Errors in near-surface and interfacial profiling of boron and arsenic

W. Vandervorst, T. Janssens, B. Brijs, T. Conard, C. Huyghebaert, J. Frühauf, A. Bergmaier, G. Dollinger, T. Buyuklimanli, J. A. VandenBerg, K. Kimura

Research output: Contribution to journalArticlepeer-review

46 Citations (Scopus)


To get an insight in the diffusion behavior of dopants such as arsenic and boron after annealing and in particular their segregation characteristics towards the interface in oxide structures on silicon, it is necessary to probe the dopant profile with very high accuracy and depth resolution. Sputter depth profiling as employed in secondary ion mass spectrometry (SIMS) is frequently used as the most suited tool for dopant profiling in view of its sensitivity and depth resolution. However in order to determine the segregated arsenic/boron peak, sub-nanometer depth resolution is required and artifacts such as beam induced broadening effects, potential ionization yield changes at interfaces, transient sputter yields need to be considered in detail. When reducing the primary beam energy the depth resolution can be improved and sub-nanometer depth sensitivity can be demonstrated. However comparisons with high resolution elastic recoil detection analysis demonstrate that it is at present impossible to obtain a reliable depth profile in the first nanometer near the surface nor even in the oxide part of the profile, where no ionization nor sputter yield transients are expected. Enhanced beam induced migration of boron during the initial phase of the bombardment needs to be invoked to explain the results. The latter appears to be enhanced under conditions (∼normal incidence) where full oxidation occurs.

Original languageEnglish
Pages (from-to)618-631
Number of pages14
JournalApplied Surface Science
Early online date24 May 2004
Publication statusPublished - 15 Jun 2004
Externally publishedYes


Dive into the research topics of 'Errors in near-surface and interfacial profiling of boron and arsenic'. Together they form a unique fingerprint.

Cite this