{"id":"abbd8e53-b78e-4081-a926-ab99cf2fead9","arxiv_id":"2501.16162","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"TranSIESTA now supports full spinor-wavefunction non-equilibrium transport, enabling first-principles simulation of spin-orbit and non-collinear-spin devices.","lead":"The authors upgraded the open-source TranSIESTA code to handle spin as a full two-component quantum object, so it can simulate devices with spin-orbit coupling and non-collinear magnetism. They validated it on iron chains, magnetic tunnel junctions, semiconductor junctions, and carbon nanotubes, and added a new way to analyze spin-dependent current.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Bulk-electrode screening assumption is the load-bearing weak point: the 1D Fe chain/domain-wall results use only 4-atom central regions with no self-energy transferability check.","rationale":"Good-faith reading: this is a useful and credible implementation paper. The spinor NEGF formalism is standard, the matrix layout and SSR reuse are sensible, and the band-structure/transmission comparisons (iron chain, TMD versus QuantumATK) show the code works in several nontrivial limits. I am not claiming the screening assumption is false, only that it is load-bearing and under-tested precisely where it is most fragile. The reader's weakest assumption pointed in this direction, but I would phrase it more narrowly: the TMD 40/56-atom test and the Fe/MgO/Fe layer tests (Fig. 15) provide partial evidence, so the open gap is specifically the 1D chain/domain-wall setup with a 4-atom device region. A buffer-length convergence test would settle whether the concern lands. I therefore keep the existing CONDITIONAL verdict; no adjustment is needed.","tokens_in":29596,"tokens_out":10883,"duration_ms":120303,"concrete_test":"Repeat the 180-degree Neel domain-wall zero-bias transmission calculation of Sec. VII.A with the same 4-atom wall, but extend the scattering region by 2 and 4 additional Fe atoms on each side, held at the electrode spin orientation. If the Fermi-level transmission, or the total transmission within +/-1 eV, changes by more than about 5% between the 4-atom and 8-atom buffers, the bulk-electrode self-energy is not converged and the reported domain-wall resistance is not reliable. For completeness, also repeat the Fe/MgO/Fe transmission with 13 vs 17 Fe layers between the MgO and the electrode boundary.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that TranSIESTA can do reliable spinor DFT+NEGF transport. Every reported transmission/current depends on the Sec. IV assumption that electrodes are bulk-like and that the scattering region screens the perturbation, so surface self-energies from the bulk electrode Hamiltonian are transferable (Eq. 5, Fig. 1). This is the least-secure link in the validation. The TMD heterojunction is explicitly tested (40 vs 56 atoms, Fig. 19), but this is a heavily doped metallic monolayer, so the conclusion does not automatically transfer. For the 1D iron-chain/domain-wall results (Sec. VII.A.3), the scattering region contains only 4 Fe atoms between semi-infinite electrodes, and no buffer-size convergence is reported. In a 1D metal screening is slow, and the domain-wall transmission could depend on how abruptly the bulk self-energy is joined to the wall. Fe/MgO/Fe has some layer-convergence data (Fig. 15), but that tests the number of relaxed device layers, not the convergence of the electrode self-energy with buffer length. Thus the most load-bearing assumption is tested in only one favorable system and is untested in the 1D case where it is most questionable.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports an implementation of spinor (non-collinear and spin-orbit) DFT+NEGF in the open-source TranSIESTA code. The authors extend the Green's function, density matrix, self-energy, and transmission routines from independent spin channels to 2x2 spin blocks, implement spin-channel projected transmission via scattering-matrix projectors, and apply the code to monatomic Fe chains with domain walls, Fe/MgO/Fe junctions, a gated MoS2/WS2 lateral heterojunction, and carbon nanotubes with magnetic clusters. The results are compared with QuantumATK and previous DFT studies.","tokens_in":29803,"tokens_out":7958,"duration_ms":78390,"significance":"If the implementation is correct, it fills a genuine gap: an open-source DFT+NEGF code capable of multi-terminal finite-bias transport with full spinor wavefunctions. The formalism is standard and the manuscript contains useful methodological contributions, including the spin-projected transmission based on the scattering matrix (Eq. 28) and the demonstration that direct broadening-matrix projection (Eq. 29) fails for non-collinear electrode states. The cross-checks against QuantumATK and existing band-structure literature are a strength. However, the validation is uneven: key quantitative claims (TMR, MAE) are not backed by direct comparisons, and the screening assumption underpinning the NEGF construction is tested only in one system.","major_comments":[{"comment":"The abstract and Sec. VII.B state that the Fe/MgO/Fe junction is used to test whether the implementation reproduces previous predictions for tunneling magnetoresistance, but no quantitative TMR ratio is reported anywhere in the manuscript. Only the parallel-state Fermi-level transmission (T about 0.0044) and spin-channel decompositions are given; the antiparallel transmission and the TMR ratio are absent. Please provide T_AP, the TMR ratio (T_P - T_AP)/T_AP at the Fermi level (or over the bias window), and compare with Refs. [74,75] and other DFT results. Without this, the headline validation claim for magnetoresistance is unsubstantiated.","section":"Sec. VII.B.2"},{"comment":"The iron-chain transport calculations use a scattering region of only 4 Fe atoms between semi-infinite electrodes (Sec. VII.A.3), with no buffer-size convergence test for the electrode self-energy. In Sec. IV the method assumes that electrodes are bulk-like and that screening regions ensure transferability of the bulk self-energy (Fig. 1, Eq. 5). For a 1D metallic chain, screening is weak and the domain wall is a non-collinear perturbation extending over several atoms; the abrupt connection to the bulk electrode could affect the transmission. Please report convergence of the domain-wall transmission with the number of buffer atoms, or state this limitation explicitly with an estimate of its effect on the reported conductances.","section":"Sec. VII.A.3 and Sec. IV"},{"comment":"The MAE of the infinite Fe chain is reported to be a factor of 2 lower than Refs. [86,87] and is attributed to bond-length sensitivity. Since no bond-length dependence is shown, this discrepancy is not actually resolved. Please provide the MAE at the same lattice constant as the reference calculations or a plot of MAE versus bond length, so the reader can judge whether the discrepancy is a parameterization effect rather than an implementation error.","section":"Sec. VII.A.1"}],"minor_comments":[{"comment":"The abstract contains a typo: \"Exisiting\" should be \"Existing\".","section":"Abstract"},{"comment":"The spin-box Hermiticity condition is stated as H^{σσ'}_{ij} = (H^{σ'σ}_{ij})^*, but the orbital indices should also be exchanged: H^{σσ'}_{ij} = (H^{σ'σ}_{ji})^*. As written, the condition is only correct for i = j.","section":"Sec. III"},{"comment":"There are typos in the text: \"QunatumATK\" should be \"QuantumATK\", and \"previously obta,ined results\" should be \"previously obtained results\".","section":"Sec. VII.C.3 and Sec. VII.B.2"},{"comment":"The 40-versus-56-atom screening test is performed on heavily hole-doped metallic monolayers; a sentence clarifying that this does not validate the undoped semiconducting case would help set expectations for the reader.","section":"Sec. VII.C.2 / Fig. 19"},{"comment":"The weighting scheme in Eqs. 22-24 uses only the charge-density variance; a sentence explaining why the spin-block off-diagonal terms can be neglected in the weights (while still being included in the density matrix) would clarify the numerical rationale.","section":"Sec. V.E"}],"recommendation":"major_revision","confidential_remarks":"The paper is a code-development manuscript whose central implementation appears sound. The main risks are validation-related: the missing quantitative TMR ratio and the untested screening assumption in the 1D iron-chain transport setup are fixable with additional calculations and would materially strengthen the paper. The authors are TranSIESTA developers; self-citations to Refs. [19,37] are appropriate. The claim that existing open-source codes cannot handle non-collinear transport is somewhat strong and could be softened without harming the paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid code paper, not a new theory. The genuinely new piece is the spin-channel projected transmission via scattering-matrix projectors (Eq. 28 with Γ^{σn}=Γ^{1/2}|σn><σn|Γ^{1/2}). The paper also shows, correctly, that the cheaper broadening-matrix projection (Eq. 29) fails when electrode states are non-collinear. That is a real methodological contribution, and the three test systems are well chosen.\n\nThe implementation itself looks sound. The 2x2 spin-block extension, reuse of the Sancho-Sancho-Rubio self-energy solver, BTD inversion, and complex-contour handling all follow standard NEGF. The cross-checks against QuantumATK and earlier DFT studies for the TMDs and bulk Fe are useful. The iron-chain AMR, domain-wall spin-flip transmission, and Fe/MgO/Fe results are qualitatively consistent with prior literature, and the authors disclose the factor-2 MAE discrepancy and attribute it to bond length, which is plausible.\n\nThe soft spots, in rough order:\n\n1. Reproducibility. No code version, input files, or data are provided. For a code paper this is a real deficiency and easy to fix with a tagged SIESTA release and test inputs.\n\n2. The bulk-electrode screening assumption is the weakest link. The TMD junction gets a 40-vs-56-atom convergence test, but the iron-chain/domain-wall calculations use only four Fe atoms between electrodes, with no buffer-size convergence check. In a 1D metal, screening is slow, so self-energy transferability is not obviously safe there. This does not prove the results wrong, but the paper should show a buffer-size test for the chain or explicitly caveat the domain-wall numbers.\n\n3. The Fe/MgO/Fe TMR is never quantified. The paper reports transmission functions but never the TMR ratio, which is the natural metric for that system.\n\n4. The text references supplementary figures and tables (Fig. S5, Tables S1, S2) that are not in the arXiv version. Minor, but sloppy.\n\n5. The abstract's claim that existing open-source DFT codes either approximate non-equilibrium or rely on collinear spins is stronger than the cited landscape. SMEAGOL and GOLLUM have spin-aware transport capabilities, even if not this exact spinor DFT+NEGF combination. The novelty is TranSIESTA's spinor DFT+NEGF, not the general idea.\n\nWho this is for: anyone doing SOC or non-collinear transport with SIESTA/TranSIESTA, and the DFT+NEGF code community more broadly. It deserves a serious referee; I would recommend 'revise and resubmit' rather than reject, with the reproducibility and buffer-convergence issues as the main conditions.","headline":"Solid spinor NEGF implementation with a genuinely new transmission projector; needs reproducibility and buffer-convergence checks before acceptance.","tokens_in":30456,"tokens_out":4616,"would_cite":true,"duration_ms":41746,"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 reports the extension of the open-source DFT+NEGF transport code TranSIESTA to full spinor wave functions, enabling first-principles calculations of transport in devices with spin-orbit coupling and non-collinear magnetism.","keywords":["spin-orbit coupling","non-equilibrium Green's function","DFT+NEGF","TranSIESTA","non-collinear magnetism","quantum transport","spin-channel projected transmission","magnetoresistance"],"falsifier":"Compute the zero-bias transmission of the Fe/MgO/Fe junction with 5, 6, and 7 fixed iron layers between the MgO barrier and the electrode region; if the transmission changes by more than the numerical tolerance instead of converging, the bulk-electrode assumption is violated and the spinor transport results are not converged. The same screening test applied to the TMD heterojunction (40 vs 56 atoms) is the only place in the paper where this assumption is explicitly checked.","tokens_in":29325,"feed_emoji":"🧲","tokens_out":9261,"duration_ms":81258,"temperature":0.7,"pith_summary":"The authors set out to remove a long-standing restriction of the open-source DFT+NEGF transport code TranSIESTA: it previously treated only spin-unpolarized or collinear-spin systems, where the Hamiltonian splits into independent spin blocks. Their new implementation promotes every NEGF matrix to 2x2 blocks in spin space, so the full spinor wave function is carried through the calculation. That makes it possible, for the first time in this code, to simulate non-equilibrium multi-terminal transport with spin-orbit coupling and non-collinear magnetic configurations from first principles. The authors demonstrate the capability on monatomic iron chains with domain walls, Fe/MgO/Fe tunnel junctions, a lateral MoS2/WS2 heterojunction, and carbon nanotubes functionalized with magnetic molecules. If correct, this closes the gap between open-source DFT transport codes and the spinor physics needed for topological materials and spintronics devices.","feed_headline":"TranSIESTA now simulates spin-orbit transport from first principles","feed_subtitle":"The open-source DFT+NEGF code drops the collinear-spin limit, enabling full 2x2 spinor device simulations.","key_machinery":"The load-bearing object is the retarded Green's function in spinor form, $G_k(z)=[zS_k-H_k-\\sum_e\\Sigma_{e,k}(z)]^{-1}$, where $S_k$, $H_k$, and the self-energies $\\Sigma_{e,k}$ are built with each orbital matrix element expanded as a $2\\times2$ block in spin space; this is what couples the spin channels. The central analysis identity is the spin-channel projected transmission $T^{\\sigma\\vec n,\\sigma'\\vec m}_{e,e'}(z)=\\int_{\\mathrm{BZ}} dk\\,\\mathrm{Tr}\\{\\Gamma^{\\sigma\\vec n}_{e,k}(z)\\,A^{\\sigma'\\vec m}_{e',k}(z)\\}$, expressed through spin-selective broadening matrices $\\Gamma^{\\sigma\\vec n}_{e,k}=\\Gamma^{1/2}_{e,k}|\\sigma\\vec n\\rangle\\langle\\sigma\\vec n|\\Gamma^{1/2}_{e,k}$ and the corresponding spectral density $A^{\\sigma\\vec n}_{e,k}=G_k\\Gamma^{\\sigma\\vec n}_{e,k}G_k^\\dagger$. This object lets the user resolve transmission into same-spin and spin-flip channels for arbitrary quantization axes in each electrode, which is the key new observable enabled by the implementation.","core_discovery":"The paper's central claim is that TranSIESTA can now perform self-consistent DFT+NEGF calculations for general spinors, not just collinear spins. In the collinear case the Hamiltonian is block diagonal in spin and the two channels can be treated independently; with spin-orbit coupling or non-collinear magnetization the off-diagonal spin blocks couple the channels, and the old code could not handle this. The implementation therefore expands each orbital matrix element $A_{ij}$ into a $2\\times2$ spin block, keeps the sparse-matrix bandwidth small by interleaving spin indices, reuses the Sancho-Sancho-Rubio algorithm for surface self-energies (which is agnostic to spin indices), and supports both full and block-tridiagonal inversion of the spinor Green's function $G_k(z)=[zS_k-H_k-\\sum_e\\Sigma_{e,k}(z)]^{-1}$. The post-processing tool TBTrans is extended with a spin-channel projected transmission that projects the scattering matrix onto spin eigenstates along arbitrary axes, and the authors show that a cheaper alternative that projects the broadening matrix instead gives unphysical results wherever the electrode spin texture is non-collinear. Validation runs reproduce reference band structures and transmissions, and reveal SOC-induced effects such as band splitting, avoided crossings, spin-flip transmission through domain walls, a 1% ballistic anisotropic magnetoresistance in bulk iron, and diode-like IV behavior in the gated MoS2/WS2 junction.","pith_inferences":["The spin-channel projection machinery could be applied beyond total transmission, for example to decompose conductances into spin-valley or spin-momentum-locking contributions in topological surface states; the paper does not attempt this, but Eq. (28) already supplies the operator form.","The demonstrated failure of broadening-matrix projection for non-collinear electrode states is a caution for any transport code that uses such a shortcut; the scattering-matrix projection may be the needed default wherever electrode spin textures are energy-dependent.","A testable extension would be to compute the domain-wall spin-flip transmission as a function of domain-wall width and SOC strength; the code now permits this systematically, and the paper only reports widths of 4-6 atoms.","The {Co4}-CNT result suggests that for molecule-nanotube hybrids, transport simulations with semi-infinite electrodes may be needed to converge magnetic properties, not merely transport functions; this could be checked by computing the PBC magnetic moment for even larger supercells."],"forward_implications":["TranSIESTA can now model finite-bias, multi-terminal devices with spin-orbit coupling or non-collinear spins without a collinear-spin approximation, so topological-material and spintronics devices become accessible to open-source first-principles transport simulation.","Spin-flip transmission between electrodes with different magnetization directions can be computed and decomposed by quantization axis, making domain-wall resistance, anisotropic magnetoresistance, and tunneling magnetoresistance calculations routine.","The implementation inherits the block-tridiagonal inversion and parallel scaling of the current TranSIESTA, so the spinor capability is available for large systems (tested up to 816 atoms and 11088 orbitals).","For semiconductor heterojunctions, the TMD example shows that uniform gating can supply enough screening for the bulk-electrode assumption to hold, allowing self-consistent NEGF studies of lateral 2D junctions with strong spin-orbit coupling.","The open-system treatment of molecule-functionalized carbon nanotubes gives a converged magnetic moment for {Co4}-CNT where periodic supercell calculations did not converge, implying that transport geometry itself can be necessary for correct magnetic ground states in these hybrids."],"supporting_citations":[{"why":"Defines the DFT+NEGF method and the original TranSIESTA formalism that this work extends to spinors.","marker":"[19]"},{"why":"Supplies the next-generation TranSIESTA algorithms, including block-tridiagonal inversion and multi-terminal support, that the spinor implementation reuses.","marker":"[37]"},{"why":"Documents the SIESTA spin methods (non-collinear and spin-orbit) whose Hamiltonian, overlap, and density-matrix structure the transport code must handle.","marker":"[39]"},{"why":"Provides the fully relativistic pseudopotential formalism used to generate spin-orbit-coupled Hamiltonians in the calculations.","marker":"[40]"},{"why":"The Sancho-Sancho-Rubio iterative algorithm used to compute electrode surface self-energies, reused unchanged because it is spin-index agnostic.","marker":"[48]"},{"why":"Supplies the NEGF scattering formalism used to define the scattering matrix and its spin projections.","marker":"[47]"},{"why":"The Fisher-Lee relation connecting the scattering matrix to Green's functions and broadening matrices, used in Eq. (11).","marker":"[58]"},{"why":"Reference DFT transport code against which the TMD monolayer and heterojunction transmissions are compared.","marker":"[100]"}],"fun_headline_variants":["TranSIESTA gets spinor power for topological transport","Full spinor transport in TranSIESTA for SOC","Noncollinear spins in TranSIESTA for spintronics","Spinor transport in TranSIESTA: no more collinear limit","TranSIESTA goes full spinor for SOC transport"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The method assumes the electrodes remain bulk-like and undisturbed by the device, so surface self-energies from a converged bulk DFT calculation stay valid; the paper directly verifies this screening only for the TMD heterojunction, not for the iron chains, Fe/MgO/Fe, or the carbon-nanotube systems.","fun_headline_variants_meta":{"raw":{"variants":["TranSIESTA gets spinor power for topological transport","Full spinor transport in TranSIESTA for SOC","Noncollinear spins in TranSIESTA for spintronics","Spinor transport in TranSIESTA: no more collinear limit","TranSIESTA goes full spinor for SOC transport"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001137,"raw_usage":{"total_tokens":4765,"prompt_tokens":1031,"completion_tokens":3734,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":647,"completion_tokens_details":{"reasoning_tokens":3656}},"tokens_in":647,"tokens_out":3734,"duration_ms":23931,"temperature":1.0,"reasoning_tokens":3656,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T13:41:28.747873+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the zero-bias transmission of the Fe/MgO/Fe junction with 5, 6, and 7 fixed iron layers between the MgO barrier and the electrode region; if the transmission changes by more than the numerical tolerance instead of converging, the bulk-electrode assumption is violated and the spinor transport results are not converged. The same screening test applied to the TMD heterojunction (40 vs 56 atoms) is the only place in the paper where this assumption is explicitly checked.","supporting_citations":[{"cited_title":"Complex magnetic structure of clusters and chains of Ni and Fe on Pt(111)","cited_arxiv_id":null,"evidence_quote":"Reference DFT transport code against which the TMD monolayer and heterojunction transmissions are compared."}],"review_version":1}