{"id":"7f6fbae8-2c52-4beb-841c-5b548a053e5e","arxiv_id":"2505.00224","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Foundry-made silicon nanobeam cavities integrated with erbium-doped TiO2 produce cavity-enhanced single photons at telecom C-band wavelengths.","lead":"Researchers built nanobeam optical cavities on a 300 mm silicon photonics foundry line and then coated them with a thin film containing erbium ions. A single erbium ion coupled to a cavity emitted single photons at telecom wavelengths, with its emission rate enhanced about 500 times.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The headline Purcell factor of ≈500 is a lifetime ratio with no independent check that the 13.8 µs cavity lifetime is radiative; surface/interface quenching could produce the same shortening, so the central enhancement claim is not yet secured.","rationale":"The reader's weakest assumption—that the ensemble waveguide lifetime equals the unenhanced lifetime of the single ion—is the same issue I consider load-bearing. The abstract's headline number 'Purcell enhancement up to about 500' is the quantitative content of the central claim; if that number is not a radiative enhancement, the paper is reduced to a foundry-cavity fabrication plus a single-ion g(2) demonstration. The paper's own description of the film supports the concern: the Er delta layer is ~10 nm from interfaces, the film is polyphase with 2.4 nm RMS roughness, and charged defects/grain boundaries are invoked to explain the 43 GHz ensemble linewidth. These are exactly the conditions under which surface/interface quenching competes with radiative decay. The authors report no quantum efficiency measurement and no lifetime measurement with the cavity detuned, so the 6.86 ms baseline is not anchored to the local environment of the measured ion. I also verified the reader's noted numerical inconsistency in the bunching analysis: for B0=1.66 and Γinhom=78.8 MHz, the quoted formula gives Γhom≈22 MHz, not 10.0 MHz (the value 1.66 is closer to what one obtains using the single-scan inhomogeneous width 50.4 MHz and homogeneous width 13.5 MHz). This inconsistency is real but ancillary; it does not by itself overturn the single-photon claim, and it reinforces keeping the verdict CONDITIONAL rather than ACCEPT. The admission in the conclusion that the films were deposited after the wafer left the foundry and are not yet optimized is an honest limitation; it weakens the word 'monolithically' in the title but not the underlying route. Because the load-bearing baseline issue is testable with the existing cryogenic gas-tuning setup, the appropriate verdict remains CONDITIONAL pending that check.","tokens_in":10843,"tokens_out":10749,"duration_ms":115052,"concrete_test":"Using the existing gas-tuning method, measure T1 of the same single ion as a function of cavity detuning over ±κ/2π (κ/2π=5.28 GHz). If the shortening is Purcell, the lifetime should lengthen as 1/[1+(2Δ/κ)^2] when the cavity is moved off the ion resonance; if T1 remains near 13.8 µs or fails to follow the cavity Lorentzian, the reduction is not radiative Purcell enhancement. As a complementary quantitative check, compute the maximum Purcell factor for a dipole in the delta-doped layer using the independently measured Q≈37,400 and simulated V<0.4(λ/n)^3, including the evanescent field overlap, and ask whether any position/orientation can reach ~500; if not, the lifetime ratio cannot be a radiative enhancement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claim, Purcell enhancement P=496(38), is computed in the single-ion characterization section as the ratio of the rutile ensemble lifetime in a waveguide, 6.86(44) ms, to the single-ion cavity lifetime, 13.80(56) µs. This ratio is a Purcell factor only if the ensemble lifetime is the intrinsic radiative lifetime of the isolated ion in the absence of the cavity. The manuscript gives no evidence for that identification. The Er delta-doped layer sits between two 10 nm TiO2 buffers, only ~10 nm from interfaces, on a polyphase film (rutile+anatase) with 2.4 nm RMS roughness, deposited over an etched nanobeam. The authors themselves attribute the 43 GHz ensemble linewidth to charged defects and grain-boundary disorder; the same environment can provide nonradiative decay channels that shorten T1 without any radiative enhancement. If the single ion's true unenhanced lifetime is shorter than 6.86 ms, P=496 overstates the Purcell factor; if the ensemble lifetime is itself nonradiatively limited, the ratio understates radiative enhancement. No quantum efficiency measurement, no off-resonance control, and no independent estimate of the maximum attainable Purcell factor from the measured cavity parameters are provided. The lifetime reduction is real, but labeling it 'Purcell enhancement up to about 500' is not yet supported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports silicon photonic crystal nanobeam cavities fabricated through AIM Photonics' 300 mm foundry platform, achieving measured quality factors above 150,000 and simulated mode volumes below 0.4 (λ/n)^3 in the telecom C-band. After receiving the wafers, the authors deposit a polyphase Er3+:TiO2 film through foundry-provided sensing trenches and characterize the resulting cavity shifts and Q-factor degradation. At T = 3.4 K, they isolate single Er3+ ions in the rutile phase, measure a cavity-shortened optical lifetime of T1 = 13.80(56) µs, compare it with the 6.86(44) ms ensemble lifetime measured in an uncavity waveguide to infer a Purcell factor of P = 496(38), and derive a single-photon coupling rate g/2π = 3.9 MHz. They also report a single-ion linewidth of 57.7 MHz (Voigt fit with 13.5 MHz homogeneous contribution), significant spectral diffusion, and a g(2)(0) = 0.27(4) (background-corrected 0.10) confirming single-photon emission. The paper concludes that this demonstrates a route toward manufacturable deterministic single-photon sources in the telecom C-band.","tokens_in":11077,"tokens_out":6271,"duration_ms":63125,"significance":"If the Purcell-factor interpretation is correct, the work represents an important step toward scalable, foundry-compatible telecom-wavelength single-photon sources: the cavity fabrication is performed on a commercial 300 mm platform with wafer-scale statistics (>5,000 devices), and the single-ion measurements include a direct lifetime ratio, a single-photon autocorrelation, and linewidth characterization. The manuscript is generally careful with error propagation on the central lifetime measurements, and the g(2)(0) result is a genuine single-photon signature. However, the quantitative claim of Purcell enhancement rests on identifying the ensemble lifetime with the intrinsic radiative lifetime, and the title's 'monolithically integrated' framing is stronger than the actual post-foundry deposition flow described in the text.","major_comments":[{"comment":"The Purcell factor P = 496(38) is computed as the ratio of the ensemble lifetime in a waveguide (6.86(44) ms) to the single-ion cavity lifetime (13.80(56) µs). This ratio rigorously quantifies lifetime shortening, but it is a Purcell factor only if the ensemble lifetime equals the intrinsic radiative lifetime of the isolated ion and if the cavity-shortened lifetime is purely radiative. The manuscript provides no quantum efficiency measurement, no off-resonant lifetime control, and no independent bound from the cavity parameters (e.g., the maximum Purcell factor implied by the measured Q ≈ 37,400 and simulated mode volume). Given the polyphase film, the 2.4 nm RMS roughness, and the delta-doped layer placed between two 10 nm TiO2 buffers, surface- and defect-mediated nonradiative decay could shorten the cavity lifetime without radiative enhancement. The authors should either supply evidence that the ensemble decay is radiative-limited (for example, temperature-dependent lifetime or comparison with the known bulk rutile radiative lifetime) or explicitly re-frame the central claim as a measured lifetime reduction rather than a Purcell enhancement.","section":"Single-ion characterization (Fig. 4b and surrounding text)"},{"comment":"The title and abstract describe the device as 'monolithically integrated' on foundry silicon photonics, but the manuscript states on page 9 that 'The results presented here are from post-processing after the wafer left the foundry' and describes front-end ALD integration as a future step. The emitter layer is therefore not part of the foundry process flow, and the device is better described as backend- or hybrid-integrated. The title and abstract should be revised to match the actual process flow, or the term 'monolithically integrated' should be defined in a way that does not imply front-end fabrication.","section":"Title, abstract, and conclusion"},{"comment":"The abstract claims a route toward 'manufacturable deterministic single photon sources,' but the data do not yet support the 'deterministic' qualifier. The observed single-ion count rate is about 170 Hz, and with a shot repetition period of approximately 4T1 (about 7.4 µs for T1 = 13.8 µs), the inferred emission probability per excitation pulse is on the order of 10^-3, far below the on-demand single-photon regime. The wording should be softened to 'promising route toward single-photon sources' or the claim should be qualified with the current count-rate-limited performance.","section":"Abstract and conclusion (deterministic single-photon sources)"}],"minor_comments":[{"comment":"The word 'linewdith' appears in the caption and should be corrected to 'linewidth'.","section":"Fig. 4c caption"},{"comment":"References 14 and 29 are the same paper (Ji et al., ACS Nano 2024) and are cited as separate entries in the reference list.","section":"References"},{"comment":"The phrase 'different reticle than those showed in Figure 2' should read 'different reticle than those shown in Figure 2.'","section":"Fig. 3b caption"},{"comment":"The manuscript reports count rates but does not specify the excitation pulse duration, repetition rate, or optical power at the device; providing these parameters would help the reader assess the single-photon count rate and the claim that spectral diffusion limits the count rate.","section":"Cryogenic characterization paragraph"}],"recommendation":"major_revision","confidential_remarks":"The main novelty of this work is the foundry-scale fabrication of the cavities, but the core Purcell-physics demonstration builds directly on prior work from the same group (refs. 12, 14, 16) in which Er ions in TiO2 on SOI were isolated and Purcell-enhanced using e-beam-fabricated cavities. The editor may wish to weigh whether the lifetime-ratio interpretation as a Purcell factor is sufficiently supported, and whether the 'monolithically integrated' terminology is acceptable for a post-foundry deposition process."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the combination: small-mode-volume silicon nanobeam cavities made on a 300 mm foundry line, with a post-foundry Er:TiO2 coating that yields single ions in the telecom C-band with g(2)(0)=0.27(4). That is a useful step beyond the EBL-only demonstrations the paper cites. The wafer-scale statistics on cavity Q and wavelength are also a real strength; this is the kind of manufacturing data the field needs.\n\nBut I read the stress-test note as landing on a real soft spot, and you should treat it as load-bearing rather than cosmetic. The Purcell factor P=496(38) is just the ratio of the ensemble waveguide lifetime to the single-ion cavity lifetime. That ratio equals a Purcell factor only if the ensemble lifetime is the intrinsic radiative lifetime of the isolated ion and if the 13.8 µs cavity lifetime is purely radiative. The film is polyphase, 2.4 nm RMS roughness, with the ions ~10 nm from interfaces, in an environment the authors themselves say contains charged defects and grain-boundary disorder. Nonradiative quenching could shorten T1 without any radiative enhancement, so the reported P is not secured. The authors give no quantum efficiency measurement, no off-resonance control, and no independent estimate from the cavity parameters. The lifetime reduction is real, but labeling it 'Purcell enhancement up to about 500' overstates what is shown.\n\nA smaller but real problem: the title says 'Monolithically Integrated,' yet the text states the films were deposited after the wafer left the foundry. That is an overclaim. The conclusion is more honest; the title should be changed to something like 'Foundry-fabricated cavities with backend-integrated quantum emitters.' Also, the bunching formula in the text is so poorly typeset that it looks like a numerical inconsistency; once you interpret it as B0 = (Γinhom/Γhom)/(2√(π ln2)) − 1, the numbers do check out. So that specific reader concern does not survive contact with the paper.\n\nNone of this destroys the main engineering message: foundry fabrication of high-Q small-mode cavities is compatible with backend rare-earth doping, and single-photon emission is observed. But the central quantitative claim needs more evidence before it can be taken at face value. I would send this to peer review, but with the clear requirement that the authors either measure or bound the nonradiative contribution to the single-ion lifetime, and fix the title. This paper is for people working on scalable quantum photonics and rare-earth emitters; they should see it, but they should not quote P≈500 as established.\n\nRecommendation: accept for review, with revisions.","headline":"Foundry-made nanobeam cavities plus backend Er:TiO2 is a real scaling step, but the headline Purcell factor of ~500 is a lifetime ratio with no proof that the short lifetime is radiative.","tokens_in":11707,"tokens_out":2351,"would_cite":true,"duration_ms":26487,"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":"Foundry-fabricated silicon cavities with erbium-doped TiO2 yield Purcell enhancement ~500 and single photons in the telecom C-band.","keywords":["silicon photonics","single photon source","erbium","Purcell enhancement","photonic crystal cavity","telecom C-band","foundry fabrication","quantum networking"],"falsifier":"Measure the lifetime of the same single ion after tuning the cavity away from resonance (or after removing the cavity) and compare it with the 6.86 ms ensemble value; if the off-resonant single-ion lifetime is not 6.86 ms, the Purcell factor estimate $P = 496$ collapses. Alternatively, measure $g^{(2)}(0)$ with pulsed excitation and background subtraction to confirm the stated single-photon purity.","tokens_in":1784,"feed_emoji":"⚛️","tokens_out":4421,"duration_ms":94376,"temperature":0.7,"pith_summary":"This paper tries to show that mass-manufacturable silicon photonic chips, fabricated in a standard 300 mm CMOS foundry, can host single erbium ions and turn them into fast, single-photon emitters in the telecom C-band. The route is a nanobeam photonic-crystal cavity with a tiny mode volume, carved into the silicon layer; after the wafer leaves the foundry, a thin erbium-doped TiO2 film is deposited through a sensing trench in the oxide cladding. A single ion coupled to such a cavity has its optical lifetime shortened from 6.86 ms to 13.8 microseconds, a Purcell factor around 500, and the autocorrelation $g^{(2)}(0)=0.27(4)$ confirms one-photon-at-a-time emission. If true, this moves deterministic single-photon sources for fiber networks from small-batch electron-beam fabrication to wafer-scale production.","feed_headline":"Foundry chips boost single erbium ions 500x for telecom quantum light","feed_subtitle":"Single erbium ions in foundry cavities emit one telecom photon at a time—a step to wafer-scale quantum networks.","key_machinery":"The carrying object is the photonic-crystal nanobeam cavity: a silicon waveguide with a parabolic taper of hole pitches forming a resonant midgap state, read out through a bus waveguide and a Sagnac loop mirror. Its combination of high quality factor and small mode volume gives a large Purcell factor, the mechanism that shortens the erbium ion's emission lifetime. The second piece is the backend deposition of an Er$^{3+}$:TiO$_2$ film through a trench in the oxide cladding, placing roughly 200 ions within the cavity mode volume and allowing individual ions to be addressed; cavities are tuned onto the rutile ensemble line at 1520.5 nm by condensing nitrogen gas onto the device.","core_discovery":"The central claim is that monolithically integrated, cavity-enhanced single erbium ions emitting in the telecom C-band can be made on a scalable 300 mm silicon photonic foundry platform. Foundry-fabricated nanobeam cavities, with quality factors above 150,000 before coating and about 37,400 for the single-ion device, and simulated mode volumes below $0.4\\,(\\lambda/n)^3$, isolate individual erbium ions in the rutile phase of a back-end-deposited Er$^{3+}$:TiO$_2$ film. Resonant single ions show a Purcell-enhanced lifetime $T_1 = 13.80(56)\\,\\mu s$ compared with the 6.86(44) ms ensemble lifetime, yielding $P = 496(38)$, an inferred single-photon coupling rate $g/2\\pi = 3.9$ MHz, and $g^{(2)}(0)=0.27(4)$, indicating single-photon emission. The paper also reports wafer-scale characterization of more than 5,000 cavities per reticle, linear tuning of resonance with hole diameter, and only modest degradation of cavity quality factor after film deposition.","pith_inferences":["A natural next step is moving the erbium-doped TiO2 deposition into the foundry's front end via atomic-layer deposition, which the paper notes is already CMOS-compatible; if film quality improves, single-ion optical linewidths could drop from tens of megahertz toward the radiative limit.","The bunching analysis attributes most of the single-ion linewidth to spectral diffusion; this suggests surface passivation or thicker buffer layers, rather than higher-Q cavities, may be the fastest route to indistinguishable photons.","Because the highest-Q cavities on the chip were outside the gas-tuning range used here, devices that land on resonance with the erbium ensemble could push the inferred single-photon coupling rate well beyond $g/2\\pi \\approx 3.9$ MHz."],"forward_implications":["Wafer-scale, reproducible fabrication of Purcell-enhanced single-photon sources in the C-band becomes possible, removing electron-beam lithography as a throughput bottleneck.","Because erbium's 1.5 micrometer transition matches the loss minimum of optical fiber, these sources could plug directly into existing fiber-based quantum networks and repeaters.","The backend trench deposition decouples foundry front-end fabrication from qubit integration, so different quantum materials could be added post-fabrication on the same platform.","With cavities retaining quality factors above 100,000 after coating and with gas-based resonance tuning, many devices on a single chip could be brought into resonance, enabling arrays of single-photon emitters."],"supporting_citations":[{"why":"Shows Purcell enhancement of erbium ions in TiO2 on silicon nanocavities, the approach this work transfers to a foundry platform.","marker":"[12]"},{"why":"Demonstrates isolation of individual erbium emitters in anatase TiO2 on silicon photonics using electron-beam lithography, the baseline for single-ion work.","marker":"[16]"},{"why":"Reports nanocavity Purcell enhancement of Er in TiO2 grown by atomic layer deposition, supporting the CMOS-compatible integration path.","marker":"[14]"},{"why":"Demonstrates high-quality small-mode-volume photonic crystal cavities fabricated in a foundry process, establishing the fabrication benchmark.","marker":"[23]"},{"why":"Provides the deterministic nanobeam cavity design with parabolic hole taper used for the resonant midgap mode.","marker":"[28]"},{"why":"Gives the Purcell effect relation used to convert the observed lifetime reduction into the enhancement factor.","marker":"[26]"},{"why":"Shows sub-MHz single erbium linewidths in a bulk crystal with a nanophotonic cavity, the linewidth target for this film.","marker":"[35]"}],"fun_headline_variants":["Foundry cavities give single erbium ions 500x emission boost","Wafer-scale quantum emitters in telecom band from foundry chips","Erbium ions in foundry chips: single photons with 500x Purcell gain","Monolithic telecom single-photon source on silicon photonics","C-band single-photon source from scalable silicon foundry"],"cache_read_input_tokens":13696,"weakest_assumption_plain":"The Purcell factor assumes the 6.86 ms ensemble lifetime of erbium in the rutile waveguide equals the intrinsic radiative lifetime of the single ion inside the cavity; if local surfaces, phase differences, or film structure change that unenhanced lifetime, the reported ~500 enhancement is not the true Purcell factor, although the lifetime reduction is still real.","fun_headline_variants_meta":{"raw":{"variants":["Foundry cavities give single erbium ions 500x emission boost","Wafer-scale quantum emitters in telecom band from foundry chips","Erbium ions in foundry chips: single photons with 500x Purcell gain","Monolithic telecom single-photon source on silicon photonics","C-band single-photon source from scalable silicon foundry"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000244,"raw_usage":{"total_tokens":1521,"prompt_tokens":923,"completion_tokens":598,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":539,"completion_tokens_details":{"reasoning_tokens":505}},"tokens_in":539,"tokens_out":598,"duration_ms":6670,"temperature":1.0,"reasoning_tokens":505,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:47:57.296540+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the lifetime of the same single ion after tuning the cavity away from resonance (or after removing the cavity) and compare it with the 6.86 ms ensemble value; if the off-resonant single-ion lifetime is not 6.86 ms, the Purcell factor estimate $P = 496$ collapses. Alternatively, measure $g^{(2)}(0)$ with pulsed excitation and background subtraction to confirm the stated single-photon purity.","supporting_citations":[{"cited_title":"M.; Solomon, M","cited_arxiv_id":null,"evidence_quote":"Shows Purcell enhancement of erbium ions in TiO2 on silicon nanocavities, the approach this work transfers to a foundry platform."},{"cited_title":"M.; Gupta, S.; Grant, G","cited_arxiv_id":null,"evidence_quote":"Demonstrates isolation of individual erbium emitters in anatase TiO2 on silicon photonics using electron-beam lithography, the baseline for single-ion work."},{"cited_title":"T.; Grant, G","cited_arxiv_id":null,"evidence_quote":"Reports nanocavity Purcell enhancement of Er in TiO2 grown by atomic layer deposition, supporting the CMOS-compatible integration path."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates high-quality small-mode-volume photonic crystal cavities fabricated in a foundry process, establishing the fabrication benchmark."},{"cited_title":"Deterministic Design of Wavelength Scale, Ultra-High Q Photonic Crystal Nanobeam Cavities","cited_arxiv_id":null,"evidence_quote":"Provides the deterministic nanobeam cavity design with parabolic hole taper used for the resonant midgap mode."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Purcell effect relation used to convert the observed lifetime reduction into the enhancement factor."},{"cited_title":"P.; Uysal, M","cited_arxiv_id":null,"evidence_quote":"Shows sub-MHz single erbium linewidths in a bulk crystal with a nanophotonic cavity, the linewidth target for this film."}],"review_version":1}