My first experience with environmental testing was when I worked for a small R&D firm that was developing radiation-measurement equipment. One project was a radiation-detection badge for the U.S. Navy. The military specs required that the plastic badge survive hot and cold storage, a 6-ft drop onto a concrete floor, and depth testing down to 100 ft. All these tests were easy to perform, and with the property data sheets from suppliers the designer easily pinpointed a suitable ABS resin.

A more-difficult project also under development was a detection system that needed flexible sensors for monitoring radiation. Manufacturing the sensor required encapsulating various radiation-sensitive materials in an adhesive that could be silk-screen printed onto flexible, thin-film substrates. The radiation-sensitive materials were thermoluminescent (TLD) phosphors that when heated from between 120 and 240°C emit light if they have been exposed to radiation. The amount of light emitted by the phosphor was proportional to how much radiation the sensor saw. Once the signal is zeroed, or read out, TLD phosphors can be used again.

Traditional ways of heating such phosphors include heated metal planchets or hot gas. Our method, however, employed a 15-W CO2 laser which sped up the heating process and let us design a system that could heat a printed array of TLD phosphor spots individually. By reading each spot and treating the corresponding signal as a pixel, an algorithm could then plot the readings as a two-dimensional contour map of exposure.

For selection of the substrate and adhesive, supplier property data sheets were only helpful in rejecting materials that could not withstand high temperatures.

With hindsight, it's easy now to see how some of the problems we encountered late in the design phase may have been avoided if we would have been more vigilant testing prototypes made with sample materials, had better methods of testing at simulated high-temperature conditions, and more importantly, understood the entire manufacturing process for the substrate material we selected.

The substrate chosen was an aluminum-coated polyimide film that had a specified service life around 350°C with a 400°C maximum. Linear shrinkage of the bare polyimide was said to be about 2% at 400°C. Substrate shrinkage was a big concern because any dimensional change in the printed array would shift spot registration inside the test equipment used to heat the printed array with the laser.

Because material selection was being performed in tandem with building the test equipment, we could only evaluate substrate samples using a conventional high-temperature oven. Doing so helped eliminated lower grade polyimides that clearly shrunk, but this also made us too complacent about our selection when a higher grade material appeared to remain stable well above our predicted temperature range.

It wasn't until we started heating sheets multiple times, that we discovered how much difference laser heating had on polyimide performance. The laser beam heated each spot separately with about a 0.1-in.-sq. beam. Because there was thermal diffusion between spots through the polyimide film, each spot needed to be isolated from its neighbors. To do this, divots were placed below each spot in the aluminum platen used to support the sheet during heating. The sheets were also held in place by vacuum holes in the platen between divots.

As the number of laser heatings on each sheet increased, problems began to arise. The first sign of trouble was when the platen no longer pulled a vacuum on some of the sheets. The polyimide was less flexible and dimples appeared underneath each spot. Repeatability tests also indicated that the reader was receiving less signal on successive readings from certain areas of the sheet for the same amount of exposure. Weeks of fruitless testing on the signal-acquisition software did not improve the results. It wasn't until someone thought of laying the silk-screen printing pattern over the sheets that we saw how shrinkage of the film drastically changed the dimensions of the printed array. The spots diagonally across the sheet shifted position by almost half a spot diameter.

This nonlinear shrinkage we believed came from the sputtering process. In order to sputter the aluminum coating, the film was stretched as the metal was deposited. Because we had such a small order, the supplier produced our material on their test machine and not the big production unit that their original samples came from.

This taught us that if you are developing parts to be used in adverse conditions, make sure you know as much as possible about the sample you are evaluating, including all the processing parameters used to make it such as special additives, how old the material is, and even the piece of equipment that it was made on. Otherwise you may be tasked with trying to salvage a shipment of material that doesn't meet your requirements.