{"id":"fb6d1e53-990a-4956-ba28-6fd324cb2f79","arxiv_id":"2502.02117","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A moving-average detrending filter with an adaptive mask size estimates RTN trap time constants from unstable filament signals, but its real-data validation is partly circular.","lead":"Researchers present an adaptive moving-average filter that flattens unstable random telegraph noise in silicon nitride resistive memory cells. The method recovers trap capture and emission time constants from drifting signals that standard time-domain analysis cannot handle.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Validation is circular: the optimization criterion (Eq. 4) includes the Lorentzian relation used as the confirmation test, so the claimed cross-validation does not independently establish accuracy.","rationale":"The reader's weakest assumption is the low-frequency separation between instability and RTN. That assumption is plausible but the paper gives some spectral evidence (Fig. 5b, lower panel) that removed components lie below f0. The circularity of the validation is more decisive: term B of the optimization criterion is the same Lorentzian relation used as the confirmation test, so the reported agreement in Fig. 5(f) is forced by the objective, not earned by measurement accuracy. This undermines the central claim's experimental support independent of whether the frequency-separation premise holds. The synthetic experiment does provide some internal validation, but the headline claim about real devices relies on the circular confirmation. The verdict remains CONDITIONAL because the method is plausible and could be validated with independent tests, but the current evidence is insufficient to establish accuracy.","tokens_in":9539,"tokens_out":4376,"duration_ms":41810,"concrete_test":"Re-run the adaptive filter on the Fig. 5(a) experimental signal with b=0 (term B removed from Eq. (4)), leaving only term A = (σc+σe)/|μc-μe| as the criterion; fit the resulting dwell times and compute f0' = (1/2π)(1/τc+1/τe). If f0' matches the Lorentzian f0 = 985 Hz within its uncertainty, the match is robust; if f0' deviates substantially, the reported agreement in Fig. 5(f) is an artifact of the objective function.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The claimed cross-validation in Section 3 is not independent. In Eq. (4), the objective C contains term B = |f0 - (1/2π)(1/τc+1/τe)|/N, where f0 is obtained by Lorentzian fitting of the PSD of the same unstable signal x(k). The confirmation in Fig. 5(f) then checks exactly this relation, Eq. (5). Since the optimizer explicitly searches for τc and τe that minimize |f0 - (1/2π)(1/τc+1/τe)|, agreement between the estimated time constants and Eq. (5) is a constrained outcome, not a verification of accuracy. The method may still work, and the synthetic experiment with known ground truth is helpful, but the experimental evidence for the central claim is currently a self-consistency check rather than a cross-validation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes an adaptive moving-average detrending filter to flatten unstable random telegraph noise (RTN) signals in silicon nitride ReRAM devices, where conductive-filament instabilities cause slow drift that obscures two-level switching. The filter optimizes the moving-average mask size by minimizing a composite criterion C (Eq. 4) that includes separation of histogram lobes (term A) and consistency with the RTN Lorentzian corner frequency f0 (term B). The method is validated on a simulated RTN signal with injected 1/f^3 instability and on one measured unstable RTN signal; the authors report that the estimated capture/emission time constants satisfy the Lorentzian relation f0 = (1/2π)(1/τc + 1/τe).","tokens_in":9735,"tokens_out":5439,"duration_ms":46351,"significance":"The problem addressed is real: RTN analysis in ReRAM multi-level cells is often corrupted by slow filament instabilities, and standard threshold-based time-domain analysis fails when levels drift. The synthetic validation with known ground truth is a strength; the paper shows that the optimizer recovers injected time constants and that the adaptive filter produces a histogram with separable Gaussian lobes. However, the experimental validation is weakened by circularity: term B of the optimization criterion (Eq. 4) directly enforces the Lorentzian relation that is later used as the 'confirmation criterion' (Eq. 5 and Fig. 5(f)), so the agreement is a constrained outcome rather than an independent cross-check. The method may well be useful, but the current experimental evidence is a self-consistency check, not a true cross-validation.","major_comments":[{"comment":"The experimental 'cross-validation' is not independent. The optimization criterion C in Eq. (4) includes term B = |f0 - (1/2π)(1/τc+1/τe)|/N, with f0 obtained from Lorentzian fitting of the PSD of the same unstable signal. Section 3 then confirms the estimated τc and τe by checking Eq. (5), which is exactly the relation minimized in term B. Consequently, Fig. 5(f) does not provide independent evidence that the method is accurate; it merely shows that the optimizer satisfied its own constraint. To support the claimed cross-validation, the authors should either (i) compare the time-domain estimates with an independently fitted Lorentzian to the stabilized signal's PSD, or (ii) use an alternative independent estimation method, or (iii) explicitly rephrase the result as a self-consistency check rather than a proof of accuracy.","section":"Section 3, Eq. (4)-(5), Fig. 5(f)"},{"comment":"The method rests on a timescale-separation assumption: the instability must have a corner frequency f0' at least an order of magnitude below the RTN corner frequency f0 and a power at least three orders of magnitude lower. This condition is verified for the simulated signal by construction, but for the measured device the paper does not quantitatively verify that the experimental drift satisfies these bounds. The bottom panel of Fig. 5(b) shows that subtracted spectral content lies below f0, but no power-ratio quantification is provided. Since the moving-average filter is not a sharp high-pass filter, partial overlap between the instability band and the RTN Lorentzian band would bias the estimated τc and τe. The authors should quantify the instability contribution (e.g., from the PSD of the subtracted signal) and show that the separation condition holds for the measured data, or state the limitation explicitly.","section":"Section 2.2"},{"comment":"The paper does not report quantitative accuracy metrics for the time-constant estimates. In the synthetic experiment (Fig. 4), the authors state that the time constants are 'calculated accurately,' but no numerical comparison with the injected values (e.g., relative errors or confidence bounds) is given. For the experimental signal (Fig. 5), no error bars or comparison with any independent estimate are provided. Given that the central claim is that the method increases accuracy, quantitative error reporting on the simulated ground truth is essential and would also make the synthetic validation more convincing.","section":"Section 3 and Fig. 4"}],"minor_comments":[{"comment":"The abstract contains typographical errors: 'tunning protocol' should be 'tuning protocol,' and 'The te and tc emission/capture time constants' should use τe and τc.","section":"Abstract"},{"comment":"The scaling parameters are introduced as 'α and b' in the text but written as 'a' and 'b' in Eq. (4); please unify the notation.","section":"Eq. (4)"},{"comment":"The notation '1/f 2' and '1/f 3' in the text and Fig. 2 should be typeset as superscripts (1/f^2, 1/f^3) to avoid confusion.","section":"Section 2.2"},{"comment":"There are grammatical errors in the sentence 'The presence of RTN is the present SiNx RRAM MIS devices have been already investigated and demonstrated' (sic); please rephrase.","section":"Section 2.1"},{"comment":"The caption of Fig. 5(b) says 'Lower figure of (b) validates that all subtracted frequencies were below corner frequency f0,' but it is helpful to state explicitly how the subtracted signal was computed and what metric is shown.","section":"Fig. 5"}],"recommendation":"major_revision","confidential_remarks":"The circularity issue is the main concern; if the authors can reframe the experimental evidence or add an independent check, the paper could be suitable. The synthetic validation is a positive aspect. The paper is within the scope of the journal as an applied physics/methods contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper offers a practical fix for a real nuisance: unstable RTN traces that drift and cannot be analyzed with simple threshold methods. The adaptive moving-average detrending with a domain-specific criterion is new in this combination, and the synthetic validation is the strongest part — the optimizer recovers the injected time constants, which shows the method can work when the assumptions hold. The writing is clear and the figures are helpful.\n\nThe soft spot is the experimental validation. The optimization criterion in Eq. (4) includes term B, which directly minimizes the deviation between f0 and (1/2π)(1/τc+1/τe). The confirmation check in Section 3 then uses exactly that relation, Eq. (5). So agreement there is constructed, not independent evidence. The text even tells the user to run the filter until the estimated constants satisfy Eq. (5), which makes the circularity explicit. That does not mean the method is wrong, but the abstract's claim of \"cross-validation proving accuracy\" is too strong. The synthetic test is the actual validation; the real-device demonstration is a self-consistency check.\n\nOther concerns are minor by comparison. Only one device signal is shown, with no error bars or sensitivity analysis. The scaling parameters a, b, and quantization N are chosen by rule of thumb, and the load-bearing assumption about frequency separation (f0' at least an order of magnitude below f0) is asserted, not quantified, for the measured device. Data and code are \"available on request,\" which makes independent reproduction harder.\n\nAll that said, the method is plausible and the synthetic results are legitimate evidence. A reader working on RTN in ReRAM would find the detrending approach worth trying. The paper deserves serious refereeing, but it needs major revision: independent experimental validation, multiple devices, parameter sensitivity, and a more careful wording of what is proven versus what is demonstrated.","headline":"Useful detrending method with a solid synthetic test, but the experimental validation is circular and needs independent confirmation before the accuracy claim can be taken at face value.","tokens_in":10279,"tokens_out":1353,"would_cite":false,"duration_ms":14214,"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":"An adaptive moving-average detrending filter can flatten unstable RTN signals in ReRAM devices, making trap capture and emission time constants measurable where standard threshold-based analysis fails.","keywords":["random telegraph noise","resistive memory","ReRAM","adaptive filter","moving-average detrending","Lorentzian fitting","trap time constants","unstable filaments"],"falsifier":"A direct test would be to synthesize unstable RTN signals with known $\\tau_c$ and $\\tau_e$, add a slow drift whose corner frequency is close to the RTN corner frequency (violating the stated separation), run the adaptive filter, and check whether the recovered time constants match the ground truth; if they do, the band-separation assumption is not necessary, and if they do not, the method's range of validity is bounded by that assumption.","tokens_in":9364,"feed_emoji":"⚡","tokens_out":4056,"duration_ms":32113,"temperature":0.7,"pith_summary":"The paper addresses a measurement problem in resistive memories (ReRAM): random telegraph noise (RTN) signals often ride on slow, unstable drift of the read current, and when that drift is strong, standard threshold-based time-domain analysis cannot separate the two current levels or recover the trap capture and emission time constants. The authors propose an adaptive filter that applies a moving-average detrending step of optimized mask size, chosen by minimizing a custom criterion built from the histogram lobe means and standard deviations together with the power-spectral corner frequency. Applying the filter to a measured unstable RTN signal from a silicon nitride ReRAM cell yields stable two-level signals from which $\\tau_c$ and $\\tau_e$ are extracted, and the estimates satisfy the expected Lorentzian relation $f_0 = (1/2\\pi)(1/\\tau_c + 1/\\tau_e)$, which the authors take as cross-validation that the method is accurate. If the method works as claimed, it extends RTN-based trap characterization to unstable filament states, where previous time-domain techniques fail.","feed_headline":"Adaptive filter recovers trap times from unstable ReRAM noise","feed_subtitle":"Moving-average detrending plus Lorentzian cross-validation extracts tau_c and tau_e where threshold analysis fails.","key_machinery":"The central object is the adaptive moving-average detrending filter, defined by $y_M(k) = x(k) - (x*G_M)(k) + \\overline{x(k)}$, where $G_M$ is an $M$-tap uniform averaging mask. The detrended signal's usefulness is steered by the cost criterion $C = a(\\sigma_c+\\sigma_e)/|\\mu_c-\\mu_e| + b\\,|f_0 - (1/2\\pi)(1/\\tau_c+1/\\tau_e)|/N$, which combines histogram separation and spread of the two RTN levels with consistency with the Lorentzian corner frequency; minimization over mask size is performed with a global optimizer (SHGO). The known identity $f_0 = (1/2\\pi)(1/\\tau_c+1/\\tau_e)$ for single-trap RTN is what connects the time-domain estimates to the frequency-domain cross-check.","core_discovery":"The central claim is that an adaptive moving-average detrending filter can flatten unstable RTN signals well enough that a simple threshold separates the two levels and the trap time constants can be measured accurately. The filter subtracts a moving average of the signal and adds back the global mean, and the mask size is not chosen by trial but by a minimization algorithm: the cost function combines the separation and spread of the two histogram lobes with a term that penalizes deviation of the estimated corner frequency from the Lorentzian prediction. On a real unstable RTN measurement from an SiN$_x$ ReRAM cell tuned to an intermediate resistance state, the filter produces a detrended signal whose histogram shows two clean Gaussian lobes, and the extracted $\\tau_c$ and $\\tau_e$ satisfy the consistency relation $f_0 = (1/2\\pi)(1/\\tau_c + 1/\\tau_e)$ with the corner frequency $f_0=985$ Hz obtained from the power spectral density. The paper claims this cross-validation demonstrates the accuracy of the proposed method.","pith_inferences":["Because the method assumes the instability lies at much lower frequencies and much lower power than the RTN corner frequency, it will likely fail when filament drift is comparable in timescale to the trap switching; a testable extension is to map the failure boundary in the $(f_0'/f_0, P'/P)$ plane on simulated signals.","The same detrending-then-fit logic could be applied to other two-level fluctuators with slow baseline drift, such as qubit charge noise or molecular conductance switches, where the Lorentzian relation plays the same validation role.","A direct comparison against Hidden Markov Model fitting on the same unstable signals would clarify whether the advantage comes from detrending or from the specific cost criterion; the paper cites HMM results but does not benchmark its own output against them."],"forward_implications":["Unstable RTN signals, previously unusable for time-domain analysis, become analyzable for single-trap capture and emission time constants.","The method gives a criterion for choosing detrending filter width without knowing the ground-truth signal, using only measurable histogram and PSD parameters.","The extracted $\\tau_c$ and $\\tau_e$ satisfy the Lorentzian consistency relation, providing a built-in validation that the measured two-level signal is from a single trap.","The approach is intended to support multi-level cell (MLC) tuning, where a failed tuning protocol leaves the resistance state drifting, by allowing trap parameters to be read from the noisy state."],"supporting_citations":[{"why":"Establishes RTN analysis in nitride memristors and its role in device tuning, motivating the need for accurate time constants.","marker":"[9]"},{"why":"Supplies the robust detrending concept from another field that the proposed adaptive filter adapts.","marker":"[19]"},{"why":"Compares RTN statistical analysis algorithms on simulated signals, serving as a baseline that does not cover unstable filaments.","marker":"[20]"},{"why":"Shows the limitations of HMM analysis on unstable RTN signals in HfOx ReRAM, the problem this paper targets.","marker":"[21]"},{"why":"Provides the Lorentzian relation used for the frequency-domain cross-validation of the extracted time constants.","marker":"[32]"},{"why":"Supplies the SHGO global optimization algorithm used to minimize the filter's cost criterion.","marker":"[34]"}],"fun_headline_variants":["Adaptive filter untangles unstable RTN to get trap times","Moving-average detrending recovers ReRAM trap constants","Stabilize noisy ReRAM signals to read trap times with accuracy","Filter flattens unstable RTN, revealing emission and capture times","New method beats threshold analysis for unstable RTN in ReRAM"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The method assumes that the slow instability in the measured signal lives at a corner frequency at least an order of magnitude below the RTN corner frequency and with at least three orders of magnitude less power, so that moving-average detrending removes the drift without touching the two-level statistics.","fun_headline_variants_meta":{"raw":{"variants":["Adaptive filter untangles unstable RTN to get trap times","Moving-average detrending recovers ReRAM trap constants","Stabilize noisy ReRAM signals to read trap times with accuracy","Filter flattens unstable RTN, revealing emission and capture times","New method beats threshold analysis for unstable RTN in ReRAM"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000556,"raw_usage":{"total_tokens":2651,"prompt_tokens":953,"completion_tokens":1698,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":569,"completion_tokens_details":{"reasoning_tokens":1611}},"tokens_in":569,"tokens_out":1698,"duration_ms":11661,"temperature":1.0,"reasoning_tokens":1611,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T13:16:13.681237+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test would be to synthesize unstable RTN signals with known $\\tau_c$ and $\\tau_e$, add a slow drift whose corner frequency is close to the RTN corner frequency (violating the stated separation), run the adaptive filter, and check whether the recovered time constants match the ground truth; if they do, the band-separation assumption is not necessary, and if they do not, the method's range of validity is bounded by that assumption.","supporting_citations":[],"review_version":1}