Item

Abstract

The concept of "tuning" semiconductor heterojunction band offsets, by inserting a very thin layer of foreign atoms at the interface which acts as an "interface dipole", has received considerable attention. Modifications of the apparent valence-band offset, as measured by photoelectron spectroscopy (PES), have been indeed observed upon Si insertion at GaAs-AlAs interfaces, and have been interpreted as real band-offset changes. However, there is an alternative explanation of the results in terms of band-bending effects. Here, we present results of PES experiments designed to test these two interpretations. We have examined the effect of Si insertion at polar (1 0 0) and nonpolar (1 10) interfaces, and we have studied the insertion of Si (n-type) and Be (p-type) intralayers. Similar results are obtained for polar and nonpolar interfaces, and effects of opposite sign are observed for Si and Be intralayers. These results can be readily interpreted in terms of a band-bending profile modification upon Si or Be insertion. We analyse the implicit assumptions often made when using photoemission for band-offset determination, and the consequences of neglecting them. Photoemission experiments performed at different substrate temperatures, which make use of the surface photovoltage induced by the incident photons, permit a test of our proposed band profiles. From these data, we obtain evidence for a sample band bending, which is consistent with the room-temperature band profiles proposed. Hence, our results can be completely understood within a band-bending interpretation, without need to invoke the interpretation of intralayer action in terms of a "band-offset tuning" effect.