE with a two-electron active space reproduces classical binding energies, and warm starting speeds it up 5-6x.","key_machinery":"The central object is a two-electron, five-orbital active space ($2e,5o$) constructed from canonical orbitals ordered by energy around the Fermi level of the periodic DFT calculation. This active-space embedding reduces the full periodic problem to a small second-quantized Hamiltonian, which is then solved by ADAPT-VQE, a variational algorithm that grows its ansatz by adding, one at a time, the fermionic excitation operators with the largest gradient. A warm-started variant, StatefulAdaptVQE, reuses information from earlier optimization steps and is the source of the 5-6× speedup. The active-space Hamiltonian is mapped to qubits with parity mapping and a two-qubit reduction, and classical parameters are optimized with SPSA.","core_discovery":"The paper reports that ADAPT-VQE with a two-electron, five-orbital active space built from canonical orbitals around the Fermi level reproduces the DFT binding energies of 1,2,4-triazole (-0.386 eV) and 1,2,4-triazole-3-thiol (-1.279 eV) on a 4×4 Al(111) supercell to within about $10^{-5}$ eV. A fixed-ansatz VQE does not, giving -2.326 eV for the unsubstituted triazole, which the authors attribute to looser embedding convergence and lack of adaptive operator selection. The thiol's stronger binding is consistent with its shorter adsorption distance and with experimental literature on sulfur-functionalized inhibitors. The warm-started StatefulAdaptVQE variant reaches the same quality of result 5-6 times faster than plain AdaptVQE in small-system benchmarks, and the paper presents the whole pipeline as a transferable hybrid quantum-classical workflow for periodic surface-adsorbate systems.","pith_inferences":["The authors do not claim a quantum advantage; the $2e,5o$ problem is small enough to be solved classically. A meaningful future test would push the active space to roughly 30 or more qubits, where classical exact diagonalization becomes prohibitive.","Part of the agreement may come from the embedding itself: the DFT calculation sets the one-electron environment, so the quantum solver only handles a nearly trivial two-electron correction. A sharper test would be an active space chosen by charge-density difference rather than orbital-energy ordering alone.","The claimed transfer to carbon capture and battery materials is plausible but untested; those systems add transition-metal d-states or charged interfaces, so the next stress test should involve stronger charge transfer than triazoles on aluminum."],"forward_implications":["The same embedding-plus-ADAPT-VQE pipeline can be applied to other inhibitor/surface pairs without changing the workflow, only the DFT input and active-space selection.","The 5-6× speedup from warm-starting makes iterative screening of many inhibitor candidates practical on simulators, since each additional candidate costs a fraction of a full VQE run.","The computed binding-energy ordering (thiol stronger than triazole) matches the experimental inhibition ranking, so the workflow can rank relative inhibitor strength, not just produce absolute energies.","Because the method uses periodic DFT embedding, it carries over to other periodic problems such as gas adsorption for carbon capture and electrode-electrolyte interfaces for batteries, as the paper explicitly suggests."],"supporting_citations":[{"why":"Supplies the general active-space embedding framework connecting periodic DFT to a quantum solver, which the workflow adapts to surface-adsorbate systems.","marker":"[29]"},{"why":"Defines the ADAPT-VQE ansatz-construction rule used for all production results.","marker":"[38]"},{"why":"Provides the interface between the periodic DFT code and the quantum solver that makes the hybrid calculation possible.","marker":"[30]"},{"why":"Provides the warm-starting technique that the faster StatefulAdaptVQE variant builds on.","marker":"[41]"},{"why":"Defines the PBE exchange-correlation functional used for the DFT reference and the embedding Hamiltonian.","marker":"[32]"},{"why":"Supplies the D3 dispersion correction needed for accurate adsorption binding energies.","marker":"[36]"},{"why":"Experimental benchmark showing sulfur-functionalized triazoles inhibit aluminum corrosion, used to validate the stronger thiol binding.","marker":"[15]"},{"why":"Machine-learning potential with D3 corrections that produces the adsorption geometries and binding distances.","marker":"[52]"}],"fun_headline_variants":["ADAPT-VQE matches DFT for corrosion inhibitor binding","Quantum-classical hybrid speeds corrosion simulation 5-6x","Warm-started VQE yields DFT-level accuracy on surfaces","Triazole corrosion inhibitors benchmarked via VQE","Quantum algorithm simulates inhibitor adsorption on Al"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a two-electron, five-orbital active space picked by orbital-energy ordering around the Fermi level captures the electronic coupling that determines inhibitor-aluminum binding.","fun_headline_variants_meta":{"raw":{"variants":["ADAPT-VQE matches DFT for corrosion inhibitor binding","Quantum-classical hybrid speeds corrosion simulation 5-6x","Warm-started VQE yields DFT-level accuracy on surfaces","Triazole corrosion inhibitors benchmarked via VQE","Quantum algorithm simulates inhibitor adsorption on Al"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000277,"raw_usage":{"total_tokens":1679,"prompt_tokens":1002,"completion_tokens":677,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":618,"completion_tokens_details":{"reasoning_tokens":600}},"tokens_in":618,"tokens_out":677,"duration_ms":6630,"temperature":1.0,"reasoning_tokens":600,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:49:57.124235+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the same two inhibitors on Al(111), but enlarge the active space to $4e,8o$ (four electrons in eight orbitals) and solve with ADAPT-VQE; compare the resulting binding energies to the DFT reference. If the match degrades beyond chemical accuracy, the $2e,5o$ selection was the reason for the reported agreement, not the quantum algorithm itself.","supporting_citations":[{"cited_title":"A general framework for active space embedding methods: applications in quantum computing","cited_arxiv_id":"2404.18737","evidence_quote":"Supplies the general active-space embedding framework connecting periodic DFT to a quantum solver, which the workflow adapts to surface-adsorbate systems."},{"cited_title":"Higgott, D","cited_arxiv_id":null,"evidence_quote":"Defines the ADAPT-VQE ansatz-construction rule used for all production results."},{"cited_title":"Rossink, qiskit-nature-cp2k, GitHub repository (2024), retrieved on 2024-11-02","cited_arxiv_id":null,"evidence_quote":"Provides the interface between the periodic DFT code and the quantum solver that makes the hybrid calculation possible."},{"cited_title":"Quantum HF/DFT-Embedding Algorithms for Electronic Structure Calculations: Scaling up to Complex Molecular Systems","cited_arxiv_id":"2009.01872","evidence_quote":"Provides the warm-starting technique that the faster StatefulAdaptVQE variant builds on."},{"cited_title":"Grimme, J","cited_arxiv_id":null,"evidence_quote":"Supplies the D3 dispersion correction needed for accurate adsorption binding energies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Experimental benchmark showing sulfur-functionalized triazoles inhibit aluminum corrosion, used to validate the stronger thiol binding."},{"cited_title":"DeNeutoy, B","cited_arxiv_id":null,"evidence_quote":"Machine-learning potential with D3 corrections that produces the adsorption geometries and binding distances."}],"review_version":1}