{"id":"82e40545-ef34-4bb3-a3db-50b147a97645","arxiv_id":"2607.18387","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":13,"one_line_summary":"A projected architecture for trapped-ion quantum modules combines single-photon heralding, integrated photonics, recoil correction, and one distillation round to deliver 99.9%-fidelity remote Bell pairs at 10^5 s^-1 cm^-2.","lead":"This paper designs a modular trapped-ion quantum computer interconnect that uses single-photon heralding, integrated photonics, and local distillation to project remote Bell-pair rates above 10^5 s^-1 cm^-2 at 99.9% fidelity. It argues that remote entanglement can stop being the bottleneck, shifting the limit to local operations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Fidelity budget relies on an undemonstrated 408 nm high-extinction fast switch; with the closest demonstrated extinction the false-herald infidelity alone breaks the 99.9% target, so the parallel all-to-all claim is not yet established.","rationale":"I read the paper as a transparent, internally consistent architecture projection, not a demonstration. The rate/fidelity model is explicit, most components are cited from existing demonstrations, and the paper honestly flags the 408 nm high-extinction fast switch as the one component exceeding demonstrated performance (Sec. VI C). The reader's CONDITIONAL verdict is appropriate: the projection is credible and actionable, but the headline numbers are not established until this enabling component is shown to meet the required extinction, speed, and phase stability. My focus on the switch network is the same load-bearing concern the reader identified among its listed fragile premises, and the quantitative check above would settle whether that concern actually breaks the 99.9% fidelity under parallel all-to-all operation. I do not see a deeper internal inconsistency that would justify REJECT; the paper's own limitation statement is correctly placed and properly weights the projection.","tokens_in":28798,"tokens_out":10754,"duration_ms":99673,"concrete_test":"Fabricate or characterize a fast (≥10 MHz) Mach–Zehnder or equivalent switch at 408 nm, measuring on-chip extinction and push–pull phase stability. If the extinction is <40 dB, recompute the Table II false-herald row as pfalse = n_X·10^{-ER/10}, where n_X is the number of switch crossings per routed photon, and compare 6 p_e pfalse with the residual-infidelity budget. If the measured extinction is, say, 30 dB and the resulting distilled fidelity is <99.9%, the parallel all-to-all claim fails and only the passive, single-pair-at-a-time variant survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of 10^5 s^-1 cm^-2 at F>99.9% with parallel all-to-all connectivity depends on the active butterfly switch network of App. C. The fidelity budget (Table II and Sec. VB) assigns only 0.024% infidelity to false heralds, using detector dark counts plus 'high-extinction switches' (App. C). But the paper itself states that the only component whose architecture requirements exceed demonstrated performance is the fast, high-extinction switch at 408 nm; the closest demonstrated device has >10 dB extinction at 420 nm [71], and a 30 dB, >10 MHz MZI mesh was shown at 737 nm [46], not at 408 nm. Quantitatively: a false-herald probability pfalse contributes infidelity ≈ 6 p_e pfalse after distillation. To keep the false-herald row at 0.024%, pfalse must be ≈2×10^-4. With one switch crossing per routed photon (M=2, D=1), a 30 dB extinction gives pfalse ≈ 10^-3, i.e. ≈0.12% infidelity; 10 dB gives ~0.1, i.e. ~12% infidelity. The required extinction is therefore ≳40 dB at 408 nm, a level not demonstrated at that wavelength and speed. Without such a switch, the architecture falls back to a passive multiport that supports only one module pair at a time (Table V), invalidating the 'parallel all-to-all' part of the headline claim. This is an explicit, acknowledged limitation, but it is load-bearing: the specific numerical claim of simultaneous all-to-all entanglement at 99.9% is contingent on a component that does not yet exist.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This Perspective proposes a modular trapped-ion quantum computing interconnect based on single-photon heralded entanglement between 88Sr+ ions in surface-electrode traps with integrated photonics. The scheme combines weak two-step excitation, coherent recoil correction, projective distillation, and leakage detection. The central projection is a distilled Bell-pair rate of 858 s^-1 per channel at fidelity ~99.96%, tiling to a remote Bell-pair rate density of ~10^5 s^-1 cm^-2, which the authors argue is sufficient for fault-tolerant operations and supports parallel all-to-all module connectivity via an active butterfly switch network. The paper emphasizes that single-photon heralding's linear scaling with detection efficiency removes the need for high-NA collection optics and allows dense integrated photonic channels, shifting the bottleneck to local operations.","tokens_in":29216,"tokens_out":6472,"duration_ms":58497,"significance":"If the projection holds, the paper would represent a substantial advance in modular trapped-ion quantum computing, potentially closing the rate/fidelity gap between remote and local entangling operations and providing a concrete architectural path toward fault-tolerant modular processors. The work is valuable for its explicit end-to-end rate model (Eqs. 2-5), a transparent fidelity budget (Table II) with derivations in App. A2, quantitative comparisons with time-bin and two-photon schemes, a novel routing proof for a rearrangeable butterfly network (App. C), and a power-requirement analysis (App. D). The authors are candid about the main open component, the fast high-extinction 408 nm switch, and about the fact that many parameters are drawn from separate state-of-the-art demonstrations rather than an integrated system. The paper makes falsifiable projections and identifies clear technical milestones, which are strengths for a Perspective.","major_comments":[{"comment":"The headline 'parallel all-to-all' claim at 99.9% fidelity depends on an active butterfly switch network with high extinction at 408 nm. With p_e=0.198, the 0.024% false-herald row in Table II requires p_false ≈ 2×10^-4 (since the residual infidelity is ≈6 p_e p_false ≈ 1.19 p_false). In the butterfly network, each crossing of a live line contributes leakage equal to the switch extinction ratio. The best demonstrated fast switch achieves 30 dB, but at 737 nm [46]; the closest 408-nm device has only >10 dB [71]. At 30 dB, p_false ≈ 10^-3 per crossing, already giving ~0.12% infidelity — above the 99.9% target; at 10 dB the infidelity is ~1-10%. The paper's own Sec. VI.C concedes that this switch 'exceeds demonstrated performances.' Therefore the specific numerical claim of simultaneous all-to-all entanglement at 99.9% fidelity is not supported by current components. Please either provide a","section":"Sec. V-B, App. C, Table II"},{"comment":"The 10^5 s^-1 cm^-2 rate density rests on the simultaneous realization of a full local-operation stack: 225 µs CNOT, 100 µs detection, 100 µs shuttling, a four-ion buffer, 375-µm zones, and an integrated collection/detection efficiency of 0.5% taken from the authors' own integrated-optics work. These have not been demonstrated together. The working point already sits at the edge of the local-operation plateau (the Bell-pair block is 359 µs vs the 325 µs CNOT block); any degradation in t_CNOT, t_detect, t_shuttle, or buffer capacity directly reduces the per-channel rate and hence the density. The claim that remote entanglement 'need not be the bottleneck' would be strengthened by a sensitivity analysis showing how the headline density depends on each Table III parameter and by discussing which combinations are plausibly co-realizable.","section":"Sec. V-B, Table III"},{"comment":"The statement that the scheme is 'local-operation-limited' and 'robust' to collection efficiency is only true under the assumed balance of the pipelined blocks. The paper's own parameters give a Bell-pair generation block (359 µs) slower than the CNOT block (325 µs); achieving the plateau requires the Bell block to be no slower than local operations. If t_CNOT or t_detect increases by even a modest factor, the rate becomes sensitive to collection efficiency again, undermining the 'need not be the bottleneck' narrative. A robustness plot with t_CNOT/t_detect variations would clarify this.","section":"Sec. V-B, Fig. 2"}],"minor_comments":[{"comment":"The preprint has several OCR/formatting errors, including 'distance<8m' (should be 'distance < 8 m'), 'M 2 log2 M' formatting in Table V, the axis label '±Á = ¼' in Fig. 10, and garbled sub/superscripts throughout. These should be corrected in the final version.","section":"General"},{"comment":"The legend 'pd = 0:5% / pd /p2 d' is difficult to read; please relabel the curves clearly (e.g., 'SE, raw', 'SE, distilled', 'SE, distilled, finite local ops').","section":"Fig. 2"},{"comment":"The 'Monte-Carlo-modeled 13.3% buffer-starvation skip fraction' is quoted as a number but the underlying simulation is not described. A brief definition of the model and its assumptions would aid reproducibility.","section":"App. A2c / Sec. V-B"},{"comment":"The notation 're E[e^{iδφ}|δt]' should be 'Re E[...]', and the definition of S_φ(f) as a 'white-frequency-noise' Wiener process is slightly confusing; please clarify the noise-model terminology.","section":"App. B"}],"recommendation":"major_revision","confidential_remarks":"The paper is a well-structured Perspective with a clear, useful quantitative framework. The main issue is that the headline 'parallel all-to-all at 10^5 s^-1 cm^-2 and 99.9% fidelity' is contingent on an active 408 nm switch whose required extinction is about 40 dB, while the closest demonstrated device is >10 dB at 420 nm. This is an acknowledged limitation, but it is load-bearing for the central claim. I recommend major revision, asking the authors to either provide a credible component development path with explicit extinction/loss/speed targets, include a sensitivity analysis, or adjust the headline claim to what is achievable with current technology (e.g., passive network with reduced connectivity). The underlying physics and the linear-scaling argument appear sound; the issue is the gap between projection and demonstrated components."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Straight up: this is a well-put-together architecture paper, and the rate/fidelity model is internally consistent, but the headline '10^5 s^-1 cm^-2 at 99.9% with parallel all-to-all connectivity' is not yet established. It depends on a fast, high-extinction 408 nm switch that has not been demonstrated. The authors say so themselves in Sec. V.C, and the stress-test note quantifies the problem: with the closest demonstrated device (10 dB at 420 nm), the false-herald infidelity alone breaks 99.9%; you need roughly 40 dB at 408 nm. Without that switch, the passive multiport fallback supports only one module pair at a time, so the all-to-all part of the headline goes away. That is a load-bearing caveat, even though it is acknowledged.\n\nWhat is genuinely new: they combine single-photon heralding, recoil correction, Campbell-Benjamin distillation, and integrated photonics into a quantitative architecture that reaches a local-operation-limited plateau. The rate model is transparent; Table III lists all input parameters; App. A derives the error budget; the math is consistent. The density estimate (zone-limited rather than optics-limited) is a real step, and the explicit comparison against time-bin schemes is fair.\n\nThe soft spots are the ones you would expect. The projection rests on many parameters that have not been demonstrated together: 0.5% end-to-end detection efficiency from grating couplers, 225 µs CNOT, 100 µs detection and shuttling, a four-ion buffer, and the 408 nm switch. The fidelity budget also excludes local gate and readout errors—reasonable if the QPU works as assumed, but it brackets the claim. The integrated-optics loss numbers come from the authors' own prior work; that is not a flaw, since those are published results, but it does mean the collection-efficiency input is not independently checked. The phase-noise estimate (σδϕ < 0.02 rad from 10 m patch cords) is plausible but unverified.\n\nNone of this is damning. The paper is honest: it flags the switch as the one component exceeding demonstrated performance, and it gives a fallback. That is the right way to write an architecture paper. The central physics claim—that single-photon heralding plus integrated optics can push remote entanglement to a local-operation limit—holds up; it's the engineering integration that is unproven.\n\nThis paper is for people working on modular trapped-ion architectures, photonic interconnects, and QEC resource estimation. It deserves a serious referee: the modeling is detailed enough to check, the parameter choices are disclosed, and the claim, if realized, matters. Send it to peer review.","headline":"A credible, transparent architecture projection for modular trapped-ion links that deserves peer review, but its headline all-to-all rate/fidelity claim hinges on a 408 nm switch that does not exist yet.","tokens_in":29841,"tokens_out":2696,"would_cite":true,"duration_ms":23604,"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":"This paper projects that a modular trapped-ion quantum computer can generate remote Bell pairs at 99.9% fidelity and 10^5 s−1 cm−2, making the intermodule link no longer the bottleneck.","keywords":["modular quantum computing","trapped ions","remote entanglement","single-photon heralding","entanglement distillation","integrated photonics","QCCD architecture","Bell pairs"],"falsifier":"Build a single-channel Bell factory using the paper's parameters (end-to-end detection efficiency ≈0.5%, excitation probability ≈0.198, 225 µs CNOT, 100 µs detection, 100 µs shuttle, four-ion buffer) and measure the delivered distilled Bell-pair rate and fidelity; if the rate falls well below 858 s−1 at 99.9% fidelity, or if the 408 nm switch cannot reach the assumed extinction and crosstalk levels, the central 10^5 s−1 cm−2 projection fails.","tokens_in":28606,"feed_emoji":"🔗","tokens_out":4265,"duration_ms":34854,"temperature":0.7,"pith_summary":"This paper argues that the intermodule link in a modular trapped-ion quantum computer—historically two orders of magnitude slower and less faithful than local gates—can be made good enough to disappear as the bottleneck. The route is to herald entanglement with a single photon instead of a two-photon coincidence, which makes the success rate scale linearly with collection efficiency and allows compact integrated optics to replace bulky lenses. The authors combine this with coherent recoil correction, leakage detection, and one round of projective distillation, then assemble the pieces into a pipelined architecture that they project delivers distilled Bell pairs at 858 s−1 per channel, 99.9%+ fidelity, and 10^5 s−1 cm−2 with all-to-all module connectivity. If right, modular trapped-ion systems could reach the rate, density, and fidelity that fault-tolerant architectures require; the remaining limit would be local operations that must improve anyway.","feed_headline":"Remote Bell pairs reach 100,000/s per cm2 at 99.9% fidelity","feed_subtitle":"Single-photon heralding plus one distillation round moves the bottleneck from intermodule links to local gates.","key_machinery":"The key mechanism is single-photon heralding on a two-step excitation: each ion is weakly driven on a narrow quadrupole transition to a metastable state and then to a short-lived state that decays by emitting a single photon; interfering the two emitted photons and detecting one photon heralds a Bell pair. Because the success probability scales linearly with detection efficiency (rather than quadratically), compact integrated optics suffice, and the architecture can be pipelined through dedicated QCCD zones whose slowest block—Bell generation, CNOT, or detection—sets the rate. Fidelity is restored by a spin-dependent displacement that undoes photon recoil, leakage checks, and one round of lo","core_discovery":"The central claim is that single-photon heralding, coherent recoil correction, a single projective-distillation round, and trap-integrated photonics compound to decouple entanglement rate, density, and fidelity in a way no previous trapped-ion interconnect achieves. With 88Sr+ ions and parameters drawn from demonstrated components, the paper projects a raw herald probability of 0.18% per 237 ns attempt, a distilled rate of 858 s−1 per channel after buffer-starvation corrections, and a delivered Bell-pair fidelity of about 99.96%, limited mostly by a 0.024% false-herald contribution. The architecture tiles six 375 µm QCCD zones per channel into a Bell-factory density of about 10^5 s−1 cm−2, a","pith_inferences":["A natural next experiment is a two-module prototype with integrated grating couplers and a single channel; confirming a distilled rate near 10^3 s−1 at roughly 99.9% fidelity would validate the linear-scaling premise before building the dense multi-channel array.","Because the rate saturates at the local-operation limit, architectures that invest in faster shuttling and readout—rather than in higher collection efficiency—would see nearly proportional gains in delivered Bell pairs.","The false-herald contribution is the largest residual infidelity, so detector technology with lower dark counts, such as superconducting nanowire detectors, could push the delivered fidelity closer to 99.99%.","The distillation stage converts most imperfections (double excitation, asymmetry, slow phase drift) into rate loss rather than error, suggesting that the link fidelity requirement could be relaxed further if raw-pair rate is plentiful."],"forward_implications":["If the projection holds, a single Bell-factory channel delivers roughly 858 distilled pairs per second at above 99.9% fidelity, sitting near the local-operation plateau where further collection improvements buy at most 1.8×.","The 10^5 s−1 cm−2 density means a few hundred channels fit within cm2-scale module footprints, meeting the per-module Bell-pair counts assumed by surface-code and qLDPC fault-tolerant architectures.","Distillation's insensitivity to slow phase drifts removes the need for global interferometric stabilization; only phase stability across consecutive heralds matters.","The scheme transfers to other trapped-ion species, and in principle to neutral atoms, though recoil heating makes the latter considerably less practical.","The remaining bottleneck shifts to local gates, detection, and shuttling, so improving those raises the performance of the whole system rather than just the link."],"fun_headline_variants":["Remote ion links hit local-gate speeds at 99.9% fidelity","Modular trapped-ion quantum computing: bottleneck shifts to local gates","Single-photon heralding + distillation: remote entanglement fixed","Dense Bell-pair factory removes modular computing bottleneck","Ion modules linked at 100k Bell pairs/s per cm2, 99.9% fidelity"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole rate-and-density projection rests on assembling several individually demonstrated components—225 µs local CNOTs, 100 µs detection windows, 100 µs shuttles, 375 µm zones, and a 408 nm high-extinction optical switch that has not yet reached the required performance—into one parallel pipeline where they operate together without degrading each other, while phase noise between consecutive Bell pairs stays below 0.02 radians.","fun_headline_variants_meta":{"raw":{"variants":["Remote ion links hit local-gate speeds at 99.9% fidelity","Modular trapped-ion quantum computing: bottleneck shifts to local gates","Single-photon heralding + distillation: remote entanglement fixed","Dense Bell-pair factory removes modular computing bottleneck","Ion modules linked at 100k Bell pairs/s per cm2, 99.9% fidelity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0003,"raw_usage":{"total_tokens":1579,"prompt_tokens":765,"completion_tokens":814,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":509,"completion_tokens_details":{"reasoning_tokens":719}},"tokens_in":509,"tokens_out":814,"duration_ms":7479,"temperature":1.0,"reasoning_tokens":719,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T15:31:58.180195+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build a single-channel Bell factory using the paper's parameters (end-to-end detection efficiency ≈0.5%, excitation probability ≈0.198, 225 µs CNOT, 100 µs detection, 100 µs shuttle, four-ion buffer) and measure the delivered distilled Bell-pair rate and fidelity; if the rate falls well below 858 s−1 at 99.9% fidelity, or if the 408 nm switch cannot reach the assumed extinction and crosstalk levels, the central 10^5 s−1 cm−2 projection fails.","supporting_citations":[],"review_version":1}