Aapo Lankinen

Low Temperature Growth GaAs on Ge

In this work, low temperature growth of GaAs epitaxial layers on Ge substrates by metalorganic vapor phase epitaxy has been studied. The experiments show that a growth temperature of 530°C and a V/III ratio of 3.5 result in smooth GaAs surfaces. Atomic force micrographs do not show any anti-phase boundaries on the surface of GaAs grown on a misoriented substrate. X-ray diffraction curves show that the layer tilt is reduced as the growth temperature is lowered. Synchrotron X-ray topography reveals very low threading dislocation densities of 300 cm-2 for the GaAs epitaxial layers. Additionally, no misfit dislocations are observed. If a single layer is deposited at low temperature, secondary ion mass spectrometry shows a considerably reduced arsenic diffusion into Ge. When an additional layer is deposited at higher temperature on top of the initial low temperature layer, a substantial increase for the deep concentration-dependent arsenic diffusion is found.

Constraints on micro-Raman strain metrology for highly doped strained Si materials

Ultraviolet (UV), low penetration depth, micro-Raman spectroscopy, and high-resolution x-ray diffraction (HRXRD) are utilized as complementary, independent stress characterization tools for a range of strained Si samples doped by low energy (2keV) Sb ion implantation. Following dopant implantation, good agreement is found between the magnitudes of strain measured by the two techniques. However, following dopant activation by annealing, strain relaxation is detected by HRXRD but not by micro-Raman. This discrepancy mainly arises from an anomalous redshift in the Si Raman peak position originating from the high levels of doping achieved in the samples. This has serious implications for the use of micro-Raman spectroscopy for strain characterization of highly doped strained Si complementary metal-oxide semiconductor devices and structures therein. We find a direct correlation between the Si Raman shift and peak carrier concentration measured by the differential Hall technique, which indicates that UV micro-R...

The evaluation of mechanical stresses developed in underlying silicon substrates due to electroless nickel under bump metallization using synchrotron X-ray topography

The switch-over to the use of flip-chip Si integrated circuit bonding techniques has been driven by a need to develop higher power and lower voltage devices, capable of carrying larger currents with greater reliability. With the increased use of solder bump interconnections, an understanding of the behaviour of commonly used electroless nickel under bump metallization (UBM) layers is becoming ever more crucial. The aim of this paper is to evaluate the usefulness of white beam synchrotron X-ray topography (WBSXRT) for non-destructive evaluation of the induced mechanical stresses on Si substrates for different Ni(P) based UBM sizes and thicknesses. It is shown that WBSXRT is a powerful tool for non-destructively mapping strain and/or defect distributions within the underlying silicon substrate. Using this technique, it was also found that the crystalline misorientation induced in the underlying silicon is increased for larger UBM diameters. Stress magnitudes in the Si substrate directly under the UBM can reach values as high as 260MPa.

Femtosecond versus nanosecond laser micro-machining of InP : a nondestructive three-dimensional analysis of strain

Ultra-fast femtosecond laser micro-machining can lead to improved surface morphology and a reduction in the heat-affected zone. In this paper, synchrotron x-ray topography (SXRT) and micro-Raman spectroscopy have been used as nondestructive tools to compare the residual strain in InP substrates after femtosecond (fs) and nanosecond (ns) laser processing. Two-dimensional stain distributions with varying probing depth and cross-section images across the four laser machined grooves were obtained. The recrystallized poly-InP layer on the laser machined groove surface has been found to be highly strained in tension, and the stress magnitude is much bigger than the shear stress introduced by crystal distortion underneath. After comparing the simulation results of SXRT orientation contrast with the topography images, the nature of the crystal plane distortion induced by both fs and ns laser machining methods was elucidated.

An X-Ray Topographic Analysis of the Crystal Quality of Globally Available SiC Wafers

We present herein a first comparative analysis of the quality of 50 mm and 75 mm diameter SiC wafers, purchased directly from vendors across the world, types including the most widely available configurations. Large Area White Beam Synchrotron Back Reflection X-Ray Topography was used to analyse selected ~1cm2 regions at various locations on up to 10 different bulk SiC wafers. The study concentrated particularly on the density and distribution of threading screw dislocations (TSDs). We also examined all wafers for basal plane dislocation (BPDs) densities and distributions. Alarmingly large variation in wafer quality was observed. TSD densities vary from a minimum of 0 cm-2 (in a-plane material) to values as large as over 2,000 cm-2 on some n-type 4H-SiC wafers. TSD densities on individual wafers can also vary by similar magnitudes, e.g. 500cm-2 to 2,500 cm-2 on two regions only 2 cm apart on a 50 mm diameter wafer. Computer-based image process analysis was used to present a statistical analysis of the distributions of defects. For example algorithms created in MATLAB®, Image Processing Toolbox, isolated possible TSD locations allowing rapid counting to be performed. These counts were confirmed by manual counting of selected unmodified images.

An evaluation of an automated detection algorithm to count defects present in X-Ray topographical images of SiC wafers

Full semiconductor wafer defect/dislocation characterization is difficult to implement manually. We present an analysis of an automated algorithm used to extract Threading Screw Dislocation defect data from Synchrotron White Beam X-Ray Topographical images of SiC wafers. This extraction involves a two-fold process; firstly the algorithm highlights the appropriate defect and secondly updates the counter to provide a final result of defect count. The result of the automated algorithm is compared to hand counts in all cases, thus allowing a critical analysis of the technique. Improvements to this algorithm have been made since last reported by the same authors [1], which are discussed. The analysis herein was also performed on a much larger sample of SiC wafer images than previously used by the same authors [1] allowing a better judgment of performance and critical evaluation. The algorithm is also compared with the original previous algorithm that was used [1]. The success of this methodology paves the way for a complete analysis of whole SiC wafers, which previously was extremely difficult due to image analysis inaccuracy or the bottleneck presented by manual counting. INTRODUCTION Silicon Carbide is a promising material for a variety of electronic applications. Properties [3, 8] include a large bandgap, high electrical breakdown, low chemical reactivity and a high operating temperature. Growth however is prone to defects, common defects include Micropipes, Basal Plane Dislocations (BPD), Stacking Faults (SF) and Threading Screw Dislocations (TSD) [2-4]. Synchrotron White Beam X-Ray Topography (SWXRT) can be used to analyze SiC wafers [4, 9, 10] allowing a large image to be created of a wafer by joining smaller images. In SWXRT TSDs and BPDs [6] show up as dots and lines [5]. Being non-destructive its advantages over other techniques, e.g. KOH etching [7], are obvious. However quantitative analysis (e.g. TSD count) is extremely time consuming and tedious making SWXRT unsuited to examination of large quantities/areas of wafers. It is therefore a massive advantage if an automated method can be established to perform quantitative analysis of SWXRT images. An algorithm to classify Basal Plane Dislocations was discussed in a previous paper [1]. However the TSD algorithm developed in [1] was only semi-automatic. The complete automation of this algorithm and a demonstration of its usefulness is outlined in this paper. Mater. Res. Soc. Symp. Proc. Vol. 994