{"id":"840f1f72-f22d-44c0-8760-6cd8fc5d2bdd","arxiv_id":"2506.01749","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The paper reports the first observation of pinched hysteresis loops in a MEMS resonator, created by parametric mode coupling, and extends this to multiple pinched loops using multiple pump signals.","lead":"Researchers show that a standard MEMS resonator, driven with parametric modulation signals and tracked by a phase-locked loop, produces pinched hysteresis loops between a stiffness perturbation and its electrical output. This is presented as the first memristor-like pinched hysteresis behavior in a MEMS device, with possible future uses in sensing, in-sensor memory, and edge AI.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The closed-loop PLL, not the resonator, may be generating the pinched hysteresis; Sec. III-B's quasi-static claim needs verification against open-loop phase-crossing branches and sweep-rate dependence.","rationale":"The reader identified the PLL quasi-static assumption as the weakest point, and I agree. The paper's proposed mechanism, phase-crossing degeneracy in virtually coupled modes, is plausible and is supported by the illustrative simulation, but the experimental branch structure is captured only through the closed loop. The quasi-static argument based on 10 mHz versus a 20 Hz bandwidth is insufficient because bandwidth is a linear small-signal property and can fail near the very degeneracies that produce the jumps. The concrete test is decisive because it compares closed-loop behavior directly against the open-loop steady-state phase-crossing branches and checks whether loop geometry changes with sweep rate or controller gains. The proxy issue is real but secondary: it limits the cross-domain physical-input claim, not whether pinched hysteresis is observed in the electrical-input experiment. Therefore the existing CONDITIONAL verdict should stand, pending open-loop verification, sweep-rate dependence checks, and data release.","tokens_in":14298,"tokens_out":3583,"duration_ms":39598,"concrete_test":"Conduct open-loop frequency sweeps of amplitude and phase at a fine voffset grid (e.g., 0.05 V steps over -2.5 V to 0.5 V) covering the regions of Figs. 8, 10, and 12, extract every frequency at which the phase response equals the 90-degree setpoint, and overlay these open-loop phase-crossing branches on the closed-loop voffset-frequency curves. Independently, repeat the closed-loop sweeps at 1, 2, 5, 10, and 50 mHz and with at least two different PID gain settings. If the closed-loop curves differ from the open-loop branches, or if loop area and discontinuity voltages change with sweep rate or PID gains, the pinched hysteresis is at least partly a PLL artifact. If they match and are rate-independent, the intrinsic interpretation is confirmed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that a MEMS resonator shows intrinsic memristor-like pinched hysteresis rests entirely on closed-loop PLL measurements in Secs. IV-A to IV-C. The only support for quasi-static tracking is the statement in Sec. III-B that the 10 mHz voffset sweep is 'well below the PLL bandwidth of 20 Hz'. This is not sufficient: PLL bandwidth is a linear small-signal metric, and near the phase-crossing degeneracies that the paper itself identifies as the switching mechanism, the loop's effective gain and settling time change, so a slow ramp can still be non-quasi-static and can produce controller-induced lag, integrator windup, and sweep-direction-dependent jump locations. The open-loop phase data in Fig. 6 are shown at only three voffset values, so they cannot rule out this artifact. If the hysteresis is a closed-loop artifact, the 'volatile memory' and multi-sensitivity conclusions in Sec. V do not follow. A secondary limitation, acknowledged in the Introduction, is that the input is an electrical offset voltage used as a proxy for physical stiffness; that weakens the cross-domain MemReSensor claim but not necessarily the observation itself.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the observation of pinched hysteresis loops in a commercial SOIMUMPs DETF MEMS resonator operated under parametric modulation. The authors apply one, two, and three parametric modulation signals to virtually couple multiple modes and, using a closed-loop PLL to track a fixed phase set-point while a 10 mHz offset voltage perturbs the stiffness, record frequency and amplitude responses that show a single hysteresis loop and multiple pinched loops. They support the interpretation with a harmonic-balance simulation of the one-PMS case and with open-loop phase-crossing measurements at three offset voltages. The paper introduces the term 'MemReSensor' and argues that the observed memory and multi-sensitivity suggest applications in in-sensor computing and cross-domain matrix multiplication.","tokens_in":14461,"tokens_out":9963,"duration_ms":97827,"significance":"If the pinched hysteresis is confirmed to be an intrinsic property of the resonator rather than an artifact of the measurement loop, this is a valuable first demonstration of memristor-like hysteresis in a micromechanical resonator. The systematic exploration of one, two, and three PMSs is a strength, and the experimental figures clearly show branch crossings and discontinuities. The HBM simulation, while qualitative and parameterized from experiment, gives a useful consistency check. However, the central claim is not yet fully supported: the closed-loop nature of the measurements and the unaddressed origin condition of the memristor fingerprint leave two load-bearing questions open. With the additional control experiments and clarifications proposed below, the paper could become a solid contribution.","major_comments":[{"comment":"The quasi-static assumption used to interpret the closed-loop measurements is not established. Stating that the 10 mHz voffset sweep is 'well below the PLL bandwidth of 20 Hz' is insufficient because the PLL bandwidth is a linear small-signal quantity and the loop dynamics near the phase-crossing degeneracies (the very points that trigger the observed jumps) can differ substantially. The open-loop data in Fig. 6 are provided for only three voffset values and therefore cannot reconstruct the branch structure over the full sweep. Please supply either (i) continuous open-loop phase/frequency responses as a function of voffset across the entire range, (ii) a sweep-rate dependence study (e.g., 1 mHz, 10 mHz, 50 mHz), or (iii) a closed-loop simulation that includes the PLL/PID controller, in order to show that the hysteresis is intrinsic to the resonator and not an artifact of the tracking loop.","section":"Section III-B, IV-A-1"},{"comment":"The claim that the loops satisfy the memristor pinched-hysteresis fingerprint of [5] is not justified. In a memristor the pinched loop passes through the origin of the input-output plane; here the crossings occur at v_offset = 0 V but at nonzero output frequency or amplitude. Please replot the responses in terms of a relative output (e.g., frequency deviation from the unperturbed f1 and amplitude relative to its value at v_offset = 0) so that the loop passes through the origin, or explicitly define the generalized pinched-hysteresis criterion being used and state how it connects to [5].","section":"Section IV-A-4, Figs. 8b, 10b, 12b"},{"comment":"The abstract and conclusion describe the demonstration as 'cross-domain' with a physical input and electrical output, but the input in all experiments is an electrical offset voltage used only as a proxy for stiffness. This limitation is acknowledged in the Introduction, but the conclusion restates the cross-domain claim without qualification. Please either temper the cross-domain wording in the abstract and conclusion or present at least one demonstration with a physical stimulus (or a physically modulated stiffness) to support the generalization.","section":"Introduction, Section V"},{"comment":"The open-loop phase-crossing analysis in Fig. 6, which is used to explain the hysteresis mechanism, does not state the PMS parameters used. If those parameters differ from the v1,3_p = 4 V, Δf1,3_p = 0 Hz used in the closed-loop data of Fig. 8, the degeneracy points at approximately ±0.3 V in Fig. 6 cannot be directly compared with the discontinuities observed at approximately +1 V and -0.75 V in Fig. 8a. Please state the parameters for Fig. 6 and reconcile this discrepancy.","section":"Section IV-A, Figs. 6 and 8"}],"minor_comments":[{"comment":"The quality-factor column labels 'Q1' and 'Q2' should be 'Q1' and 'Q3' to match the first and third modes used in the simulation.","section":"Table I"},{"comment":"The sentence 'The PMSs used have frequencies determined as follows, and Δf1,jp ...' is awkward; please rephrase.","section":"Section III-B"},{"comment":"The phrase 'This sweeping process is again comparable' uses 'again' without a prior comparison; consider removing it.","section":"Section IV-A-1"},{"comment":"The sentence 'the phenomenon that satisfies the definition of pinched hysteresis [5] occurs, even though a pinched intermediate branch, rather than a crossing, exists' is self-contradictory in its use of 'pinched'; clarify whether the branch overlap is intended to count as a pinch.","section":"Section IV-B-2"},{"comment":"Please specify the PLL/PID controller settings (e.g., proportional/integral gains and the method used to determine the 20 Hz bandwidth) so that the quasi-static claim can be reproduced.","section":"Section III-B"}],"recommendation":"major_revision","confidential_remarks":"The key missing evidence is a set of open-loop or sweep-rate control experiments. If those experiments reveal PLL-induced hysteresis, the paper's central claim would collapse; if they confirm intrinsic hysteresis, the paper is likely acceptable after revisions. The journal should ask for these experiments before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuine first — pinched hysteresis loops in a MEMS resonator, with multiple loops when two or three parametric pumps are applied. The phase-crossing explanation is coherent, and the single-pump harmonic balance simulation matches the experiment's shape. I'd send it to a serious referee. But the paper's central interpretation as a memristor-like sensor is ahead of the evidence. The biggest unresolved issue is the measurement chain: everything is recorded through the MFLI phase-locked loop, and the only support for calling the response quasi-static is that the 10 mHz perturbation is below the quoted 20 Hz PLL bandwidth. That is a linear small-signal metric; near the phase-crossing degeneracies identified in the paper, loop gain and settling change. A slow ramp can still produce controller-induced lag, integrator windup, and sweep-direction-dependent jumps. The open-loop phase data in Fig. 6 are shown at only three voffset values, so they cannot rule out a closed-loop artifact. I don't think this kills the observation — the open-loop response clearly shows phase-crossing count changing with voffset, so the system genuinely has multivalued branches — but the hysteresis loop shapes, pinch locations, and 'memory' claims need confirmation at different sweep rates and ideally open-loop tracking of the branches. \n\nSecond soft spot: the memristor classification. A loop that crosses at the origin is necessary but not sufficient without checking the frequency dependence of the loop area, the standard fingerprint. No such test appears here. Also, the input is an electrical offset voltage used as a proxy for physical stiffness, so the cross-domain 'MemReSensor' claim is not actually demonstrated. The paper acknowledges the proxy, but the title and abstract push the sensor angle harder than the data support. \n\nWhat the paper does well: careful experimental setup, temperature stabilization, feedthrough checks, clear description of the mode coupling and phase-degeneracy mechanism, and a plausible simulation for the one-pump case. The two- and three-pump data are novel and will be useful to people working on coupled MEMS dynamics. \n\nWho this is for: MEMS resonator people and anyone building parametric sensors; the memristor analogy is more of a hook than a working device. If I were the editor, I'd send it out — the core observation is new and deserves scrutiny — but the authors should be told that the PLL artifact needs to be addressed with open-loop measurements and sweep-rate dependence, and that the memristor claims should be softened until the fingerprint test is done.","headline":"First pinched hysteresis in a MEMS resonator, measured through a PLL; the observation is plausible and new, but the memristor claim needs controls against loop-controller artifacts.","tokens_in":15015,"tokens_out":3021,"would_cite":false,"duration_ms":31111,"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":"A MEMS resonator shows memristor-like pinched hysteresis between a stiffness input and its electrical output.","keywords":["MEMS resonator","pinched hysteresis","memristor","parametric modulation","phase-locked loop","mode coupling","MemReSensor"],"falsifier":"Repeat the closed-loop stiffness sweep at several rates (for example 1 mHz to 1 Hz) and with different PLL bandwidth settings; if the loop area, crossing location, or discontinuity voltages change systematically with rate or controller gain, the pinched hysteresis is partly a control-loop artefact. Alternatively, map the open-loop phase response as a function of offset voltage and verify that the number of phase crossings changes exactly at the measured discontinuity voltages; if it does not, the PLL itself is generating the bistability.","tokens_in":14034,"feed_emoji":"🔁","tokens_out":9059,"duration_ms":85010,"temperature":0.7,"pith_summary":"This paper reports the first observation of pinched hysteresis — the self-crossing loop that defines a memristor — in a silicon MEMS resonator, with no material switching involved. The device is a generic double-ended tuning-fork resonator whose vibration modes are virtually coupled by parametric pump signals; a phase-locked loop tracks its resonance while a slow electrical offset voltage perturbs the stiffness. Sweeping this stiffness input up and down produces hysteresis loops in the output frequency and amplitude, and the loops cross without exchanging branches, which is precisely the pinched-hysteresis fingerprint. The authors argue that this makes resonant sensors capable of volatile memory of past input and programmable sensitivity, and could lead to sensors that compute within the physical domain.","feed_headline":"MEMS resonator shows memristor-like pinched hysteresis","feed_subtitle":"Slow stiffness sweeps create pinched hysteresis loops in a silicon resonator, giving it memory of past inputs.","key_machinery":"The load-bearing mechanism is parametric modulation (red-detuned pumping) combined with phase-locked-loop tracking. Parametric pumps at frequencies near the differences between the first mode and higher modes, $f_p^{1,j}=f_j-f_1+\\Delta f_p^{1,j}$, virtually couple the modes, modifying the phase transition from $180^\\circ$ to $0^\\circ$ so that it crosses the $90^\\circ$ set-point multiple times. As the electrostatic offset voltage $v_{\\text{offset}}$ changes the stiffness, these phase crossings move; when two crossings merge, the PLL discontinuously switches modes. The set of branches traced by upward and downward sweeps, with crossings that do not allow branch exchange, constitutes the pinched hysteresis. The paper also uses the conversion $\\Delta k_{e,1}/\\Delta v_{\\text{offset}} \\approx 0.126\\ \\mathrm{N/m/V}$ from electrostatic parallel-plate formulas, validated against the experimental value $0.121\\ \\mathrm{N/m/V}$, to state the input in stiffness units.","core_discovery":"The central claim is that the pinched hysteresis fingerprint can be transferred from the electrical domain into a mixed physical-electrical domain: an electrostatic stiffness perturbation $\\Delta k_{e,1}$ acts as the input and the resonator's tracked frequency or amplitude acts as the output, without invoking a memristive material. Under the parametric-modulation operating scheme, a pump at $f_p^{1,j}=f_j-f_1+\\Delta f_p^{1,j}$ creates virtual coupling between mode 1 and mode j. This changes the phase response near $f_1$ so that the $90^\\circ$ phase set-point of the phase-locked loop is crossed at one or three frequencies depending on stiffness. At two stiffness values the three crossings degenerate to one; the PLL then jumps to another coupled mode, and the upward and downward sweeps follow different branches. The branches cross at zero offset voltage but the device does not switch there — the defining property of a pinched hysteresis loop. With one pump this appears in the amplitude response; with two pumps, in the frequency response; with three pumps, multiple pinched loops appear in both, because intermediate branches are accessible in either sweep direction.","pith_inferences":["Beyond the paper, a direct test of the cross-domain claim would be to couple the same parametrically modulated resonator to a genuine acceleration or magnetic-field input and check that the same pinched loops survive without the electrical offset-voltage proxy.","The phase-crossing explanation predicts that the pinched crossing point coincides with the stiffness where the open-loop phase response changes from three intersections with the set-point to one; a precomputed open-loop phase map over stiffness would test this.","The mechanism is generic to any weakly coupled oscillator under phase tracking, so similar pinched hysteresis may appear in photonic, phononic, or macroscopic mechanical systems, not only in MEMS.","The memory demonstrated here is volatile and depends on the phase-locked loop staying active; a non-volatile version would require a state variable that persists after the loop is turned off, which this paper does not claim."],"forward_implications":["A resonant sensor can hold two or more stable sensitivity values at the same input, with the past input deciding which one is active; this is a volatile, mechanical memory of the sort memristors provide.","With two or three parametric pumps, the device shows multiple pinched hysteresis loops, so a single resonator can encode multi-state input–output relations rather than a single fixed sensitivity.","Because an electrostatic offset voltage is a proxy for stiffness changes caused by acceleration, magnetic field, or temperature, the same scheme should transfer to real physical sensing axes.","The presence of hysteresis gives the resonator the short-term memory needed for reservoir computing, making in-physical-sensor computation a plausible next step.","Arranged in an array, such resonators could multiply a vector of physical inputs by programmable sensitivity values, including negative and zero entries, in analogy to a memristor crossbar."],"supporting_citations":[{"why":"Defines the pinched hysteresis criterion used to identify the loops as memristor-like.","marker":"[5]"},{"why":"Supplies the parametric modulation scheme that creates virtual coupling between modes.","marker":"[18]"},{"why":"Provides the phase-locked-loop tracking method and parametric modulation context for resonant sensors.","marker":"[19]"},{"why":"Shows the standard PLL operation in practical resonant sensors that the experiment mirrors.","marker":"[20]"},{"why":"Establishes the electrostatic stiffness perturbation as a proxy for physical inputs and the Δk/k relation used for calibration.","marker":"[21]"},{"why":"Provides the coupled-mode equations of motion and the electrostatic stiffness conversion used in the theory.","marker":"[29]"},{"why":"Supplies the harmonic balance method used to simulate the one-pump case and predict the loop shapes.","marker":"[30]"},{"why":"Explains the phase-crossing degeneracy in two-mode-coupled oscillators that underlies the discontinuities.","marker":"[36]"}],"fun_headline_variants":["First pinched hysteresis loop in a MEMS resonator","MEMS resonator remembers: pinched hysteresis observed","Memristor-like memory in a silicon resonator","Resonator mimics memristor with pinched hysteresis","MEMS resonator shows memristor-like memory effect"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the phase-locked loop, locked to a fixed $90^\\circ$ phase set-point, follows the resonator's mode branches quasi-statically as the 10 mHz offset-voltage sweep changes stiffness, so the recorded frequency and amplitude curves are the resonator's intrinsic response rather than a product of the PLL/PID controller dynamics or sweep speed.","fun_headline_variants_meta":{"raw":{"variants":["First pinched hysteresis loop in a MEMS resonator","MEMS resonator remembers: pinched hysteresis observed","Memristor-like memory in a silicon resonator","Resonator mimics memristor with pinched hysteresis","MEMS resonator shows memristor-like memory effect"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000175,"raw_usage":{"total_tokens":1302,"prompt_tokens":979,"completion_tokens":323,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":595,"completion_tokens_details":{"reasoning_tokens":248}},"tokens_in":595,"tokens_out":323,"duration_ms":3572,"temperature":1.0,"reasoning_tokens":248,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T11:34:23.793708+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the closed-loop stiffness sweep at several rates (for example 1 mHz to 1 Hz) and with different PLL bandwidth settings; if the loop area, crossing location, or discontinuity voltages change systematically with rate or controller gain, the pinched hysteresis is partly a control-loop artefact. Alternatively, map the open-loop phase response as a function of offset voltage and verify that the number of phase crossings changes exactly at the measured discontinuity voltages; if it does not, the PLL itself is generating the bistability.","supporting_citations":[{"cited_title":"Toward high-resolution inertial sensors employing paramet- ric modulation in coupled micromechanical resonators,","cited_arxiv_id":null,"evidence_quote":"Provides the phase-locked-loop tracking method and parametric modulation context for resonant sensors."},{"cited_title":"If it’s pinched it’s a memristor,","cited_arxiv_id":null,"evidence_quote":"Defines the pinched hysteresis criterion used to identify the loops as memristor-like."},{"cited_title":"Dynamic modulation of modal coupling in microelectromechanical gyroscopic ring resonators,","cited_arxiv_id":null,"evidence_quote":"Supplies the parametric modulation scheme that creates virtual coupling between modes."},{"cited_title":"A vibrating beam MEMS accelerometer for gravity and seismic measurements,","cited_arxiv_id":null,"evidence_quote":"Shows the standard PLL operation in practical resonant sensors that the experiment mirrors."},{"cited_title":"En- hancing parametric sensitivity in electrically coupled MEMS resonators,","cited_arxiv_id":null,"evidence_quote":"Establishes the electrostatic stiffness perturbation as a proxy for physical inputs and the Δk/k relation used for calibration."},{"cited_title":"On enhancing the sensitivity of resonant thermometers based on parametric modulation,","cited_arxiv_id":null,"evidence_quote":"Provides the coupled-mode equations of motion and the electrostatic stiffness conversion used in the theory."},{"cited_title":"Computational and quasi-analytical models for non-linear vibrations of resonant MEMS and NEMS sensors,","cited_arxiv_id":null,"evidence_quote":"Supplies the harmonic balance method used to simulate the one-pump case and predict the loop shapes."},{"cited_title":"Higher-order singularities in phase-tracked electromechanical oscillators,","cited_arxiv_id":null,"evidence_quote":"Explains the phase-crossing degeneracy in two-mode-coupled oscillators that underlies the discontinuities."}],"review_version":1}