Optical components makers are only too familiar with the problems of thermal management — controlling the temperature of transmission sources like lasers and light-emitting diodes. Device operating temperatures need to be stabilized to ensure correct performance and ultimately to prevent failure in the field.
This issue is particularly important for telecoms optical components, which can be installed in a range of environments — from enclosed central offices to anywhere that you can place a telephone pole.
The telecoms industry has traditionally employed active cooling solutions based on the thermocouple — a device that consists of an n-type and a p-type semiconductor connected by a metal plate. Electrical connections at the opposing ends of the p- and n-type material complete an electric circuit. When current is supplied to the two dissimilar materials, the thermocouple cools on one side and heats on the other by what is known as the Peltier effect.
However, thermoelectric coolers (TECs) manufactured using sintered pressed powders have a number of limitations. For a start, the manufacturing processes are not compatible with high-volume implementation. Second, reliability has historically been poor — even the best modules today still fall short of accepted semiconductor reliability standards. And third, existing TECs are relatively large. This means that the heat-pumping density (a measure of the heat extracted per unit area) is low, which restricts miniaturization.
To further complicate matters, optical components packaging must typically integrate the solid-state device with geometric optics. The integration of these two dissimilar technologies requires manufacturing tolerances in the order of a few microns. This requirement exists not just during the manufacturing phase but throughout the operating lifetime of the device. To maintain these assembly tolerances and to prevent contamination, hermetically sealed metal and ceramic packages have been used.
These packages are relatively expensive and one way to reduce the cost of a component is to shrink the packaging. Figure 1 clearly shows how the package size of optoelectronic modules has decreased over time, but what can’t be seen is the increasing power levels of the devices contained therein. But while there has been a trend towards higher-power devices and smaller packages, the scalability of the TECs has been slow to follow. As a result, the device design has been constrained by the need to shrink the space that the device occupies while maintaining a sealed environment.
To address these constraints, Nextreme has developed a new, semiconductorbased manufacturing approach to create microscale thermocouples. The process integrates thin-film bismuth telluride (Bi2Te3) into a conventional solder bump process to produce a thermally active solder bump. The process consists of three key steps: first a solder bump is put down by electroplating metals, consistent with widely accepted manufacturing processes; meanwhile, Bi2Te3 thin films are produced by metal-organic chemical vapour deposition; finally, the Bi2Te3 thin film is integrated into the solder bump using standard pick-and-place production equipment.
Nextreme has used thermal bumping to fabricate thin-film devices that are, in comparison with conventional technology, roughly six times smaller in the x-y dimensions and 18 times smaller in the z dimension. On a volume basis Nextreme’s thin-film TEC is about 110 times smaller than conventional TECs. The smallest devices can even fit inside a TO-56 can. Measuring just 5.6 mm wide, the TO-56 can represents the next step in shrinking optical transmitters.
This is a win–win situation. Not only can smaller TECs be placed closer to the source of the heat so that they operate more efficiently, they can also extract more heat overall. The amount of heat that can be extracted is inversely proportional to the material thickness. Bulk TECs typically pump 10–13 W/cm2, while Nextreme’s devices can pump up to 10 times as much.
Modules can be fabricated using different manufacturing processes to achieve different performance characteristics. The Nextreme UPF process is based on a 238 μm diameter bump to deliver a relatively low-voltage, high-current specification. The Nextreme HV process is based on a 100 μm bump to deliver a high-voltage, low-current specification that is ideal for low-power applications.
These thin-film versions of TECs offer the scalability required in an industry driven by size and cost. The devices have passed initial reliability testing and are currently entering into full GR-468 reliability testing. Nextreme has previously sampled engineering prototypes and is in pilot production.