Correcting for Thickness in Micro X-Ray Fluorescence

Abstract

Correcting for Thickness in Micro X-Ray Fluorescence

Sarah Guinn

Office of Science, Science Undergraduate Laboratory Internship Program

Austin College, Sherman, TX

SLAC National Accelerator Laboratory

Menlo Park, California

July 24, 2012

Prepared in partial fulfillment of the requirement of the Office of Science, Department of

Energy’s Science Undergraduate Laboratory Internship under the direction of Sam Webb in the Stanford Synchrotron Radiation Lightsource at SLAC National Accelerator Laboratory.

Introduction

Micro X-ray fluorescence (μXRF) has developed over the years due to the use of synchrotron radiation, such as that which is used at the Stanford Synchrotron Radiation Lightsource (SSRL) at the SLAC National Accelerator Laboratory. μXRF Spectroscopy is used to make images of the samples’ chemical composition. This ability to know not only what metals are present in samples, but also where they are present is a very important and helpful tool. This research is particularly interesting because it has such broad implications ranging from environmental and medical fields to archeological and cultural heritage fields due to its basically non-destructive nature1. As an example, this can be helpful when trying to look at how arsenic is stored in a contaminated rice plant2. Currently at Beam Line 2-3 at the SSRL at SLAC, there are experiments being conducted using synchrotron-based hard X-rays in this variety of fields.

This technology is very helpful; however, creating a quantification procedure for the spectroscopy can be quite difficult. While quantification for thin samples has been established, corrections must be made for thicker samples because of the absorption of X-rays as they pass through the sample. The focus of this research is forming and testing correction factors for quantification of thick samples.

Methods and Techniques

The first approach at creating a correction factor for quantification is made assuming a single, homogeneous material with a defined thickness and linear mass attenuation coefficient, as well as an angle of incidence set as 45º. This particular situation will be helpful for research in biological medical research where more homogeneous material is studied.Figure 1.1 [RS1][RS2]shows an example of a sample where the incident beam hits the sample at 45º and then fluoresces. Length D will be the thickness used in calculations because it is the thickness that the beam is passing through, while d (equal to D times the square root of 2)is the actual thickness of the samples.

After a correction factor is established for this situation, the next step is to create a more general equation for material with multiple layers of different material that is studied using a confocal set-up for 3D imaging. Confocal μXRF for 3D imaging is an exciting branch of micro X-ray fluorescence that uses a second X-ray optical element placed in the detection channel to filter the X-rays that reach the detector allowing for a 3D image to be created of the object being studied. Mantouvalou et al. have developed a reconstruction algorithm for stratified materials3. The work this summer is to test and develop this for use at SSRL. If time permits, this work will also include developing an equation to find correction factors as the angle of incidence changes.

Figure 1.2 Schematic [RS3]sketch of the confocal set-up. 3

After finding literature on the procedures already in place for quantification with considerations for thicker samples, the sensitivity of each part of the equation is tested for sensitivity [RS4]and the equations are adjusted for the specific approach for this research. To test the validity of these correction factors, tests are run at Beam Lines 2-3 using gelatin molds with a specific known concentration of metal ions.

A ~25 mL stock solution is made four different salts (Cu, Zn, KBr, KCl) and water with ~1M. ~10 mL of the stock solution is mixed with gelatin powder and hot water to form ~40 mL gelatin solutions with ~.25 M. These bulk gelatin samples are then mixed in 11 different ~12 mL samples. This includes a mixture that contains Copper and Zinc, Copper and Potassium Bromide, Zinc and Potassium Chloride, Zinc and Potassium Bromide, and a mixture of all four gelatins.A a sample of Potassium Chloride with Zinc sporadically swirled through but not mixed is also made to model a brain-like sample because brain samples are frequently examined using μXRF Spectroscopy. Along with the 12 mL samples, washers are used as molds of ~1 mm thickness and ~1 cm diameter to create another set of samples. The thicknesses of these molded samples are smaller after the samples solidify and are used as thinner samples.The ~12 mL samples are also put in washer molds once the gelatin solidifies and 5μm film is used to protect the samples and keep the samples compacted in a space with a width of ~1 cm and a thickness of ~1 mm. These are used as thicker samples.Data is taken as all samples are run at Beamline 2-3 with a regular set-up and a few are run at 10-2 with a confocal set-up.

Using the data summary tool from Sam’s Microprobe Analysis Kit (SMAK) and Beer-Lambert’s Equation rearranged in equations (1) and (2), the mean intensity before the sample (I0) and the mean intensity after the sample (I1) are used to determine the actual thickness of each sample. The symbol λin equations (1) and (2) represents the absorption length which in this case is 4500 μm in brain matter ( C1.4HO7.1N0.2) at an energy of 13,450 eV. Brain matter is used for this calculation because gelatin is of a similar chemical composition and the absorption length is known.D represents the thickness of the sample that the beam passes through (depicted in Figure 1.1).

1. Malzer, W.; Kanngiesser, B., Spectrochim. Acta, Part B, 60, 1334-1341(2005)
2. Seyfferth, A.; Webb, S.; Andrews, J.; Fendorf, S., Environ. Sci. Technol. 44, 8108 – 8113 (2010)
3. Mantouvalou, I.; Malzer, W.; Schaumann, I.; Lühl, L.; Dargel, R.; Vogt, C.; Kanngießer, B., Anal. Chem. 80, 819 (2008)

[RS1]Figures labelded consectively 1,2,3,...

[RS2]Figure should be bigger so thaat it can be read.

[RS3]Bigger figure so we can see it

[RS4]Too many sensitivity