ADAS Sensor Window Coating: Hydrophobic, Anti-Fog & Optical Cost

Functional coatings for intelligent-driving perception windows are thin functional layers applied to the outer surfaces of camera cover lenses and protective glazing, LiDAR transmit/receive windows, millimeter-wave radomes, and in-cabin sensing-module glass. Their job is to control how water, fog, frost, and soil interfere with these optical and electromagnetic channels while keeping the cost in visible transmittance, haze, near-infrared transmittance, and millimeter-wave insertion loss inside the perception system's budget. What they deliver is not "it sheds water" or "it does not fog" but locking several mutually repelling sets of physical quantities inside one budget on the same surface: wetting state (contact angle, roll-off angle, contact-angle hysteresis), optical quality (total transmittance, haze, residual reflection color), electromagnetic transmission (permittivity, loss tangent, sheet resistance), mechanical durability (post-wear performance retention) and chemical-climate durability (salt, alcohol, ammonia, UV).

A typical field case happens on an elevated road in the rain. A car with 11 cameras and 1 forward LiDAR runs a navigation-assist feature in a downpour; the cockpit warns that perception is limited and asks the driver to take over. In the returned data, the front telephoto camera's stable recognition distance shrinks from roughly 180 m to the 90 m order, the point cloud's range jitter on low-reflectivity targets grows, and ten minutes after the rain stops everything is normal again. The hardware check finds no damage — the problem sits on the window surface: the aftermarket water-repellent coating, worn thin within half a year through tens of car washes and tens of thousands of wiper sweeps, no longer lets water bead and roll off but smears it into a water curtain. The curtain itself is a lens and a scatterer: imaging contrast falls, distant target edges go soft, and the perception model first degrades, then exits. Another version of the same accident chain is winter: a layer of fog condenses on the inner side of an exterior camera's glazing, or snow outside is pressed into an ice shell by body heat — and no matter how "hydrophobic" the coating is, it cannot save you, because hydrophobicity governs liquid water while ice is a solid-adhesion problem. This article takes the window apart as a coating-design problem: first why the requirement matrix fights itself, then three executable sets of quantified criteria (surface, optical, electromagnetic), then durability test combinations organized by failure mode, and finally a vehicle-qualification and aftermarket judgment process that can actually be run.

TL;DR: conclusions first

• Hydrophobicity and anti-fog point in opposite surface-energy directions; one layer cannot do both. Hydrophobicity relies on low surface energy (fluoro-silicone systems at 16～24 mN/m) plus micro/nano roughness to produce contact angles above 150° and roll-off angles below 10° in the Cassie state; anti-fog relies on superhydrophilicity that spreads water into a continuous thin film that does not refract (contact angle approaching 0°). "Bead" and "film" are mutually exclusive states at the same point, so engineering leaves only three routes: zoning (a hydrophobic area plus a hydrophilic area, or micro-patterned surfaces), removing fog with heat and structure instead of expecting it from the coating, or explicitly sacrificing one of the two.
• The first acceptance item for a window coating is optics, not contact angle. Total transmittance and haze are measured on the integrating-sphere basis of GB/T 2410-2008 (modified adoption of ASTM D1003:2007): haze is the percentage of transmitted light scattered more than 2.5° away from the incident direction relative to total transmission. Transparent superhydrophobicity and low haze are naturally at odds (once scattering centers approach 0.4～0.7 μm, haze rises), so acceptance should write "post-coating haze increase" into the specification rather than "contact angle".
• The "quarter wavelength" differs by three orders of magnitude across physical windows, and that decides whether a coating may go on a radome. n·d=λ/4: at 550 nm of visible light, a coating of refractive index 1.38 needs only 99.6 nm; at 905 nm it needs 164 nm; at 77 GHz (free-space wavelength 3.893 mm) a matching layer of the same class of material would need 0.56～0.70 mm. Liquid coatings cannot produce — and should not produce — that thickness: millimeter-wave impedance matching is structural-part language, and at radar frequencies a coating is merely an "electrically ultra-thin dielectric film" (30 μm is about λ/130).
• The real risk on the millimeter-wave side is not film thickness but dielectric loss and a continuous conductive network. Estimated from Debye parameters, water's absorption coefficient at 77 GHz is about 32.6 dB/mm; a continuous 50 μm water film alone costs 1.63 dB one-way, which converts via the radar equation into roughly 18% off the detection-range ceiling; a low-loss coating of the same thickness sits on the 0.01 dB order. Conversely, once a coating forms a continuous conductive path at 10²～10³ Ω per square, the transmission-line model gives an insertion loss of 9.20～1.50 dB — functional fillers such as graphene, carbon nanotubes, and silver nanowires are inputs that "must be re-reviewed as a shielding layer" on radar windows.
• Bands and regulatory bases must not be written wrong. China's automotive radar follows 工信部无〔2021〕181 号, the Interim Provisions on Automotive Radar Radio Management: 76-79 GHz is planned, effective 2022-03-01, with 76-77 GHz limits of 55 dBm peak / 50 dBm average / 28 dBm/MHz spectral density, and 77-79 GHz at 39 / 34 dBm with 7 dBm/MHz; Europe runs 77-81 GHz on ETSI EN 302 264 (with Commission Decision 2004/545/EC). ETSI EN 302 858 governs short-range automotive radar at 24.05-24.25 / 24.05-24.50 GHz, not 77 GHz, and citing the wrong standard voids the whole RF report.
• Durability must be verified as a set; the initial contact angle alone proves nothing. Post-abrasion hydrophobic retention uses ASTM D8380-21 (dry abrasion of water-repellent and omniphobic coatings) plus Taber abrasion (GB/T 1768-2006, ASTM D4060, CS-17 / 1 kg) plus GB/T 45017-2024, Test methods for mechanical stability of superhydrophobic surfaces; what the anti-fog layer fears is organic contamination — fresh sol-gel TiO₂ films show a 0° water contact angle, turn hydrophobic after storage in laboratory air as airborne organics adsorb, and can be regenerated photocatalytically (Science in China Series B, 2002 No. 5) — so "night, tunnels, no UV in the cabin" is the true failure window of hydrophilic anti-fog coatings.
• Vehicle language is automotive environmental testing, not cosmetics-style claims. The verification combination should cite the GB/T 28046 series (equated to ISO 16750: environmental conditions and testing for electrical and electronic equipment of road vehicles, including climatic and chemical loads), the IP codes of GB/T 30038-2013 (IPX9K is 80±5 ℃ high-pressure hot water at 8～10 MPa), and the abrasion test of GB/T 5137.1-2020; while GB 11555-2025 governs only the driver's vision: ice formed at -18±3 ℃ by spraying 0.044 g/cm² × glass area, defrosting 80% of Zone A in 20 min; steaming at -3±1 ℃ with 70 g/h × number of seats, demisting 90% of Zone A in 10 min — sensor windows are outside it. That is both the boundary and the opportunity.
• Do not casually add a coating on camera / radar / LiDAR windows. Post-hoc coating changes residual reflection and color calibration, introduces a haze and stray-light baseline, adds one more reflecting interface, and alters millimeter-wave and near-infrared transmission — and all of these drift with aging; Chapter 7 gives the executable flow: baseline first (transmittance, haze, contact angle) → small-area samples → bench regression → whole-vehicle perception regression → only then talk about production. The data-sheet fields the supplier must hand over are in Table 7 at the end.

1. First draw the boundary: is this window "glass", "a structural part", or "an optical part"

1.1 Three kinds of sensing windows, three sets of physical quantities

Talking about "sensor coatings" as one category is where most misjudgments begin. The three types take different inputs, and a coating acts in different directions for each:

• Camera (visible + near-infrared). The main pass band is 400～700 nm; modules are assembled with visible-transmit / near-infrared-cut filters (650 / 700 nm cutoffs are common), and night vision and in-cabin monitoring often use 850 / 940 nm active IR illumination. The coating is responsible for: transmittance not dropping, haze not rising, water beading and leaving the field of view quickly, and oil soil and insect residue being easy to clean off.
• LiDAR (near-infrared laser). 905 nm and 1 050 nm fall in the high-response band of silicon-based detectors (SPAD / APD); 1 550 nm is absorbed by the cornea and the anterior aqueous of the human eye, so a higher peak power is allowed at the same eye-safety class, giving a longer detection-range ceiling (the detector must switch to InGaAs, and cost moves up). Window coating here is essentially "dual-band anti-reflection + laser-damage resistance + fog and frost protection", and the design wavelength is not the camera's 550 nm (see the order-of-magnitude calculations in Chapter 4).
• Millimeter-wave radar (76-81 GHz). What matters is not refractive index but complex permittivity, loss tangent, wall thickness, and incidence angle; a coating is only a dielectric micro-perturbation, and the main risks come from conductive fillers, water uptake, and a continuous water film on the surface. This site already has a full treatment for these windows: Wave-transmitting nano anti-icing/snow coating for vehicle radomes: a cross-disciplinary engineering scheme from millimeter-wave transmission to zero-power anti-icing.

1.2 Division of labor with the site's three existing articles (explicit boundary)

To avoid repetition, the division of labor with three existing site articles is stated here: the radome anti-icing/snow article covers the anti-icing mechanism (supercooled droplets, ice-layer adhesion, SLIPS and electrothermal de-icing) and the millimeter-wave transmitting material system (low-dielectric resins, hollow fillers, wave-transmittance assessment); its object is "the radome body itself". This article covers the general criteria and failure modes of the window: why the requirement matrix conflicts, how to measure contact angle and roll-off angle per the national standards, on what basis transmittance and haze are accepted, why optics and millimeter wave fight at the order-of-magnitude level, how the automotive environmental tests combine, and where the vehicle-qualification and aftermarket boundaries lie. The two articles cross-link and neither replaces the other. In addition, the Superhydrophobic Nano Water-Repellent Coating for Automotive Glass and the Anti-Fog Nano Hydrophilic Coating for Automotive Rearview Mirrors articles are general material articles for the two functions respectively; this article does not repeat their routes or film chemistry, and discusses only the conflict question — "what to do when both ideas are wanted on the same sensor window" — with a quantified trade-off table. Related published articles: Superhydrophobic nano water-repellent coating for automotive glass: a technological leap from the lotus effect to driving safety, Anti-fog nano hydrophilic coating for automotive rearview mirrors: a complete technical system from surface physicochemistry to superhydrophilic anti-fog, Anti-fog and anti-condensation nano coatings: superhydrophilic mechanisms and the boundaries of durability.

1.3 Three fingerprints of failure scenarios

In the field you can triage the failure type without disassembly: appears only in rain and clears when the rain stops points to the wetting state being lost (degraded hydrophobicity; water sheeting into a curtain or beading that stays in the field of view); appears at dawn, in rain, or after the heater runs, and clears when wiped points to fogging (condensation on the inner surface or a dew point on the outer surface under a thermal gradient); deteriorates monotonically with mileage and does not recover when wiped points to optical or mechanical degradation of the film itself (rising haze, wear tracks, yellowing, embedded contamination). The countermeasures for the three are completely different; mixing them into "the coating is bad" is why rework cost runs out of control.

2. Why the requirement matrix fights itself: surface energy, roughness, and the three states of water

2.1 The physics: bead and film are two solutions of the same equation

Young's equation cosθ = (γ_sv − γ_sv′)/γ_lv sets the intrinsic contact angle; roughness amplifies the result — the Wenzel state gives cosθ* = r·cosθ (hydrophilic becomes more hydrophilic, hydrophobic more hydrophobic), while the Cassie-Baxter state locks air with micro/nano structures so the droplet sits on a composite face of "solid peaks + air"; this is the origin of contact angles above 150° and roll-off angles below 10°. What engineering uses to judge "will this surface be wetted and spread on" is the critical surface tension (Zisman method) and dyne values: spreading happens only when the liquid's surface tension is below the solid's critical surface tension. Water at 20 ℃ is about 72.8 mN/m; common oil soil (n-hexadecane) is about 27.5 mN/m; fluoro-silicone and polysiloxane low-surface-energy systems typically sit at 16～24 mN/m, so both water and oil bead up. The anti-fog criterion is the exact opposite: water must spread completely on the surface (contact angle approaching 0°), which requires pulling the surface energy far above water's surface tension and ensuring enough hydrophilic sites (hydroxyl groups, PEG segments, photo-generated hydroxyls on TiO₂ surfaces).

The contradiction is therefore hard at the materials level: low and high surface energy cannot coexist at the same point; what can coexist is only "sequence in time" or "zoning in space". There is also an optical constraint that is often overlooked: anti-fog requires the water film's thickness variation to be far smaller than the wavelength (ideally a molecular-level adsorbed film plus a uniform water film), while superhydrophobicity requires micro/nano roughness, and once the roughness approaches the visible-light wavelength order (0.4～0.7 μm), the coating surface itself becomes a scatterer and haze rises. This is the same side of one coin as the site article on mechanical durability and wear failure of superhydrophobic coatings — "the more transparent, the harder it is to keep enough roughness": Mechanical durability and wear failure of superhydrophobic coatings, Nanocoating wettability and contact angle: Young, Wenzel, Cassie-Baxter and surface energy, Coating surface tension and wetting-spreading theory.

2.2 The requirement matrix: five functions, four kinds of cost

How to read the table: treat every "×" at a row-column crossing as a design constraint, not a defect. The only items a window coating can genuinely co-optimize are hardness and weatherability (switching aromatic resins to aliphatic, replacing organic polymer backbones with siloxane / sol-gel inorganic networks); everything else is a trade-off. The site's published articles on the boundaries of low-surface-energy systems and fluorine-free routes: Fluorine-free water-repellent coatings: PFAS ban timelines, low-surface-energy design, and durability boundaries.

2.3 There are only three engineering routes, each with its own applicability

• Route A: zoning (coexistence in space). Pattern the wetting on one window: make the field-of-view center anti-fog superhydrophilic (so imaging is not prism-ified by droplets), and the areas outside the center — the edges and outer ring — hydrophobic (so rainwater and wash water drain quickly without leaving a curtain); or use micro-patterned structures to achieve "pinning anisotropy", letting droplets roll only along the wipe / gravity direction (front-low, back-high wedge micro-grooves). Applicability: the window is planar, the field-of-view geometry is well defined, coating or lithography / imprint steps are available, and two-step cost is acceptable. Limits: the zoning boundary is itself an optical discontinuity; the haze and residual-reflection step at the crossing must enter acceptance; when the pattern size is smaller than the droplet diameter, the pattern stops working.
• Route B: de-fog with energy and structure, not the coating. A transparent conductive heating layer (ITO, silver nanowires, aluminum-doped zinc oxide, conductive polymer) holds the surface at 40～70 ℃, lifting the inner surface above the dew point; or use air ducts / blowing, or a sealed cavity with desiccant and a vent membrane to control cavity humidity; frost on the outer surface goes to thermal management or an anti-icing system. Applicability: the perception system tolerates added power draw and control logic, the window geometry allows conductor routing, and the supplier accepts a "coating + functional film" hybrid structure. This is the mainstream half of today's mainstream mass-production practice — on Route B the coating only carries "easy cleaning, anti-soiling, reduced ice adhesion"; it does not carry the anti-fog criterion.
• Route C: sacrifice one item explicitly. Either go hydrophobic-only and accept the fogging risk in high humidity (leaving fog to the perception algorithm's self-check and degradation strategy), or go hydrophilic anti-fog-only and accept the water-curtain problem of raindrops on windows / glazing (with wipers or air-jet washing). Applicability: cost-sensitive, no heating resource, or the window is only a redundant / blind-spot-fill sensor in the system (for example a surround-view filler camera whose algorithm is already designed for "usable under partial occlusion"). The key is writing the trade-off into the specification and the after-sales script, not stacking two coatings and gambling.

Beyond the three routes there is no fourth layer of magic: painting a hydrophobic layer outside a hydrophilic one redefines the wetting state at the outer surface — the inner layer's function no longer acts on water; painting a hydrophilic layer outside a hydrophobic one lets water spread on the outer hydrophilic layer and creep along the interface, forming atomized "sandwich water" that is even harder to remove. The site's lens-level article has a corresponding structural discussion: Anti-reflective superhydrophobic nano coatings for premium camera lenses: from Fresnel reflection to high-performance optical coatings for outdoor photography.

3. Quantifiable criteria (I): how to measure surface properties so acceptance means something

3.1 Contact angle: the measurement protocol matters more than the number

The current method standard that can be cited directly in China is GB/T 30693-2014, Measurement of water contact angle on plastic films (issued 2014-12-31, implemented 2015-03-02, administered by the National Quality Supervision Key Product Inspection Method Standardization Technical Committee): it specifies the optical method for measuring the water-droplet contact angle on polymer film surfaces and applies to polymer films other than those that develop chemical affinity with water. The testing laboratories' published execution practice for this standard gives three key conditions: the probe liquid is deionized water; droplet volume 2～5 μL (an oversized droplet deforms under gravity and the reading skews); environment 23±2 ℃ / relative humidity 50±5%; the reading is taken 5～30 s after the drop has settled; at least 5 different positions per sample, averaged arithmetically; the conditioning temperature and humidity can be traced back to GB/T 9278-2008, Paints and varnishes — Standard atmosphere for conditioning and testing of test specimens. For glass substrates, GB/T 24368-2009, Detection of hydrophobic contaminants on glass surfaces — Contact-angle measurement method, can also be used. What must be recorded alongside is the advancing angle, receding angle, and contact-angle hysteresis: large hysteresis means the surface chemistry or topography is non-uniform (strong pinning); the static contact angle masks the real difference — which is exactly why "130° contact angle but the water will not roll off at all" happens.

3.2 Roll-off angle and the superhydrophobicity criterion

Roll-off angle (tilted-plate method): place the droplet on a horizontal stage and raise the tilt at a stable low rate (published working practice commonly takes 0.5°/s with a 5 μL or 10 μL droplet), recording the tilt at the instant the droplet starts to roll as a whole; the corresponding "slide angle" is the critical angle for lateral sliding. The industry's prevailing superhydrophobicity criterion is static contact angle greater than 150° and roll-off angle less than 10°, and both conditions must be met simultaneously — this criterion was written into the project drafting note of GB/T 45017-2024, Test methods for mechanical stability of superhydrophobic surfaces (public record on the National Standards Information Public Service Platform: failing to meet both conditions simultaneously is judged as loss of superhydrophobicity), and international standards such as ISO/TS 10689:2023, Nanotechnologies — Superhydrophobic surfaces and coatings: characterization and performance evaluation, adopt the same logic. A published explanation of the characterization methods is at Hydrophobic angle and roll-off angle of nano coatings: characterization methods and their self-cleaning significance.

3.3 Critical surface tension, dyne values, and the "should this be coated at all" judgment

Dyne pens / mixed-liquid tests (an industry rule of thumb; no specific standard number is cited here) give the threshold judgment of "can this be wetted": the wet film can only spread, and adhesion can only have a foundation, when the substrate surface energy is higher than the coating's surface tension. Clean metal surfaces are typically 40～45 mN/m; polypropylene-type low-energy surfaces 29～33 mN/m; a fluoro-silicone contamination layer can drop to 20～24 mN/m or lower — published industry practice treats a surface below 38 mN/m as a cratering-risk zone, and the same logic applies to window rework: for a piece of glazing contaminated with silicone oil or coating residue, whether a fresh anti-fog layer will succeed can already be predicted at the dyne-value step.

After Table 2 there is one caution this article wants to keep in view: pencil hardness, contact angle, and transmittance carry no comparability once separated from their test conditions. The site's published articles on the same issue: Abrasion and hardness testing of nano coatings: pencil hardness, Taber, and scratch, Controlling judgment error in ISO 15184 pencil-hardness testing for high-hardness abrasion-resistant baked metal paints.

4. Quantifiable criteria (II): transmittance, haze, anti-reflection, and scattering

4.1 Total transmittance and haze: definitions and test methods

The current national standard is GB/T 2410-2008, Determination of haze and transmittance of transparent plastics (issued 2008-08-04, implemented 2009-04-01, replacing GB/T 2410-1980, modified adoption of ASTM D1003:2007): Method A is the haze-meter method (integrating sphere), Method B the spectrophotometer method, applicable to transparent plastics in plate, sheet, and film form. At the definition level: total transmittance Tt = total luminous flux transmitted through the specimen / incident flux × 100% (direct component Tp + diffuse component Td); haze Haze = Td / Tt × 100%, where Td specifically means light scattered more than 2.5° away from the incident direction — the 2.5° threshold is set by ASTM D1003; ISO 13468 / ISO 14782 use a compensated geometry and JIS K 7105 is a separate system, so cross-laboratory comparisons must state which set is being used. The instrument pairs an integrating sphere with a light trap: total transmission is measured first, then the trap absorbs the direct beam in the original direction, and what the sphere walls collect is the large-angle scattered light.

The three quantities carry different engineering meanings and cannot substitute for one another: transmittance governs "is it bright" (absorption + reflection losses); haze governs "is it milky" (wide-angle forward scattering, which directly suppresses contrast); clarity / distinctness of image governs "are the edges sharp" (near-axis small-angle scattering under 2.5°, which does not enter the haze figure). A frosted surface may show modest haze yet blur the image — that is why "clarity ≠ 100% − haze". For camera windows, haze and clarity are the lethal items: a droplet or scratch array that deflects light beyond 2.5° smears the edge of a road sign, while total transmittance only drops 1～2 percentage points — the report shows no problem while the model has already degraded.

4.2 The quarter-wave condition of AR coatings and its three costs

The low-reflection condition of a single anti-reflection layer is an optical thickness n·d＝λ/4, and choosing the film index n₁＝√(n₀·n_s) can theoretically drive reflection to zero; in practice magnesium fluoride with n₁≈1.38 is common, the glass substrate n_s≈1.52, air n₀＝1, and the single-side reflectance floor is computed as R_min＝[(n₀n_s−n₁²)/(n₀n_s＋n₁²)]²:

• Uncoated glass reflects about 4.26% per surface; after passing through both surfaces of a plate, total transmittance is about 91.7% (absorption ignored).
• Both sides coated with λ/4 magnesium fluoride (550 nm design, film thickness 99.6 nm): single-side reflection falls to 1.26%, and the two-surface total transmittance is about 97.5% — that is the order of "one layer buys six percentage points of transmittance".
• Cost 1: residual reflection color. A single layer cancels reflection only at the design wavelength; at 450 nm and 650 nm reflection climbs back to 1.6% and 1.4%, which is why the enhanced-transmission face looks violet-blue or blue-green to the naked eye — this is single-layer physics, not a defect. Broadband multilayer stacks use high/low index pairs to suppress the whole visible band, but the cost moves to the two items below.
• Cost 2: angular color shift. At oblique incidence the phase thickness inside the film scales with cosθ and the reflectance minimum shifts toward shorter wavelengths. For a film with n＝1.38: at 45° incidence (about 31.1° refraction inside the film) the design point moves from 550 nm to about 500 nm; at 60° incidence it moves to about 428 nm. Edge pixels of wide-angle and fisheye lenses see large incidence angles, so reflectance and reflection color at the periphery differ from the center — showing up as peripheral color cast and uneven stray-light distribution. This is why "one window needing both a large field of view and anti-reflection" requires a wide-angle coating design.
• Cost 3: band mismatch. Take the 137.5 nm optical thickness optimized for 550 nm into the near-infrared: per thin-film matrix calculation, single-side reflection climbs back to 2.28% at 905 nm (two-surface transmittance about 95.5%) and 3.44% at 1 550 nm. In other words, the visible-band AR coating of a camera window is only "half good" for the LiDAR bands; the λ/4 thickness at 905 / 1 050 nm is 156～164 nm (for n＝1.45 / 1.38), and 1 550 nm needs higher-index materials (e.g. 193.8 nm when n≈2.0). Shared-window or shared-radome designs must deliver the reflectance minimum in multiple bands at once, or explicitly accept the rebound in one band.

Comparable calculations and process boundaries from photovoltaics and lens coating are already on the site: Anti-reflective nano coatings for solar panels: optical engineering from Fresnel reflection to PV efficiency gains, Functional coatings for photovoltaic modules: synergistic protection through anti-reflection, self-cleaning, and weatherability.

5. Quantifiable criteria (III): the optical vs. millimeter-wave order-of-magnitude gap decides whether a coating belongs on a radome

5.1 Bands and regulatory bases (including one common citation error)

China: 工信部无〔2021〕181 号, the Interim Provisions on Automotive Radar Radio Management (issued 2021-11-16, effective 2022-03-01) plans 76-79 GHz for automotive radar, and from the effective date no type approval is accepted for vehicle radars in 24.25-26.65 GHz; its Annex 1, "RF technical requirements for automotive radar", sets the limits: 76-77 GHz at 55 dBm peak, 50 dBm average, and 28 dBm/MHz e.i.r.p. spectral density; 77-79 GHz and 76-79 GHz at 39 dBm peak, 34 dBm average, and 7 dBm/MHz spectral density. The same document gives the engineering context: long-range radar (LRR) covers about 250 m with an 18° detection angle; short-range radar (SRR) about 30 m with 110°. The Rules and Regulations on Radio Spectrum Allocation of the People's Republic of China (MIIT Order No. 62), effective 2023-07-01, states in a footnote that the 79-81 GHz radiolocation allocation is intended for priority use by automotive radar and similar applications, though published industry research reports note that it has not entered actual deployment before compatibility and coexistence conditions are settled.

Europe and the world: WRC-15 identified 76-81 GHz as the globally harmonized automotive radar band. On the EU side, 76-77 GHz ground-based vehicle radar uses ETSI EN 301 091-1, and 77-81 GHz short-range traffic radar uses ETSI EN 302 264 (the harmonized standard covering Article 3.2 of the Radio Equipment Directive 2014/53/EU; the recent version published 2017-05, applicable to wideband devices with an occupied bandwidth of at least 50 MHz), with Commission Decision 2004/545/EC as its regulatory basis (the "79 GHz band" is 77-81 GHz, with a mean power-density cap of −3 dBm/MHz e.i.r.p. and a peak of 55 dBm e.i.r.p.).

The citation error that must be called out: ETSI EN 302 858 applies to "radar equipment operating in 24.05-24.25 GHz or 24.05-24.50 GHz" (V2.1.1, 2016-12) — part of the 24 GHz short-range family, and has nothing to do with 77 GHz. Writing EN 302 858 as the basis for 76-81 GHz is a frequent reject item in RF and coating reviews. When qualifying window coatings in China, the technical agreement should state "whether the coating affects the whole-machine type-approval RF metrics" instead of vaguely citing an ETSI number.

5.2 The order-of-magnitude case: can a coating go on a radome

The free-space wavelength at 77 GHz is λ₀＝c/f＝3.893 mm; inside a dielectric it shortens to λ₀/√εr. Put the optical and the millimeter-wave quarter-wavelengths in one table and the answer speaks for itself:

The conclusion is clear: liquid functional coatings at 76-81 GHz are merely "electrically ultra-thin dielectric films" — 30 μm is about λ/130, nowhere near any matching condition, and their impact on the radar link comes from three things rather than from thickness:

• Interface reflection is set by the dielectric constant. The normal-incidence single-interface power reflectance R＝[(√εr−1)/(√εr＋1)]²: Dk 2.1 (PTFE-class fluoropolymers) gives about 3.36%; Dk 3.0 (PC / ABS class) about 7.18%; Dk 4.0 (glass-fiber composite or heavily filled coating) about 11.11%; Dk 6.0 (glass) about 17.66%. If the coating loads large amounts of high-refractive / high-dielectric inorganic particles to chase hardness, the effective Dk shifts up and reflection plus multi-pass ripple follows.
• Absorption is set by the loss tangent, but the coating is too thin to eat much energy. Estimated with α≈8.686·π√εr·tanδ/λ₀, a 30 μm coating with Dk 4.0 and tanδ 0.02 costs only about 0.008 dB one-way; at tanδ 0.10 it is 0.042 dB; with water uptake pushing tanδ near 0.30, 0.126 dB. At the same thickness, a continuous water film — from water's Debye parameters (20 ℃: ε_s≈80.6, ε_∞≈4.9, relaxation time τ≈8.2 ps) the 77 GHz values are ε′≈9.4, ε″≈17.9, absorption about 32.6 dB/mm — costs 0.33 dB at 10 μm, 1.63 dB at 50 μm, and 3.26 dB at 100 μm; ice (Dk≈3.2, tanδ≈0.02) at 100 μm is only about 0.025 dB. That explains the true value of the hydrophobic layer on a radome: it is not about "transmitting waves" — it keeps water beading and rolling off so that the radome face is never blanketed by a dB-scale absorbing film.
• A continuous conductive network is the standout upper risk among the three inputs. Treating the coating as a resistive thin film in the transmission-line model, the one-way insertion loss changes sharply with sheet resistance: 100 Ω/sq about 9.20 dB, 300 Ω/sq about 4.23 dB, 1 000 Ω/sq about 1.50 dB, 3 000 Ω/sq about 0.53 dB, 10 000 Ω/sq about 0.16 dB. Fillers such as graphene, carbon nanotubes, conductive polymers, and silver nanowires are the protagonists of anti-corrosion and antistatic systems (site articles: Graphene anti-corrosion nano coating for steel, MXene (Ti₃C₂Tₓ) anti-corrosion coating: the lamellar barrier mechanism and engineering boundaries, Conductive and antistatic coatings: building three conductive networks with carbon nanotubes, graphene, and conductive mica, and precisely controlling 10⁴-10⁹Ω sheet resistance), but once they form a continuous path on a radar window they turn from "functional fillers" into "partial shielding". The executable criterion: coatings used on radomes / LiDAR windows should have outgoing inspection that includes sheet resistance, with the target range no lower than 10⁴ Ω/sq; if it cannot meet that, treat the part as a shielding item and re-run the whole-machine RF retest.

How loss folds back into perception capacity follows directly from the radar equation's fourth-root relation: 0.25 dB one-way → about −2.8% detection-range ceiling; 0.5 dB → −5.6%; 1.0 dB → −10.9%; 2.0 dB → −20.6%. Likewise the window's two-way transmittance must be squared: if total transmittance drops from 0.92 to 0.85 because of contamination or film degradation, LiDAR echo power loses −0.72 dB to −1.41 dB via T². The converse view of bandwidth gains: 3 GHz bandwidth gives 5 cm theoretical range resolution, 4 GHz 3.75 cm, 5 GHz 3 cm — the centimeter-level resolution bought at such effort can be eaten by a 50 μm water film (1.63 dB).

5.3 Division with the radome anti-icing article, and this article's increment

The site's Wave-transmitting nano anti-icing/snow coating for vehicle radomes: a cross-disciplinary engineering scheme from millimeter-wave transmission to zero-power anti-icing has already finished the materials main line — "low-dielectric resin system + hollow fillers + anti-icing strategies (superhydrophobic / SLIPS / electrothermal) + wave-transmittance assessment" — and readers choosing radome formulations should start there. Outside that line this article adds three blocks: first, it treats the window as a joint optics-and-electromagnetics budget problem (the same layer has completely different sensitivity at 550 nm and at 77 GHz, see Chapter 5); second, it gives executable wetting and optics test bases (Tables 2 and 3 in Chapters 3 and 4), turning "hydrophobic / anti-fog / abrasion-resistant" from adjectives into criteria; third, it chains the failure modes with automotive environmental testing (Chapter 6) and the vehicle-qualification and aftermarket boundaries (Chapter 7) into one executable verification flow. Together the two articles cover "material selection — criteria — verification — vehicle qualification" without repeating content: Anti-icing coatings: mechanisms and applications of the three strategies — superhydrophobic, SLIPS lubricating layer, and electrothermal.

6. Durability and failure modes: six decay paths reduced to reproducible test combinations

6.1 Post-abrasion hydrophobic failure: structure collapse, loss of low-surface-energy components, and "recovery" illusions

Hydrophobicity stands on two things holding at once: micro/nano roughness maintaining the Cassie state, and low-surface-energy components covering the outer surface. Wipers, sand-laden rain wash, brush car washes, finger wipes, and de-icing actions attack these two things separately. The directional figures from published literature and patents are worth citing but must be read with their conditions: under Taber CS-17 / 1 kg, transparent superhydrophobic coatings commonly show "CA still >90° at 200 revs, dropping from 156° to the 58° order by 500 revs"; a patent background instead writes "after 500 Taber abrasion cycles the CA falls from 140° to below 120°"; rotating rubber wheels (about 20 kPa-class loads, thousands of cycles) and #0000 steel-wool wipes under fixed load are another abrasion basis commonly used for anti-fog and hardened layers. The scrap logic should follow GB/T 45017-2024 and ASTM D8380-21 as "post-abrasion CA + roll-off angle re-measurement", not "how many revs before the substrate shows through". There is also a trap: migration of low-surface-energy small molecules creates a false "hydrophobic recovery" — the CA partially rebounds after resting several days post-abrasion, and this time dependence must be flagged with the read-out moment in the report. The site has dedicated treatments of Cassie→Wenzel collapse and "transparency versus sustainable roughness": Mechanical durability and wear failure of superhydrophobic coatings, Nano SiO₂ hydrophobic modification: from hydrophilic surfaces to superhydrophobic coatings, Scratch-resistant automotive clearcoats: the formulation logic from friction coefficient to critical load.

6.2 Anti-fog layer failure: organic contamination and the dark-state problem of photocatalysis

The main reason superhydrophilic anti-fog layers fail is usually not wearing off but being contaminated. The literature evidence is direct: TiO₂ films prepared by the sol-gel route on soda-lime glass show a fresh-surface water contact angle of 0°, but after storage in laboratory atmosphere the hydrophilicity gradually converts toward hydrophobicity; XPS and thermal-desorption analysis attributes this to adsorbed trace airborne organic contaminants, and the lost hydrophilicity can be regenerated by photocatalysis (Yin Haoyong et al., Science in China Series B, 2002 No. 5, 413-419; the original report of light-induced amphiphilic surfaces is Wang, Hashimoto, Fujishima et al., Nature 1997, 388:431). Later work adds two engineering details: post-contamination UV regeneration depends on the film's photocatalytic activity; a SiO₂ intermediate separation layer or more layers suppress Na⁺ migration from the glass substrate toward the surface and improve the recovery rate after contamination (Fang Min et al., Journal of Soochow University (Natural Science), 2012 No. 2). Translated into vehicle language:

• No UV, no regeneration. Cabin cameras, glazing shaded by the instrument panel, tunnels and night driving sit long-term in "contamination accumulating + no photocatalytic regeneration"; a TiO₂-type anti-fog layer without heat or airflow is certain to decay.
• The contamination list must be concrete. Fingerprint oils, dashboard volatiles (plasticizers), cleaner and water-repellent residues, surfactants in windshield washer fluid, and organic aerosols leaking past the cabin filter all cover the hydrophilic sites — therefore an anti-fog window must never share the same cleaning consumables with a water-repellent treatment.
• Hydrophilic gel layers are mechanically weak. PEG / polyacrylic-acid hydrophilic layers bind water through hydrogen bonds; they are soft and are carried away by wiping — exactly why the site's anti-fog articles keep emphasizing "durability boundaries": Anti-fog and anti-condensation nano coatings: superhydrophilic mechanisms and the boundaries of durability, Self-cleaning nano coating principles: the lotus effect and TiO₂ photocatalysis, two pathways.

6.3 Frost–defrost cycles, salt and washer fluid, UV yellowing

Frosting and defrosting is a process that repeatedly "tears the coating open": water first seeps into the surface microstructure and then freezes; freeze–thaw cycles permanently squeeze the Cassie air layer out, and the expanding ice layer pries at film defects. GB 11555-2025's icing basis can be borrowed directly for window-part testing: cold room at −18 ℃±3 ℃, horizontal wind speed below 2.2 m/s, water sprayed into a uniform ice layer over twice the glass area at 0.044 g/cm² with a spray gun (nozzle bore 1.7 mm, working pressure 350 kPa±20 kPa, liquid flow rate 0.395 L/min, 200～250 mm from the glass), then soak 30～40 min before starting the clock; the same standard also prescribes the pre-test cleaning procedure — first degrease thoroughly with a methanol-containing alcohol detergent, dry, then wipe with 3%～10% aqueous ammonia, and finish with a dry cotton cloth. These two facts are extremely valuable to coating suppliers: the chemical tolerance spectrum of a window coating must at least cover methanol (or ethanol / isopropanol) and aqueous ammonia as a pair, because both vehicle-level inspection and aftermarket cleaning use them.

Chemical loads have a dedicated standard: Part 5 of the GB/T 28046 series (against ISO 16750-5) sets the exposure-test logic (immersion and spraying) for materials and equipment against engine oil, transmission oil, brake fluid, battery fluid, cleaning agents, urea solution, and windshield washer fluid. That is the route for "will washer fluid eat the coating", not an invented "wipe 100 times". The aggressive components of washer fluid are alcohols (anti-freeze and oil dissolving) + surfactants (wetting and penetration) + corrosion inhibitors and dyes: on a hydrophobic layer, surfactant residues "disguise" the low-surface-energy surface as wettable; on a hydrophilic anti-fog layer, non-ionic surfactants adsorb and directly block the hydrophilic sites — this is the mechanism behind "the more you wash, the less it anti-fogs". On the de-icing-salt side, run neutral salt spray per GB/T 10125-2021 / ISO 9227 (50 g/L NaCl, 35 ℃, pH 6.5～7.2), though a combined salt + damp-heat + freeze–thaw cycle is closer to real window duty.

UV aging and yellowing deducts from the window at the short end: once the chromophore absorption edge red-shifts, 400～450 nm transmittance drops first — the eye sees yellowing, the instrument sees the yellowness index and ΔE rise, while the broadband average may fall only a little — so reporting only a single 550 nm transmittance will miss yellowing; both the 400 nm value and the full-band curve must be given. The formulation-side countermeasures share their roots with vehicle topcoats: aliphatic (HDI / IPDI) hardeners, UV absorbers and hindered-amine light stabilizers, and the UV-screening action of nano TiO₂ (30～40 nm-class particles build broad UV absorption inside the resin film); yet the same TiO₂ is both the anti-fog photocatalytic component and a UV absorber — the two functions compete for photons within one layer, an inherent trade-off in photocatalytic anti-fog film design. Related articles: Clearcoat gloss loss and yellowing: causes and prevention — UV aging and weatherable clearcoats, Weatherability of polyurethane topcoats: a complete analysis from chemical structure to QUV accelerated testing, Anti-aging UV-resistant nano paints for automotive plastic parts: from photo-oxidative polymer aging to nano UV shielding.

6.4 Vehicle-grade verification combinations: how GB/T 28046 and the IP codes pair up

GB/T 28046, Road vehicles — Environmental conditions and testing for electrical and electronic equipment, spans five parts: Part 1 general provisions; Part 2 electrical loads; Part 3 mechanical loads; Part 4 climatic loads; Part 5 chemical loads (the series structure appears in the foreword of GB/T 28046.2-2019; each part modified-adopts the corresponding ISO 16750 part, e.g. Part 4, GB/T 28046.4-2011, against ISO 16750-4:2006). Its real function is the severity framework graded by mounting location and the functional-state classification: Class A full function throughout; Class B transient deviation allowed during the test with automatic recovery afterwards; Class C function lost during the test but recoverable after; Class D requires repair or replacement (class definitions in Part 1). Temperature ranges, cycle counts, and per-item severity levels are agreed by the OEM in the DVP according to part location — the standard itself does not pick one set of numbers for every product, and this must be written into the technical agreement, otherwise "passed per GB/T 28046" is an invalid statement. Testing laboratories' published execution practice adds two cautions: climatic tests must use series-state units (retaining housing, seals, vent valves, and harness assembly), and storage-type tests are commonly agreed as Class C while operating-type tests as Class A — writing storage as Class A causes misjudgment.

The four load families directly relevant to coatings: climatic (high/low-temperature storage and operation, temperature step, temperature cycling, rapid change of temperature, ice-water shock, constant and cyclic damp heat, salt spray, mixed-gas corrosion, solar radiation, dust and water protection), mechanical (sine and random vibration, mechanical shock, free fall, gravel and chippings impact), chemical (automotive fluid and vapor exposure), and ingress protection. Dust and water protection should go through the vehicle-specific standard GB/T 30038-2013, Road vehicles — Degrees of protection provided by enclosures (IP Code) for electrical and electronic equipment (identically adopted ISO 20653, implemented 2014-07-01), where IPX9K uses 80 ℃±5 ℃ hot water at 80～100 bar (8～10 MPa), roughly 15～16 L/min flow, the nozzle about 0.175 m from the specimen, sprayed face by face at multiple angles — exactly the duty that destroys the edges and adhesive lines of externally mounted camera glazing, so the coating system must be reviewed together with the sealing design. Optical and mechanical acceptance of glass parts has its own product and method standards (GB 9656-2021, Safety specifications of automotive glazing; GB/T 5137.1-2020, Test methods for automotive safety glass — Part 1: Mechanical properties); rework and release cannot be judged only by the paint trade's cross-cut and pencil hardness. The site's acceptance-map articles: Automotive paint test standards (adhesion / weathering / stone chipping): an acceptance map from GB/T to ISO, Corrosion validation of automotive aluminum-alloy wheel coatings: salt spray, stone chipping, and high-pressure wash compared.

7. Vehicle-qualification and aftermarket boundaries: why nobody should casually coat a sensor window

7.1 Aftermarket DIY coating liquids and factory-cured coatings are not one durability class

The difference between the two product classes is not "whether SiO₂ is included" but the film-formation and bonding mechanism: aftermarket coating liquids are mostly dilute solutions of solvent-borne siloxanes / fluoro-silanes that cure by room-temperature condensation and physical adsorption — low solids, film thickness commonly below the micron level, no primer anchoring — and they decay after car-wash and wiper cycles; factory or pre-assembly coatings usually run the full chain of cleaning — primer (silane coupling or anchoring layer) — functional layer — curing (bake or UV / LED) — post-clean, with adhesion and hardness carried by chemical bonding and crosslink density. The site's measured accounts for ceramic coatings and paint care serve as analogies: 9H nano ceramic coating for automotive paint: the definitive technical analysis of paint protection, Hydrophobic mechanisms of automotive ceramic nano coatings: a full analysis from contact angle to durability, Automotive ceramic coating vs. paint protection film (PPF): the difference, Everyday automotive paint care: washing / waxing / sealing / ceramic coating. The point to stress: perception windows do not accept "appearance-trim" aftermarket treatment — a failed coating on a painted panel is merely unsightly; a failed coating on a window is a degraded sensor.

7.2 Five auditable risk classes (not "might have an effect")

• Optical transmission and color calibration. The residual reflection color of the film changes the ISP's white-balance and color-matrix calibration baseline; yellowing eats 400～450 nm first, drifting the blue-channel sensitivity. Criterion: color-chart ΔE and ΔTt at 400 nm, before and after coating, same light source and same target.
• Stray light and ghosting. Every additional air–film interface adds one more reflection path; when the AR layer is off-design (thickness deviating from λ/4) or locally missing, backlit scenes and oncoming high beams at night show ghosts and veiling halos. Criterion: stray-light distribution comparison under a dark-room point source.
• Image shift and calibration drift. Per this article's estimate, every 1 μm of film-thickness gradient across a 4 mm clear aperture corresponds to about 0.125 mrad of beam deviation; a wide-angle camera (FOV 100° / 1 920 px) sees about 0.14 pixel, a tele (FOV 30° / 8 MP) up to about 0.9 pixel; a uniform 30 μm slab introduces a focal-plane shift of about d(1−1/n) ≈ 10 μm. None of these values is large by itself, but they change with temperature, humidity, and aging, while extrinsic calibration is done only once — and that is the real reason against casual coating.
• LiDAR and radar echoes and multipath. Window transmission loss folds back into the signal as the two-way square (T from 0.92 to 0.85 → echo −0.72～−1.41 dB); on the millimeter-wave side, beyond the dB-scale losses from continuous water films and conductive fillers (see 5.2), the more subtle issue is the radome phase and antenna pattern: the angle-error correction table in the covered state was calibrated against the factory radome, and once a coating adds interfaces and dielectric micro-perturbation, the table no longer holds — the edge of the field of view can show a stable angular bias and phantom targets (related terms in published research reports: Grating Ghost and Mirror Ghost).
• Aftermarket irreversibility. Stripping a coating requires chemistry or abrasion, which simultaneously damages the hardened layer and the seal edges; once the "the more you clean, the blurrier it gets" loop forms, the only practical exit left is part replacement.

7.3 One executable judgment flow (five gates; missing any one means no production)

• Gate 1 — Baseline: on the uncoated state, measure total transmittance and haze (GB/T 2410-2008 or ASTM D1003 basis, light source noted), contact angle and roll-off angle (GB/T 30693-2014, droplet 2～5 μL, 23±2 ℃ / 50±5% RH, at least 5 points), and the band transmittance in the visible and at 850 / 905 / 940 / 1 550 nm; for millimeter-wave parts, additionally measure the covered-state antenna pattern and the angle-error baseline (bare reference + covered delta).
• Gate 2 — Samples: apply the coating on equivalent small coupons (same substrate, same cure conditions, same edge-seal form), re-measure everything from Gate 1, and report ΔTt, Δhaze, ΔCA, ΔRAA, and Δinsertion loss; at the same time run durability-target abrasion (ASTM D8380-21 or Taber basis) and chemical-media mini-trials (alcohols, ammonia, washer fluid).
• Gate 3 — Bench: test the full assembly (with housing, seals, venting components, and harness) per the GB/T 28046 series climatic and mechanical combination; IP protection per GB/T 30038-2013 (including IPX9K); glass abrasion per GB/T 5137.1-2020; after each stage return to the Gate 1 re-test items.
• Gate 4 — Whole-vehicle perception regression: regress across six scene families — rain, fog, night, backlight, tunnel, wet-cold salt, plus defrost / demist in working and non-working states; for millimeter-wave and LiDAR, compare covered-state angle error and detection-range ceiling; for cameras, run calibration residuals and recognition-distance curves (fixed 10% and 90% reflectance targets suggested). The criterion must be a "scenario pass set", not "visible once to the naked eye".
• Gate 5 — Production and aftermarket documents: bring the coating into change management (resin, filler, or cure-condition changes all trigger re-test), publish the rework / removal work instruction and the banned-consumables list (abrasives, silicone-containing polishes, high-surfactant cleaners), and define the aftermarket re-verification items.

7.4 The data-sheet fields a supplier should hand over

Graphene, MXene, and other 2D materials have clear value in the anti-corrosion and barrier directions (site articles: Graphene-modified nano-composite automotive topcoats: a technological leap from laboratory to industrialization, MXene (Ti₃C₂Tₓ) anti-corrosion coating: the lamellar barrier mechanism and engineering boundaries), but when putting them into a perception-window formulation they must be re-audited against the sheet-resistance criterion of 5.2: what pays off as conductive anti-corrosion on a painted panel may cost dB on a radar window.

8. Key figures at a glance (anchor list for this article)

• 72.8 mN/m and 16～24 mN/m: water's surface tension at 20 ℃ versus the surface-energy window of fluoro-silicone low-surface-energy systems; oil soil (n-hexadecane) is about 27.5 mN/m — oleophobicity is harder than hydrophobicity because the probe liquid's surface tension is lower.
• 2～5 μL / 23±2 ℃ / 50±5%RH: the published execution basis of the GB/T 30693-2014 sessile-drop method (an oversized droplet is pulled out of shape by gravity), reading 5～30 s after equilibration, at least 5 points averaged per sample.
• Contact angle above 150° and roll-off angle below 10°: the general dual criterion of superhydrophobicity; GB/T 45017-2024's project drafting note makes it explicit that failing to meet both conditions at once is judged as superhydrophobicity failure.
• 4.26%→1.26%, 91.7%→97.5%: the drop of glass's single-side Fresnel reflection after λ/4 magnesium fluoride (99.6 nm), and the two-surface total transmittance change of one plate — one layer buys about six percentage points.
• 2.28% and 3.44%: the single-side reflection rebound at 905 nm and 1 550 nm of the 137.5 nm optical-thickness layer designed for 550 nm; the λ/4 thicknesses for 905 / 1 550 nm are 164 nm (n＝1.38) and 193.8 nm (n＝2.0) respectively.
• 2.5°: the scattering-angle threshold specified by ASTM D1003 — haze only counts scattering more than 2.5° away from the incident direction; near-axis small-angle scattering enters "clarity", not haze.
• 3.893 mm / 0.562 mm / 1.124 mm: the 77 GHz free-space wavelength, the quarter-wave matching-layer thickness in a Dk 3.0 medium, and the half-wave wall thickness — millimeter-wave "film thickness" is millimeter-scale, three orders above optical films (99.6 nm on the 550 nm side).
• 32.6 dB/mm and 1.63 dB: the Debye-parameter estimate of water's absorption coefficient at 77 GHz and the one-way loss of a 50 μm continuous water film (10 μm about 0.33 dB, 100 μm about 3.26 dB); a low-loss coating of the same thickness is about 0.008 dB.
• 9.20 / 4.23 / 1.50 / 0.16 dB: the one-way insertion loss of resistive thin films at sheet resistances of 100, 300, 1 000, and 10 000 Ω/sq under normal incidence (transmission-line estimate); the conductivity gate for window coatings is suggested no lower than 10⁴ Ω/sq.
• −2.8% / −5.6% / −10.9% / −20.6%: the shrink of the detection-range ceiling for one-way losses of 0.25 / 0.5 / 1.0 / 2.0 dB, converted by the radar equation's fourth-root relation.
• 55 dBm / 50 dBm / 28 dBm/MHz and 39 / 34 dBm / 7 dBm/MHz: the peak, average, and e.i.r.p. spectral-density limits of 工信部无〔2021〕181 号 for 76-77 GHz and 77-79 GHz (and 76-79 GHz) automotive radar; Europe's 77-81 GHz runs on ETSI EN 302 264, and Commission Decision 2004/545/EC gives −3 dBm/MHz mean density with a 55 dBm peak.
• 5 / 3.75 / 3 cm: the theoretical range resolution at 3, 4, and 5 GHz bandwidth (c/2B) — also the conversion landing point for "what the dB that the coating saves actually bought".
• 0.044 g/cm², −18 ℃±3 ℃, 20 min / 80%: GB 11555-2025's icing mass and defrost criterion (Zone A 80% in 20 min, Zone A′ 80% in 25 min, Zone B 95% in 40 min); on the demist side −3 ℃±1 ℃, steam output 70 g/h±5 g/h × number of seats, Zone A 90% / Zone B 80% at 10 min.
• 80 ℃±5 ℃ / 8～10 MPa / 15～16 L/min: the IPX9K high-temperature high-pressure jet test conditions per GB/T 30038-2013 (IDT ISO 20653) — the real destruction duty on external-window edges and adhesive lines.
• 3.87 mm and 0.62 mm: the contact diameters of the same 1 μL water droplet at contact angles of 10° and 150° (an area difference of about 39×) — film-spreading and beading mean completely different things optically: the former scatters little but cannot stop fog; the latter drains away easily but acts as a micro-prism.
• 0°→hydrophobic→0°: the measured chain of the sol-gel TiO₂ film — fresh surface at 0°, converting to hydrophobic after storage in laboratory atmosphere through adsorption of trace organics, regenerable by photocatalysis (Science in China Series B, 2002 No. 5).

9. Frequently asked questions

Q1　Can one camera cover glass carry both superhydrophobicity and superhydrophilic anti-fog?

Not at the same location at the same time. Hydrophobicity requires low surface energy (fluoro-silicone systems at 16～24 mN/m) plus micro/nano roughness to hold the Cassie state and lift water into beads with a contact angle above 150° and a roll-off angle below 10°; anti-fog requires high surface energy plus a sufficient density of hydrophilic sites so that water spreads into a continuous film whose thickness variation is far smaller than the visible wavelength — the two directions are opposite solutions of Young's equation. Only three routes work: spatial zoning (anti-fog at the field-of-view center, draining hydrophobicity on the outer ring; or pinned-anisotropy micro-grooves that let water travel only along the wipe direction), handing defogging to energy and structure (a transparent conductive heating layer lifting above the dew point, air blowing, a dehumidified sealed cavity) while the coating only manages cleanability, and explicitly sacrificing one item with the trade-off written into the specification and the algorithm's degradation strategy. Stacking the two coatings directly is ineffective: the outer layer decides the wetting state, the inner layer's function stops acting on water, and the extra interfaces introduce haze and ghosts.

Q2　The customer only asks "what is the contact angle" — how do I pull acceptance back on track?

Replace the question with three sentences. First, the contact angle must be reported together with the roll-off angle and the advancing / receding hysteresis, and per the GB/T 30693-2014 basis it must state the probe liquid, droplet volume 2～5 μL, environment 23±2 ℃ / 50±5% RH, equilibration time, and number of points (at least 5, averaged) — otherwise cross-lot data are not comparable. Second, the superhydrophobic criterion is a contact angle above 150° and a roll-off angle below 10°, both holding at the same time (the failure logic of GB/T 45017-2024); reporting only the static angle misreads a heavily pinned 130° surface as a self-cleaning one. Third, for a window part the functional endpoint should be "post-abrasion Δhaze and ΔCA", not the ex-factory CA. Write these three into the technical agreement and the discussion returns from adjectives to data.

Q3　Does a hydrophobic coating actually do anything good for a millimeter-wave radar?

Yes — but the benefit lies in water management, not in "transmitting waves". A 30 μm coating at 77 GHz (λ₀＝3.893 mm) is an electrically ultra-thin dielectric film of about λ/130 and cannot perform impedance matching; the loss formula puts a low-loss coating in the 0.01 dB one-way order, whereas the same 30～50 μm as a continuous water film costs roughly 0.33～1.63 dB per water's Debye parameters, and 100 μm reaches 3.26 dB — the radar equation's fourth root then cuts 10%～20% off the detection-range ceiling. The hydrophobic layer's payoff is making water bead and roll off quickly so the radome face is never covered by a water sheet, while lowering the adhesion of soil and salt frost. The cost is equally clear: if a continuous conductive network is introduced to raise hardness or to dissipate static, dropping the sheet resistance to 10²～10³ Ω/sq brings 9.20～1.50 dB of insertion loss — worse than the water film. So the criterion is not "add hydrophobicity or not" but "after adding it, take measured Dk, tanδ, sheet resistance, and the covered antenna pattern".

Q4　Why can't windshield water repellent be used on camera or LiDAR windows?

All three reasons are auditable. Optically, repellent buys beading, and a bead inside the field of view is equivalent to a micro-prism and micro-lens; wide-angle and telephoto lenses show scatter blooms against backlight, while a water sheet on the windshield outside, though unfavorable, distributes its imaging impact more evenly and the wiper can clear it. On durability, repellents are mostly solvent-borne siloxanes that film-form at room temperature through physical adsorption and decay after wash and wiper cycles; as they decay the wetting inverts into zonal contamination of "half shedding, half hanging water", which is harder to live with than uniformly uncoated glass. On the aftermarket, removal requires chemistry or abrasion, which simultaneously damages the hardened layer and the edge seals. The site's two general articles address the glass and mirror scenarios (Superhydrophobic nano water-repellent coating for automotive glass); porting their applicability straight onto a perception window is out-of-scope use.

Q5　Why does the anti-fog coating get "worse the more it is used" — is it a formulation-quality problem?

In most cases it is contamination rather than failure. Fresh TiO₂-type photo-induced superhydrophilic surfaces can reach a 0° water contact angle, but after storage in laboratory atmosphere they adsorb trace airborne organics and progressively turn hydrophobic; UV illumination is needed to photocatalytically decompose the contaminants and restore hydrophilicity (Science in China Series B, 2002 No. 5; light-induced amphiphilicity: Nature 1997, 388:431). The real duty of cabin-mounted and interior cameras is exactly "contamination keeps arriving, UV never does": dashboard volatiles, fingerprint oils, organic aerosols carried back by the HVAC, and non-ionic surfactant residues all cover the hydrophilic sites; night, tunnels, and shaded positions leave almost no regeneration window. The engineering countermeasures in priority order: move the defogging responsibility to heating and airflow (the coating only does easy-clean), choose formulations that remain recoverable after contamination (published routes such as SiO₂ separation layers suppressing Na⁺ migration), and require the "post-contamination recovery rate" as a specification item instead of reporting only the initial fog-free duration.

Q6　How large can the coating's influence on camera calibration be? Give an order of magnitude.

Estimate sensitivity from the film-thickness gradient: every 1 μm of thickness difference across a 4 mm clear aperture approximates a wedge angle, producing about 0.125 mrad of beam deviation at refractive index 1.5; that works out to roughly 0.14 pixel on a wide-angle camera with FOV 100° and 1 920 pixels, and about 0.9 pixel on a telephoto camera with FOV 30° and 8 MP; a uniform 30 μm slab film shifts the focal plane by about 10 μm. None of these numbers is fatal by itself; the real problem is that they are not constants: water uptake, temperature, UV aging, and contamination make thickness, refractive index, and residual reflection vary over time, while intrinsic-focus fine-tuning and extrinsic calibration are completed once on the assembly line. That is why industry practice puts window optical parts on the "calibration-invariant list": any coating change triggers recalibration and residual re-measurement in the covered (with-window) state, rather than just filing one more coating report.

Q7　The supplier handed me a report saying "weathering 1 000 h passed" — is that enough?

No — three pieces of information are missing. The first is the aging type and spectrum: xenon arc and fluorescent UV follow very different acceleration rules for short-wave yellowing, so the report must state the lamp, irradiance, black-panel temperature, and cycle mode, and give ΔTt at 400 nm with ΔYI / ΔE; giving only "60° gloss retention" is appearance-part language. The second is what was actually aged: the vehicle-grade climatic and chemical loads of the GB/T 28046 series require series-state assemblies (housing, seals, vent components, and harness included); results from bare boards or disassembled samples cannot represent the installed condition. The third is the classification basis: the functional state must be agreed as A / B / C / D per Part 1 — storage-type items are commonly recorded as Class C and operating-type items as Class A; writing storage as Class A causes misjudgment, and writing a Class D as "minor cosmetic change" can let sub-film failures pass.

Q8　We make the decorative covers and trim parts outside the radome — how should the coating be specified?

First separate "radome body" from "trim". A radome body carrying RF functionality must be designed by electrical thickness (half-wave wall thickness, measured Dk, wall tolerance, and assembly spacing); the criteria and material system for that layer are already laid out in the site's radome article (Wave-transmitting nano anti-icing/snow coating for vehicle radomes). A trim coating handles only appearance and weatherability — but you must confirm that when it sits outside the radome it does not enter the main beam, or, if it does, it is re-tested electromagnetically at whole-machine level. The requirements to the coating shop should include: target-band measured Dk and tanδ (cavity-resonator or transmission-line method; no 1 kHz / 1 MHz nominal values accepted), sheet resistance, water absorption and post-absorption re-measurement, covered-pattern and angle-error deltas (give separate ceilings for gain loss, beam-pointing shift, and near-sidelobe rise), and a description of the phase-calibration procedure in the covered state. Write all of these and "can we coat it" stops being folklore.

Q9　After going fluorine-free, can hydrophobicity still hold up to a vehicle window's life?

Budget for it. The long-term stability of low surface energy comes from C—F bonds; fluorine-free routes (siloxanes, long-chain alkyls, waxes, nano-structural types) can reach 110°～140° initial contact angle, but their resistance to hydrolysis, UV, and oil is usually weaker than fluorinated systems, and they lean more on roughness to hold the Cassie state — they fall back faster after abrasion. The workable combination: shift the functional center of gravity from "superhydrophobic self-cleaning" to "easy clean + low adhesion" (reduce soil and ice adhesion instead of chasing 160°), raise the inorganic fraction with sol-gel or silsesquioxane networks to stabilize the roughness, and write both the fluorine-free surface energy and the PFAS compliance timetable into the technical agreement with corresponding acceptance items (details: Fluorine-free water-repellent coatings: PFAS ban timelines, low-surface-energy design, and durability boundaries). Make the trade-off on paper, not in the aftermarket.

10. References and basis notes

• GB/T 30693-2014, Measurement of water contact angle on plastic films (issued 2014-12-31, implemented 2015-03-02, administered by the National Quality Supervision Key Product Inspection Method Standardization Technical Committee; specifies the optical method for the water-droplet contact angle on polymer film surfaces); GB/T 24368-2009, Detection of hydrophobic contaminants on glass surfaces — Contact-angle measurement method; GB/T 26490-2011, Test method for the super-omniphobic performance of nanomaterials; GB/T 45017-2024, Test methods for mechanical stability of superhydrophobic surfaces (its public project drafting note states the superhydrophobicity failure criterion: failing to meet both static contact angle above 150° and roll-off angle below 10°); GB/T 42270-2022, Test method for the hydrophobic performance of porous hydrophobic membranes; ISO/TS 10689:2023, Nanotechnologies — Superhydrophobic surfaces and coatings: characterization and performance evaluation.
• GB/T 2410-2008, Determination of haze and transmittance of transparent plastics (issued 2008-08-04, implemented 2009-04-01, replacing GB/T 2410-1980, modified adoption of ASTM D1003:2007; Method A haze meter, Method B spectrophotometer); ASTM D1003 (haze and transmittance, scattering-angle threshold 2.5°); ISO 13468 / ISO 14782 (compensated-method basis); JIS K 7105; JJF 1303, Calibration specification for haze meters (instrument traceability).
• GB/T 9278-2008, Paints and varnishes — Standard atmosphere for conditioning and testing of test specimens, and GB/T 2918-2018, Plastics — Standard atmospheres for conditioning and testing of specimens (issued 2018-12-28, implemented 2019-11-01, modified adoption of ISO 291:2008); the ambient basis for contact-angle and haze testing (23±2 ℃ / 50±5% RH) comes from these two standard families and the testing laboratories' published execution documents.
• Abrasion methods: GB/T 1768-2006, Paints and varnishes — Determination of abrasion resistance (rotating rubber-wheel method); ASTM D4060 (Taber abraser, CS-10 / CS-17 wheels); ASTM D1044 (abrasion resistance of clear plastics, expressed as transmittance loss); ASTM D8380-21, Standard test method for dry abrasion resistance of water-repellent and omniphobic coatings; GB/T 6739-2022, Paints and varnishes — Determination of hardness with pencil method (its scope explicitly states the method is not suitable for comparing the pencil hardness of different coatings); GB/T 9286-2021 (cross-cut test); GB/T 5137.1-2020, Test methods for automotive safety glass — Part 1: Mechanical properties (issued 2020-03-31, implemented 2021-02-01, modified adoption of the corresponding ISO specifications, covering impact, penetration, abrasion resistance, fragmentation state and other methods).
• Vehicle-grade environmental conditions: GB/T 28046, Road vehicles — Environmental conditions and testing for electrical and electronic equipment, in 5 parts (Part 1 general provisions; Part 2 electrical loads — GB/T 28046.2-2019, issued 2019-10-18, implemented 2020-05-01, replacing the 2011 edition, modified adoption of ISO 16750-2:2012; Part 3 mechanical loads; Part 4 climatic loads — GB/T 28046.4-2011 against ISO 16750-4:2006; Part 5 chemical loads); GB/T 30038-2013, Road vehicles — Degrees of protection (IP Code) for electrical and electronic equipment (identically adopted ISO 20653, implemented 2014-07-01; IPX9K conditions: 80 ℃±5 ℃, 80～100 bar, about 15～16 L/min, nozzle about 0.175 m from the specimen, multi-angle face-by-face spray — as reprinted from testing laboratories' public technical articles).
• Vision-recovery and glazing regulations: GB 11555-2025, Technical specification of defogging and demisting systems for motor vehicle windscreen (issued 2025-04-15, enforced for newly approved vehicle types from 2026-01-01, replacing GB 11555-2009; defrost: −18 ℃±3 ℃, icing 0.044 g/cm² × area, spray gun bore 1.7 mm / pressure 350 kPa±20 kPa / flow 0.395 L/min, Zone A 80% in 20 min, Zone A′ 80% in 25 min, Zone B 95% in 40 min; demist: −3 ℃±1 ℃, steam output 70 g/h±5 g/h × number of seats, Zone A 90% in 10 min, Zone B 80%; the pre-test cleaning uses a methanol-containing alcohol detergent and 3%～10% aqueous ammonia); GB 7258-2017, Safety specifications for power-driven vehicles operating on roads (visible-light transmittance of the front windscreen and the driver's vision zone ≥70%); GB 9656-2021, Safety specifications of automotive glazing; GB 11562-2014, Motor vehicle driver's forward vision — Requirements and measurement method.
• Millimeter-wave bands and RF regulation: 工信部无〔2021〕181 号, Interim Provisions on Automotive Radar Radio Management (issued 2021-11-16, effective 2022-03-01; plans 76-79 GHz for automotive radar; Annex 1 gives the peak / average power and e.i.r.p. spectral-density limits, LRR about 250 m / 18°, SRR about 30 m / 110°); the Rules and Regulations on Radio Spectrum Allocation of the People's Republic of China (MIIT Order No. 62, effective 2023-07-01; a footnote states the 79-81 GHz radiolocation service is prioritized for automotive radar and other applications); WRC-15 identified 76-81 GHz as the globally harmonized automotive radar band; ETSI EN 302 264 (77-81 GHz short-range traffic radar harmonized standard, recent version 2017-05; with Commission Decision 2004/545/EC: −3 dBm/MHz mean power density, 55 dBm peak e.i.r.p.); ETSI EN 301 091-1 (76-77 GHz ground-based vehicle radar); ETSI EN 302 858 V2.1.1 (2016-12) applies to 24.05-24.25 GHz or 24.05-24.50 GHz and does not cover 76-81 GHz; the Automotive Radar 76-81 GHz Spectrum Requirements Research Report (jointly published by the Transporter Information Industry Application Alliance and the State Radio Monitoring Center Testing Center, 2026-03) as background for the bandwidth-to-range-resolution conversion.
• Wetting and optics literature: the Young, Wenzel (cosθ*＝r·cosθ) and Cassie-Baxter models and the Zisman critical-surface-tension method; water surface tension at 20 ℃ of 72.8 mN/m, n-hexadecane about 27.5 mN/m; hydrophilicity loss and photocatalytic regeneration of sol-gel TiO₂ films — Yin Haoyong et al., Science in China Series B, 2002 No. 5, 413-419; light-induced amphiphilic TiO₂ surfaces — Wang, Hashimoto, Fujishima et al., Nature 1997, 388:431-432; post-contamination UV recovery and Na⁺-migration suppression in TiO₂ / SiO₂ composite films — Fang Min et al., Journal of Soochow University (Natural Science), 2012 No. 2.
• Published engineering bases (non-standard): typical radome Dk and tanδ ranges (PTFE about 2.05 / 0.0005, ABS about 2.7～3.2 / 0.006～0.01, PC about 2.9～3.2 / 0.006～0.01, glass-fiber composites about 3.5～4.5 / 0.01～0.02, foam sandwich 1.1～1.5 / below 0.002), the half-wave wall formula t＝nλ₀/(2√εr), and covered-state acceptance items (insertion loss, beam-pointing shift, 3 dB beamwidth, near sidelobes, angle error, environmental stability) — reprinted from chip vendors' publicly summarized radome design guides; post-abrasion contact-angle fall-back of superhydrophobic coatings (Taber CS-17 / 1 kg: still >90° at 200 revs, from 156° down to the 58° order at 500; a patent-background figure of CA dropping from 140° to below 120° after 500 cycles) — reprinted from public literature and patents; the sharp wave-transmittance drop of waterborne transmitting coatings under low-temperature icing and the "low-Dk resin + hollow ceramic microsphere" route (77 GHz targets Dk 2.10～2.25, tanδ≤0.005) — reprinted from public patent backgrounds and examples.
• This article's own calculations (re-checkable): λ₀＝c/f; the in-medium wavelength λ₀/√εr and the λ/4, λ/2 thicknesses; Fresnel power reflectance R＝[(n_s−n₀)/(n_s＋n₀)]², λ/4 film R_min＝[(n₀n_s−n₁²)/(n₀n_s＋n₁²)]², and single-layer matrix reflectance from 450～1 550 nm; absorption coefficient α≈8.686·π√εr·tanδ/λ₀; Debye-relaxation estimate for water (20 ℃: ε_s≈80.6, ε_∞≈4.9, τ≈8.2 ps → at 77 GHz ε′≈9.4, ε″≈17.9, α≈32.6 dB/mm); resistive-film normal-incidence transmission T＝[1＋(Z₀/2R_s)]⁻² (Z₀≈377 Ω); radar-equation R∝P^(1/4) loss-to-range conversion; c/2B range resolution; wedge deviation δ≈(n−1)α and slab focal shift d(1−1/n); spherical-cap droplet contact-diameter geometry. All of the above are substitutions into formulas and represent no standard's limits.

Basis note: the standard numbers, names, and issue / implementation dates in this article are checked against the public records of the National Standards Information Public Service Platform, the Standard Information Service Network, and the standards' own forewords; the temperature ranges, cycle counts, and severity levels of vehicle-grade tests are agreed by the OEM's DVP according to mounting location — this article gives only the framework and item names, and does not specify numbers on any product's behalf. Values marked "public basis / reprinted" (radome Dk and tanδ, post-abrasion contact-angle fall-back, DIY coating life ranges, etc.) float with grade, batch, and duty; they serve selection communication and criterion design and are not an arbitration basis. Test conditions such as IPX9K are reprinted from testing laboratories' public technical articles; formal testing follows the standard text itself. This article uses no unverified AEC device-grade basis and cites no standard number that could not be confirmed (for instance, the search found GB/T 30691 — often mistaken for a contact-angle method — is actually "Conveyor belts — Test environments and conditioning times", so it is not cited here). The supplier data-sheet fields and the vehicle-qualification judgment flow are this article's recommended engineering practice, not mandatory regulatory requirements.

Last updated: 2026-11-03 | By: Kexin New Materials (Guangdong) Co., Ltd. (B2B technical literature; trade-terms basis EXW / FOB)