Carbon nanotube patterning without the etch: a pressed filter attacks a 90% material loss

Take the loss figure at face value and the arithmetic runs in the right direction: if only about a tenth of a deposited film survives etching into the shape you actually wanted, a patterning step that wastes nothing returns on the order of ten times the usable film per gram of nanotube bought ^[17]. That is the entire commercial case, and it rests on one number supplied by the authors. Dmitry Krasnikov of Skoltech puts the loss at "as much as 90%" and adds that etching degrades what is left ^[4]; the announcement does not tie that figure to a measured baseline, so read it as the team's characterisation of common practice rather than an audit of anyone's line.

The mechanism is worth separating from the marketing framing of cheaper, faster and more resource-efficient ^[16], because what changed is which object carries the pattern. The mask and the collection substrate are now the same piece of nitrocellulose filter: a laser-cut hot metal stencil bearing the inverse pattern is pressed into it at roughly 200 megapascals, which crushes the pores shut in the regions that should stay bare ^[6]. That is about 2,000 times atmospheric pressure ^[18], which is why a conventional press does the job rather than a semiconductor tool. Deposition then relies on an aerosol fact rather than a chemical one: single-walled nanotubes are light enough that their inertia is negligible, so they go where the carrier gas goes, and the gas can only go through the pores that survived ^[8]. Nothing is removed afterwards, and no organic solvent enters the process, which is what preserves film quality and eases the later transfer ^[9].

The delta against the group's own earlier attempt is the part that tells you this is engineering rather than a press release. That version blocked deposition by sputtering copper onto selected areas ^[10]. It was already single-stage, but the blocking agent was foreign matter: patterns had to be continuous, with no isolated or weakly connected features ^[11], and copper migrated onto the nanotubes and degraded them ^[12]. Hot pressing swaps an added layer for a mechanical state of the substrate, and both problems go with it.

What has been shown is fidelity, not devices. Optical microscopy, scanning electron microscopy and electrical measurement put the lines and gaps on the stencil geometry ^[13], and the regions meant to stay empty stayed very nearly empty ^[14]. Reusability and scale were probed with larger stencils and repeated-use tests, reported as tests conducted rather than as a cycle count ^[15]. The patterned film also still has to come off the membrane and onto a working substrate ^[7], and that transfer is where any yield claim would be settled or lost.

For the applications named, advanced optics and strain sensors for structural integrity monitoring ^[2], material cost is a per-unit cost, so a tenfold change in nanotube consumption per pattern compounds with however many units get installed. That is a different kind of improvement from a better film, and a harder one to argue with.