{"id":"d7038433-782d-4eb6-ac19-a27b135d386a","arxiv_id":"2505.01905","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Hybrid-MBE-grown SrRuO3/BaTiO3/SrRuO3 capacitors show ferroelectric hysteresis with a remnant polarization of about 15 uC/cm2 measured by PUND from 500 Hz to 20 kHz.","lead":"This paper reports the first direct measurement of remnant polarization in BaTiO3 films grown by hybrid molecular beam epitaxy, using a SrRuO3/BaTiO3/SrRuO3 capacitor stack. The measured remnant polarization of about 15 uC/cm2 across 500 Hz to 20 kHz confirms that MBE-grown BaTiO3 can switch ferroelectricity, supporting a scalable route to lead-free ferroelectric devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"PUND subtraction may leave a residual, polarization-state-dependent leakage in the N/D branch, so the 15 uC/cm2 Pr is not yet fully isolated from non-ferroelectric currents.","rationale":"The reader's weakest assumption identifies exactly the load-bearing point: PUND subtraction is only as good as the equality of non-ferroelectric currents between switching and non-switching pulses. I agree with that identification. I would tighten the wording to emphasize that the risk is not merely voltage-dependent leakage, but leakage that differs between the two polarization states encountered in P and U (or N and D), plus any slow leakage transient with a time constant exceeding 10 us. The manuscript's own Figure 3g shows strongly asymmetric leakage and resistance degradation under negative bias, which is the branch most vulnerable to this artifact. The reported frequency independence of Pr and the near-ideal capacitive phase up to 40 kHz are supportive, but they do not quantify leakage-current mismatch integrated over the full pulse. Because the paper's novelty is precisely a first quantitative Pr for MBE-grown BaTiO3, the claim should be conditional on a null test: a subcoercive PUND measurement on the same devices, or an equivalent non-ferroelectric reference capacitor, would distinguish a genuine ferroelectric response from a leakage-subtraction artifact. If that test passes, the ACCEPT verdict is justified; if it fails, the central claim would need substantial revision. Thus the appropriate verdict is CONDITIONAL rather than a flat rejection, since no concrete error in the reported data has been demonstrated.","tokens_in":16837,"tokens_out":11725,"duration_ms":139587,"concrete_test":"Perform the identical PUND pulse train and analysis on the same devices with the maximum pulse voltage reduced below the coercive voltage (e.g., V_max ~ 0.3 V, well below E_c*d ~ 0.5 V indicated by Figure 3e), so no ferroelectric switching should occur. If the same P-U/N-D subtraction and integration yields an apparent Pr larger than the integration noise floor (say >1 uC/cm2), then non-ferroelectric currents are not cancelled by the subtraction and the 15 uC/cm2 claim is not isolated. A positive control at V_max above coercive should still reproduce ~15 uC/cm2.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim rests entirely on the PUND subtraction in Section 4.4 and Figure 3a-c: Pr is obtained by integrating (P-U) and (N-D), which is valid only if the U and D pulses reproduce the same non-ferroelectric currents that occur during the switching P and N pulses. The paper itself reports in Section 2.3 and Figure 3g that the static I-V shows 'a greater leakage current and resistance degradation under negative bias.' This is precisely the branch where N-D subtraction is used. If the leakage current depends on instantaneous polarization state (ferroelectric-diode behavior) or has a slow component with time constant longer than the 10 us delay, then the second pulse does not reproduce the first, and the residual integrated current contributes directly to the reported 15 uC/cm2. The stated checks (t_d=t_w=10 us and no back-switching, Figures S2-S3) address transient decay and imprint back-switching, not state-dependent leakage; impedance phase near -90 deg up to 40 kHz reduces but does not eliminate this concern because PUND integrates any residual current over the whole pulse. No null measurement on a non-switching control is reported, so the subtraction is uncalibrated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports the hybrid molecular beam epitaxy (MBE) growth of an all-epitaxial SrRuO3 (16 nm)/BaTiO3 (40 nm)/SrRuO3 (16 nm) heterostructure on Nb-doped SrTiO3 (001) substrates and its ferroelectric characterization. The central claim is that Positive-Up-Negative-Down (PUND) measurements yield hysteretic polarization-electric field loops with a remnant polarization Pr ~ 15 uC/cm2 that is stable from 500 Hz to 20 kHz, which the authors present as the first direct measurement of remnant polarization in MBE-grown BaTiO3 films. Supporting evidence includes phase-pure epitaxial X-ray diffraction, reciprocal space mapping showing a partially relaxed BTO layer with tetragonality of about 1.022, impedance spectroscopy showing near-ideal capacitive behavior up to about 40 kHz, and a benchmarking comparison with PLD- and sputtered-grown films. The paper also proposes defect-dipole origins for the observed switching asymmetry and leakage.","tokens_in":17010,"tokens_out":5700,"duration_ms":51276,"significance":"If the central claim holds, this is an important step for MBE-grown ferroelectrics: it provides the missing electrical verification of ferroelectric switching in a synthesis technique that is otherwise well established for epitaxial oxide heterostructures, and it demonstrates a practical hybrid MBE route with adsorption-controlled growth of both the electrodes and the ferroelectric layer. The paper is commendable for reporting averaged results over four devices with standard deviations, giving complete PUND timing parameters, and being transparent about the unresolved top-electrode peak and the leaky negative-bias branch. The structural and dielectric data (XRD, RSM, impedance phase near -90 degrees) are consistent with a high-quality ferroelectric capacitor. However, the electrical claim rests on the validity of the PUND subtraction, which warrants careful scrutiny.","major_comments":[{"comment":"The PUND subtraction of the 'Up' and 'Down' currents from the 'Positive' and 'Negative' currents is valid only if the non-ferroelectric current is identical in the switching and non-switching pulses. Figure 3g shows a strong leakage current and resistance degradation under negative bias, and the N-D branch is precisely that bias region. The stated checks (t_d = t_w = 10 us and no back-switching, Supplementary Figures S2-S3) address transient decay and imprint-driven back-switching, but not polarization-state-dependent leakage (e.g., a ferroelectric diode effect) or leakage components with time constants longer than 10 us that would not fully decay during the delay. Because no control measurement on a non-switching or paraelectric capacitor is reported, the residual non-ferroelectric charge integrated into the PUND loops is uncalibrated and could bias the reported 15 uC/cm2. This is a load-bearing issue for the paper's central claim.","section":"Section 4.4, Figure 3c and 3g"},{"comment":"The sentence 'We shift the polarization values such that the magnitudes of the saturation polarization at both peak voltages are similar' describes an unspecified adjustment of the integration constant. If this shift is chosen to force matching of the saturation magnitudes, it can impose a symmetry on the P-V loop that partly predetermines the extracted Pr and imprint voltage. Moreover, the equations defining Pr, Ec, and V_offset are garbled in the manuscript, so it is unclear whether Pr is computed from the difference of the two remanent values (which would be independent of a constant shift) or from the shifted zero-field intercept (which would not). The authors should state the exact definitions and report the magnitude of the shift; without this, the reported Pr is not uniquely defined.","section":"Section 4.4, paragraph defining Pr, Ec, and V_offset"}],"minor_comments":[{"comment":"The word 'thaat' appears in the sentence 'We conclude thaat growth on the more closely lattice-matched scandate substrates...' and should be corrected to 'that'.","section":"Section 2.3, typos"},{"comment":"The abstract states Pr ~ 15 uC/cm2, while Section 2.3 reports 'Pr ~ 15.2 uC/cm2'; these values should be made consistent.","section":"Abstract and Section 2.3"},{"comment":"The equivalence between the PUND pulse train and a four-quadrant triangular sweep is stated as f = 1/[2(t_r + t_f)], but the actual waveform includes wait times t_w = 10 us, making the pulses rectangular. The authors should clarify how the frequency f is defined for the rectangular PUND train and how this affects the interpretation of the frequency-dependent data in Figure 3e.","section":"Section 4.4, pulse-train definition"},{"comment":"The measurement temperature is not explicitly stated; the authors should confirm that all electrical measurements were performed at room temperature.","section":"General, measurement conditions"},{"comment":"Several supporting claims (Figures S2-S3, Tables S2-S5) are made with reference to supplementary material; if the supplementary information is not part of the review package, the authors should ensure the main text is self-contained or provide the supplementary material.","section":"Supplementary Information"}],"recommendation":"major_revision","confidential_remarks":"The authors cite several of their own previous papers on MBE growth windows, but these are not used to support the ferroelectric-switching result itself, so I see no circularity concern. The main risk is the PUND subtraction: the paper's own static I-V data show strong asymmetry under negative bias, and the subtraction assumes state-independent leakage with a fast transient. This warrants a major revision, ideally with an additional control experiment or a quantitative leakage assessment. The paper is otherwise well-organized and clearly reported, and the structural and dielectric characterization is solid."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new thing here is the claim itself: a direct remanent polarization of about 15 uC/cm2 in an all-epitaxial SrRuO3/BaTiO3/SrRuO3 capacitor grown by hybrid MBE. If that claim holds, it fills a real gap: MBE-grown BTO has been characterized by PFM and electro-optics, but not by direct polarization hysteresis. The paper does solid work. It gives explicit PUND timing parameters, shows data from four devices with standard deviations, verifies phase-pure epitaxy by XRD, and uses RSM to document partial relaxation with a tetragonality of 1.022. The benchmarking against PLD and sputtered films is useful, and the authors are honest about unresolved features, notably the top SrRuO3 peak. The main soft spot is exactly what the stress-test note flags. The PUND subtraction assumes the Up and Down pulses reproduce the non-ferroelectric currents during the switching Positive and Negative pulses. The paper itself reports greater leakage and resistance degradation under negative bias (Figure 3g), which is the branch where the N-D subtraction matters. With only 10 us delay and wait, slow or state-dependent leakage could leave a residual that integrates into the reported Pr. The authors do not report a null measurement on a non-switching control or a leakage-compensated cross-check. That is a legitimate concern, but I would not call it a load-bearing flaw. The normalized switching peaks in Figure 3c are clear and look like ferroelectric switching, not just leakage; the impedance data support a capacitive response up to 40 kHz; and the claimed Pr is plausible for a 40 nm partially strained BTO film. The absence of a resolved top SRO peak is a minor structural gap, and the saturation-matching shift used in the integration is not fully justified, but these do not undermine the central result. For the ferroelectric and oxide MBE communities, this is a useful letter: it establishes that hybrid MBE can produce a working ferroelectric capacitor, and it gives a clear recipe for others to try. I would cite it if I were working on BTO thin films. It deserves a serious referee who can pressure-test the PUND analysis and ask for the additional control measurements; that is a normal part of the process. The verdict from the reading group, accept with moderate confidence, seems right to me. The paper is not overhyped and the claims are calibrated to what is shown.","headline":"First direct remnant polarization in MBE-grown BaTiO3 is reported with careful PUND methodology; the PUND subtraction is the main caveat, but the paper deserves a serious referee.","tokens_in":735,"tokens_out":916,"would_cite":true,"duration_ms":27884,"reading_group":"yes","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 reports the first direct measurement of remnant polarization in MBE-grown BaTiO3 films, using an all-epitaxial SrRuO3/BaTiO3/SrRuO3 capacitor grown by hybrid MBE.","keywords":["BaTiO3","SrRuO3","hybrid molecular beam epitaxy","remnant polarization","PUND","ferroelectric switching","epitaxial heterostructure","lead-free ferroelectric"],"falsifier":"Repeat the PUND measurement on the same devices with delay and wait times of 1, 10, and 100 µs while keeping rise and fall times fixed; if the extracted remnant polarization changes by more than the device-to-device scatter, the 10 µs subtraction does not fully separate ferroelectric from non-ferroelectric charge. Independently, measure hysteresis above the BaTiO3 Curie temperature: a genuine ferroelectric loop should collapse, while leakage artifacts would survive.","tokens_in":16629,"feed_emoji":"⚡","tokens_out":9005,"duration_ms":82364,"temperature":0.7,"pith_summary":"The paper reports an all-epitaxial capacitor, 16 nm SrRuO3 / 40 nm BaTiO3 / 16 nm SrRuO3, grown on Nb-doped SrTiO3 (001) by hybrid molecular beam epitaxy with metal-organic precursors. It claims that Positive-Up-Negative-Down (PUND) measurements isolate a genuine ferroelectric switching current, yielding hysteretic polarization–electric-field loops with a remnant polarization of about 15 µC/cm² at frequencies from 500 Hz to 20 kHz. This is presented as the first direct measurement of remnant polarization in MBE-grown BaTiO3, a synthesis route previously limited to indirect evidence such as piezoresponse force microscopy and electro-optic hysteresis. If correct, the result establishes hybrid MBE as a practical route to ferroelectric devices, not just to high-quality epitaxial films.","feed_headline":"First direct remnant polarization measured in MBE-grown BaTiO3","feed_subtitle":"All-epitaxial SrRuO3/BaTiO3/SrRuO3 capacitors show a remnant polarization near 15 µC/cm² from 500 Hz to 20 kHz.","key_machinery":"The load-bearing object is the PUND pulse train, a five-pulse sequence (poling, Positive, Up, Negative, Down) in which the Positive and Negative pulses contain both ferroelectric switching and non-ferroelectric currents while the Up and Down pulses, arriving after a 10 µs delay, are assumed to contain only the non-ferroelectric parts. Subtracting Up from Positive and Down from Negative isolates the switching current without assuming a functional form for leakage or capacitance. The second piece of machinery is hybrid MBE growth of both SrRuO3 electrodes and the BaTiO3 layer using Ru(acac)3 and titanium tetraisopropoxide precursors, which the paper argues enables adsorption-controlled, stoichiometric growth with low defect densities.","core_discovery":"The central claim is that an all-epitaxial SrRuO3/BaTiO3/SrRuO3 heterostructure grown by hybrid MBE on Nb:SrTiO3 (001) shows true ferroelectric switching. Using the PUND pulse train, which compares switching pulses (Positive and Negative) with non-switching pulses (Up and Down), the authors subtract leakage and capacitive currents and integrate the remaining current to obtain polarization–voltage hysteresis. They report a remnant polarization of about 15 µC/cm² that stays roughly constant between 500 Hz and 20 kHz, with coercive field and imprint voltage increasing with frequency. The asymmetry between up-to-down and down-to-up switching is attributed to structurally and chemically dissimilar electrodes, and the voltage-dependent leakage is hypothesized to come from gradients of defect dipoles, most likely oxygen or barium vacancies.","pith_inferences":["A natural test of the subtraction logic is to vary the PUND delay and wait times from 1 µs to 100 µs; if the extracted remnant polarization drifts, the 10 µs choice is not neutral and the quoted number would need revision.","Because the paper's own static I–V data show strongly voltage-dependent leakage under negative bias, an independent method such as switched-charge measurement or temperature-dependent hysteresis could determine whether the cancellation is biased.","If the defect-dipole-gradient hypothesis is correct, oxygen annealing under an applied bias or at different temperatures should change the imprint voltage and leakage asymmetry in a predictable, testable way.","The same all-epitaxial platform could be extended to pyroelectric energy harvesting studies, since bulk BaTiO3's sharp Curie transition is already of interest and a direct remnant polarization gives a baseline for switchable charge."],"forward_implications":["If correct, MBE-grown BaTiO3 joins the list of epitaxial ferroelectric films with directly verified non-zero remnant polarization, closing a long-standing gap between MBE growth studies and electrical characterization.","The reported value, about 15 µC/cm², is roughly five times the saturation polarization of the only earlier MBE-grown BaTiO3 film that showed P–E hysteresis, suggesting that electrode engineering rather than the growth technique was the main obstacle.","Because the remnant polarization is flat from 500 Hz to 20 kHz, the switched charge is not a low-frequency artifact and could persist at frequencies relevant to memory and capacitor applications.","The frequency-dependent coercive field and imprint voltage imply that switching from the up to the down state becomes progressively harder at higher frequencies, so device designs must budget for asymmetric switching kinetics.","The benchmark comparison indicates that scandate substrates and smaller capacitor areas are concrete levers that could raise the remnant polarization toward the values reported for PLD- and sputter-grown films."],"supporting_citations":[{"why":"Supplies the PUND measurement protocol that the paper relies on to separate ferroelectric switching current from non-ferroelectric contributions.","marker":"[96]"},{"why":"The only prior MBE-grown BaTiO3 film with hysteretic P–E curves but no reported remnant polarization; provides the baseline this work exceeds.","marker":"[37]"},{"why":"Shows that SrRuO3 electrodes and depolarization-field minimization enable low-voltage switching, motivating the all-epitaxial electrode choice.","marker":"[25]"},{"why":"Demonstrates strain-enhanced remnant polarization in BaTiO3 and provides a benchmark for coercive-field trade-offs.","marker":"[18]"},{"why":"Establishes hybrid MBE growth of SrRuO3 using the Ru(acac)3 precursor, the method used for the electrodes.","marker":"[89]"},{"why":"Provides the optimized adsorption-controlled growth window for SrRuO3 that the heterostructure growth relies on.","marker":"[90]"},{"why":"Supplies the hybrid MBE growth window for BaTiO3 using titanium tetraisopropoxide, the method used for the ferroelectric layer.","marker":"[91]"},{"why":"Identifies oxygen vacancies as negative charge traps that impede switching, supporting the defect-dipole interpretation of leakage and asymmetry.","marker":"[102]"}],"fun_headline_variants":["First direct ferroelectric switching in MBE-grown BaTiO3","MBE BaTiO3 shows real remnant polarization via PUND","Hybrid MBE BaTiO3: Pr = 15 µC/cm² confirmed","Ferroelectric switching proven in MBE-grown BaTiO3"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole measurement rests on the assumption that, after a 10 microsecond delay, the Up and Down pulses carry exactly the same non-ferroelectric current as the Positive and Negative pulses; the paper's own I–V data (Figure 3g) show leakage that changes steeply with voltage and bias direction, so the cancellation could be incomplete.","fun_headline_variants_meta":{"raw":{"variants":["First direct ferroelectric switching in MBE-grown BaTiO3","MBE BaTiO3 shows real remnant polarization via PUND","Hybrid MBE BaTiO3: Pr = 15 µC/cm² confirmed","Ferroelectric switching proven in MBE-grown BaTiO3"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000156,"raw_usage":{"total_tokens":1204,"prompt_tokens":917,"completion_tokens":287,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":533,"completion_tokens_details":{"reasoning_tokens":210}},"tokens_in":533,"tokens_out":287,"duration_ms":3168,"temperature":1.0,"reasoning_tokens":210,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:06:48.797457+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the PUND measurement on the same devices with delay and wait times of 1, 10, and 100 µs while keeping rise and fall times fixed; if the extracted remnant polarization changes by more than the device-to-device scatter, the 10 µs subtraction does not fully separate ferroelectric from non-ferroelectric charge. Independently, measure hysteresis above the BaTiO3 Curie temperature: a genuine ferroelectric loop should collapse, while leakage artifacts would survive.","supporting_citations":[],"review_version":1}