Sunteți pe pagina 1din 20

Journal of Microscopy, Vol. 232, Pt 1 2008, pp.

726 Received 12 December 2007; accepted 22 February 2008

Contribution of electron precession to the study of perovskites displaying small symmetry departures from the ideal cubic ABO3 perovskite: applications to the LaGaO3 and LSGM perovskites
J . - P . M O R N I R O L I , G . J . A U C H T E R L O N I E , J . D R E N N A N & J . Z O U

Laboratoire de M etallurgie Physique et G enie des Mat eriaux, UMR CNRS 8517, USTL, ENSCL, B atiment C6, Cit e Scientifique, 59500 Villeneuve dAscq, France Centre for Microscopy and Microanalysis School of Engineering, The University of Queensland, 4072, Queensland, Australia

Key words. Electron precession, LaGaO 3 , lanthanum gallate, LSGM crystal structure, perovskite.

Summary Electron microscopy and electron diffraction are well adapted to the study of the fine-grained, faulted pure and doped LaGaO 3 and LSGM perovskites in which the latter is useful for fuel cell components. Because these perovskites display small symmetry departures from an ideal cubic ABO 3 perovskite, many conventional electron diffraction patterns look similar and cannot be indexed without ambiguity. Electron precession can easily overcome this difficulty mainly because the intensity of the diffracted beams on the precession patterns is integrated over a large deviation domain around the exact Bragg condition. This integrated intensity can be trusted and taken into account to identify the ideal symmetry of the precession patterns (the symmetry which takes into account both the position and the intensity of the diffracted beams). In the present case of the LaGaO 3 and LSGM perovskites, the determination of the ideal symmetry of the precession patterns is based on the observation of weak superlattice reflections typical of the symmetry departures. It allows an easy and sure identification of any zone axes as well as the correct attribution of hkl indices to each of the diffracted beams. Examples of applications of this analysis to the characterizations of twins and to the identification of the space groups are given. This contribution of electron precession can be easily extended to any other perovskites or to any crystals displaying small symmetry departures.

Introduction There are many new materials, which are either in the market place or under development, that are based upon high-temperature electrochemical processes. Some examples of these materials are batteries (Kao et al., 1992), catalysts (Lahousse et al., 1998), epitaxial substrates (Christen et al., 1997), gas separation membranes (Balachandran et al., 1998), high-temperature electrodes (Kawada et al., 2006), oxygen pumps (Yuan & Kroger, 1969), oxygen sensors (Weissbart & Ruka, 1961), radioactive waste containment (Ringwood et al., 1979) and solid oxide fuel cells (Minh, 1993). With such diverse categories of materials and applications one can see the importance of these materials to our modern technological society (Stlen et al., 2006). Some of these new materials are based upon the ABO 3 perovskite-type structure where several inner transition (rare earth) element gallates are isostructural with the archetypal ABO 3 perovskite, GdFeO 3 . With A- and/or B-site doping, these perovskite-based oxides transform from insulators into conductors. Rather than being electron conductors these oxides are either mixed electron/ion conductors or ion conductors. Because ions are much larger in size than electrons, their migration pathways become far more critical because these ions must jostle past cations in a solid lattice. It is therefore important to study the basic migration mechanisms for electronic, ionic and mixed conduction in these complex ABO 3 perovskite-type oxides. The ABO 3 perovskite-type structure is both geometrically and chemically stable giving it the advantage of great chemical flexibility. That is, the ability of the lattice to accommodate many different types of dopant atoms with both wide-ranging atomic sizes and chemical valencies. Thus, in these doped ABO 3 perovskites one must correctly identify the crystal

Correspondence to: J.-P. Morniroli. Tel: 33320436937; fax: 33320434040; e-mail: Jean-Paul.Morniroli@univ-lille1.fr

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society

J.-P. MORNIROLI ET AL.

structure at both room and operating temperatures, especially as at high temperatures, phase transitions occur which yield the desired operating properties of these materials. Once the crystal structure has been correctly identified, one can then hopefully determine the anion conduction pathway through the solid. The structure determination of these compounds using X-ray patterns has been used routinely for many years; for example see Wang et al. (1991). It indicates that some perovskites adopt an ideal cubic aristotype structure but most of them exhibit some small departures from this ideal cubic symmetry leading to a lower symmetry to form orthorhombic, rhombohedral, tetragonal or monoclinic structures. However, as research moves into the nanoworld, small and/or heavily faulted perovskite crystals are often encountered that cannot be studied by X-ray or neutron diffraction. The electron microscope is thus well adapted, not only for imaging these fine-grained materials but also for identifying or characterizing their crystallographic structure by means of electron diffraction. A major difficulty is then encountered due to this symmetry lowering: that is many zone axis diffraction patterns are very similar to each other and cannot be indexed without ambiguity with conventional electron diffraction. Nevertheless, the correct identification of a zone axis pattern (ZAP) as well as the correct attribution of hkl indices to each diffracted beam are required in many material science fields, for example the characterization of crystal defects (stacking faults, dislocations, twins. . .). It is also required in electron crystallography in order to identify many crystallographic features (the crystal system, the Bravais lattice, the Laue class and the point and space groups). One solution to overcome this difficulty consists in using convergent-beam electron diffraction or large-angle convergent-beam electron diffraction which gives more accurate and useful line patterns but these patterns can be very complex and their interpretation is usually tedious and time consuming. In addition, these methods require a specimen of optimal thickness and, for the large-angle convergentbeam electron diffraction technique, relatively large and nondistorted crystals (Morniroli, 2002a). Recently, the electron precession method, which was proposed by (Vincent & Midgley, 1994) became commercially available and can be fitted on most modern transmission electron microscopes. In this technique (Fig. 1a), a parallel or a nearly parallel incident beam is rapidly rotated by means of the pre-specimen deflection coils of the microscope on the surface of a hollow cone whose axis is directed along the optical axis and whose semi-angle , in the range 0 to 3 , is the precession angle. When a [uvw] zone axis of the studied crystal is set as close as possible along the optical axis, the diffraction pattern located in the back focal plane is made of circles, which depending on the precession angle, are more or less superimposed (Fig. 1b). The transmitted circle gives the transmitted intensity as a function of the orientation of the incident beam during the

precession movement whereas each hkl diffracted circle gives the diffracted intensity as a function of the orientation of the incident beam with respect to the corresponding (hkl) lattice planes. This circle pattern is transformed into a spot pattern by means of the post-specimen deflection coils which act in a synchronized way and in opposite direction with respect to the pre-specimen deflection coils. The precession patterns, thus observed on the microscope screen (Fig. 1c) look similar to selected-area electron diffraction patterns but they display four main advantages: (1)The diffracted intensity of each hkl spot is integrated over the corresponding hkl circle present in the back focal plane, i.e. along a large deviation domain on each side of the exact Bragg conditions. As a matter of fact, this integrated intensity is directly connected with the area located under the rocking curve (the diffracted intensity as a function of the deviation from the exact Bragg condition). As a result, the integrated intensities are not very sensitive to a slight crystal misorientation and a [uvw] zone axis precession pattern always looks well aligned even if the [uvw] crystal zone axis is not exactly located parallel with the incident beam. This also means that the intensities of the diffracted beams can be taken into account and trusted when analysing a precession pattern. Thus, the ideal symmetry of the precession patterns becomes available. This symmetry takes into account both the position and the intensity of the diffraction beams present on a pattern. It can be observed for both the zero-order Laue zone (ZOLZ) and the whole pattern (WP) 1 reflections and it is connected with the Laue class (Morniroli & Steeds, 1992). As will be illustrated in Section 4, very small differences of intensities can be observed on precession patterns, which can be used to identify the presence or the absence of symmetry elements and to surely infer this ideal symmetry. It is a crucial and critically important advantage with respect to selected area-electron diffraction patterns or microdiffraction patterns whose diffracted intensities are extremely sensitive to slight misorientations and usually only provide the net symmetry (the symmetry which only considers the position of the reflections). (2)The precession patterns display a larger number of reflections in the ZOLZ and in the high-order Laue zones (HOLZs) than a selected-area diffraction pattern and this number increases with an increase in precession angle. This important property is very useful in electron crystallography (Morniroli & Redjaimia, 2007; Morniroli et al., 2007).

A WP displays the ZOLZ and at least one HOLZ. Throughout this paper, the notations

for the ZOLZ and WP ideal symmetries will be in accordance with (Morniroli & Steeds, 1992); they are underlined and given between parentheses for the ZOLZ.

Journal compilation

C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

CONTRIBUTION OF ELECTRON PRECESSION

Fig. 1. Schematic description of the electron precession technique. (a) Electron ray-path in the column of a transmission electron microscope. For the sake of simplicity, only one hkl diffracted beam is drawn and the column contains only one intermediate lens and no projector lens. (b) Circle pattern observed in the back focal plane of the objective lens. (c) Spot pattern observed on the microscope screen.

(3)During the precession movement, the incident beam is never aligned along the zone axis where the strongest dynamical interactions occur. Thus, the precession patterns are less dynamical than the conventional diffraction patterns. (4)A few-beam behaviour is encountered with large precession angles. Because the multiple diffraction paths to the forbidden reflections are unlikely to occur when this few-beam behaviour prevails, forbidden reflections disappear or become very weak on large-angle precession

patterns. This property allows the identification of the kinematical forbidden reflections (Morniroli & Redjaimia, 2007). The purpose of this paper is to demonstrate that these remarkable and useful precession features can be used to identify, without ambiguity, any [uvw] zone axes and to assign the correct hkl indices to the diffracted beams of perovskite crystals displaying some small symmetry departures. To illustrate this possibility, LaGaO 3 perovskite specimens were selected and observed. LaGaO 3 perovskite is a

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

10

J.-P. MORNIROLI ET AL.

Fig. 2. Description of the ideal cubic (a) and orthorhombic LaGaO 3 (b) perovskites.

good example because both its room and high-temperatures crystal structures exhibit a small departure from a cubic symmetry to an orthorhombic and to a rhombohedral symmetry, respectively. Some applications to the Sr and Mg doped LaGaO 3 perovskites (LSGM perovskites) will also be given.

Description of the room temperature LaGaO 3 crystal structure The ABO 3 perovskite-type oxides consist of a trivalent transition ion located in a A-site coupled with a trivalent transition metal ion located in a B-site, which forms a BO 6 octahedra (Glazer, 1972; Fig. 2a). The arrangement of the BO 6 octahedra in an ideal cubic perovskite is shown in Fig. 2a. As reported by Marti et al. (1994), Slater et al. (1998) and Lerch et al. (2001), in the room temperature LaGaO 3 perovskite, these octahedra are slightly tilted and rotated (Fig. 2b). The corresponding structure is then described by an orthorhombic unit cell with the space group Pnma. As a matter of fact, this pseudo-cubic LaGaO 3 perovskite displays some very small symmetry departures from the ideal cubic perovskite, which means that its diffraction patterns will be very close to cubic. To quantify this aspect, let us consider a <uvw> zone axis form of the ideal cubic perovskite. In the general case, observed when u = v = w and non-zero, this form contains 48 equivalent [uvw] directions (directions having the same parameter P [uvw] but different orientations because, in the cubic crystal system, P [uvw] is not modified if the u, v, and w signs are positive or negative and if the u, v and w indices are interchanged). Let us consider the case of the <123> zone axes. In the case of the Laue class m3m, it gives two types of non-superimposable and mirror-related diffraction patterns labelled <123> A and <123> B in Fig. 3 where, for the sake of simplicity, only the ZOLZ reflections are shown. With the orthorhombic perovskite, the situation becomes more complex because the 48 <123> equivalent cubic directions are transformed into six orthorhombic zone axis

forms : <133>, <331>, <551>, <115>, <422> and <224> each of them containing eight equivalent directions (in the orthorhombic crystal system, the parameter P [uvw] is not modified if the u, v and w signs are positive or negative but the u, v and w indices cannot be interchanged). As a result, in the Laue class mmm, two sets of six different and mirrorrelated diffraction patterns are obtained as shown in Fig. 3. They differ from the <123> cubic patterns by the presence of weak extra reflections (to be more visible, they are magnified 10 times in Fig. 3), which are located at two different positions: (1)In the middle of the small edge of the parallelograms drawn with respect to the cubic reflections for the <133>, <331>, <511> and <115> zone axis forms. Note that the intensity of the extra reflections is different and typical in each of these four patterns (2)In the middle of both sides of the parallelograms as well in the middle of the parallelograms for the two <422> and <224> zone axis forms. The intensity of the extra reflections is also typical of the zone axis.

How to correctly identify these patterns? With conventional electron diffraction (selected-area electron diffraction or microdiffraction), the diffracted intensities are too strongly modified by dynamical effects (multiple diffraction) and/or by thickness variations and crystal misorientations in the diffracted area, so that only the positions of the reflections and the net symmetry can be trusted. This means that a zone axis cannot be surely identified among the four <133>, <331>, <511> and <115> or between the two <422> and <224> zone axis forms. This is no longer the case with electron precession because the intensity and the ideal symmetry can be taken into account. Thus, the observation of the intensity of these typical additional reflections is the basis of the zone axis identification described in the present paper. To this aim, we describe, the pseudo-cubic LaGaO 3 perovskite with respect to the ideal cubic perovskite and, in analogy with the ordered structures commonly observed
C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

Journal compilation

CONTRIBUTION OF ELECTRON PRECESSION

11

Fig. 3. Description of the diffraction patterns produced by: the 48 equivalent <123> zone axes from the ideal cubic perovskite the corresponding <133>, <331>, <511>, <115>, <422> and <224> zone axes from the orthorhombic perovskite. For the sake of simplicity, only the ZOLZ reflections are shown. To increase their visibility, the weak extra reflections are magnified 10 times.

in metal alloys, we consider the corresponding diffraction patterns as made of fundamental reflections common to the cubic and pseudo-cubic perovskites and of superlattice reflections typical of the new periodicities and thus typical of the pseudo-cubic perovskite. Figure 4 illustrates this analogy. In order to make a direct comparison between the ideal and pseudo-cubic perovskites and to describe the ZAPs with the same [uvw] indices, both structures must be described by means of comparable unit cells. For this reason, the ideal cubic perovskite is not described by its conventional primitive cubic unit cell but by means of a multiple tetragonal unit cell (Fig. 4a) whose contour (Fig. 4b and c) is close to the orthorhombic LaGaO 3 unit cell (Fig. 4a , b and c ). The corresponding reciprocal lattices are

also given in Fig. 4d and d . Both reciprocal lattices display the same fundamental reciprocal nodes whereas the pseudo-cubic one displays additional superlattice nodes, which occur when h + l is odd or when h + l is even and k odd. Throughout the present text, the subscripts c, t, and o will refer to the cubic, multiple tetragonal and orthorhombic unit cells, respectively. In Fig. 5, the main <uvw> zone axis diffraction patterns (those which correspond to the six <001> c , twelve <110> c and eight <111> c ZAPs of the ideal cubic perovskite) are displayed and arranged on a stereographic projection so that their mutual orientations are preserved. These patterns were kinematically simulated by means of the electron diffraction software (Morniroli, 2002b). The patterns from the ideal cubic

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

12

J.-P. MORNIROLI ET AL.

Fig. 4. Comparative description of the ideal cubic and pseudo-cubic LaGaO 3 perovskites. a, b, c, d Projection of the structure (a), projection of the direct lattice (b), direct lattice (c) and reciprocal lattice (d) of the ideal cubic perovskite. a , b , c , d Projection of the structure (a ), projection of the direct lattice (b ), direct lattice (c ) and reciprocal lattice (d ) of the orthorhombic LaGaO 3 perovskite.

perovskite (Fig. 5a) only display fundamental reflections. One notices the absence of kinematical forbidden reflections because the corresponding space group Pm3m does not contain any screw axes or glide planes. As expected, most of the simulated patterns from the pseudo-cubic perovskites (Fig. 5b) display weak additional superlattice reflections. For the sake of clarity, the intensity of these extra reflections is exaggerated (their diameter is magnified 10 times). Some couples of superlattice reflections located on each side of the mirrors of the ideal cubic perovskite display a typical difference in intensity connected with the lowering of symmetry (see, e.g. the couples of reflections (arrowed) on the [111] o , [111] o , [111] o and [111] o ZAPs and located on each side of the pseudo-mirrors m 1 and m 2 ). Some patterns also display kinematical forbidden reflections connected with the glide planes and screw axes of the Pnma space group of the orthorhombic perovskite. With conventional electron diffraction (selected-area electron diffraction or microdiffraction), the experimental observation of these typical intensity differences between some couples of superlattice reflections as well as the identification of forbidden reflections is usually impossible due to multiple diffraction paths connected with the strong dynamical behaviour of electron diffraction. This is possible with electron precession because the precession patterns are less dynamical and they display, at least with large precession angle, a few-beam behaviour.

Experimental procedures Sample preparation for TEM observations LaGaO 3 perovskite specimens were prepared from powders of 99.99% pure La 2 O 3 and Ga 2 O 3 . The powders were weighed and mixed in a ball mill for 48 h in an alcohol slurry. They were then calcined, ball milled for another 24 h and finally sintered at 1200 C in air for 12 h. For LSGM, the raw powders capable of adsorbing H 2 O and CO 2 from the atmosphere were calcined at 1000 C in air and cooled in a dry, nitrogen atmosphere prior to weighing. Powder milling was performed in a vibratory mill for 6 h using stabilized zirconia in propan-2-ol. Milled powders were dried in a vacuum oven at 6080 C for 1012 h, and then passed sequentially through 300-, 150- and 75-m stainless steel sieves. After final sieving the powders were calcined at the required temperature (10001200 C) in air and then remilled. After passing through the 75-m sieve, the powder was uniaxially pressed into either bar shapes for electrical conductivity evaluation or pellets for phase assemblage and microstructure characterization. The green specimens were isostatically pressed then fired at 14001450 C for 15 h in air. The resulting LaGaO 3 and LSGM powders were pressed into pellets at 200 kPa. Crushed pellets were mixed with a
C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

Journal compilation

CONTRIBUTION OF ELECTRON PRECESSION

13

Fig. 5. Schematic description of the main zone axis patterns for the ideal cubic (a) and pseudo-cubic (b) LaGaO 3 perovskites. For the cubic unit cell, the subscript c refers to the conventional cubic unit cell. The subscript o is used for the indices of the orthorhombic LaGaO 3 unit cell. Mirrors present on the patterns are indicated by bold lines and pseudo-mirrors by dotted lines.
C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

14

J.-P. MORNIROLI ET AL.

20-m aluminium powder with a ratio of about 10% perovskite and 90% metal powder. A few tens of grams of this powder mixture was then placed between two flat iron sheets and laminated with a rolling mill at a pressure near the maximum of the press in order to obtain a metal foil of thickness of about 50 m in which the crushed perovskite crystals are embedded. Three millimetre discs were then punched from the foil with a Gatan disc punch (Gatan, Pleasanton, CA, USA). These thin foils were ion beam thinned on a Gatan 600 DuoMill at room temperature for 2 h with both the upper and lower Ar+ ion beams operated at 4 kV, with a 10 mA ion current per gun and an angle of incidence of 10 both above and below the foil, until the foil was electron optically transparent. These specimens have a much better quality than the crushed specimens deposited on a carbon film where the transparent areas are usually very small because they result from a cleavage mechanism which produces some rapid variations of the specimen thickness. On the contrary, the ion thinning of the Al/perovskite foils produces relatively large electronoptically transparent areas with relatively small thickness variations of the embedded crystals. Because aluminium is a very malleable metal, these specimens are not brittle and can be observed many times. In addition, aluminium is relatively transparent to electrons meaning that there is no risk of confusion with the studied perovskite crystals. Aluminium is also a very good thermal conductor well adapted to high temperature experiments up to 500 C. Transmission electron microscopy Room temperature experiments on the perovskite samples were performed with a Philips CM30 TEM operated at 300 kV and equipped with the electron precession Spinning Star equipment from Nanomegas (Brussels, Belgium). The specimens observed in the present study are usually heavily faulted and contain many twins. In order to avoid artefacts due to these defects and to probe only defect free areas, the precession patterns were obtained in microdiffraction mode. In this mode, the electron beam is a nearly parallel electron beam produced by a 10-m C 2 condenser aperture. This beam is focused on the specimen with a probe diameter of between 10 and 50 nm. All the precession patterns were recorded on a 1k 1k Gatan CCD camera. An electron micrograph of a LaGaO 3 grain coming from a thin area of the sample is shown in Fig. 6. It shows typical contrast connected with approximately 1020 nm domain sizes. Experimental results All the ZAPs considered in the present study only display the ZOLZ and therefore only give the ZOLZ ideal symmetry. The main zone axes of the ideal cubic perovskite are described, i.e. the <001> c , <110> c and <111> c ZAPs.

Fig. 6. Typical electron micrograph of a LaGaO 3 grain.

Identification of the [uvw] zone axis from precession patterns <001> c zone axes As shown in the theoretically simulated patterns in Fig. 7, the six equivalent <001> c zone axes of the ideal cubic perovskite with ideal (4 mm) symmetry (Fig. 7a) give, in the orthorhombic perovskite, two different patterns: (1)The [010] o and [010] o patterns (type A) with (2 mm) symmetry. They display additional superlattice reflections located in the centre of the square drawn with respect to the fundamental reflections. Typical kinematical forbidden reflections are also located along the two remaining mirrors m 3 and m 4 (Fig. 7b). (2)The [101] o , [101] o , [101] o and [101] o patterns (type B) with (2 mm) symmetry where the additional reflections are located along one of the edges of the square. Some kinematical forbidden reflections are located along its mirror m 2 (Fig. 7c). There is no special difficulty to identify these two A and B types of zone axes by electron precession (Fig. 7b and c ) even with conventional electron diffraction (Fig. 7b and c ) because the position of the superlattice reflections on both types are very different and cannot be modified by multiple diffraction. One notices that the kinematical forbidden reflections are clearly visible on all these patterns. Nevertheless, we can use electron precession to detect these forbidden reflections. As indicated in Section 1, the kinematical forbidden reflections can be identified by using a large precession angle (3 was found to be a good value); with these experimental conditions, a few-beam behaviour is observed during the precession movement of the incident beam so that the probability of having multiple diffraction paths to the forbidden reflections is unlikely to occur. With these conditions, a forbidden reflection disappears or becomes
C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

Journal compilation

CONTRIBUTION OF ELECTRON PRECESSION

15

Fig. 7. Electron diffraction patterns of the six <001> c ZAPs for the pseudo-cubic perovskite. They are sorted into two types of patterns A and B. a Kinematical simulated pattern, with (4 mm) symmetry, of the six equivalent <001> c for the ideal cubic perovskite. b, b , b [010] o and [010] o diffraction patterns with (2 mm) symmetry (type A). Kinematical simulated patterns (b), experimental precession pattern (b ) and experimental selectedarea electron diffraction pattern (b ). The forbidden reflections are visible on the experimental patterns. c, c , c [101] o , [101] o , [101] o and [101] o diffraction patterns with (2 mm) symmetry (type B). Kinematical simulated pattern (c), experimental precession pattern (c ) and experimental selectedarea electron diffraction pattern (c ). The forbidden reflections are visible on the experimental patterns. d, d Experimental precession pattern of type A obtained with a 1 precession angle (d) and corresponding dynamical simulation performed with jEMS (d ). The kinematical forbidden reflections (circled reflections) are visible on both patterns. e, e Experimental precession pattern of type A obtained with a 3 precession angle (e) and corresponding dynamical simulation performed with jEMS (e ). The kinematical forbidden reflections are invisible on both patterns. f, f Experimental precession pattern of type B obtained with a 1 precession angle (f) and corresponding dynamical simulation performed with jEMS (f ). The kinematical forbidden reflections (circled reflections) are visible. g, g Experimental precession pattern of type B obtained with a 3 precession angle (g) and corresponding dynamical simulation performed with jEMS (g ). The kinematical forbidden reflections are invisible. Mirrors present on the simulated patterns are indicated by bold lines.

very weak. This is what is observed on the patterns in Fig. 7e and g, which correspond to the patterns with types A and B. One notices that these patterns are in excellent agreement with the corresponding dynamical simulations (Fig. 7d , e , f and g ) performed with the jEMS software (Stadelmann, 1987, 2007). <110> c zone axes The 12 equivalent <110> c zone axes of the cubic perovskite (Fig. 8a), with symmetry (2 mm), give four different diffraction patterns (Fig. 8be) in the orthorhombic LaGaO 3 perovskites:

(1)The [001] o and [001] o patterns (type C) with (2 mm) symmetry. The superlattice reflections are located on the small edge of the rectangle drawn with respect to the fundamental reflections. Some forbidden reflections are located along the m 2 mirror (Fig. 8b). (2)The [111] o , [111] o , [111] o and [1 1 1] o patterns (type D 1 ) with (2) symmetry. Some superlattice reflections are located along the diagonal d 1 and some forbidden reflections along the other diagonal d 2 (Fig. 8c).

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

16

J.-P. MORNIROLI ET AL.

Fig. 8. Electron diffraction patterns of the twelve <110> c ZAPs for the pseudo-cubic perovskite. They are sorted into four types of patterns C, D 1 , D 2 and E. a Kinematical simulated patterns of the twelve equivalent <110> c , with (2 mm) symmetry, for the ideal cubic perovskite. b, b [001] o and [001] o diffraction patterns with (2 mm) symmetry (type C). Kinematical simulated pattern (b) and experimental precession pattern (b ). The forbidden reflections are visible on the experimental pattern. c [111] o , [111] o , [111] o and [1 1 1] o kinematical simulated patterns with (2) symmetry (type D 1 ). d [111] o , [111] o , [11 1] o and [111] o kinematical simulated patterns with (2) symmetry (type D 2 ). e [100] o and [100] o kinematical simulated patterns with (2 mm) symmetry (type E). f Experimental precession pattern performed with a 1 precession angle. It is in agreement with the types D 1 , D 2 or E. g, g Experimental precession patterns performed with 1 (g) and 3 (g ) precession angles. The absence of kinematical forbidden reflections along the d 1 diagonal in figure g (circled reflections) proves that this pattern belongs to the D 2 type. h, h Experimental precession patterns performed with 1 (g) and 3 (g ) precession angles. This pattern belongs to the type E because no forbidden reflections are identified on the 3 precession pattern. The bold lines indicate mirrors.

(3)The [111] o , [1 11] o , [11 1] o and [1 1 1] o patterns (type D 2 ) with (2) symmetry. The superlattice and forbidden reflections also are along the d 1 and d 2 diagonals but these patterns are mirror related with respect to the patterns of type D 1 (Fig. 8d). (4)The [100] o and [100] o patterns (type E) with (2 mm) symmetry. Some superlattice reflections are located along both the d 1 and d 2 diagonals. Forbidden reflections are not present on these zone axes (Fig. 8e). Type C patterns are easily distinguished from the three other types D 1 , D 2 and E because the position of the superlattice reflections on these patterns is typical. Thus, the experimental precession pattern in Fig. 8b belongs to this type. On the other hand, the types D 1 , D 2 and E are very difficult to be distinguished from each other by conventional methods, or in low angle precession patterns because the forbidden reflections

will appear by multiple diffraction (Fig. 8f). Actually, they can be distinguished by large-angle electron precession as shown on the examples in Fig. 8g, g , h, and h . Large-angle precession patterns prove that there are no forbidden reflections in Fig. 8h , whereas kinematical forbidden reflections are identified along the d 1 diagonal in Fig. 8g . Therefore, the precession patterns in 8g and 8h belong to the types D 2 and E, respectively. Another way to make the distinction among the three E, D 1 and D 2 patterns consists of tilting the specimen along the m 1 and m 2 mirrors or pseudo-mirrors until some more typical patterns are observed and to compare them with the corresponding simulated patterns. The two sets of experimental patterns shown in Fig. 9a and b were obtained in that way. Each set display some strong differences connected with the positions or the intensity of the superlattice reflections
C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

Journal compilation

CONTRIBUTION OF ELECTRON PRECESSION

17

Fig. 9. Sets of electron precession patterns located along the m 1 and m 2 mirrors around the <110> c ZAPs. a, b Experimental precession patterns. Some typical superlattice reflections are circled. c, d, e Corresponding simulated patterns located around the [100] o (or [100] o ) (type E) (c), [111] o (or [111] o , [111] o , [1 1 1] o ) (type D 1 ) (d) and [111] o (or [1 11] o , [11 1] o , [111] o ) (type D 2 (e). The arrows indicate some typical features. The bold and dotted lines indicate mirrors and pseudo-mirrors, respectively.

(see especially the circled superlattice reflections which prove the presence of two m 1 and m 2 mirrors in Fig. 9a and the loss of these mirrors in Fig. 9b). The two sets are, in a unique way, in perfect agreement with the corresponding simulated patterns from type E and D 1 , respectively. Note that the simulated sets of precession patterns also indicate a 2 mm, 1 and 1 WP ideal symmetry for the types E, D 1 and D 2 , respectively. <111> c ZAPs The eight equivalent <111> c ZAPs of the ideal cubic perovskite (Fig. 10a) with (6 mm) symmetry give two types of patterns in the orthorhombic perovskite: (1)The [012] o , [012] o , [012] o and [01 2] o patterns (type F) with (2 mm) symmetry (Fig. 10b). (2)The [2 10] o , [2 10] o , [2 10] o and [210] o patterns (type G) with (2 mm) symmetry (Fig. 10c). Both types display forbidden reflections and very weak superlattice reflections located at the same positions. They also display the two same mirrors m 1 and m 4 meaning that the four other m 2 , m 3 , m 5 and m 6 mirrors present in the cubic perovskite are lost in the orthorhombic perovskite and are only pseudo-mirrors. These two types of patterns are too

close to be easily identified even with electron precession. To make the distinction between them, the solution consists again in tilting the specimen along the m 1 mirror and the two pseudo-mirrors m 3 and m 5 in order to reach some more useful ZAPs like the ones shown in Fig. 10b , b , b and c , c , c . The positions and the intensity of some superlattice reflections present on these patterns is very typical of the types F and G which can be identified without ambiguity in that way (see, especially, the superlattice reflections indicated by an arrow on the patterns b , b , c and c ). It is clear that the experimental precession patterns shown in Fig. 10d and d display superlattice reflections (marked with an arrow) in agreement with type F patterns and in disagreement with type G patterns. Note that the WP symmetry of the type F and G patterns is m. Identification of the other <uvw> c zone axes Any other <uvw> c zone axes can always be correctly indexed with respect to the three main <100> c , <110> c and <111> c zone axes described above.

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

18

J.-P. MORNIROLI ET AL.

Fig. 10. Electron diffraction patterns of the eight <111> c ZAPs for the pseudo-cubic perovskite. They are sorted into two types of patterns F and G. a Kinematical simulated patterns of eight equivalent <111> c ZAPs for the ideal cubic perovskite. They display a (6 mm) symmetry. b [012] o , [012] o, [012] o and [01 2] o kinematical simulated diffraction patterns with (2 mm) symmetry (type F). c [210] o , [210] o, [2 10] o and [210] o kinematical simulated diffraction patterns (type G) with (2 mm) symmetry. b , b , b Simulated diffraction patterns obtained after a 19.4 specimen tilt along the m 1 , m 3 and m 5 mirror and pseudo-mirrors of the pattern of type F. c , c , c Simulated diffraction patterns obtained after a 19.4 specimen tilt along the m 1 , m 3 and m 5 mirror and pseudo-mirrors of the pattern of type G. d, d , d , d - Experimental precession patterns in agreement with the simulated patterns b, b , b and b (type F). The bold and dotted lines indicate mirrors and pseudo-mirrors, respectively.

Interpretation of a <uvw> o ZAP To fully interpret a [uvw] zone axis diffraction pattern, hkl indices must be attributed to each of the diffracted spots. This attribution is not always an easy task with the pseudo-cubic LaGaO 3 perovskite due to symmetry lowering. For example, let us consider the case of the [010] o (or [010] o ) pattern (type A) (Fig. 11a). The (4 mm) symmetry (mirrors m 1 , m 2 , m 3 and m 4 ) which would be observed on this pattern for the ideal cubic perovskite is decreased to a (2 mm) symmetry (mirrors m 2 and m 4 ) for the pseudo-cubic LaGaO 3 perovskite. As a result, subtle and very weak differences of intensities affect some couples of superlattice reflections located on each side of the two pseudomirrors m 1 and m 3 . They are visible on the theoretical pattern in Fig. 11a (see, e.g. the couples of reflections marked with an arrow), but they are too weak to be surely distinguished on the corresponding experimental precession pattern. This means that it is nearly impossible to make an experimental

distinction between two [010] o patterns rotated by 90 . To remove this ambiguity, the solution is to observe more typical ZAPs located around the studied pattern (a solution already described earlier to make the distinction between the <110> c and <111> c zone axes). This is the case with the three patterns shown in Fig. 11bd located at about 18.5 from [010]. A clearly visible difference of intensity is observed between some couples of superlattice reflections (see the circled reflections in Fig. 11) located on each side of the pseudo-mirrors m 1 and m 3 . These patterns are in agreement in a unique way with the [131] o , [131] o and [131] o dynamical simulations in Fig. 11b , c d (or [131] o , [13 1] o , [1 3 1] o if the pattern in Fig. 11a is interpreted as being the [010] ZAP). Therefore, the hkl indices of the [010] o ZAP (or [010] o ZAP) can be attributed without ambiguity (Fig. 11a and a ). A second example concerns the [101] o ZAP (or [101] o , [101] o and [1 0 1] o ) (type B) (Fig. 12a). This pattern displays

Journal compilation

C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

CONTRIBUTION OF ELECTRON PRECESSION

19

Fig. 11. Interpretation of the [010] o (or [010] o ) zone axis pattern (type A). a, a [010] o (or [010] o ) experimental (a) and simulated (a ) precession patterns with (2 mm) symmetry. b, b [131] o (or [131] o ) experimental (b) and simulated (b ) precession patterns with (2) symmetry. c, c [131] o (or [13 1] o ) experimental (c) and simulated (c ) precession patterns with (2) symmetry. d, d [131] o (or [1 3 1] o ) experimental (d) and simulated (d ) precession patterns with (2) symmetry. The bold and dotted lines indicate mirrors and pseudo-mirrors, respectively.

a (2 mm) symmetry with two mirrors m 1 and m 3 whereas it would display a (4 mm) symmetry (m 1 , m 2 , m 3 and m 4 mirrors) in the ideal cubic perovskite. As a matter of fact, its WP symmetry is not (2 mm) but only m with a unique mirror m 1 meaning that two [101] o ZAPs rotated by 180 are different. This 180 ambiguity can be removed by tilting the specimen around the [101] o and along the m 1 mirror and m 3 pseudo-mirror in order to observe three more typical ZAPs like the ones shown in Fig. 12bd. Again, a clear difference of intensity among some couples of superlattice reflections located on each sides of the pseudo-mirror m 3 is visible (see the circled and arrowed reflections in Fig. 12). Note that the couples of superlattice reflections located on each side of the m 1 mirror display the same intensity. The comparison with simulated patterns allows the unique attribution of the [313] o , [201] o and [313] o indices to the experimental patterns b,

c and d, respectively. Thus, the correct hkl indices can be assigned to the reflections of the [101] o pattern as shown in Fig. 12a . Applications Identification of pertinent ZAPs for electron crystallography The identification of both the Laue class and the possible space groups of a crystal can be obtained, at microscopic and nanoscopic levels, from observations of three features available on conventional microdiffraction patterns: (1)The net and ideal symmetries displayed by some specific microdiffraction patterns provided the diffraction patterns display at least one HOLZ. They are connected with the crystal system and the Laue class.

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

20

J.-P. MORNIROLI ET AL.

Fig. 12. Interpretation of the [101] o (or [101] o , [101] o , [101] o ) zone axis pattern (type B). a, a [101] o (or [101] o , [101] o , [101] o ) experimental (a) and simulated (a ) precession patterns with (2 mm) symmetry. b, b [313] o (or [3 13] o , [31 3] o , [313] o ) experimental (b) and simulated (b ) precession patterns with (2) symmetry. c, c [201] o (or [210] o , [201] o , [201] o ) experimental (c) and simulated (c ) precession patterns with (2 mm) symmetry. d, d [313] o (or [313] o , [313] o , [3 13] o ) experimental (d) and simulated (d ) precession patterns with (2) symmetry. The bold and dotted lines indicate mirrors and pseudo-mirrors, respectively.

(2)The shifts between the reflections located in the first-order Laue zone (FOLZ) with respect to the ones located in the ZOLZ. They are connected with the Bravais lattice. (3)The periodicity differences between the reflections located in the FOLZ and in the ZOLZ. They are connected with the glide planes. A systematic method based on these features was proposed by (Morniroli & Steeds, 1992) and was successfully applied to various crystals (Redjaimia & Morniroli, 1994; Mateo et al., 1997; Huv e et al., 2000; Wei et al., 2000; GomezHerrero et al., 2001; Ranjan et al., 2001; Meshi et al., 2002; Quarez et al., 2003; Tarakina et al., 2003; Labidi et al., 2005; Meshi et al., 2005). Nevertheless, some experimental difficulties are encountered with the ideal symmetry because its

identification requires a perfect alignment of the incident beam along a zone axis. Other difficulties occur when the studied crystal is not thin enough so that its patterns only display a small number of reflections in the HOLZs. The FOLZ/ZOLZ shifts and periodicity differences are then very difficult to observe. These difficulties are easily overcome with electron precession mainly because the integrated intensities on the precession patterns is directly connected with the ideal symmetry and because the number of reflections present in each of the Laue zones is larger than the one encountered on conventional diffraction patterns. In the present case of the orthorhombic LaGaO 3 , this method requires one to observe the three following ZAPs: <100> o (patterns with type E), <010> o (patterns with type A) and
C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

Journal compilation

CONTRIBUTION OF ELECTRON PRECESSION

21

Fig. 13. Identification of the LaGaO 3 space group from electron precession patterns. a [100] o zone axis precession pattern. Only the ZOLZ reflections are visible and can be characterized by mean of a centred rectangular unit cell with sides parallel to the m 1 and m 2 mirrors. b, c Electron precession patterns obtained when the crystal is tilted a few degrees away from [100] o along the m 1 and m 2 mirrors. The FOLZ reflections are characterized by means of a rectangular unit cell and some FOLZ reflections are located on the m 1 and m 2 mirrors (circled reflections). d Theoretical [100] o ZAP obtained in the case of a primitive orthorhombic lattice (oP) with a diagonal n glide plane parallel to the (100) lattice planes. The corresponding partial extinction symbol is Pn.. .

<001> o (patterns with type C). As mentioned previously, there is no difficulty to identify the <010> o and <001> o ZAPs because the corresponding patterns display typical superlattice reflections. This is no longer the case with the <100> o ZAP (type E) which can be easily confused with the patterns with the types D 1 or D 2 . We have also indicated previously how to identify this zone axis by using a large-angle precession pattern or by specific specimen tilts. The pattern in Fig. 13a was identified in this way. Nevertheless, its FOLZ is not visible because the FOLZ radius is too large for the acceptance angle of the microscope. To observe it, the specimen is tilted a few degrees away from the

zone axis along the mirror m 1 and m 2 until some reflections located in the FOLZ are observed as shown in Fig. 13b and c. Two useful features are visible on these misoriented patterns: (1)Some reflections are located on the mirrors m 1 and m 2 (circled reflections); they reveal the absence of shift of the FOLZ reflections with respect to the ZOLZ reflections in agreement with a primitive P Bravais lattice. (2)More reflections are observed in the FOLZ than in the ZOLZ. This feature can be easily quantified by drawing, in the FOLZ and in the ZOLZ, the smallest rectangular unit cell describing the 2D lattice of reflections. In the FOLZ, this unit cell is a rectangle whose sides are parallel to the m 1 and

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

22

J.-P. MORNIROLI ET AL.

m 2 mirrors whereas in the ZOLZ the unit cell is a centred rectangle whose sides are two times larger. This FOLZ/ZOLZ periodicity difference is connected with the presence of a glide plane perpendicular to the [100] zone axis, i.e. along the (100) lattice planes. Comparison with the theoretical pattern in Fig. 13d (Morniroli & Steeds, 1992) indicates that these features correspond, in a unique way, to a n glide plane parallel to the (100) lattice planes and to a partial extinction symbol Pn.. (Hahn, 2002). The same experiments performed along the two other [010] and [001] zone axes allow the identification of the Pnma space group for the LaGaO 3 structure. Identification of a twin Due to symmetry lowering, twins are very frequently observed in perovskites and were described by many authors in the case of LaGaO 3 (Wang et al., 1991; Yao et al., 1991; Fink-Finowicki et al., 1992; Bdikin et al., 1993; Wang & Lu, 2006; Wang & Lu, 2007). The identification of the twin law, i.e. the identification of a rotation axis [uvw] and a rotation angle around it, can be inferred from diffraction patterns (Morniroli & Gaillot, 2000). In principle, two couples of parallel [uvw] A //[uvw] B directions coming from the two crystals A and B located on each side of the studied twin need to be identified. It could be two couples of [uvw] zone axes or one couple of zone axes [uvw] and one couple of parallel lattice planes (hkl). The knowledge of the correct [uvw] or hkl indices are crucial for this identification. Let us consider the diffraction pattern in Fig. 14a. It was obtained when the electron beam is focused on a LaGaO 3 twin and it is made of the superimposition of a [010] o ZAP (type A) and a [101] o ZAP (type B). In the previous sections, we indicate that 90 and 180 ambiguities exist for these [010] o and [101] o patterns, respectively, and that these ambiguities can be removed by observing, on some ZAPs located around the [010] o and [101] o , some typical superlattice reflections located on each side of the pseudo-mirrors present on the patterns. Actually, the patterns in Figs 11 and 12 were obtained with the incident beam located on each side of the studied twin. Both figures are arranged in a coherent way so as to preserve their mutual orientation. The correct [uvw] and hkl indices of these patterns were demonstrated in the previous sections. From these patterns, many couples of parallel [uvw] ZAPs can be obtained, for example [010] oA //[101] oB and [131] oA //[201] oB , which give the following twin law: a 120 rotation around the common rotation axis [210]. This result is obtained from the mathematical calculation described by Morniroli & Gaillot, 2000. A schematic description of this twin with {121} twin plane is given in Fig. 14b. It is in agreement with previous studies.

Fig. 14. Characterization of a LaGaO 3 twin. a Electron precession pattern obtained when the incident beam is located on the twin. It is made of the superimposition of the [010] o and [101] o ZAPs. b Schematic description of the twin. The two lattices A and B on each side of the twin plane (121) oA //(121) oB are rotated by a 120 angle around the common direction [210] o .

Another way to obtain the same twin law is to consider the pattern in Fig. 14a, where in addition to the couple of parallel ZAPs [010] oA //[101] oB it also gives (101) oA //(020) oB . It is pointed out, that the correct interpretation of the [010] and [101] ZAPs is required to identify the actual twin law. Especially, the 90 and 180 ambiguity for the [010] o and [101] o ZAPs should be taken into account. If not, wrong twin laws like 90 around [101] or 180 around [111] are obtained. Identification of the crystal structure at microscopic and nanoscopic level The comparison between experimental precession patterns and simulated patterns allows an unambiguous crystal
C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

Journal compilation

CONTRIBUTION OF ELECTRON PRECESSION

23

Fig. 15. Identification of the LSGM space group. a Experimental precession patterns located around a <001> c ZAP. b, c Simulated diffraction patterns located around the [010] o (or [010] o )(a) and [101] o (or [101] o , [101] o , [101] o ) (b) for the orthorhombic Imma space group. d, e, f Simulated diffraction patterns located around the [010] o (or [010] o )(d), [101] o (or [101] o ,) (e) and [101] o (or [101] o ) (f) for the monoclinic I112/b space group. The mirrors are indicated by a bold line and the pseudo-mirrors by a dotted line.

identification, especially if the weak superlattice reflections are taken into account. In the case of LaGaO 3 there is no special difficulty because most of the authors agree with the orthorhombic structure. As a matter of fact, all the experimental patterns from our specimens are in perfect agreement with this structure. This is no longer the case with the LSGM perovskite. At room temperature, two different structures based on X-ray and neutron diffraction experiments were proposed: (1)An orthorhombic structure with space group Imma (Lerch et al., 2001). (2)A monoclinic structure with space group I112/b (Slater et al., 1998). Both of them are very close and only differ by some very slight changes and by different loss of symmetry elements (there is less symmetry elements in the monoclinic structure than in the orthorhombic). The identification of the real LSGM space group can be obtained from a careful examination of the superlattice reflections present on some diffraction patterns located around a <001> c ZAP (Fig. 15a). In this figure, it is clear that all

the couples of reflections (circled and arrowed spots) located on each side of the m 1 , m 2 , m 3 and m 4 mirrors of the ideal cubic structure display a clear difference of intensity, which proves that these patterns do not contain any mirror. This absence of mirror is in agreement in a unique way with the corresponding simulated patterns located around the [101] m (or [101] m ) ZAP from the monoclinic I112/b space group (Fig. 15e). It is in disagreement with all the other <001> c theoretical patterns from the two possible space groups I112/b and Imma. As a result, the LSGM space group is identified as I112/b.

Discussion This analysis is based on a qualitative observation of the symmetry elements present on the precession patterns and the deduction of the ZOLZ and WP ideal symmetry. Table 1 summarizes these ideal symmetries for the main zone axes of the cubic perovskite as well as the resulting symmetries in the orthorhombic perovskite. A 180 ambiguity occurs with the patterns displaying an m or 1 symmetry. A special

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

24

J.-P. MORNIROLI ET AL.

Table 1. ZOLZ and WP ideal symmetries for the main zone axis patterns of the cubic and the orthorhombic LaGaO3 perovskites. Ideal cubic LaGaO 3 perovskite Laue class m-3 m Ideal symmetry ZOLZ WP Orthorhombic LaGaO 3 perovskite Laue class mmm Ideal symmetry ZOLZ WP <111> cs <100>c <110>c

(6 mm) 3m Type F [012] o [01-2] o [0-12] o [0-1-2] o (2 mm) m

Type G [210] o [-210] o [-2-10] o [2-10] o (2 mm) m

(4 mm) 4 mm Type A [010] o [0-10] o

Type B [101] o [-10-1] o [-101] o [10-1] o (2 mm) m

(2 mm) 2 mm Type C [001] o [00-1] o

Type D 1 [1-11] o [11-1] o [-111] o [-1-1-1] o (2) 1 s 180 ambiguity

Type D 2 [111] o [-1-11] o [1-1-1] o [-11-1] o (2) 1

Type E [100] o [-100] o

(2 mm) 2 mm very close to (4 mm) 4 mm 180 ambiguity

(2 mm) 2 mm

(2 mm) 2 mm

180 ambiguity

90 ambiguity

180 ambiguity

90 ambiguity is observed for the [010] and [010] zone axes because their corresponding 2 mm symmetry is very close to 4 mm. The ideal symmetry is easily observed on the experimental precession patterns. This is a major advantage with respect to the conventional electron diffraction where only the less useful net symmetry is usually inferred with surety. On most patterns, only the ZOLZ reflections are considered to identify the ZOLZ ideal symmetry and the WP ideal symmetry was deduced from specific specimen tilt experiments. Another way to get this WP symmetry is to consider the HOLZ reflections. This approach is also possible but it is usually more complex to perform especially with high-symmetry zone axes which have a HOLZ radius too large to be accepted by the microscope. Using electron precession to identify the zone axis could be applied to any zone, but it is pertinent to use high-symmetry patterns like <001> c , <110> c or <111> c because these patterns are easily recognizable and these zone axes can be used as a starting point to reach any other less symmetrical zone axes. The transition from one high-symmetry ZAP to another one less symmetrical is easily made by observing the Kikuchi lines which are visible on the corresponding convergent-beam electron diffraction patterns. Some experimental difficulties are connected with twins, which are very frequent in these materials. These twins are only visible for some particular crystal orientations and the contrast differences between the two crystals located on each side of the twin is usually weak. As a result, twin free areas are not easy to locate and have a small size of about 0.1 m so that the analysis requires very accurate positioning of the incident beam. Other limitations are connected with the possible tilt angles of the specimen holder. The double tilt specimen holder used in the present study allows 45 and 30 and tilt angles,

respectively. The required zone axes should be located within these angular domains. We indicate that the kinematical forbidden reflections can be identified on large-angle precession patterns (about 3 ) because a few-beam behaviour prevails, which strongly decreases the possibility of multiple diffraction. Nevertheless, most of the precession patterns given in the present paper were obtained with a precession angle of about 1 . The alignments are then easy to perform even for very small spot sizes down to 50 nm and the resulting patterns display a high quality. With large precession angle (the maximum value is about 3 with our precession equipment) the perfect scan and descan alignments are more difficult to achieve and the resulting spot size is larger due to spherical aberration so that it could be difficult to focus the beam on a twin free area. Then, the quality of large-angle precession patterns is usually poorer. A microscope equipped with a spherical aberration corrector should allow one to obtain larger precession angles without decreasing the spot size. Most of the simulated patterns given in the present paper result from kinematical calculations, where the intensity of the diffracted beams is connected with the square modulus of the structure factor. These calculations are well adapted to the weak superlattice reflections, which have a kinematical behaviour especially at a large precession angle. The latest version of the jEMS software from Pierre Stadelmann allows one to simulate dynamical precession patterns. They were particularly useful to interpret the effect of the precession angle on the forbidden reflections on Fig. 7. Conclusions Electron precession is very useful for the analysis of these pseudo-cubic perovskites. This property is mainly due to the integrated intensity of the diffracted beams meaning that the diffracted intensities can be taken into account and trusted.
C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

Journal compilation

CONTRIBUTION OF ELECTRON PRECESSION

25

The more useful ideal symmetry is available instead of the net symmetry. Very small differences of intensity are also detectable and are of very great help when dealing with the identification of the presence or the absence of mirrors on the precession patterns. The possibility of identifying, on large precession patterns, the kinematical forbidden reflections due to glide plane or screw axes is also very useful. A method to identify any zone axis without ambiguity is described and some applications of this method to the characterization of twins and to the identification of the space group are given. This method is general and can be easily extended to other perovskites or crystals displaying small symmetry departures from a high symmetry. Acknowledgements We thank Justin Kimpton (RMIT) for supplying the LaGaO 3 and LSGM; the French Embassy and the Australian Science and Technology (FEAST) program for J.-P.M. to travel to UQ (October 2003, February 2006) and for G.J.A. to travel to Lille (July 2005, July 2006) and The University of Queensland for a UQ Travel grant for G.J.A. to Lille (July 2007).
References
Balachandran, U., Ma, B., Maiya, P.S. et al. (1998) Development of mixedconducting oxides for gas separation. Solid State Ion. 108, 363370. Bdikin, I.K., Shmytko, I.M., Balbashov, A.M. & Kazansky, A.V. (1993) Twinning of LaGaO 3 single-crystals. J. Appl. Crystallogr. 26, 7176. Christen, H.M., Boatner, L.A., Budai, J.D., Chisholm, M.F., Gerber, C. & Urbanik, M. (1997) Semiconducting epitaxial films of metastable SrRu 0.5 Sn 0.5 O 3 grown by pulsed laser deposition. Appl. Phys. Lett. 70, 2147-2149. Fink-Finowicki, J., Berkowski, M. & Pajaczkowska, A. (1992) Twinning structure of LaGaO 3 grown by the Czochralski method. J. Mater. Sci. 27, 107110. Glazer, A.M. (1972) Classification of tilted octahedra in perovskites. Acta. Crystallogr B B 28, 33843392. Gomez-Herrero, A., Landa-Canovas, A.R., Johnson, A.W.S. & Otero-Diaz, L.C. (2001) Transmission electron microscopy study of Y 1x x Cr 2 S 4 , x similar to 1/3 phase. J. Alloys Compd. 323, 8690. Hahn, T. (2002) International tables for crystallography. Brief Teaching Edition of Volume A, Space-Group symmetry, 5th edition. Published for the International Union of Crystallography by Kluwer Academic Publishers, Dordrecht, Boston. Huv e, M., Vannier, R.N. & Mairesse, G. (2000) EDS and TEM study of the family of compounds with a structure based on [Bi 12 O 14 ] columns in the Bi 2 O 3 -MoO 3 binary system. J. Solid State Chem. 149, 276283. Kao, W.H., Haberichter, S.L. & Bullock, K.R. (1992) Corrosion-resistant coating for a positive lead-acid-battery electrode. J. Electrochemical Soc. 139, L105L107. Kawada, T., Sase, M., Kudo, M. et al. (2006) Microscopic observation of oxygen reaction pathway on high temperature electrode materials. Solid State Ion. 177, 30813086. Labidi, O., Roussel, P., Huve, M., Drache, M., Conflant, P. & Wignacourt, J. P. (2005) Stabilization of a new < >< /> polymorph in P-

substituted Pb 2 BiVO 6 : single crystal structure of Pb 2 Bi(V 0.84 P 0.16 )O 6 and conduction properties of related materials. J. Solid State Chem. 178, 22472255. Lahousse, C., Bernier, A., Grange, P., Delmon, B., Papaefthimiou, P., Ioannides, T. & Verykios, X. (1998) Evaluation of g-MnO 2 as a VOC removal catalyst: comparison with a noble metal catalyst. J. Catalysis 178, 214225. Lerch, M., Boysen, H. & Hansen, T. (2001) High-temperature neutron scattering investigation of pure and doped lanthanum gallate. J. Phys. Chem. Solids. 62, 445455. Marti, W., Fischer, P., Altorfer, F., Scheel, H. & Tadin, M. (1994) Crystal structures and phase transitions of orthorhombic and rhombohedral RGaO 3 (R = La, Pr, Nd) investigated by neutron powder diffraction. J. Phys. Condens. Matter 6, 127135. Mateo, A., Llanes, L., Anglada, M., Redjaimia, A. & Metauer, G. (1997) Characterization of the intermetallic G-phase in an AISI 329 duplex stainless steel. J. Mater. Sci. 32, 45334540. Meshi, L., Zenou, V.Y., Ezersky, V., Munitz, A. & Talianker, M. (2002) Identification of the structure of a new Al-U-Fe phase by electron microdiffraction technique. J. Alloys Compd. 347, 178183. Meshi, L., Zenou, V., Ezersky, V., Munitz, A. & Talianker, M. (2005) Tetragonal phase in Al-rich region of U-Fe-Al system. J. Alloys Compd. 402, 8488. Minh, N.Q. (1993) Ceramic fuel-cells. J. Am. Ceramic Soc. 76, 563588. Morniroli, J.-P. (2002a) Large-Angle Convergent Beam Diffraction. Applications to Crystal Defects, SF, Paris. Morniroli, J.-P. (2002b) Electron Diffraction, a Software to Simulate Electron Diffraction Patterns. USTL & ENSCL, Lille. Morniroli, J.P. & Gaillot, F. (2000) Trace analyses from LACBED patterns. Ultramicroscopy 83, 227243. Morniroli, J.P. & Redjaimia, A. (2007) Electron precession microdiffraction as a useful tool for the identification of the space group. J. Microsc.(Oxford) 227, 157171. Morniroli, J.-P. & Steeds, J.W. (1992) Microdiffraction as a tool for crystalstructure identification and determination. Ultramicroscopy 45, 219 239. Morniroli, J.P., Redjaimia, A. & Nicolopoulos, S. (2007) Contribution of electron precession to the identification of the space group from microdiffraction patterns. Ultramicroscopy 107, 514522. Quarez, E., Huve, M., Abraham, F. & Mentre, O. (2003) From the mixed valent 6H-Ba 3 Ru 25.5 +NaO 9 to the 6H-Ba 3 (Ru 1.69 C 0.31 )(Na 0.95 Ru 0.05 )O 8.69 oxycarbonate compound. Solid State Sci. 5, 951963. Ranjan, R., Pandey, D., Schuddinck, W., Richard, O., De Meulenaere, P., Van Landuyt, J. & Van Tendeloo, G. (2001) Evolution of crystallographic phases in (Sr 1x Ca x )TiO 3 with composition (x). J. Solid State Chem. 162, 2028. Redjaimia, A. & Morniroli, J.-P. (1994) Application of microdiffraction to crystal-structure identification. Ultramicroscopy 53, 305317. Ringwood, A.E., Kesson, S.E., Ware, N.G., Hibberson, W. & Major, A. (1979) Immobilization of high-level nuclear-reactor wastes in SYNROC. Nature 278, 219223. Slater, P.R., Irvine, J.T.S., Ishihara, T. & Takita, Y. (1998) Hightemperature powder neutron diffraction study of the oxide ion conductor La 0.9 Sr 0.1 Ga 0.8 Mg 0.2 O 2.85 . J. Solid State Chem. 139, 135 143. Stadelmann, P.A. (1987) EMS a software package for electrondiffraction analysis and HREM image simulation in materials science. Ultramicroscopy. 21, 131145.

C 2008 The Authors Journal compilation C 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

26

J.-P. MORNIROLI ET AL.

Stadelmann, P.A. (2007) jEMS a Software Package to Simulate Dynamical Electron Precession Patterns. EPFL, Lausanne. Stlen, S., Bakken, E. & Mohn, C.E. (2006) Oxygen-deficient perovskites: linking structure, energetics and ion transport. Phys. Chem. Chem. Phys. 8, 429447. Tarakina, N.V., Tyutyunnik, A.P., Zubkov, V.G., DYachkova, T.V., Zainulin, Y.G., Hannerz, H. & Svensson, G. (2003) High temperature/high pressure synthesis and crystal structure of the new corundum related compound Zn 4 Nb 2 O 9 . Solid State Sci. 5, 459463. Vincent, R. & Midgley, P.A. (1994) Double conical beam-rocking system for measurement of integrated electron-diffraction intensities. Ultramicroscopy 53, 271282. Wang, W.L. & Lu, H.Y. (2006) Phase-transformation-induced twinning in orthorhombic LaGaO 3 : {121} and [010] twins. J. Am. Ceramic Soc. 89, 281291.

Wang, W.L. & Lu, H.Y. (2007) <111> rotation twins in an orthorhombic LaGaO 3 perovskite. J. Am. Ceramic Soc. 90, 264271. Wang, Y., Liu, X., Yao, G.-D., Liebermann, R. & Dudley, M. (1991) High temperature transmission electron microscopy and X-ray diffraction studies of twinning and the phase transition at 145 C in LaGaO 3 . Mater. Sci. Eng. A 132, 1321. Wei, Q., Wanderka, N., Schubert-Bischoff, P., Macht, M.P. & Friedrich, S. (2000) Crystallization phases of the Zr 41 Ti 14 Cu 12.5 Ni 10 Be 22.5 alloy after slow solidification. J. Mater. Res. 15, 17291734. Weissbart, J. & Ruka, R. (1961) Oxygen gauge. Rev. Sci. Instrum. 32, 593. Yao, G.D., Dudley, M., Wang, Y., Liu, X. & Liebermann, R.C. (1991) Synchrotron radiation topography studies of the phase-transition in LaGaO 3 crystals. Nucl. Instrum. Methods Phys. Res. B 56(7), 405408. Yuan, D. & Kroger, F.A. (1969) Stabilized zirconia as an oxygen pump. J. Electrochem. Soc. 116, 594600.

Journal compilation

C 2008 The Authors 2008 The Royal Microscopical Society, Journal of Microscopy, 232, 726

S-ar putea să vă placă și