{"id":"b70676a4-8226-49b6-b1dd-6a9b31f4be6c","arxiv_id":"2504.20175","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The authors present simulated and one measured D-band RIS unit-cell designs using five switching technologies, reporting insertion losses and beam steering patterns.","lead":"This paper evaluates five switching technologies for reconfigurable intelligent surfaces (RIS) at D-band (110-170 GHz), using simulations and one experiment. It compares Schottky diodes, memristors, liquid metal, phase-change materials, and RF-SOI switches for reflective and transmissive designs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Simulation-only 'potential' claim rests on undisclosed switch models; experimental support covers only liquid metal.","rationale":"The reader's weakest-assumption correctly identifies the load-bearing dependence on unverified simulations and switch models. I agree that the liquid-metal measurement is genuine experimental support, but it cannot validate the Schottky, memristor, PCM, or RF-SOI designs. My concern sharpens the issue: not only are the simulations unmeasured, but the text omits the actual switch model parameters for three of the four simulation-only technologies, making the reported insertion losses and phase shifts impossible to assess or reproduce from the paper alone. The RF-SOI section gives Ron, Con, Roff, and Coff, but even these values are approximate and no measurement is reported. The central claim is phrased as 'demonstrate the potential,' which is modest, but the conclusion in Section III states 'This work has demonstrated significant progress in switching technologies' without distinguishing validated from simulated. That overstatement supports a conditional rather than full accept. The internal inconsistency between a '30 GHz' bandwidth and a '121-158 GHz' (37 GHz) range in II-B2 is a minor flaw and not load-bearing. The recommended verdict remains CONDITIONAL, matching the reader; no change is needed because the conditionality already captures the concern.","tokens_in":7187,"tokens_out":5709,"duration_ms":64384,"concrete_test":"Fabricate the PCM-based T-RIS unit-cell of Section II-B1 and measure S21 at 140 GHz in both GeTe states, comparing with the simulated 0.69 dB insertion loss and 180-degree phase difference shown in Fig. 9. If the measured insertion loss exceeds about 2 dB or the phase difference deviates from 180 degrees by more than 30 degrees, the central claim of PCM viability is not supported by current evidence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract concludes that Schottky diodes, memristors, PCM, and RF-SOI technology show potential for D-band RIS alternatives. For Schottky (II-A1), memristor (II-A2), and PCM (II-B1), the evidence is entirely full-wave simulation, and the text does not report the switch equivalent-circuit or material-state parameters used for the diode, memristor, or GeTe switches. The simulated insertion losses, 180-degree phase differences, and RCS patterns therefore cannot be checked against realistic D-band device behavior. The only measured design is the liquid-metal beam splitter (II-A3); that experiment validates the grating structure, not the electronic or phase-change switch models. If those models idealize switch performance, the central claim of overcoming traditional-technology limitations is not established for four of the five technologies. The RF-SOI switch values in II-B2 are stated, but the ON/OFF contrast at 140 GHz is modest and no fabricated RF-SOI RIS is measured. The liquid-metal result is solid independent support, but it does not carry the other four simulations.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript evaluates five switching technologies for D-band (110-170 GHz) reconfigurable intelligent surfaces (RISs): Schottky diodes, memristor switches, liquid metal, phase-change materials (PCM), and 45 nm RF-SOI. It presents unit-cell and small-array full-wave simulations for reflective (Schottky, memristor, liquid metal) and transmissive (PCM, RF-SOI) designs, with 1-bit phase control in the electronic and PCM cases, and reports one experimental validation for a liquid-metal-based reconfigurable beamsplitter at 150 GHz. The abstract and conclusion claim that these advanced components are viable alternatives to traditional PIN-diode or varactor approaches for sub-THz RISs.","tokens_in":7344,"tokens_out":3315,"duration_ms":34783,"significance":"If the reported simulations faithfully represent realistic device behavior, the paper provides a useful comparative view of emerging switching technologies for D-band RIS design, a topic of current interest for 6G. The liquid-metal beamsplitter is a concrete, measured demonstration at 150 GHz, and the RF-SOI design gives explicit switch equivalent-circuit values. However, the central claim of viability rests primarily on simulation-only results: four of the five technologies have no experimental backing, and for three of those the switch models are not disclosed. The paper's value as an 'evaluation' is therefore limited unless the simulation inputs are provided and the conclusions are scaled to match the evidence.","major_comments":[{"comment":"The text states that Fig. 12(a) indicates \"a wide transmission bandwidth of 30 GHz with less than 2 dB return loss over 121 GHz-158 GHz with absolute bandwidth of 37 GHz.\" This is internally inconsistent: the range 121-158 GHz spans 37 GHz, and in a transmissive unit-cell the relevant loss metric is insertion loss (|S21|), not return loss. A return loss below 2 dB would indicate poor matching. The sentence needs to be rewritten to specify which S-parameter is meant (presumably |S21| < 2 dB over 121-158 GHz) and to reconcile the 30 GHz and 37 GHz bandwidth numbers.","section":"Section II-B2, last paragraph before Fig. 12"},{"comment":"The Schottky diode, memristor, and PCM switch models are not specified. For the Schottky design the text mentions only that a diode is integrated; for the memristor no equivalent-circuit parameters are given; for the PCM design the GeTe switches are described qualitatively (amorphous/crystalline) but the electrical conductivity or sheet resistance in each state is not reported. These parameters are direct inputs to the full-wave simulations and determine the simulated insertion loss and phase response. Without them, the reader cannot assess whether the simulated performance reflects realistic D-band devices or idealized switch behavior, which is load-bearing for the claim that these technologies are viable alternatives. Please report the switch models and parameter values for all three cases.","section":"Sections II-A1, II-A2, II-B1"},{"comment":"The conclusion states \"We provided both simulation and experimental validation of various RIS architectures,\" but the only experimental validation in the paper is the liquid-metal beamsplitter at 150 GHz. The Schottky, memristor, PCM, and RF-SOI designs are simulation-only. This overstatement should be corrected so that the conclusion accurately reflects the validation status of each technology, and the abstract should be aligned as well if it implies broader experimental support.","section":"Section III, Conclusion"}],"minor_comments":[{"comment":"The text says \"losses below 1.5 dB over a 27% bandwidth\" and also reports 0.69 dB at 140 GHz; please state explicitly whether the 27% bandwidth refers to the condition that insertion loss stays below 1.5 dB or to another criterion, and define the center frequency used for the percentage.","section":"Section II-B1, Fig. 9 description"},{"comment":"The phrase \"each liquid channel has an in/outlet\" should read \"inlet/outlet,\" and the sentence \"which is much higher than water and ten times below copper\" is awkward; it should be rewritten for clarity, e.g., \"its conductivity is much higher than that of water and about one order of magnitude below that of copper.\"","section":"Section II-A3, Fig. 5"},{"comment":"The caption in the text reads \"(a) Unit-cell with an air-gap of 50 µm (b) S-parameter over 110-180 GHz\"; it would be clearer to state which S-parameters are shown in (b) (e.g., |S21| and phase) for the two switch states.","section":"Section II-B2, Fig. 12 caption"},{"comment":"The claim that memristors provide \"non-volatile switching ... eliminating the need for continuous voltage application\" could be misread as eliminating any voltage during operation; non-volatile devices still require a programming pulse to change state. The wording can be made more precise.","section":"Section II-A2, memristor paragraph"}],"recommendation":"major_revision","confidential_remarks":"The paper is broad in scope, covering five technologies in a short format, which limits depth. The main concerns are fixable: adding the missing switch-model parameters, correcting the RF-SOI bandwidth/insertion-loss statement, and aligning the conclusion with the actual evidence. I would not reject, but the current version does not fully support the claimed breadth of validation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the measured liquid-metal reconfigurable beamsplitter at 150 GHz, where simulation and experiment agree on the transition from one beam to three and five beams as the period changes. That is a real result and worth building on. The three other reflective/transmissive designs (Schottky, memristor, PCM, RF-SOI) are unit-cell simulations with no fabricated hardware, and the PCM cell is close to prior work [19] except for an O-slot and GeTe switches instead of whatever [19] used. So the paper is best read as a design-evaluation study with one experimental anchor.\n\nThe liquid-metal section is the strongest: the grating concept is simple, the fabricated structure is described with enough detail to reproduce, and the measured patterns match simulation well. The simulations for the other technologies look professionally done, but they are only as credible as the switch models embedded in them, and the paper does not report the equivalent-circuit or material-state parameters for the Schottky diode, the memristor, or the GeTe PCM switches. That makes it impossible to check whether those insertion-loss and phase-difference numbers are realistic. The RF-SOI section at least gives Ron/Roff/Con/Coff values, but it contains a concrete internal inconsistency: the text claims a 30 GHz transmission bandwidth with less than 2 dB return loss over 121–158 GHz, which is a 37 GHz span, and the 'return loss' wording is suspect for a transmission response. That needs fixing in revision.\n\nThe abstract's sentence about 'demonstrating the potential' of all five technologies goes beyond the evidence. The measured liquid-metal result demonstrates the grating concept; the other four demonstrate only that good full-wave models with optimistic switch parameters can produce nice-looking RCS patterns. The paper would be more honest as 'design studies of candidate switching technologies, with one experimental proof-of-concept.' The citation pattern is fine; the TERRAMETA context explains the author overlap with earlier transmitarray work, and the prior PCM paper is cited.\n\nFor a reader in the sub-THz RIS hardware community, this is a useful snapshot of what is on the bench. It deserves serious peer review, but with a request for the switch-model parameters and a corrected RF-SOI bandwidth statement. I would not cite the simulation-only sections as evidence for any specific technology, but I would cite the liquid-metal beamsplitter as a reproducible measured data point.","headline":"One measured liquid-metal beamsplitter plus four simulation-only switch studies: the measurement is solid, the simulations are plausible but the 'demonstrated potential' claim overreaches, and the RF-SOI bandwidth numbers are internally inconsistent.","tokens_in":7982,"tokens_out":1857,"would_cite":true,"duration_ms":20927,"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 Schottky diodes, memristors, liquid metal, phase-change materials, and RF-SOI switches can replace PIN diodes in D-band reconfigurable intelligent surfaces.","keywords":["reconfigurable intelligent surface","D-band","switching technologies","Schottky diode","memristor","liquid metal","phase change material","RF-SOI"],"falsifier":"Fabricate the Schottky, memristor, PCM, and RF-SOI unit cells and measure their two-state S-parameters at 140 GHz. If the ON/OFF phase difference deviates substantially from 180 degrees or insertion/return loss exceeds the simulated values—for example, PCM insertion loss above 1.5 dB at 140 GHz or RF-SOI bandwidth narrower than 121–158 GHz—the paper's central viability claim would be refuted.","tokens_in":6997,"feed_emoji":"📡","tokens_out":6976,"duration_ms":62807,"temperature":0.7,"pith_summary":"The paper argues that the usual RIS switches—PIN diodes and RF-MEMS—become lossy and slow at D-band, and that five alternatives can take their place: Schottky diodes, memristors, liquid metal, phase-change materials (GeTe), and 45 nm RF-SOI CMOS switches. For each technology it designs a unit cell and checks reflection or transmission across 110–170 GHz with full-wave simulations; for the liquid-metal design it also fabricates a beamsplitter and measures it at 150 GHz. If the simulations are right, these results give hardware designers a menu of practical bit-level phase controls for sub-THz reconfigurable surfaces, which matters because 6G networks are expected to rely on such surfaces for beam steering.","feed_headline":"Five switching technologies open the D-band for smart surfaces","feed_subtitle":"Simulations and one measured 150-GHz liquid-metal test back five D-band switch alternatives to PIN diodes","key_machinery":"The carrying objects are the unit cells themselves, each paired with a different switching mechanism. Reflective cells use a planar tightly coupled bowtie dipole with the switch bridging the two halves, so 'ON' and 'OFF' states produce about 180 degrees of phase difference; the liquid-metal cell instead changes the period of a strip grating by pumping EGaIn in and out of microchannels. Transmissive cells use a GeTe-based O-slot patch that toggles between amorphous and crystalline states, and an RF-SOI Fabry-Perot resonator whose two I-shaped rotators are shunted by CMOS switches. In every case the mechanism is 1-bit phase quantization: the surface steers beams by flipping each element between two phase states, and the paper's simulations measure the insertion loss and phase error this quantization introduces.","core_discovery":"The paper's central claim, stated in the abstract, is that advanced switching components are viable alternatives to traditional technologies in D-band RISs. Concretely: a Schottky-diode reflective cell and a memristor reflective cell each steer a $20\\times20$ aperture to 30 degrees; a liquid-metal grating acts as a reconfigurable beamsplitter, splitting one normally incident beam into three or five beams as channel spacing grows from 2 mm to 4 or 6 mm, confirmed by measurement at 150 GHz; a GeTe (germanium telluride) phase-change transmissive cell gives two states separated by 180 degrees with 0.69 dB insertion loss at 140 GHz and under 1.5 dB over 27% bandwidth; and an RF-SOI (silicon-on-insulator) transmissive cell offers a 37 GHz transmission band with less than 2 dB return loss. All non-liquid-metal results are simulation-based.","pith_inferences":["Because all non-liquid-metal results rest on full-wave simulation with assumed switch equivalent circuits, a natural next step is to fabricate the Schottky, memristor, PCM, and RF-SOI cells and compare measured reflection/transmission phase and loss; this would tighten or revise the switch models.","The technologies differ sharply in control speed: Schottky and RF-SOI switch in nanoseconds or faster, while liquid metal and PCM are slower but non-volatile, suggesting hybrid RIS designs could pair a fast electronic array with a non-volatile reconfiguration layer.","The same unit-cell topologies might extend beyond the D-band, but switch parasitics and substrate losses grow with frequency, so the viability ranking could change near 300 GHz; the Fabry-Perot transmitarray reference already points to 300 GHz operation with PCB technology."],"forward_implications":["D-band RISs can be designed without PIN diodes or MEMS: Schottky and memristor cells show simulated beam steering to 30 degrees with only the phase error expected from 1-bit quantization.","Phase-change transmissive RISs can reach low insertion loss at 140 GHz—0.69 dB at center, below 1.5 dB over a 27% bandwidth—while retaining non-volatile state retention.","RF-SOI CMOS switches support a wide 37 GHz transmission band (121–158 GHz) with less than 2 dB return loss, making them suitable for broadband D-band operation.","Liquid-metal gratings can serve as reconfigurable beamsplitters at 150 GHz, with the number of output beams controlled by the channel period; this is the only design in the paper confirmed by measurement."],"supporting_citations":[{"why":"Frames the switching-technology comparison for THz RISs and identifies the hardware-impairment challenges.","marker":"[7]"},{"why":"Establishes the limitation of PIN diodes at THz and sub-THz frequencies that motivates all five alternatives.","marker":"[10]"},{"why":"Supplies the non-volatile memristor-based 1-bit RIS concept and the energy-efficiency rationale.","marker":"[11]"},{"why":"Justifies Schottky diodes as low-parasitic high-frequency switches for tunable metamaterial phase shifters.","marker":"[16]"},{"why":"Supports Schottky-based sub-THz transmissive reconfigurable surfaces.","marker":"[17]"},{"why":"Provides the basis for reversible amorphous-crystalline switching in phase-change materials.","marker":"[18]"},{"why":"Offers the PCM-based programmable transmitarray approach that the paper's T-RIS extends.","marker":"[19]"},{"why":"Demonstrates RF-SOI switches operating up to 220 GHz and supplies the ON/OFF circuit values used in the unit-cell model.","marker":"[20]"}],"fun_headline_variants":["Liquid metal joins five-way race for D-band smart surfaces","D-band RISs get five new switch options, one measured at 150 GHz","Schottky, memristor, liquid metal, GeTe, RF-SOI: D-band switch alternatives","Beyond PIN diodes: five technologies for D-band reconfigurable surfaces","Simulations plus one 150-GHz test back five RIS switch technologies"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The weakest load-bearing premise is that the full-wave simulations, with their assumed substrate losses, metal properties, and switch equivalent circuits, accurately predict how real D-band devices behave; only the liquid-metal beamsplitter has measurement behind it.","fun_headline_variants_meta":{"raw":{"variants":["Liquid metal joins five-way race for D-band smart surfaces","D-band RISs get five new switch options, one measured at 150 GHz","Schottky, memristor, liquid metal, GeTe, RF-SOI: D-band switch alternatives","Beyond PIN diodes: five technologies for D-band reconfigurable surfaces","Simulations plus one 150-GHz test back five RIS switch technologies"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000415,"raw_usage":{"total_tokens":2097,"prompt_tokens":851,"completion_tokens":1246,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":467,"completion_tokens_details":{"reasoning_tokens":1157}},"tokens_in":467,"tokens_out":1246,"duration_ms":10140,"temperature":1.0,"reasoning_tokens":1157,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T05:34:42.678577+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fabricate the Schottky, memristor, PCM, and RF-SOI unit cells and measure their two-state S-parameters at 140 GHz. If the ON/OFF phase difference deviates substantially from 180 degrees or insertion/return loss exceeds the simulated values—for example, PCM insertion loss above 1.5 dB at 140 GHz or RF-SOI bandwidth narrower than 121–158 GHz—the paper's central viability claim would be refuted.","supporting_citations":[{"cited_title":"Non-volatile memristor-based 1-bit reconfigurable intelligent surface towards a greener 6g,","cited_arxiv_id":null,"evidence_quote":"Supplies the non-volatile memristor-based 1-bit RIS concept and the energy-efficiency rationale."},{"cited_title":"Air-bridged schottky diodes for dynamically tunable millimeter- wave metamaterial phase shifters,","cited_arxiv_id":null,"evidence_quote":"Justifies Schottky diodes as low-parasitic high-frequency switches for tunable metamaterial phase shifters."},{"cited_title":"Sub-terahertz transmissive reconfigurable in- telligent surface for integrated beam steering and self-ook-modulation,","cited_arxiv_id":null,"evidence_quote":"Supports Schottky-based sub-THz transmissive reconfigurable surfaces."},{"cited_title":"Reversible switching in phase-change materials,","cited_arxiv_id":null,"evidence_quote":"Provides the basis for reversible amorphous-crystalline switching in phase-change materials."},{"cited_title":"Design of a binary programmable transmitarray based on phase change material for beam steering applications in D- band,","cited_arxiv_id":null,"evidence_quote":"Offers the PCM-based programmable transmitarray approach that the paper's T-RIS extends."},{"cited_title":"140–220 GHz SPST and SPDT switches in 45 nm CMOS SOI,","cited_arxiv_id":null,"evidence_quote":"Demonstrates RF-SOI switches operating up to 220 GHz and supplies the ON/OFF circuit values used in the unit-cell model."}],"review_version":1}