{"id":"3e1848f6-4842-4f8c-b1fb-db66258e8955","arxiv_id":"2607.28943","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"The hyperfine structure of the excited 1s:3T2 state of the singly ionized interstitial aluminum donor in 28Si has been optically resolved, giving a contact hyperfine coupling of 2.75 µeV.","lead":"This paper reports the sharpest optical look yet at a little-studied defect in silicon, showing that its telecom-band luminescence carries readable atomic-scale structure from the defect's nucleus. It makes the defect a new candidate for silicon quantum memory, where light pulses could read and write a single nuclear spin.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"a_iso extraction depends on unvalidated g_L=0; free-g_L refit of Fig. 3 data is the decisive check.","rationale":"The reader's weakest_assumption points to the state assignment and g_L deviation. I agree that the g_L≈0 assumption is a genuine load-bearing concern for the quantitative parameter a_iso, but the state assignment itself is strongly supported by the data: the three zero-field lines match the F=3/2,5/2,7/2 hyperfine levels of a J=1 manifold, and the 110 mT spectrum resolves six m_I states. That is clear evidence of I=5/2 hyperfine in an excited state, regardless of whether the model is exchange-split triplet or some other triplet-like level. The earlier spin-orbit-split interpretation of Latushko et al. predicts a different number of zero-field lines (e.g., two or four for J=1/2 or 3/2), so it is disfavored by the observation. The remaining weak point is the fixed g_L: the fit never tests this parameter, and the quoted uncertainty in a_iso is purely statistical. A free-g_L fit is a feasible and decisive check. If g_L turns out to be nonzero, the a_iso value would need revision, but the headline claim of optically-resolved hyperfine would still stand. Therefore, the verdict remains CONDITIONAL, and this stress-test does not change the reader's assessment.","tokens_in":17433,"tokens_out":12188,"duration_ms":111553,"concrete_test":"Re-fit the high-field spectra of Fig. 3 (and the hyperfine data of Fig. 4) with g_L and g_s as free parameters in Eq. (1), then re-extract a_iso with g_L fixed at the best-fit value. If the best-fit g_L is within ±0.05 of 0 and a_iso shifts by less than the reported 0.03 µeV, the concern is resolved. If the fit prefers a nonzero g_L or the a_iso shift exceeds the quoted uncertainty, the model is incomplete and a_iso must be re-reported with the appropriate systematic error.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline claim—first optically-resolved excited-state hyperfine structure—is supported by the observation of three zero-field lines and a six-line magnetic fan-out, which is a robust fingerprint of an I=5/2 nucleus coupled to a J=1 manifold. However, the quantitative result (a_iso = 2.75 ± 0.03 µeV) is extracted from a fit that fixes g_L = 0 and g_s = 2 a priori (Sec. IV, Eq. 1), citing literature for neutral chalcogens rather than testing this against the Al_i+ data. The magnetospectroscopy in Fig. 3 was fit with g_L fixed; a nonzero g_L would alter the Zeeman shifts of the J=1,2 sublevels, and the subsequent hyperfine fit (which fixes λ, g_L, g_s from that fit) would need a different a_iso to reproduce the 110 mT spectrum. The quoted 0.03 µeV uncertainty reflects only peak-position statistics, not systematic uncertainty from the g_L assumption. If g_L is actually, say, 0.1 (plausible for a T2 orbital with incomplete quenching), the Zeeman shift at 110 mT is ~0.6 µeV, a sizable fraction of a_iso, so the fitted a_iso could shift beyond the quoted error. This does not threaten the existence of the hyperfine structure, but it undermines the specific a_iso value and the associated quantum-number labels.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports an extensive spectroscopic study of the singly ionized interstitial aluminum donor (Al_i^+) in isotopically enriched 28Si. The authors obtain the He+-like AM1 absorption series with improved precision and several new Rydberg transitions, measure the photoluminescence spectrum, and determine a Debye-Waller factor of 47±1%. They attribute the 774.87 meV emission to the 1s:3T2(J=1) → 1s:1A1 transition of an exchange-split two-electron excited state, rather than to the earlier spin-orbit-split interpretation. From magnetospectroscopy they fit an effective spin-orbit parameter λ = 47.6 ± 0.9 µeV with g_s = 2 and g_L = 0 fixed. At zero field they resolve three lines within the J=1 manifold, and at 110 mT a six-line fan-out, which they fit with an isotropic contact hyperfine interaction a_iso = 2.75 ± 0.03 µeV. They claim this is the first optically resolved hyperfine structure in the excited state of a telecom-band silicon colour centre and use the fitted Hamiltonian to predict >99% nuclear-spin-preserving decay at fields ≥ 606 mT.","tokens_in":17827,"tokens_out":6456,"duration_ms":55897,"significance":"If substantiated, the result is a valuable step for silicon-based quantum networking: it would provide an optical interface to an I = 5/2 nuclear spin in a centre with a diamagnetic ground state, with emission in the telecom L band. The observation of three zero-field lines and a six-line magnetic fan-out is a strong, model-independent fingerprint of hyperfine coupling to a J=1 manifold, and the measured 25.5 ms triplet lifetime is notable. The paper also provides a careful refinement of the AM1 series and independent lifetime determinations. The main caveats are that the quantitative a_iso and the F labels depend on assuming g_L = 0 and g_s = 2 (Table I) and on the exchange-triplet assignment, which is justified by analogy rather than by an independent calculation. These caveats are addressable and do not by themselves invalidate the central observation.","major_comments":[{"comment":"The quoted a_iso = 2.75 ± 0.03 µeV is obtained with g_L fixed to zero and g_s fixed to 2; the error bar includes only peak-position statistics. At B = 110 mT, a nonzero g_L of 0.1 would produce an orbital Zeeman shift of roughly 0.6 µeV—about 20% of a_iso. The magnetospectroscopy in Fig. 3 should be refit with g_L (and ideally g_s) as free parameters, and the resulting joint uncertainty on (λ, g_L, a_iso) reported. Without this, the specific value of a_iso and the F = 3/2, 5/2, 7/2 labels are not robustly established.","section":"Sec. IV/V, Eq. (1), Table I"},{"comment":"The assignment of the 774.87 meV transition to an exchange-split 1s:3T2 state, with the fine structure described by λ(L·S), is made by analogy to chalcogen and magnesium double-donor systems (Refs. 35–43) and is not validated by an independent calculation or by a direct measurement of the singlet/triplet character. The temperature-dependent intensity ratio in Sec. A4 and the observed ΔE_st = 21.26 meV are consistent with the exchange model, but they do not uniquely exclude the earlier spin-orbit-split interpretation of Latushko et al. Because the hyperfine quantum-number labels and a_iso are defined within the new assignment, this is a load-bearing assumption. A concrete test—for example, the Zeeman response of the 796 meV singlet transition or a stress/polarization study of the two transitions—should be provided or explicitly identified as future work.","section":"Sec. II and IV"}],"minor_comments":[{"comment":"Typos: 'seperations' in Fig. 1; 'qudrupole-like' should be 'quadrupole-like'; 'particularity' should be 'particularly'; inconsistent spelling 'Latushko' vs 'Latuschko' between main text and Refs. 32–34; 'Abromosimov' and 'T_Hewalt' in Refs. 2–3.","section":"Throughout"},{"comment":"The 5σ discrepancy between the two singlet lifetime determinations is attributed to a 'forbidden 1s:E exchange-split dark state' without direct evidence. This is a plausible conjecture but should be flagged as such rather than stated as an explanation.","section":"Sec. A4"},{"comment":"The 'partial inner product' P(m_J, m_I; m'_I) is not defined precisely; please give the overlap expression in terms of Clebsch–Gordan coefficients or eigenvector components. Also, the >99% nuclear-spin-preservation statement is a calculated consequence of the fitted Hamiltonian, not an independent prediction; the conclusion should not present it as a confirmation of the model.","section":"Sec. VI, Eqs. (3)–(4)"},{"comment":"'Available from the authors upon reasonable request' is weaker than current norms for reproducibility. Please deposit the raw spectra and fitting code in a public repository.","section":"Data Availability"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a solid experimental contribution with a clear central observation. The g_L = 0 assumption is the main technical risk; I would like the authors to address it with a free-g_L fit and to report the resulting systematic uncertainty on a_iso. The level-assignment question is also important but can be handled by framing and a proposed test. I would not reject."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the headline is real. The zero-field three-line spectrum and the field-dependent fanout in Fig. 4 are a textbook fingerprint of I=5/2 nuclear spin coupled to a J=1 manifold. That alone justifies the claim of first optically-resolved excited-state hyperfine structure in a telecom-band silicon colour centre. The paper also reports genuinely new data: previously unobserved transitions up to 6p±, refined energies, first lifetimes, and a plausible reassignment of the emitting state.\n\nThe strong part is the spectroscopy. The linewidths in 28Si are good enough to see the hyperfine splitting, and the field dependence is consistent with the Hamiltonian. The authors are careful about fitting peak positions and report uncertainties that reflect line-position statistics.\n\nThe soft spots are real but not fatal. The biggest is g_L=0 fixed from literature for neutral chalcogens, not tested against these data. At 110 mT a g_L of 0.1 changes Zeeman shifts by ~0.6 µeV, comparable to a_iso. So the quoted 2.75±0.03 µeV is precise but not accurate unless g_L is truly zero. That doesn't threaten the existence of the hyperfine structure, but it does mean the specific value and the F labels should be treated as model-dependent. A free-g_L refit of Fig. 3 data would settle it; the authors should do that or provide a bound.\n\nSecond, the exchange-split triplet assignment replacing Latushko's spin-orbit interpretation is built on analogy to chalcogen/Mg donors, not ab initio calculation. It's plausible, but if wrong, the quantum-number labels change. The authors acknowledge this.\n\nThird, the two independent singlet-lifetime determinations disagree by 5σ. They invoke an unobserved dark state. This is not central to the hyperfine claim but suggests one of the two methods has a systematic error. They should address it.\n\nData are only available on request. That's a minor complaint relative to the above.\n\nBottom line: this is a solid experimental contribution that deserves a serious referee. The central claim holds up; the quantitative hyperfine parameter needs more careful treatment of systematic uncertainties. The paper will be of interest to anyone working on silicon colour centres and spin-photon interfaces. I'd send it out, with a request for a free-g_L fit or a strong justification for fixing it.","headline":"First optically-resolved hyperfine structure in a telecom silicon colour centre is real, but the quoted a_iso carries unquantified systematic error from a fixed g_L=0.","tokens_in":18383,"tokens_out":2321,"would_cite":true,"duration_ms":20660,"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":"Telecom-band silicon colour centre shows optically resolved hyperfine structure in its excited state, yielding a contact hyperfine coupling of 2.75 µeV.","keywords":["silicon colour centre","interstitial aluminum donor","hyperfine structure","excited state","spin-photon interface","telecom band","nuclear spin qubit","photoluminescence"],"falsifier":"Measure the Zeeman dispersion of the three zero-field hyperfine lines up to about 1 T: the Hamiltonian predicts a specific fan-out of the J=1 manifold into six nuclear-spin branches for each m_J substate. If the number of branches or their field-dependent slopes disagrees with the J=1 assignment, or if an independent technique such as optically detected magnetic resonance resolves a different hyperfine pattern, the assignment and a_iso would be refuted.","tokens_in":17319,"feed_emoji":"🔬","tokens_out":2465,"duration_ms":24429,"temperature":0.7,"pith_summary":"The paper characterizes the singly-ionized interstitial aluminum donor in isotopically purified silicon-28 and reports the first optical resolution of hyperfine structure in the excited state of a telecommunications-band silicon colour centre. It identifies the 774.87 meV emission as an exchange-split triplet level of the 1s:T2 excited state, whose long lifetime (25.5 ms) and narrow linewidth allow the fine and hyperfine splittings to be seen directly in photoluminescence. The authors extract a spin-orbit coupling strength of 47.6 µeV and an isotropic contact hyperfine parameter of 2.75 µeV, and they argue these transitions can be used to read out the aluminum nuclear spin optically. If correct, this opens a route to nuclear-spin quantum memories in silicon that avoid ground-state electron decoherence.","feed_headline":"Silicon defect shows hyperfine lines in telecom band","feed_subtitle":"Optically reading an aluminum nuclear spin via a metastable triplet could enable silicon quantum memories.","key_machinery":"The central object is an effective Hamiltonian for the two-electron 1s:T2 excited state: H = λ(S·L) + μB gs (S·B) + μB gL (L·B) + a_iso(I·S) – μn gI(I·B). Here L=1 is a fictitious orbital angular momentum representing the threefold degeneracy of the T2 manifold, S=1 is the electron spin of the triplet, I=5/2 is the aluminum-27 nuclear spin, and λ is the effective spin-orbit strength. This Hamiltonian organises the levels into J=0,1,2 manifolds and then into hyperfine states, and it is used to fit the observed optical spectra and predict nuclear-spin-preserving transition probabilities.","core_discovery":"The paper establishes that the singly-ionized interstitial aluminum donor in 28Si has a metastable spin-triplet excited state, 1s:3T2(J=1), whose optical transition to the diamagnetic 1s:1A1 ground state is narrow enough to resolve hyperfine structure. Three zero-field transitions are observed within the J=1 manifold, and fitting an effective Hamiltonian with spin-orbit coupling, electron Zeeman, and an isotropic contact hyperfine term gives a_iso = 2.75 ± 0.03 µeV. The authors also show that at magnetic fields above about 606 mT the emission decays become more than 99% nuclear-spin-preserving, meaning a single optical photon can projectively measure the aluminum-27 nuclear spin state.","pith_inferences":["If the assignment holds, the same optical-resolution technique could be applied to other diamagnetic-ground-state centres (e.g., G, C, or interstitial carbon) to look for hyperfine structure that has so far been invisible in photoluminescence.","The near-zero orbital g-factor assumption (g_L ≈ 0) could be tested directly by measuring the Zeeman dispersion at higher fields; if g_L is non-negligible, the extracted a_iso would shift, but the qualitative nuclear-spin-preserving behaviour may persist.","One could test the pumping scheme by resonantly exciting the 1s:1T2 singlet and observing whether the hyperfine populations in the triplet become non-thermal, which would be a direct signature of nuclear initialization.","A natural extension is to search for the analogous hyperfine-resolved triplet in isotopically purified 28Si doped with indium or thallium, where the heavier nucleus and stronger spin-orbit coupling could yield faster, brighter emission in the same telecom bands."],"forward_implications":["Optical readout of the aluminum-27 nuclear spin becomes possible via spin-selective decays from the J=1 triplet manifold, with a predicted >99% nuclear-spin-preserving branching ratio above 606 mT.","The long-lived triplet (25.5 ms) could serve as a metastable communication qubit interfaced to a nuclear memory qubit in the ground state, avoiding electron-spin-induced decoherence.","The measured hyperfine splitting of about 2.75 µeV (roughly 665 MHz at the relevant transitions) is resolvable with standard fibre Fabry-Perot cavities, enabling photon-mediated readout.","Heavier group-III interstitials, such as indium or thallium, may exhibit faster triplet emission due to stronger spin-orbit coupling, pointing toward faster networking rates.","The refined AM1 donor-series energies and newly observed excited states provide a more accurate benchmark for theoretical models of this defect."],"fun_headline_variants":["First telecom-band silicon color center with resolved hyperfine lines","Silicon defect's excited hyperfine split seen in telecom band","Optical readout of aluminum nuclear spin via silicon defect","Metastable triplet in silicon yields hyperfine-resolved telecom photons","Aluminum donor in silicon shows optically resolved hyperfine structure"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire hyperfine analysis rests on assigning the 774.87 meV emission to the J=1 level of an exchange-split 1s:3T2 triplet, with the electron g-factors fixed at g_s=2 and g_L≈0 by analogy to other donors; if that level ordering or the g-factor values are wrong, the quantum-number labels and the extracted hyperfine coupling would not be valid.","fun_headline_variants_meta":{"raw":{"variants":["First telecom-band silicon color center with resolved hyperfine lines","Silicon defect's excited hyperfine split seen in telecom band","Optical readout of aluminum nuclear spin via silicon defect","Metastable triplet in silicon yields hyperfine-resolved telecom photons","Aluminum donor in silicon shows optically resolved hyperfine structure"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000362,"raw_usage":{"total_tokens":1792,"prompt_tokens":748,"completion_tokens":1044,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":492,"completion_tokens_details":{"reasoning_tokens":974}},"tokens_in":492,"tokens_out":1044,"duration_ms":8023,"temperature":1.0,"reasoning_tokens":974,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T16:46:27.139889+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the Zeeman dispersion of the three zero-field hyperfine lines up to about 1 T: the Hamiltonian predicts a specific fan-out of the J=1 manifold into six nuclear-spin branches for each m_J substate. If the number of branches or their field-dependent slopes disagrees with the J=1 assignment, or if an independent technique such as optically detected magnetic resonance resolves a different hyperfine pattern, the assignment and a_iso would be refuted.","supporting_citations":[],"review_version":1}