UpNano has introduced 3D Adaptive Resolution, or 3DAR. The company says the voxel formed during two-photon polymerisation can be shaped in all three spatial directions according to the geometry being printed. The objective is not merely smaller detail, but optical-grade curved surfaces together with large layer spacing for micro-optics, photonics packaging and semiconductor applications.

A voxel is a volumetric reaction zone

2PP uses a focused pulsed laser to trigger polymerisation in a minute region of a photoresist. Because the reaction has a threshold, a solidified zone forms around the focus; this is the voxel. Its dimensions depend on optics, laser power, exposure, material and scanning. A voxel is therefore not a rigid printing pixel but a process feature controlled by settings.

UpNano's earlier Adaptive Resolution divided a component into fine and coarse modes: the shell and critical details could use a small voxel while bulk used a larger one. The announced 3DAR promises more continuous control. According to the company, XYZ voxel dimensions follow the current contour so curvature is not approximated only by many closely spaced planar layers.

Why curved micro-optics matter

With conventional layer approximation, large layer spacing leaves stair steps on a curve. Smaller spacing reduces them but requires more paths and time. In micro-optics, surface is functional rather than cosmetic: form error and roughness can cause scatter, wavefront error or coupling loss. The commercial promise is therefore to remain close to the target contour with fewer layers.

The principle is related to grayscale lithography, which varies height through dose control. UpNano distinguishes 3DAR by applying voxel tuning not only to a 2.5D topography above a plane but inside freeform three-dimensional surfaces. Other 2PP vendors offer related dynamic-voxel approaches, so this is best described as UpNano's implementation on its platform, not the general invention of tunable voxels.

Published figures and measurement boundaries

The company reports Sq surface roughness of 3 nm and form accuracy better than 20 nm. It demonstrates a 125 µm spiral phase plate in under ten seconds and a 7 µm aspheric microlens in two to three seconds. These are notable demonstrations, but the announcement does not provide a full protocol, sample count, material and optical configuration or independent laboratory result.

Sq is areal root-mean-square roughness, not the same metric as line-based Ra. Field size, separation of waviness, filtering, instrument and preparation matter. Form accuracy likewise needs a defined CAD registration, uncertainty and evaluated area. The numbers are meaningful for comparison only when the methodology is held constant.

Direct printing on a substrate

UpNano also positions 3DAR for direct printing on chips, optical facets and fibre arrays. It describes automatic marker recognition that aligns a microlens or another element to an existing component. In this workflow, positioning error, substrate flatness, resin shrinkage, refractive index and development can matter more than nominal printer resolution.

The first 3DAR-capable systems were purchased by an unnamed Asian company, and UpNano reports interest in more systems. That is an early commercial signal, not evidence of installed volume or production yield. The buyer, application and acceptance criteria remain undisclosed.

What a production introduction must prove

A user should measure contour error, surface roughness, optical function and part-to-part variation on a reference geometry. Testing must separate printing from washing, development, drying and any post-cure. Direct-on-substrate production also needs alignment repeatability, adhesion and stability after thermal or humidity exposure.

Seconds of demonstration exposure do not necessarily equal total cycle time. File preparation, substrate loading, alignment, resin handling, development, inspection and scrap belong in unit cost. In series production, parallelisation and utilisation can matter as much as the exposure time for one optic.

Verified facts, company claims and conclusion

Verified facts are the 6 October 2026 announcement, the described principle of dynamic XYZ voxel control and the named application areas. Nanometre roughness, form accuracy, printing times and early customer interest are UpNano's own claims; the source contains no independent confirmation. LAYERTIMES concludes that the important proposition is not one record but a new way to manage the compromise among layer spacing, surface and throughput. Repeated optical measurements, yield and total cycle time must establish industrial impact.