During the past fifteen years wavelength dispersive x-ray fluorescence (WDXRF) spectrometry has been the primary analytical technique for quantitative elemental analysis of automotive catalyst precious metals, contaminants, and substrate materials. While extensive work has been devoted to improving the accuracy of WDXRF quantitative procedures, minimal attention has been given to the calculation of critical sample thickness (primary x-ray beam depth of penetration) and total volume of material excited for each element in the catalyst. However, with the increasing use of WDXRF for measuring and comparing elemental concentrations at the inlet, middle, and outlet surfaces of catalysts, critical sample thickness and volume of material excited becomes very important for accurate interpretation of results.

Additive elements are now included in manufacturing specifications for greases. Texaco uses X-ray fluorescence at lubricants manufacturing locations; however, analyses of greases in the neat form are difficult due to problems with uniform and repeatable filling of sample cups and outgassing of greases during measurements. This preparation procedure allows quantitative determinations of additive elements in a variety of greases using a single analytical program. Having a single method for all greases is important at busy manufacturing locations.

The preparation method involves dilution of samples in a binder powder. This approach eliminates outgassing and overcomes differences among greases such as type of soap, presence of MoS2 and graphite, and varying oil content. The powder absorbs oil and creates a uniform particle structure. Thorough mixing and careful cup packing yield repeatable measurements. The method allows the use of liquid oil standards for calibration, eliminating the need to obtain analyzed greases. Accurate calibrations have been created for both wavelength-dispersive and energy-dispersive spectrometers.

Federal laws require that urban redevelopment projects characterize the Pb hazard of old painted surfaces. In large metropolitan surveys, the Pb analyses are commonly obtained with a portable XRF unit; however, for small surveys, capital investments in such instrumentation may be unwarranted. In 1988, the city of Las Cruces, NM, requested our help in providing needed Pb in paint analyses. We developed an economical XRF method for Pb analysis suitable for small scale urban surveys and the requirements of individual home owners, it is described below.

1) A standard wavelength dispersive XRF spectrometer is used to determine the CI countrate from a 3.6 cm KCl disk and the countrate per square cm is computed. 2) The paint chip is weighed to the nearest 0.1 mg and cemented onto the disk. 3) The prepared disk is counted for CI, Pb, and Rb using the same spectrometer. 4) The CI countrate deficit is used to compute the area of the paint chip. 5) The mass of Pb corrected for Rb interference is determined from an appropriate calibration curve, and the results are reported in mg Pb/cm2.

The Pb analysis is calibrated against geochemical standards. Spex Mix containing Pb is sprinkled on double faced mylar tape of known area and used as an independent check on the calibration curve. We present the results of our calibration curves and our analytical experience over the last 5 years.

A review of the methods of characterization of materials using X-rays incident at grazing angles is presented. The rationale of all such methods is the need to obtain information from near-surface regions. The methods include grazing incidence diffraction, reflectivity, diffuse scatter and fluorescence. The experimental techniques are outlined, and the information obtainable and the methods of interpretation are discussed.

X-ray diffraction measurements were performed on CoαPt1-α/Pd, Co/Pd, Co/Fe, and Co/W multilayer samples with different structures, such as CoαPt1-α alloy layer composition α, bilayer thickness, and number of bilayers. Multilayer samples were made by magnetron sputtering in a chamber with multi-parallel guns and a position controllable substrate. CoαPt1-α alloy layers were deposited by cosputtering from Co and Pt targets mounted on guns tilted towards a common substrate. Compositions of Co and Pt in CoαPt1-α layers were varied by use of different sputtering power. The thicknesses of magnetic and non-magnetic layers in multilayered samples were also systematically changed to investigate the relationship between x-ray diffraction lines and crystalline structures of multilayered films. It was found that the position of the main diffraction peak from multilayered films was solely determined by the crystalline structures within bilayers rather than bilayer thickness. A model was introduced to calculate microstructural parameters such as the thickness of interfaces and compositions at interfaces.

In recent years, the microstructure of Co-based alloy magnetic thin film disks has been studied extensively by using a conventional x-ray diffraction (XRD) technique(1-4). Efforts to understand the relationship between the microstructure and the magnetic properties of disk had been limited due to the lack of depth resolution in the conventional theta-2theta scanning technique. In this paper, a fixed 0.5 degree of grazing incidence(5) 2-theta scan X-ray diffraction (GIXRD) technique was used to characterize the thin film disk microstructure and therefore to correlate with the magnetic properties.

Polyimide films are becoming more important in multi-layer structures, acting as dielectric films to separate conductive layers. High thin film residual stresses can cause problems with the polyimide film, leading to loss of adhesion between the metal film and the polyimide or can cause cracking in the polyimide.

In this paper, we examine the residual stresses in “as deposited” metal films evaporated onto BTDA-APB polyimide films and follow the change in residual stress of the nickel layer during subsequent thermal processing. We will show that the change in residual stresses can be correlated to the glass transition temperature of the polyimide film.

Residual stresses present in thin films evaporated on ceramic substrates, can lead to loss of adhesion of the metal features to the ceramic, or, between metal films in multi-level stacks. If the adhesion is good, ceramic cracking may occur around the metal features. Long term one must avoid stress corrosion cracking in the ceramic. These residual stresses are generated by a variety of causes. In general, defect incorporation during the deposition process and thermal expansion mismatch between thin film and substrate play major roles.

In this paper, we describe a method for determining the residual stresses that may be detrimental, or more accurately, the residual stress induced load that may cause reliability problems. This method involves evaporating metal pads or various thicknesses onto the ceramic to the point of failure, and measuring the residual stress loading on the substrate. A direct measure of this loading is the stress intensity factor which takes into account both the residual stress and pad or film thickness. Thus, a critical “load” or stress intensity at which substrate cracking or thin film delamination occurs can be estimated. This approach can be used to select the optimum metal-thickness combination for various substrates.

Double Crystal X-Ray Diffraction (DCD) is often used to determine the Al content of AlxGa1-xAs/GaAs epitaxial layers. Assessing composition from a measurement of mismatch is problematic because it invokes a number of assumptions. This study bypasses these difficulties by comparing the measurement of mismatch directly with Al composition measurements made by electronprobe microanalysis. A study of coherent epitaxial AlxGa1-xAs layers showed that mismatch varies linearly with composition. The equation Al (x) = |ΔΘ| / 368 summarizes the relationship over the coherent range, where |ΔΘ| is measured in arc seconds.

The development of high density interconnection (HDI) technology in multichip modules (MCM's) will establish a new level in the hierarchy of electronic systems. The modules use organic insulating layers which, because of their low dielectric permittivity and loss, enable circuits with maximum density and speed. However, differences in the coefficients of thermal expansion (CTE's) for the insulating layers, the interconnects, and the substrates, produce residual stresses in the various components during processing. These stresses must be understood to engineer reliable designs for MCM's.

Two proposed methods are described for obtaining the depth profiles (τ-profiles) of strains and stresses as a function of the 1/e penetration depth τ, using GIXD in the asymmetric geometry, without the need for layer removal. In the first, the ψ-method: ψ is varied, for a given reflection, while holding τ constant, by varying the wavelength λ and adjusting the incident angle α so as to maintain τ constant. This allows dϕ,ψ vs Sin2ψ plots to be obtained at constant τ. In the second, the ϕ-integral method: interplanar spacings dϕ,ψ are measured for ϕ values from 0 to 2π, at constant ψ and τ by holding α fixed. Two ψ values, i.e. two wavelengths, are needed to obtain the whole strain tensor, but if the sample is quasi-isotropic and the stress state is biaxial, only one wavelength is needed. A direct method is also described for obtaining the profiles as a function of depth z (z-profiles) from the corresponding τ-profiles using inverse Laplace transforms. Application of the method to the residual strain vs τ data of Doerner and Brennan for an Al film on Si is presented.

Methods are described which make use of x-ray absorbing masks to either limit beam penetration in the sample (Type I), or confine the diffracting volume within definite limits (Type II). The masks consist of a metal film of Au or U, containing many long, parallel apertures of effective width w and spacing c. The mask is either applied directly to the sample surface (Type I), or to a weakly absorbing film (Be or Kapton) on the sample surface (Type II). With the Type I mask, the diffracting volume can be limited to a surface layer whose maximum depth zm at ψ=0 varies linearly with w, irrespective of sample or wavelength. The masks can be used with ψ-tilts in an Ω-type goniometer, and strains and stresses can be determined by the Sin2ψ method. Depth profiles as a function of depth z (z-profiles) can be obtained with layer removal if gradients over the depth Zm can be neglected. With the Type n mask, the diffracting volume can be limited to a discrete subsurface layer whose average depth depends on (c+w) and the Bragg angle 2θ, and whose thickness depends on w and 2θ. z-Profiles can be obtained by varying the wavelength for a given reflection using a synchrotron source, without the need for layer removal. The methods have not yet been tested experimentally.

Near-surface-layer analysis by x-ray fluorescence critical takeoff-angle detection was examined theoretically and experimentally. The takeoff-angle dependence of Cr K-L x-ray fluorescence from Cr thin films was measured with practical analyzed depth varying from several thousand to less than one hundred angstroms. Comparison of Cr K-L and K-M takeoff-angle dependence shows the analyzed depth depends on the observed x-ray energy even with identical takeoff angles. The potential of nondestructive depth profiling is also discussed.

Depth profiling is performed by monochromatic primary excitation using variable incidence and take-off angles, A proper choice of the photon energy and the angular range of incident radiation allows depth profiling of layered structures without comparison to reference samples. The method has been verified for two different systems. Thin Al(x)Ga(1-x)As-layers on GaAs substrates are characterized by thickness D and atomic ratio x. For an excitation of AlKα-radiation by SiKα-radiation from a secondary target, the lower limit of measurement is layer thicknesses of 40 to 80 nm at x=0.6 to 0.4. Zn-coatings on steel have been investigated with CuKα radiation for selective excitation of FeKα-radiation from the substrate and with GeKα radiation for the excitation of ZnKα radiation from the layer. The scatter of the results required a measurement of the lateral homogeneity of the coating thickness. These experiments have been performed by step scanning in steps of 1 mm under fixed incidence and take-off geometry over an area of 30mm*30mm and excitation with unfiltered radiation from a Cu target diffraction tube with a beam cross section of less than 1 mm2. The obtained thickness distribution of ±20% has been confirmed by electron micrographs.

NIST SRM 2708 is a thin film of zinc sulfide approximately 0.02 μm thick that was sputter deposited on polycarbonate substrates using a NIST ion-beam instrument. It is intended for the standardization of x-ray fluorescence spectrometers, especially for analysis of air particulates or similar material collected on filter media.

Since more than a decade ago, the topic of multilayer thin film analysis using Fundamental Parameters (FP) method has been addressed by many authors. However, as a general approach to the implementation of this method, an iteration must be applied directly with the theoretical equations. In many cases, this procedure can become very time-consuming and sometimes cause divergence problems, especially when a complicated layer structure is involved or some special requirement for the analysis must be satisfied. In this work we adopted a new approach to the analysis of some types of multilayer thin films, which is conceptually similar to die well known theoretical alphas method used in bulk sample analysis in the sense that they all use FP theoretical calculations to establish mathematical relationships between the XRF intensity and the quantities of interest.

The use of whole-pattern matching techniques for phase identification provides increased confidence in the phases determined compared with d-I matching and enhances the potential for determining the existence of phases present in low concentrations. Reference patterns from a database of selected phases were compared with the experimental pattern obtained from an unknown, and the results were ranked using figures-of-merit designed to distinguish the best pattern matches. Several different figures-of-merit have been evaluated, all of which proved successful in recognizing the strongest phase but varied with respect to the other phases.

Low-concentration phases are revealed when the patterns of the more abundant phases are stripped from the experimental trace using the best-fit scaled reference traces. Pattern stripping is improved by pattern shifting and profile shape matching which are provided for in the matching program.

Intensity aberrations in Bragg-Brentano X-ray diffractometers reduce the measured intensities at low angles relative to higher angles, and the correction factor can be large at low 2θ angles. Aberration-free data is of great importance in full-profile Rietveld quantitative analysis or structure refinement. An algorithm is given which corrects the intensity aberrations for particular instrument dimensions and sample absorbance, with a fixed divergence slit. Variable divergence slits increase the Bragg-Brentano aberrations and skew the data. The importance of an aberration correction curve produced with the computer program BBCCURV is described in the multiphase Rietveld quantitation of a montmorillonite clay with the SIROQUANT quantitative analysis program system; good quantitation of the montmorillonite can only be obtained if the aberration corrections are applied.

The multiphase Rietveld method has been used in the determination of phase abundance in six standard Portland cements, using the SIROQUANT personal computer software package for XRD quantitative analysis. Because of the extreme amount of line superposition and weakness of the cement patterns care must be taken in the selection of standard XRD profiles and in the refinement of unit cell dimensions and linewidths. Comparison with round-robin microscope assays of the same cements shows r.m.s. deviations of assays between the two methods of 2.1% for alite, 3.1% for belite, 1.7% for aluminate and 1.3% for ferrite. Precision, studied by repeated SIROQUANT analyses of nine XRD patterns from the same Australian cement show esd's of 1.3% for the alite assay, 0.8% for belite, 0.5% for aluminate and 0.5% for ferrite.

Rietveld X-ray powder diffraction (XRD) and uniaxial stress measurement procedures are being evaluated as a means of correlating the crystalline structure properties of zinc-nickel electroplate with observed plating properties. The study is being performed on a suite of six samples of varying composition (6 to 20%Ni) electrodeposited on mild steel. Rietveld refinement of the XRD data (Rietveld, 1969) has been performed on a specially prepared ingot (9%Ni:91%Zn) to determine the structural parameters and relative abundances for the γ(Ni4Zn22), δ(Ni3Zn22) and η(Zn) phases which could possibly be present in the electroplate. Relative phase abundances were verified by Atomic Absorption Spectrometry (AAS) and electron probe microanalysis (EPMA) image mapping techniques.

The structural parameters for the ingot zinc-nickel phases were subsequently used as initial parameters in Rietveld calculations for the six electroplate samples. Structure refinement of the electroplate XRD data was hampered by extremely high preferred orientation and the corresponding lack of quality diffraction peaks. However, Rietveld refinements produced accurate cell dimensions (which correlate linearly with nickel composition), a means of quantifying the degree of preferred orientation, and crystalline phase verification. For residual stress determination, d-sin2ψ plots were produced for each electroplate sample. The plots were found to be linear for Ni < 12%, allowing calculation of residual stress, and distinctly curved for Ni > 12%.