X-Ray Fluorescence Spectroscopy (XPS)

Submitted To: Dr. Muhammad Nadeem Badani

Submitted By: M. Ishtiaq Anjum (2018-UET/MS-PHY-15)

Syed Danish Hussain Rizvi (2018-UET/MS-PHY-10)

Introduction
X-Rays were discovered in 1895 by the German scientist, Wilhelm Conrad Roentgen. This
discovery opened doors for the development of X-Ray Fluorescence (XRF) spectroscopy which
has now become a powerful and versatile technique for the analysis and characterization of
materials.X-ray fluorescence is a well-established and powerful tool fornondestructive elemental
analysis of virtually any material.
XRF has been applied during the 1970s to 1990s as a versatile tool to many analytical problems.
The analysis of major, minor and trace elements in various kinds of samples can be performed
qualitatively as well as quantitatively. It is widely used for environmental, industrial,
pharmaceutical, forensic, and scientific research applications to determine the presence or
absence and in some cases to measure the concentration of elemental constituents or
contaminants.

Theoretical Background

X-Rays:
X-rays can be seen as electromagnetic waves with their associated wavelengths, or as beams of
photons with associated energies. The wavelengths of x-rays are in the range from 0.01 to 10 nm,
which corresponds to energies in the range from 0.125 to 125 KeV. The wavelength of x-rays is
inversely proportional to its energy, according to E*λ = hc. E is the energy in kev and λ the
wavelength in nm.

Interaction of X-rays with Matter:

There are three main interactions when X-ray contact matter:
i. Fluorescence ii. Compton Scatter iii. Rayleigh Scatter

If a beam of X-ray photons is directed towards a slab of material a fraction will be transmitted
through, a fraction is absorbed (producing fluorescent radiation) and a fraction is scattered back.
Scattering can occur with a loss of energy and without a loss of energy. The first is known as
Compton scatter and the second is known as Rayleigh scatter. The fluorescence and the scatter
depend on the thickness (d), density (ρ) and composition of the material, and on the energy of the
x-rays. The following figure shows the three phenomenon which occur when X-rays interact
with matter.

X-ray Fluorescence:

X ray fluorescence is the emission of characteristics or secondary x rays from a material that has
been excited by bombarding with high energy electrons, or gamma ray photon.

Basic principle:

XRF is based on a fundamental principle that individual atoms, when excited by external energy
source, emit X-ray photons of a characteristics energy or wavelength. By estimating the energy
of photon and counting the number of photon of individual energy emitted from a sample, the
elements present may be identified and quantified.

Theory:

When an electron beam of high energy strikes a material, one of the results of the interaction is
the emission of photons which have a broad continuum of energies. This radiation,
called bremsstrahlung, or “braking radiation”, is the result of the deceleration of the electrons
inside the material. The bremsstrahlung continuum is illustrated as a function of electron
acceleration voltages for a molybdenum target.

Another result of the interaction between the electron beam and the material is the ejection of
photoelectrons from the inner shells of the atoms making up the material. These photoelectrons
leave with a kinetic energy (E-φ) which is the difference in energy between that of the incident
particle (E) and the binding energy (φ) of the atomic electron. This ejected electron leaves a
“hole” in the electronic structure of the atom, and after a brief period, the atomic electrons
rearrange, with an electron from a higher energy shell filling the vacancy. By way of this
relaxation the atom undergoes fluorescence, or the emission of an X-ray photon whose energy is
equal to the difference in energies of the initial and final states.

INSTRUMENTATION:

Most of the XRF instruments in use today fall into two categories: energy-dispersive (ED) and
wavelength-dispersive (WD) spectrometers. Within these two categories is a tremendous variety
of differing configurations, X-ray sources and optics, and detector technologies. This paper will
cover the most common types of instruments, with forays into the more advanced or specialized
components where they are of interest.

A diagram of a WD system is shown in Figure 2. The instrument operates based on the principle
of Bragg diffraction of a collimated X-ray beam, in this case the beam emanating from the
sample. A detector is angularly scanned relative to the analyzing crystal, registering the
spectrum.

Figure 2: Diagram of a wavelength-dispersive spectrometer.

A typical ED spectrometer is diagrammed in Figure 3. Here the entire polychromatic spectrum

from the sample is incident upon a detector that is capable of registering the energy of each
photon that strikes it. The detector electronics and data system then build the X-ray spectrum as a
histogram, with number of counts versus energy.
X-Ray Source
The “front end” of both types of instruments is the X-ray source. The source consists of an
evacuated chamber with a heated cathode, which is usually a tungsten wire, and an anode, which
is held at a potential difference of several tens of kilovolts relative to the cathode. Thermal
electrons are released from the cathode and accelerated toward the anode. When the electron
beam impinges upon the anode, bremsstrahlung radiation as well as X-ray lines characteristic of
the anode material are emitted. These photons escape through a beryllium window built into the
side of the tube.

There is no one-size-fits-all approach for source selection in XRF. X-ray tube powers may be set
up at very different levels, from a fraction of a watt for EDXRF instruments with high detection
efficiencies to several kilowatts for WDXRF instruments. In this latter case, the tube must be
liquid-cooled since the majority of the power is dissipated as heat. The anode materials must be
carefully chosen as well, since the wavelength of their characteristic lines is important for proper
excitation of the sample. Some example single-element anode materials are aluminum,
chromium, tungsten, palladium, or gold. For detection of light elements, a high intensity of low
energy, i.e. 1-10 keV, radiation must be available, while heavy elements require excitation at
higher energies up to 50 keV (Jenkins 1995: 43-47, Skoog 1998: 274). It is also important to
keep in mind that the primary source of detector background will be the intense primary
radiation from the tube, above which the secondary sample radiation must be detected. The use
of secondary targets, or filters, can greatly reduce the background and improve sensitivity for
specific portions of the spectrum.

For instruments that are designed to acquire the entire spectrum with good sensitivity on light as
well as heavy elements, a different approach is taken. A tube anode material is chosen to give a
high bremsstrahlung or continuum output, which is used to excite a secondary fluorescence,
or target, which gives off its own characteristic lines without the continuum. The sample is then
excited by the emission from the target, which is chosen to efficiently excite elements in a
certain Z range. A system may be set up to change targets automatically during the analysis so
that the low, middle, and high end of the sample spectrum may be sequentially boosted. When an
EDXRF instrument uses such a system, tube powers must be increased dramatically since most
of the original X-ray intensity is lost.

Certainly XRF excitation is not limited only to X-ray tubes. Less common excitation sources
include gamma-emitting radioisotopes (241Am, 109Cd, 153Gd, and others), electron sources
where the sample is the tube anode, and synchrotrons, which produce highly intense, coherent,
monochromatic X-ray beams (Jenkins 1988: 56, Jenkins 2000).

Sample Chamber
Up to this point, little has been said regarding what kinds of samples may be analyzed by XRF.
The development of portable XRF (PXRF) instruments has greatly expanded the range of
samples suitable for analysis. There is no longer a need to fit a sample into a small chamber. In
the case of PXRF, it is possible to analyze the samples with the instrument in a stand or the
instrument can be moved to the sample, as in the case of analyzing a exposed rock outcrop or a
large painting. The majority of the samples analyzed by XRF at MURR are small obsidian
artifacts that are placed on the nose of the instrument. We have recently acquired an automated
sample changer that allows us to load up to 20 samples at a time and operated much like the
sample changers on large lab-based instruments.

Detector Systems
The two main types of XRF spectrometers (WD and ED) differ completely in their detection
systems. EDXRF systems depend on semiconductor-type detectors which receive the entire
emitted spectrum from the sample and decode it into a histogram of number of counts versus
photon energy. WDXRF spectrometers, however, use an analyzing crystal to disperse the emitted
photons based on their wavelength and place the detector in the correct physical location to
receive X-rays of a given energy.

Wavelength-Dispersive Spectrometers
Collimators, usually made from a series of closely spaced parallel metal plates, are needed to
direct the beam in order to closely control the diffraction angle of all detected photons. The
analyzing crystal angularly disperses incident radiation of wavelength l according to Bragg’s
Law (nλ = 2d sinΘ, n = 1, 2, 3,… where d is equal to the crystal lattice spacing). In the
instrument shown, the analyzing crystal may be rotated with the detector assembly
simultaneously revolving around it to scan through the possible wavelengths. To resolve
wavelengths in all regions, different crystals must be used, since crystals with large spacing must
be used for long wavelengths but they make the short wavelengths irresolvable at low q.

The system in the diagram utilizes two detectors in series. The first, a gas-flow proportional
detector, is efficient for detecting long-wavelength radiation (> 0.15 nm). Most high-energy X-
rays pass through it, however, and are counted by the NaI(Tl) scintillation detector. The gas-flow
proportional detector works by placing a high voltage across a volume of gas (usually Ar with
methane). An X-ray photon will ionize a number of Ar atoms proportional to its energy. The
freed electrons are accelerated in the high voltage, ionizing other Ar atoms and creating an
electron cascade which is controlled by the quench gas methane. The freed charges are measured
in the circuitry as a voltage pulse whose height is proportional to the energy of the photon that
initiated the cascade.

An NaI(Tl) detector contains a large single crystal of sodium iodide that has been doped with
thallium. This crystal is sealed from light by a Be window. When an X-ray photon enters the
crystal, it places primarily the I atoms in an excited state, in numbers again proportional to its
energy. These excited states decay exponentially with time, giving off a flash of light or
scintillation when they go. The summed intensity of light strikes a photocathode, which releases
photoelectrons that are amplified in a discrete dynode detector. The pulse height measured from
this detector is proportional to the energy of the original X-ray photon (Jenkins 1995: 96, Knoll
2000).

One may wonder why these detectors need to have any energy resolution at all, since the X-ray
energies are supposed to be dispersed by the Bragg crystal. In order to improve the performance
of the spectrometer, it is useful to be able to discard X-rays from higher order (n>1) diffraction,
hence the need to resolve out photons of different energies.

With WDXRF systems, it may be possible to have several detector assemblies placed at fixed
angular locations in order to analyze for a few selected elements over and over. WDXRF
spectrometers often offer more flexibility for the researcher as well as very good sensitivities.
The detector outputs are also simpler to use directly and do not generally require heavy use of
electronics and computer algorithms in order to deconvolute. Disadvantages include the inability
to quickly acquire the entire X-ray spectrum for full-element analyses, higher hardware costs,
and a larger instrumental footprint when compared to EDXRF systems.

Energy-Dispersive Spectrometers
While simpler in terms of the positioning of the detector versus the sample, EDXRF
spectrometers require sophisticated electronics and computer software in order to interpret the
detector output. Nowadays this is less complicated, though, due to important technological
advances in multichannel analyzers and faster computers, and EDXRF is often the technique of
choice for fast multi-element analyses.
Dispersion in EDXRF is accomplished through the use of a semiconductor-type detector (Knoll
2000). Although germanium detectors are utilized, the most common type in service is the
Si(Li), or lithium-drifted silicon, detector. A semiconductor detector operates based on the
principle that an X-ray photon incident upon the diode material will give up its energy to form
electron-hole pairs, the number of which is proportional to the energy of the photon. The high
voltage applied across the diode quickly collects the released charge on a feedback capacitor, and
the resulting proportional voltage pulse amplified by a charge-sensitive preamplifier. The output
of the preamp is fed to a main amplifier system. The pileup rejector, part of this system, deals
with the probable event that two pulses will arrive very close together in time. From this point,
the pulse is converted to a digital signal and processed in the multichannel analyzer (MCA)
(Jenkins 1995). In the MCA, dead time, caused by high counting rates, must be corrected. Peaks
in the energy spectrum, once acquired, are subject to a large degree of massaging by the software
in the connected computer. Sophisticated algorithms sense and quantitatively correct for high
backgrounds due to Compton scattering from low atomic number matrices (Metz 1994).
Spectrometers that use secondary targets may acquire several energy spectra for each sample,
one from each target. Since each target yields better sensitivity in one part of the spectrum, the
information from the energy spectra is combined to quantitate each element being analyzed.
Accurate quantitative data on the entire mass spectrum may be obtained in a matter of minutes
using EDXRF. However, in general the detection limits are higher than those for WDXRF.

For both of the Bruker Tracer instruments we use we have incorporated a secondary target made
of thin sheets of copper, aluminum and titanium to optimize the spectra for the analysis of
obsidian and any other analyses focusing on elements with fluorescent energies between about
10 and 20 kV.

We have developed a world-renowned set of obsidian calibration standards that we have used to
calibrate our own instruments and Bruker now runs this calibration on all portable XRF
instruments heading out to museums and archaeologists. With this calibrations it is possible to
acquire quantitative concentrations for many elements that are comparable to data acquired by
more costly and destructive neutron activation analysis (NAA)

Example Instrument: The Bruker Tracer III-SD

The primary source of X-rays for the Tracer-III is an X-ray tube with a Rh exit window. The
tube voltage can be varied up to 45 kV, although we generally analyze the obsidian with a setting
of 40kV. The secondary target, or filter, primarily used includes a 6 mil thick sheet of copper
(used to block X-rays below about 20kV) a 2 mil sheet of titanium (added to remove the
secondary copper X-rays) and a 12 mil sheet of aluminum (to absorb the titanium X-rays).

The newly developed sample changer (designed by Dewitt Systems) has a carousel that holds 20
samples . This sample changer has greatly increased our analytical efficiency, allowing us to
analyze up to 800 samples per week. An example sample spectrum is presented in Figure 4.

Sample Preparation for XRF

Sample preparation is highly variable depending on the matrix and goals of the analysis. Most of
the materials we analyze (obsidian, metals, and ceramic paints) do not require any sample
preparation. The choice of sample preparation depends on the nature of the X-ray beam relative
to the sample. For example, a piece of obsidian that is 1 cm thick and has a clean, flat surface
will provide ideal results. As sample sget smaller, thinner, or less homogenous it is necessary to
understand the nature of the X-ray beam and how it interacts with the sample.

The Bruker Tracer III instruments have a beam size of only about 2 x 3 mm. This small beam is
fine for homogenous materials, but heterogenous material such as crystalline rocks and tempered
pottery may need to be analyzed multiple times in numerous areas to generate a representative
average composition. The small beam size is ideal for isolating specific painted elements on the
surface of ceramics and also aids in the analysis of very small obsidian artifacts.

Perhaps even more important than the area of the beam is the depth of analysis. As a general
rule, the higher up the energy spectrum, the greater the depth of X-ray penetration in the sample.
For example, the analysis of iron (6.4 kV) in obsidian is primarily a surface analysis while
zirconium (15.78 kV) is measured from up to almost a centimeter deep. In thick homogenous
samples this depth of analysis makes little difference, but if samples are thinner, it effects to
resulting spectrum in different ways depending on the specific sample thickness and particular
element of interest. Ferguson (in press) addresses a number of approaches to quantitative
analysis of thin samples.

The ability to analyze samples without destructive sample preparation procedures has been a
great advancement for archaeologists. We can now analyze large and valuable artifact
assemblages that would have been off-limits to destructive proceedures. However, for non-
archaeological applications of XRFthe most common method of sample prep is pelletizing,
which can be made to work for most matrices that can be ground into an homogeneous powder,
including soil, minerals, and dried organic matrices such as tissues or leaves. Difficult grinding is
accomplished with a hard agate mortar and pestle but many samples can be adequately
homogenized by placing into a hard plastic vial, adding a plastic mixing ball, and violently
shaking in a mixer mill. A powdery binder containing cellulose, starch, polyvinyl alcohol or
other organics is usually weighed in and blended thoroughly with the sample, and the resulting
mixture added to a deformable aluminum cup. (Buhrke p. 39) The sample and cup are pressed in
a cylindrical die to form a supported pellet which ideally has a smooth, homogeneous sample
surface and good physical stability.

Here particle size and homogeneity play a big factor. The thin surface layer must be
representative of the bulk sample, and studies have shown that large errors (up to 50%) in
quantitation can occur for some elements when particle sizes exceed 50 µm. This is due to the
variance in X-ray penetration depths with energy. Particles may be inhomogeneous also, having
a different surface composition than their bulk. For example, copper sulfides may become
partially oxidized at the surface, causing the relative absorption for Cu K lines to differ from that
of the L lines. The L line photons will not penetrate as deeply and will tend to be emitted more
from the oxide layer. By monitoring the ratio of the Cu Kα line to the Cu Lα, the analyst can use
this phenomenon to an advantage, allowing him or her to deduce some of the chemistry of the
sample.

One way to get around sample grinding is to fuse the sample at high temperatures with sodium or
lithium tetraborate and then to pour this glass-like mixture into a mold. Chemical reactions occur
within the melt which dissolve particles and create a homogeneous liquid that hardens upon
cooling. The disadvantages to this technique include the additional time to prepare the melt and
the possibility of the sample reacting with even inert crucible materials such as platinum.

Homogeneous solid samples such as metals may be machined and smoothed to form disks.
Whatever type of preparation is done, the surface roughness of the sample should be taken into
account. A rough surface causes the penetration layer to look heterogeneous to the spectrometer.
In general, the surface roughness (measured in µm) should not exceed the path length for 10%
absorption of the radiation being measured.

Applications of XRF Spectrometry

Currently XRF spectrometry is very widely applied in many industries and scientific fields. The steel and
cement industries routinely utilize XRF devices for material development tasks and quality control.
(Anzelmo 1987 Part 1) NIST utilizes XRF as one technique to quantitatively analyze and acceptance-test
many of its standard reference materials (SRMs), from spectrometric solutions to diesel fuel to coal to
metal alloys (Sieber 2000). The plastics industry is looking at a modified XRF spectrometer as an on-line
wear monitor, taking advantage of its ability to detect particles of worn-off metal in extruded plastic
pieces (Metz 1994). Polish scientists are accomplishing XRF analyses on very thin films by placing the
source and detector at very low angles with respect to the sample. This technique is being applied to trace
element determinations in water samples that have been evaporated to a thin film of residue (Holynska
1998).

XRF has been one of the tools of choice for geologists for many years, so much so that graduating
geologists usually receive practical training with these devices, whereas graduating chemists probably
haven't even heard of the technique. For geologists, the ability to determine major and trace components
in one quick analysis with relatively little sample preparation has been a boon (Anzelmo 1987 Part 1, Part
2). Current basic research aimed at improving XRF analyses for geological and ecological samples
focuses on methods for correcting for matrix effects, in which major components absorb some of the X-
rays emitted from trace components (Revenko 2002).

Archaeometrists have applied XRF in order to solve their ancient mysteries. An example of this was the
study of the composition of blue soda glass from York Minster, England, which distinguished three
compositional groups, indicating this number of possible sources for the glass. Trace metal signatures
also can effectively differentiate genuine artifacts from modern copies (Jenkins 2000).

Forensic scientists utilize XRF spectrometry to match samples associated with suspects (i.e. dirt or sand
on clothing or shoes) to samples from crime scenes. As for other applications, here XRF can help
elucidate an elemental fingerprint, without need to analyze the evidence destructively (Jenkins 2000).