{"id":"d9a85a9c-bb85-433f-a003-6eb5d6d6b10d","arxiv_id":"2607.08447","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Self-consistent 6×6 k·p SP + CI simulations of a triangular Si FinFET DQD yield gate-tunable exchange J and magnetic orientations that favor high-fidelity SWAP or CZ operations.","lead":"A GPU-accelerated Schrödinger-Poisson plus configuration-interaction framework computes exchange coupling J for hole spins in a 5-gate Si FinFET double quantum dot. The work maps voltage and magnetic-field sweet spots for native SWAP and CZ two-qubit gates on a geometry close to real devices.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.5","headline":"CI truncation to four single-particle states above EC is the load-bearing untested assumption for both J and the magnetic sweet spots.","rationale":"The Reader correctly isolates the CI truncation as the weakest assumption. The manuscript supplies no convergence table or supplementary figure that enlarges the single-particle manifold beyond the EC cutoff, so the quantitative reliability of both the exchange spectrum (Fig. 5) and the angular maps (Fig. 7) remains unproven. All other elements of the pipeline (6×6 k·p SP, strain, Slater–Condon construction) are standard and internally consistent; the experimental g-tensor mismatch is already acknowledged by the authors and does not undermine the internal logic of the simulation. Because the Reader already assigned CONDITIONAL precisely on this ground, the stress-test does not alter the verdict. The concrete basis-enlargement test above would settle the residual uncertainty.","tokens_in":22971,"tokens_out":587,"duration_ms":6006,"concrete_test":"Re-run the CI pipeline of Fig. 2 at the bias points of Fig. 6(b) and at the two magnetic orientations marked by the orange and red dots in Fig. 7, once with the original four-state cutoff and once with the next two doubly-degenerate single-particle states included (K=66). If either |J| changes by more than ~10 % or the (θ,ϕ) locations of the J⊥=0 / max-J⊥ extrema move by more than a few degrees, the sweet-spot claims are basis-size sensitive and the truncation assumption fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim rests on a reduced CI basis of the four highest single-particle states (K=28 determinants) lying above the charging energy EC≈3.34 meV (bonding-to-first-excited gap, Fig. 3(b) and §II.C). The paper asserts this truncation is “physically meaningful” and “sufficient,” yet never reports a systematic enlargement of the basis (e.g., six or eight states, or a lower energy cutoff) to show that the extracted exchange J(εVP,VB) and, more critically, the angular locations of the J⊥-dominated SWAP and J∥-dominated CZ sweet spots (Fig. 7) remain stable. Because the subsequent effective Hamiltonian (Eq. 17) and the anisotropic exchange matrix ˜J are built from these CI energies and from g-tensors of the same truncated single-particle manifold, any contamination of the singlet–triplet gap or of the bonding/antibonding character by higher-lying states would shift both the magnitude of J and the magnetic-field orientations claimed to be optimal. The truncation is therefore the single least-secure condition on which the design conclusions depend.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript presents a GPU-accelerated 6\times6 k·p Schrödinger–Poisson solver coupled to a configuration-interaction (CI) treatment of two-hole states for a 5-gate triangular Si FinFET double quantum dot. Self-consistent electrostatic potentials and single-particle eigenstates (including cooling-induced strain via Pikus–Bir) are used to build a reduced Slater-determinant basis; diagonalization yields the singlet–triplet spectrum and exchange J as functions of plunger detuning and barrier voltage. An effective anisotropic-exchange Hamiltonian (Eq. 17) constructed from the same single-particle manifold and experimental spin-orbit parameters is then used to map magnetic-field orientations that separately favor native SWAP (J⊥-dominated) and CZ (J∥ finite, J⊥≈0) gates. Qualitative comparison with the experimental g-tensors and δEZ of Ref. [42] is provided.","tokens_in":23278,"tokens_out":960,"duration_ms":11011,"significance":"If the reduced-basis CI results and the resulting magnetic sweet spots are robust, the work supplies a concrete, device-level design tool for hole-spin two-qubit gates in industrially relevant FinFET geometries. The combination of a full 3-D SP solver, strain, and CI is a non-trivial technical advance over purely phenomenological Fermi–Hubbard fits, and the explicit identification of field orientations that suppress leakage for CZ while preserving flip-flop for SWAP is of immediate experimental interest. The GPU acceleration and the transparent documentation of the Slater–Condon construction further enhance the paper’s utility as a methodological reference.","major_comments":[{"comment":"Section II.C and Figs. 3–5: the central design claims (J(εVP,VB) surfaces and the angular locations of the SWAP/CZ sweet spots in Fig. 7) rest on a CI space truncated to the four highest single-particle states above the charging energy EC≈3.34 meV (K=28 determinants). No systematic enlargement of the basis (six or eight states, or a lower energy cutoff) is reported to demonstrate that the extracted singlet–triplet gap and, more critically, the bonding/antibonding character that enters the g-tensors and the anisotropic ˜J matrix remain stable. Because Eq. (17) and the subsequent J⊥/J∥ maps are built directly from this truncated manifold, a convergence test is load-bearing for the claimed optimal operating conditions.","section":null},{"comment":"Section III.C and Fig. 8: the simulated left- and right-dot g-tensors are nearly identical, while the experimental tensors of Ref. [42] are almost orthogonal; the resulting δEZ angular maps therefore differ quantitatively. The paper attributes the discrepancy to fabrication imperfections but does not quantify how such asymmetries would shift the sweet-spot locations of Fig. 7. Without at least a sensitivity analysis (e.g., artificially tilting one g-tensor), the claim that the simulated orientations remain optimal for real devices is under-supported.","section":null}],"minor_comments":[{"comment":"Fig. 5 caption and surrounding text: the numerical noise of order 10−8 eV near zero detuning is acknowledged but not quantified for the exchange values used later; a short statement of the practical lower bound on reliable J would help.","section":null},{"comment":"Eq. (17) and the paragraph that follows: the spin-orbit length λso and axis nso are taken from experiment without listing their numerical values or uncertainty; a brief table or appendix entry would improve reproducibility.","section":null},{"comment":"Fig. 4: the energy labels and the gray-scale coding of the six single-particle states become hard to read once the detuning exceeds ~15 mV; a clearer legend or separate panels would help.","section":null},{"comment":"Throughout: several typographical slips remain (“BowenmV”, “SW AP”, “CP HASE”, inconsistent spacing around units). A careful proof-reading pass is needed.","section":null}],"recommendation":"major_revision","confidential_remarks":"The methodological core is solid and the GPU-SP + CI pipeline is a genuine contribution. The two major points (CI convergence and g-tensor sensitivity) are fixable with additional calculations that fit within the existing framework; once addressed, the paper should be suitable for a specialized quantum-device or condensed-matter journal. Fit to a broad-interest venue is more marginal because the quantitative experimental agreement remains only qualitative."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The new material here is concrete: J(VB, ε) surfaces and the full angular maps of J⊥ and J∥ that pick out separate (θ, ϕ) orientations for native SWAP versus CZ on this specific 5-gate triangular Si FinFET. That is useful design data the community does not already have for this geometry.\n\nWhat they do well is the pipeline. GPU-accelerated 3-D 6×6 k·p SP with cooling strain, then a carefully documented CI construction via Slater–Condon rules, produces the expected singlet–triplet spectrum, anti-crossing, and exchange tuning. The reduced basis (four single-particle states above EC ≈ 3.34 meV → K = 28) is motivated by the charging energy and by the clear energy gaps in Fig. 4; the resulting spectrum in Fig. 5 looks textbook. They also own the quantitative mismatch with the experimental g-tensors (Fig. 8) and correctly attribute it to real-device asymmetry rather than claiming victory.\n\nThe stress-test concern is fair but overstated. The paper never shows a systematic basis enlargement, so we do not know how stable the sweet-spot angles in Fig. 7 are under six or eight states. That is the softest load-bearing assumption. Still, the energy hierarchy they plot makes contamination of the low-lying S–T gap unlikely to be large, and the qualitative design message (there exist orientations where J⊥ vanishes while J∥ remains finite, and vice versa) is robust. Other free parameters (λso, nso, strain model) are taken from the literature or the same experiment; that is normal for this class of work and does not circularize the J surfaces themselves.\n\nMath and citation pattern look solid; no invented entities. This is for people who actually design or simulate hole-spin FinFET qubits. I would bring it to reading group, cite the sweet-spot maps if I were working on similar devices, and send it to peer review. Ask the authors for a short basis-size check and, if possible, code release; otherwise accept with minor revision.","headline":"Solid device-level CI maps of J and magnetic sweet spots for a realistic hole FinFET DQD; the truncation is a real but not fatal soft spot, and the experimental g-tensor mismatch is already owned.","tokens_in":23868,"tokens_out":539,"would_cite":true,"duration_ms":6391,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Reduced-basis simulations of a 5-gate Si FinFET double quantum dot map exchange coupling and locate magnetic orientations that separately enable native SWAP and CZ two-qubit gates.","keywords":["hole spin qubits","double quantum dot","exchange coupling","FinFET","configuration interaction","Schrödinger-Poisson","two-qubit gates","silicon"],"falsifier":"Re-running the identical bias and field scan with a substantially larger CI basis (or with measured device-to-device geometric disorder) would show whether the predicted J values and the angular locations of the SWAP/CZ sweet spots remain stable or shift outside experimental error bars.","tokens_in":23880,"feed_emoji":"🔬","tokens_out":647,"duration_ms":6740,"temperature":0.7,"pith_summary":"This paper builds a GPU-accelerated, self-consistent Schrödinger-Poisson solver for hole states in a realistic 5-gate silicon FinFET and feeds the resulting single-particle orbitals into a configuration-interaction two-particle Hamiltonian. The goal is to compute the exchange coupling between the two quantum dots and to show that a deliberately truncated basis (states above the charging energy) already recovers the expected singlet-triplet spectrum and its dependence on detuning and barrier voltage. With that exchange in hand, the authors map how an external magnetic field orientation splits the interaction into transverse and longitudinal pieces, thereby identifying field directions that favor either an exchange-dominated SWAP gate or a Zeeman-dominated CZ gate with suppressed leakage. The same framework is compared with experimental g-tensors and exchange data from a closely related device. The practical payoff is a concrete in-silico route for choosing bias and field conditions that turn a CMOS-compatible FinFET into a high-fidelity two-qubit building block.","feed_headline":"Simulations map SWAP and CZ sweet spots in a Si FinFET qubit","feed_subtitle":"A reduced-basis model of hole exchange shows which magnetic angles favor each native two-qubit gate","key_machinery":"The configuration-interaction Hamiltonian built from Slater determinants of the four highest single-particle hole states lying above the charging energy; its diagonalization supplies the singlet-triplet gap (exchange J) that is then decomposed into transverse and longitudinal components under an applied magnetic field.","core_discovery":"A reduced-basis configuration-interaction treatment constructed on self-consistent 6\times6 k·p Schrödinger-Poisson solutions reproduces the magneto-electrostatic singlet-triplet spectrum and exchange coupling of a 5-gate Si FinFET double quantum dot and locates magnetic-field orientations that separately favor native SWAP (J_perp-dominated) and CZ (J_parallel-dominated with J_perp near zero) two-qubit operations.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Reduced-basis CI maps SWAP and CZ angles in Si FinFET DQD","k·p SP solver finds magnetic sweet spots for FinFET hole qubits","Exchange coupling model locates native SWAP vs CZ orientations","Self-consistent DQD simulations optimize Si FinFET two-qubit gates","Magnetic-field angles separate SWAP and CZ regimes in FinFET dots"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"The two-particle Hilbert space can be cut off after the four highest single-particle states above the charging energy without changing the extracted exchange or the locations of the magnetic sweet spots.","fun_headline_variants_meta":{"raw":{"variants":["Reduced-basis CI maps SWAP and CZ angles in Si FinFET DQD","k·p SP solver finds magnetic sweet spots for FinFET hole qubits","Exchange coupling model locates native SWAP vs CZ orientations","Self-consistent DQD simulations optimize Si FinFET two-qubit gates","Magnetic-field angles separate SWAP and CZ regimes in FinFET dots"]},"model":"grok-4.5","effort":"low","cost_usd":0.003978,"raw_usage":{"total_tokens":1251,"prompt_tokens":786,"num_sources_used":0,"completion_tokens":75,"cost_in_usd_ticks":39780000,"prompt_tokens_details":{"text_tokens":786,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":390,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":786,"tokens_out":75,"duration_ms":4250,"temperature":1.0,"reasoning_tokens":390,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-10T07:18:25.995846+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Re-running the identical bias and field scan with a substantially larger CI basis (or with measured device-to-device geometric disorder) would show whether the predicted J values and the angular locations of the SWAP/CZ sweet spots remain stable or shift outside experimental error bars.","supporting_citations":[],"review_version":1}