The implementation of a DSSSD in the upgraded boron analysis at LIBAF for applications in geochemistry

Interest in high spatial resolution boron analyses from a geochemical perspective is related to the recognition of boron as an important tracer of chemical recycling in the Earth, due to the high solubility of boron in aqueous fluids and silicate melts. Although boron is not a nominal component in common silicates they may still contain significant B-concentrations and hence constitute important boron reservoirs in the deeper parts of the Earth. Boron analyses have been performed at the Lund Ion Beam Analytical Facility for almost 20 years. For the analysis the nuclear reaction p+B is used with beam energy just below 700 keV where the reaction has a broad resonance. In this paper we describe an upgrade of the system with a double sided silicon strip detector, which allows for much higher count rates compared to the old annular surface detector based system. A gain close to 20 in the data rate allows for high resolution mapping of boron distributions in crystals. This is illustrated by a number of examples. In addition, the detection limits for boron in geological samples are improved, now around 5 ppmw. In this work we address issues with data quality, especially charge normalization, lifetime correction and subtraction of different background components. 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/).


Introduction
Boron and boron related analyses have been performed at LIBAF (Lund Ion Beam Analysis Facility) for almost 20 years. For the analysis the nuclear reaction p+ 11 B is used with beam energy just below 700 keV where the reaction has a broad resonance. This reaction emits three alpha particles with energies much higher than the elastically scattered proton, which can easily be discriminated and counted as a function of beam charge. In the original set-up an annular surface barrier was used to detect the alpha particles 1,2 . Then an analysis currents of 100-500 pA could be used together with the focused ion beam. This kept the count rate below 10 kHz, which was considered acceptable for this kind of analysis. In this paper we will describe and evaluate the upgrade of the system from a single detector to a DSSSD (double-sided silicon strip detector) 3 .

Boron as a geochemical tracer
The interest in high-resolution boron analyses from a geochemical perspective is due to the recognition of boron as an important tracer of chemical recycling in the Earth, as a result of the high solubility of boron in aqueous fluids and silicate melts. Although boron is not a nominal component in common silicates (e.g. pyroxene, olivine, sillimanite) they may contain enhanced B-concentrations (≤1 wt% B 2 O 3 ) and hence constitute important boron reservoirs in the deeper parts of the Earth. The nature of B-substitution mechanisms in these minerals is not fully understood and partly contradictory models have been suggested. Isotopic fractionation of boron is strongly dependent on how boron is coordinated by its nearest anions 7 .

The Samples
In the experiments two sets of samples have been analyzed, one set mounted on an In-backing and one set on a epoxy backing. Each set had between 15 and 20 crystals and also one tourmaline crystal was included on each sample for internal normalization.

Conclusion
Ø New detector set-up tested for boron analysis -improved the system significantly. Ø Can be operated with very high count-rate -up to 200 kHz or higher. Ø Good detection limits -depends on sample composition and beam energy. Ø High resolution mapping possible with good result.

Results
Apart from analyzing two sets of crystals regarding boron concentration and distribution the performed experiment was intended as an evaluation of the upgraded system with the new DSSSD detector. In figure 1 energy spectra from analyzes of an tourmaline standard are shown. The upper curve is the raw spectrum taken with a current of 10 nA, the second curve (scaled a factor of 10) is the same set of data but gated on the multiplicity condition 1. The third curve is a spectrum taken with 1 nA and shows the same shape as the high current spectrum. Count/charge (a.u.)

Division of Nuclear Physics, Department of Physics, Lund University, Lund, Sweden 2 Swedish Museum of Natural History, Department of Geosciences, Stockholm, Sweden 3 Caltech, Division of Geological and Planetary Sciences, Pasadena, CA, USA
In figure 2 the energy spectra from a tourmaline are shown for two different beam energies (above and below a 18 O resonance). The difference in yield is about a factor of two. In figure 3 a background spectrum and one spectrum from a low boron concentration crystal are shown (X-X3 from the table). The solid line is a the tourmaline yield normalized to the boron in X-X3. The overlay of background on the distribution is clear. Here the background corresponds to roughly 150 ppm. The background will be lower below the 18 O resonance energy, but so will the yield.
An illustration of what can be achieved with a high resolution run with good statistics. The zonation pattern in the crystal is nicely shown in figure 4 . Figure 5 demonstrates another application of the scanning method. A fast scan gives the gross features of the crystal (a). An area is selected and the step is decreased and a high resolution map over the interesting part of the crystal is collected (b). Finally a line scan is performed with optimal resolution (c).
Quantitative analysis is the primary goal of the presented technique. In the table the different steps going from counts to ppm are illustrated for the epoxy sample. First a suitable area from a scan is selected and the number of B-counts is extracted from this area. The corresponding charge is extracted and the data is normalized. The background is subtracted and the concentration is normalized to the tourmaline standard.