Quality assurance in an implantation laboratory by high accuracy RBS

https://doi.org/10.1016/j.nimb.2006.03.171Get rights and content

Abstract

A time series of 5 × 1015 As/cm2 100 keV implantations into 100 mm silicon wafers has been analysed by RBS, using the amorphous Si yield to determine the charge solid-angle product for each spectrum. Wafers are also annealed and the resistivity characterised by the four point probe. The implants are demonstrated by four point probe to be uniform at 0.7%. Very high precision down to 0.5% in the determination of implanted average fluence is obtained by analysing the sum of RBS spectra obtained over the whole wafer, with total collected charges of up to about 1 mC. An accurate pileup correction does not contribute significantly to this uncertainty which is dominated by the electronic gain. A full uncertainty budget is presented. The absolute accuracy is still dominated by the uncertainty of the silicon stopping force (2%). The implanted fluence is demonstrated to be consistently within about 3% of nominal.

Introduction

It is a common misconception that because electrical currents down to pA can be measured very accurately with appropriate instrument calibration, and because implantation uses a charged particle beam, the implanted fluence is easy to determine from Faraday cup measurements. Unfortunately the electrical environment in an implantation chamber is complex. For a heavy ion such as As the number of secondary electrons generated can exceed the number of incident ions by orders of magnitude. Some of these electrons have high energy and are very hard to suppress. For these and other reasons quantitative implantation has remained elusive.

Rutherford backscattering spectrometry (RBS) is a standard method for absolute measurement of implanted fluence. Jeynes et al. have discussed accurate RBS in some detail [1]. Boudreault et al. have previously [2], [3] shown fluence measurements by RBS with an uncertainty of about 1.4%.1 We present here new results at unprecented precision, with a full uncertainty analysis [4].

Section snippets

Implantation

Implants were carried out on a 200 kV DF1090 [5] from Danfysik with electrostatic scanning (X: 1000 Hz, Y: 1004.5 Hz). For dosimetry we rely entirely on the performance of Faraday cups. This is a challenging application for them since the scanning beam strikes the Faraday cups only intermittently, and usually only partially. The Faraday cup assembly has four cups with reamed apertures each of diameter 0.282 ± 0.002 cm2. Each Faraday cup has two cylindrical electron suppressor electrodes to separately

Results

1.557 MeV 4He+ from the 2 MV Tandetron [6] was used to collect RBS spectra with two detectors A and B of scattering angles 172° (Cornell geometry) and 149° (IBM geometry) and solid angles 1.0 and 3.1 msr, respectively. The beam was channelled on (1 0 0), normal to the wafers, to reduce the count rate and guarantee the geometry. Typically 25 μC was collected on each of 24 spots distributed over the wafer with a beam current of 40 nA.

The scattering cross-section has a screening correction due to

Discussion and conclusions

The measurements are demonstrably self-consistent at the 0.5% level. The four point probe results demonstrate a combined standard uncertainty (including the annealing process) of about 0.6%, and indicate that the uniformity of the implantation is 0.7%.

The implanted fluence is measured by RBS with a combined standard uncertainty of just over 0.5% at best: this is dominated by the uncertainty in the electronic gain. There is a further uncertainty of about 2% in the value of the stopping force of

Acknowledgement

We greatly appreciate the help of the Ion Beam Centre staff: Mark Browton, Adrian Cansell and Alex Royle; and the support of the IBC by EPSRC under contract number GR/R50097/01.

References (12)

There are more references available in the full text version of this article.

Cited by (5)

  • "total IBA" - Where are we?

    2012, Nuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and Atoms
    Citation Excerpt :

    It should be mentioned that XRF is not an effective competitor technique for IBA in the measurement of aerosols (pace [2]) considering that an IBA measurement of a few minutes is really “total”, that is, it measures up to 80% of the total mass of the sample (using PIGE for light elements, BS for C and O and ERD for H) where XRF is only partial (and takes much longer). Moreover, the accuracy of benchtop XRF is intrinsically limited by the difficulty of characterising the spectrum of the X-ray tube, but IBA can easily be run at ∼2% precision (see [30,31]). There is a new approach to IBA which may prove remarkably valuable.

  • Ion beam analysis: A century of exploiting the electronic and nuclear structure of the atom for materials characterisation

    2012, Reviews of Accelerator Science and Technology: Accelerator Applications in Industry and the Environment
View full text