Contamination is avoided through:
high temperature outgassing of furnaces during bakeout
high temperature outgassing of substrate heater during bakeout
After baking the machine, furnaces are operated approximately
10% above their highest anticipated operating point for 18-24 hours
to remove the last traces of impurities that might be likely to outgas
and contaminate epitaxial wafers. The background impurity density in GaAs
is then inferred from 77K Hall mobility measurements made on a wide spacer
2 dimensional electron gas (2DEG) structure. An undoped AlGaAs spacer
150 Angstroms thick is used to create a 2DEG which will have a mobility that
is predominantly limited by scattering due to unintentional impurities.
The structure is shown below.
This structure has been in use for many years to detect the presence of
unanticipated contamination of the MBE machine. This baseline characterization
is performed each time the machine is baked. The results which have been
obtained following bakeouts over a several year period are shown above. There
are 2 points omitted where the mobility was found to be lower than
100,000 as a result of contamination, or leaks, which then had to identified
and removed.
It can be seen that there has been a long term increase in the mobilities measured just after bakeout. This is likely to be the result of:
A jump in material purity occured in 1988 when the substrate heater was baked during bakeout for the first time. Previously, it was not baked because it was not originally wired with bakeable heater and thermocouple cables. Approximately 2,500 wafers were grown prior to this change in procedure. Since then, more than 2,500 have been grown including modulation doped field effect transistors (MODFET)s and quantum well lasers which have demonstrated record high frequency performance. Both classes of devices require low unintentional impurity concentrations.
The same baseline structure above has been grown following bakeout procedures at two other labs. One lab measured a mobility of 145,000 in a turbomolecular pumped system, while the other lab measured only 90,000-110,000. The second lab did not follow the bakeout procedure outlined above.
The mobility that is measured 30-60 days after the initial background check will be almost 200,000 at 77K. When the machine exhibits this level of purity, a 300 Angstrom spacer structure with half as much doping has been shown to exhibit mobility of 2,000,000 at 50mK.
February 20, 1977
Please send your comments and suggestions to: schaff@iiiv.tn.cornell.edu.