In the last years Total Reflection X-Ray Fluorescence spectroscopy (TXRF) has shown to be an analytical technique for trace element analysis as well as surface quality control. Detection limits in the range of pg or ng/g (ppb) level in concentration values of aqueous solutions or 1011 -109 atoms/cm2 are achieved with new excitation sources. The range of detectable elements has been extended to low and high 2 elements. Instrumental developments and results of the Atominstitut as well as from some other working groups are presented.

The use of glancing angles of incidence enables X-ray Absorption Spectroscopy to be measured as a function of depth from the surface of a material into the bulk. As x-rays rather than photoelectrons are detected, a UHV environment is not required and instead surfaces and interfaces can be examined under realistic operational conditions. Whilst the reflected beam carries the fine structure of concentrated species in the imaginary part of the refractive index, this is obscured by the contribution from the real part for angles greater than ϕc, the critical angle for total external reflection. Measuring the x-ray fluorescence offers more flexibility, particularly for dilute systems. The use of synchrotron radiation in conjunction with a multi-element Solid State Detector enables impurity loadings down to a few 1019 cm-3 to be measured which for ion implants is equivalent to around half-monolayer coverage at the surface. This sensitivity makes it practical to examine impurities in semiconductors at realistic dopant levels.

Recently, external X-ray total reflection, or the grazing-incidence condition, is being widely used for the surface characterization in various research fields. Surface X-ray diffraction, or grazing-incidence diffraction/scattering, and total-reflection fluorescence XAFS (X-ray absorption fine structure) are typical, powerful techniques for surface characterization. X-ray fluorescence (XRF) analysis under the total-reflection condition has also been attracting much attention regarding analytical applications. Two types of XRF experiments have been carried out under the grazing-incidence condition. One is trace-element analysis of a dried solution sample or a semiconductor wafer; another is an analysis of the elemental concentration-profile in depth. Both experiments are closely related to each other, and are valuable and promising because of their non-destructive nature and the high precision of XRF analysis.

The angular dependence of Total-reflection X-Ray Fluorescence intensities can behave in various ways. The variety is discussed by examining the x-ray intensity distribution in the material under investigation. It is shown that for thin layers on solids, interference fringes are present due to x-ray standing waves. This phenomenon is exploited to determine the depth distribution of elements in layered specimens.

We have developed a technique to determine the dissolved solid components of water and other liquids by Energy- Dispersive X-Ray Fluorescence (EDXRF) spectrometry. The liquid samples are presented to the spectrometer as small dried spots of the residue remaining after evaporation of the liquid. The dried residues are mounted on thin plastic films cemented to 35 mm plastic slide mounts. Elements from sodium through uranium are detected with detection limits less than one nanogram in favorable cases. Precision of the measurement is better than 2 % relative in favorable cases for secondary-target excitation, and is typically 5 to 10 % for direct excitation given the geometry of the instrument. Detection limits are as low as a part per billion relative to the original liquid for solutions whose total dissolved solid content is 1000 part per million or less.

Critical to “Trace Analysis” measurement are the concepts of detection and quantification. Detection means positively identifying a chemical component. The term Minimum Detection Limit (MDL), used here, follows that of Currie, and is the smallest concentration that yields a net count, above system background, which will be detected with a 95% probability (with only a 5% probability of false detection).

X-ray fluorescence excited by a monochromatic collimated Mo-Kα beam at grazing incidence is measured as a function of the angle of incidence. Monochromatic excitation guarantees a well-defined penetration depth and enables a simple analytical description of the fluorescence intensity. This method is applied to a system of thin Cu and Ti metallization layers on a Si wafer and to As dopant concentration profiles in Si wafers.Thereby, the effect of annealing can be analyzed non-destructively.

Recently, semiconductors have been integrated and densified to much higher levels. Under this circumstance, control of surface impurity on the Si substrate has become more important. We have attempted to increase analytical sensitivity of the transition metals, which have the greatest influence on productivity.

For the last few years, x-ray fluorescence analysis, using total reflection as a non-destructive and speedy method, has been applied for surface impurity analysis of Si wafers. However, this method is insufficiently sensitive (to the level required for surface impurity analysis) and the lower detection limit is still 1011 to 1012 atoms/cm2 order because the continuous x-ray excitation causes lots of scattering which makes the background higher.

This paper deals with soft X-ray spectra of compounds which have one or more spectator hole(s) in the valence orbitals. The aim is to give a simple picture of the complicated chemical effects and to describe implications of the chemical effects to know the chemical state from the complicated spectra of materials. The first example of the system which has a spectator hole in the valence orbital is the intensity modifications of the shake-off satellites (KM-→LM) in chlorine Kα spectra. The second example of the system which has spectator holes in the valence orbitals is the multiplet structure of Lα spectra of the transition-metal compounds. It is described that the chemical effects of these two systems are clearly interpreted by the avoided crossing [1] of the molecular orbitals, one of which has a localized hole. We also describe the systems which have no spectator holes. The interpretation of the X-ray spectra of these systems are easy by a molecular-orbital calculation at the ground state.

In the field of material analysis and characterization interest has considerably shifted over the last few decades from bulk to surface and very thin film problems. At the present state a wide range of surface analytical techniques - such as x-ray photoelectron (XPS), Auger electron (AES), secondary ion mass (SIMS), ion scattering (ISS) spectroscopies - have become available but every one of them exhibits specific analytical features and information content. Within the context of this paper the main parameter to be considered is the information depth i.e the layer thickness from which the majority of information-bearing particles escape and hence are detected. For XPS and AES, this parameter is associated with the mean-free path of photoelectrans or Auger electrons and typically is in the range from 0.5 to 4 nm. In SIMS the ejected secondary ions are emitted from the outer 2 or 3 atomic layers (i.e. from about 1 nm) while the single-collision binary process occuring in ISS is restricted to atoms from the top most atomic layer (0.2-0.3 nm).

High resolution X-Ray Fluorescence Spectroscopy(HRXRFS) has been extensively used as a chemical bonding probe for both solids and gases for quite some time. However there have been virtually no HRXRFS studies reported on liquid phase systems and particularly on solution systems. This seems somewhat surprising not only because this is an unexplored area, but because much of chemistry and chemical bonding occurs in solution. Furthermore, since HRXRFS is a “photon in - photon out” technique, providing one uses relatively energetic x-rays, the experimental techniques for performing these types of measurements are relatively straightforward, especially when compared to techniques which utilize charged particles.

Over recent years, the availability of man made “crystals“ (Layered Synthetic Microstructures or Multilayers) has led to a significant extension of the capabilities of WD XRF for the lighter elements. In particular, the elements from Atomic Number 5 (Boron) to Atomic Number 12 (Magnesium) have benefitted.

This paper will cover the lower end of this range and show how selection of multilayers is made. It will then consider the optimization of collimation to suit The performance characteristics of the multilayer device. One of many new applications opened up by these advances will be covered in detail (analysis of Boron in BPSG coatings on Silicon Wafers). Finally, a word of warning will be issued to those seeking to apply this newfound performance in routine XRF environments.

X-ray emission spectrometry is of great practical interest because the method is rapid, non destructive and quantitative. However, the analysis of light elements which emit highly absorbed long wavelength radiations is always difficult. Although recent development of multilayered microstructures as dispersive devices has brought higher count rates for light elements, the improvement of performance in the relevant energy range depends, largely, on the improvement of the excitation. For this purpose, an electronic source providing a non-focused low energy electron beam constitutes a means of obtaining a very efficient excitation. As a result, we have used such an excitation to compare the performances of some layered synthetic microstructures (LSM) and conventional crystals for the detection of soft and ultra-soft xrays, to show up a few spectral interferences and to solve them by choosing the best dispersive device. Among the cases studied, we have shown how to carry out an accurate analysis of boron in borophosphosilicate glass (BPSG) samples. Finally a survey of the lower detection limits obtained for different elements using different dispersive devices was done.

The rapid progress of x-ray fluorescence analysis spectrometers has been closely associated with advances in computers. Due to the power of computers, it is possible to acquire the data automatically and interpret complex information accurately and quickly, so as to provide both qualitative and quantitative analyses. It is now about thirty years that computers have been applied to X-Ray Fluorescence Analysis (XRFA). Few workers have discussed the style of the various approaches, although many different software packages have been used for XRFA. Requirement analysis has been performed for Energy Dispersive (ED) XRFA, and preliminary designs are given for three types of structures: i. Sequence structure, ii. Tree structure, and iii. Net structure.

A modern theoretical method using the Fundamental Algorithm to correct for matrix effects in X-Ray Fluorescence (XRF) analysis is described. This powerful quantitative method combines the practical flexibility of influence coefficient concepts and the theoretical exactness of the fundamental parameter technique. This method is in full agreement with the treatment of the physics as proposed by Sherman and can be applied to the analysis of any sample types. It offers the maximum of accuracy limited only by the quality of sample preparation. The special calibration procedure associated with the Fundamental Algorithm is explained. This procedure allows the matching of theoretical formalism to experimental data of each individual laboratory. Finally and obviously, this approach requires a computer program to be applied. Some important guidelines are given to help XRF analysts to select the XRF program appropriate to their needs.

Increasing demands are being placed on the supervisors and operators of laboratories using XRF to establish the capability of their instruments in order to comply with internal requests and to meet customer quality assurance requirements through an audit procedure. To quantify just how accurate or precise the instrumentation is, a baseline must first be established on the instrument capability. Once the capability is documented, the precision of the instrument can be established and detection limits can be calculated. Establishing the capability of the instrument can provide a basis for developing a quality assurance procedure.

It is to be expected that drift will occur in the energy calibration of Energy Dispersive X-Ray Fluorescence Analysis (EDXRFA) spectrometers. The error thus caused in the measured x-ray intensity is evaluated in this paper. In order to minimize the error, two steps are necessary. The first is to make a new energy calibration to follow the trend of the drift. The second step involves moving the regions of the characteristic peaks to accommodate the drift. Software has been developed to accomplish both steps automatically.