This experimental study measures lateral shear adhesion of IP-S two-photon-polymerized microstructures on plasma-activated silicon, ITO-coated glass, MAPTMS-silanized silicon and SU8-coated silicon. SU8 interlayers give higher peak forces and retain more normalized adhesion after acetone exposure, while rounded structures improve adhesion on substrates showing interfacial failure. Finite-element simulations connect the geometry effect to residual stress. The measured conclusions concern the tested resin, geometries and processing protocols; SU8 geometry comparisons were excluded because partia

TWO-PHOTON POLYMERIZATION · MATERIALS & INTERFACES

Printing a tiny structure is only the beginning. What keeps it attached through the next step?

This study measures the bond beneath two-photon printed microstructures. Surface chemistry, a polymer buffer and even the shape of the footprint change how that bond holds.

A tightly focused laser writes a polymer in three dimensions. But every printed structure meets the rest of the device at one small interface.

01 / BREAK THE BOND

How hard is it to push off?

Choose a surface. Increase the sideways force. Watch where the structure fails.

YOUR TEST SPECIMEN

Oxygen-plasma-activated silicon gives the lowest reported mean peak force.

50 µm IP-S cube · blade speed 5 µm/s. Failure occurs at the rounded mean peak in this teaching animation.

THE SURFACE COMPARISON

A buffer changes the breaking point.

Swipe to explore the full chart.

Dots: individual tests in the open dataset. Dark marks: rounded mean peaks reported in the article. SU8 also changes the failure mechanism.

INSIDE ONE TEST

The moment of failure.

Swipe to explore the full chart.

Representative silane test: peak 30.5 mN.

Individual measured traces, separate from the surface means. The sharp drop marks failure. Elapsed time is from the recorded test; curves are condensed for display.

Stronger attachment can move the failure into the polymer.

On silicon, ITO and silane, the microstructure detaches at its base. With SU8, material can remain on the surface as part of the structure fractures. The measured force therefore captures a different failure mechanism.

02 / KEEP IT THROUGH PROCESSING

A strong start is not the whole story.

Acetone is common in microfabrication. Explore what remains of the bond after exposure and drying.

FOLLOW THE PROCESS

Two interfaces. One solvent.

Each series is normalized to its own initial mean peak force.

16% vs 50%. The SU8 buffer retains roughly half of its initial adhesion; silane retains roughly one sixth.

Measured point · means and standard errors from the open dataset.

THE FULL EXPOSURE SERIES

The bond changes before the shape does.

Swipe to explore the full chart.

Connecting lines guide the eye. Between sampled exposure times, the slider shows a linear interpolation, identified in the readout. These are separate conditioning tests, not a real-time measurement of a single bond.

A CLOSER LOOK / SUPPLEMENTARY FIGURE S2

Looks intact. Bonds differently.

The microscopy shows the same kind of printed cube after heating and then acetone. Its visible geometry remains intact. A force measurement reveals changes that appearance alone does not.

The paper also exposes samples to heat for two minutes and tests them after cooling. The SU8 series increases to about 1.4 times its initial adhesion, while the silane series stays near its initial value.

Interpolated separately from the measured temperature points. No combined solvent-and-heat prediction is implied.

Swipe to explore the full chart.

03 / CHANGE THE FOOTPRINT

The corners carry a cost.

With the same contact area, a round footprint distributes shrinkage stress more evenly.

REDESIGN THE BASE

21.7 MPa for cuboids. 28.2 MPa for cylinders.

Force estimates use the paper’s fitted adhesion stress on silane. SU8 was excluded from the shape comparison because structures partly fractured.

Finite-element simulations in the paper predict maximum residual stresses of 0.09 Eα for a square footprint and 0.05 Eα for a circular one, where E is Young’s modulus and α is isotropic shrinkage. The display uses E = 1 GPa, as in the paper. Its 2–12% shrinkage range comes from supplier data cited by the authors. The FEM represents shrinkage through thermal contraction in isotropic, linear elastic materials. Changing shrinkage changes this model stress; the measured force fit is determined by contact area and shape. The glowing rim illustrates the stress pattern qualitatively.

FROM LASER TO MEASUREMENT

A small structure, a complete fabrication workflow.

Write the polymer.

Two-photon polymerization confines curing to the focus of a laser. The study prints IP-S shells with an internal tetrahedral scaffold using a Nanoscribe Photonic Professional and a 25× lens.

Prepare the interface.

The substrates range from bare silicon and ITO-coated glass to MAPTMS silane and SU8 buffer layers. Development and UV post-curing complete the prints.

Measure the peak.

A blade advances laterally at 5 µm/s. The largest force before failure measures how strongly the structure is held. Subsequent tests probe compatibility with solvent and heating.

Slicing: 1 µm. Hatching: 0.5 µm. Base writing speed: 50,000 µm/s; shell writing speed: 100,000 µm/s. The article describes averaging three nominally identical structures for each condition; the open surface-comparison dataset includes more individual trials. The virtual shear test uses article mean peaks and illustrative motion, while the plots use measured source data.

THE ORDER OF OPERATIONS MATTERS

When should SU8 be baked?

Cross-linking the buffer changes its ability to bond with the print. The supplementary study compares different bake sequences.

~124 mN. Thin SU8, without a pre-print post-exposure bake.

Supplementary Note 1. For thick SU8, a post-print post-exposure bake was selected as a compromise between surface bonding and layer stability.

WHY THE INTERFACE MATTERS

Prints that belong in a larger process.

Stronger, more resilient attachment helps two-photon printed parts participate in the rest of microfabrication: lift-off, PDMS molding and multilayer structures. This paper makes the surface beneath the print a measurable design choice.

Research, made explorable.

THE SOURCE PAPER

Enhanced adhesion of high-resolution two-photon polymerization 3D-printing: a quantitative study on substrate treatments and process compatibility for microfabrication

Mathilde Roig, Leïla Ade, Pierre-François Calmon, Adrian Laborde, Samuel Charlot, Fabien Mesnilgrente, Christian Bergaud, Dina N Arvanitis and Vincent Mansard.

Additive Manufacturing Letters 19 (2026), 100415. doi:10.1016/j.addlet.2026.100415 ↗

Paper figures and supplementary microscopy © 2026 The Authors, CC BY 4.0. Supplementary microscopy is resized; Figure 2 is cropped and resized. The graphical abstract is an AI-generated conceptual illustration.

Measured data: Roig et al., open research dataset ↗, Licence Ouverte / Etalab 2.0. Conditioning means and standard errors are normalized to each series’ initial mean; force traces are condensed for display. Interactive animations are conceptual; numerical outcomes distinguish measured values, interpolation and model estimates.