The acceptance angle of the detector of a double axis diffractometer is designed such that all of the diffracted beam is recorded for a rocking curve. When this acceptance angle is reduced, as in the triple axis diffractometer, two dimensional diffraction data can be recorded. In the resulting diffraction space maps each diffraction feature has a shape and a position from which the unit cell dimensions of a heteroepitaxial layer can be derived as well as information about relative tilts, curvature, lattice parameter variations and defect densities. Applications of diffraction space mapping using high and low resolution optics are discussed

A detailed discussion is given for the x-ray rocking curve analysis of semiconductor heterointerfaces and strained quantum wells using x-ray interferometry. It is shown that for a single heteroepitaxiai layer sample and a multiple quantum well sample, the heterointerface where the composition changes in both anion and cation sublattices can be studied with a half monolayer sensitivity by analyzing the x-ray rocking curve profile. Such sample systems include GaInAs/InR InAs/GaSb, GalnP/GaAs, ZnSe/GaAs, etc. A simple kinematical diffraction principle involved in the x-ray analysis is explained. Namely, the relative phase difference between the reflection amplitudes of any two layers depends on the total sum of strain (or mismatch)-thickness products over the intervening layers and interfaces. The x-ray interferometry is also effective in measuring the strain-thickness product of strained quantum wells. Experimental data and analysis results are presented for a series of functional device structures having a strained quantum well. For a more complicated sampie structure which involves both a (strained) quantum-effect layer and strained interfaces, it is shown that the x-ray interferometry is only capable of providing the total strain-thickness product over the quantum layer and the surrounding interfaces, but no detailed information about the interface grading can be gained from the rocking curve analysis alone if onhy one isolated sample is measured. An experimental data of such a sample, 0.3μm-GaAs/100Å-GaInP/ GaAs(001), is analyzed.

The interfacial structure of a lattice-matched InGaAs/InP/(100)InP superlattice with a long period of ∼630Å has been studied by fully dynamical simulations of high-resolution x-ray diffraction curves. This structure exhibits a very symmetrical x-ray pattern enveloping a large number of closely spaced satellite intensities with pronounced maxima and minima. It appears in the dynamical analysis that the position and shape of these maxima and minima is extremely sensitive to the number N of molecular layers and atomic spacing d of the InGaAs and InP layer and in particular the presence of strained interfacial layers. The structural model of strained inrerfaces was also applied to an epitaxial lattice-matched 700Å InP/400Å InGaAsP/(100)InP heterostructure.

Both double-crystal and triple-axis x-ray diffraction techniques have been used to study complex SiGe/Si structures. A novel method for measuring the nucleation activation energy of dislocations in strain relaxed SiGe/Si structures is presented to illustrate the usefulness of these techniques.

High resolution Bragg-case X-ray double and triple axis diffractometry and Laue-case white beam synchrotron X-ray topography experiments have been performed on undoped [001] oriented float-zone GaAs crystals have been grown under microgravity conditions in space on the D2 mission. Near the seed, excellent anomalous transmission was achieved and a clear cellular structure of dislocations observed. The double and triple axis rocking curves were comparable with those from semi-insulating terrestrial material. Following a heater failure, the molten zone height dropped and reciprocal space maps revealed a long ridge of scatter transverse to the diffraction vector direction. This corresponds to the presence of a distribution of sub-grains containing little internal strain. Continued growth resulted in twin formation.

Reciprocal space mapping can be efficiently carried out using a position-sensitive x-ray detector (PSD) coupled to a traditional double-axis diffractometer. The PSD offers parallel measurement of the total scattering angle of all diffracted x-rays during a single rocking-curve scan. As a result, a two-dimensional reciprocal space map can be made in a very short time similar to that of a one-dimensional rocking-curve scan. Fast, efficient reciprocal space mapping offers numerous routine advantages to the x-ray diffraction analyst. Some of these advantages arc the explicit differentiation of lattice strain from crystal orientation effects in strain-relaxed heteroepitaxial layers; the nondestructive characterization of the size, shape and orientation of nanocrystalline domains in ordered-alloy epilayers; and the ability to measure the average size and shape of voids in porous epilayers. Here, the PSD-based diffractometer is described, and specific examples clearly illustrating the advantages of complete reciprocal space analysis are presented.

Synchrotron white beam transmission topography of GaAs as previously reported by the authors relied on scanning specimen and film synchronously through the incident x-ray beam to record transmission topographic images en film. Sometimes the total dose required for reasonable contrast on film carried with it enough thermal deposition to cause elastic warping of the wafer. To escape these problems, a real time system was assembled. This system included an image intensifier, a solid state camera, a computer board to frame-grab and digitize images, and appropriate image processing software. With this system, a three inch specimen was scanned from edge to edge in one minute. At this scan rate, the incident x-ray beam had to be significantly attenuated to avoid saturating the intensifier output.

We investigated strain relaxation in (001) InGaAs/GaAs structures using both double and triple axis high resolution x-ray diffraction techniques. We determined diat broadening which is observed in double axis scans stews pnmanly from mosaic spread and not from lattice constant variations in the layer, demonstrating that relaxation is uniform along the growth direction. These observations held for layers with both low and high indium content and extents of relaxation. Triple axis measurements showed that the peak broadening was due exclusively to mosaic spread for the low indium content samples and also confirmed earlier double axis measurements that a crystallographic tilt of the epitaxial layer was attributed to substrate miscut. The ability to distinguish the source of peak broadening and crystallographic tilts makes triple axis diffraction a powerful characterization technique for the study of mismatched epitaxial layers.

X-ray diffraction topography is the name given to several x-ray diffraction techniques where large area x-ray beams diffracted from a crystal provide detailed information about the surface structure and internal perfection of crystal microstructures. Since x-ray topographic techniques are based on Bragg (reflection) or Laue (transmission) diffraction from a crystal lattice, they are extremely sensitive to any atomic lattice imperfections and strains. Alterations of the interplanar spacing as small as one part in ten thousand extending over a reasonable number of atomic ce11 lengths can be recorded as a corresponding change in the diffracted beam intensity. Line Modified-Asymmetric Crystal Topography (LM-ACT) is one such reflection technique that shows particular promise in Che field of microelectronics. The LM-ACT system is designed with low angular divergence in the x-ray beam probe. Low probe beam divergence allows details of device geometries on the order of microns to be resolved in the recorded x-ray intensity variation of the diffracted beam.

The LM-ACT system was applied here to the study of integrated circuits (IC) after specific processing steps were accomplished during IC fabrication and in the final product condition. Topographs obtained from specular crystal surfaces that were implanted through a patterned mask showed contrast variations between the implanted and non-implanted regions; details of the mask patterns have been resolved on the order of a few microns. LMACT topographs from annealed, and unannealed, Implanted specimens showed marked differences and as a result it is suggested that LM-ACT would be beneficial in optimizing the processing schedule for a particular wafer/electronic system. A significant feature of the LM-ACT technique is the capability for producing high resolution stereo-pair topographs that provide quantitative information through the depth of individual process layers in an integrated circuit.

X-ray diffraction techniques were utilized to study the film stress effects in silicon wafers subjected to several processing conditions. The wafer processing matrix consisted of three thicknesses of polysilicon deposited on (100) silicon with a 25 nm layer of Si02. The polysilicon was doped with phosphorous and arsenic after which the samples were implanted with fluorine at 30 KeV at a dose of 6 × 1015 cm2. A synchrotron radiation source at Brookhaven National Laboratory on beamline X-14 was employed to determine the residual stresses in the polysilicon film by a powder diffraction technique ordinarily used on poly crystal line material X-ray rocking curve experiments were employed to determine the stress in the film from measurements of the curvature of the substrate. These two stress measurements allow the estimation of the stress gradient in the intermediate amorphous oxide layer. Correlations are made between the stress gradients, the fluorine distribution in the wafer and the electrical characteristics of the structures.

We have conducted in-situ, real-time x-ray diffraction experiments to probe the dynamic structural changes occurring in copper during loading and then on relaxation. The 331 KαI, KαII peaks were used to monitor the development of elastic strains during loading, and their response during relaxation. The peak width was studied to better understand the structural changes that occur during loading, and more importantly on relaxation, since it is these structural rearrangements that reduce the overall strain in the system and allow the stress to relax.

The results revealed that the structure is highly mobile immediately following the start of stress relaxation. The mobility decreases with time, scales with the magnitude of the applied strain and is highly dependent upon the applied strain rate. In addition, it was apparent that the KαI and KαII peaks do not respond in the same way to the elastic strains and that they also show different structural rearrangements. This suggests an in homogeneous distribution of displacements within the sample.

A numerical procedure known as the constrained linear inversion method is employed for obtaining the z-profiles of diffraction data from their τ-profiles. The method was first tested with known z-profiles and gave good agreement with the starting z-profiles even with up to 5% random noise introduced into the τ-profile or up to 10% error in specimen thickness. The procedure was then applied to an experimentally obtained strain profile and a non-oscillatory z-profile was calculated. Although the z-profile obtained could not be verified by an independent method, the success of tests with known z-profiles leads to increased confidence in the numerical method described.

Two different inversion methods, the inverse Laplace method and the linear constrained numerical method, for retrieving the z-profiles of diffraction data from experimentally obtained τ-profiles were compared using tests with a known function as the original z-profile. Two different real data situations were simulated to determine the effects of specimen thickness and missing τ-profiles data at small τ-values on the retrieved z-profiles. The results indicate that although both methods are able to retrieve the z-profiles in the bulk specimens satisfactorily, the numerical method can be used for thin film samples as well. Missing τ-profile data at small τ values causes error in the retrieved z-profiles with both methods, particularly when the trend of the τ-profile at small τ is significantly changed because of the missing data

A method for using grazing incidence x-ray diffraction (GIXD) for profiling composition changes with depth of photovoltaic quality thin films is presented. The average thickness of the first layer in a multi-layer film of CuIn2Se3.5/CuInSe2/Mo and the variation in solid solution composition of a Cu(In1-xGax)Se2 (CIGS) film with depth are solved using this method. The phase volume fraction and the phase composition profiles are developed from peak intensity and d-spacing measurements respectively at a series of fixed incident angles corresponding to a set of increasing 1/e penetration depths, τ. Inverse Laplace and numerical methods are applied to the τ profiles converting them to true depth profiles. Vegard's law is applied to the d-spacing vs z-profile to obtain x in the formula Cu(In1-xGax)Se2. The results show that an ∼1 μm thick layer of CuIn2Se3.5 is present on the surface of the multi-layer film and that the CIGS film consists of a Ga rich surface layer ∼2000 Å thick followed by a gradual decrease in Ga content with increasing depth. This gradient appears to be desirable for producing photovoltaic quality CIGS films.

(ω, φ) X-ray Diffraetometry is a method of XRD based on two independent variables, ω–the rotation of the sample around an axis contained in the plane of the sample and normal to the incident beam, and-the rotation of the sample in its own plane. Each crystallographic plane has an φ peak position, i.e. diffracted intensity is a maximum at a particular ω position for a given φ position. The incident X-ray beam is considered to be parallel, (a divergence of 0.1 degrees or less is accepted), very narrow (the width of the beam is not critical, a width of 0.1 mm or less is accepted), and monochromatic. The position of the detector is not a necessary variable in this system. Its open X-ray sensitive area collects the photons from the diffracting planes at different positions on that sensitive area. Using the method, the diffraction angles 2θ of different crystallographic planes can be determined without the need for knowing the position of the detector, which is a large area detector, free of slits between the sample and the area sensitive to the X-ray photons. Consequently the detector can be placed in a favorable position such as close to the sample.

The method can be applied to any crystal symmetry. If the symmetry of the crystal is known then a theoretical ω(φ) curve can be generated and compared with the experimental data. The difference between these two curves can be easily computed and attributed to the crystallographic lattice distortions of the sample.

X-ray fluorescence analysis using a synchrotron x-ray microprobe has become an indispensable technique for non-destructive micro-analysis. One of the most important parameters that characterize the x-ray microbeam system for x-ray fluorescence analysis is the beam size. For practical analysis, however, the photon flux, the energy resolution and the available energy range are also crucial. Three types of x-ray microbeam systems, including monochromatic and continuum excitation systems, were compared with reference to the sensitivity, the minimum detection limit and the applicability to various types of x-ray spectroscopic analysis.

We have developed an algorithm for calculating the x-ray tube continuum based on the depth distribution function (DDF) proposed by Pochou and Pichoir, extended the description of white and characteristic radiation given by Wiederschwinger et al. to the low energy range from 5 to 30 keV and compared the results frorn these algorithms to the signals obtained from algorithms using the absorption correction of Philibert and of Sewell et al. Preceding calculations the measured spectra were separated into characteristic peak spectra and into the corresponding white spectra, where the background below the peak areas was numerically interpolated.

Widespread interest in light element analysis using XRF has stimulated the development of thin x-ray tube windows. Thinner windows enhance the soft x-ray output of the tube, which more efficiently excite the light elements in the sample. A computer program that calculates the effect of window thickness on light element sample fluorescence has been developed. The code uses an NIST algorithm to calculate the x-ray tube spectrum given various tube parameters such as beryllium window thickness, operating voyage, anode composition, and take-off angle. The interaction of the tube radiation with the sample matrix is modelled to provide the primary and secondary fluorescence from the sample. For x-rays in the energy region 30 - 1000 eV the mass attenuation coefficients were interpolated from the photo absorption data compilation of Henke, et al. The code also calculates the x-ray background due to coherent and incoherent scatter from the sample, as well as the contribution of such scatter to the sample fluorescence. Given the sample fluorescence and background the effect of tube window thickness on detection limits for light elements can be predicted.

Several interesting phenomena involving ultra-soft X-rays and synthetic multilayer crystals were studied as a result of the on-going process of improving the Rigaku Mode] 3630 Wafer Analyzer for the measurement of BPSG (1000-2500 Å) and other thin films.1-3 These phenomena can be divided into four categories; “ghost” peaks, diffraction from the substrate, fluorescence from the multilayer and higher order lines from the multilayer. Each of these is a potential snurce nf error in the measurement of ultra-soft X-rays, Fortunately, as will be shown, each can be readily dealt with.

Total reflection X-ray fluorescence (TXRF) analysis of “low Z” elements, Ti, Fe, Ni, Cu, Zn and “high Z” elements, Zr, W, Au, Pb, are required for micro scale semiconductor device production processes. In the past, the W–Lβ X-rays have been used for the analysis of S to Zn.

This paper describes a new three crystal monochromator coupled with a gold direct-drive rotating anode target. This development has produced significant improvements in the sensitivities and LLD's for all measured elements. The monochromator makes the Au Lα (9.71 keV), Au Lβ (11.44 keV) and Au Lγ (13.33 keV) available for routine use in TXRF analyses. The Au Lα X-ray is best used for the lower Z elements, Na to Cu. The best result for W, Cu and Zn are obtained with the Au Lβ X-ray. This is the result of both increased peak intensity and lower background. For the heavier elements, such as Au and Pb, the Au L X-ray yields intensities 100 times higher than what has been achieved in the past.