{"id":"d46924e2-faf2-4ad8-84f5-19f046c0868b","arxiv_id":"2507.14724","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A hybrid silicon and ferroelectric nematic liquid crystal Mach-Zehnder modulator achieved 102 Gbit/s PAM-4 transmission with VπL(AC)=0.3 V·cm and >67 GHz bandwidth.","lead":"Researchers demonstrated a silicon photonic modulator that uses a ferroelectric nematic liquid crystal to reach 102 Gbit/s data transmission. The device works with a low driving voltage and very wide bandwidth, showing promise for faster optical interconnects.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 102 Gbit/s PAM-4 demonstration rests on an equalized eye and TDECQ only; without a BER measurement the quantitative headline claim is not established.","rationale":"The reader's weakest_assumption focuses on the FNLC monodomain and Pockels mechanism, but the more load-bearing gap for the stated headline 'demonstrated at 102 Gbit/s PAM-4' is the complete absence of bit-error-ratio data. An equalized open eye with TDECQ = 2.56 dB is positive evidence of a working modulation scheme, but TDECQ is not a link-performance certificate; with heavy receiver equalization and substantial device loss, an open eye can coexist with raw BER above the KP4-FEC limit. The paper explicitly gives no error statistics, so the quantitative claim is not yet verified. The S21 flatness to 67 GHz and the V-pi extraction at 25 MHz are useful independent support for the high-speed and reasonably linear nature of the modulation, which is why the missing BER does not by itself invalidate the mechanism claim or the device's potential. It does mean the manuscript should remain CONDITIONAL, requiring a BER measurement or a narrowing of the claim to an eye/TDECQ demonstration. This is consistent with the reader's CONDITIONAL verdict, hence the verdict is UNCHANGED, but the reason is now focused on the transmission metric rather than on the material-state assumption.","tokens_in":5223,"tokens_out":6367,"duration_ms":78184,"concrete_test":"Repeat the 51-GBd PAM-4 transmission experiment with the same 7-tap transmitter FFE and 42-tap receiver FFE settings, and measure the bit-error ratio directly over at least 10^6 bits. Compare the raw pre-FEC BER with the KP4-FEC threshold of 2.4e-4 and report the post-FEC BER as well. If the pre-FEC BER cannot meet 2.4e-4, the 102 Gbit/s PAM-4 demonstration is not experimentally supported and the claim should be reworded accordingly. Also report repeated V-pi measurements to supply the error bars currently missing from the stated 0.3 V-cm modulation efficiency.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Fig. 5 is the sole evidence for the central quantitative claim of 102 Gbit/s PAM-4 modulation. It is an eye diagram obtained after a 7-tap transmitter FFE and a 42-tap receiver FFE, accompanied by a reported TDECQ of 2.56 dB. No BER, pre-FEC error count, or Q factor is reported anywhere in the paper. TDECQ is a standardized transmitter-side stress metric: it quantifies how well a transmitter could in principle work with an ideal reference receiver, but it does not certify that the actual link, with its measured phase-shifter loss of roughly 6.5 dB/mm and the receiver used here, operates at 102 Gbit/s with acceptable errors. The title and abstract claim '102 Gbit/s PAM-4' as a demonstrated result, and that claim depends on error statistics, not merely on an open equalized eye. This is the load-bearing missing quantity: if the raw BER is above the KP4-FEC threshold of 2.4e-4, then the device has not been demonstrated at that line rate even if the Pockels mechanism and flat S21 are exactly as reported. The fix is straightforward and should determine whether the headline claim is kept or weakened to a modulation-eye demonstration.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a hybrid silicon-organic modulator integrating a ferroelectric nematic liquid crystal (FNLC) into a silicon slot waveguide Mach-Zehnder modulator. The device is fabricated on a standard SOI MPW platform, and the authors report a modulation efficiency VπL(AC) = 0.3 V·cm, a 6-dB bandwidth exceeding 67 GHz, and a PAM-4 eye diagram at 102 Gbit/s with a TDECQ of 2.56 dB. The authors claim the first demonstration of an FNLC-based silicon photonic modulator at this data rate, attributing the high-speed response to the Pockels effect in the FNLC.","tokens_in":5499,"tokens_out":5420,"duration_ms":56754,"significance":"If the central claim is fully substantiated, this would be a noteworthy advance for liquid-crystal-based photonic modulators, showing that FNLC can support multi-100-Gbit/s modulation in a silicon-compatible platform. The work benefits from being performed in a commercial foundry MPW, with TEM-based structural verification, measured S21 up to 67 GHz with a noise floor control, and a direct eye-diagram measurement. However, as detailed below, the absence of bit-error-rate data currently prevents the headline data-rate claim from being considered a demonstrated error-free or FEC-correctable transmission.","major_comments":[{"comment":"The only evidence for the 102 Gbit/s PAM-4 claim is an equalized eye diagram and a reported TDECQ of 2.56 dB. No BER, pre-FEC error count, or Q-factor is provided. TDECQ is a transmitter-side stress metric that does not certify that the full link (including the modulator's ~6.5 dB/mm phase-shifter loss and the receiver used) operates at the claimed line rate. Please provide BER measurements (e.g., back-to-back and with the device) or, if unavailable, weaken the headline claim to \"eye diagram demonstration at 102 Gbit/s with TDECQ of 2.56 dB\" rather than \"demonstrated 102 Gbit/s PAM-4 modulation.\"","section":"Experimental Results, Fig. 5"},{"comment":"The Vπ(AC) = 3 V is extracted by fitting modulated waveforms at different wavelengths, but the fitting procedure, the number of wavelengths, and any error estimates are not reported. Since VπL(AC) = 0.3 V·cm and the inferred r33 = 29.5 pm/V are central quantitative claims, a more detailed description of the extraction method and its uncertainty is needed. If the fit is not robust, the quoted values could be inaccurate.","section":"Experimental Results, Fig. 3(a)"},{"comment":"The paper attributes the modulation to the Pockels effect based on the known properties of FNLC and states that the high-speed results confirm Pockels-based modulation. However, no direct evidence (e.g., linear scaling of the EO response with applied field, or a measurement of the electro-optic coefficient) is presented for this device. Given that the headline claim rests on the Pockels mechanism, a simple linearity test (e.g., measuring the modulation depth versus drive amplitude) would strengthen the attribution. As written, the mechanism is an assumption inferred from material properties rather than demonstrated in the device.","section":"Introduction and Experimental Results"}],"minor_comments":[{"comment":"There is a typo in the abstract: \"Gb it/s\" should be \"Gbit/s\".","section":"Abstract"},{"comment":"The phrase \"PRBS equal to 11\" should be clarified as a PRBS pattern of length 2^11−1 (PRBS11).","section":"Experimental Results"},{"comment":"The sentence \"While some unwanted ripple was observed in the S21 data, and the S11 results indicated higher reflections at lower frequencies\" is awkward; consider splitting it into two sentences for clarity.","section":"Experimental Results"},{"comment":"The statement that \"the S21 signals remained approximately 20 dB above the noise floor up to 65 GHz\" appears to conflict with the earlier statement that noise became significant above 45 GHz; please reconcile these two observations.","section":"Experimental Results"},{"comment":"The caption should specify what the \"input signal\" and \"modulated waveform\" are, and indicate at which wavelengths the measurements were taken.","section":"Fig. 3(a) caption"},{"comment":"Reference [17] lists the journal as \"Optics Letter\" but the correct title is \"Optics Letters\".","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is a conference-style report with a promising device, but the headline claim of 102 Gbit/s PAM-4 is not fully supported by the data as presented. I recommend requesting BER measurements or a revised claim that accurately reflects the evidence. The VπL and r33 extraction also need more detail to be fully credible. If the authors cannot provide BER data, the paper could still be published as a demonstration of eye opening, but the title and abstract should be adjusted accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the good: this is the first FNLC-based silicon photonic modulator to run PAM-4 at 102 Gbit/s, with a VπL of 0.3 V·cm and an EO response that is flat to at least 67 GHz. That is a legitimate new data point, built on prior integration work from the same group (ref. 9) but going well beyond it in signaling speed. The device is fabricated on a standard MPW run, the eye diagram is measured with real equalization, and the authors are upfront about the 6.5 dB/mm phase-shifter loss and the instrumentation-limited bandwidth. Credit where due: the claim is specific, the demonstration is reproducible in principle if you have access to PM616, and the paper frames itself clearly against the EO polymer SOH literature.\n\nThe soft spot is exactly what the stress-test flags: the 102 Gbit/s headline rests on a TDECQ of 2.56 dB from an equalized eye (7-tap TX FFE, 42-tap RX FFE) and no BER or pre-FEC error count anywhere. TDECQ is a legitimate transmitter quality metric, but it measures what an idealized reference receiver could do, not what this link actually does. The title says 'demonstrated at 102 Gbit/s'; a strict reading requires error statistics. That is a small fix — add a BER curve or at least a measured BER below KP4-FEC threshold — and it does not change the underlying value of the result. Other minor issues: the r33 of 29.5 pm/V is back-extracted from Vπ and simulation, so treat it as an estimate; the Pockels mechanism is inferred from the flat S21 and material properties, not directly proven; and the proprietary FNLC mixture makes independent replication harder. None of these are fatal, but they cap the paper's current evidentiary level.\n\nWho is this for? The silicon photonics / optical interconnect crowd, especially people tracking SOH-type modulators. It deserves a serious referee: the novelty is real, the design is clearly described, and the missing BER is easily supplied. I would send it to review with a request for error statistics or a softened headline, rather than desk reject it.","headline":"Real first for FNLC modulators at 102 Gbit/s, but the headline claim needs a BER number or a softer verb; otherwise, the paper earns a fair but conditional review.","tokens_in":6051,"tokens_out":2207,"would_cite":true,"duration_ms":26662,"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":"A silicon photonic modulator using a ferroelectric nematic liquid crystal demonstrates 102 Gbit/s PAM-4 at 3.5 V bias, with VπL(AC)=0.3 V·cm and bandwidth above 67 GHz.","keywords":["ferroelectric nematic liquid crystal","silicon photonics","optical modulator","Pockels effect","PAM-4","slot waveguide","Mach-Zehnder modulator","C-band"],"falsifier":"Measure the small-signal EO response at 25 MHz while sweeping the DC bias from -5 V to +5 V and simultaneously monitor the optical phase shift; also heat the device through the FNLC's phase transition. A Pockels response must reverse sign when the bias polarity reverses, vanish as the polar order is destroyed at the transition, and reproduce the same extracted r33 tensor, whereas a reorientation-driven or quadratic response would not.","tokens_in":5068,"feed_emoji":"⚡","tokens_out":8882,"duration_ms":92476,"temperature":0.7,"pith_summary":"This paper aims to establish that a ferroelectric nematic liquid crystal (FNLC) can serve as the active material in a silicon photonic modulator, reaching data rates that were previously the preserve of organic-polymer or plasma-dispersion devices. The authors demonstrate a hybrid silicon slot-waveguide Mach-Zehnder modulator in which the FNLC fills the slot and acts as the electro-optic cladding, achieving 102 Gbit/s PAM-4 transmission at a 3.5 V DC bias. The reported modulation efficiency is VπL(AC)=0.3 V·cm at 25 MHz, with an electrical 6-dB bandwidth above 67 GHz limited by the test equipment and an extracted in-device r33 of 29.5 pm/V. The significance, if the result holds, is that FNLCs offer Pockels-effect modulation without the high-field poling step required by organic electro-optic polymers, making them scalable and repeatable for future silicon photonic transceivers.","feed_headline":"First ferroelectric-nematic silicon modulator hits 102 Gbit/s","feed_subtitle":"Ferroelectric nematic liquid crystals give low-voltage Pockels modulation without poling, on a standard silicon photonics process.","key_machinery":"The load-bearing object is the hybrid Si-FNLC slot-waveguide phase shifter inside a ground-signal-ground-signal-ground (GSGSG) traveling-wave Mach-Zehnder modulator. The ferroelectric nematic liquid crystal, a liquid crystal phase with spontaneous electric polarization discovered in 2020, is both the dielectric cladding and the electro-optic material in the roughly 125 nm slot, where the TE0 optical mode and the applied horizontal electric field overlap. A modest 3.5 V DC bias aligns the FNLC's spontaneous polarization into a monodomain, and the linear Pockels response then provides high-speed index change. Push-pull driving of two phase shifters doubles the efficiency and gave measured VπL(AC)=0.3 V·cm at 25 MHz, from which the authors extract an in-device r33 of 29.5 pm/V using TEM-measured waveguide dimensions and mode simulations.","core_discovery":"The central claim is that a ferroelectric nematic liquid crystal, used as both the cladding and the slot-filling active medium of a silicon slot waveguide, supports linear Pockels-effect modulation fast enough for 100 Gbit/s-class data while keeping the drive voltage and phase-shifter length small. With a 3.5 V DC bias establishing a monodomain and a 1.9 Vpp differential signal in push-pull, the modulator produced an open 102 Gbit/s PAM-4 eye after equalization, an extinction ratio of 21.8 dB, and a free spectral range of 7.5 nm. The authors assign the high speed to the intrinsic Pockels response of the FNLC phase rather than to the kHz-scale molecular reorientation that limits ordinary nematics.","pith_inferences":["A direct transfer of this design to micro-ring modulators or directional couplers could trade some of the demonstrated bandwidth for a much smaller device footprint, though the paper does not test those geometries.","Because the FNLC mixture is proprietary, independent replication would require either a supplier of the same material or a published reference mixture with known r33, something the paper does not provide.","The claimed thermodynamic stability of the monodomain under DC bias is not backed by long-term or temperature-cycling data, so reliability over time is an open question the demonstration does not address.","A head-to-head reference measurement using the same test electronics on a bare silicon MZM would clarify how much of the 102 Gbit/s result is attributable to the FNLC device itself."],"forward_implications":["100 Gbit/s-class PAM-4 is achievable with an FNLC-loaded silicon slot waveguide, with modulation efficiency VπL(AC)=0.3 V·cm and a drive signal of only 1.9 Vpp after equalization.","Because FNLC alignment requires only a modest DC bias instead of polymer poling, fabricating these modulators can follow a standard silicon photonics MPW flow, including an oxide-open step, with post-foundry FNLC loading.","The f6dB measurement exceeding 67 GHz and limited only by the network analyzer means the intrinsic speed of the FNLC Pockels effect is not observed to roll off within the measured band.","An in-device r33 of 29.5 pm/V extracted from the measured Vπ is already comparable to some poled organic electro-optic polymers, and the paper forecasts further gains from next-generation FNLC mixtures.","The current -6.5 dB/mm phase-shifter loss is identified as the main efficiency limiter; reducing sidewall roughness and shortening the phase shifter are concrete routes to lower insertion loss."],"supporting_citations":[{"why":"Reports the discovery of ferroelectric nematic liquid crystals and their spontaneous polar order, the physical basis for poling-free Pockels modulation.","marker":"[8]"},{"why":"Demonstrates the same hybrid silicon-slot/FNLC integration for poling-free Pockels-effect modulation that this work extends to 102 Gbit/s PAM-4.","marker":"[9]"},{"why":"Shows the ultra-high electro-optic activity achievable in silicon-organic hybrid modulators, the efficiency/bandwidth benchmark FNLC devices are compared with.","marker":"[4]"},{"why":"Reports 100 GBd PAM4 signaling in an SOH Mach-Zehnder modulator, the closest prior speed/efficiency benchmark this device aims to match without poling.","marker":"[5]"},{"why":"Established the electro-optic polymer-clad silicon slot waveguide geometry that yields low VπL and GHz bandwidth, which the FNLC device inherits.","marker":"[12]"},{"why":"Documents the high-field poling required for organic electro-optic polymers, the step that FNLC's spontaneous polarization avoids.","marker":"[13]"},{"why":"Describes ferroelectric nematic materials for high-speed electro-optic applications, supporting the material-level feasibility of the Pockels mechanism.","marker":"[10]"}],"fun_headline_variants":["Ferroelectric nematic silicon modulator hits 102 Gbit/s PAM-4","Silicon + ferroelectric nematic: 102 Gbit/s PAM-4","Pockels silicon modulator reaches 102 Gbit/s with FNLC","Low-V silicon modulator: 102 Gbit/s PAM-4 achieved","First FNLC silicon photonic modulator does 102 Gbit/s"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the proprietary PM616 mixture stays in a stable single-domain ferroelectric nematic state under the 3.5 V bias, so that the measured high-speed modulation is a genuine Pockels effect rather than a reorientation or thermal response.","fun_headline_variants_meta":{"raw":{"variants":["Ferroelectric nematic silicon modulator hits 102 Gbit/s PAM-4","Silicon + ferroelectric nematic: 102 Gbit/s PAM-4","Pockels silicon modulator reaches 102 Gbit/s with FNLC","Low-V silicon modulator: 102 Gbit/s PAM-4 achieved","First FNLC silicon photonic modulator does 102 Gbit/s"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000655,"raw_usage":{"total_tokens":2915,"prompt_tokens":775,"completion_tokens":2140,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":391,"completion_tokens_details":{"reasoning_tokens":2037}},"tokens_in":391,"tokens_out":2140,"duration_ms":16451,"temperature":1.0,"reasoning_tokens":2037,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:48:55.644682+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the small-signal EO response at 25 MHz while sweeping the DC bias from -5 V to +5 V and simultaneously monitor the optical phase shift; also heat the device through the FNLC's phase transition. A Pockels response must reverse sign when the bias polarity reverses, vanish as the polar order is destroyed at the transition, and reproduce the same extracted r33 tensor, whereas a reorientation-driven or quadratic response would not.","supporting_citations":[{"cited_title":"Hybrid integration of silicon slot photonics with ferroelectric nematic liquid crystal for poling-free Pockels-effect modulation,","cited_arxiv_id":null,"evidence_quote":"Demonstrates the same hybrid silicon-slot/FNLC integration for poling-free Pockels-effect modulation that this work extends to 102 Gbit/s PAM-4."},{"cited_title":"Ultra- high electro-optic activity demonstrated in a silicon-or- ganic hybrid modulator,","cited_arxiv_id":null,"evidence_quote":"Shows the ultra-high electro-optic activity achievable in silicon-organic hybrid modulators, the efficiency/bandwidth benchmark FNLC devices are compared with."},{"cited_title":"Silicon-organic hybrid (SOH) Mach-Zehnder modulators for 100 GBd PAM4 signaling with sub-1 dB phase-shifter loss,","cited_arxiv_id":null,"evidence_quote":"Reports 100 GBd PAM4 signaling in an SOH Mach-Zehnder modulator, the closest prior speed/efficiency benchmark this device aims to match without poling."},{"cited_title":"Demonstration of a low VπL modulator with GHz band- width based on electro-optic polymer-clad silicon slot waveguides,","cited_arxiv_id":null,"evidence_quote":"Established the electro-optic polymer-clad silicon slot waveguide geometry that yields low VπL and GHz bandwidth, which the FNLC device inherits."},{"cited_title":"Electric field poled organic electro-optic materials: state of the art and future prospects,","cited_arxiv_id":null,"evidence_quote":"Documents the high-field poling required for organic electro-optic polymers, the step that FNLC's spontaneous polarization avoids."},{"cited_title":"Ferroelectric nematic materials for high-speed electro-optic applications,","cited_arxiv_id":null,"evidence_quote":"Describes ferroelectric nematic materials for high-speed electro-optic applications, supporting the material-level feasibility of the Pockels mechanism."}],"review_version":1}