{"id":"1a3eb671-c63a-464a-8a6b-a12a699eb758","arxiv_id":"2608.12718","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A combined finite-element and quantum simulation attributes the large gate-voltage tunability of the g-factor in germanium hole spin qubits to inhomogeneous thermal-contraction strain and wavefunction averaging.","lead":"Germanium quantum dot qubits show large g-factor swings when gate voltages change. This paper uses strain simulations and quantum dot wavefunction modeling to argue that thermal-contraction strain creates a spatially varying g-tensor landscape, and that gate voltages shift the dot within it, producing the observed tunability.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 80%/150% g-factor predictions rest on an unvalidated pointwise application of uniform-strain g-tensor formulas and an undefined coefficient κ in Eqs. (9)-(12); a full 4-band check is needed.","rationale":"I agree with the reader's weakest_assumption. The single most load-bearing concern is the step from Eqs. (9)-(12) to the numbers in Fig. 10: the coefficient κ is not defined, and the uniform-strain formula is applied locally in a strongly inhomogeneous strain field with no validation against the full 4-band Hamiltonian. This is not a dispute with a consensus model; it is a missing scale and an uncontrolled approximation in the derivation of the headline values. I did not make the absence of a direct experimental overlay the primary concern, because the paper is explicitly hedged ('may account for') and the central claim is about a plausible mechanism; overlaying the predicted curves on Ref. [9] data would nevertheless be a valuable secondary check. The FEM strain magnitudes matching x-ray measurements and the visually validated Gaussian fits are genuine supports for the strain and wavefunction parts, but neither addresses the κ/local-averaging step. The reader's CONDITIONAL verdict remains appropriate: the paper becomes convincing only if κ is specified and the local-averaging approximation is shown to agree with a direct inhomogeneous calculation.","tokens_in":16716,"tokens_out":13278,"duration_ms":176689,"concrete_test":"Set κ to the value defined in Refs. [12] and [16] (state the number explicitly) and recompute Fig. 10 by direct numerical diagonalization of the 4×4 Luttinger-Kohn Hamiltonian (Eq. 4) with the full simulated inhomogeneous strain tensor and the MaSQE potential for the same gate configurations, instead of using the pointwise local g-tensor averaged over the Gaussian wavefunction. Compare the resulting g*-versus-Brm curves with the published ones. If the curves differ by more than ~30%, or if the 80%/150% variations disappear, the central claim is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equations (9)-(12) are the only bridge between the FEM strain field and the reported g-factor changes. They contain the constant κ, which is never defined. Eq. (9) makes the diagonal correction proportional to 6bvκ/ΔLH (⟨εyy⟩−⟨εxx⟩), and Eq. (12) makes the off-diagonal term proportional to 4√3 dvκ/ΔLH ⟨εxy⟩; the final effective-g curves in Fig. 10 are linear in these corrections. Without κ, a reader cannot reproduce the magnitude of any predicted swing, including the headline 80% (single dot) and 150% (singlet-triplet) changes. The paper also applies these uniform-strain formulas pointwise and then averages over a Gaussian hole, although the simulated strain varies by ~30% across the 16-nm well and on tens of nanometers laterally, i.e., on the same scale as the wavefunction. No convergence test or comparison against a direct solution of the inhomogeneous Luttinger-Kohn Hamiltonian (Eq. 4) is provided. If κ is the standard Luttinger Zeeman parameter, stating its value would remove the ambiguity; if not, the formula is incomplete. Either way, the central quantitative claim is currently unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes that the strong gate-voltage tunability of hole g-factors in germanium quantum dots is caused by device-induced thermal-contraction strain. The authors solve a 3D thermoelastic finite-element problem for a realistic multilayer gate stack, extract inhomogeneous strain fields with magnitudes of order 1e-4, combine these with strain-induced g-tensor corrections taken from Refs. [12,16], and average the corrections over hole wavefunctions obtained from the MaSQE Schrödinger-Poisson solver and then approximated as multivariate Gaussians. For a range of Brm gate voltages and in-plane magnetic-field orientations, they report that the effective g-factor of a single dot varies by over 80% and that of a singlet-triplet qubit by more than 150%, with features such as non-monotonic gate-voltage dependence. The paper frames this as a quantitative explanation of the gate-tunable g-factor observed in recent germanium qubit experiments.","tokens_in":16976,"tokens_out":5423,"duration_ms":68610,"significance":"If the quantitative result held, this would be an important contribution: it would connect an experimentally observed qubit parameter (gate-tunable g-factor) to a concrete physical mechanism (inhomogeneous thermal-contraction strain) and would identify in-plane g-tensor averaging as design-relevant. The paper has genuine strengths: it is a device-scale forward simulation rather than a fit to the target experiment, the strain magnitudes are checked against x-ray measurements, the δg formulas are imported from prior theory rather than tuned here, and the boundary-condition choice is benchmarked against Ref. [12]. However, the headline numbers currently rest on an undefined coefficient κ and on a pointwise application of uniform-strain formulas, so the central quantitative claim is not yet established. The paper also does not quantitatively compare its predicted sensitivity with the experimental one it claims to explain.","major_comments":[{"comment":"The coefficient κ appearing in Eqs. (9) and (12) is never defined anywhere in the manuscript, and the text as typeset uses a lower-case k in Eqs. (10) and (11). No numerical value, material parameter, or reference is given. Because the diagonal correction δgxx and the off-diagonal correction δgxy are both linear in this coefficient, every reported percentage change in Fig. 9 and Fig. 10 scales with κ. The authors must state what κ is (for example, the Luttinger Zeeman parameter), give its value, and cite its source; otherwise the central quantitative claim is unreproducible.","section":"Section III.B, Eqs. (9)-(12)"},{"comment":"The g-tensor correction is computed pointwise from uniform-strain formulas, Eqs. (7)-(12), and then averaged over a Gaussian hole wavefunction, but the simulated strain varies by about 30% across the 16-nm well (Fig. 4) and on lateral scales of tens of nanometers, i.e., on the same scale as the dot wavefunction. The paper provides no convergence test and no comparison against a direct solution of the inhomogeneous Luttinger-Kohn Hamiltonian, Eq. (4), in the simulated strain field. It is therefore not demonstrated that the pointwise uniform-strain approximation is quantitatively accurate; this directly affects the 80% and 150% figures presented in Section V.","section":"Section IV.B and IV.E"},{"comment":"The gate operating points are explicitly selected by 'tun[ing] the gate voltages in order to move the dots into regions of large g-factor correction,' and the plotted Brm range (0.4-0.7 V) is then the range over which the effect is largest. The authors should show that this voltage range corresponds to the experimentally relevant operating range and report how the predicted sensitivity changes if a broader or unbiased range of gate voltages is used. Without this, the strength of the predicted tunability may be an artifact of the chosen operating window rather than a robust property of the device.","section":"Section III.C and Section V"},{"comment":"The paper claims that the mechanism 'may account for' the experimentally observed tunability, but it makes no quantitative comparison with the experiment in Ref. [9], which reports a nearly order-of-magnitude change in the singlet-triplet qubit frequency for a 12 mV change in a barrier gate. In contrast, the model shown in Fig. 10(c) gives roughly a factor-of-two change over a 0.3 V range of Brm. The authors should compare the predicted and measured sensitivity in common units (e.g., fractional change per mV) and discuss any discrepancy, or they should moderate the claim from 'quantitative explanation' to a proof-of-principle mechanism.","section":"Section V and Fig. 10"}],"minor_comments":[{"comment":"The text refers to 'Fig. 8' twice when describing the z-direction projection of the hole density and then again for the horizontal planar cross-section; the figure numbers appear to be mismatched, with Fig. 7 being the z-direction profile and Fig. 8 the planar cross-section.","section":"Section IV.E and Fig. 7/8"},{"comment":"The sentence 'expanding Eq. (3)' appears to refer to the effective g-factor definition, Eq. (14), not to the weak-form equation Eq. (3); please correct the cross-reference.","section":"Section V, after Eq. (16)"},{"comment":"There is a typo in 'tuned for for each value'; the word 'for' is repeated.","section":"Section III.C"},{"comment":"The text says the g-tensor correction is calculated 'using Eq. 9', but Fig. 6 shows all four correction components, which require Eqs. (9)-(12); please clarify.","section":"Section IV.D"},{"comment":"The Conclusion states that within a 0.3 V gate change and a π/2 rotation of the in-plane field the singlet-triplet g-factor changes by a factor of 2, whereas Section V reports a 145% change for Brm from 0.4 to 0.55 at fixed φ=3π/2; these statements should be reconciled or explicitly distinguished.","section":"Conclusion"}],"recommendation":"major_revision","confidential_remarks":"This is a device-theory paper with an interesting and plausible mechanism, and the forward-modeling structure is a strength. The main barrier to publication is the undefined κ in Eqs. (9)-(12) and the lack of validation of the pointwise uniform-strain approximation; both are fixable within the scope of a major revision. The authors should also be asked to check, before resubmission, that the experimental comparison with Ref. [9] is stated quantitatively and not overclaimed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague —\n\nThe thing to know: this paper gives a concrete, device-realistic mechanism for gate-tunable g-factors in germanium hole qubits — thermal-contraction strain, spatially varying g-tensor, dot displacement under gate voltage, wavefunction averaging — and it is plausible. But the two headline numbers (80% single-dot, 150% singlet-triplet) rest on an undefined parameter κ, so the quantitative claim is not yet reproducible.\n\nWhat is genuinely new: the full 3D FEM strain solution for this specific gate geometry, the Gaussian-averaged treatment of the hole over an inhomogeneous strain field, and the inclusion of in-plane field mixing that pulls out-of-plane g-corrections into the effective in-plane response. The forward pipeline is coherent: thermoelasticity with temperature-averaged CTEs, Luttinger-Kohn form for the g-corrections, MaSQE Schrödinger-Poisson for the wavefunctions. The strain magnitudes (~1e-4) match the x-ray data they cite, and they checked their boundary condition against Ref. [12]. No free parameters are fitted to the target experiment, which is the right direction for a mechanism paper.\n\nThe soft spots are real and concentrated in one place. Equations (9)–(12) are the bridge between FEM strain and g-factor change, and κ (and also k in Eqs. 10–11) is never defined. The second-order formulas are imported from prior work, but to claim \"over 80%\" and \"more than 150%\" you need the numerical value of κ. The reported curves are linear in these corrections, so a different κ scales the entire effect. The paper also applies uniform-strain formulas pointwise and then averages over a Gaussian hole in a strain field that varies by ~30% across the well and on tens of nanometers laterally — same scale as the wavefunction. No comparison against a direct solution of the inhomogeneous Luttinger-Kohn Hamiltonian is provided. That is a legitimate concern, not a nitpick. The operating points are chosen to sit in regions of large correction, which the authors admit; that does not sink the mechanism but it means the magnitude is an upper-bound-ish estimate. There is no overlay with the Rooney et al. data, so the \"may account for\" language is doing important work.\n\nThe paper is honest about its limitations, the math is coherent, and the citation pattern looks fine — the strain-g-tensor framework is properly credited to Abadillo-Uriel and Mauro et al.\n\nWho gets value: germanium hole qubit experimentalists and device simulators. It deserves a serious referee, but only with a request to define κ (and k), share code/data, and either validate the pointwise-then-average approximation or show how much it matters. If those are fixed, this could be a useful reference.\n\nI would send it to review, not desk reject, but I would not cite the numbers until the parameters are on the table.","headline":"A plausible strain-based mechanism for gate-tunable g-factors in Ge hole qubits, but the headline percentages hinge on an undefined parameter (κ) and an unvalidated pointwise-averaging step; worth serious review after those are fixed.","tokens_in":17484,"tokens_out":2495,"would_cite":false,"duration_ms":29843,"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":"Device-induced thermal-contraction strain creates a spatially varying g-tensor landscape that explains why germanium qubit g-factors swing by more than 80% under gate voltage changes.","keywords":["g-factor","germanium hole spin qubits","singlet-triplet qubit","strain-induced g-tensor correction","thermal contraction strain","quantum dot","finite-element simulation","gate voltage tunability"],"falsifier":"Measure the effective g-factor of a single germanium dot as a function of in-plane magnetic field azimuth at fixed gate voltages and compare the resulting map with Fig. 10; the predicted 80% range for a single dot and 150% swing for a singlet-triplet qubit are sharp signatures. A cleaner control is a device whose gate metal has nearly the same thermal expansion as germanium, where the model predicts gate-voltage g-factor tunability should largely disappear.","tokens_in":16512,"feed_emoji":"🧲","tokens_out":11963,"duration_ms":118622,"temperature":0.7,"pith_summary":"The paper sets out to explain a puzzling experimental fact: in germanium quantum dot qubits, small gate-voltage changes produce large shifts in the g-factor, sometimes near an order of magnitude. Its central claim is that cooling the device from room temperature to cryogenic operation leaves a permanent, spatially nonuniform strain field in the germanium layer, and this strain field reshapes the hole's g-tensor locally. A gate voltage does not change the strain; it shifts the quantum dot's wavefunction to a new position inside the strain landscape, and the effective g-factor is the strain correction averaged over that wavefunction. The paper reports that this mechanism makes the effective g-factor of a single dot vary by over 80% and that of a singlet-triplet qubit by more than 150% as gate voltage and in-plane magnetic field direction are tuned. If correct, this turns g-factor variability into a predictable, engineerable handle for qubit control.","feed_headline":"Thermal strain drives 80% g-factor swings in germanium qubits","feed_subtitle":"Finite-element strain maps plus hole wavefunctions link gate voltage and field angle to qubit frequency.","key_machinery":"The central object is the strain-induced g-tensor correction field, a spatially nonuniform tensor $\\delta g(\\mathbf{r})$ obtained from the local strain components through the correction formulas of Eqs. (9)-(12). The argument is carried by a chain: a finite-element solution of the linear thermoelasticity equation gives the displacement and hence the strain; the strain corrections are divided by the heavy-hole/light-hole gap $\\Delta_{\\mathrm{LH}}$; a Schrödinger-Poisson hole wavefunction, fit to a three-dimensional Gaussian, averages the correction over the dot; and the effective g-factor vector is formed as $\\delta \\vec{g}^{\\,*} = \\delta g \\cdot \\vec{B}/B$. The last identity is the pivot: even though the out-of-plane corrections $\\delta g_{zx}$ and $\\delta g_{zy}$ are individually below 1% of the unstrained $g_\\perp$, an in-plane magnetic field mixes them into the measured response, so the large swings are not simply echoes of the in-plane strain components.","core_discovery":"The paper's central discovery is that the observed gate tunability of the g-factor in germanium hole spin qubits can be accounted for without invoking any change in the material's intrinsic spin properties. Device-induced strain, arising from differential thermal contraction of the aluminum gates, the Al$_2$O$_3$ cap, and the Ge/GeSi heterostructure as the device is cooled from 300 K to 20 K, produces a spatially varying g-tensor correction whose components are proportional to local strain differences and shear strains divided by the heavy-hole/light-hole splitting $\\Delta_{\\mathrm{LH}}$. The hole wavefunction, computed from a Schrödinger-Poisson solver and approximated as a three-dimensional Gaussian, averages these corrections over the dot. Because gate voltages move the dot through the strain landscape, the averaged corrections change; the in-plane correction components $\\delta g_{xx}$ and $\\delta g_{xy}$ are of the same order as the unstrained in-plane g-factor $g_\\parallel = 0.15$, and through the relation $\\delta \\vec{g}^{\\,*} = \\delta g \\cdot \\vec{B}/B$, the out-of-plane corrections $\\delta g_{zx}$ and $\\delta g_{zy}$ mix into the in-plane response. The result is that single-dot and singlet-triplet effective g-factors swing by over 80% and more than 150%, respectively, across the explored gate-voltage and field-angle range; within a 0.3 V gate change and a $\\pi/2$ field rotation, the swings are about 60% and a factor of two.","pith_inferences":["If this mechanism is correct, a device whose gate metal has nearly the same thermal expansion coefficient as germanium should show strongly suppressed gate-voltage g-factor tunability; the paper does not state this prediction, but it follows directly from the thermal-contraction origin of the strain.","The same mechanism implies that thermal history could matter: repeated cooling cycles or different cooling rates might change the residual strain field and slightly shift qubit frequencies, so a device's g-factor map may not be perfectly reproducible run to run.","Wider quantum dots should average over more of the strain landscape and show smaller g-factor swings, so measuring g-factor tunability versus dot size is a testable way to distinguish wavefunction averaging over strain from other mechanisms.","Because the simulated strain pattern inherits the x-reflection symmetry of the gate layout, breaking that symmetry or choosing an asymmetric magnetic-field angle should produce left-right asymmetric tunability that could be used to address individual dots in a pair."],"forward_implications":["A 0.3 V change in the middle barrier gate combined with a $\\pi/2$ rotation of the in-plane magnetic field can change a single dot's effective g-factor by about 60% and a singlet-triplet qubit's by a factor of two relative to $g_\\parallel = 0.15$.","Because the correction is a wavefunction average over the strain landscape, the qubit frequency's sensitivity to gate noise is itself position-dependent, and moving the dot to a strain extremum or saddle point can suppress or amplify that sensitivity.","The model produces the non-monotonic dependence of the singlet-triplet g-factor on barrier gate voltage seen in the experiments, indicating that strain, rather than an intrinsic electrostatic effect, underlies that behavior.","Contour lines in the voltage-versus-field-angle plane give practical recipes for holding the g-factor constant while tuning the dot, or for deliberately sweeping the qubit frequency over a wide range.","Strain becomes an engineering variable: choosing gate materials, layer thicknesses, and operating temperature changes the thermal-contraction strain pattern and therefore the achievable g-factor tunability in a predictable way."],"supporting_citations":[{"why":"This experiment supplies the gate-voltage-modulated singlet-triplet qubit frequency data and the non-monotonic behavior that the model is built to explain.","marker":"[9]"},{"why":"This work supplies the strain-induced g-tensor correction framework, the unstrained g-factor values $g_\\perp = 13.5$ and $g_\\parallel = 0.15$, and the geometry used to validate the boundary condition.","marker":"[12]"},{"why":"This work gives the strain-modified heavy-hole/light-hole gap and the correction formulas of Eqs. (9)-(12) that convert local strain into g-tensor shifts.","marker":"[16]"},{"why":"This Schrödinger-Poisson solver computes the hole wavefunction for each gate configuration, and the Gaussian fits of those wavefunctions carry the averaging step.","marker":"[17]"},{"why":"This work provides the temperature-averaged thermal expansion coefficient approach and earlier gate-geometry strain results used to contextualize the simulated strain patterns.","marker":"[18]"},{"why":"This experiment maps strain tensors in a similar quantum dot device, and its measured strain magnitudes support the simulated strain values around $10^{-4}$.","marker":"[19]"}],"fun_headline_variants":["Strain, not spins, explains germanium qubit g-factor shifts","Cooling-induced strain tunes g-factor in germanium dots","Device strain accounts for gate-tunable g-factor in Ge qubits","Thermal strain shifts g-factor by 80% in germanium qubits","Gate voltage moves dot through strain to alter qubit frequency"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the strain-induced g-tensor correction formulas used here, including the unspecified coefficient $\\kappa$ in Eqs. (9)-(12), remain quantitatively accurate when the local strain is averaged over a Gaussian hole wavefunction in the sharply inhomogeneous strain field directly under the gates.","fun_headline_variants_meta":{"raw":{"variants":["Strain, not spins, explains germanium qubit g-factor shifts","Cooling-induced strain tunes g-factor in germanium dots","Device strain accounts for gate-tunable g-factor in Ge qubits","Thermal strain shifts g-factor by 80% in germanium qubits","Gate voltage moves dot through strain to alter qubit frequency"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000621,"raw_usage":{"total_tokens":2918,"prompt_tokens":1021,"completion_tokens":1897,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":637,"completion_tokens_details":{"reasoning_tokens":1806}},"tokens_in":637,"tokens_out":1897,"duration_ms":13772,"temperature":1.0,"reasoning_tokens":1806,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:49:15.045332+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the effective g-factor of a single germanium dot as a function of in-plane magnetic field azimuth at fixed gate voltages and compare the resulting map with Fig. 10; the predicted 80% range for a single dot and 150% swing for a singlet-triplet qubit are sharp signatures. A cleaner control is a device whose gate metal has nearly the same thermal expansion as germanium, where the model predicts gate-voltage g-factor tunability should largely disappear.","supporting_citations":[{"cited_title":"Rooney, Z","cited_arxiv_id":null,"evidence_quote":"This experiment supplies the gate-voltage-modulated singlet-triplet qubit frequency data and the non-monotonic behavior that the model is built to explain."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"This work supplies the strain-induced g-tensor correction framework, the unstrained g-factor values $g_\\perp = 13.5$ and $g_\\parallel = 0.15$, and the geometry used to validate the boundary condition."},{"cited_title":"Strain engineering in Ge/GeSi spin qubits heterostructures","cited_arxiv_id":"2407.19854","evidence_quote":"This work gives the strain-modified heavy-hole/light-hole gap and the correction formulas of Eqs. (9)-(12) that convert local strain into g-tensor shifts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"This Schrödinger-Poisson solver computes the hole wavefunction for each gate configuration, and the Gaussian fits of those wavefunctions carry the averaging step."},{"cited_title":"Corley-Wiciak, C","cited_arxiv_id":null,"evidence_quote":"This experiment maps strain tensors in a similar quantum dot device, and its measured strain magnitudes support the simulated strain values around $10^{-4}$."}],"review_version":1}