{"id":"ea8270a1-81c1-4c36-b5b4-1ce870f615f8","arxiv_id":"2505.16272","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Splitting a superconducting chip into physically separated dies contains cosmic-ray phonon bursts to a single die, with a small residual rate of direct particle hits across both dies.","lead":"Researchers split a superconducting circuit into two separate silicon dies with a small gap and watched cosmic-ray bursts with microwave detectors. Bursts stayed on the die they hit, suggesting physical separation could stop the correlated errors cosmic rays cause in quantum computers.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The blocking claim assumes unverified same-die burst detection efficiency; with only two MKIDs per die monitored, weak cross-die bursts could be missed and misclassified as containment.","rationale":"The reader's weakest assumption identifies the same sensitive point; I agree. The paper's conclusion would require that the pairwise detector trigger condition is a reliable proxy for 'any burst on a die,' but the experiment never tests that proxy. The explicit limitation that only four MKIDs were measured and the concession that some events are too small to detect make this more than a hypothetical. Note that the geometric estimate of 4% direct double-die events matching the observed 2.8% is real supporting evidence for the direct-traversal explanation of the simultaneous events, so the paper is not merely circular. However, the central 'conclusively demonstrated' claim outruns the data. This does not change the reader's CONDITIONAL verdict; it strengthens the stated conditions: add a sensitivity calibration or full coverage and a control comparison, and report Poisson/statistical uncertainties on the 10/352 simultaneous-event fraction.","tokens_in":7072,"tokens_out":5061,"duration_ms":43680,"concrete_test":"Monitor all eight MKIDs simultaneously and inject calibrated pulses (pulsed resistive heater or focused laser) into one die. For each pulse, record how many of the four MKIDs on the source die trigger and whether any MKID on the other die triggers. If same-die detection is not near 100% at relevant burst energies, or if any cross-die trigger occurs above noise, the blocking conclusion needs revision; if same-die detection is complete and no cross-die trigger occurs, the sensitivity loophole is closed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central conclusion rests on a null result: no cross-die phonon bursts were seen. That null result only proves blocking if a burst on either die would always be detected on the monitored MKIDs of the other die. The paper's own statement that 'a particle impact anywhere on a die should be detectable in all four MKIDs on the same die' is an assumption, not a calibration. Only D04/D05 and D09/D10 were read out; D1, D6, D8 (and dead D7) were not. A phonon burst that crosses the 0.2 mm gap but arrives with low amplitude at the two monitored MKIDs on the far die would be below the fluctuation-derived trigger threshold and would be recorded as a single-die event. The text even concedes that 'the signal generated by some events is small, and they are therefore not measured.' Without a calibrated energy source, a monolithic control, or full MKID coverage, the absence of cross-die triggers cannot distinguish true phonon blocking from incomplete detection. The same ambiguity affects the 10 simultaneous four-MKID events: their interpretation as direct particle traversals is plausible and the geometric estimate is a genuine non-circular check, but it is not a calibrated discrimination against phonon leakage through the common copper mount and PCB ground.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experiment using microwave kinetic inductance detectors (MKIDs) on two separate silicon dies, mounted adjacently with a ~0.2 mm gap and no galvanic connection, to test whether die separation blocks phonon bursts induced by cosmic rays or background radioactivity. The authors observe that most detected bursts appear on only one die (in both monitored MKIDs of that die), and interpret the absence of cross-die bursts as evidence that phonon propagation is stopped by the physical gap. They also observe 10 simultaneous events in all four monitored MKIDs (2.8% of 352 bursts) and argue, based on a geometric model of straight-line particle trajectories, that the expected rate of a single particle traversing both dies is 4.0%, supporting the interpretation that these are direct particle hits rather than phonon leakage. The paper concludes that die separation contains burst effects to a single die and is a viable mitigation for correlated errors in multi-die quantum processors.","tokens_in":7268,"tokens_out":2881,"duration_ms":22828,"significance":"If the central claim is correct, the work offers a simple, practical mitigation for correlated quasiparticle-burst errors that threaten quantum error correction in multi-die superconducting processors. The geometric double-hit estimate is a genuine, parameter-free prediction (aside from the trigger threshold) and the lack of observed steady cross-die bursts is suggestive. The experiment uses a real device with continuous simultaneous readout, and the paper is honest about the existence of simultaneous events. However, the strength of the conclusion rests on a null result whose validity depends on unverified detection efficiency, and the statistical support for the geometric match is not quantified. These gaps are load-bearing and require additional measurements or analysis before the blocking claim can be considered conclusive.","major_comments":[{"comment":"The claim that 'the propagation of bursts is effectively stopped by the physical separation' rests on the absence of cross-die triggers, but only D04/D05 and D09/D10 were monitored, while D1, D6, D8 (and dead D7) were not read out. The paper's assertion that 'a particle impact anywhere on a die should be detectable in all four MKIDs on the same die' is an assumption, not a calibration. Without a control experiment on a monolithic chip, a calibrated energy source, or full MKID coverage, the null result cannot distinguish true phonon blocking from a weak cross-die burst falling below the threshold of the two monitored MKIDs on the far die. This is the central load-bearing point of the paper.","section":"Experimental setup and Fig. 3 (paragraph beginning 'Due to technical limitations')"},{"comment":"The agreement between the observed 2.8% (10/352) and the predicted 4.0% is presented as support for the direct-traversal interpretation, but no statistical uncertainty is given. With Poisson counting statistics, the 95% confidence interval for 10 events spans roughly 5-18 events, i.e., 1.4%-5.1%, which comfortably includes 4.0%. The match is therefore not statistically constraining at the claimed level of precision, and the conclusion that these events are 'likely from a single cosmic particle' should be stated with appropriate uncertainty or with a quantitative likelihood comparison.","section":"Probability calculation (paragraph 'We model the trajectory of a particle...' and the sentence 'of the total 352…"},{"comment":"The statement that 'We have ruled out the significance of potential energy transfer mechanisms between the dies such as energy transfer through radiation bouncing inside the sealed copper box... and phonon propagation through the bulk of the copper box and through the PCB' is too strong for the data presented. The experiment does not separately test each channel; it only observes no above-threshold coincident events. To rule out these mechanisms, one would need a measurement with known injected energy on one die and a calibrated sensitivity on the other, or a monolithic control chip. As it stands, the absence of coincidences is consistent with blocking but also with detection inefficiency, so the 'ruled out' language overstates the claim.","section":"Conclusion (paragraph beginning 'We have ruled out the significance...')"}],"minor_comments":[{"comment":"The figure captions contain corrupted character sequences (e.g., '/uni00000017/uni0000008b/uni0000008f...') that should be repaired before publication.","section":"Fig. 3 and Fig. 4 captions"},{"comment":"There are typographical errors, including 'after loading' (should be 'After loading'), 'ge varnish' (should be 'GE varnish'), and 'were r2 is' (should be 'where r2 is').","section":"Throughout"},{"comment":"The trigger threshold is described only as 'a predefined threshold value, which we determined from the signal fluctuations.' A quantitative statement (e.g., threshold in units of standard deviation of the noise) would help readers assess the sensitivity and the likelihood of missing weak events.","section":"Trigger threshold description"},{"comment":"The paper cites Ref. [6] (Moshel et al., Applied Physics Letters 2024) from the same group but does not explicitly state what is new in this work relative to that prior study; a brief statement of the advance would be helpful.","section":"Relation to Ref. [6]"}],"recommendation":"major_revision","confidential_remarks":"The manuscript addresses a timely and practical question in quantum hardware, and the geometric estimate is a nice non-circular check. However, the central claim of phonon blocking is supported mainly by a null result with unverified detection efficiency; this needs to be made convincing through a control measurement, full-die coverage, or explicit sensitivity calibration. The statistical comparison of the double-die rate is also weaker than presented. I would lean toward major revision rather than rejection because the limitations are clearly identifiable and potentially fixable, and the core idea is sound."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nThe short version: this is a credible, useful experiment, but the 'conclusively demonstrated' in the abstract overstates what the data show. The new thing is direct: two dies separated by a 0.2 mm gap, with simultaneous MKID readout on both sides, and almost every burst lands on one die. The 10/352 (2.8%) simultaneous events track a parameter-free geometric estimate of 4.0% for a single particle crossing both dies—a genuine non-circular check.\n\nCredit where due: the MKID characterization is thorough, the trigger and readout scheme is described in enough detail to reproduce, the geometry model is explicit, and the prior mitigation literature (shielding, traps, post-selection) is cited fairly. The paper also candidly notes that small signals are not measured and that only four MKIDs could be read at once.\n\nThe soft spot is the blocking claim itself. It rests on a null result—no phonon bursts crossing the gap—but only D4/D5 and D9/D10 were monitored; D1, D6, D8 were silent and D7 was dead. The paper assumes a burst anywhere on a die is detectable in all four MKIDs on that die, so two monitored MKIDs per die should be enough. That assumption is not calibrated. The text's own admission that 'the signal generated by some events is small, and they are therefore not measured' means a weak burst crossing the gap could fall below trigger on the far die and look like containment. So a fair conclusion is: for bursts above threshold, die separation blocks cross-die propagation. 'Conclusively demonstrated' goes beyond that.\n\nTwo smaller issues: no monolithic-chip control, though prior work on substrate phonon propagation makes this less critical; and no error bars on the 10/352 vs 4.0% comparison, which is still consistent under Poisson statistics (2.8% ± 0.9%).\n\nWho should read this: groups building multi-die superconducting processors and anyone working on correlated-error mitigation. It deserves a serious referee. I'd send it out and ask for a softer claim, plus either a calibrated phonon source, full MKID coverage, or a monolithic control. The mechanism is credible and the experiment is honest; the overclaim is the main thing to fix.","headline":"A credible packaging-level test of die separation for phonon-burst containment, with an overclaimed 'conclusive' verdict that needs a calibration or control.","tokens_in":7821,"tokens_out":4999,"would_cite":true,"duration_ms":42024,"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":"Separating a superconducting chip into distinct dies with a small gap blocks phonon bursts from propagating between dies, containing cosmic-ray-induced errors in a single die.","keywords":["cosmic rays","phonon bursts","quasiparticle poisoning","kinetic inductance detectors","superconducting qubits","correlated errors","multi-die processors","quantum error correction"],"falsifier":"Inject known-energy phonon bursts at controlled positions on one die while monitoring both dies; if any injection produces a trigger on the far die's MKIDs, or fails to trigger at least one same-die MKID, the blocking claim is contradicted. Alternatively, measure the cross-die coincidence fraction with much larger statistics and compare to the ray-geometry model: a statistically significant excess over the predicted 4.0% would indicate an unblocked transfer channel.","tokens_in":6847,"feed_emoji":"⚛️","tokens_out":9031,"duration_ms":69596,"temperature":0.7,"pith_summary":"Cosmic rays and radioactive decays hitting a superconducting circuit create high-energy phonon bursts that spread through the substrate, produce quasiparticles, and cause correlated errors across many qubits; standard quantum error correction struggles with such correlated errors. This paper asks whether physically cutting a single chip into adjacent dies, separated by a small gap and no galvanic connection, stops the phonon bursts from propagating between dies. Using kinetic-inductance detectors on two dies, the authors find bursts appear in the two detectors of one die but not on the other, and conclude the gap blocks cross-die phonon propagation. They also observe a small number of simultaneous bursts on both dies (10 of 352, 2.8%), which a geometric model attributes to a single particle passing through both dies. If correct, die separation localizes burst-induced errors to one die, where spectator detectors, post-selection, or error correction can manage them.","feed_headline":"Die separation blocks phonon bursts between qubits","feed_subtitle":"A 0.2 mm gap with no galvanic connection keeps cosmic-ray burst errors on one die, preserving quantum error correction.","key_machinery":"The operating detector is a microwave kinetic inductance detector (MKID), a $\\lambda/4$ superconducting transmission-line resonator with resonance frequency $\\omega_0 = 1/\\sqrt{C(L_g + L_k)}$; a burst breaks Cooper pairs and raises the kinetic inductance $L_k$, shifting $\\omega_0$ downward and changing the measured $S_{21}$ phase. The separator is the physical die gap itself: two silicon dies held at their corners with roughly $0.2\\text{ mm}$ between them, wired to the readout through aluminum wirebonds, so the die circuits have no direct galvanic connection. The comparison model is a ray-geometry calculation (Eqs. 3-5) that gives the expected probability a random cosmic ray intersects both dies, which is used to decide whether simultaneous two-die events are leakage or direct traversal.","core_discovery":"The central claim is that separating a superconducting circuit into distinct dies with a roughly 0.2 mm physical gap and no galvanic connection blocks the propagation of pair-breaking phonon bursts, so a burst is contained within the die it strikes. The authors demonstrate this with two silicon dies, each carrying four kinetic-inductance detectors driven and read out simultaneously; of 352 bursts recorded, the typical signature is a pair of detections in the two MKIDs on one die, with none on the other die. They rule out transfer through the copper mounting base, the common circuit ground, and radiation re-emitted inside the sealed box. Ten events were detected in all four MKIDs simultaneously; the authors model particle trajectories as random rays and compute an expected double-die fraction of 4.0%, close to the measured 2.8%, supporting the interpretation that these rare events come from a single cosmic particle traversing both dies rather than from phonons crossing the gap. The conclusion is that die separation is a viable containment strategy for burst-induced correlated errors in multi-die superconducting processors.","pith_inferences":["A natural extension, not tested in the paper, is varying the gap width and die thickness to see whether the two-die coincidence fraction follows the geometric ray model; a mismatch would indicate an additional transfer channel.","The method is orthogonal to quasiparticle and phonon traps, so a combined strategy could block cross-die bursts and also shorten in-die recovery; the paper does not test this combination.","Because per-die detection sensitivity is not calibrated against a known source, a controlled injection measurement would be the cleanest way to bound missed weak bursts and sharpen the blocker's claimed efficiency.","Scaling to many dies turns the residual risk into shallow-angle particles that traverse several dies; staggering or tilting dies could reduce that fraction beyond the two-die result."],"forward_implications":["A multi-die processor built with high-proximity capacitive interconnects can exchange quantum information between dies while preventing phonon-burst errors from spreading, so burst-induced errors are localized to a single die.","Adding MKIDs as spectator resonators to each die gives real-time burst detection, allowing post-selection or targeted quantum error correction at a detection time estimated below $1\\,\\mu\\text{s}$ in an optimized system.","The measured 2.8% simultaneous two-die bursts imply that direct particle traversal remains a residual correlated-error channel; any protocol using die separation must account for it.","The results rule out the copper mounting base, the PCB common ground, and re-emitted radiation as significant cross-die burst pathways for this geometry."],"supporting_citations":[{"why":"Establishes that cosmic-ray and radioactivity impacts create high-energy phonon bursts in superconducting substrates, the phenomenon being mitigated.","marker":"[1–3]"},{"why":"Explains how burst-generated quasiparticles poison qubits and cause correlated errors, the threat die separation addresses.","marker":"[4]"},{"why":"Provides prior experimental characterizations of burst dynamics and timescales that motivate the MKID trigger and recovery-time analysis.","marker":"[5–9]"},{"why":"Defines the microwave kinetic inductance detector principle used to sense burst-induced quasiparticles.","marker":"[21]"},{"why":"Supplies the resonator physics linking kinetic inductance changes to resonance frequency shifts, the basis of burst detection.","marker":"[22, 23]"},{"why":"Gives the $S_{21}$ transmission formula used to convert measured quadratures into resonance shifts and triggers.","marker":"[25]"},{"why":"Provides the cosmic-ray angular distribution used in the geometric estimate of double-die event probability.","marker":"[26]"}],"fun_headline_variants":["Die gap blocks phonon bursts in qubit chips","Separated dies contain cosmic-ray burst errors","0.2 mm gap stops phonon spread between dies","Phonon bursts trapped by die separation","Die isolation halts burst propagation in qubits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The blocking claim rests on the assumption that a phonon burst anywhere on a die is always large enough to trigger the monitored kinetic-inductance detectors on that die, so no weak burst crossing the gap is missed.","fun_headline_variants_meta":{"raw":{"variants":["Die gap blocks phonon bursts in qubit chips","Separated dies contain cosmic-ray burst errors","0.2 mm gap stops phonon spread between dies","Phonon bursts trapped by die separation","Die isolation halts burst propagation in qubits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000194,"raw_usage":{"total_tokens":1311,"prompt_tokens":863,"completion_tokens":448,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":479,"completion_tokens_details":{"reasoning_tokens":376}},"tokens_in":479,"tokens_out":448,"duration_ms":3891,"temperature":1.0,"reasoning_tokens":376,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T15:03:33.316286+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Inject known-energy phonon bursts at controlled positions on one die while monitoring both dies; if any injection produces a trigger on the far die's MKIDs, or fails to trigger at least one same-die MKID, the blocking claim is contradicted. Alternatively, measure the cross-die coincidence fraction with much larger statistics and compare to the ray-geometry model: a statistically significant excess over the predicted 4.0% would indicate an unblocked transfer channel.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Explains how burst-generated quasiparticles poison qubits and cause correlated errors, the threat die separation addresses."}],"review_version":1}