3D Printing Treatment Two: Two-Photon Lithography for Micro-Optics

Two-photon polymerization, often shortened to 2PP or MPL, uses a femtosecond laser focused through an objective. At the focal point the intensity is high enough that two photons are absorbed at once. The resist only polymerizes there. Away from the focus the light passes through without curing anything, so you get a solid voxel inside a liquid bath rather than a cured layer on a surface.

That voxel is small. Practical systems produce features around 100 to 500 nm with surface roughness in the tens of nanometers, and a single lens can be a few hundred micrometers across. The laser is scanned or the stage is moved, and the part is built as a stack of overlapping voxels. A 200 µm microlens might take a few minutes; a full lens array on a fiber tip can run into hours.

The build is not the finished part. After development in a solvent, the structure is a soft polymer with low crosslink density. It absorbs moisture, creeps under load, and will not pass a thermal cycling test. The 3d printing treatment two step is what turns that green structure into something you can mount, bond and ship.

• 1
  Feature sizeRoughly 100–500 nm lateral, set by the objective NA and laser power.
• 2
  Build volumeMillimeter scale in X and Y, sub-millimeter in Z for high-NA optics.
• 3
  ThroughputMinutes for one lens, hours for a dense array on a fiber tip.

Treatment is not one operation. It is a short sequence, and each step has a failure mode. The first is a UV or thermal flood cure that raises crosslink density. Undercure leaves the structure tacky and dimensionally unstable; overcure shrinks it. A 1 to 3 percent linear shrink is normal, which matters when the lens is designed to a focal length tolerance.

The second step is a thin inorganic coating, usually by atomic layer deposition or sputtering. A 10 to 50 nm layer of oxide or nitride blocks moisture uptake and raises the refractive index contrast at the surface. Without it, a hybrid resist lens will haze after weeks in a humid lab. With it, transmission stays stable and the surface can be cleaned.

The third step is mounting. Most 2PP optics are too small to clamp, so they are bonded to a ferrule, a fiber tip or a machined holder. Adhesive choice matters: a rigid epoxy holds alignment but can crack the lens during cure shrinkage, while a soft adhesive survives thermal cycling but drifts. We usually machine the holder to a flatness under 5 µm so the bond line stays uniform.

• 1
  Flood cureUV or thermal; 1–3% linear shrink is expected.
• 2
  Barrier coating10–50 nm ALD oxide or nitride for moisture resistance.
• 3
  MountingBond to ferrule or machined holder; holder flatness under 5 µm.

The clearest fit is a freeform surface that cannot be turned on a lathe. An off-axis micro-lens, a lens with a non-rotationally symmetric correction, or a stack of two surfaces separated by a few micrometers. Diamond turning needs a rotationally symmetric surface, so an off-axis design forces you into a printed or lithographic route.

Fiber coupling is the second common case. A lens printed directly on a fiber tip removes two air-glass interfaces and the alignment step that goes with them. The printed lens sits within a few micrometers of the core, so the coupling tolerance stack is shorter. This is where 2PP beats a separately machined ball lens on both alignment and part count.

The third case is a small batch. If you need 10 to 200 units of a micro-optic for a prototype line or a research instrument, printing avoids the tooling cost of molding. Above a few thousand units, molding wins on unit cost, and the printed part becomes the master for a mold insert. The insert itself is often machined or electroformed, which is where our shop comes in.

There are cases where printing is the wrong answer. A lens larger than about 1 mm across takes too long to print and will not hold its figure. A part that must survive 150 °C continuously will creep in most resists. And a simple plano-convex lens in PMMA is cheaper to diamond turn or mold than to print.

• 1
  Freeform surfacesOff-axis or asymmetric profiles that turning cannot produce.
• 2
  Fiber-tip lensesFewer interfaces and a shorter alignment stack.
• 3
  Low volume10–200 units avoid mold tooling cost.
• 4
  Wrong fitParts over 1 mm, hot environments, simple symmetric lenses.

A printed lens rarely ships alone. It sits in a holder, a ferrule or a lens barrel that has to hold alignment over temperature. Those parts are machined. We run 16 simultaneous 5-axis machining centers and 16 mill-turn centers, which covers both the prismatic holder and the turned ferrule in one setup where possible.

Tolerances on the holder drive the optical result. A lens bonded into a bore with 20 µm of clearance will move off-axis by that much, and the coupling loss follows. We hold ±0.005 mm on critical diameters and Ra 0.8–1.6 μm on bond surfaces, so the adhesive layer stays thin and repeatable. For a lens array, the pitch between bores matters more than any single diameter.

Material choice for the holder follows the thermal budget. Aluminum 6061-T6 is the default for lab hardware. For a fiber-coupled package that sees thermal cycling, 17-4PH stainless or Inconel holds alignment better because the expansion coefficient is closer to that of the glass. We machine all three, plus titanium TC4 when weight matters.

The workflow we see most often: the customer prints the optical surface, sends us the holder drawing, and we machine the metal, anodize or black oxide it, and ship both. Final bonding and optical test stay with the customer because it needs their light source. When the customer wants a single source, we can do the bond and the dimensional check, but the optical performance test still needs their setup.

• 1
  Holder tolerance±0.005 mm on critical bores keeps the lens on axis.
• 2
  Bond surface finishRa 0.8–1.6 μm for a uniform adhesive layer.
• 3
  Thermal match17-4PH or Inconel for cycling; 6061-T6 for lab use.

Printed micro-optics are hard to inspect after the fact. A 200 µm lens with a 300 nm surface figure is beyond most shop-floor metrology, so the design should be verified on a sample print before the holder is machined. Print one lens, coat it, and measure focal length and wavefront. Only then cut the metal.

For the machined holder, the checks are conventional. Bore diameter, concentricity, and the flatness of the bond face. We inspect 100% of parts before shipment and can supply reports on request. Raw material certificates come with the lot, and in-process monitoring catches a drifting bore before the run finishes.

One practical note on cleaning. A coated printed lens should not go into an ultrasonic bath, because the coating can lift at the edge. The machined holder can be cleaned normally. Keep the two processes separate and bond after cleaning, not before.

If the part is going into a medical or automotive program, the paper trail matters as much as the dimensions. We hold ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022, so the holder can be supplied under the same quality system as the rest of the assembly.

• 1
  Print firstVerify focal length and wavefront before cutting the holder.
• 2
  Holder checksBore diameter, concentricity, bond-face flatness.
• 3
  CleaningNo ultrasonic bath on coated printed lenses.