{"id":"616ab6da-3a9c-4316-b35b-21f95010d093","arxiv_id":"1908.03365","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"In a Kondo-regime quantum dot with two ferromagnetic leads and a superconductor, the linear cross conductance becomes negative when the superconducting coupling exceeds the normal tunnel coupling, and antiparallel spin polarization increases the negative value.","lead":"This paper predicts that a quantum dot connected to two magnetic leads and a superconductor can show negative nonlocal conductance when the superconducting coupling is strong. The effect is stronger when the two magnetic leads are polarized in opposite directions, because crossed Andreev reflection becomes the dominant current channel.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sign of the reported negative cross conductance depends on an unresolved current-direction convention: Eq. (30) is inconsistent with Eq. (24) and with the stated right-lead current formulas.","rationale":"The reader's weakest assumption is the uncontrolled slave-boson mean-field approximation. That is a legitimate concern about quantitative accuracy, and I agree it should be benchmarked. However, I found a more directly load-bearing internal inconsistency: the sign of the cross conductance, which is the central observable, is not unambiguously defined in the manuscript. Equation (24) defines the current as flowing from each lead into the dot, while Eq. (30) and the plotted positive G_C at Gamma_s=0 are only consistent with the opposite convention for the right lead. This is not a mere typo in notation: the physical interpretation of the headline result, namely that crossed Andreev reflection is responsible for negative nonlocal conductance, changes sign depending on which convention is used. If Eq. (24) is taken literally, the negative values in the figures correspond to ET-dominated rather than CAR-dominated response. Thus the central claim cannot be assessed until the authors clarify the sign convention and correct the mismatch between Eq. (24), Eqs. (26)-(28) after L/R exchange, and Eq. (30). The slave-boson approximation remains a secondary concern that would affect the quantitative magnitude and possibly the sign boundary, but the current-direction inconsistency is more fundamental because it prevents an unambiguous reading of the reported result even within the mean-field calculation. I therefore recommend keeping the CONDITIONAL verdict, but the condition should explicitly require resolving the sign convention and verifying that the corrected cross conductance reproduces the reported negative LCC.","tokens_in":12691,"tokens_out":10177,"duration_ms":121872,"concrete_test":"Recompute dI_R/dV analytically from Eq. (24) using the same Green's functions and the exchanged versions of Eqs. (26)-(28). Evaluate at Gamma_s=0, epsilon_d=-U/2, p=0: the literal Eq. (24) convention gives G_C = -G0, whereas Eq. (30) gives +G0. Then check whether the plotted curves in Figs. 2-5 use the literal Eq. (24) convention or the conventional right-lead current convention (current from dot into right lead). If the latter, replace Eq. (30) by G_C = GET - GCAR and verify that all reported negative values are obtained with that sign; if the literal convention is retained, the reported negative LCC values become positive and the headline claim fails.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Equation (24) defines I_eta as the current flowing from lead eta into the quantum dot: I_eta = -e d/dt sum_k c^dagger c. For the right lead at Gamma_s=0 and positive left bias, electrons flow from the dot into the right lead, so the literal I_R is negative and dI_R/dV should be -GET, not +GET. Exchanging L and R in Eqs. (26)-(28) gives dI_ET_R/dV = -GET, dI_CAR_R/dV = +GCAR, and dI_DAR_R/dV = 0, so with Eq. (24) the cross conductance is G_C = -GET + GCAR. The paper instead states Eq. (30), G_C = GET - GCAR, and plots G_C = +G0 at Gamma_s=0. This is internally inconsistent: either Eq. (24) is wrong and the intended I_R is the conventional current from the dot into the right lead, or the sign of the headline negative LCC is reversed. Because the central claim is specifically that CAR makes the cross conductance negative, the attribution of the sign to CAR depends entirely on this convention. The quantitative value G_C ~ -G0/5 in the AP configuration cannot be interpreted until the convention is fixed.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the zero-temperature subgap transport of a quantum dot connected to two ferromagnetic leads and one superconducting lead in the Kondo regime. Using the finite-U Kotliar-Ruckenstein slave-boson mean-field approach combined with the nonequilibrium Green function technique, the authors compute local and cross linear conductances as functions of the dot level, the superconducting proximity coupling Γ_s, and the spin polarization p of the ferromagnetic leads. They report that a sufficiently strong superconducting coupling suppresses the local conductance and can make the linear cross conductance negative, that the antiparallel lead configuration enhances this negative cross conductance through crossed Andreev reflection, and that the parallel configuration instead splits the Kondo resonance and suppresses the negative nonlocal response. The paper also presents nonlinear differential conductance results showing Kondo zero-bias anomalies and their splitting at large Γ_s.","tokens_in":13002,"tokens_out":21458,"duration_ms":220120,"significance":"The central prediction—that the nonlocal current response changes sign when crossed Andreev reflection dominates elastic cotunneling—is a concrete and falsifiable statement for three-terminal hybrid devices. A clear strength is the explicit decomposition of the local and cross conductances into ET, DAR, and CAR components in Eqs. (29)-(33), which gives a transparent physical mechanism and shows that the DAR process drops out of the cross conductance while CAR enters with the opposite sign to ET. The calculation has no fitted parameters beyond the model parameters (U, Γ_s, p, ε_d, Γ), and the parameter study spans the Kondo, mixed-valence, and empty-orbital regimes. However, the significance is limited by an internal sign-convention inconsistency in the definition of the right-lead current, which must be resolved before the headline negative-cross-conductance claim can be interpreted; in addition, the slave-boson mean-field treatment is uncontrolled in the simultaneous presence of strong correlations, superconductivity, and ferromagnetism, so the quantitative values (e.g., G_C ≈ -G0/5 in Fig. 4) should be taken with caution.","major_comments":[{"comment":"The sign convention for the right-lead current is inconsistent, and this directly affects the central claim. Equation (24) defines I_eta = -e d/dt sum_k c†_{eta k sigma} c_{eta k sigma}, so positive I_eta corresponds to electrons leaving lead eta (the usual 'current from the lead into the dot' in the NEGF convention). Exchanging L and R in Eqs. (26)-(28) gives I_ET_R = -GET and I_CAR_R = +GCAR in the linear-response limit, and therefore G_C = dI_R/dV = -GET + GCAR, not GET - GCAR as printed in Eq. (30). The explanatory sentence after Eq. (30)—'a hole entering the right lead is physically equivalent to an electron injecting into the QD from the right lead'—indicates that the authors actually intend I_R to be the current from the QD into the right lead, which is the negative of the quantity defined by Eq. (24). Under that alternative convention Eq. (30) is correct, but then the derivation from Eqs. (26)-(28) is missing a minus sign and the definition in Eq. (24) is mislabeled. Because the abstract and Figs. 2-5 use the sign of Eq. (30) to conclude that 'when Γ_s is bigger than Γ the LCC becomes negative,' the authors must state the current-direction convention unambiguously, correct Eq. (24) or Eq. (30) accordingly, and re-examine whether their stated condition for negative cross conductance is preserved.","section":"II.C, Eqs. (24)-(30)"},{"comment":"The self-consistent equations are the basis for all numerical results, but they are stated without derivation after 'a lengthy and tedious calculation,' and Eqs. (18)-(19) appear to contain index errors. In Eq. (18), the first square bracket contains ~Γ_L1 f_L(ω) + ~Γ_R1 f_R(ω), while the second square bracket mixes ~Γ_L2 with ~Γ_R1 rather than following the same η pattern; Eq. (19) has a similar structure. As printed, these expressions cannot be checked by a reader, and the reliability of Figs. 2-6 rests on them. I request that the authors provide the derivation of the slave-boson equations of motion or at least state the origin of each term, and correct the indices and prefactors in Eqs. (18)-(19).","section":"II.B, Eqs. (9)-(19)"},{"comment":"The SBMF approximation replaces all slave-boson operators by their expectation values, an approximation that is uncontrolled when the dot is simultaneously subject to strong Coulomb interaction, superconducting pairing with Γ_s ~ Γ, and ferromagnetic leads with p = 0.5. Since the quantitative prediction G_C ≈ -G0/5 in Fig. 4 and the sign of the cross conductance in the strong-coupling regime are central claims, the paper should benchmark the approach against known limits (for example, the normal-lead Kondo limit, or available numerical renormalization-group results for a dot coupled to a superconductor) and discuss how the mean-field decoupling may affect the ET-CAR competition.","section":"II.B and III"}],"minor_comments":[{"comment":"There are numerous typographical errors ('Konod', 'supercoducting', 'Adnreev', 'injuring', 'diffenertial'); the text should be carefully proofread.","section":"Throughout"},{"comment":"The y-axis of Fig. 4(b) is labeled 'Conductance (2e2/h)' but should be labeled 'Cross conductance (2e2/h)'.","section":"Fig. 4"},{"comment":"The statement that 'if Γ_s = 0 ... the cross conductance reduces to the local conductance' is only correct for the convention in which I_R is the current from the QD into the right lead; please make this convention explicit at that point.","section":"After Eq. (30)"},{"comment":"The final term f_η(ω)G^A_d11(ω) in Eq. (18) appears to lack a prefactor relative to the other terms in the integrand; please verify the expression.","section":"Eq. (18)"},{"comment":"Reference 41 appears with corrupted text ('Micha/suppress lek') in the compiled version; please ensure all bibliographic entries are typeset correctly.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main obstacle is the sign-convention issue. I believe it is fixable: if the authors intend G_C = ∂J_R/∂V with J_R the current from the QD into the right lead, they should correct Eq. (24) and the text, and verify that the numerical plots were generated with Eq. (30). If they retain Eq. (24), the sign of the headline result reverses for Γ_s > Γ, which would be much more serious. I recommend asking them to trace the sign explicitly through Eqs. (26)-(28) and to state the convention prominently. The self-consistent equations should also be made checkable; the current presentation is a barrier to verification."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead the arXiv paper 1908.03365. The new piece is a finite-U slave-boson mean-field treatment of a quantum dot attached to two ferromagnetic leads and one superconducting lead, focused on the linear cross conductance in the Kondo regime. The parameter maps – Γ_s vs ε_d for parallel and antiparallel configurations – are new and presented cleanly. The extension beyond the master-equation work of Futterer et al. and the normal-lead work of Michalek et al. is reasonable, and the citations to prior work look right.\n\nThe main problem is the sign convention for the right-lead current. Eq. (24) defines I_η as the current from lead η into the QD. Exchanging L and R in Eqs. (26)-(28) gives, for positive left bias at Γ_s=0, a negative I_R (electrons flow from the dot into the right lead), so dI_R/dV = -GET and the cross conductance under that definition is GC = -GET + GCAR. The paper instead gives Eq. (30) as GC = GET - GCAR and plots GC = +G0 at Γ_s=0. Both cannot be right. The interpretation that CAR makes the cross conductance negative depends entirely on which convention is meant. If the intended I_R is the current from the QD into the right lead, then Eq. (30) is correct but Eq. (24) is wrong. This is a load-bearing inconsistency in the central claim, not a cosmetic issue.\n\nSecondary issues: the self-consistent equations are delegated to a 'lengthy and tedious calculation' and the printed forms appear to contain index slips in Eqs. (18)-(19). The SBMF is uncontrolled in the presence of both pairing and ferromagnetism, so quantitative sizes like GC ≈ -G0/5 should be taken as indicative. Still, the qualitative trends could survive a more exact treatment.\n\nThe paper is written in a standard style, the numerics are reproducible in principle, and the physics is relevant to Cooper pair splitting experiments.\n\nMy bottom line: this deserves a round of peer review, but only on the condition that the authors clarify the current sign convention and fix the definition/formula mismatch. Without that fix, the headline result cannot be interpreted. A serious referee would catch this immediately.","headline":"Useful but conventional SBMF numerics for a three-terminal dot; the headline negative cross conductance is currently ambiguous because the paper's current definition contradicts its own formula and plots.","tokens_in":13484,"tokens_out":5990,"would_cite":false,"duration_ms":62507,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["74.70.-b","74.45.+c","72.15.Qm","73.23.Hk","75.47.-m"],"model":"deepseek-v4-flash","headline":"The paper claims that strong superconducting coupling makes the nonlocal conductance negative in a Kondo quantum dot, with antiparallel ferromagnetic leads deepening the effect.","keywords":["quantum dot","Kondo effect","crossed Andreev reflection","nonlocal conductance","ferromagnetic leads","superconducting proximity effect","slave-boson mean-field","three-terminal transport"],"falsifier":"A numerically exact treatment of the same model (beyond the slave-boson mean-field level) or a low-temperature measurement on a tunable three-terminal quantum dot should check whether $G_C$ stays negative for $\\Gamma_s > \\Gamma$ in the Kondo plateau, whether $G_C$ passes through zero at $\\Gamma_s \\approx \\Gamma$ at particle-hole symmetry, and whether the antiparallel polarization $p=0.5$ indeed deepens $G_C$ to $\\approx -G_0/5$; any of these failing would falsify the mean-field prediction.","tokens_in":12513,"feed_emoji":"⚛️","tokens_out":7569,"duration_ms":64286,"temperature":0.7,"pith_summary":"This paper claims that in a quantum dot coupled to two ferromagnetic leads and one superconducting lead, the nonlocal (cross) conductance becomes negative in the Kondo regime once the superconducting proximity coupling $\\Gamma_s$ exceeds the normal tunnel coupling $\\Gamma$. The negative value arises because crossed Andreev reflection, which transfers a Cooper pair from the dot into the superconductor while reflecting a hole into the opposite lead, opposes the normal single-particle current at the right contact. In the antiparallel magnetization configuration, increasing spin polarization suppresses the local conductance but makes the cross conductance more negative, reaching $G_C \\simeq -G_0/5$ at polarization $p=0.5$; in the parallel configuration, polarization instead destroys the crossed Andreev channel and the cross conductance returns toward the normal positive value. The result matters because a negative nonlocal conductance is a direct transport signature of Cooper-pair splitting in a three-terminal device.","feed_headline":"Kondo dot with superconductor: nonlocal conductance goes negative","feed_subtitle":"Antiparallel magnetizations deepen the negative signal, a fingerprint of crossed Andreev reflection.","key_machinery":"The central object is the decomposition of the current into three subgap tunneling processes: normal electron transfer ($I_{\\rm ET}$), direct Andreev reflection ($I_{\\rm DAR}$), and crossed Andreev reflection ($I_{\\rm CAR}$). The linear cross conductance formula $G_C = G_{\\rm ET} - G_{\\rm CAR}$ is the load-bearing identity, because it shows the same crossed-Andreev amplitude that adds to the local current subtracts from the nonlocal response. These quantities are evaluated with renormalized Green functions obtained from the finite-$U$ slave-boson mean-field equations, where the proximity pairing appears as an effective $\\Gamma_s$ term mixing empty and doubly occupied dot states into Andreev bound states. The competition between $\\Gamma_s$ and the renormalized tunnel couplings controls whether $G_{\\rm CAR}$ overtakes $G_{\\rm ET}$ and thereby flips the sign of $G_C$.","core_discovery":"Working with a finite-$U$ slave-boson mean-field description and nonequilibrium Green functions, the paper shows that the linear cross conductance can be written as $G_C = G_{\\rm ET} - G_{\\rm CAR}$, while the local conductance is $G_L = G_{\\rm ET} + G_{\\rm DAR} + G_{\\rm CAR}$. Since the crossed Andreev reflection (CAR) hole current in the right lead flows opposite to the electron transfer current, the cross conductance turns negative when $G_{\\rm CAR} > G_{\\rm ET}$; this happens for $\\Gamma_s > \\Gamma$ inside the Kondo plateau. At the particle-hole symmetric point $\\epsilon_d = -U/2$ and $\\Gamma_s = \\Gamma$, the calculation gives $G_{\\rm DAR} = G_0/2$, $G_{\\rm CAR} = G_{\\rm ET} = G_0/4$, so the local conductance returns to $G_0$ while the cross conductance vanishes. In the antiparallel configuration the spin-dependent availability of up and down electrons makes the CAR channel more efficient as polarization grows, producing $G_C \\simeq -G_0/5$ at $p=0.5$, and in the fully polarized limit the paper argues only CAR survives, leaving $G_L = -G_C = G_{\\rm CAR} = G_0/2$. In the parallel configuration, polarization splits the Kondo peak and suppresses CAR, so $G_C$ stays weakly negative and the local conductance develops a four-peak structure.","pith_inferences":["A natural extension is to read the sign and magnitude of $G_C$ as a direct measure of the renormalized crossed-Andreev amplitude, so the same setup could be used as a tunable detector of nonlocal pairing correlations in quantum-dot devices.","The mean-field prediction at $p \\to 1$ in the antiparallel configuration suggests a symmetry-protected regime in which the device acts as a pure Cooper-pair splitter with equal and opposite local and nonlocal conductances; verifying this limit beyond the slave-boson approximation would test whether the mechanism survives strong correlations.","The same decomposition $G_C = G_{\\rm ET} - G_{\\rm CAR}$ should apply more generally to any three-terminal hybrid with proximity pairing, so the qualitative sign-change criterion $\\Gamma_s > \\Gamma$ is a testable design rule for experiments and for other theoretical approaches."],"forward_implications":["If the central claim is right, a negative linear cross conductance in the Kondo regime is a clean experimental marker for crossed Andreev reflection, distinguishing it from sequential-tunneling mechanisms that also produce negative nonlocal responses at higher bias.","In the antiparallel configuration the predicted polarization dependence gives a direct handle: raising $p$ toward unity should drive the device toward a pure crossed-Andreev state with $G_L = -G_C = G_0/2$.","The vanishing of $G_C$ and the return of $G_L$ to $G_0$ at $\\Gamma_s = \\Gamma$ and $\\epsilon_d = -U/2$ provides a specific quantitative check that could be tested by tuning gate voltage and tunnel barriers.","Because the bias voltage destroys the negative response, finite-bias differential-conductance measurements should show the negative region only within the Kondo window, with recovery at large $V$.","In the parallel configuration, the predicted drop of the cross conductance back toward the positive normal value quantifies how strongly ferromagnetism in parallel alignment suppresses proximity-induced pairing."],"supporting_citations":[{"why":"Supplies the three-terminal model Hamiltonian and the prior master-equation conclusion that strong on-dot interactions, not crossed Andreev reflection, dominate the negative nonlocal response at high bias; the present work directly extends this setup to the low-bias Kondo regime.","marker":"[38]"},{"why":"Establishes that crossed Andreev reflection is the dominant nonlocal subgap channel at low bias in two normal leads, yielding the negative cross-conductance mechanism whose formulas are adapted here.","marker":"[41]"},{"why":"Extends the crossed-Andreev analysis to a similar three-terminal geometry and provides the current formulas (ET, DAR, CAR) on which the present decomposition is based.","marker":"[42]"},{"why":"Introduces the finite-$U$ slave-boson representation and mean-field constraints used to treat the on-dot Coulomb interaction.","marker":"[43]"},{"why":"Supplies the slave-boson mean-field plus nonequilibrium Green function machinery for Kondo transport through a single quantum dot.","marker":"[44]"},{"why":"Documents the finite-$U$ slave-boson approach for conductance over mixed-valence to empty-orbital regimes, supporting the method's validity.","marker":"[45]"},{"why":"Applies the same approach to spin-polarized transport, providing the renormalization equations used for ferromagnetic leads.","marker":"[46]"},{"why":"Provides the multi-level formulation of the slave-boson approach used for the self-consistent equations.","marker":"[47]"},{"why":"Reports the experimental observation of crossed Andreev reflection and Cooper-pair splitting that motivates the nonlocal conductance sign as a probe.","marker":"[31]"}],"fun_headline_variants":["Crossed Andreev reflection flips conductance sign in Kondo dot","Antiparallel magnets deepen negative cross conductance in QD","Kondo dot with superconducting lead: CAR beats ET, conductance negative","Negative nonlocal conductance from crossed Andreev reflection in QD","CAR dominates ET: Kondo dot cross conductance goes negative"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the slave-boson mean-field approximation, which replaces all boson operators by their expectation values, remains quantitatively reliable for the Kondo regime when superconducting pairing and ferromagnetic leads are both present; if strong correlations renormalize the pairing or the Kondo resonance beyond this mean-field level, the predicted values, and possibly the sign, of the cross conductance could change.","fun_headline_variants_meta":{"raw":{"variants":["Crossed Andreev reflection flips conductance sign in Kondo dot","Antiparallel magnets deepen negative cross conductance in QD","Kondo dot with superconducting lead: CAR beats ET, conductance negative","Negative nonlocal conductance from crossed Andreev reflection in QD","CAR dominates ET: Kondo dot cross conductance goes negative"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000421,"raw_usage":{"total_tokens":2234,"prompt_tokens":1085,"completion_tokens":1149,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":701,"completion_tokens_details":{"reasoning_tokens":1060}},"tokens_in":701,"tokens_out":1149,"duration_ms":8580,"temperature":1.0,"reasoning_tokens":1060,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:15:07.598523+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A numerically exact treatment of the same model (beyond the slave-boson mean-field level) or a low-temperature measurement on a tunable three-terminal quantum dot should check whether $G_C$ stays negative for $\\Gamma_s > \\Gamma$ in the Kondo plateau, whether $G_C$ passes through zero at $\\Gamma_s \\approx \\Gamma$ at particle-hole symmetry, and whether the antiparallel polarization $p=0.5$ indeed deepens $G_C$ to $\\approx -G_0/5$; any of these failing would falsify the mean-field prediction.","supporting_citations":[{"cited_title":"Futterer , author M","cited_arxiv_id":null,"evidence_quote":"Supplies the three-terminal model Hamiltonian and the prior master-equation conclusion that strong on-dot interactions, not crossed Andreev reflection, dominate the negative nonlocal response at high bias; the present work directly extends this setup to the low-bias Kondo regime."},{"cited_title":"Micha ek , author B","cited_arxiv_id":null,"evidence_quote":"Establishes that crossed Andreev reflection is the dominant nonlocal subgap channel at low bias in two normal leads, yielding the negative cross-conductance mechanism whose formulas are adapted here."},{"cited_title":"Micha ek , author T","cited_arxiv_id":null,"evidence_quote":"Extends the crossed-Andreev analysis to a similar three-terminal geometry and provides the current formulas (ET, DAR, CAR) on which the present decomposition is based."},{"cited_title":"Dong and author X","cited_arxiv_id":null,"evidence_quote":"Supplies the slave-boson mean-field plus nonequilibrium Green function machinery for Kondo transport through a single quantum dot."},{"cited_title":"Dong and author X","cited_arxiv_id":null,"evidence_quote":"Documents the finite-$U$ slave-boson approach for conductance over mixed-valence to empty-orbital regimes, supporting the method's validity."},{"cited_title":"Dong and author X","cited_arxiv_id":null,"evidence_quote":"Applies the same approach to spin-polarized transport, providing the renormalization equations used for ferromagnetic leads."},{"cited_title":"Ma , author B","cited_arxiv_id":null,"evidence_quote":"Provides the multi-level formulation of the slave-boson approach used for the self-consistent equations."},{"cited_title":"Beckmann , author H","cited_arxiv_id":null,"evidence_quote":"Reports the experimental observation of crossed Andreev reflection and Cooper-pair splitting that motivates the nonlocal conductance sign as a probe."}],"review_version":1}