Photonic wire bonding

A photonic wire bond is a three-dimensional polymer waveguide written directly between two photonic ports, chip to chip or chip to fiber, by two-photon polymerization inside a drop of photoresist. The machine first images the actual positions of the two facets, then computes a smooth trajectory with mode-matched taper ends and writes the bridge in place. The defining move is that alignment happens after placement, in software: components can be attached with a few microns of placement error and the written bond absorbs it, where a butt-coupled assembly would demand sub-micron active alignment.

The numbers have matured from demonstration to product. Chip-to-chip insertion losses of a fraction of a decibel and chip-to-fiber losses around half a decibel to one decibel are reported routinely, with broadband, polarization-tolerant behavior set by the taper design. Commercial writers from the Karlsruhe lineage now ship, and the technique sits in the packaging toolbox alongside microlens arrays and evanescent attaches for exactly the assemblies driving current demand: laser and semiconductor optical amplifier attach to silicon and silicon nitride platforms, fiber array attach for co-packaged optics, and mixed-platform modules where each die is grown in its native material and joined afterward.

The lab-practice angle is why this entry sits in two categories. Wire bonds change what a coupling budget looks like: the interface loss becomes a written-component spec with a distribution, verified per bond by the tool's cameras and post-write transmission checks, rather than an alignment-rig outcome that drifts with epoxy cure. The failure modes to qualify are polymer aging, thermal cycling, and high-power handling, and serious deployments qualify them the way any packaging step is qualified, with Telcordia-style stress batteries and witness samples.

References: N. Lindenmann et al., Opt. Express 20, 17667 (2012); M. Blaicher et al., Light Sci. Appl. 9, 71 (2020).