| Popis: |
Summary form only given. It has been pointed out by Henderson (1996) and others that HBTs are vulnerable to recombination-enhanced defect reactions (REDRs), which can lead to formation of crystal defects with activation energies (E/sub a/'s) as low as 0.35 eV. It is therefore possible that when conventional lifetests with junction temperatures from 200 to 300/spl deg/C are run, other high-E/sub a/ phenomena occur first, and the tests are subsequently stopped. This may have the result that the low-E/sub a/ phenomena are never observed. Thus these lifetests may fail to characterize the mechanism that will actually occur first at the low temperature of a typical application, and may therefore greatly overestimate the real reliability. Conventional lifetests have been run on the HRL Laboratories Generation 1 AlInAs/GaInAs HBTs, on InP substrates, with good results: the beryllium base dopant appears stable, there appear to be increases in the resistances of the emitter contacts, but there is no indication of increases in base currents, that would be a sign of defect generation. So in parallel with these tests, we have been running long-term lifetests at the very low junction temperatures of 125 and 160 C, in an attempt to detect low-Ea phenomena. This experiment was reported in detail when we were at about 14,000 hrs. But it has been continuing since then, and has now reached a power-on time of over 19,000 hrs (accumulated over an elapsed time of almost 3 years). We continue to see no failures, so we can now rule out low-E/sub a/ failures with a higher confidence level. To our knowledge, this is the only lifetest on HBTs of any composition, that has been run with temperatures well below 200 C. |
