{"id":"2a3f53e8-a0c8-4fe5-b1a2-f60ce3539e59","arxiv_id":"2608.10802","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Electrically injected magnon spin transport in CrPS4 is anisotropic, with diffusion length at least 2.7 times longer and spin conductivity at least 2.2 times larger along the b-axis than the a-axis.","lead":"In the layered magnetic insulator CrPS4, magnon spin currents travel at least 2.7 times farther and with at least 2.2 times higher conductivity along the crystal's b-axis than along its a-axis. The work also warns that thermal techniques commonly used to measure magnon diffusion can badly overestimate the distance, which may require revisiting earlier claims.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Abstract's 'λ_b at least 2.7× longer' is not supported by the same-flake S1 data (2.22±0.89); the combined fit assumes thickness-independent λ despite a-axis samples disagreeing.","rationale":"The reader's weakest_assumption emphasizes the isotropic-interface assumption in Eq. 2. That is a legitimate concern about σ_m extraction, but the distance-dependent decay that yields λ_m is largely governed by the bulk diffusion length; even with anisotropic interface conductances, the functional form of the distance decay would still be set by the bulk λ. The more load-bearing concern is that the headline λ ratio of 'at least 2.7' is not statistically robust. The paper itself reports the same-flake ratio as 2.22±0.89 and the combined ratio as 2.72±0.96, and the two a-axis samples (S1 and S3) give inconsistent λ_a values that the authors cannot explain except by thickness effects. The S5 off-axis sample provides a direct contradiction to the angular model and is dismissed via thickness, showing that thickness is a strong confounding variable. Because the abstract states 'at least 2.7 times longer' as a central result, and because the most controlled data support only '~2.2' with a large uncertainty, the quantitative claim is overstated. This does not invalidate the qualitative finding of b-axis-favored transport, so the verdict should remain CONDITIONAL, as the reader already concluded. My concern differs from the reader's in identifying the statistical/thickness issue as the single most load-bearing problem rather than the interface anisotropy, hence 'partial' agreement.","tokens_in":38244,"tokens_out":12889,"duration_ms":127973,"concrete_test":"Recompute the λ_b/λ_a ratio and its confidence interval using only the same-flake S1 data from Table S2 (λ_a = 273±25.7 nm, λ_b = 606±236 nm) with standard error propagation. If the 95% lower bound falls below 2.7—which it will, given the reported uncertainties—the abstract's 'at least 2.7 times longer' is not supported. Additionally, refit the combined a-axis dataset excluding S3 (the thinner sample) to see whether λ_a moves to ~273 nm and the ratio drops to ~2.2, confirming that the 2.7 value is an artifact of thickness pooling.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim for the spin diffusion length, λ_b/λ_a ≥ 2.7, rests on the combined-sample fit in Supplemental V (Table S2/S3). That fit assumes the magnon diffusion length is independent of CrPS4 thickness in the 40–64 nm range and pools samples with different thicknesses. The data contradict this assumption. For the a-axis, the two samples give λ_a = 273±25.7 nm (S1, 64 nm) and λ_a = 165±33 nm (S3, 40 nm), which disagree at roughly the 2.5σ level. The most controlled comparison, on the same flake S1, yields λ_b/λ_a = 606/273 = 2.22±0.89, not 2.7. The combined fit's central ratio of 2.72 arises from λ_a=211±72 nm (a thickness-averaged value) and λ_b≥575 nm. Even taking λ_b at its lower bound, the 95% upper bound on λ_a (283 nm) gives a ratio of 2.03, so 'at least 2.7' is not a conservative lower bound. The off-axis sample S5 (14.7 nm) gives λ=216 nm, which is far below the angular-model prediction (~468 nm) and is attributed by the authors to thickness effects, further confirming that λ depends strongly on thickness. Since the abstract advertises a λ anisotropy factor of 'at least 2.7', this statistical and thickness-dependence issue directly undermines a headline number, even though a real b-axis enhancement remains plausible from the same-flake data.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports nonlocal magnon spin transport measurements on the van der Waals antiferromagnet CrPS4 at 25 K, with Pt injector/detector strips aligned to the crystallographic a and b axes. From distance-dependent first-harmonic nonlocal resistances the authors extract magnon spin diffusion lengths λ_a ≈ 211 nm and λ_b ≥ 575 nm, and magnon spin conductivities whose ratio σ_b/σ_a is at least 2.2; the same-flake S1 data give σ_b/σ_a = 2.22 ± 0.90 and λ_b/λ_a = 2.22 ± 0.89, while the combined-sample fit gives λ_b/λ_a = 2.72 ± 0.96. Thermally excited nonlocal second-harmonic resistances are larger along b by a factor of about 7 at 8 T on S1, but the authors argue that an extended temperature profile and anisotropic thermal conductivity (κ_bb < κ_aa, established by 3ω and nonlocal thermometry) prevent reliable extraction of λ_m or of the spin Seebeck anisotropy from the second-harmonic data. The first-harmonic anisotropy is interpreted as originating from the anisotropic in-plane exchange couplings J1, J2, J3 obtained from neutron scattering.","tokens_in":38561,"tokens_out":6736,"duration_ms":68856,"significance":"If the central anisotropy claim survives scrutiny, the paper establishes crystalline exchange anisotropy as a control parameter for magnon spin currents and introduces a magnonic analogue of anisotropic, planar-Hall-like transport. The manuscript has several genuine strengths: the b-axis values of λ_m and σ_m are explicitly treated as lower bounds; the ordinary Nernst effect is excluded by comparing local and nonlocal second-harmonic signs; the interpretation invokes independent neutron-scattering exchange constants rather than transport-derived parameters; and the thermometry data are internally consistent and support κ_bb < κ_aa. The authors also correctly warn against extracting λ_m from nonlocal SSE data. These elements make the qualitative conclusion—larger magnon spin transport along b than along a—plausible and worth publishing after revision. However, the headline quantitative claim 'spin diffusion length at least 2.7 times longer along b' is not supported as stated, and the interface-isotropy assumption underlying the σ_m extraction is not tested.","major_comments":[{"comment":"The abstract claims a spin diffusion length 'at least 2.7 times longer' along b, with λ_m^a ~ 211 nm and λ_m^b ≥ 575 nm. This ratio is not a conservative statement. The same-flake comparison on S1 gives λ_b/λ_a = 606/273 = 2.22 ± 0.89, not 2.7. The combined fit that yields 2.72 ± 0.96 assumes λ_m is independent of CrPS4 thickness in the 40–64 nm range, yet Table S2 shows λ_a = 273 ± 25.7 nm for S1 (64 nm) and λ_a = 165 ± 33 nm for S3 (40 nm), which disagree at roughly the 2.5σ level; the off-axis sample S5 (14.7 nm) gives λ = 216 ± 44 nm, far below the angular-model prediction (~468 nm), and the authors attribute this to thickness effects. Even using the reported λ_b lower bound (575 nm) together with the 95% upper bound of λ_a (211 + 72 = 283 nm), the ratio lower bound is about 2.03, so 'at least 2.7' is not supported. The abstract and conclusion should be revised to quote either the same-flake ratio with its uncertainty or a lower bound computed with the upper bound on λ_a.","section":"Abstract; §4 (Fig. 4); Supplemental V C, Table S2"},{"comment":"Equation (2) and the extraction of σ_m and λ_m assume a negligible and isotropic interfacial spin resistance at the Pt/CrPS4 interface, with the Pt detector acting as an ideal magnon sink. The manuscript itself states that dc sputtering of Pt removes the top CrPS4 layers and forms an interfacial PtS_x layer (Supplemental I, Ref. S1), but no test is provided for whether the interface transparency or spin-mixing conductance is isotropic along the a and b directions. If the interface is anisotropic, the reported σ_b/σ_a ≥ 2.2 would be partly an interface artifact. The authors should either provide an interface-controlled test or explicitly qualify the σ_m anisotropy as a lower bound that assumes an isotropic interface.","section":"Eq. (2); Supplemental V A; Supplemental I"},{"comment":"Table S1 lists t_CrPS4 = 64 nm for the S1 (k∥a) electrodes and 52.9 nm for the S1 (k∥b) electrodes, while the main text describes S1 as a single CrPS4 flake used to reduce device-to-device variations. If the two electrode sets are on the same flake, the thickness should be identical; if the flake has a step or the thickness differs between the two measured regions, the 'same-flake' comparison and the extraction of σ_m (through C = σ_m t_CrPS4 η_Pt²) need to be re-stated with this caveat. Please clarify whether the quoted σ_m values account for the local thickness at each electrode set, and whether the λ ratio on S1 is affected by the different thicknesses.","section":"Table S1; §4 ('same-flake' comparison)"}],"minor_comments":[{"comment":"Equation (4) appears to have a typographical error: the second term should read (λ_m^b)^2 sin²θ rather than λ_m^b sin²θ.","section":"§4, Eq. (4)"},{"comment":"The sentence 'the magnon spin diffusion length is does not vary significantly...' contains a grammatical error and should read 'does not vary significantly'.","section":"§4, 'Nonlocal first-harmonic response'"},{"comment":"The figures and captions display 'λ_b = 575 nm' where the text consistently states 'λ_b ≥ 575 nm'; the inequality should be shown in the figures for consistency.","section":"Fig. 4 and Fig. S6"},{"comment":"The discussion of Fig. S6 would be easier to follow if the fitted curves for fixed λ_b values were shown with confidence bands, so that the reader can see how similar the fit quality is for λ_b = 600 nm and λ_b = 1250 nm.","section":"Supplemental V D"}],"recommendation":"major_revision","confidential_remarks":"I see no grounds for rejection: the qualitative b-axis enhancement is supported by the same-flake data and the authors are appropriately cautious about λ_b being a lower bound. The revision should focus on replacing the 'at least 2.7' headline with a statistically defensible statement, clarifying the thickness discrepancy in Table S1, and adding an explicit caveat (or control) for the interface-isotropy assumption in Eq. (2)."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the take: this is a real anisotropic magnon transport effect in CrPS4, with the b-axis favored for both spin conductivity and diffusion length. The data support that. But the abstract's headline number 'at least 2.7 times longer' for λ_b is not conservative; the same-flake comparison gives 2.22±0.89, and the combined-sample fit that produces 2.7 assumes thickness-independent λ in a range where the data actually vary.\n\nWhat's new: previous work on CrPS4 anisotropy used thermally driven magnons (Ref 23), and electrical anisotropy was seen in non-vdW magnets. Here they inject and detect electrically in a vdW antiferromagnet and separately estimate σ_m and λ_m. The distance-dependent nonlocal first-harmonic data clearly show the b-axis signal larger at comparable separations. They are also honest about the b-axis λ being a lower bound, since a 1/d decay fits. The thermal conductivity anisotropy measurements (κ_bb < κ_aa) are careful, and the warning that nonlocal SSE cannot be used to extract λ_m without knowing the full temperature profile is useful for the whole magnon transport community.\n\nSoft spots: First, the extraction uses Eq. 2, which assumes a negligible and isotropic Pt/CrPS4 interface spin resistance. Sputter deposition is known to damage the top layers (their own Ref S1), and they provide no test of interface transparency anisotropy. That concern lands lightly because the effect is large, but it should be acknowledged as a caveat on σ_m, not λ_m.\n\nThe bigger issue is statistical. The same-flake S1 data give λ_b/λ_a = 2.22±0.89. The combined fit yields 2.72 by averaging λ_a across S1 (273 nm, 64 nm thickness) and S3 (165 nm, 40 nm), which disagree at about 2.5σ, and then assuming λ is thickness independent. The off-axis sample S5 (14.7 nm) gives λ=216 nm, far below the angular-model prediction, which the authors themselves attribute to thickness effects. So the thickness dependence is not negligible, and the combined lower bound 'at least 2.7' is not supported by the confidence intervals. A conservative phrasing would be 'λ_b/λ_a ≥ 2' based on the same-flake comparison, or the authors should propagate the uncertainty properly.\n\nThat said, the qualitative conclusion survives. The paper is for experimentalists in vdW magnonics and anyone who uses nonlocal SSE to estimate diffusion lengths. It deserves serious peer review; the authors should be asked to fix the abstract and statistics, and to state the interface assumption as a limitation.","headline":"Real anisotropic magnon transport effect in CrPS4, but the 'at least 2.7×' diffusion-length claim overstates the combined-sample fit; the same-flake data support a 2.2× effect.","tokens_in":39154,"tokens_out":3343,"would_cite":true,"duration_ms":33026,"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":"Electrically driven magnons in CrPS4 travel at least 2.7 times farther along the b-axis than the a-axis, so crystal orientation can steer magnon spin currents.","keywords":["magnon spin transport","spin diffusion length","magnon spin conductivity","CrPS4","van der Waals antiferromagnet","crystal anisotropy","spin Seebeck effect","thermal conductivity anisotropy"],"falsifier":"Measure the nonlocal first-harmonic decay on CrPS4 flakes thinner than 20 nm and on devices with a different interface metal; if $\\lambda_b/\\lambda_a$ or $\\sigma_b/\\sigma_a$ shrinks toward 1 while the crystal axes are unchanged, the anisotropy is not an intrinsic bulk property. A more direct check would be to measure the Pt/CrPS4 spin-mixing conductance along both axes and see whether it is anisotropic by a comparable factor.","tokens_in":37991,"feed_emoji":"🧲","tokens_out":11375,"duration_ms":106379,"temperature":0.7,"pith_summary":"In the layered antiferromagnet CrPS4, magnons—the quanta of magnetic spin waves—carry spin currents whose range and conductance depend on which crystallographic direction they travel. This paper claims that electrically injected magnons diffuse at least 2.7 times farther and with at least 2.2 times higher spin conductivity along the crystallographic b-axis than along the a-axis, with spin diffusion lengths $\\lambda_m^b \\geq 575$ nm and $\\lambda_m^a \\approx 211$ nm at 25 K. The same crystal anisotropy shows up in the thermally excited signal, which is about 7 times larger along b at 8 T, but the paper argues that thermal data alone cannot yield clean transport parameters because the temperature profile and heat conductivity of CrPS4 are themselves anisotropic. If the claim holds, the crystal lattice becomes a built-in compass for magnon spin currents, a practical control knob for directional magnonic devices.","feed_headline":"Magnons travel 2.7x farther along one axis of CrPS4","feed_subtitle":"Crystal orientation alone steers how far spin information flows, a control knob for directional magnonic devices.","key_machinery":"The load-bearing measurement is the lateral nonlocal geometry: a platinum strip injects a spin current via the spin Hall effect, magnons diffuse through CrPS4, and a second platinum strip detects the arriving spin current via the inverse spin Hall effect. The analysis runs through the diffusive transport relation $R_{\\mathrm{NL}} = \\sigma_m t_{\\mathrm{CPS}} \\eta_{\\mathrm{Pt}}^2 \\, \\mathrm{csch}(d/\\lambda_m)/\\lambda_m$, which converts the decay of the nonlocal first-harmonic resistance with injector–detector spacing $d$ into the magnon spin conductivity $\\sigma_m$ and the magnon spin diffusion length $\\lambda_m$. Its anisotropic extension, a conductivity tensor diagonal in the a/b crystal axes, predicts that transport along an oblique direction produces both longitudinal and transverse magnon spin currents. Local 3$\\omega$ and nonlocal thermometry supply the complementary piece, showing $\\kappa_{bb} < \\kappa_{aa}$, which is needed to separate thermal from magnonic anisotropy in the spin Seebeck channel.","core_discovery":"In the monoclinic van der Waals antiferromagnet CrPS4, where ferromagnetic layers stack antiferromagnetically, electrically injected magnons diffuse anisotropically once the magnetic field drives the spins into a collinear state above the spin-flop transition. From the distance dependence of the nonlocal first-harmonic resistance, the authors extract magnon spin conductivities and spin diffusion lengths along the two principal in-plane axes, finding $\\sigma_m^b/\\sigma_m^a \\geq 2.2$ and $\\lambda_m^b/\\lambda_m^a \\geq 2.7$, with $\\lambda_m^a \\approx 211$ nm and $\\lambda_m^b \\geq 575$ nm. Because the easy axis lies along the c-axis, perpendicular to the propagation plane, the anisotropy cannot come from the orientation of the easy axis; the authors attribute it to the strongly anisotropic in-plane exchange couplings. The paper also shows that the nonlocal second-harmonic resistance from thermally excited magnons is about 7 times larger along b at 8 T and 25 K, but argues that extracting a diffusion length from this signal would be unreliable because the response convolves the extended temperature profile with magnon transport. Thermometry measurements give $\\kappa_{bb} < \\kappa_{aa}$, so the thermal gradient that drives the spin Seebeck response is itself anisotropic. The authors conclude that electrical injection and detection provides the unambiguous transport channel, and that the anisotropic magnon spin conductivity tensor implies transverse magnon spin currents for off-axis gradients, a magnon analog of the planar Hall and transverse Seebeck effects.","pith_inferences":["The transverse magnon current predicted by the anisotropic conductivity tensor has not been measured here; a detector placed off-axis from the injector would directly test Eq. 3 and would separate a bulk tensor effect from any interface anisotropy.","The reported neutron-scattering exchange constants make a quantitative consistency check possible: if the Ref. [37] scaling $\\sigma_m \\propto J_S$ and $\\lambda_m \\propto \\sqrt{J_S}$ holds, the a/b ratios of $\\sigma_m$ and $\\lambda_m^2$ should track the relevant exchange-stiffness anisotropy, so future work could identify which CrPS4 bonds control relaxation.","The off-axis sample S5, only 14.7 nm thick, yields $\\lambda \\approx 216$ nm despite its oblique orientation, closer to the a-axis value than the b-axis value; the $\\lambda_m^b \\geq 575$ nm bound may therefore hold only for flakes in the 40–64 nm range, and thickness-dependent measurements would show whether the anisotropy ratio survives in the ultrathin limit."],"forward_implications":["A single CrPS4 flake can act as a directional magnon channel: with both spin conductivity and diffusion length larger along b, the orientation of the crystal axes relative to the injector–detector line sets how much spin current arrives.","Thermally driven nonlocal spin Seebeck measurements of CrPS4, including the previously reported 1.6 µm diffusion length, overestimate $\\lambda_m$ because the response is a convolution of magnon diffusion with an extended, anisotropic temperature profile rather than pure magnon transport.","Because $\\kappa_{bb} < \\kappa_{aa}$, the roughly 7 times larger second-harmonic signal along b at 8 T is not evidence for an equally large anisotropy in the spin Seebeck coefficients; separating the two requires the full thermal conductivity tensor.","An oblique magnon chemical potential gradient should generate a transverse magnon spin current from the lattice anisotropy alone, giving CrPS4 a magnon analog of the planar Hall and transverse Seebeck effects.","Gate-tunable magnetism in CrPS4 combined with this intrinsic directional anisotropy opens a route to electrically switching or steering magnon spin transport in one material."],"supporting_citations":[{"why":"Supplies the nonlocal magnon injection-detection geometry and the diffusive transport model on which Eq. 2 is based.","marker":"[3]"},{"why":"Provides the anisotropic exchange constants J1, J2, J3, Joop and K that the paper uses to attribute the transport anisotropy to exchange coupling.","marker":"[36]"},{"why":"Establishes the scaling $\\lambda_m \\propto \\sqrt{J_S}$ and $\\sigma_m \\propto J_S$, linking exchange stiffness to the two transport parameters extracted here.","marker":"[37]"},{"why":"Earlier report of anisotropic thermally driven magnon diffusion in CrPS4, with a thermal-conductivity anisotropy of opposite sign to the present thermometry.","marker":"[23]"},{"why":"Earlier extraction of a 1.6 µm magnon spin diffusion length from nonlocal second-harmonic data, which the paper argues is an overestimate caused by the extended temperature profile.","marker":"[24]"},{"why":"Demonstrates electrically generated magnon spin transport in CrPS4, the framework this work extends to the anisotropic case.","marker":"[9]"}],"fun_headline_variants":["Magnon spin flow in CrPS4 is 2.7x longer along b-axis","Crystal orientation tunes magnon spin transport in CrPS4","Anisotropic magnons: 2.2x conductivity, 2.7x diffusion along b","CrPS4 magnons show strong direction-dependent spin transport","Magnon spin diffusion prefers b-axis in CrPS4"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The numbers rest on the assumption that the Pt/CrPS4 contact lets spins through equally along the a and b axes and that flake thickness between 40 and 64 nm does not change the diffusion length; if either fails, part of the reported anisotropy could come from the interface or the sample rather than from the bulk crystal.","fun_headline_variants_meta":{"raw":{"variants":["Magnon spin flow in CrPS4 is 2.7x longer along b-axis","Crystal orientation tunes magnon spin transport in CrPS4","Anisotropic magnons: 2.2x conductivity, 2.7x diffusion along b","CrPS4 magnons show strong direction-dependent spin transport","Magnon spin diffusion prefers b-axis in CrPS4"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000661,"raw_usage":{"total_tokens":3130,"prompt_tokens":1165,"completion_tokens":1965,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":781,"completion_tokens_details":{"reasoning_tokens":1866}},"tokens_in":781,"tokens_out":1965,"duration_ms":14505,"temperature":1.0,"reasoning_tokens":1866,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:06:21.861901+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the nonlocal first-harmonic decay on CrPS4 flakes thinner than 20 nm and on devices with a different interface metal; if $\\lambda_b/\\lambda_a$ or $\\sigma_b/\\sigma_a$ shrinks toward 1 while the crystal axes are unchanged, the anisotropy is not an intrinsic bulk property. A more direct check would be to measure the Pt/CrPS4 spin-mixing conductance along both axes and see whether it is anisotropic by a comparable factor.","supporting_citations":[{"cited_title":"Physica Status Solidi (RRL)--Rapid Research Letters , volume=","cited_arxiv_id":null,"evidence_quote":"Supplies the nonlocal magnon injection-detection geometry and the diffusive transport model on which Eq. 2 is based."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the anisotropic exchange constants J1, J2, J3, Joop and K that the paper uses to attribute the transport anisotropy to exchange coupling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the scaling $\\lambda_m \\propto \\sqrt{J_S}$ and $\\sigma_m \\propto J_S$, linking exchange stiffness to the two transport parameters extracted here."},{"cited_title":"Nano Letters , volume=","cited_arxiv_id":null,"evidence_quote":"Earlier report of anisotropic thermally driven magnon diffusion in CrPS4, with a thermal-conductivity anisotropy of opposite sign to the present thermometry."},{"cited_title":"Nano Letters , volume=","cited_arxiv_id":null,"evidence_quote":"Earlier extraction of a 1.6 µm magnon spin diffusion length from nonlocal second-harmonic data, which the paper argues is an overestimate caused by the extended temperature profile."},{"cited_title":"Advanced Functional Materials , volume=","cited_arxiv_id":null,"evidence_quote":"Demonstrates electrically generated magnon spin transport in CrPS4, the framework this work extends to the anisotropic case."}],"review_version":1}