ELECTRON ROTATION CAMERA

A. Prevoiusly known technique

5 1. X-ray diffraction

The best way to determine the atomic structure of a crystalline material is by diffraction techniques. X-ray diffraction was the first and is still the dominating method for obtaining diffraction patterns. However, due to the weak interaction between X-rays and matter, it is not possible to collect data on crystals smaller o than about 1 μm ³ , even using a synchtrotron with a very bright X-ray source. A sample of one 1 μm ³ contains about 100 000 000 000 atoms.

In order to collect a complete 3D data set of diffraction spots, it is necessary to rotate the crystal in the X-ray beam. There are different geometries for doing this, such as Weissenberg, precession and rotation (also called 5 oscillation). The dominating method today is to use a synchrotron as a fixed X-ray light source, and then rotate the crystal around an axis perpendicular to the narrow beam of X-rays. The rotation may be up to or even over 90 degrees.

2. Electron diffraction 0 In many modern industrial applications it is necessary to be able to study samples in the nanometer scale. Such specimens may contain from 1 to 10 000 atoms. The only way to study such minute samples is by electrons, which interact about a million times stronger with matter. However, in the traditional mode of electron diffraction, SAED, both the electron beam and crystal are fixed, thereby 5 limiting the amount of data that is recorded in any one image. In order to collect a 3D data set, it is necessary to collect a large number of diffraction patterns at different angles (see Figure 1).

In fig. 1 there is disclosed an example of a collection of electron diffraction patterns from many directions on the same crystal. Each diffraction pattern is in ø fact a section through 3D space, as illustrated by the set of five lines at the bottom left of fig, 1.

This procedure is not only very time consuming, it does not collect the complete set of diffraction points. Furthermore, the relative intensities of the diffraction spots (which carries the information about the atomic structure) is

unreliable, due to multiple scattering of the electrons as they pass through the sample.

3. Electron precession Recently Paul Midgley and Roger Vincent in Bristol, UK invented the precession method for electron diffraction. Here the electron beam is tilted by a small angle, typically 1-3 degrees, and then rotated around the optical axis. In this way a volume of reciprocal space is recorded, rather than just single sections.

Furthermore, the multiple scattering is greatly reduced, since at any one moment only a small number of reflections are excited. The beam tilt is accomplished by adding small currents to the lens(es) above the sample in the electron microscope.

In order to compensate for the diffraction pattern moving around below the sample

(as a result of the incident beam coming from different directions), the electromagnetic coils below the sample are used to descan. The precession technique has recently been improved and commercialized by NanoMEGAS, protected by several patents.

B. Limitations of present techniques

X-rays diffract so weakly with matter that samples smaller than 1 μm ³ cannot be studied. The SAED method of electron diffraction is very time- consuming and demanding on the operator of the electron microscope, and it does not integrate through the full 3D volume of reciprocal space. For any crystals more than a few nanometers thick, the multiple scattering of electrons will make data collected by SAED of limited quality.

Electron precession solves all these problems, but one important thing remains. It is very difficult, not to say impossible to collect full 3D data with the precession geometry. There will be many partially recorded reflections at the borders of the scanned space. In fig. 2 there is disclosed an example of a crystal illuminated by a stationary electron beam, called SAED for Selected Area Electron Diffraction, which gives a small circular area of ZOLZ (for Zero Order Laue Zone) and a very narrow ring of FOLZ (for Second Order Laue Zone).

C. Solution of the problems

In X-ray crystallography, precession was a standard technique until about 1975, when the oscillation or rotation method was developed by Arndt and Wonacott in Cambridge, UK. The main reason for going from precession to rotation was the simplified geometry, especially the fact that one film recorded by rotation could be directly added on to the following, allowing partially recorded reflections on one film to be filled in on the next film. For the same geometricai reasons, full high-quality electron diffraction data should also be collected by a rotation method.

Eiectron rotation can be achieved by a device rather simiiar to the one making electron precession. The main difference is that the electron beam (and its corresponding descanning unit) should not go around in a circle, but rather follow a straight line, like a pendulum. This Sine can be along the x-direction, along the y- direction or along any diagonal in between.

In order to handle partially recorded reflections, the data should be collected in a number of small steps. For example, each scan may have a rotation of only +/- 0.5 degrees along a line. The next scan will follow on from exactly where the previous stopped, i.e. from +0.5 to +1.5 degrees, with the next one + 1.5 to +2.5 degrees etc. One such series of rotation images can total up to about 6 degrees (the exact range is limited by the design of the specific model of electron microscope).

In fig. 3 there is disclosed a description (side view) of how the electron beam is rotated from one side to the other using the proposed rotation method. Different reflections are excited as the incident angle is altered. The incident electron beam comes from above and excites reflections on the surface of a large sphere (the EwaSd sphere). The reflections lie on planes, numbered / = 0, 1 , 2 etc. All reflections within the grey region are seen in the diffraction pattern (see Figure 2). Reflections with / = 0 are said to belong to the ZOLZ (for Zero Order Laue Zone), / = 1 are FOLZ (for Second Order Laue Zone), / = 2 SOLZ (Second order Laue Zone) (not shown in the Figure) and so on.

All reflections in the ZOLZ between +/- R ₀ ,o _υt and in the FOLZ between Ri _]jn and R _{1 out} are recorded on the detector. The reflections have a smai in all three dimensions, here illustrated by a line called g, called excitation

This patent is for making a device that can scan and descan the eiectron beam by a predetermined but variable angular rotation range, and aiong any direction. St will also have the possibility of taking a series of scans, pair wise exactly fitting back-to-back. Furthermore, it comprises computer programs for simulating the rotation data sets and for determining the exact orientations at the start and end of each rotation pattern. The computer programs aiso include functions for identifying partially recorded reflections and adding them on consecutive recordings.

The diffraction patterns may be recorded on any media, including photographic film, imaging plates or CCD cameras.

In fig. 4 there is disclosed an example when the beam is rotated or osciliated +/- a few degrees, the number of reflections that are recorded is greatly increased in the ZOLZ and even more markedly in the FOLZ. Reflections at the borders of the recorded area (marked by arrows) are oniy partially recorded on this image which has been obtained by rotation from -1 to +1 degree. The remaining parts of the reflections are recorded on a consecutive image, for example with rotation from +1 to +3 degrees.

The device and related technique described in this Patent uses a combination of specific scanning protocol of ED intensities (including beam rotation) that can be further and simultaneously scanned across a specific nanocrystai surface observed in TEM. This may require the presence of a scan generator configured to control scan coils for scanning an ED pattern, said scanning generator may exist in the TEM or can be built as external device to synchronize beam displacement over the sample with simultaneous acquisition of ED patterns. A computer may collect such patterns (through the frame grabber) and store them for further on-line or off-Sine analysis.

Such ED patterns generated at each sample point of the nanocrystai under study may reveal crystal structure details such as crystal phase and local orientation after comparison with recalculated theoretical kinematical ED patterns of known crystal phases under different crystaSlographic orientations.

Such ED patterns obtained under specific scanning protocols (ϋke precession or rotation or oscillation) may be useful as quasi-kinematicai ED intensity fingerprints of known phases and can be used as such for search/comparison with known crystaSiographic databases (for example ICDD,

FIZ, CSD or others). Such comparison can be performed visually or by image processing techniques. Rotation, oscillation and alternative scanning protocois (inciuding but not limited to beam precession) are producing ED intensities close the their ideal kinematical values, aSiowing therefore direct comparison with simulated kinematical ED patterns of known compounds without taking in account crystal thickness effects.