{"id":"312210ec-1798-4577-b9fe-342798915563","arxiv_id":"2608.05349","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"UV-written FBGs in high-NA fiber have the lowest cladding-mode loss (0.5 dB) but high insertion loss (4.6 dB), while fs-IR-written FBGs in SMF-28 have insertion loss below 0.05 dB and cladding-mode loss 0.93 dB; a bridge-fiber design is proposed for integration.","lead":"Tests of fiber Bragg grating filters for astronomy show a trade-off: filters written with infrared laser pulses in ordinary telecom fiber lose almost no light at the splice, while filters written with ultraviolet light in a special high-NA fiber lose less light in the cladding. The paper proposes a bridge-fiber layout that keeps total insertion loss below 1 dB and could cut splice junctions in half for future telescope instruments.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The IL<1 dB system-level viability claim for UV-inscribed SM1500(4.2) filters is inferred by halving the measured two-sided bridge loss (0.88 dB) rather than measured directly, so the headline 'viable after bridging' is not yet established.","rationale":"The stand-alone loss comparison is a legitimate, transparent experimental contribution; the measured ranking of CM loss and IL is plausible and internally consistent, and the paper discloses the aperture-shaping dependence of the fs-IR CM loss. The fragile part of the central claim is the step from measured components to the system-level statement 'both platforms remain viable ... with IL below 1 dB' (§4.1, §5). That step relies on a symmetry/independence assumption for splice loss that is neither measured nor argued beyond dividing a two-sided result by two, plus manufacturer bulletin data for the reduced-clad option. The 1 dB margin is thin: 0.88 dB + 0.05 dB + any PL/VGA coupling margin leaves <0.1 dB tolerance. The reader's weakest assumption (five-line representativeness) is also legitimate but is a scope limitation the paper explicitly acknowledges; the bridge extrapolation is a factual support gap in the paper's main integrative claim. A direct one-sided bridge measurement would settle it. I therefore do not move the verdict: the conditional status should remain, with the one-sided bridge and full-chain measurements made explicit conditions.","tokens_in":20298,"tokens_out":8569,"duration_ms":72891,"concrete_test":"Fabricate at least four one-sided bridged assemblies SMF-28 – B2 – B1 – SM1500(4.2) – SM1500(4.2/80), with and without the five-line FBG array, and measure IL at 1550 nm and 1570 nm using the setup of Fig. 1(b). Compare the measured one-sided IL with half of the Table 3 two-sided 2-bridge value (0.88 dB). If the measured mean plus one standard deviation exceeds 1.0 dB, or if it differs from 0.88 dB by more than ~0.2 dB, the IL<1 dB claim should be reclassified as an unverified estimate and the conclusion made conditional on a direct system-level demonstration.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.1 concludes that UV-inscribed SM1500(4.2) filters will incur IL<1 dB when placed between a photonic lantern and a V-groove array/AWG. The number 0.88 dB comes from dividing the measured two-sided 2-bridge IL of 1.77±0.22 dB (Table 3) by two, assuming each side contributes equally and independently. That assumption is not tested. The two sides are separate splices with uncontrolled angular misalignment in the range 0.2°–1°, and the Marcuse estimate (Eq. 1) is sensitive to the exact MFD ratio and angular term; the uncertainty on the half-bridge value alone (±0.11 dB if the two sides are independent) leaves no guaranteed margin under 1 dB once the 0.05 dB direct splice to SM1500(4.2/80) is added. The proposed reduced-clad configuration (Table 4) is taken from a manufacturer technical bulletin and the paper explicitly lists experimental validation as future work (§4.2). No measurement of the complete PL-output → bridge → FBG filter → reduced-clad fiber → AWG chain is presented. If the one-sided bridge loss is higher than ~0.9 dB, the 'viable with IL below 1 dB' conclusion fails even for the five-line test array, independent of whether the five lines are representative of the full 105-line set.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a comparative study of insertion loss (IL) and cladding-mode (CM) loss in five-line fiber Bragg grating (FBG) OH-suppression filters fabricated by continuous-wave UV inscription in three photosensitive fibers (PS1250, SM1500(4.2), CMS2) and by femtosecond-IR inscription in SMF-28 and CMS2. The central finding is a trade-off: UV-inscribed filters in SM1500(4.2) show the lowest CM loss among the photosensitive fibers (about 0.5 dB) but the highest IL (4.6 dB), whereas fs-IR-inscribed filters in SMF-28 show IL below 0.05 dB with a higher CM loss of 0.93 dB. The paper then evaluates tapering and bridging approaches to reduce the SM1500(4.2) splice loss, reporting that a two-bridge configuration reduces IL from about 4.76 dB to 1.75 dB, and uses this to infer that both filter platforms can be integrated into an astrophotonic system with IL below 1 dB. A compact reduced-clad bridge configuration is proposed to further lower IL below 0.5 dB. The full H-band set of about 105 OH lines is deferred to future work.","tokens_in":20542,"tokens_out":4710,"duration_ms":41020,"significance":"If the central results hold, the manuscript provides a useful quantitative trade-off for instrument designers choosing between UV- and fs-IR-inscribed FBG filters for OH suppression, and it addresses the current unavailability of the CMS8 fiber used in earlier GNOSIS/PRAXIS demonstrations. The paper's strengths include direct comparative measurements across multiple fibers and inscription methods, replicate samples for the bridging configurations (Table 3), a clear statement of the controlled five-line test case, and an honest identification of several open issues (e.g., CM-loss behavior with reduced-clad fibers, validation of the proposed bridge scheme). The main limitation is that the headline system-level 'viability with IL below 1 dB' claim rests on an inferred, rather than directly measured, one-sided bridge loss and on manufacturer technical-bulletin data for the proposed <0.5 dB configuration.","major_comments":[{"comment":"The claim that UV-inscribed SM1500(4.2) filters will incur IL<1 dB when placed between a photonic lantern and a V-groove array is supported only by halving the measured two-sided two-bridge loss of 1.75±0.22 dB (Table 3) and adding a 0.05 dB direct-splice estimate. This assumes the two bridge chains contribute equally and independently, but the angular misalignment per splice is uncontrolled within 0.2°–1° and the Marcuse equation (Eq. 1) has an exponential angular term, so the one-sided loss is not guaranteed to be half of the two-sided mean. Taking the measured standard deviation at face value, the one-sided value has about ±0.16 dB uncertainty, and adding the 0.05 dB direct splice leaves little margin below the 1 dB threshold. The authors should either measure the one-sided bridge configuration directly or report a conservative upper bound; as written, the IL<1 dB system-level claim is not established.","section":"§4.1, Fig. 9, Table 3"},{"comment":"The system-level 'viability' discussion treats IL and CM loss separately, but for an OH-suppression filter the CM loss is also a real throughput loss. For the fs-IR SMF-28 platform, IL<0.05 dB plus CM loss 0.93 dB is just below 1 dB total; for the UV SM1500(4.2) platform, the inferred one-sided IL of about 0.88 dB plus the 0.05 dB direct splice plus the measured CM loss of 0.49 dB exceeds 1 dB. The manuscript should state explicitly whether the '<1 dB' viability threshold applies to IL alone or to total filter-induced loss, and if the latter, the SM1500(4.2) conclusion needs revision.","section":"§4.1/§5; Table 2"},{"comment":"The headline loss values in Table 2 are single-point measurements with no error bars or replicate information, while Table 3 reports means and standard deviations only for the bridging study. This matters for the comparative ranking: for example, the CM losses for SM1500(4.2) and CMS2(UV) are 0.49 dB and 0.52 dB, respectively, a difference comparable to typical measurement repeatability. Without uncertainties or at least a statement of measurement reproducibility, the claim that SM1500(4.2) achieves the lowest CM loss among photosensitive fibers is not yet quantitatively supported.","section":"§2.4, Table 2"},{"comment":"The proposed compact bridge configuration with IL<0.5 dB is based entirely on average splice-loss values from a manufacturer's technical bulletin (reference 62), and the manuscript explicitly lists experimental validation as future work. This is therefore a projection, not a demonstrated result. The abstract and Section 5 present the '<0.5 dB' figure without this caveat; the wording should be adjusted so that the distinction between measured and proposed performance is clear.","section":"§4.2, Table 4"}],"minor_comments":[{"comment":"The five-line test array is explicitly a controlled case, but several statements in the abstract and conclusions ('both filter platforms remain viable for integration into an astrophotonic system') could be read as applying to the complete 105-line OH-suppression filter set. A sentence clarifying that viability is currently demonstrated only for the five-line subset would prevent over-interpretation.","section":"§1, §4.2"},{"comment":"The notation for fiber types is inconsistent: SM1500(4.2) is sometimes written as SM1500(4.2/125), and the table uses 'SMF' while the text uses 'SMF-28'. Please standardize the nomenclature.","section":"Table 2"},{"comment":"The footnote for CMS2(UV) correctly states that its performance is not directly comparable because the gratings are weaker (reflectivity about 82% versus >96%); this caveat should also be mentioned in the abstract where CMS2 results are summarized or omitted from the summary if not essential.","section":"§2.4.3"},{"comment":"The text says the two-bridge configuration reduces IL 'from 4.7 dB to 1.77 dB' at 1570 nm, while Table 3 reports 4.76±0.24 dB and 1.75±0.22 dB at 1550 nm. Please clarify whether the figure is at 1570 nm and whether the difference from the table is due to wavelength dependence or to using a different sample.","section":"§3.2, Fig. 7"},{"comment":"The sentence 'It is not straightforward to assess how incorporating a reduced-clad fiber at the interface will affect CM losses for both fiber platforms, warranting further study' is an important limitation that appears only in Section 4.1. It should be reflected in the conclusions as well, since it directly affects the system-level viability assessment.","section":"§4.1"},{"comment":"Reference 62 is cited as a technical bulletin with a URL but no accession date or document number beyond 'TN34.1'. Including the full title, version, and date of the bulletin would improve reproducibility.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a solid comparative study with useful measurements, and I do not see any integrity or novelty-disclosure concerns. The main issue is that the abstract's headline 'viability with IL below 1 dB' claim rests on an inferred halving of a two-sided bridge measurement and on vendor technical-bulletin data, with no direct measurement and no uncertainty analysis. This is fixable within the paper's scope by measuring the one-sided bridge configuration (or providing a conservative bound) and by explicitly separating measured from proposed performance, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real contribution is the side-by-side loss data: five-line FBG arrays, four fiber types, two inscription methods, measured IL and CM loss in one place. That table is useful, especially now that CMS8 is gone and ELT-era instruments need a fiber choice. The trade-off is clear: SM1500(4.2) gives the lowest CM loss among photosensitive fibers (0.49 dB) but high IL to standard fiber (4.6 dB); SMF-28 written with fs-IR gives IL below 0.05 dB but 0.93 dB CM loss with the aperture-shaped inscription. The bridging study with four samples per configuration and standard deviations is a step above single-point reporting, and the tapering section is an honest negative result—simulations don't reproduce the experiment, and they say so.\n\nSoft spots, in proportion. Table 2's headline values are single measurements with no error bars. The fs-IR CM loss is inflated by the shaping aperture, which they disclose, but it means the comparison is between one particular inscription configuration, not the method in general. The bigger issue is Section 4.1: the claim that UV filters in SM1500(4.2) will have IL under 1 dB after bridging comes from halving the measured two-sided 1.77 ± 0.22 dB loss. That assumes the two sides are equal and independent, which is not tested. One-sided uncertainty alone is roughly ±0.11 dB, and adding the 0.05 dB reduced-clad splice leaves no guaranteed margin. The proposed 0.5 dB compact bridge is taken from a manufacturer bulletin and explicitly lacks experimental validation. So the 'viable after bridging' conclusion should be read as a plausible projection, not a demonstrated result. To their credit, the paper lists validation as future work, but the abstract and conclusions state the 1 dB figure as a finding.\n\nWho this is for: people building OH-suppression systems, photonic-lantern integrators, anyone choosing between UV and fs-IR platforms. The comparative data are the takeaway; the system-level estimates need to be labeled as estimates. The paper is transparent, internally consistent, and the central empirical trade-off holds up. It deserves a serious referee—send it to review, and ask the authors to add error bars to Table 2 and reframe the IL<1 dB claim as an inferred projection rather than a measured outcome.","headline":"The measured loss trade-off between UV- and fs-IR-inscribed FBG filters is the real contribution; the system-level IL<1 dB claim is an unverified estimate that needs softening.","tokens_in":650,"tokens_out":1413,"would_cite":true,"duration_ms":26519,"reading_group":"maybe","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 claims that UV- and fs-IR-written fiber Bragg grating OH filters split losses: UV high-NA fiber wins on cladding-mode loss (~0.5 dB) but loses on insertion loss (4.6 dB), while fs-IR SMF-28 wins on insertion loss (<0.05 dB)…","keywords":["fiber Bragg gratings","OH suppression filters","astrophotonics","cladding-mode loss","insertion loss","UV inscription","femtosecond inscription","bridging fibers"],"falsifier":"A concrete check would be to inscribe the full ~105-line H-band filter set, or a dense block of at least 20–30 lines, in SM1500(4.2) and SMF-28 and measure accumulated insertion loss and cladding-mode loss; if per-line losses compound, the sub-1 dB viability conclusion would fail. A second check is to measure the splice loss of the proposed reduced-clad bridging configuration in-house, since the 0.44 dB figure comes from a manufacturer's technical bulletin rather than the authors' own measurements.","tokens_in":20103,"feed_emoji":"🔭","tokens_out":8241,"duration_ms":60093,"temperature":0.7,"pith_summary":"This paper compares two ways of writing fiber Bragg grating filters that block atmospheric hydroxyl (OH) emission lines in the near-infrared, with an eye to the photon-starved instruments on next-generation ground-based telescopes. The authors measure five representative filter lines and find a clean trade-off: UV-written gratings in a high-numerical-aperture photosensitive fiber (SM1500(4.2)) suffer the lowest cladding-mode loss (~0.5 dB) but the highest insertion loss (4.6 dB), while femtosecond-written gratings in ordinary SMF-28 fiber have insertion loss below 0.05 dB but a higher cladding-mode loss (0.93 dB). They then show that stepwise bridging fibers cut the insertion loss of the high-NA platform by ~3 dB, and argue that inside a photonic-lantern-to-chip architecture both platforms can meet an insertion-loss budget below 1 dB. If the trade-off holds across the full ~105-line H-band filter set, instrument designers can choose the fiber and inscription method by whether cladding-mode purity or raw throughput matters more.","feed_headline":"UV and fs-IR fiber filters trade cladding-mode loss for insertion loss","feed_subtitle":"High-NA UV gratings cut cladding-mode loss to 0.5 dB; fs-IR SMF-28 keeps insertion loss below 0.05 dB.","key_machinery":"The argument is carried by a loss decomposition into two mechanisms. Cladding-mode loss arises when the index perturbation written into the fiber core couples the guided core mode to counter-propagating cladding modes, producing short-wavelength dips; high-NA fibers confine the mode and reduce this coupling. Insertion loss is dominated by splice loss from mode-field-diameter mismatch, modeled by the Marcuse power-transmission coefficient $T=(2w_1w_2/(w_1^2+w_2^2))^2\\exp[-2(\\pi n_2 w_1 w_2 \\theta)^2/((w_1^2+w_2^2)\\lambda^2)]$, plus fiber attenuation from heavy dopants. The enabling fabrication tools are a CW 244 nm UV laser writing multi-channel gratings through a complex phase mask with overlapping channels, and an 800 nm femtosecond-IR laser writing through a phase mask with integrated shaping apertures that control the spectral profile. The mitigation machinery is bridging: intermediate fibers with intermediate mode-field diameters step the mode field from 4.2 µm to 10.4 µm, converting a 4.76 dB direct splice loss into 1.75 dB, and a reduced-clad SM1500(4.2/80) fiber directly interfaces the filter to an AWG chip with an estimated ~0.44 dB splice loss.","core_discovery":"The central discovery, on the paper's own terms, is a quantitative loss trade-off between inscription platforms rather than a single winner. For five notch filters in the 1546–1552 nm range, UV-inscribed FBGs in SM1500(4.2)—a high-NA, bend-insensitive photosensitive fiber—reach a cladding-mode loss of about 0.5 dB at 1547.43 nm, the lowest among photosensitive fibers and lower than the fs-IR aperture-shaped filters in SMF-28 (0.93 dB at 1544.07 nm), but they pay for this with an insertion loss of 4.6 dB at 1550 nm, dominated by mode-field mismatch at splices to standard fiber. fs-IR-inscribed gratings in non-photosensitive SMF-28 have insertion loss below 0.05 dB, because no mode-field mismatch is involved, and a cladding-mode loss of 0.93 dB. The paper reports that a two-bridging-fiber connection reduces the high-NA insertion loss by ~3 dB (4.76 dB to 1.75 dB), and that once the filters are integrated with a photonic lantern upstream and an arrayed-waveguide-grating chip downstream, both platforms achieve insertion loss below 1 dB; a proposed reduced-clad bridge could push this below 0.5 dB while halving the number of splice junctions.","pith_inferences":["If the loss-per-line result extends, the full 105-line filter set will likely need a longer fiber and more complex phase masks; cladding-mode loss could scale with the number of notches and accumulated UV fluence, so the sub-1 dB budget is an extrapolation until the full set is written.","The 0.2 dB cladding-mode loss quoted for standard fs-IR phase-mask inscription suggests the aperture-shaped profile is the price of spectral control; a two-step inscription that shapes the spectrum but expands the index modification into the cladding could plausibly recover both low cladding-mode loss and low insertion loss.","The proposed compact bridge could be tested immediately without an astronomical spectrograph: splice SMF-28 to SM1500(7.8/80) to SM1500(4.2/80), measure insertion loss and cladding-mode loss at 1550 nm, and compare against the manufacturer's 0.44 dB figure.","Because CMS2 under UV inscription only reached 82% reflectivity, the comparison excludes the strongest gratings in a cladding-mode-suppressed fiber; hydrogenation studies would be needed before ruling out that platform."],"forward_implications":["Astrophotonic instrument designers get a quantitative decision rule: choose UV-inscribed SM1500(4.2) when cladding-mode purity is the priority, and accept bridging to manage insertion loss.","fs-IR inscription in SMF-28 is the lower-insertion-loss route, but the 0.93 dB cladding-mode loss here is higher than the 0.2 dB reported for standard fs-IR phase-mask writing, with shaping apertures identified as the cause.","Two-stage bridging cuts the SM1500(4.2)-to-SMF-28 splice loss from about 4.8 dB to about 1.8 dB, and one-sided bridging in a photonic-lantern chain brings total insertion loss below 1 dB for the UV platform.","Using a reduced-clad bridge fiber can bring both platforms below 0.5 dB insertion loss and halve the splice-junction count from 4 to 2 per channel, a meaningful saving for a 19-port photonic lantern (76 to 38 junctions).","When filters sit between a photonic lantern and an AWG spectrograph, the insertion-loss difference between UV and fs-IR platforms shrinks below the level that decides platform choice, leaving cladding-mode loss as the remaining discriminator."],"supporting_citations":[{"why":"GNOSIS, the first instrument to use fiber Bragg gratings for OH suppression, defines the full 105-line context and the five-line subset source.","marker":"[9]"},{"why":"PRAXIS, the first demonstration of OH suppression in a high-efficiency spectrograph, is the benchmark system these filters must feed.","marker":"[10]"},{"why":"Goebel et al. showed direct femtosecond-IR writing as an alternative route for multi-notch OH filters, setting up the fs-IR platform comparison.","marker":"[24]"},{"why":"Krämer et al. describe the shaping-aperture method used for fs-IR spectral control and explain why it increases cladding-mode loss.","marker":"[34]"},{"why":"Dong et al. supply the mechanism by which high-NA mode confinement suppresses cladding-mode coupling loss.","marker":"[35]"},{"why":"Grobnic et al. provide the 0.2 dB cladding-mode-loss benchmark for standard fs-IR phase-mask writing in SMF-28 that frames the paper's 0.93 dB result.","marker":"[36]"},{"why":"Marcuse's splice-loss equation is the model used to estimate insertion loss from mode-field mismatch and angular misalignment.","marker":"[41]"},{"why":"The manufacturer's bridging-fiber technical bulletin supplies the splice-loss data for the proposed reduced-clad bridge and the below-0.5 dB claim.","marker":"[62]"}],"fun_headline_variants":["UV gratings: 0.5 dB cladding loss, 4.6 dB insertion; fs-IR flips","Astrophotonics filters: trade low insertion for low cladding loss","Bridge cuts UV grating insertion loss by ~3 dB","fs-IR SMF-28 insertion <0.05 dB; UV high-NA cladding 0.5 dB","Fiber grating loss trade-off: choose your poison for astronomy"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the performance of five closely spaced test lines, written in short fiber lengths, predicts the loss budget of the full ~105-line H-band OH-suppression filter set with realistic fiber lengths and many notches.","fun_headline_variants_meta":{"raw":{"variants":["UV gratings: 0.5 dB cladding loss, 4.6 dB insertion; fs-IR flips","Astrophotonics filters: trade low insertion for low cladding loss","Bridge cuts UV grating insertion loss by ~3 dB","fs-IR SMF-28 insertion <0.05 dB; UV high-NA cladding 0.5 dB","Fiber grating loss trade-off: choose your poison for astronomy"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000498,"raw_usage":{"total_tokens":2577,"prompt_tokens":1218,"completion_tokens":1359,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":834,"completion_tokens_details":{"reasoning_tokens":1249}},"tokens_in":834,"tokens_out":1359,"duration_ms":11126,"temperature":1.0,"reasoning_tokens":1249,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T14:50:03.875992+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check would be to inscribe the full ~105-line H-band filter set, or a dense block of at least 20–30 lines, in SM1500(4.2) and SMF-28 and measure accumulated insertion loss and cladding-mode loss; if per-line losses compound, the sub-1 dB viability conclusion would fail. A second check is to measure the splice loss of the proposed reduced-clad bridging configuration in-house, since the 0.44 dB figure comes from a manufacturer's technical bulletin rather than the authors' own measurements.","supporting_citations":[{"cited_title":"and Ellis, Simon C","cited_arxiv_id":null,"evidence_quote":"GNOSIS, the first instrument to use fiber Bragg gratings for OH suppression, defines the full 105-line context and the five-line subset source."},{"cited_title":"Monthly Notices of the Royal Astronomical Society , volume =","cited_arxiv_id":null,"evidence_quote":"PRAXIS, the first demonstration of OH suppression in a high-efficiency spectrograph, is the benchmark system these filters must feed."},{"cited_title":"Goebel and Gayathri Bharathan and Martin Ams and Maximilian Heck and Ria G","cited_arxiv_id":null,"evidence_quote":"Goebel et al. showed direct femtosecond-IR writing as an alternative route for multi-notch OH filters, setting up the fs-IR platform comparison."},{"cited_title":"and Smelser, C.W","cited_arxiv_id":null,"evidence_quote":"Krämer et al. describe the shaping-aperture method used for fs-IR spectral control and explain why it increases cladding-mode loss."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Dong et al. supply the mechanism by which high-NA mode confinement suppresses cladding-mode coupling loss."},{"cited_title":", title =","cited_arxiv_id":null,"evidence_quote":"Grobnic et al. provide the 0.2 dB cladding-mode-loss benchmark for standard fs-IR phase-mask writing in SMF-28 that frames the paper's 0.93 dB result."},{"cited_title":"Electronics Letters , year=","cited_arxiv_id":null,"evidence_quote":"Marcuse's splice-loss equation is the model used to estimate insertion loss from mode-field mismatch and angular misalignment."}],"review_version":1}