The story in four numbers
The firm reads this result as almost certainly real and almost certainly narrowband, and those two characteristics are not in tension because they are linked by a physical constraint rather than by an engineering shortcoming. A causality bound published by Konstantin Rozanov in 2000 fixes the maximum product of absorption depth and bandwidth available to any passive absorber of a given thickness, and it follows from the same Kramers-Kronig relations that govern every causal material response, which means it cannot be engineered around by any choice of geometry, metamaterial patterning, or resonance structure. Running an illustrative 70-micrometre non-magnetic film through that bound yields roughly four tenths of one percent of fractional bandwidth at ninety decibels of suppression at X-band, which describes a laboratory demonstration rather than a stealth coating. The identical film operated at a more modest twenty decibels, which is still 99 percent absorption, buys illustratively 15 percent fractional bandwidth at automotive radar frequencies, enough to span the entire 76 to 81 GHz band on which every advanced driver assistance system deployed today depends. Our view is that the defence headline is the least commercially interesting reading of this result, and that the millimetre-wave commercial market is where a hair-thin conformal absorber has a defensible economic case.
The claim and the units it is stated in
Absorber performance reported in percent is an unhelpful convention that obscures more than it conveys, and the reporting around this film illustrates the problem precisely: the headline figure and the accompanying description differ by a factor of one hundred thousand in reflected power, while appearing to a general reader as though they describe roughly the same thing.
The headline characterises the film as stopping 99.99 percent of radar waves. The supporting description appends five further nines to it. Expressed in percent the two look like adjacent claims separated by a handful of decimal places. Expressed in the unit that electromagnetics actually uses, they are forty decibels and ninety decibels of suppression respectively, a difference of fifty decibels, which corresponds to a factor of one hundred thousand in reflected power and a factor of roughly three hundred and sixteen in reflected field amplitude. The gap between those two claims is larger than the entire performance range separating a mature fielded absorber from an untreated metal panel.
This is not a criticism of the reporting so much as an observation about why the firm treats percent-denominated absorber claims as requiring conversion before they can be assessed. Decibels are the correct unit because absorber performance is logarithmic in nature and because the physical bounds that govern absorbers are stated as integrals of the logarithm of reflectivity. A claim of 99 percent absorption is twenty decibels and is unremarkable, achievable with commodity carbonyl iron loaded paint. A claim of ninety decibels is extraordinary and, as the arithmetic below establishes, is purchasable only at a very specific and very narrow slice of spectrum.
The second load-bearing figure is the thickness. Human hair spans illustratively 17 to 180 micrometres, with a commonly cited central value near 70 micrometres, so the description places the film somewhere in the tens of micrometres. For the analysis that follows the firm adopts 70 micrometres as a working figure, noting that the conclusions strengthen rather than weaken if the true thickness is lower, because the governing bound scales linearly with thickness.
What ninety decibels of suppression actually means
Ninety decibels is not an incremental improvement on the performance of fielded radar absorbing materials. It is roughly three orders of magnitude beyond what operational coatings deliver, and understanding why that gap exists is the entry point to understanding what the result is and is not.
A radar absorber is assessed by its reflectivity, conventionally expressed as the ratio of reflected to incident power in decibels and conventionally negative. Fielded absorbing materials on operational aircraft deliver illustratively ten to twenty decibels of reflectivity reduction across their design band, which is to say they return between one percent and ten percent of the energy that strikes them. Laboratory anechoic chamber materials, which face none of the aerodynamic, thermal, or erosion constraints of a flight surface and may be several centimetres thick, reach illustratively thirty to fifty decibels. A claim of ninety decibels sits far outside both of those ranges.
The reason such a claim is nonetheless credible is that deep absorption at a single frequency is not difficult to engineer. Any resonant structure can be tuned so that the reflected field from one path destructively interferes with the reflected field from another with near-perfect cancellation at the resonance. At exact resonance, with the loss tangent tuned to the impedance-matching condition, the cancellation can be made arbitrarily deep in principle and is limited in practice by manufacturing tolerance and measurement noise floor rather than by physics. Ninety decibels is a plausible measured figure at a carefully chosen frequency, normal incidence, and a single polarisation.
The difference between forty decibels of suppression and ninety reads, in percentage terms, as a rounding refinement. It is a factor of one hundred thousand in reflected power. Every decibel of suppression past roughly twenty is purchased with bandwidth, and the exchange rate is fixed by causality rather than by engineering skill.
The question that determines whether a deep absorption figure has operational meaning is therefore never how deep, but how wide. An absorber delivering ninety decibels across forty megahertz at 10 GHz is a laboratory curiosity for defence purposes, because a fire-control radar can hop outside that window trivially and because the emitter frequency is not a parameter the target controls. An absorber delivering twenty decibels across four gigahertz is an engineering product. The published material describes the depth and, on the available reporting, does not disclose the width, which is the single disclosure that would allow the result to be priced.
The bound that prices thinness against bandwidth
The exchange rate between absorption depth and bandwidth is not a rule of thumb or an empirical observation about existing materials. It is a theorem, derived from causality and passivity alone, and it applies to every passive absorber regardless of its internal construction.
In 2000, Konstantin Rozanov published a result establishing the ultimate thickness-to-bandwidth ratio for radar absorbers backed by a conducting plane. The Rozanov bound states that the integral of the absolute value of the natural logarithm of reflectivity, taken over wavelength across the entire spectrum, cannot exceed a constant multiplied by the static permeability and the physical thickness of the absorber. For a non-magnetic absorber, where the static relative permeability equals one, the available budget is simply two pi squared multiplied by the thickness. The derivation rests on the Kramers-Kronig relations, which are a direct mathematical consequence of the requirement that a material cannot respond before it is excited, combined with the requirement that a passive material cannot generate energy. Neither assumption can be relaxed by clever structuring.
The practical consequence is that thickness buys a fixed budget of absorption-bandwidth product, and the designer may spend that budget on deep absorption across a narrow band or shallow absorption across a wide one, but not both. A 70-micrometre non-magnetic film has a budget of illustratively 1.38 millimetres when expressed in the units of the bound. Ninety decibels of suppression corresponds to a logarithmic reflectivity magnitude of roughly 10.4, so the widest wavelength interval over which that depth can be sustained is illustratively 1.38 divided by 10.4, or roughly 0.133 millimetres. At 10 GHz, where the free-space wavelength is close to 30 millimetres, that interval represents a fractional bandwidth of illustratively four tenths of one percent, or roughly forty megahertz.
| Suppression | Max interval | Frac. BW at 10 GHz | Frac. BW at 77 GHz | Read |
|---|---|---|---|---|
| 10 dB | ~1.20 mm | ~4.0% | ~31% | Broadly useful |
| 20 dB | ~0.60 mm | ~2.0% | ~15% | Covers 76-81 GHz |
| 30 dB | ~0.40 mm | ~1.3% | ~10% | Chamber grade |
| 40 dB | ~0.30 mm | ~1.0% | ~7.7% | Single-band only |
| 90 dB (the claim) | ~0.13 mm | ~0.4% | ~3.4% | Point-frequency |
The one legitimate route around the non-magnetic constraint is magnetic loading, because the bound scales with static permeability and a magnetically loaded absorber therefore commands a proportionally larger budget. This is precisely why ferrite tiles and carbonyl iron loaded coatings have dominated practical absorber engineering for decades despite their mass penalty. The limitation is that Snoek's law imposes its own trade between static permeability and the frequency at which magnetic response persists, so magnetic loading that delivers a large budget at one gigahertz delivers very little at ten and effectively nothing at seventy-seven. At the frequencies where a hair-thin film is most interesting, magnetic loading is not available to relax the bound.
Absorption is not radar cross section
Even granting a broadband absorber of exceptional performance, the firm's framework treats surface absorption as addressing only one contributor to an aircraft's radar signature, and generally not the dominant one. The relationship between coating performance and platform observability is far weaker than the framing around stealth materials usually implies.
An aircraft's radar cross section is set first and overwhelmingly by its geometry. The faceting of the F-117, the planform alignment of the B-2 and B-21, the serpentine inlet ducts and sawtooth panel edges of the F-22 and F-35 all exist to redirect specular reflection away from the emitter rather than to absorb it. Redirection is enormously more efficient than absorption because it costs nothing in mass, imposes no bandwidth penalty, and degrades gracefully. Absorbing material is applied to clean up the residue that shaping cannot eliminate, which is concentrated at edges, at panel gaps, at access-door seams, around apertures, and at the discontinuities where travelling surface waves shed energy.
This distinction matters commercially because it determines where an absorber has to perform. The specular return from a large flat treated surface, which is the geometry an absorber test coupon actually measures, is precisely the contributor that shaping has already suppressed by thirty or forty decibels. Improving it further produces little change in the total. The contributors that dominate the residual signature are edge diffraction and structural resonances, which are governed by the geometry and electrical size of the feature rather than by the impedance of the surface, and which a thin absorbing layer influences only weakly.
Specular return from treated surfaces, travelling and creeping surface waves shed at discontinuities, cavity resonance inside inlets and apertures where multiple bounces let even modest per-bounce absorption compound, and the multipath and ghost-target artefacts that degrade a commercial radar sensor's own picture. These are real contributors and a good absorber measurably improves them.
Edge diffraction from wings and control surfaces, which is set by geometry and electrical size. Resonance-region scattering when the emitter wavelength approaches airframe feature dimensions, the physics that makes VHF and UHF radar a persistent counter-stealth avenue. Returns from unavoidable apertures. And the infrared, acoustic, and emission-control signatures that are not electromagnetic scattering problems at all.
An aircraft's radar cross section is written by its geometry and merely edited by its coating. A material that improves specular absorption by fifty decibels improves the contributor that airframe shaping already reduced to insignificance, while leaving the edge diffraction and resonance-region scattering that actually set the residual signature substantially where it was.
Where the economics actually work
The firm's assessment is that a hair-thin conformal absorber with tunable resonance has a considerably stronger commercial case outside defence than within it, and that the constraint driving this conclusion is qualification burden and unit volume rather than technical performance.
The defence path is slow and narrow. A material applied to a flight surface must survive illustratively twenty years of thermal cycling, rain and particulate erosion at transonic speed, ultraviolet exposure, hydraulic fluid and fuel contact, and repeated field repair, while retaining its electromagnetic properties throughout. Qualification for a manned combat platform is conventionally measured in years and the addressable fleet is small in unit terms. Where the firm does see genuine defence value is in sustainment rather than in signature improvement, because low-observable restoration has historically been a substantial driver of stealth platform maintenance burden and of the mission-capable rates that have attracted sustained congressional attention. A film that could be applied and replaced as a conformal appliqué without a climate-controlled facility and a multi-day cure would have an economic argument based on availability and cost per flight hour, which is a more tractable and more defensible case than one based on incremental signature reduction.
The commercial case is the stronger one and the arithmetic in section two is the reason. At the millimetre-wave frequencies where the automotive, telecommunications, and test-instrumentation markets operate, the same physical thickness purchases roughly eight times the fractional bandwidth it does at X-band, because fractional bandwidth for a fixed wavelength interval scales inversely with the operating wavelength. A 70-micrometre film delivering twenty decibels of absorption covers illustratively fifteen percent fractional bandwidth at 77 GHz, which spans the entire 76 to 81 GHz automotive radar allocation with margin. Twenty decibels is not a headline figure, but it is the figure the application actually needs.
The engineering problem that creates the demand is that automotive radar sensors sit behind bumper fascia and emblems, and every reflective discontinuity in that path generates multipath returns and ghost targets that the sensor's processing must then reject. A thin conformal absorber applied to structures inside the radar's field of view is a direct fix for a real and growing problem, on a per-vehicle basis across a global light-vehicle market measured in the tens of millions of units annually. Adjacent demand exists in millimetre-wave test and measurement, where compact absorbers reduce chamber size and cost, in electromagnetic interference control inside dense electronics packaging as switching frequencies rise, and in the base-station and repeater infrastructure being deployed for high-frequency telecommunications. None of these carries a stealth headline and all of them have shorter qualification cycles, larger unit volumes, and buyers who will accept twenty decibels because twenty decibels is what the problem requires.
What would have to be true
The firm's position is not that the reported result is overstated. It is that a single number describing absorption depth is insufficient to price the technology, and that four specific disclosures would move the assessment materially in either direction.
The first and most consequential is bandwidth. If the ninety decibel figure holds across a fractional bandwidth materially wider than the causality bound permits for the stated thickness, then either the film is thicker than described, or it is magnetically loaded, or it is not backed by a conducting plane in the measured configuration, or the result implies something genuinely novel about the material's dispersion. Each of those is a different investment proposition and the disclosure distinguishes between them. The second is angular performance, because absorbers tuned at normal incidence degrade as incidence angle increases, and an operational surface presents a wide and continuously varying range of angles. The third is polarisation dependence, since many patterned resonant absorbers perform asymmetrically and an uncontrolled threat environment offers no polarisation guarantee. The fourth is durability under environmental cycling, which is the variable that has historically separated laboratory absorber results from fielded products and which almost never appears in initial announcements.
Absent those disclosures, our framework treats this as a credible narrowband resonant absorber demonstration of the kind that the metamaterial literature has produced steadily for roughly two decades, distinguished by an impressive thickness figure rather than by a departure from known physics. That is a real engineering achievement with a real commercial market attached to it. It is not, on the evidence currently available, a development that changes the stealth balance, and the firm would treat any valuation premised on the latter framing as mispriced.
The most useful discipline in assessing any absorber claim is to convert the percentage into decibels, divide the causality budget by the result, and ask what bandwidth remains. That arithmetic takes under a minute and it separates the announcements that describe a product from the announcements that describe a resonance. In this instance it points away from the stealth aircraft in the headline and toward the bumper of an ordinary car, which is where the units are, where the qualification cycle is survivable, and where twenty decibels is not a disappointment but a specification.
Sources: Interesting Engineering (interestingengineering.com), reporting by Mrigakshi Dixit on the radar-absorbing film. Analytical framework references K. N. Rozanov, Ultimate thickness to bandwidth ratio of radar absorbers, IEEE Transactions on Antennas and Propagation (2000); standard treatments of Kramers-Kronig relations, Snoek's law, and radar cross section decomposition in the open electromagnetics literature. All bandwidth arithmetic is the firm's own illustrative scenario calculation from published bounds and is not a measurement of the material described. This note is for informational purposes only and does not constitute investment advice.
