The study asks how two-photon lithography can fabricate preassembled microscale joints with moving parts without sacrificial supports. Experiments show that a high-yield-stress epoxy photoresist temporarily immobilises unsupported components, while fluorosilane surface grafting reduces friction after development. Cone–sphere tests establish a reliable 2.5 μm clearance under selected printing conditions; printed hair-inspired joints move reversibly under fluid drive, and drying dice arrays yield approximately uniform face frequencies. Fluid–structure responses are also simulated. The authors id

Two-photon lithography · Ma et al., 2026

A microscale machine. Printed already assembled. With room to move.

A laser writes the parts. The surrounding epoxy holds them in place. A molecular surface layer helps them move after release.

Three experiments. One new way to make tiny mechanisms.

Minimum reliable clearance under selected printing conditions

Lower measured friction after surface treatment

A printed object can be intricate. Can it also be a mechanism?

The idea

Print the parts together. Keep them apart.

A kinematic pair is two connected parts that can move relative to one another—like a ball in a socket.

Two-photon lithography uses a tightly focused pulsed laser to trigger chemistry in tiny volumes of a light-sensitive resin. Scanning the focus writes the chosen three-dimensional geometry.

At this scale, two problems meet: an unconnected part needs support during printing, and it must avoid sticking once that support is removed. This paper combines a solid-like epoxy photoresist with a low-energy surface coating to tackle both.

The uncured resin itself provides temporary support. No sacrificial printed pillars are needed.

01 / The release experiment

Can you free the sphere?

Choose a clearance and laser power. Then remove the uncured resin. Will the sphere detach—or stay stuck?

Your printing recipe

Try a lower power. Can you release the sphere without making the gap larger?

Replay the observations

A little space makes a big difference.

At small gaps, resin removal and overexposure can leave parts adhered. At larger gaps, gravity released the sphere into the hollow cone.

The map reproduces the paper's tested combinations (Figure 5g). The 3D geometry, clearance and release motion are schematic.

The smallest stable printed lines were 390.2 ± 4.5 nm wide. Reliable moving clearances were larger: 2.5 μm at 15 mW and 800 μm/s. A fine line and a free joint are different tests.

How it works

Support, then release.

One material holds the structure while it is written. A molecular coating prepares it for motion.

Hold

Solvent removal gives the epoxy photoresist a solid-like state. Its yield stress lets it immobilise features that have no printed support.

Write

A 780 nm femtosecond laser focuses two-photon excitation into the resin. The focus scans each layer; a bake advances epoxy polymerisation.

Release

Development removes uncured resin. FDTS fluorosilane grafting forms a self-assembled surface layer that lowers surface energy and friction.

Photoresist. TPEO-C combines SU-8 and EPON 1031 epoxy monomers at a 1:3 ratio, ITX sensitiser and PYZ photoacid generator. Across formulations, measured yield stress ranged from 1,048 to 9,084 Pa.

Laser writing. 780 nm, 80 MHz, 100 fs pulses through a 63×, NA 1.4 oil-immersion objective. Hatching: 300 nm; layer spacing: 150 nm. Meandering scans fill each layer sequentially from the substrate upwards.

After exposure. A 95°C, 30-minute bake advances cationic epoxy ring opening. PGMEA–silane developer is used for at least 30 minutes, followed by cleaning in pure PGMEA.

Optical uniformity. The reported refractive-index change after curing was 0.0045 at 780 nm, helping preserve a uniform optical path during fabrication.

The surface matters

A molecular layer. A measurable change.

FDTS molecules bond to the epoxy surface. Their fluorinated ends make it less prone to adhesion. Compare the measured friction of untreated and coated samples.

These are sample friction measurements, not predicted torque or friction for the animated microjoint.

Data from Ma et al. (2026), supplementary source workbook, Figures 3h–i. Redrawn from the supplied data; error bars retained.

02 / Motion by design

Give it a nudge.

Move the fluid. Reverse the flow. Change the hair. The socket and shaft shape set the room for motion.

Choose an architecture

Measured swing limits

The substrate constrains movement to its plane. Split tips can entangle, limiting swing. The experiment used a 120 rpm shaker and a fluid speed of 0.01–0.1 m/s.

Illustrative motion interpolated within measured swing limits; not a fluid-dynamics calculation. The paper separately models fluid–structure interactions in Figure 7.

03 / Motion becomes information

Let drying decide.

Freed components do more than swing. During solvent drying, tiny captured dice can tumble into new orientations.

Watch the actual experiment

In the microscope, the receding liquid boundary changes the forces acting on each die.

Supplementary video 1, Ma et al. (2026). Excerpt 64–88 s; brightness adjusted and compressed, with no speed change. Source · CC BY 4.0.

Small fluctuations. Different final faces.

The authors propose that tiny initial thermal fluctuations are amplified by capillary forces during drying. They examined ten 4 × 4 arrays—160 dice . Face frequencies were close to the uniform value of 16.7% , with a slight excess of 3 and 5.

The applet uses software randomness to illustrate changing orientations; it does not reproduce the physical randomness of drying.

What this opens up

From printed shapes to tiny moving systems.

The study demonstrates preassembled joints, reversible fluid-driven hairs and drying dice arrays. Geometry becomes a way to tune what a microscale structure can do.

The authors propose uses in microrobotics, hair-inspired flow or touch sensing, microparticle manipulation and physical random-number generation. Improving printing and gap precision remains a route towards smaller mechanisms.

The source

Self-assembled monolayer enhanced support-free two-photon polymerisation for microscale kinematic pairs

Zhiyuan Ma, Bingkun Cheng, Yu Long, Xiaobing Wang, Tao Wang, Kun Xu, Bo Song, Yu Lei, Yan Wang, Kashif Ali Khan, Lei Zhang, Biwei Deng and Cuifang Kuang.

Virtual and Physical Prototyping 21(1), e2665992 (2026). doi:10.1080/17452759.2026.2665992 ↗

Article figures and supplementary data: © 2026 The Authors, CC BY 4.0. Figures reproduced and resized; friction data redrawn. Video and source data · Supplementary Methods and figures. The graphical abstract is an AI-generated conceptual illustration. Interactive geometry and motion are educational schematics; observed release outcomes, swing limits and friction values come from the paper.