Search

Spherical aberration (Cs) correctors were demonstrated in the last years of the twentieth century and became commercially available a few years later. In Grenoble, we received our first probe corrector on a TEM/STEM machine in 2006. Cs-correctors have allowed us to improve the spatial resolution and the contrast of high resolution images both in TEM and STEM. The aim of the article is not to give a detailed description of Cs-correctors or a thorough analysis of their pros and cons but to illustrate what the benefits of the Cs-correctors have been in four areas: (i) atomic structure determination, (ii) polarity measurement, (iii) strain determination and (iv) interface analysis. Emphasis is put on the probe corrector although some comments on image correctors are given as well.

Si/SiGeC superlattices can be used as core structure for the fabrication of new microelectronic architectures with enhanced electrical properties such as multichannel transistors. The introduction of carbon in SiGe allows for compensation of the compressive strain and to avoid plastic relaxation. However, the formation of incoherent ?-SiC clusters during annealing at high temperature limits the processability of SiGeC. This precipitation leads to a strong modification of the stress in the alloy due to the reduction of the substitutional carbon content. Here, we investigated the variation of the substitutional C content in annealed Si/SiGeC superlattices using a comparison of dark-field electron holography and finite element simulations.

Nanobeam electron diffraction (NBED or NBD) is a transmission electron microscope (TEM) based technique able to measure strain with nanometre spatial resolution. This technique is especially well designed for nanostructures as reference diffraction pattern can be acquired anywhere on the sample. NBED is here applied on AlN/GaN superlattices grown on SiC and results compared to finite element simulations in order to assess the accuracy of the method. Strain profiles were then acquired with success on GaN nanowires presenting 3 nm GaN inclusions separated by 3 nm thick AlN.

The Geometrical Phase Analysis (GPA) is an efficient method to measure strain from High Resolution Transmission Electron Microscopy (HRTEM) images. Here we show that it can also be applied efficiently to High Resolution Scanning Transmission Electron Microscopy (HR-STEM) images, which have several advantages: (i) patterns in HR-STEM do not change with thickness and chemical composition (ii) thicker samples can be analysed and (iii) strain and composition can be simultaneously determined. In many situations, the distortions due to the scanning of the beam can be corrected. The strain fields around different threading dislocations in an AlInN layer have been determined from plan view samples prepared by focus ion beam (FIB). Experimental strain maps were compared to analytical calculations that take into account the strain field of dislocations and of the In segregation. Mixed type dislocations are always terminated by an inverse hexagonal pyramidal pit, at the sample surface. The edges of the inverse pyramid are indium rich. The dislocation core is not situated at the centre of the inverse pyramid, which is indium-rich, but slightly shifted.

There is a need to measure the dopant potentials and strain fields in semiconductor materials with nm-scale resolution. Here we show that off-axis electron holography is a powerful technique that can be used to measure the fields present in a high-k metal gate 28-nm node nMOS device with a contact etch stop liner stressor. Off-axis electron holography has been used to map the positions of the active dopants with a spatial resolution of 1 nm. The experimental results have been compared to electron energy loss spectroscopy maps. Finally, dark field electron holography has also been used to provide strain maps and the experimental results have been verified using nanobeam electron diffraction.

Strain is routinely used in state-of-the-art semiconductor devices in order to improve their electrical performance. Here we present experimental strain measurements obtained by two different Transmission Electron Microscopy (TEM) based techniques. Dark field electron holography and nanobeam electron diffraction (NBED) have been applied to a range of semiconductor devices including recessed SiGe source and drain devices and SiGe pFETs on SOL Here we demonstrate how the spatial resolution and sensitivity of these two techniques has been improved by using a state-of-the-art, probe corrected Titan TEM, and have benchmarked the capabilities of the two different techniques and highlight which method is most appropriate for different types of semiconductor specimens.

High resolution Z-contrast STEM imaging is used to study the Tm doping of GaN quantum dots grown in AlN by molecular beam epitaxy (MBE). High-angle annular dark field (HAADF) imaging allows us to visualize directly individual Tm atoms in the AlN matrix and even to count the number of Tm atoms in a given AlN atomic column. A new visibility coefficient to determine quantitatively the number of Tm atoms in a given atomic column is introduced. It is based on locally integrated intensities rather than on peak intensities of HAADF images. STEM image simulations shows that this new visibility is less sensitive to the defocus-induced blurring or to the position of the Tm atom within the thin lamella. Most of the Tm atoms diffuse out of GaN dots. Tm atoms are found at different positions in the AlN matrix, (i) Above the wetting layer, Tm atoms are spread within a thickness of 14 AlN monolayers (MLs). (ii) Above the quantum dots all the Tm are located in the same plane situated at 2-3 MLs above the apex of the GaN dot, i.e. at a distance of 14 MLs from the wetting layer, (iii) In addition, Tm can diffuse very far from the GaN dot by following threading dislocations lines.

Nano beam electron diffraction (NBD or NBED) is applied on a well controlled sample in order to evaluate the limit of the technique to measure strain. Measurements are realised on a 27nm thick Si0.7Ge0.3 layer embedded in a silicon matrix, with a TITAN microscope working at 300kV. Using a standard condenser aperture of 50μm, a probe size diameter of 2.7 nm is obtained and a strain accuracy of 6×10−4 (mean root square, rms) is achieved. NBED patterns are acquired along a [110] direction and the bidimensionnal strain in the (110) plane is measured. Finite element simulations are carried out to check experimental results and reveal that strain relaxation and probe averaging in a 170nm thick TEM lamella reduces strain by 15%.

The atomic structure of an incommensurate (001)/(110) Si grain boundary (GB) or 90° Si GB has been studied by transmission electron microscopy (TEM) and refined by atomistic simulations (Stillinger-Weber potential). Samples were made by bonding one (001) Si wafer with one (110) Si wafer and carefully orienting the 2 wafers in order that they have a common [10] direction. In the interfacial direction perpendicular to [10], the [110]I direction of grain I is parallel to the [001]II direction of grain II and, as the ratio of these 2 vectors is , it is impossible to find 2 integers n and m such that n[110]I=m[001]II. The structure is incommensurate in this direction. Z-contrast images obtained in an FEI-Titan microscope equipped with a probe Cs-corrector easily resolve the Si dumb-bells in the two grains and allow us to determine the complex atomic structures of the interface. On the other hand, near on-axis high resolution TEM images obtained in a JEOL 4000EX microscope are very efficient to analyse the long range order of the interface.