Kim and Saha investigate how to preserve nanoscale printing fidelity while projecting many laser focal spots in parallel. Their custom grayscale projection two-photon lithography system uses binary, sub-diffraction motifs to independently tune spot intensity. Fourier-optics simulations select the patterns; optical measurements and fabricated polymer structures demonstrate control of linewidth, compensation of proximity and illumination effects, and continuous single-pulse-per-layer printing. Motif palettes depend on the optical configuration and require recalibration when the number of project

NATURE COMMUNICATIONS / 2026

Tiny on/off patterns tune thousands of laser spots at once—so nanoscale structures can print quickly and keep their pores.

Based on research by Harnjoo Kim & Sourabh K. Saha

01 THE PRINTING LAB

Rescue the pores.

A dense center can grow too much. Change its brightness and discover why the same laser exposure can make a very different structure.

A locally dimmer mask limits growth in the crowded center, preserving the open architecture.

YOUR EXPERIMENT

Where should the light go?

Keep the boundary bright. Reduce the light in the crowded center.

The paper’s mask uses 60 / 80 / 100% from center to boundary. The model illustrates local overgrowth; it does not calculate resin chemistry.

At full brightness, excess growth in the dense central region merges neighboring features. Try lowering the center toward 60%.

THE REAL EXPERIMENT

Same exposure. Open architecture.

The experimental woodpile on the left has a closed center. On the right, a grayscale mask lowers the central intensity and preserves the pores.

Both structures in Figure 4a were printed with the same laser power and 5 ms per layer . The fine binary patterns inside the mask provide the local brightness control.

THE CENTRAL IDEA

The mirrors are binary. The light doesn’t have to be.

Each micromirror is either on or off. But seven-by-seven groups of mirrors form patterns smaller than the optical detail the lens can resolve. The lens filters those patterns into smooth spots—with different brightness.

02 THE MOTIF PALETTE

An on/off mask. 32 shades of light.

Choose a pattern from the paper’s exact palette. Watch its brightness change while the optical spot stays almost the same width.

The optical footprint stays similar. Changing the peak intensity changes how much resin cures.

CHOOSE A MOTIF

Same scale. A different dose.

The pattern’s spatial frequencies determine which light reaches the focal plane.

RELATIVE PEAK INTENSITY

100.0% measured · motif 1

Simulated: 100.0%

851 nm ± 65 nm · 2 standard deviations

Width is the mean of five measurements. A missing entry means no width is reported for that motif and exposure.

Light intensity changes the printed width.

Measured wire widths at the selected exposure. Error bars: ±2 s.d. Gaps: no reported width.

Smaller prints, similar light spots.

Across the palette, measured optical widths average 830 nm , with a 14 nm standard deviation between motifs.

Less light means less resin receives enough dose to become solid. The printed feature can therefore be much narrower than the optical spot.

The rendered profiles are Gaussian illustrations scaled to the measured width and peak intensity. The paper’s measured profiles and simulations appear in Figures 2, 3 and S5.

Motifs redrawn from Supplementary Table S3; measurements from Source Data, Figures 3a, 3b and S5a. © Kim & Saha (2026), CC BY 4.0. Each projected pixel is 114 nm wide; the first Airy-disk diameter is 785 nm.

03 THE SPEED OF A LAYER

One pulse. One whole layer.

Brightness is encoded in the pattern itself. No sequence of differently timed masks is needed to give each spot its own intensity.

One patterned pulse exposes an entire layer. The light sheet advances through the resin as the architecture grows.

BUILD AN ARCHITECTURE

Project the whole pattern at once.

A patterned femtosecond light sheet cures a thin layer. Move through the resin and repeat.

4.8ms

24 layers × 1 pulse ÷ 5,000 pulses/s = 4.8 ms of ideal pulse-train time.

Illustrative 24-layer schedule at the printer’s 5 kHz repetition rate. The time counts delivery at 5 kHz; motion, settling and post-processing are separate. Animation is slowed for inspection.

One-pulse layers were demonstrated in the porous structures of Figure 5d–e. The finest nanowire used five pulses per layer.

Minimum wire width reported in the abstract. Figure 5c shows a 53 nm example.

Focal spots whose intensity can be independently tuned within one projection.

One layer per pulse at the printer’s 5 kHz laser repetition rate.

The abstract reports a volume rate of about 215 mm³/h . The paper’s calculation uses the projected area, layer spacing and 5,000 layers per second, giving about 217 mm³/h for that geometry.

Counting each focal spot as one voxel gives 79 million voxels/s . The reported 1.7 billion voxels/s instead uses a 77 nm voxel based on the minimum feature dimensions and the nanowire’s five-pulse exposure. The voxel definition and print settings determine the rate.

These are distinct demonstrations and operating measures; the finest wire and the one-pulse structures used different print settings.

INSIDE THE PRINTER

A digital pattern. A physical transformation.

The contribution is local intensity control inside one binary projection—preserving the speed of printing many features together.

Pattern a femtosecond pulse

A custom printer uses an 804 nm, 35 fs Ti-sapphire laser. A digital micromirror device redirects the beam through exact on/off motifs.

Focus in space and time

A 60×, 1.25 NA objective projects 114 nm pixels. Spatial and temporal focusing localizes curing around the patterned light sheet. A light-sensitive initiator absorbs two photons in one excitation, starting polymerization. This nonlinear absorption is strongest in intense focal light.

Build through liquid resin

The substrate moves through a custom acrylate resin as new masks are projected. Uncured resin is solvent-washed away; the structure is UV post-cured.

WHAT THIS OPENS UP

Speed and precision, in the same projection.

Porous architectures need their empty spaces as much as their solid parts. Independent brightness control offers a route to fast printing with fine features and more faithful geometry.

The paper demonstrates correction of proximity effects, uneven illumination and stitching artifacts. A second objective lens and gold-film ablation also show that the motif approach can extend beyond one printing configuration.

Applications in photonics, metamaterials and miniature devices motivate this control. Here, the advance is the fabrication capability: choosing the light intensity locally without giving up whole-layer projection.

THE SOURCE PAPER

Grayscale projection two-photon lithography using sub-diffraction motifs for ultrafast and precise nanoscale 3D printing

Harnjoo Kim & Sourabh K. Saha

Nature Communications 17 , 6981 (2026) · DOI: 10.1038/s41467-026-73782-x

The palette contains 32 selected motifs. Four-fold symmetry reduces the search over 49 binary pixels to 8,192 patterns, which the authors evaluate with Fourier optics simulations.

RESULTS · METHODS · SOURCE DATA

This experience is based on the article and its Supplementary Information and Source Data. The woodpile model illustrates local overgrowth and brightness correction from Figure 4a. The motif shapes are exact redrawings of Supplementary Table S3; intensity, optical width and printed wire measurements come from Figures 3a, 3b and S5a Source Data. Pulse timing is an illustrative 24-layer pulse-train calculation using the reported 5 kHz laser repetition rate. Each pulse itself lasts 35 fs.

Figures 2, 4 and 5 are reproduced from Kim & Saha (2026). The Figure 4a comparison is cropped; original figures are available in the figure viewer. Motifs and measurement charts are redrawn. Source material is used under Creative Commons Attribution 4.0 International.

The graphical abstract is an AI-generated conceptual illustration , informed by Figures 2, 4 and 5. It is a visual explanation of the motif-to-light-to-print principle.

Explore the science. Read the paper.

Figure 4

Original figure reproduced from Kim & Saha (2026), Nature Communications. CC BY 4.0.