{"id":"625d08bb-99cb-41a4-8f74-3e41c737bb2a","arxiv_id":"1908.06528","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Adding sub-monolayer titanium layers to platinum raises the spin-orbit torque efficiency to about 0.35 with only 90 μΩ cm resistivity, limited by a tradeoff between spin Hall conductivity and carrier lifetime.","lead":"This paper shows that inserting thin titanium layers into platinum films can double the spin-orbit torque efficiency while keeping the material's electrical resistivity low. The result points to a practical platinum-based material for energy-efficient magnetic memory and logic devices.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Claimed θSH≈0.8 depends on a model-calibrated interface transparency Tint; plausible interface-parameter variations lower it to about 0.5-0.6.","rationale":"The paper's central contribution has two parts: a robust materials result, namely ξ_DL^j≈0.35 at 90 μΩ cm in [Pt0.75/Ti0.2]7/Pt0.75, and a model-dependent interpretation, namely the internal spin Hall ratio θSH≈0.8 and the claim that carrier-lifetime shortening caps θSH. The harmonic-orbit torque measurement and the resistance characterization are described carefully and have small stated measurement uncertainties, so I do not object to the directly measured efficiency. However, converting ξ to θSH requires the interface transparency model of Eq. (1) and the spin-memory-loss calibration, with λs derived from an assumed Elliott-Yafet mechanism and G↑↓ taken from theory rather than measured on these stacks. No uncertainty or independent cross-check is provided. Numerically, the claimed 0.8 is consistent with Tint≈0.43; plausible upward variations in G↑↓ reduce Tint enough to put θSH in the 0.5-0.6 range or lower. The paper itself concedes that the exact theoretical limit remains unsettled. This does not refute the materials result, but it means the headline quantitative claim is conditional on an unvalidated interface model. A sensitivity analysis or a thickness-dependent fit with free interface parameters would settle the matter, so I keep the reader's conditional verdict rather than moving to acceptance or rejection.","tokens_in":10094,"tokens_out":10215,"duration_ms":105921,"concrete_test":"Recompute the m=7 point in the bulk limit of Eq. (1) as θSH = ξ_DL^j × [1 + (ρxx λs)^{-1}/(2G↑↓)] × (1−0.23Ks)^{-1}, sweeping G↑↓ over 0.59−1.77×10^15 Ω^-1 m^-2, Ks over 0.36−0.65 erg/cm^2, and λs over 0.8−1.2 nm. If the resulting θSH spans below 0.6, the paper must report θSH as a range and soften the 0.8 claim. Ideally, also measure ξ_DL^E on a thickness series at fixed m and fit Tint and λs freely, without imposing Elliott-Yafet.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing quantitative claim is the internal spin Hall ratio θSH≈0.8 for [Pt0.75/Ti0.2]7/Pt0.75 (Fig. 3(b)). This value is not measured directly: the measured quantity is the torque efficiency ξ_DL^j≈0.35, and θSH is obtained by dividing by Tint = Tint^SBF × Tint^SML. Tint is fixed through Eq. (1) and the relation Tint^SML≈1−0.23Ks from Ref. [30]. In the bulk limit used for m≥3, Eq. (1) gives Tint^SBF = [1 + (ρxx λs)^{-1}/(2G↑↓)]^{-1}. With ρxx=90 μΩ cm, λs of order 1 nm, and G_Pt/Co↑↓ = 0.59×10^15 Ω^-1 m^-2, this is about 0.5; with Ks up to 0.65 erg/cm^2, Tint^SML≈0.85, so θSH≈0.35/0.43≈0.8. The fragility is that λs is imposed by assuming a dominant Elliott-Yafet mechanism, G_Pt/Co↑↓ is a first-principles value not measured on these stacks, and the SML coefficient is calibrated in earlier work by the same group. If G↑↓ is doubled, Tint rises to about 0.58 and θSH drops to about 0.6; if Tint were unity, θSH would be 0.35. The paper provides no uncertainty analysis for these interface parameters, and its own text states that the exact theoretical limit remains unsettled. Because the general θSH≥0.8 bound inherits this model dependence, the central quantitative conclusion is conditional on an unvalidated interface model.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports harmonic Hall measurements of the dampinglike spin-orbit torque in [Pt d/Ti 0.2]m/Pt d/Co stacks with sub-monolayer Ti insertions. The directly measured torque efficiency per applied field, ξDL^E, decreases monotonically with insertion number m, while the torque efficiency per unit current density, ξDL^j, increases from about 0.16 at m=0 to a peak of about 0.35 at m=7 (ρxx≈90 μΩ cm) and then declines slightly. Using a spin-transparency model Tint = Tint^SBF × Tint^SML, the authors convert these measured efficiencies into the spin Hall ratio θSH and spin Hall conductivity σSH, obtaining θSH≈0.8 for [Pt 0.75/Ti 0.2]7/Pt 0.75. They attribute the decrease of σSH with increasing resistivity to shortening of the carrier lifetime, rather than to strain or loss of crystalline order, and they conclude that this trade-off sets a practical upper bound of θSH≥0.8 for Pt-based spin Hall materials.","tokens_in":10479,"tokens_out":6097,"duration_ms":60364,"significance":"If the quantitative claim θSH≈0.8 is accepted, the paper provides both a practically useful spin Hall metal and an important limit for the design of Pt-based spin-torque devices. The paper has notable strengths: the harmonic Hall data are direct and internally consistent; the non-monotonic evolution of ξDL^j and the σSH-versus-σxx scaling are robust to the spin-transparency model because Tint is nearly constant for m≥3 in the bulk limit; and the demonstration of ξDL^j≈0.35 at a moderate resistivity of 90 μΩ cm is a valuable materials result independent of the absolute θSH calibration. The main caveat is that the headline θSH≈0.8 and the upper-bound statement rest on model parameters taken from earlier work, with no uncertainty propagation. The paper would be substantially strengthened by a quantitative sensitivity analysis and by tempering the upper-bound language.","major_comments":[{"comment":"The conversion from the directly measured torque efficiency ξDL^j to the headline θSH≈0.8 relies entirely on the model Tint = Tint^SBF × Tint^SML, with Tint^SBF given by Eq. (1) using λs obtained by assuming a dominant Elliott-Yafet mechanism and G_Pt/Co^↑↓ = 0.59×10^15 Ω^-1 m^-2, and Tint^SML from the linear relation of Ref. [30]. No uncertainty or sensitivity analysis is provided for these interface parameters. If G_Pt/Co^↑↓ were doubled, Tint would increase and θSH would drop to approximately 0.6; in the limiting case Tint=1, θSH would equal the directly measured ξDL^j≈0.35. Because the central quantitative claim and the abstract's upper bound inherit this model dependence, the authors should provide a sensitivity analysis over G^↑↓, λs, and Ks, or an independent experimental determination of Tint for this specific Pt/Ti/Co interface, before claiming θSH≈0.8.","section":"Spin transparency model and Fig. 3(b)"},{"comment":"The attribution of the σSH decrease to carrier lifetime is stated as 'unambiguous,' but the supporting comparison between Pt/Ti and Pt/Hf multilayers is not controlled: the two series differ in total Pt thickness (6 nm versus 4 nm), insertion-layer density, and degree of structural disorder. The dashed line in Fig. 4(d) is a guide to the eye rather than a quantitative fit to the predicted dirty-metal σSH(σxx) scaling. The data are consistent with the carrier-lifetime mechanism, but the wording overstates the evidence; a quantitative comparison with the theoretical σSH(σxx) curve, or additional samples that vary disorder independently of resistivity, would strengthen the mechanistic conclusion.","section":"Mechanistic discussion and Fig. 4(d)"},{"comment":"The phrase 'upper bound of θSH ≥ 0.8' is internally inconsistent, as is the conclusion's 'effective upper bound of ξDL^j ≥ 0.4 (θSH ≥ 0.8)': an achieved maximum at m=7 is evidence for a lower bound on the attainable maximum, not for an upper bound. The text also acknowledges that 'the exact theoretical limit of θSH and the corresponding resistivity have remained unsettled.' A discrete scan of m values cannot establish a practical upper bound unless the authors argue, or demonstrate, that no other insertion density, layer sequence, or microstructure can improve on m=7. Please revise the claim to 'maximum measured value' or provide the missing argument for why this is a true upper bound.","section":"Abstract and Conclusion"}],"minor_comments":[{"comment":"The caption labels the last two panels both as '(d)'; the panel showing spin transparency should be labeled '(e)'.","section":"Fig. 3 caption"},{"comment":"The word 'Unambigous' should be corrected to 'Unambiguous'.","section":"Main text, Section 4"},{"comment":"The phrase 'the the derivative' should be corrected to 'the derivative'.","section":"Main text, Section 2"},{"comment":"The Ks values used to estimate Tint^SML are not reported in the main text; the authors should give the measured Ks values and their uncertainties for each m, since this is a central input to the absolute θSH calibration.","section":"Spin transparency model"},{"comment":"The statement that the uncertainty of ξDL^E is 'less than 2%' appears to refer to statistical fitting precision; please clarify whether systematic uncertainties from the harmonic-analysis model, such as the neglect of fieldlike torque contributions or the macrospin assumption, are included in this estimate.","section":"Measurement uncertainty"},{"comment":"The wording 'practical upper bound of θSH ≥ 0.8' should be harmonized with the actual data and with the conclusion; as written, 'upper bound' and '≥' are contradictory.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"For the editor: This manuscript is likely to be of interest to the spin-orbit torque community, and the direct experimental trend—especially the peak in ξDL^j at moderate resistivity—is credible and useful. My main reservation is that the headline θSH≈0.8 and the upper-bound statement depend on interface spin-transparency parameters taken from the authors' own earlier publications, with no uncertainty propagation in the present dataset. I recommend major revision rather than rejection because this can be addressed by adding a sensitivity analysis, an independent Tint check, and revised upper-bound language. The underlying materials demonstration (a Pt-based multilayer with ξDL^j≈0.35 at 90 μΩ cm) does not require the contested absolute value of θSH and should be preserved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: this is a useful materials paper with a robust directly measured trend, and an overreaching model-dependent claim about an 'upper bound' that the data don't establish. The specific [Pt 0.75/Ti 0.2]7/Pt 0.75 stack with ξ_DL^j ≈ 0.35 at 90 μΩ cm is a genuinely better efficiency-resistivity combination than W, Ta, or the earlier Pt/Hf multilayers. The harmonic Hall data are credible, and the direct comparison of Ti vs Hf insertions showing the same σSH–σxx scaling is a nice experimental check that the degradation tracks the carrier lifetime rather than strain or disorder.\n\nThe soft spot is the conversion from the measured torque efficiency to the internal spin Hall ratio. θSH ≈ 0.8 is obtained by dividing by a spin transparency Tint built from the Elliott-Yafet assumption for λs, a first-principles G↑↓, and a spin-memory-loss coefficient calibrated in the authors' prior papers. None of those parameters carries an uncertainty, and they all move θSH substantially. At Tint=1 the same data give θSH≈0.35. The paper even notes the theoretical limit is 'unsettled', yet the conclusion asserts a universal upper bound. That is a load-bearing mismatch.\n\nThere is also a wording problem in the abstract and conclusion: 'upper bound of θSH ≥ 0.8' is an inequality pointing the wrong way for an upper bound. What the data show is a peak of about 0.8 in this one series, not a bound for all Pt-based materials.\n\nNone of this kills the paper. The central qualitative result—σSH falls with increasing resistivity roughly as expected from carrier-lifetime shortening—doesn't depend much on Tint, because Tint is nearly constant for m ≥ 3. The materials claim stands on the directly measured ξ_DL^j.\n\nBottom line: send this to a serious referee. It deserves publication after the authors either drop the 'upper bound' framing or actually bound it, and add an honest error budget for Tint. Spintronics materials people will want the data.","headline":"Solid materials result with a directly measured efficiency-resistivity improvement; the θSH≈0.8 and 'upper bound' claims are model-dependent and overstated but not fatal.","tokens_in":11034,"tokens_out":2727,"would_cite":true,"duration_ms":26979,"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":"The paper establishes a practical upper bound near 0.8 on the spin Hall ratio achievable in platinum by shortening carrier lifetime, and shows that a Pt/Ti multilayer reaches it at 90 μΩ cm.","keywords":["spin-orbit torque","spin Hall effect","spin Hall conductivity","spin Hall ratio","Pt/Ti multilayers","carrier lifetime","interfacial spin transparency","dirty-metal regime"],"falsifier":"Measure the dampinglike torque on the same $[\\mathrm{Pt}\\,0.75/\\mathrm{Ti}\\,0.2]_7/\\mathrm{Pt}\\,0.75$ stack with a technique that does not assume the interfacial spin transparency, for example a thickness-series spin-torque ferromagnetic resonance analysis that extracts the bulk $\\sigma_{\\mathrm{SH}}$; if the resulting $\\theta_{\\mathrm{SH}}$ is near 0.35 rather than 0.8, the $T_{\\mathrm{int}}$ model used for the headline number is the limiting assumption.","tokens_in":9889,"feed_emoji":"🧲","tokens_out":13868,"duration_ms":123418,"temperature":0.7,"pith_summary":"The paper tries to establish that there is a practical upper bound, at least about 0.8, on how much the spin Hall ratio of platinum can be increased by making the metal more resistive. It shows that inserting sub-monolayer titanium layers into Pt raises the resistivity through interfacial scattering while leaving the face-centered-cubic structure of platinum largely intact, and that the intrinsic spin Hall conductivity survives strain and moderate disorder but falls rapidly once carrier lifetime shortens enough to enter the dirty-metal regime. The optimum film, $[\\mathrm{Pt}\\,0.75\\ \\mathrm{nm}/\\mathrm{Ti}\\,0.2\\ \\mathrm{nm}]_7/\\mathrm{Pt}\\,0.75\\ \\mathrm{nm}$, reaches $\\theta_{\\mathrm{SH}}\\approx 0.8$ and a dampinglike torque efficiency per unit current density $\\xi_{\\mathrm{DL}}^{j}\\approx 0.35$ at a resistivity of 90 μΩ cm. The authors argue that this sets an upper bound near 0.8 for any heterogeneous Pt-based material whose resistivity is raised by shortening carrier lifetime, and that the multilayer is a strong candidate for energy-efficient spin-orbit-torque memory and logic.","feed_headline":"Pt/Ti multilayers cap spin Hall ratio at 0.8","feed_subtitle":"Thin titanium layers raise resistivity without destroying spin Hall flow, giving strong torque at low impedance.","key_machinery":"The load-bearing object is the identity $\\xi_{\\mathrm{DL}}^{j}=(2e/\\hbar)\\,T_{\\mathrm{int}}\\,\\sigma_{\\mathrm{SH}}\\,\\rho_{xx}$, which connects the measured torque efficiency to the bulk spin Hall ratio through the interfacial spin transparency $T_{\\mathrm{int}}$, together with $\\theta_{\\mathrm{SH}}=\\sigma_{\\mathrm{SH}}/\\sigma_{xx}$. The experimental mechanism is the $[\\mathrm{Pt}\\,d/\\mathrm{Ti}\\,0.2]_m/\\mathrm{Pt}\\,d$ multilayer, where each sub-monolayer Ti insertion acts as a strong interfacial scatterer that raises $\\rho_{xx}$ while leaving the fcc order of Pt mostly intact, and where the spin diffusion length is connected to resistivity through an Elliot-Yafet assumption. The argument then rests on the intrinsic spin Hall conductivity of Pt being robust against strain and moderate disorder but sensitive to carrier lifetime, so the scaling of $\\sigma_{\\mathrm{SH}}$ with $\\sigma_{xx}$ identifies the dirty-metal regime as the limiting factor.","core_discovery":"The central claim is that $\\theta_{\\mathrm{SH}}=\\sigma_{\\mathrm{SH}}/\\sigma_{xx}$ for Pt-based spin Hall materials cannot be pushed much beyond roughly 0.8 by resistivity engineering, because the same carrier-lifetime shortening that raises $\\rho_{xx}$ eventually degrades the intrinsic spin Hall conductivity $\\sigma_{\\mathrm{SH}}$. The paper supports this with a series of Pt/Ti multilayers: increasing the number of 0.2 nm Ti insertions raises $\\rho_{xx}$ from 26.5 to 192 μΩ cm while the measured dampinglike torque efficiency per applied field falls by more than a factor of two. Converting those data through an assumed interfacial spin transparency yields an internal $\\theta_{\\mathrm{SH}}$ that rises from about 0.46 for pure Pt to about 0.8 at the optimum and then declines. The decline of $\\sigma_{\\mathrm{SH}}$ tracks the decline of electrical conductivity on the same curve for both Ti and Hf insertions, which the authors take as evidence that shortened carrier lifetime, rather than strain or disrupted crystal order, is the dominant degradation mechanism.","pith_inferences":["A natural next test is to extract $\\sigma_{\\mathrm{SH}}$ with an interface-independent method, such as a thickness-series spin-torque ferromagnetic resonance analysis, to check whether $\\theta_{\\mathrm{SH}}\\approx 0.8$ survives without the assumed $T_{\\mathrm{int}}$.","If the carrier-lifetime ceiling is universal for intrinsic spin Hall metals, beating $\\theta_{\\mathrm{SH}}\\approx 0.8$ would require a different lever: raising $\\sigma_{\\mathrm{SH}}$ itself through band-structure engineering or a different crystal phase, not adding more resistivity.","Because the paper estimates spin-memory loss reduces the interface transmission by at most about 15%, an interface engineered to weaken interfacial spin-orbit coupling could raise $\\xi_{\\mathrm{DL}}^{j}$ by roughly that fraction without changing the bulk bound.","The same tradeoff should appear in other intrinsic spin Hall metals, so measuring $\\sigma_{\\mathrm{SH}}$ as a function of $\\sigma_{xx}$ in Pd-based or other 5d alloys would test whether their practical ceilings follow the same curve."],"forward_implications":["Raising the resistivity of a Pt-based spin Hall material by shortening carrier lifetime has diminishing returns: past the optimal insertion density, $\\xi_{\\mathrm{DL}}^{j}$ falls because $\\sigma_{\\mathrm{SH}}$ drops faster than $\\rho_{xx}$ rises.","A device built on $[\\mathrm{Pt}\\,0.75/\\mathrm{Ti}\\,0.2]_7/\\mathrm{Pt}\\,0.75$ can deliver torque efficiency comparable to $\\beta$-W while operating at lower resistivity, which lowers write energy, device impedance, and Joule-heating endurance problems.","The collapse of the Ti and Hf insertion data onto one $\\sigma_{\\mathrm{SH}}$-versus-$\\sigma_{xx}$ curve means the mechanism and the ceiling are not specific to titanium; other Pt-based multilayers should obey the same scaling.","Adding more scattering layers beyond the optimum is counterproductive: it continues to raise $\\rho_{xx}$ but lowers $\\xi_{\\mathrm{DL}}^{j}$, so the optimum is a genuine maximum of torque per current density."],"supporting_citations":[{"why":"Establishes that the giant intrinsic spin Hall conductivity of Pt degrades as carrier lifetime shortens, the mechanism this paper extends.","marker":"[23]"},{"why":"Prior Pt/Hf multilayer study that introduced sub-monolayer insertions to raise resistivity; provides the comparison data for the σSH versus σxx scaling.","marker":"[24]"},{"why":"Supplies the spin diffusion length and spin conductance of Pt used to compute the spin-backflow transparency.","marker":"[35]"},{"why":"Establishes how spin Hall torque efficiency depends on Pt/ferromagnet interface transparency and provides the spin-mixing conductance basis.","marker":"[29]"},{"why":"Provides the spin-memory-loss scaling with interfacial spin-orbit coupling used for the second factor of the transparency.","marker":"[30]"},{"why":"Theoretical prediction of the intrinsic spin Hall effect in transition metals and its reduction in the dirty regime, the expected behavior the data are compared against.","marker":"[19]"},{"why":"Earlier Pt alloy work showing impurity-induced resistivity gains degrade spin Hall efficiency, motivating the multilayer approach.","marker":"[22]"}],"fun_headline_variants":["Spin Hall ratio ceiling 0.8 from Pt/Ti trade-off","Pt/Ti multilayers optimize spin Hall efficiency to 0.8","Carrier lifetime caps Pt spin Hall ratio at 0.8","Optimal Ti inserts yield Pt spin Hall ratio 0.8"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline $\\theta_{\\mathrm{SH}}\\approx 0.8$ rests on the model's interfacial spin transparency of roughly 0.5; if the true transparency were perfect, the same torque data would give only about $\\theta_{\\mathrm{SH}}\\approx 0.35$.","fun_headline_variants_meta":{"raw":{"variants":["Spin Hall ratio ceiling 0.8 from Pt/Ti trade-off","Pt/Ti multilayers optimize spin Hall efficiency to 0.8","Carrier lifetime caps Pt spin Hall ratio at 0.8","Optimal Ti inserts yield Pt spin Hall ratio 0.8"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000254,"raw_usage":{"total_tokens":1607,"prompt_tokens":1020,"completion_tokens":587,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":636,"completion_tokens_details":{"reasoning_tokens":510}},"tokens_in":636,"tokens_out":587,"duration_ms":5369,"temperature":1.0,"reasoning_tokens":510,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:41:53.994575+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the dampinglike torque on the same $[\\mathrm{Pt}\\,0.75/\\mathrm{Ti}\\,0.2]_7/\\mathrm{Pt}\\,0.75$ stack with a technique that does not assume the interfacial spin transparency, for example a thickness-series spin-torque ferromagnetic resonance analysis that extracts the bulk $\\sigma_{\\mathrm{SH}}$; if the resulting $\\theta_{\\mathrm{SH}}$ is near 0.35 rather than 0.8, the $T_{\\mathrm{int}}$ model used for the headline number is the limiting assumption.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes that the giant intrinsic spin Hall conductivity of Pt degrades as carrier lifetime shortens, the mechanism this paper extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior Pt/Hf multilayer study that introduced sub-monolayer insertions to raise resistivity; provides the comparison data for the σSH versus σxx scaling."},{"cited_title":"N guyen, D","cited_arxiv_id":null,"evidence_quote":"Supplies the spin diffusion length and spin conductance of Pt used to compute the spin-backflow transparency."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes how spin Hall torque efficiency depends on Pt/ferromagnet interface transparency and provides the spin-mixing conductance basis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the spin-memory-loss scaling with interfacial spin-orbit coupling used for the second factor of the transparency."},{"cited_title":"Tanaka, H","cited_arxiv_id":null,"evidence_quote":"Theoretical prediction of the intrinsic spin Hall effect in transition metals and its reduction in the dirty regime, the expected behavior the data are compared against."},{"cited_title":"Nguyen, M","cited_arxiv_id":null,"evidence_quote":"Earlier Pt alloy work showing impurity-induced resistivity gains degrade spin Hall efficiency, motivating the multilayer approach."}],"review_version":1}