SURFACE AND INTERFACE ANALYSIS Surf. Interface Anal. 2002; 33: 478–486 Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/sia.1235 Accurate RBS measurement of ion implant doses in silicon † G. Boudreault, 1* C. Jeynes, 1 E. Wendler, 2 A. Nejim, 1 R. P. Webb 1 and U. W ¨ atjen 3 1 University of Surrey Ion Beam Centre, Guildford, Surrey GU2 7XH, UK 2 Institut f ¨ ur Festk ¨ orperphysik, Friedrich-Schiller-Universit ¨ at, D-07743 Jena, Germany 3 EC-JRC, Institute for Reference Materials and Measurements (IRMM), B-2440 Geel, Belgium Received 9 November 2001; Revised 1 February 2002; Accepted 7 February 2002 We demonstrate very accurate ion implant dose measurements using Rutherford backscattering spectrometry (RBS) traceable to a certified reference material from IRMM, Geel and the Bundesanstalt f ¨ ur Materialforschung (BAM), Berlin. The measurements have an absolute accuracy of better than 1.4% and a precision of better than 1.25%. The certified standard sample is compared directly with recent absolute determinations of the energy loss of He in Si, and also with a sample calibrated against the Harwell Bi standard. We determine the dose in a series of three In implants and six As implants of various doses and energies. Some of the samples were amorphized to eliminate channelling effects. A double detector geometry was used, giving pairs of spectra with a common incident charge but where the solid angle and the electronic gain were determined for each detector channel independently. The statistical uncertainty is reduced to <1%. The non-linear pile-up background is determined carefully. The errors are determined critically. The experiments were carried out at different dates and different places, so time and space reproducibility of the results is confirmed. The IBA DataFurnace is used for analysis of the certified standard reference material and compared with a transparent manual data reduction method: the use of this code for routine data analysis at the highest accuracy is validated. Copyright  2002 John Wiley & Sons, Ltd. KEYWORDS: Rutherford backscattering spectrometry; ion implantation; dosimetry; standards; silicon energy loss INTRODUCTION Ion implantation processes are universally used in semi- conductor processing, and especially in the fabrication of integrated circuits on silicon. Rutherford backscattering spectrometry (RBS) is an important technique for charac- terizing the absolute implant dose because it is based on analytical cross-sections and because it does not have matrix- dependent sensitivity factors. Its disadvantage is that it is relatively insensitive because the cross-sections are quite low, and also it is only for heavy implants in a light substrate that the signal from the implant is relatively background-free. The technique is only one of a cluster of related ion beam analysis (IBA) techniques—these include non-Rutherford scattering, forward recoil spectrometry, nuclear reaction analysis and particle-induced x-ray emission—but of these only RBS has (nearly) analytical (i.e. Rutherford) cross-sections, and it is therefore RBS that we use for this standards work. In this work we are aiming to establish the instrumen- tation for dosimetry on our ion implanters at the 1% level L Correspondence to: G. Boudreault, University of Surrey Ion Beam Centre, Guildford, Surrey GU2 7XH, UK. E-mail: g.boudreault@surrey.ac.uk † Paper presented at: QSA-11, University of Surrey, UK, 3–7 July 2000. Contract/grant sponsor: EPSRC; Contract/grant number: GR/L 78512. for high-dose heavy implants. This is expected to be a worst case for the implanter instrumentation because secondary electron generation is reduced for light ions and low beam currents. Rutherford backscattering spectrometry is often cited as a technique with 1% accuracy: however, this depends on the accuracy of the knowledge of the product of the detector solid angle and the collected charge, i.e. the number of ions incident on the sample. We are aware of no critical reports of RBS systems with charge collection better than 1%, and measurements of solid angle at this accuracy are notoriously difficult. Therefore, accurate work must be validated by stan- dards for the charge ð solid angle product. The Harwell Bi implant standard has been determined absolutely and was quoted at 2% accuracy, 1 but the remaining samples from this standard should soon be re-released with individual certifi- cates based on high-precision RBS measurements relative to weighed evaporated films. 2 The same method together with two completely independent methods, namely instrumental neutron activation analysis (INAA) and inductively-coupled plasma isotope-dilution mass spectrometry (ICP-IDMS), are used to certify new Sb implants into Si substrates at about ¾0.6%, 3–5 and we use one of these reference samples (now designated IRMM-302/BAM-L001) in this work. The only critical RBS that we are aware of near 1% accuracy, apart from that by W¨ atjen & Bax, 2 is the Copyright  2002 John Wiley & Sons, Ltd.