In a nutshell, normal components (ground to 1mm think where needed) are mounted on an 0.3mm elastic substrate and connected together by serpentined-serpentine tracks. The substrate and circuit is capped with a flexible lid, and lubricating silicone oil injected (see image).
However, there are subtleties.
The main one being that only the ICs and components are actually mounted on the substrate - on short elastomer posts in each case. Tracks are completely free-floating - leading to at least 10x more stretchability compared to similar tracks either bonded to, or embedded in, the same elastic substrate.
To keep the fragile copper in the neutral plane of the track structure, the tracks are a polyimide-copper-polyimide sandwiches. These were patterned and etched elsewhere, transferred to the substrate, then oven re-flowed with low-temperature solder after components were hand-positioned.
So successful has this floating track strategy been that the substrate loses less than 10% stretchability after the components and tracks are mounted. Maximum strain in the copper is under 0.2% for 50% stretch in both directions at once.
Uni-directional stretching is in some ways more difficult to cope with that two-dimensional stretching - because the ICs tend to collide and tracks tangle.
This particular device has been stretched uni-directionally by 30% (40% is the upper limit before collisions) 6,000 times with no ill effects.
The layout is optimised to minimise the chance of conductor tangling, with low walls built on the substrate in critical places to keep conductors apart. The substrate is also roughened to stop the tracks sticking.
Computer models agree almost completely with the actual devices and, in one place, predicted a substrate anchor in the middle of a particular track would reduce tangling, which it did in practice.
The device is a working prototype - having been demonstrated sensing electrocardiograms (ECGs), electromyograms (EMGs), electrooculograms (EOGs) and electroencephalograms (EEGs) through external electrode meshes (also serpentined-serpentine) - see photo left.
This coil has to be within a few millimetres of the patch to couple in enough power.
Researchers came from Northwestern University and University of Illinois at Urbana-Champaign, and the work is published in Science. For more details click here.