We discuss the use of Fourier transform techniques to extract layer thickness from the interference fringes observed in high resolution X-ray diffraction rocking curves of pseudomorphic HEMT structures. The interference structure is extracted by cubic spline fitting to the extrema of the data, thereby obtaining a background envelope which is used to normalise the data. The resulting constant background is subtracted from the data and the residual Fourier transformed. Auto correlation of the residual significantly improves the result from noisy data. Satisfactory results are obtained only when the Bragg peak from the substrate is windowed out. With a limited dynamic and angular range, there is often insufficient data to separate the two closely spaced periods arising from the total layer thickness and that excluding the quantum well. The result then corresponds to the average of these two thicknesses.

We show that a very low noise, high dynamic range scintillation detector has major advantages over conventional detectors for characterization of pseudomorphic HEMT structures by high resolution X-ray diffraction. We show that the reduced background enables a second modulation period to be detected, enabling the thickness and composition to be determined independently. Using a conventional X-ray generator and diffractometer we demonstrate that, in a single scan taking only 10 seconds, we are able to obtain sufficiently good data to provide quality assurance.

The work describes a method for determining the orientation of crystallographic planes in thin samples using both asymmetric Bragg reflection and asymmetric Laue transmission method.

The method is based on the extension of a system of nonlinear equations for the reflection case. The solutions of the system of nonlinear equations in question can be positive, negative or mixed with physical significance. The positive solutions correspond to reflection case, while the negative solutions are for transmission case. The mixed solutions were generated for the planes {111} in a cubic structure.

A theoretical test is given for a silicon sample. The results are in good agreement with the experimental results given in literature.

Residual stresses in thin films are important technologically due to their impact on the reliability of these structures. Various techniques are used to measure these stresses, and the results of these techniques do not always agree. In some cases, formation of interphase stresses were postulated to account for such discrepancy, but no definitive microstructural analysis was given. In this study, the residual stress response of evaporated films of 9 1% Al - 9% Hf (atomic %) deposited on 0.4 mm thick (100) Si wafers was studied as a function of thermal treatment. The stresses were measured by radius of curvature and d vs. sin2ψ techniques. Concurrently, the microstructure evolution was tracked using powder diffraction and metallographic techniques. As previously reported in the literature, residual stresses were tensile in all cases for both techniques. However, as expected, the stress results from both techniques did not always agree. The delta between the residual stresses measured by the radius of curvature and d vs. sin2ψ techniques were found to correlate well with the development of a 2nd phase (metastable cubic Al3Hf).

Grazing incidence x-ray diffraction (GIXRD) using long soller slit assemblies and a flat crystal monochromator, sometimes referred to as thin film optics, are used to study poly crystal line thin films, polymers, metals, etc. The unique ability of GIXRD to characterize crystalline materials as function of depth and provide accurate strain measurements has lead to the growing popularity of this technique. This non-focusing pseudo-parallel beam geometry can produce various effects on diffraction data which can make interpretation difficult Artifacts such as peak splitting, kα2 distortion, and peaks which shift dramatically as a function of grazing incidence angle are observed when using GIXRD. These artifacts can be related to grazing incidence angle, the divergence of the soller plates, the thickness of the soller plates, the substrate material, and the type of monochromator used. This paper briefly describes some of the topics discussed at the 1992 & 1993 Denver Conferences workshops on parallel beam optics.

TiAIN is a protective coating for cutting and reshaping tools, since it reduces wear, increases the cutting velocity of cutting tools, and increases the lifetime of reshaping tools. Thus many investigators studied the structure and properties of thin TiN films1-6. Due to the complicated geometry of the tools, the coating thickness may vary at different locations on a tool. For quality assurance thickness measurements are necessary. In this work-as a first step-only a simple geometry was considered.

Precise thin-film thicknesses have been obtained from X-ray specular and off-specular reflectivity data collected with a conventional powder diffractometer. An analysis of the specular reflectivity curve for a 565.9-Å thick pt film showed the results agreed with those determined previously from high-resolution reflectometer data to within 6.6 Å or 1.2%. An analysis of the off-specular reflectivity curves which have well-defined interference fringes showed that the results were insensitive to the surface alignment. Values of Pt thickness for the off-specular reflectivity curves agreed with that of the specular reflectivity curve to within 6.2 Å or 1.1%. The insensitivity in film-surface misalignment makes conventional powder diffractometer attractive for film-thickness determination and opens this techniques to many laboratories.

Intrinsic stresses as a function of σ, the 1/e penetration depth were measured for a smooth, 1μm thick, fine grained, cylindrical post magnetron sputtered molybdenum film deposited on a vycor glass substrate in the dynamic deposition mode. Using grazing incidence diffraction and the Mo (321) reflection, lattice spacing profiles were determined for τ values from 200-4400 Å. The in-plane intrinsic stresses parallel and perpendicular to the post axis were determined employing the ϕ-integral method and assuming elastic isotropy. The results were related to the surface structure and composition profiles via atomic force microscopy (AFM) and auger electron spectroscopy (AES) respectively.

A numerical procedure is described and demonstrated for determining z-profiles from measured τ-profiles of diffraction data such as may be obtained by grazing incidence x-ray diffraction (GIXD). The z-profile was approximated by a piece wise linear function and the problem of solving the integral equation for the z-profile was thus converted to one of solving a set of linear equations. With a sufficient number of the linear pieces, the z-pTofile could be accurately determined. The procedure and its sensitivity to data errors was tested with several arbitrarily chosen known functions. It was found that when errors were introduced into either the τ-profile or the sample thickness D the resulting z-profiles become oscillatory.

The characterization of multi-layer thin film properties, such as thicknesses of layers or concentrations of elements, using the fundamental parameter method is discussed. Experimental results are shown for repetitive multi-layers of up to 80 layers. Furthermore the impact of using different characteristic lines on the information depth and the resolving power is discussed for single layers and repetitive multi-layers. Finally, thin films containing light elements are presented; in this case the conventional fundamental parameter theory fails to provide accurate results, perhaps due to the neglect of the impact of the secondary excitation by photoelectrons.

The presence of one or more common elements in different layers often complicates or hampers the analysis of a multilayer thin film system. Some constraints or preconditions are usually required in order to reach a mathematical solution. For example, either the concentration(s) of the common element(s) in, or the thickness(es) of one of the relevant layers must be known.

An alternative approach is the use of more than one analyzing line for the element in question. Although this approach has long been proposed, there have been no detailed studies seen in literature. An attraction of using this approach is that it requires no prior knowledge of either the concentration or the thickness. The author has implemented this approach, and studied in detail its feasibility and pitfalls for practical applications.

Thin film technology has become an integral part of semiconductor multichip ceramic packages. Characterizing the thickness of multimetal multilayer thin film structures that combine both thin and thick films is an important parameter for manufacturing process control and development. Accuracies in the range of 3-5% and precisions of 1.3% were attained on multilayer metal films ranging from a few hundred angstroms to tens of thousands of angstroms.

The wafer analyzer has been used to fulfil many applications needs in the semiconductor industry. The prominent features of the XRF method for the semiconductor industry are:analysis of many types of films, e.g., oxides, silicides and metallic alloys, and simultaneous analysis of film thickness and compositions.

In the past, the analysis results of BPSG (Boron-doped Phospho-Silicate Glass) films, with thicknesses greater than 4000 Å, were reported. With the recent increased demand for larger scale and higher quality semiconductor devices (larger than 64 Mbit), more accurate analysis with high precision has been required.

Conventional X-ray stress analysis is based on the so called "linear sin2ψ" law and yields average two axial stresses in the irradiated surface layer of polycrystalline specimens. Three examples are given in this paper where either strong deviations from the linear law are observed due to stress gradients or where the linear law yields wrong results because of three axial stresses. Special experimental techniques are described and formalisms are given for the evaluation of the corresponding stress fields which are in the described cases more complicated than usual.

A formalism which was proposed by Ruppersberg (1992) for a self consistent evaluation of stress gradients in the case of strongly varying σ33(T) stresses is applied for studying the variation of the stress components with distance z from the irradiated surface of a ground hardened steel plate.

Neutron diffraction has proved to be a useful tool for measuring residual and applied stresses in materials. Aspects of experimental design and methods for the mapping of complete triaxial stress tensors in a large component are discussed in the context of a comprehensive set of stress measurements made within the interior of the circumferential weld on a 0.91 m diameter cylinder. The cylinder is a subscale model of the casings for the Advanced Solid Rocket Motor being designed for the space shuttle. The subscale model was fabricated in order to characterize the residual stresses created in the welding process and to determine the effect of post weld heat treatment on the residual stresses. Control of the residual stresses in this application is necessary to eliminate possible problems with stress corrosion cracking in a salt water environment. Some illustrative results are also presented.

The state of stress in a material, as represented by the stress tensor, can be measured using x-ray or neutron diffraction techniques. A stress tensor measured using diffraction represents an experimental estimate of the true state of stress in the material. The measured stress tensor will include both instrumental and counting statistical errors. With careful measurement techniques, instrumental errors can be minimized, and accurate results can be obtained. The errors in the measured stress tensor that are due to counting statistics can be estimated using well established error propagation techniques. Unfortunately, these errors cannot be analytically propagated through the solution to the eigenvalue problem which yields the principal stresses and directions. Without estimates of the errors associated with the principal stresses and directions, the values determined for these quantities are of limited value.

Conventional stress analysis with diffraction techniques is based on the equations of continuum mechanics, but only few materials encountered in X-ray stress analysis satisfy this assumption. In this paper, a new method is formulated yielding quantitative strain tensor fixed to the specimen coordinates. The method enables independent in-situ determination of Poisson's ratio which serves as a self control mechanism for measurement and evaluation. The analysis has been extended from flat to curved surfaces. Thus an exact analysts of spherical or cylindrical shaped specimens is now possible. The method has the capability to evaluate stress tensor and stress free lattice spacing at the surface as well as their variation with depth. This is demonstrated by two experimental data sets.

X-ray residual stress analysis was used to investigate the role of various phases in the formation of residual stress in laser-scribed alumina microelectronic substrates. Previous work by our group shows that the stress induced in the vicinity of a laser-scribe line is 20 to 50 percent of the modulus of rupture of this multiphase material. Our concern with the resulting stress is fracture of the substrate during subsequent firing processes and in-service fracture of the finished microelectronic device.

The phases investigated, in addition to the alumina matrix, included a glassy phase (7% by volume), a spinel structure (3%), and porosity (4%) in 96% alumina. Electron probe microanalysis (EPMA), electron spectroscopy for chemical analysis (ESCA). scanning electron microscopy (SEM), and quantitative metallography were used to characterize the microstructure and composition of these phases. A Nd:YAG laser was used to scribe lines which raised the state of stress in the substrate.

X-ray residual stress analysis shows that a uniform residual stress exists across the laser-scribed substrate, contrary to expected results. X-ray composition analysis suggests that the spinel phase of the material is vitrified in the heat affected zone of the laser scribe. The formation of this brittle glass phase, mainly at triple points and grain boundaries, may act as a pathway for inter-granular fracture and lead to subsequent failure of the finished microelectronic device.

As known well, non-linear sin2ϕ diagrams are observed from materials having either steep gradients of residual stresses paralell to the surface normal of the specimen or tri-axial stress components. Many authors have reported on this problem for several processing such as polishing and grinding etc. The authors also obtained such non-linear data from an electro-plated zinc—nickel alloy. For these cases, the sin2ϕ method is inadequate for the stress evaluation. Much attention has been given to the stress analysis from about 1970s as compared to the X-ray elastic constants for this phenomenon.