{"id":"b8c676d0-a7b6-4f69-b16e-1024ea80c25a","arxiv_id":"2507.14838","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"THz emission from topological material and ferromagnet bilayers depends strongly on surface preparation, so thickness dependence alone cannot identify the spin-to-charge conversion mechanism, and NbP works as a broadband emitter.","lead":"Scientists made thin sandwiches of topological materials and magnetic metals that emit terahertz pulses when hit by laser light. They found that how the layers are prepared, especially keeping the interface free of oxidation, changes the signal more than layer thickness does, and that a Weyl semimetal can also emit these pulses efficiently.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The SCC thickness trends in Figs. 2d/3b are not normalized by pump power absorbed in the CoFeB source; d- and oxidation-dependent optical absorption could mimic the reported interface-sensitive thickness dependence.","rationale":"The reader's weakest assumption concerns the symmetry decomposition in Eqs. 1-2. I agree that the decomposition is unverified, but I find a more direct, testable gap: the excitation-side normalization. The paper's rejection of thickness as a mechanism fingerprint depends on comparing SCC amplitudes across d and across interface preparation; without correcting for the pump power actually absorbed in the CoFeB source, those amplitudes are not established as SCC efficiencies. This is not an internal contradiction or a charge of misconduct; it is an omitted control. Because the authors already perform transmission/impedance corrections, adding an absorption normalization and a bare-CoFeB or spacer control is a tractable revision. The central claim may well survive, but it is not yet demonstrated to the level claimed. Therefore the reader's CONDITIONAL verdict stands.","tokens_in":11327,"tokens_out":16538,"duration_ms":182567,"concrete_test":"Measure the 0.8 eV pump reflectance R(d) and transmittance T(d) for the complete CoFeB(4 nm)|Bi2Se3(d QL)|sapphire stack at d = 6, 16, 32, 64 QL, and for in-situ versus ex-situ 16 QL samples. Use a transfer-matrix model with literature optical constants to determine the absorbed power in the CoFeB layer, then re-plot the extracted SCC signal and sheet current in Figs. 2d and 3b normalized by this absorbed power. If the nonmonotonic peak at 16 QL and the ex-situ suppression survive normalization, the trends reflect SCC physics; if they flatten or shift, the central thickness/interface comparison is an optical artifact.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the thickness dependence of spin-to-charge conversion changes with interface quality, so thickness is not a reliable ISHE/IREE fingerprint. The evidence is the SCC component extracted from the B-odd THz signal as a function of Bi2Se3 thickness and growth condition (Figs. 2d, 3b). Even granting the symmetry decomposition of Eq. (1), the measured THz amplitude is proportional to the pump power absorbed in the 4-nm CoFeB spin source times spin-injection efficiency, SCC efficiency, and outcoupling. The paper quotes a fixed 7 mW pump power but reports no measurement or model of the absorbed pump power in CoFeB for each d, nor for oxidized versus protected interfaces. Optical interference in the CoFeB|Bi2Se3|sapphire stack changes with d over 6-64 QL, and surface oxidation before CoFeB deposition changes the interface reflectance; both affect the 0.8 eV pump absorption in CoFeB. The authors normalize the emitted THz field by the THz impedance to extract a sheet current, but that corrects the detection side, not the excitation side. Hence the nonmonotonic peak at 16 QL in the in-situ series and the strong suppression of ex-situ samples could be partially or wholly artifacts of d-dependent or oxide-dependent pump absorption rather than genuine changes in SCC efficiency. This is load-bearing because the central claim compares exactly these thickness and interface trends.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports time-domain THz emission spectroscopy of ferromagnet/topological-material heterostructures, including Bi2Se3, Pb-doped Bi2Se3, (Bi1-xSbx)2Te3, and the Weyl semimetal NbP. The authors decompose the measured signal into a magnetic-field-odd term attributed to spin-to-charge conversion (SCC) and a field-even term attributed to a shift current, using azimuthal fits described by Eqs. (1) and (2). They compare SCC amplitudes as a function of Bi2Se3 thickness for in-situ and ex-situ prepared samples, study composition dependence of (Bi,Sb)2Te3, and demonstrate broadband emission up to about 8 THz from both TI and WSM bilayers. The central claim is that the thickness dependence of SCC changes with interface quality, so thickness trends alone are not a reliable way to distinguish inverse spin Hall effect from inverse Rashba-Edelstein effect, and that in-situ growth and proper composition are essential for efficient SCC.","tokens_in":11578,"tokens_out":6417,"duration_ms":83473,"significance":"If the conclusions hold, the paper delivers a useful caution against reading TI thickness trends as a mechanism fingerprint and extends spintronic THz emission to Weyl semimetal/ferromagnet bilayers. The authors should be credited for a clear symmetry decomposition, for correcting the detected THz signal by the stack impedance when extracting sheet currents, for including two nominally identical samples at several thicknesses in the in-situ series (Fig. 2d), and for correcting the broadband spectra by the GaP detector response (Fig. 5b). The main limitation is that the reported SCC amplitudes are not normalized by the pump power actually absorbed in the CoFeB spin source, and several key series consist of single samples without error bars; these issues currently weaken the quantitative support for the central thickness-dependent and interface-dependent claims.","major_comments":[{"comment":"The central thickness comparison is based on emitted THz amplitudes measured at a fixed incident pump power of 7 mW at 0.8 eV, without normalization to the pump power absorbed in the 4-nm CoFeB spin source. Varying the Bi2Se3 thickness from 6 to 64 QL changes the optical interference in the CoFeB/Bi2Se3/sapphire stack, and ex-situ surface oxidation changes the interface reflectance; both effects can alter the absorbed fraction in CoFeB and mimic the nonmonotonic peak near 16 QL or the strong suppression in ex-situ samples. The impedance correction described in the text corrects the detection/outcoupling side only, not the excitation side. The authors should measure or model the absorbed pump fraction as a function of thickness and surface condition, or normalize the extracted SCC amplitudes by that fraction.","section":"§Results (Figs. 2d, 3b)"},{"comment":"In the ex-situ Bi2Se3 thickness series, only the d = 16 QL point is averaged over two samples; the points at d = 6, 32, and 64 QL each come from a single sample and carry no error bars. Given the large batch-to-batch variation shown in Fig. 3a, the conclusion that interface quality changes the thickness dependence is under-supported. Replication of at least the key thicknesses is needed before this central claim can be considered robust.","section":"§Results (Fig. 3b)"},{"comment":"The composition dependence of (Bi1-xSbx)2Te3 and the Pb-doped Bi2Se3 comparison are based on one sample per composition without error bars and without normalization by absorbed pump power. Since the optical constants and therefore the pump absorption in the CoFeB layer vary with Sb fraction and Pb doping, the minimum near x = 0.9 and the suppression in Pb-Bi2Se3 could partly reflect absorption changes rather than genuine SCC changes. The statement that the same trend is observed at 0.2 eV is not supported by shown data, so it cannot serve as a control in its current form.","section":"§Results (Figs. 4b, 4c)"},{"comment":"The claim that NbP|Py emits about one-fifth as efficiently as in-situ Bi2Se3|CoFeB is based on a single NbP sample and raw waveform amplitudes, with no error bars and no correction for pump absorption differences between the two stacks. The comparison also uses different ferromagnets (Py vs CoFeB) and different film thicknesses. At present this result should be presented as a proof-of-principle demonstration; the quantitative efficiency ratio is not established without additional samples and a common normalization.","section":"§Results (Fig. 5)"}],"minor_comments":[{"comment":"In the Summary, 'spin to charger conversion' should read 'spin to charge conversion'.","section":"Summary"},{"comment":"The zero-field fit introduces an offset phi0' for the remanent magnetization direction; please state whether the magnetization was independently characterized or whether phi0' is a free parameter, and show fit residuals or parameter values to demonstrate that the two-offset model is not overfitting.","section":"Eq. (2) and Fig. 2c"},{"comment":"The main text does not explicitly state how many independent growths were used for each thickness in the in-situ series. Please state this in the text rather than only in the figure caption, and clarify whether the two-sample average in Fig. 2d applies to every thickness.","section":"Figs. 2d and 3b"},{"comment":"When contrasting the present thickness trend with the monotonic decrease reported in the previous Bi2Se3|Co study, please note explicitly that the previous work used Co rather than CoFeB and different growth conditions, since these differences are part of the interface-quality argument.","section":"Comparison with Ref. [9]"},{"comment":"The data availability statement says that all data are in the paper and additional data may be requested. For reproducibility, providing the raw time-domain waveforms and azimuthal scans in a repository would be helpful.","section":"Data availability"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope and the qualitative story is plausible, but the normalization and replication issues directly affect the central thickness-dependence and interface-quality claims. I recommend major revision rather than rejection because the problems are addressable with additional measurements or clear normalization arguments. No circularity or novelty concerns were identified."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe thing to know: this paper has three genuinely new experimental results—the in-situ/ex-situ interface comparison, the (Bi1-xSbx)2Te3 composition series, and the first THz emission from NbP|Py—and the central message about thickness-dependent SCC being an unreliable ISHE/IREE fingerprint is plausible. But the abstract oversells NbP as 'comparable' when the body says one-fifth, and the thickness-dependence argument misses a pump-absorption normalization.\n\nThe real strengths: the azimuthal decomposition (Eqs. 1/2) cleanly separates SCC from shift current; the ex-situ versus in-situ comparison is a practical warning for the community; and the THz-impedance correction means they extract sheet currents rather than raw field amplitudes. The composition series with a minimum near x=0.9 is a nice observation, and repeating it at 0.2 eV adds confidence.\n\nThe soft spots: the pump-absorption issue is load-bearing. They quote a fixed 7 mW pump but never measure or model the power absorbed in the 4-nm CoFeB source. The Bi2Se3 thickness changes the optical interference, and oxidation before CoFeB deposition changes interface reflectance; both affect the 0.8 eV pump field at CoFeB. Correcting THz impedance fixes the out-coupling side, not the excitation side. So the nonmonotonic SCC peak at 16 QL, and the strong suppression of the ex-situ samples, could be partly artifacts. The conclusion that SCC is not surface-dominated rests on that peak; a referee should ask for pump reflection/transmission measurements or a transfer-matrix calculation. The abstract overstatement about NbP is minor in comparison but should be fixed. Several figures use single samples without error bars (ex-situ series, Fig. 4b, Fig. 5), and the oxidation is inferred, not chemically verified. Raw data are not deposited.\n\nThese are addressable, not fatal. The paper is a net positive for the subfield: it gets people thinking about interface preparation and adds a Weyl material to the emitter list. I'd send it for peer review, but the load-bearing thickness dependence should be made robust to pump-absorption variations before the fingerprints are discarded.\n\nFor you: if you work on THz spintronics, this is worth a look; I'd cite the NbP result and the composition series once they're firmed up.","headline":"Useful new THz-emission data on topological-material|ferromagnet bilayers, but the thickness-dependence argument needs a pump-absorption check and the abstract overstates NbP efficiency.","tokens_in":12249,"tokens_out":3849,"would_cite":true,"duration_ms":40726,"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 argues that in topological-material/ferromagnet bilayers, the thickness trend of spin-to-charge conversion depends on interface quality, so thickness measurements alone cannot distinguish inverse spin Hall from inverse…","keywords":["terahertz emission spectroscopy","spin-to-charge conversion","topological insulator","Weyl semimetal","inverse spin Hall effect","inverse Rashba–Edelstein effect","Bi2Se3","NbP"],"falsifier":"Apply the static field along the x direction instead of y on an in-situ Bi2Se3(16 QL)|CoFeB sample while detecting the x-polarized THz field: SCC predicts the emission direction rotates to y and the x component vanishes, leaving only the threefold shift current. A detectable field-antisymmetric x component in that geometry would reveal a magnetic-field-dependent channel beyond SCC, invalidating the decomposition behind the thickness-comparison conclusion.","tokens_in":11067,"feed_emoji":"🧲","tokens_out":8772,"duration_ms":101077,"temperature":0.7,"pith_summary":"This paper studies ultrafast spin-to-charge conversion (SCC) in topological-material/ferromagnet bilayers by time-domain terahertz emission spectroscopy. It separates the emitted signal into a field-antisymmetric SCC part and a field-symmetric shift-current part, and tracks both as the topological layer thickness changes. The central finding is that the SCC-versus-thickness trend is not fixed: in-situ grown Bi2Se3 layers show a nonmonotonic trend peaking near 16 QL, while ex-situ (air-exposed) samples show suppressed and inconsistent signals. The paper concludes that thickness dependence alone is not a reliable fingerprint for distinguishing inverse spin Hall effect from inverse Rashba–Edelstein effect, and that in-situ growth and composition control are what determine efficient SCC. It also reports that NbP|permalloy (a Weyl semimetal bilayer) emits THz at about one-fifth the efficiency of Bi2Se3|CoFeB, with spectra extending to about 8 THz.","feed_headline":"Interface quality, not thickness, sets THz spin conversion trend","feed_subtitle":"Clean in-situ Bi2Se3|CoFeB bilayers show a nonmonotonic conversion peak; air-exposed samples are suppressed and inconsistent.","key_machinery":"The load-bearing object is the azimuthal-angle decomposition of the emitted THz peak signal into a field-antisymmetric SCC term and a field-symmetric threefold shift-current term. Under a static field $\\vec{B}$ along $\\hat{y}$, the spin current $\\vec{j}_s$ along $\\hat{z}$ with polarization $\\vec{\\sigma}$ along $\\hat{y}$ produces SCC emission fixed along $\\hat{x}$ with no azimuthal dependence, while the Bi2Se3 surface shift current follows $\\sin[3(\\phi-\\phi_0)]$. Fitting the measured peak signal versus sample azimuth to Eqs. 1 and 2 separates the two channels, and THz transmission and impedance measurements convert emitted fields into transverse sheet currents. This decomposition is what allows the paper to track SCC and shift current separately as functions of TI thickness, composition, and growth condition.","core_discovery":"The core claim, stated on the paper's own terms, is that the magnetic-field-antisymmetric THz signal in Bi2Se3|CoFeB bilayers is a direct measure of SCC, while the field-symmetric part is a threefold-symmetric shift current from the Bi2Se3 surface. Using an azimuthal-angle decomposition $S_{\\pm B}(\\phi)=\\pm S_{\\mathrm{SCC}}+S_{\\mathrm{shift}}\\sin[3(\\phi-\\phi_0)]$ and the corresponding zero-field form, the authors find that the SCC component varies nonmonotonically with Bi2Se3 thickness, peaking near 16 QL, whereas the shift current is nearly thickness independent. In air-exposed ex-situ samples, the SCC signal is strongly reduced and the thickness trend becomes inconsistent. From this they conclude that the thickness dependence of SCC is controlled by interface quality, so thickness trends by themselves cannot distinguish bulk inverse spin Hall from interfacial inverse Rashba–Edelstein conversion; both mechanisms likely contribute. They additionally show that (Bi1−xSbx)2Te3|CoFeB emission is weakest near x≈0.9, where the bare TI is near charge neutrality, and that Pb-doped (bulk-insulating) Bi2Se3 emits much less than bulk-conducting Bi2Se3.","pith_inferences":["Beyond the paper, this result implies that previously published thickness trends in TI-based THz emitters should be re-read with attention to sample history, since air exposure alone changes the extracted SCC trend.","A testable extension would be a controlled oxidation series: identical Bi2Se3 thickness with different air-exposure times before capping, compared against surface-sensitive probes, to quantify how interface oxidation suppresses spin injection.","The NbP|Py result suggests that other Weyl semimetals with stronger Berry-curvature effects might outperform this emitter, but the paper does not claim that."],"forward_implications":["THz emission efficiency of Bi2Se3|CoFeB is not governed by TI thickness alone; in-situ growth that avoids surface oxidation is required for high spin-to-charge conversion.","In (Bi1−xSbx)2Te3|CoFeB, SCC efficiency varies strongly with Sb content and is minimized near x≈0.9, where the bare TI sits near charge neutrality; Pb-doped bulk-insulating Bi2Se3 likewise shows suppressed emission.","The Weyl semimetal NbP, in a NbP|Py bilayer, can act as a spin-to-charge conversion layer with THz emission efficiency about one-fifth that of Bi2Se3|CoFeB.","Both TI|FM and WSM|FM heterostructures emit THz pulses with spectral weight from roughly 0.2 to 8.3 THz, comparable to established spin-Hall emitters.","Thickness-dependence studies alone cannot be used to claim a particular SCC mechanism; interface preparation and composition must be controlled and reported."],"supporting_citations":[{"why":"Earlier Bi2Se3|Co THz study reporting a monotonic decrease in signal beyond 6 QL; it is the central comparison the paper overturns.","marker":"[9]"},{"why":"Establishes THz emission spectroscopy as a contact-free probe of spin-to-charge conversion and the emission direction rule $\\propto \\vec{j}_s \\times \\vec{\\sigma}$ used in the decomposition.","marker":"[7]"},{"why":"Identifies the threefold-symmetric shift current from the bare Bi2Se3 surface, which the paper separates from the SCC component.","marker":"[42]"},{"why":"Provides the azimuthal behavior of the surface shift current that supports Eq. 1 of the decomposition.","marker":"[43]"},{"why":"Growth of single-crystalline NbP thin films on MgO, the Weyl semimetal layer used in the WSM|FM bilayer.","marker":"[39]"},{"why":"Theoretical work showing that metallic overlayers induce charge transfer that shifts the TI Fermi level, motivating the interface-quality and composition conclusions.","marker":"[24–29]"},{"why":"Shows that Sb substitution tunes the chemical potential toward the Dirac point at x≈0.9 in (Bi,Sb)2Te3, anchoring the composition-dependence argument.","marker":"[49–53]"},{"why":"Characterizes the GaP electro-optic detection response used to convert time-domain traces into broadband spectra extending to about 8 THz.","marker":"[63,64]"}],"fun_headline_variants":["Interface quality, not thickness, sets THz spin conversion","Topological stacks emit broadband THz via spin-to-charge","Clean interfaces key for efficient THz spin conversion","In-situ growth crucial for topological THz emitters","Weyl semimetal matches topological insulator in THz output"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the field-antisymmetric THz signal is entirely spin-to-charge conversion and the field-symmetric signal is entirely surface shift current, with no other magnetic-field-dependent channel contributing; the paper does not independently measure either component in the bilayers.","fun_headline_variants_meta":{"raw":{"variants":["Interface quality, not thickness, sets THz spin conversion","Topological stacks emit broadband THz via spin-to-charge","Clean interfaces key for efficient THz spin conversion","In-situ growth crucial for topological THz emitters","Weyl semimetal matches topological insulator in THz output"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000243,"raw_usage":{"total_tokens":1601,"prompt_tokens":1088,"completion_tokens":513,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":704,"completion_tokens_details":{"reasoning_tokens":432}},"tokens_in":704,"tokens_out":513,"duration_ms":6542,"temperature":1.0,"reasoning_tokens":432,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:45:59.468900+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Apply the static field along the x direction instead of y on an in-situ Bi2Se3(16 QL)|CoFeB sample while detecting the x-polarized THz field: SCC predicts the emission direction rotates to y and the x component vanishes, leaving only the threefold shift current. A detectable field-antisymmetric x component in that geometry would reveal a magnetic-field-dependent channel beyond SCC, invalidating the decomposition behind the thickness-comparison conclusion.","supporting_citations":[{"cited_title":"Wang , author L","cited_arxiv_id":null,"evidence_quote":"Earlier Bi2Se3|Co THz study reporting a monotonic decrease in signal beyond 6 QL; it is the central comparison the paper overturns."},{"cited_title":"Seifert , author S","cited_arxiv_id":null,"evidence_quote":"Establishes THz emission spectroscopy as a contact-free probe of spin-to-charge conversion and the emission direction rule $\\propto \\vec{j}_s \\times \\vec{\\sigma}$ used in the decomposition."},{"cited_title":"Braun , author G","cited_arxiv_id":null,"evidence_quote":"Identifies the threefold-symmetric shift current from the bare Bi2Se3 surface, which the paper separates from the SCC component."},{"cited_title":"Stensberg , author X","cited_arxiv_id":null,"evidence_quote":"Provides the azimuthal behavior of the surface shift current that supports Eq. 1 of the decomposition."},{"cited_title":"Bedoya-Pinto , author A","cited_arxiv_id":null,"evidence_quote":"Growth of single-crystalline NbP thin films on MgO, the Weyl semimetal layer used in the WSM|FM bilayer."}],"review_version":1}