Two-photon lithography: printing tiny 3D objects with light

What the method does

Most 3D printers build objects layer by layer from fairly coarse material. Two-photon lithography, also called two-photon polymerisation, works at a much finer scale. A liquid resin is exposed to a tightly focused femtosecond laser beam. The resin hardens only where two photons are absorbed together, which happens only at the very centre of the beam where the light is most intense. By moving that focal point through the resin in three dimensions, the machine draws a solid structure one tiny point at a time.

Where the method came from

The principle was shown in 1997. Shoji Maruo, Osamu Nakamura and Satoshi Kawata reported three-dimensional microfabrication by photopolymerisation stimulated by two-photon absorption. The abstract describes a Ti:sapphire laser at 790 nm with 200 femtosecond pulses, and a resin of photoinitiator with urethane acrylate monomers and oligomers (Optics Letters, 1997). The abstract does not state a feature size.

A 2023 review in iScience by Jaiswal and colleagues, titled "Two decades of two-photon lithography", looks back over about two decades of the method. It describes two-photon lithography as a versatile technology for additive manufacturing (Jaiswal and colleagues, 2023).

Evidence, sorted by strength

Established. The basic mechanism is well understood: two-photon absorption confines hardening to the laser focus, which is what allows fine features. Jaiswal and colleagues list among its drawbacks that it is not a scalable commercial technology yet.

Demonstrated.

• Maruo, Nakamura and Kawata (1997) demonstrated 3D microfabrication by the process described above.
• Gu, Mao and colleagues (Nature, December 2025) used a metalens array to produce more than 120,000 cooperative focal spots. Their abstract reports feature sizes down to 113 nanometres, a throughput exceeding 100 million voxels per second, and centimetre-scale 3D architectures (Caltech record).
• Binder and colleagues (Additive Manufacturing, 2025) built a two-photon system with a resonant scanner. Their abstract claims throughput at least one order of magnitude higher than existing commercial products (Binder et al., 2025). This is a single study, and the abstract gives no specific write times.

Reported or speculative.

• Otuka and colleagues (2021) say two-photon polymerisation has mainly been used for microscaffolds that culture cells, tissues and microorganisms. The review describes micro-devices such as micro-turbines that can be moved with an external magnet, and does not use the term microrobot (Otuka et al., 2021). Other applications discussed in the literature should be read as proposals rather than demonstrated results.

Practical limits

Throughput. This is the main barrier. Jing and colleagues (2022) report "a limited throughput resulting from the low scanning speed in the order of less than 1 mm per second". They add that a millimetre-scale construct with submicron accuracy "might take from a few hours to days" (Jing et al., 2022).

Materials. Few resins are suited to the process. Jaiswal and colleagues call the lack of two-photon polymerisable resins a "bottleneck" to the growth of the method. Jing and colleagues note a shortage of biomaterials for scaffolds, and say the toxicity of unreacted monomers, degradation products and photoinitiators cannot be ignored.

Shrinkage. Resins shrink as they harden, which can distort fine structures. Jing and colleagues describe shrinkage during processing that caused aberrations in one hybrid material. Jaiswal and colleagues mention an ultra-low shrinkage hybrid resin and note that positive-tone materials can avoid shrinkage during fabrication. The sources give no general figure, so the effect depends on the material.

Equipment cost. Jaiswal and colleagues describe femtosecond pulsed lasers as an "expensive commodity". No verified price for a commercial system was found in the sources.

Size of printed objects. Jaiswal and colleagues say it is "extremely difficult to fabricate structures with heights beyond few hundred micrometers" using this method. Jing and colleagues name the limited overall size of structures as a main limitation. The 2025 metalens work addresses printable area by writing across centimetre-scale write fields. The Lawrence Livermore National Laboratory (LLNL) announcement of December 2025 says earlier systems, which relied on microscope objectives, confined the printable area to a few hundred microns. This is a press claim and has not been checked against the paper.

Limits and open problems

• Speed against detail. Finer voxels mean longer print times. Jaiswal and colleagues say the raster scanning of tiny voxels results in longer printing time, and Binder and colleagues link long processing times to high resolution.
• Parallel writing is not yet shown at commercial scale. The LLNL announcement says that further gains will come "as higher-power lasers, larger metalens wafers and faster modulators become available", which is a statement about the future. The sources reviewed give no measured cost for the parallel approach.
• Resin supply and safety. New resins are needed for both mechanical and biomedical uses, and biocompatibility is a concern for scaffolds.
• Independent checking. Most of the headline speed and size results come from single studies or from press material. Replication by other laboratories has not yet been reported in the sources reviewed.

What to watch next

• Whether other groups reproduce the metalens throughput and the 113 nanometre feature size.
• Whether the resonant-scanner speed-up from Binder and colleagues is matched or exceeded in other systems.
• New low-shrinkage and biocompatible resins, which the reviews identify as a bottleneck.
• Verified prices and specifications for commercial machines.
• Working microrobots made by this method, moving beyond demonstrations of steered micro-machines.