{"id":"9c06465e-a837-41e7-87fd-19c70c4aa976","arxiv_id":"2507.18770","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":12,"one_line_summary":"Ultrafast space- and time-resolved imaging reveals two propagating neutral collective modes in twisted WSe2, one consistent with the Goldstone mode of an intervalley-coherent (IVC) state.","lead":"Researchers imaged the motion of two new neutral spin-valley excitations in twisted WSe2 using ultrafast laser pulses, resolving their real-space propagation. The fast mode behaves like the long-sought Goldstone mode of an intervalley-coherent state, a hallmark of this correlated quantum phase.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Goldstone-mode assignment rests on a single ballistic velocity estimate; no direct measurement of the mode's dispersion or its softening near Tc is presented, leaving alternative low-energy spin-valley excitations viable.","rationale":"The reader's conditional verdict is well calibrated: the experimental campaign is careful, the controls are strong, and the observed modes are very likely collective spin-valley excitations. My stress-test sharpens the same load-bearing assumption rather than raising a new one. The fast mode's non-diffusive behavior is inferred from a spatial-profile anomaly, and its Goldstone nature is inferred from a velocity estimate that is consistent with theory only at the order-of-magnitude level. The B-field suppression and VHS proximity establish a correlated collective origin, but they do not distinguish an IVC Goldstone mode from other low-energy spin-valley collective modes. A direct dispersion measurement would be the decisive falsifiable test and would convert the central interpretation from a phenomenological fit into a verified prediction. Since the reader already requested exactly this kind of strengthening, the conditional verdict should stand unchanged.","tokens_in":18060,"tokens_out":7604,"duration_ms":95840,"concrete_test":"Perform a transient spin-valley grating measurement in the same VHS regime: create a sinusoidal Sz modulation with wavevector q along the transport direction using a spatially modulated pump, and record the time evolution of the grating amplitude for q from roughly 0.5 to 5 μm^-1 at n = -8.3e12 cm^-2, E = 0.1 V/nm, and Bz = 0.5 T. A Goldstone mode should show a propagating, oscillatory response with ω(q) = v q, v≈3 km/s, and no gap as q→0; extract v(q) = ω/q. Repeat near T≈8-10 K to check that v softens as the mode disappears. If the response is purely diffusive or v(q) is not approximately constant and gapless, the central Goldstone assignment is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the fast propagating object is the neutral Goldstone mode of an IVC state. The experimental evidence is indirect: a non-diffusive component in the first ~2 ns, a velocity estimate v≈3 km/s, emergence near the VHS, Bz-dependent suppression, and below-gap excitation. What is not measured is the defining property of a Goldstone mode: a gapless, linear ω(q) branch with v = ∂ω/∂q approximately constant at small q. The quoted velocity is obtained from real-space wave-packet motion over a few nanoseconds, and the hydrodynamic model (Methods, Eq. 7) is fitted to reproduce the two-mode structure, with non-determinable parameters set to order 1 (Extended Data Fig. 7). A gapped collective mode with large group velocity and weak damping, or a single-particle drift channel with energy-dependent mobility, could also produce a fast non-diffusive front at early times. The paper itself concedes that the non-diffusive component cannot be quantitatively isolated and that the specific IVC/IKS type cannot be pinpointed. The interpretation is plausible and internally consistent, but the evidence does not uniquely constrain the Goldstone identification.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports ultrafast optical imaging of spin-valley transport in twisted WSe2 bilayers (twist angles 3.5°, 3.8°, and 5.0°). Using circularly polarized and unpolarized pump pulses with a wide-field probe, the authors identify an ordinary single-particle spin-valley mode that exists throughout the phase diagram, and two additional 'exotic' modes that appear only near the van Hove singularity. The exotic modes have opposite signs of out-of-plane spin polarization, are excited by below-gap light, and propagate with markedly different speeds: a fast non-diffusive component estimated at about 3 km/s and a slow diffusive component (D ≈ 2 cm2/s). On the basis of the large velocity, low-energy excitation, VHS proximity, magnetic-field dependence, and a minimal hydrodynamic model, the authors interpret the fast mode as the Goldstone mode of an intervalley-coherent (IVC) state and the slow mode as a gapped amplitude (Higgs-like) mode. The paper claims this is the first imaging of propagation of a neutral Goldstone mode in a condensed matter system.","tokens_in":18333,"tokens_out":4709,"duration_ms":57177,"significance":"If the interpretation holds, this is a substantial advance: it would provide the first direct imaging of a neutral Goldstone mode, offer a new space-and-time-resolved probe of charge-neutral collective modes, and supply experimental evidence for IVC order in twisted WSe2. The experimental core is strong: the existence of the two exotic modes with opposite Sz is supported by reproducible imaging in three devices, with controls including unpolarized vs. circular pump, below-gap pumping, magnetic-field dependence, and polarization-selective detection. The paper also ships a concrete hydrodynamic model and honestly states several limitations. The weakest load-bearing point is the identification of the fast mode as an IVC Goldstone mode; that identification rests on a rough velocity estimate and on consistency with a model whose parameters are partly set by the data, rather than on a direct measurement of the mode's dispersion or of the IVC order parameter. The manuscript therefore merits publication only after the interpretive claim is either substantially strengthened or appropriately weakened.","major_comments":[{"comment":"The central identification of the fast mode as an IVC Goldstone mode rests on a velocity estimate of about 3 km/s taken from the first two nanoseconds of the spatial profile, but the Methods explicitly state that the non-diffusive component cannot be quantitatively isolated and that 'we cannot reliably determine the width or shape of the two modes.' The supporting hydrodynamic model then uses this velocity as an input to fix the spin compressibility (Extended Data Fig. 7: ℏχs = ρs/(ℏvG^2)), so the agreement between simulation and experiment is not an independent confirmation. The defining property of a Goldstone mode—a gapless, linear ω(q) with approximately constant v at small q—is not measured. The authors do not report a temperature-dependent velocity near the supposed transition, nor a momentum-resolved dispersion. I recommend either adding a test that can distinguish a Goldstone branch from a gapped but weakly damped mode (for example, temperature dependence of the velocity and damping across Tc, or a transient-grating measurement), or explicitly reframing the claim as a plausible hypothesis rather than a demonstrated identification.","section":"Origin of the exotic modes; Methods, 'Non-diffusive transport'"},{"comment":"The nonmonotonic magnetic-field dependence is cited as strong evidence for an IVC ground state, but the argument is one of consistency rather than uniqueness. A mode whose excitation amplitude is proportional to the field-induced Sz imbalance and whose underlying order parameter is suppressed at large Bz would show a similar nonmonotonicity for several valley-ordered or spin-ordered ground states, not only IVC. The paper correctly notes that it cannot distinguish IVC from IKS order, but the same reasoning applies to the more basic question of whether the order is IVC at all. A quantitative comparison with a field-dependent microscopic or Ginzburg-Landau theory—for instance, predicting how the mode velocity or damping should evolve with Bz—would make this evidence more discriminating.","section":"Magnetic field dependence (Fig. 5 and Methods)"},{"comment":"The below-gap pump experiment rules out interband single-particle excitations as the source of the exotic modes, but it does not rule out intraband single-particle channels or hot-carrier drift. The paper asserts that the large velocity and weak scattering 'require' a Goldstone or near-gapless mode, but this assertion is not proven; a gapped collective mode with weak damping and a large group velocity, or a single-particle drift channel with energy-dependent mobility, could also produce a fast non-diffusive front at early times. The authors should either provide a control that excludes such alternatives (for example, varying pump fluence to check the amplitude scaling expected for a condensate versus single-particle heating, or comparing the below-gap response at different doping levels within the VHS region) or state more cautiously that the single-particle origin is excluded only for interband processes.","section":"Methods, 'Below-gap pump measurements' and 'Origin of the exotic modes'"}],"minor_comments":[{"comment":"The caption lists 'Spin polarization damping rate ΓΔ ≈ 0.01 ns-1', but in the Methods the spin relaxation rate is denoted Γ_S and the amplitude-mode damping is ΓΔ. The symbol appears to be a typographical error and should be corrected to Γ_S.","section":"Extended Data Fig. 7 caption"},{"comment":"The diffusion-decay model uses the symbol Δp(x,t) but the main text discusses δSz; using a consistent notation for the measured spin-valley polarization would improve readability.","section":"Methods, Eq. (3)"},{"comment":"The abstract states the fast mode 'is consistent with' a Goldstone mode, but the Discussion later states the results 'provide strong evidence for Goldstone modes and IVC states.' Given that the direct Goldstone signature is not measured, the stronger wording should be softened to match the level of evidence actually presented.","section":"Abstract and Discussion"},{"comment":"In the Lagrangian (Eq. 7), the Berry-phase term is written ℏ(δSz)θ̇ while the continuity equation then uses ℏ∂tδSz. The equations are consistent with the canonical structure, but a brief comment on the dimensions of δSz and on the choice ℏ = 1 or not would help the reader avoid unit confusion.","section":"Methods, 'Analogy between IVC states and superfluid'"}],"recommendation":"major_revision","confidential_remarks":"This is a strong experimental paper that will attract wide interest. The existence of two neutral spin-valley modes with opposite Sz and very different velocities appears well supported by the imaging data and controls. The main barrier is the identification of the fast mode as an IVC Goldstone mode: the velocity estimate is rough, no dispersion or softening measurement is presented, and the hydrodynamic model is partly fitted. The authors could reasonably address this by either adding a discriminating experiment or, at minimum, reframing the central claim as a well-motivated but provisional interpretation. I would not reject the paper, but I would not accept it in its current form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Chenhao Jin and co-workers have pulled off an impressive experiment: space-and-time-resolved imaging of two neutral spin-valley collective modes in three twisted WSe2 devices. The observation that one mode propagates ballistically at ~3 km/s with one sign of Sz while a slower, diffusive mode carries the opposite Sz is new and, as far as I can tell, solid. The controls—unpolarized pump, below-gap pump, magnetic-field dependence, temperature dependence, and the distinction from the ordinary single-particle spin diffusion—are careful and mutually consistent. If the fast mode is what they say it is, this is the first direct imaging of a neutral Goldstone mode in a solid.\n\nWhat is not yet nailed down is the identification of the fast mode as the IVC Goldstone mode. The evidence is indirect: the mode appears only near the VHS, it is excited by below-gap photons, it is suppressed at Bz>2T, and its velocity is consistent with theory. But there is no direct measurement of a gapless linear dispersion, and the quoted velocity comes from the first two nanoseconds of a wave packet that cannot be quantitatively isolated from the coexisting slow mode. The hydrodynamic model in the Methods takes the measured velocities and diffusion constants as inputs, so the agreement is a fitting exercise, not a prediction. The paper itself concedes it cannot pinpoint the IVC/IKS type and cannot isolate the non-diffusive component. A gapped collective mode with a large group velocity, or a single-particle drift channel with unusual energy-dependent mobility, would also produce a fast non-diffusive front at early times.\n\nThat said, the soft spots are in proportion: the central observation stands, and the interpretation is the most economical one available. The magnetic-field dependence in particular does a lot of work: the exotics saturate at low field and are suppressed at high field, while equilibrium MCD saturates—that is a strong argument against a paramagnetic single-particle origin. The temperature dependence (modes disappear at ~10–20 K) also fits a collective-mode picture.\n\nWho is this for? Anyone working on moiré TMDs, IVC order, or neutral collective modes. It deserves a serious referee; the authors have done the heavy experimental lifting and the interpretation, while not conclusive, is well-argued. In review, I would ask for: (1) an attempt to measure the spatial profile with higher signal-to-noise to better separate the ballistic and diffusive components; (2) a more quantitative error budget on the 3 km/s estimate; (3) a simulation that starts from a microscopic model rather than fitted parameters, or at least a clear statement of which predictions would falsify the IVC picture.","headline":"First imaging of propagating neutral spin-valley modes in twisted WSe2; the IVC Goldstone assignment is plausible but not yet proven.","tokens_in":18940,"tokens_out":2014,"would_cite":true,"duration_ms":20369,"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":"The paper reports ultrafast imaging of two neutral spin-valley collective modes in twisted WSe2 and identifies the faster one, at about 3 km/s, as the Goldstone mode of an intervalley-coherent state.","keywords":["twisted WSe2","moire superlattices","intervalley coherence","Goldstone mode","spin-valley transport","ultrafast imaging","van Hove singularity","collective modes"],"falsifier":"Measure the fast mode's dispersion directly: an IVC Goldstone mode should show gapless linear behavior, $\\omega=v|k|$ with $v\\approx 3$ km/s, and its velocity should drop as the magnetic field approaches the transition to the valley-polarized state, whereas a gapped or single-particle mode would show a gapped or quadratic dispersion with no such softening; independently, a probe sensitive to the IVC spin texture (for example spin-polarized scanning tunneling microscopy or magnetic circular dichroism imaging) could confirm that the ordered ground state exists where the fast mode appears.","tokens_in":17836,"feed_emoji":"🌀","tokens_out":11899,"duration_ms":113219,"temperature":0.7,"pith_summary":"This paper reports space-and-time-resolved imaging of two charge-neutral spin-valley collective modes propagating through twisted WSe2 moiré devices. Near the van Hove singularity, a line-shaped pump launches two waves that ordinary electrical transport would miss: a fast mode moving at about 3 km/s that diffusion cannot explain, and a slow, diffusive mode carrying the opposite out-of-plane spin-valley polarization. The authors identify the fast mode as the Goldstone mode of an intervalley-coherent (IVC) state, the spontaneously broken valley-U(1) order predicted across moiré graphene and TMD systems, and the slow mode as its gapped amplitude (Higgs-like) counterpart. If this identification is correct, it is the first direct observation of the defining neutral excitation of IVC order and the first imaging of propagating superfluid-like collective modes in a condensed-matter system.","feed_headline":"Imaged: a neutral Goldstone mode moving at 3 km/s in twisted WSe2","feed_subtitle":"Two collective spin-valley waves appear only near the van Hove singularity, pointing to intervalley-coherent order.","key_machinery":"The central object is the intervalley-coherent order parameter $\\Delta(\\mathbf{r},t)=|\\Delta|e^{i\\theta}$, defined by $\\langle c_K^\\dagger c_{K'}\\rangle$; spontaneous fixing of the phase $\\theta$ breaks the approximate valley U(1) symmetry and yields a gapless Goldstone mode, while fluctuations of $|\\Delta|$ produce a gapped amplitude mode. The argument is carried by a minimum hydrodynamic Lagrangian that couples $\\theta$, the out-of-plane spin-valley polarization $S_z$, and the amplitude fluctuation $\\delta=|\\Delta|-\\Delta_0$, with a Berry-phase term $\\hbar\\,\\delta S_z\\,\\dot{\\theta}$ supplying the canonical structure and a $B_z$-linear mixing term that makes the two modes carry opposite $S_z$ signatures. Linearizing these equations gives one fast, nearly ballistic phase mode and one slow, damped amplitude mode, reproducing the two observed waves and their opposite spin-valley signs.","core_discovery":"Near the van Hove singularity of twisted WSe2, under a small out-of-plane magnetic field, unpolarized pump light produces two propagating neutral modes that are absent elsewhere in the doping–electric-field phase diagram: a fast mode with speed about 3 km/s and a slow mode with the opposite out-of-plane spin-valley polarization. The fast mode propagates non-diffusively, can be excited by below-gap light, and is suppressed when the magnetic field exceeds about 2 T, while the slow mode diffuses and vanishes near 20 K. The paper's central claim is that these are respectively the Goldstone (phase) mode and the gapped amplitude (Higgs-like) mode of an intervalley-coherent state, the spin-valley analogues of the two collective modes of a superfluid: $S_z$ plays the role of particle density, and a gradient of the IVC phase $\\theta$ carries a spin-valley supercurrent. Because a gapless Goldstone mode follows generically from spontaneous breaking of valley U(1) symmetry, the authors present the fast mode as the first detection of the hallmark neutral excitation of IVC order.","pith_inferences":["If the identification is right, the fast mode's dispersion should be gapless and linear, and its velocity should soften as $B_z$ approaches the transition into the valley-polarized state; measuring $\\omega(k)$ directly would test the Goldstone assignment.","The measured velocity, about three times the theoretical estimate the authors cite, suggests the superfluid stiffness or spin compressibility in these devices differs from current model values; a velocity-versus-twist-angle study could separate parameter error from a different ordering wavevector.","The featureless spin-valley response at $\\nu=-1$, despite the insulating state in charge transport, suggests the insulator coexists with IVC order or forms through degrees of freedom decoupled from $S_z$; combining this imaging with a local charge probe could test that coexistence."],"forward_implications":["The defining neutral excitation of intervalley-coherent order—the Goldstone mode—becomes experimentally accessible in solids, giving a direct probe of spontaneously broken valley U(1) symmetry.","The IVC region hugs the van Hove singularity and overlaps or adjoins the doping and field ranges where superconductivity was previously reported in twisted WSe2, supporting theories that connect IVC order to pairing.","Because the two modes are separated in space and time, the technique can resolve coexisting charge-neutral modes that ensemble or electrical measurements would average together or short-circuit.","The fast and slow modes reproduce the phase and amplitude collective modes of a superfluid in the spin-valley channel, with $S_z$ acting as the conserved density and a phase gradient acting as the supercurrent."],"supporting_citations":[{"why":"Supplies the theoretical picture of the IVC state as a spin-valley superfluid whose phase mode is a gapless Goldstone mode carrying spin-valley current.","marker":"Ref. 36"},{"why":"Predicts spin-valley locked instabilities, including intervalley coherence, near the van Hove singularity in moiré transition metal dichalcogenides.","marker":"Ref. 27"},{"why":"Recent theory placing IVC order and topological superconductivity in twisted WSe2, used to connect the observed IVC region to pairing.","marker":"Ref. 29"},{"why":"Predicts IVC-type ground states with a Goldstone-mode group velocity of order 1 km/s, the reference value the measured 3 km/s is compared against.","marker":"Ref. 31"},{"why":"Provides the easy-plane-magnet and superfluid transport analogy by which the fast mode is interpreted as a spin-valley supercurrent.","marker":"Ref. 42"},{"why":"Defines gapped amplitude (Higgs) collective modes, the identification assigned to the slow mode.","marker":"Ref. 19"},{"why":"Establishes the ultrafast imaging of pure spin-valley diffusion in TMD heterostructures that the ordinary-mode baseline and detection scheme rely on.","marker":"Ref. 39"},{"why":"Earlier direct imaging of intervalley-coherent order in twisted graphene through charge textures, contrasting with the neutral spin-valley signature reported here.","marker":"Ref. 35"},{"why":"Prior report of superconductivity in 5.0° twisted bilayer WSe2, situating the sample and linking the IVC region to the superconducting dome.","marker":"Ref. 17"}],"fun_headline_variants":["Imaged: spin-valley Goldstone mode at 3 km/s in WSe2","Two neutral collective waves appear in twisted WSe2","Fast and slow spin-valley modes seen near VHS","Goldstone and Higgs-like modes imaged in WSe2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the fast-moving signal really is the massless Goldstone wave of a valley-ordered ground state, because the ground-state order is never directly measured; a different valley-ordered phase or a single-particle origin for the fast signal would overturn the central claim.","fun_headline_variants_meta":{"raw":{"variants":["Imaged: spin-valley Goldstone mode at 3 km/s in WSe2","Two neutral collective waves appear in twisted WSe2","Fast and slow spin-valley modes seen near VHS","Goldstone and Higgs-like modes imaged in WSe2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000211,"raw_usage":{"total_tokens":1469,"prompt_tokens":1054,"completion_tokens":415,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":670,"completion_tokens_details":{"reasoning_tokens":341}},"tokens_in":670,"tokens_out":415,"duration_ms":4197,"temperature":1.0,"reasoning_tokens":341,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:08:32.584760+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the fast mode's dispersion directly: an IVC Goldstone mode should show gapless linear behavior, $\\omega=v|k|$ with $v\\approx 3$ km/s, and its velocity should drop as the magnetic field approaches the transition to the valley-polarized state, whereas a gapped or single-particle mode would show a gapped or quadratic dispersion with no such softening; independently, a probe sensitive to the IVC spin texture (for example spin-polarized scanning tunneling microscopy or magnetic circular dichroism imaging) could confirm that the ordered ground state exists where the fast mode appears.","supporting_citations":[],"review_version":2}