Growth related interference effects in band edge thermometry of semiconductors

During epitaxial growth on GaAs and InP substrates two effects were observed that can interfere with correct substrate temperature measurements when using band edge thermometry. The first effect was due to specular reflection from a hot effusion cell opposite the optical detector. The worst case observed was a false temperature drop of similar to 30 degrees C upon starting growth. This effect was eliminated by moving the detector to a port that doesn't allow specular reflection from any cell. The second effect is due to the development of constructive and destructive interference fringes while growing heterostructures. False temperature oscillations as large as +/- 5 degrees C were seen when growing a thick AlGaAs/GaAs heterostructure on GaAs. The solution to reducing the false T-S oscillations is the use of a third derivative of the spectra, making the location of the band edge much less susceptible to interference effects. This technique reduced the false oscillations to +/- 1.5 degrees C. However, the second interference effect also opens up new possibilities for adapting band edge thermometry to extracting real time growth rate and composition information by monitoring oscillations at several fixed wavelengths both above and below the band edge. (c) 2005 American Vacuum Society.