{"id":"d859673b-74db-437f-9597-7ded12727de6","arxiv_id":"2607.25062","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.5,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A wafer-scale hybrid 2D–3D photonic circuit focuses 405–880 nm light onto three ion sites at 5 µm pitch with −27 dB average intensity crosstalk and shows selective Ca+ repumping.","lead":"A chip combines waveguide lenses with 3D-printed micromirrors to send many laser colors to individual trapped ions. That could ease a main bottleneck in scaling ion-based quantum computers.","discovery_kind":"new_method","skeptic_critique":{"model":"moonshotai/kimi-k3","headline":"The −27 dB individual-addressing figure is an offline profiler result at two wavelengths; at the ion plane the claim is supported only by a 120 µm-off-focus Ba+ stray-light scan and one limited-contrast single-λ Ca+ repump — the 5 µm-pitch on-ion crosstalk that anchors the abstract was never measure","rationale":"The reader's weakest_assumption already identifies exactly this load-bearing gap: offline crosstalk plus a limited-contrast single-λ demo and off-focus Ba+ data do not yet establish practical multi-λ individual control, given the charging and trap-instability failures. My independent read of the full text reaches the same conclusion and finds no additional concern of comparable weight. Secondary points — the \"average\" crosstalk metric (peak field, ~4.5% residual amplitude at −27 dB intensity, is what enters coherent spectator errors), the 50% FWHM excess over design, and 35 dB loss at 405 nm — are disclosed and attributed to correctable causes (aberrations, waveguide platform), so they are not load-bearing for the claim as written. The fabrication and optical characterization are thorough and credible; the paper is forthright about its ion-side limitations. The reader's CONDITIONAL verdict with HIGH confidence is correct, and the condition (fix charging/ground plane, demonstrate quantified multi-λ on-ion addressing at the focus) is precisely the concrete test above. No verdict change.","tokens_in":16486,"tokens_out":1939,"duration_ms":65579,"concrete_test":"On a device with the proposed fixes (ITO-coated planar-lens facet and ground plane), trap a 5 µm-spaced 40Ca+ pair stably at the designed focus and drive the 729 nm qubit transition through one entrance waveguide; measure the addressed vs spectator excitation ratio (Rabi-frequency squared ratio or shelving probability asymmetry) and compare against the −27 dB offline prediction; repeat at ≥2 wavelengths (e.g., 866 nm). If on-ion crosstalk matches, claim (b) is confirmed; if it degrades by >6 dB or ions cannot be held at the focus, the abstract's addressing claim should be scoped to offline optics only.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim bundles three assertions: (a) broadband foci 405–880 nm, (b) −27 dB average intensity crosstalk at 5 µm pitch enabling three-ion individual addressing, (c) ion-trap compatibility. For the central engineering claim (b), the only 5 µm-pitch crosstalk evidence is offline confocal-profiler beam maps at 515 and 730 nm (Fig. 3c) plus simulation; \"broadband\" addressing at other wavelengths is inferred from FWHM retention (Fig. 3b), not crosstalk. The in-situ check that should have closed the loop never happened at the relevant location: dielectric charging (Supp. Note 1, several-thousand V/m stray fields) blocked Ba+ shuttling to the focus, so Fig. 4b measures intensity 120 µm from beam center — far-sidelobe falloff, not nearest-neighbor crosstalk — referenced to an offline-inferred peak. The only ion-plane addressing demonstration (Fig. 4d) is single-wavelength (866 nm repump), two ions, unquantified contrast, with the paper itself citing unoptimized axial position, saturation, and radial misalignment; the Ca+ trap then failed (suspected photo-induced Si conductivity change) before any other wavelength could be tested on ions. So the inferential step \"offline −27 dB implies individual multi-λ qubit control on ions\" is untested precisely where aberrations, pointing drift, charging-induced position shifts, and per-wavelength behavior could degrade it. The SI's own ~100 µs spectator coherence estimate concedes residual light is marginal. This is a completeness gap, not an inconsistency: the optical core (FWHM vs λ, loss budget, −27 dB offline) looks solid and internally consistent, and the authors disclose the failures honestly.","agreement_with_reader":"agree"},"referee_report":{"model":"moonshotai/kimi-k3","summary":"The manuscript presents a monolithic photonic integrated circuit for trapped-ion light delivery that replaces diffractive grating couplers with a hybrid 2D/3D optical system: a lithographically defined planar waveguide lens (Al2O3/SiO2 platform) feeds a biconic-aspherical micromirror printed by two-photon polymerization (2PP) and metallized with the same Al layer that forms the surface-trap electrodes. The authors demonstrate wafer-scale fabrication on 6-inch wafers, including (to my knowledge novel) lithography and etch post-processing directly on 2PP-printed optics. Offline confocal-profiler characterization shows tight focusing at h = 75 µm across 405–880 nm (axial FWHM 0.67–1.46 µm), simulated and measured nearest-neighbor intensity crosstalk of about −27 dB at 5 µm pitch at two wavelengths (515/730 nm measured; 532/729 nm simulated), simultaneous three-spot illumination, and a full fiber-to-ion loss budget. Ion experiments trap 138Ba+ and 40Ca+ above the devices: dielectric charging prevented shuttling Ba+ to the beam focus, so the on-ion optical measurement is a stray-light scan 120 µm from beam center; with Ca+, selective 866 nm repumping of one ion of a 5 µm-spaced pair is shown qualitatively before trap failure ended further tests. Charging mechanisms are characterized in the Supplementary Information and mitigation strategies (ITO coating, ground plane) are proposed.","tokens_in":16968,"tokens_out":3453,"duration_ms":145059,"significance":"If the device-level results hold, this is a genuine architectural advance for integrated-photonics ion traps: a single shared optical element delivering individually addressable foci over 405–880 nm removes the per-wavelength grating multiplier that currently dominates PIC footprint in QCCD-style architectures. Specific strengths worth naming: (i) the first demonstration of high-resolution lithography and wet etch on top of 2PP micro-optics, enabling monolithic integration of printed optics with trap electrodes; (ii) broadband focal maps at six wavelengths with quantitative FWHM and a complete, packaged-as-trapped fiber-to-ion loss budget (Fig. 3e), which is rare and useful; (iii) measured (not merely simulated) nearest-neighbor crosstalk at two wavelengths and simultaneous three-spot operation; (iv) UHV packaging surviving four bakeouts to 125 °C; and (v) unusually candid reporting and quantitative SI characterization of the dielectric-charging failure mode (Supp. Note 1), which is valuable to the community independent of the headline device. The ion-plane validation is, however, substantially weaker than the abstract implies, and the paper's impact statement should be calibrated","major_comments":[{"comment":"The headline claim — 'address three individual ions from λ = 405–880 nm with −27 dB average intensity crosstalk at 5 µm pitch' — bundles three assertions whose evidential status differs sharply. The −27 dB figure and the three-spot demonstration are offline confocal-profiler measurements at 515/635/730 nm (Fig. 3c,d); crosstalk at the other four wavelengths is inferred only from FWHM retention (Fig. 3b), not measured. On ions, the only addressing datum is a single-wavelength (866 nm), two-ion repump with unquantified contrast (Fig. 4d), and the Ba+ measurement samples the beam 120 µm off focus. The abstract and Discussion should state plainly that individual-addressing crosstalk is a beam-profiler result validated on ions at one wavelength, or the authors should supply additional on-ion data. As written, a reader of the abstract would conclude that multi-wavelength on-ion addressing at −","section":"Abstract; §6 Discussion"},{"comment":"The selective-repump demonstration is the manuscript's only on-ion test of individual addressing, yet the contrast is reported only qualitatively ('weaker fluorescence', two Gaussian fits without quoted amplitudes). The authors themselves cite three contrast-limiting factors (axial mispositioning, saturation, radial misalignment), each of which decouples fluorescence contrast from the underlying intensity crosstalk. Please report the fitted peak ratio with uncertainties, the estimated on-ion intensity ratio after accounting for saturation, and the integration time/repetitions. Without a number, this figure cannot support or bound the on-ion crosstalk claim at all.","section":"§5, Fig. 4d"},{"comment":"The Ba+ measurement is described as characterizing 'crosstalk from the integrated 455 nm delivery path', but the data span positions ~65–185 µm from the cutout center, with the nearest point ~120 µm from the expected focus — i.e., far-sidelobe falloff, referenced to a peak intensity inferred from offline loss measurements (Fig. 3e) rather than measured on-ion. This is a legitimate and interesting stray-light measurement (relevant to spectator-zone isolation, as in Supp. Note 2), but it does not constrain the 5 µm-pitch nearest-neighbor crosstalk that anchors the abstract. The text should be explicit about which claim this measurement supports. Relatedly, the intensity inference chain (bright-time binning → scattering rate → saturation parameter with a ~50% S1/2 population correction under strong 493 nm saturation) carries model dependence; please quantify the uncertainty on the −57 dB fi","section":"§5, Fig. 4b (crosstalk scan, mislabeled — see minor comments)"},{"comment":"Measured axial FWHM is on average 50% larger than the geometric-NA prediction, attributed to non-uniform aperture illumination and spherical aberration. Since crosstalk at 5 µm pitch depends on the full point-spread function — including aberration-driven skirts — the −27 dB value measured at 515/730 nm need not transfer to wavelengths where the aberration balance differs (e.g., 405 nm and 880 nm, the band edges that motivate the 'broadband' claim). Either measured crosstalk at additional wavelengths or full-wave simulated crosstalk versus wavelength (the FDTD capability used for Fig. 3c presumably permits this) is needed to support the band-wide addressing claim.","section":"§4, Fig. 3b,c"},{"comment":"The spectator-coherence estimate assumes 'the same relative crosstalk profile applies at 493 nm' and arrives at T_coh ≈ 100 µs, which the authors concede is marginal for gate operations. Given that measured crosstalk and FWHM vary across the band (Fig. 3b,c), and that the 455 nm profile was measured far off focus, this extrapolation needs justification or error bars; otherwise the SI's own numbers suggest the device as-built does not yet protect spectator ions, which bears directly on the scalability argument in §6.","section":"Supp. Note 2, Eq. (3)"}],"minor_comments":[{"comment":"Panel lettering is inconsistent: the caption assigns (b) to Ca+ levels, (c) to the 455 nm crosstalk scan, and (d) to selective repumping, but the main text says 'Figure 4b shows the measured 455 nm intensity' and 'Trapping of 40Ca+ ... (Fig. 4d)'. Please reconcile.","section":"Fig. 4 caption vs. main text"},{"comment":"The Methods refer to packaging photographs as 'Fig. 4(a)' and 'Fig. 4(b)', which do not correspond to the Fig. 4 caption (ion data). Presumably a separate packaging figure was intended; please fix the cross-references.","section":"Methods, UHV packaging"},{"comment":"Please define 'average intensity crosstalk' precisely: is −27.6 dB the ratio of intensity integrated over the neighboring spot's FWHM area to the addressed-spot peak, a peak-to-peak ratio, or an average over the two neighbors? The number's meaning (and comparability with the −36 dB grating result of Ref. 17) depends on this.","section":"§4 / Fig. 3c"},{"comment":"Typos: 'The ions fluorescence contrast' (missing apostrophe); 'florescence' in the Fig. 4d caption. Also 'All scale bars are 5 µm' in the Fig. 4 caption conflicts with the '200 um' scale bar visible in Supp. Fig. 5a.","section":"§5"},{"comment":"The 35 dB fiber-to-ion loss at 405 nm deserves one sentence of discussion: for Ba+ this band covers photoionization/shelving transitions, and 35 dB may be prohibitive for some uses even if 'not a limit of the architecture'. A power budget for the most demanding wavelength would strengthen the broadband claim.","section":"Fig. 3e"},{"comment":"The claim 'Alignment between the substrate and the printed optics is ≤1 µm in all directions' should state how this was measured (alignment marks? confocal metrology?) and give the observed distribution across the 50 mirrors per wafer, since pointing error directly maps to on-ion crosstalk via the 5 µm pitch.","section":"§2"},{"comment":"Ref. 24's DOI ('10.1103/l1cn-28kv') appears malformed; please verify. Several arXiv-only references (4, 25, 32, 33) may have since appeared in print.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The device physics and fabrication work are strong and, in my view, publishable in principle; the gap is between the abstract's framing and the ion-plane evidence, which the authors themselves partially acknowledge in §5. Since the Ca+ trap failure and Ba+ charging likely preclude quick acquisition of new on-ion data, a revision that tempers the headline claims and quantifies Fig. 4d may be the realistic path; the editor may wish to signal that such a revision would be acceptable in lieu of new experiments."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The new piece here is the hybrid stack: Al2O3 planar waveguide lens plus a wafer-scale 2PP biconic micromirror under a monolithic surface trap, giving one focusing path from 405–880 nm instead of a forest of narrowband gratings. That architecture is the real contribution, and the optical core is done carefully.\n\nFig. 3 is the strong part. Multi-λ focal maps, axial/radial FWHM across the band, simulated and measured nearest-neighbor intensity near −27 dB at 5 µm (515/730 nm), three-spot images, and a fiber-to-ion loss budget with the mirror itself only ~2 dB. Post-2PP lithography and Al etch on the printed optics is nontrivial process work and they document it. They also trap both Ca+ and Ba+, show selective 866 nm repump on a two-ion crystal, and are frank in the Discussion and SI about dielectric charging (kV/m-scale stray fields) and the Ca+ trap dying mid-campaign.\n\nThe soft spot is exactly what the stress-test flags, and it is real but proportional: the abstract’s “address three ions… −27 dB… individual repumping” bundles offline profiler crosstalk with incomplete on-ion closure. Ba+ intensity data are 120 µm off the designed focus because charging blocked shuttling; the Ca+ demo is one wavelength, limited contrast, and they never got other λ on ions. SI’s own ~100 µs spectator estimate shows residual light is still a systems issue. So “practical multi-λ individual qubit control on ions” is not fully demonstrated yet—it is a credible optical platform plus partial ion validation. Math/design (FDTD, Zemax, BEM) and citations look standard and honest; no circularity games.\n\nThis is for ion-trap hardware and quantum-photonics people who care about QCCD optical delivery scaling. Worth a serious referee. I would engage: cite the optics/architecture, push them on ground-plane/ITO fixes and a cleaner multi-λ on-ion crosstalk measurement. Send to peer review.","headline":"Real broadband hybrid PIC with solid offline optics; ion-side individual-addressing claim is only partly closed and the abstract overreaches a bit.","tokens_in":17854,"tokens_out":532,"would_cite":true,"duration_ms":13987,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A single on-chip planar lens and 3D-printed micromirror can address three trapped ions across 405–880 nm with −27 dB crosstalk at 5 µm pitch.","keywords":["trapped ions","photonic integrated circuit","two-photon polymerization","broadband addressing","surface electrode trap","optical crosstalk","QCCD","micromirror"],"falsifier":"Load a stable multi-ion crystal directly at the designed focus, deliver several control wavelengths through the same path, and measure per-ion gate or repump fidelity with nearest-neighbor intensity crosstalk remaining at or below −27 dB without dielectric-charging or substrate-photoconductivity failures.","tokens_in":17455,"feed_emoji":"⚛️","tokens_out":927,"duration_ms":19445,"temperature":0.7,"pith_summary":"Trapped-ion quantum computers need many laser wavelengths delivered with micron precision to every qubit, but free-space optics do not scale and on-chip grating couplers are narrowband, so each wavelength needs its own footprint-hungry coupler. This paper shows a hybrid photonic circuit that routes light in a broadband alumina waveguide, expands it with a planar waveguide lens, and focuses it with a wafer-scale two-photon-polymerized micromirror coated as part of the surface-electrode trap. One shared optical path produces three individually addressable foci from 405 to 880 nm with about −27 dB average intensity crosstalk at 5 µm spacing—the pitch of a typical ion chain. The authors trap calcium and barium ions above the devices, measure stray light with barium, and show selective on-chip 866 nm repumping of one calcium ion in a pair. If the approach holds up under full multi-wavelength operation, optical control layers need only scale with the number of ion registers, not with the number of wavelengths.","feed_headline":"One on-chip mirror addresses three ions across 405–880 nm","feed_subtitle":"Planar lens plus 3D-printed micromirror hits −27 dB crosstalk at 5 µm pitch for trapped-ion control","key_machinery":"The hybrid imaging path: a lithographic planar waveguide lens that expands and collimates the guided mode, paired with a wafer-scale 2PP biconic-aspherical micromirror that redirects and focuses it ~75 µm above the chip, mapping each entrance waveguide to a discrete ion-plane focus.","core_discovery":"A monolithic 2D–3D photonic integrated circuit—planar waveguide lens plus two-photon-polymerized biconic micromirror integrated with a surface Paul trap—delivers individually addressable beams for three ions over λ = 405–880 nm with −27 dB average intensity crosstalk at 5 µm pitch, supports trapping of 40Ca+ and 138Ba+, and demonstrates individual on-chip 866 nm repumping of calcium.","pith_inferences":["If ITO facet coatings and ground planes fix charging as proposed, the same stack could host full Ca+ or Ba+ gate sets without free-space delivery for those wavelengths.","Correcting coma with extra printed elements could extend the three-spot field of view, enabling longer chains per zone without new grating banks.","The architecture’s value compounds most in multi-species or multi-zone machines where wavelength count multiplies faster than zone count."],"forward_implications":["One broadband emitter can replace many wavelength-specific grating couplers per zone, shrinking the photonic footprint of each QCCD interaction region.","Optical delivery element count can scale with the number of ion registers rather than with the number of control wavelengths.","Fewer transport steps per clock cycle become feasible if a single zone can address more ions with the same optics.","Wafer-scale additive manufacturing of 3D micro-optics becomes a practical route to custom free-space beam shaping inside surface-electrode traps."],"fun_headline_variants":["On-chip lens plus 3D micromirror addresses three ions at 405–880 nm","Planar waveguide and printed mirror hit −27 dB crosstalk at 5 µm","Monolithic 2D–3D PIC delivers broadband beams to individual ions","Wafer-scale biconic micromirror enables 405–880 nm ion addressing","2D lens with 3D-printed mirror traps and repumps ions on chip"],"cache_read_input_tokens":128,"weakest_assumption_plain":"That offline beam maps and a limited single-wavelength ion demo are enough to claim practical multi-wavelength individual control, even though charging blocked full shuttling to the focus and trap instability stopped further wavelength tests on ions.","fun_headline_variants_meta":{"raw":{"variants":["On-chip lens plus 3D micromirror addresses three ions at 405–880 nm","Planar waveguide and printed mirror hit −27 dB crosstalk at 5 µm","Monolithic 2D–3D PIC delivers broadband beams to individual ions","Wafer-scale biconic micromirror enables 405–880 nm ion addressing","2D lens with 3D-printed mirror traps and repumps ions on chip"]},"model":"grok-4.5","effort":"low","cost_usd":0.004538,"raw_usage":{"total_tokens":1308,"prompt_tokens":773,"num_sources_used":0,"completion_tokens":94,"cost_in_usd_ticks":45384000,"prompt_tokens_details":{"text_tokens":773,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":441,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":773,"tokens_out":94,"duration_ms":8224,"temperature":1.0,"reasoning_tokens":441,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T02:17:30.642960+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Load a stable multi-ion crystal directly at the designed focus, deliver several control wavelengths through the same path, and measure per-ion gate or repump fidelity with nearest-neighbor intensity crosstalk remaining at or below −27 dB without dielectric-charging or substrate-photoconductivity failures.","supporting_citations":[],"review_version":1}