{"id":"5f7e40e5-90fb-4280-827d-73f5bd78e268","arxiv_id":"2608.13177","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The authors demonstrate entanglement between a telecom photon and a stored 979 nm photon across more than 16,000 temporal modes, including over the Geneva metropolitan fiber network.","lead":"Researchers entangled a telecom photon that traveled through 25.3 km of fiber with a 979 nm photon stored in a rare-earth crystal memory for 125 microseconds. The system supports more than 16,000 temporal modes, a record for matter-telecom entanglement, and was tested over the Geneva metropolitan fiber network.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'stored modes' count is the Schmidt number of the source biphoton JTA, not a directly measured number of independently usable memory modes; the paper itself leaves the operational link to repeater speed-up open.","rationale":"The paper's strongest measured result is the CHSH violation after storage and fiber propagation; that part is well supported by the reported S values and the consistency check S = 2 sqrt(2) V. The weakest load-bearing step is the identification of the Schmidt mode number of the biphoton state with the number of temporal modes stored and usable for multiplexing. This is exactly the assumption the reader flagged, and the authors themselves state in the Discussion that the speed-up interpretation is open. I do not see a fatal flaw: Eq. (3) is a legitimate lower bound for the Schmidt number, the memory's time-bandwidth product makes a large mode count plausible, and the paper is appropriately cautious in the Discussion. However, because the headline 'more than 8000 modes' is not directly witnessed by the CHSH data and the relation to repeater performance is unresolved, a CONDITIONAL verdict is appropriate. My read does not move the verdict; it confirms the reader's conditional assessment. I would not raise this to REJECT because the central entanglement demonstration is credible and the mode count, while overstated operationally, is a defensible quantitative measure of the source/memory time-frequency capacity.","tokens_in":13920,"tokens_out":10802,"duration_ms":114728,"concrete_test":"Use the rate formalism of Ref. [52] with the Table I parameters (250 MHz memory bandwidth, storage times 63 and 125 us, g2 ~ 100, rectangular 7.65 ns post-filter) to compute the continuous-pump DLCZ-type repeater rate relative to an ideal single-mode pulsed source. If the resulting speed-up is significantly less than N = 8235 or 16340, the mode count should be presented as a source Schmidt number rather than an operational multiplexing gain. As a complementary check, bootstrap the raw g2 histograms to attach an uncertainty to T_mode and N and verify whether the field-test value remains above 8000 at the upper end of that uncertainty.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline figure rests on N = T_pump/T_mode with T_mode = 7.65 ns extracted from the measured g2 width and applied in Table I to every storage time. This N is a property of the SPDC two-photon state, computed via Eq. (3) under the assumption that the JTA has factorizable phase; it is not a measurement of the number of temporal modes that the AFC memory can store and retrieve independently. The CHSH tests are insensitive to this number: they post-select a single 7.65 ns detection window inside the central Franson peak, so each S value certifies entanglement of the post-selected time-bin subspace, not the presence of many independent entangled modes. The paper's own Discussion states that 'it is an open question if this number represents the expected speed-up with respect to a single-mode repeater,' and notes that filtering the conditional modes is infeasible in practice. A further concrete fragility is that Table I quotes N without uncertainty; the field-test value 8235 = 63 us / 7.65 ns is only about 4% above the 'more than 8000' threshold, so a modest systematic error in the mode-duration calibration would invalidate that specific headline number. None of this weakens the CHSH entanglement evidence, which is credible; it weakens the interpretation of the mode count as demonstrated multiplexing capacity.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports a quantum repeater node built from a 171Yb3+:Y2SiO5 atomic frequency comb (AFC) memory and a bandwidth-matched SPDC entangled-photon source. The authors demonstrate energy-time entanglement between a telecom photon at 1553 nm transmitted through optical fiber (0.5–25.3 km spools and a 5.66 km deployed Geneva metropolitan link) and a 979 nm photon stored in the AFC memory for up to 125 µs, using Franson interferometry and CHSH Bell tests with S values between 2.65 and 2.73. They introduce a Schmidt decomposition of the joint temporal amplitude to estimate the effective temporal mode number N = T_pump/T_mode, with T_mode = 7.65 ns, and report up to 16,340 modes at 125 µs storage in the lab and 8,235 modes at 63 µs in the field. The paper also characterizes memory efficiency (η0 = (19±1)%) and coherence time (TM = (76.6±11.0) µs), and discusses the path toward spin-wave AFC repeaters.","tokens_in":14163,"tokens_out":6801,"duration_ms":55998,"significance":"Assuming the results hold, the work is significant: it provides credible CHSH-based evidence for light-matter entanglement after 25 km propagation and microsecond-scale storage in a rare-earth memory, and the deployed urban fiber test is an important step toward practical repeater nodes. The Schmidt-decomposition methodology is a useful quantitative tool, and the authors are careful to compare a measured lower bound with theory (Fig. 2(c)) and to report memory efficiency and lifetime in detail. However, the headline mode-count claim is not directly supported by the Bell tests: the CHSH analysis post-selects a single 7.65 ns window, and the N values are computed from the source joint temporal amplitude rather than measured as independently addressable storage modes. The authors themselves flag this as an open question in the Discussion. Because the paper's central contribution is advertised as 'more than 8000 modes,' the relationship between N and demonstrated multiplexing capacity needs to be clarified before publication.","major_comments":[{"comment":"The headline 'storage of more than 8000 modes' is derived from N = T_pump/T_mode, where T_mode = 7.65 ns is extracted from the measured g2 correlation width. This N is the Schmidt number of the SPDC two-photon state, not a directly measured number of independently addressable temporal modes stored in the AFC memory. The CHSH tests in Table I post-select a single detection window of duration T_mode, so each S value certifies entanglement of the post-selected time-bin subspace and does not certify that all N computed modes are entangled or independently usable. Please either soften the wording to 'source Schmidt modes' or provide explicit evidence that the memory can store this many independent modes.","section":"Multimode analysis through Schmidt decomposition, Eq. (2)–(3), Fig. 2(c), Table I"},{"comment":"N is reported without any uncertainty. The field-test headline value N = 8235 = 63 µs / 7.65 ns exceeds the 'more than 8000' threshold by only about 4%, so a modest systematic error in T_mode (whose extraction shows a binning dependence in Fig. 2(c)) could invalidate the specific claim. Please provide an uncertainty analysis for T_mode and propagate it to N, or state the claim with a conservative margin.","section":"Table I and Fig. 2(c)"},{"comment":"The authors state that 'it is an open question if this number represents the expected speed-up with respect to a single-mode repeater' and that filtering conditional modes is infeasible in practice. This caveat is not reflected in the abstract or title, which present the mode number as an achieved storage capacity. Since the abstract currently says 'storing 8235 modes,' I recommend the caveat be moved forward or the claim be explicitly qualified as a source-level estimate.","section":"Discussion, first paragraph"}],"minor_comments":[{"comment":"'based on measurements as shown in (e) and (f)' should probably refer to (d) and (e), since (f) is the resulting efficiency plot.","section":"Fig. 1 caption, panel (f)"},{"comment":"The dashed line is described as 'the expected theoretical value of N,' while the text refers to it as 'the theoretical lower bound N'; please align the terminology.","section":"Fig. 2(c) caption"},{"comment":"The notation for the lower bound N is not visually distinct from the Schmidt number N in Eq. (2); consider using an underline or a different symbol to avoid confusion.","section":"Notation, Eq. (3)"},{"comment":"The row for the 32 µs field experiment lacks idler/signal/coincidence rates; please state explicitly in the caption why these are omitted.","section":"Table I"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about the mode-count claim is valid and should be addressed. The authors' own caveat is a strength, but the framing in the abstract overstates the mode-count claim. No concerns about academic integrity; the self-citations are to prior characterization work and are appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe stuff you should know first: the entanglement results are credible and the paper is worth reading. They show CHSH violations S=2.65–2.73 for telecom photons after up to 25.3 km fiber and 979 nm photons stored up to 125 µs in a 171Yb:Y2SiO5 AFC memory, plus a field test over 5.66 km of Geneva fiber. The memory characterization is careful: efficiency versus storage time, lifetime 76.6±11 µs, and the measured visibilities are consistent with the g2 values. That part is solid.\n\nWhat is genuinely new is the combination: broadband AFC memory (250 MHz) plus bandwidth-matched SPDC source, operated as an elementary repeater node, with a Schmidt-decomposition mode number as capacity measure. They get to four-digit mode counts, far above earlier quantum storage demonstrations.\n\nThe soft spot, and the stress-test note has it right, is that the headline 'modes' are not directly measured. N is computed as T_pump/T_mode with T_mode=7.65 ns extracted from one correlation width, applied without uncertainty to all storage times. The CHSH test post-selects a single 7.65 ns window inside the central Franson peak, so it certifies entanglement of one time-bin pair, not thousands of independent entangled modes. The Schmidt number is a property of the biphoton state—a theoretical capacity bound, not a demonstrated count of addressable storage modes. The field value 8235 is within ~4% of the 'more than 8000' threshold, so calibration error in T_mode could move that particular headline. The authors themselves flag in the Discussion that whether Schmidt number equals repeater speed-up is open and that conditional mode filtering is impractical. So the reader's CONDITIONAL verdict is fair: accept the entanglement evidence, treat the mode count as an optimistic capacity claim.\n\nWhat could improve the paper: give T_mode uncertainty and propagate to N, distinguish 'Schmidt capacity' from 'demonstrated multiplexed entanglement storage' in the abstract, and perhaps present the mode count as a capacity rather than a direct measurement. The citation pattern is fine; the main self-refs are to their own prior characterization work, which is appropriate. None of this undermines the core experimental achievement.\n\nWho it's for: quantum repeater and rare-earth memory people. It deserves a serious referee, and I'd take it after revision. I would cite it for the memory characterization and the Schmidt capacity framing, with a caveat about what N means.","headline":"Credible entanglement storage with honest caveats; the 8,000-mode headline is a derived capacity, not a directly measured multiplexing count, and the paper's own discussion says so.","tokens_in":14775,"tokens_out":2980,"would_cite":true,"duration_ms":24969,"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":"A quantum memory stored one photon of an entangled pair across 16,340 temporal modes while its partner travelled 25.3 km of fiber.","keywords":["quantum repeaters","multimode quantum memory","atomic frequency comb","entanglement distribution","temporal modes","Schmidt decomposition","CHSH Bell test","rare-earth-doped crystals"],"falsifier":"Set up a mode-selective readout that addresses one narrow time bin within the AFC echo and measure the CHSH parameter conditioned on that bin; if only a few bins violate the inequality while the rest contribute noise, the usable multiplexing capacity is far below the reported Schmidt count.","tokens_in":13671,"feed_emoji":"🔗","tokens_out":8284,"duration_ms":70070,"temperature":0.7,"pith_summary":"Entanglement distribution over fiber is limited by the fact that a single-mode quantum memory can only hold one attempt at a time. The paper's goal is to show that an atomic frequency comb memory in a 171Yb3+:Y2SiO5 crystal can break that bottleneck by storing one photon of an entangled pair across many temporal modes at once, while its telecom partner travels through the fiber. To that end, it demonstrates experimentally that a 979 nm photon remains entangled with a 1553 nm telecom photon after storage for 125 microseconds across 16,340 temporal modes, with the telecom photon having passed through 25.3 km of fiber; a field test over 5.66 km of the Geneva metropolitan fiber network stored 8,235 modes for 63 microseconds. The paper also introduces the Schmidt mode number, computed from the joint temporal amplitude, as a quantitative measure of how many modes are available for multiplexing. If correct, this is the highest temporal-mode count demonstrated for an entangled light-matter quantum repeater node and points a clear way toward practical multiplexed quantum repeaters.","feed_headline":"Quantum memory stores entanglement across 16,340 time modes","feed_subtitle":"One photon stays in a crystal while its entangled partner crosses 25 km—more modes means faster quantum repeaters.","key_machinery":"The central element is the atomic frequency comb (AFC) memory in a 171Yb3+:Y2SiO5 crystal: light is absorbed by a comb of narrow spectral peaks burned into an inhomogeneously broadened transition, and the comb spacing Δ dictates that the collective excitation re-emits after a time 1/Δ. The quantitative engine is the Schmidt decomposition of the joint temporal amplitude of the SPDC two-photon state; because the pump pulse is long and slowly varying, the Schmidt mode number reduces to N = T_pump / T_mode, with T_mode = 7.65 ns extracted from the measured g^(2) correlation width. The lower bound on N, computed from the measured joint temporal intensity, is what the paper reports as the stored mode count. A Franson-type setup with a Michelson interferometer at 1553 nm and a double-AFC analyzer at 979 nm certifies energy-time entanglement through a CHSH Bell test.","core_discovery":"The paper reports that a crystal-based atomic frequency comb memory can store a 979 nm photon entangled with a telecom photon at 1553 nm while that telecom photon propagates through optical fiber, and that the number of temporal modes over which this works can be quantified by a Schmidt decomposition. In the lab, with fiber spools up to 25.3 km, storage for 125 microseconds across 16,340 modes was achieved, with CHSH Bell parameters between S = 2.65 and 2.73 for all tested lengths, proving that entanglement survives both storage and propagation. In a field test using a 5.66 km dark fiber through the Geneva metropolitan network, storage for 63 microseconds across 8,235 modes gave S = 2.67 ± 0.04. The memory has a bandwidth of 250 MHz and a measured lifetime of (76.6 ± 11.0) microseconds, corresponding to an effective AFC coherence time of about 307 microseconds.","pith_inferences":["An implication the paper stops short of claiming: if the Schmidt modes are independently usable, a repeater node could schedule one entanglement attempt per ~7.65 ns bin, so a 125 microsecond storage window holds 16,340 parallel attempts instead of one.","A testable extension: measure the CHSH violation for photons selected in narrow time bins across the echo window; if the violation and cross-correlation stay constant bin-by-bin, the operational mode count tracks the Schmidt count.","A rate calculation that folds in the 76.6 microsecond memory decay and the detection window is still needed to convert 8,000+ modes into an actual repeater speed-up, as the paper notes when it calls the speed-up question open."],"forward_implications":["A quantum repeater node can herald storage of a 979 nm photon entangled with a telecom photon that has propagated 25.3 km, holding 16,340 temporal modes for 125 microseconds with CHSH S = 2.70.","At 10 microseconds storage and 2 km fiber, the memory still stores 1,307 modes at 88% of its maximum efficiency, so short-distance entanglement distribution pays almost no efficiency penalty for high multiplexing.","In a metropolitan field test on 5.66 km of fiber, entanglement survives with 8,235 stored modes for 63 microseconds (S = 2.67), showing that a real fiber network does not destroy the effect.","The measured AFC lifetime of 76.6 microseconds corresponds to an effective coherence time of about 307 microseconds, roughly one-third of the material's optical coherence time; closing that gap would directly raise the attainable storage time and mode count.","The Schmidt mode number provides a simple quantitative benchmark for multimode capacity that can be computed from measured cross-correlation data, which the paper proposes as a standard comparison across repeater nodes."],"supporting_citations":[{"why":"Provides the 171Yb3+:Y2SiO5 AFC memory with 250 MHz bandwidth and 125 microsecond storage; the memory preparation sequence is described there.","marker":"[35]"},{"why":"Establishes the long optical coherence time of 171Yb3+:Y2SiO5 clock states that underlies the memory lifetime.","marker":"[34]"},{"why":"Introduces the AFC storage mechanism whose temporal multimode capacity is exploited here.","marker":"[30]"},{"why":"Gives the bandwidth-times-storage-time scaling that sets the number of temporal modes an AFC memory can hold.","marker":"[31]"},{"why":"Supplies the SPDC source design that generates the bandwidth-matched 979 nm / 1553 nm entangled photon pairs.","marker":"[23]"},{"why":"Defines the single-photon-entanglement (DLCZ-type) repeater setting whose rate the multimode memory is meant to accelerate.","marker":"[4]"},{"why":"Introduces the double-AFC scheme used as the 979 nm analyzer in the Franson-type entanglement measurement.","marker":"[47]"},{"why":"Gives the relation linking the measured decay time to the effective AFC coherence lifetime used to report 307 microseconds.","marker":"[37]"},{"why":"Supports the energy-time entanglement and Franson-type interferometric analysis used to certify entanglement after storage.","marker":"[44]"}],"fun_headline_variants":["16,340-mode quantum memory stores entanglement","8,235 entangled modes stored on Geneva metro fiber","Crystal memory holds entangled photon while twin travels","125 µs storage across 16,340 entangled modes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline claim rests on treating the Schmidt mode number of the SPDC state, T_pump / T_mode, as the number of temporal modes the memory can store and a repeater could use for multiplexing; the paper itself notes that it is an open question whether this number equals the actual repeater speed-up.","fun_headline_variants_meta":{"raw":{"variants":["16,340-mode quantum memory stores entanglement","8,235 entangled modes stored on Geneva metro fiber","Crystal memory holds entangled photon while twin travels","125 µs storage across 16,340 entangled modes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00145,"raw_usage":{"total_tokens":5834,"prompt_tokens":935,"completion_tokens":4899,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":551,"completion_tokens_details":{"reasoning_tokens":4838}},"tokens_in":551,"tokens_out":4899,"duration_ms":35143,"temperature":1.0,"reasoning_tokens":4838,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:00:04.023855+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Set up a mode-selective readout that addresses one narrow time bin within the AFC echo and measure the CHSH parameter conditioned on that bin; if only a few bins violate the inequality while the rest contribute noise, the usable multiplexing capacity is far below the reported Schmidt count.","supporting_citations":[{"cited_title":"Canteri, Z","cited_arxiv_id":null,"evidence_quote":"Introduces the AFC storage mechanism whose temporal multimode capacity is exploited here."},{"cited_title":"Sangouard, C","cited_arxiv_id":null,"evidence_quote":"Defines the single-photon-entanglement (DLCZ-type) repeater setting whose rate the multimode memory is meant to accelerate."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the long optical coherence time of 171Yb3+:Y2SiO5 clock states that underlies the memory lifetime."},{"cited_title":"Tittel, J","cited_arxiv_id":null,"evidence_quote":"Introduces the double-AFC scheme used as the 979 nm analyzer in the Franson-type entanglement measurement."},{"cited_title":"Rielander, K","cited_arxiv_id":null,"evidence_quote":"Gives the relation linking the measured decay time to the effective AFC coherence lifetime used to report 307 microseconds."}],"review_version":1}