Comparison of B2O3 and BN deposited by atomic layer deposition for forming ultrashallow dopant regions by solid state diffusion

In this study, the authors investigated atomic layer deposition (ALD) of B2O3 and BN for conformal, ultrashallow B doping applications and compared the effect of dopant-containing overlayers on sheet resistance (R-s) and B profiles for both types of films subjected to a drive-in thermal anneal. For the deposition of B2O3, tris(dimethylamido) borane and O-3 were used as coreactants and for the deposition of BN, BCl3 and NH3 were used as coreactants. Due to the extreme air instability of B2O3 films, physical analysis was performed on B2O3 films, which were capped in-situ with similar to 30 angstrom ALD grown Al2O3 layers. For the BN films, in-situ ALD grown Si3N4 capping layers (similar to 30 angstrom) were used for comparison. From spectroscopic ellipsometry, a thickness decrease was observed after 1000 degrees C, 30 s anneal for the B2O3 containing stack with 60 ALD cycles of B2O3, whereas the BN containing stacks showed negligible thickness decrease after the annealing step, regardless of the number of BN cycles tested. The postanneal reduction in film thickness as well as decrease in R-s for the B2O3 containing stack suggests that the solid state diffusion dopant mechanism is effective, whereas for the BN containing stacks this phenomenon seems to be suppressed. Further clarification of the effectiveness of the B2O3 containing layer compared to the film stacks with BN was evidenced in backside secondary ion mass spectrometry profiling of B atoms. Thus, B2O3 formed by an ALD process and subsequently capped in-situ followed by a drive-in anneal offers promise as a dopant source for ultrashallow doping, whereas the same method using BN seems ineffective. An integrated approach for B2O3 deposition and annealing on a clustered tool also demonstrated controllable R-s reduction without the use of a capping layer. (C) 2015 American Vacuum Society.