Comparison of convergent beam electron diffraction and geometric phase analysis for strain measurement in a strained silicon device

Abstract

Key words. Convergent beam electron diffraction (CBED), geometric phase analysis (GPA), high-angle annular-dark-field scanning transmission electron microscopy (HAADF STEM), local strain measurement. SummaryConvergent beam electron diffraction and geometric phase analysis were used to measure strain in the gate channel of a p-type strained silicon metal-oxide-semiconductor field-effect transistor. These measurements were made on exactly the same transmission electron microscopy specimen allowing for direct comparison of the relative advantages of each technique. The trends in the strain values show good agreement in both the [110] and [001] directions, but the absolute strain values are offset from each other. This difference in the absolute strain measured using the two techniques is attributed to the way the reference strain is defined for each.Because strained silicon is being used on a highly localized scale to enhance the electrical performance of complementary metal-oxide-semiconductor (CMOS) devices (Chidambaram et al., 2004;Thompson et al., 2004), techniques for the quantitative measurement of elastic strain with high spatial resolution are increasingly necessary. Although convergent beam electron diffraction (CBED) and geometric phase analysis (GPA) have been used previously by several groups for these types of measurements (Toda et al., 2001;Hÿtch et al., 2003;Benedetti et al., 2004;Toh et al., 2005;Chung et al., 2008;Hüe et al., 2008), there have been no reports directly comparing the strain measurements from the two techniques for the same device structure. Because there are relative advantages and limitations of these techniques for strain measurement, a direct comparison of these two techniques should provide a better understanding of these. In CBED, a small, convergent probe is used to produce diffraction from both the zero-order and higher-order Laue zones (ZOLZ and HOLZ, respectively). The HOLZ diffraction appears as pairs of lines in the diffraction patterns -rings of excess lines diffracted to a relatively large angle and corresponding deficit lines in the central disk. Because these lines occur due to Bragg diffraction at relatively large angles, their positions are very sensitive to small changes in the crystal lattice. The sensitivity of lattice measurements using this technique can be on the order of 10 −4 . (Twigg et al., 1987;Kramer & Mayer, 1999) However, most silicon-based devices are oriented along 110 directions, and CBED patterns along this direction do not produce HOLZ lines under typically utilized CBED-operating conditions due to the large spacing between planes in the reciprocal lattice along this direction. Therefore, measurements using this technique are performed by tilting away from the 110 axis, typically along the 004 Kikuchi band, resulting in a loss of lateral resolution. This is particularly important as the scale of microelectronic devices continues to decrease. The degree of resolution loss depends on the angle of tilt from the [110] direction, so efforts have been ...

Strain fields around dislocation arrays in a Σ9 silicon bicrystal measured by scanning transmission electron microscopy

“…Since strain relaxation occurring at the two specimen surfaces modifies the system studied, thicker region are generally more representative of the original bulk sample. GPA in STEM has previously been used to study structural distortions in charge-ordered manganite [24], as well as strain fields in SiGe/Si heterostructures [25,26] and surrounding dislocations [27,28,29]. Because images in STEM are acquired pixelby-pixel, instabilities and scanning effects introduce distortions that ultimately limit the precision of strain measurements.…”

Quantitative measurement of strain field in strained-channel-transistor arrays by scanning moiré fringe imaging

“…These effects can give rise to a fluctuation in the strain value when the TEM image is converted into the strain map, which results in a degradation of the measurement precision. 5 In contrast, the position of the atomic column intensity in high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) images is not shifted by the local variation in sample thickness because the atomic column intensity is obtained by collecting the electrons incoherently scattered at high angles (Z-contrast imaging). 10 Furthermore, in Z-contrast imaging, the contrast reversal does not occur as the sample thickness varies.…”

“…The nano-scale measurement of the strain field of the channel in a transistor has been performed by applying geometrical phase analysis to the TEM image (GPA-TEM). [4][5][6][7][8][9] However, the drawback of TEM images is that the phase contrast is greatly affected by a sample condition such as the local variation in thickness, which leads to a phase shift or contrast reversal in the TEM image. These effects can give rise to a fluctuation in the strain value when the TEM image is converted into the strain map, which results in a degradation of the measurement precision.…”

High‐Resolution Scanning Transmission Electron Microscopy (HRSTEM) Techniques: High‐Resolution Imaging and Spectroscopy Side by Side

“…[79] High-resolution applications of diffraction mapping techniques include local Ge concentration determination in Si/SiGe nanostructures, [80] the evaluation of lattice distortions on InP nanowires containing an axial screw dislocation, [81] and strain mapping in metal oxide semiconductor field-effect transistors (MOSFET) devices. [82,83] In addition, improved grain orientation mapping on polycrystalline materials has been demonstrated by the use of precession-enhanced electron diffraction. [84] Moreover, diffraction analysis on STEM mode currently provides new information on electron-scattering processes of crystalline materials.…”