5+ oxy-chloride, solved by precession electron diffraction: Electric and magnetic behavior

Journal of Solid State Chemistry 212 (2014) 99–106

Contents lists available at ScienceDirect

Journal of Solid State Chemistry journal homepage: www.elsevier.com/locate/jssc

Sr4Ru6ClO18, a new Ru4 þ /5 þ oxy-chloride, solved by precession electron diffraction: Electric and magnetic behavior Pascal Roussel a,n, Lukas Palatinus b, Frédéric Belva a, Sylvie Daviero-Minaud a, Olivier Mentre a, Marielle Huve a a Université Lille-Nord de France, UCCS—Unité de Catalyse et Chimie du Solide, UMR CNRS 8181, Ecole Nationale Supérieure de Chimie de Lille, 59652 Villeneuve d’Ascq, France b Institute of Physics of the Czech Academy of Sciences, Department of Structure Analysis, Na Slovance 2, 182 21 Praha 8, Czech Republic

art ic l e i nf o

a b s t r a c t

Article history: Received 31 October 2013 Received in revised form 6 January 2014 Accepted 9 January 2014 Available online 15 January 2014

The crystal structure of Sr4Ru6ClO18, a new Ru4 þ /5 þ oxo-chloride, has been determined from Precession Electron Diffraction (PED) data acquired on a nanocrystal in a transmission electron microscope using the technique of electron diffraction tomography. This approach is described in details following a pedagogic route and a systematic comparison is made of this rather new method with other experimental methods of electron diffraction, and with the standard single crystal X-ray diffraction technique. Both transport and magnetic measurements, showed a transition at low temperature that may be correlated to Ru4 þ /Ru5 þ charge ordering. & 2014 Elsevier Inc. All rights reserved.

Keywords: Ruthenium oxide Precession Electron crystallography Mixed valence Transport properties

1. Introduction Due to the offered mean valence states and specificities of 4d electrons, the ruthenates transition-metal oxides are intriguing systems which exhibit a variety of interesting properties covering unconventional superconducting behavior, as observed in Sr2Ru þ 4O4 [1], high TC itinerant metallic ferromagnetism in Sr4Ru3þ 4O10 [2] and in SrRu þ 4O3 [3], and even non Fermi-liquid behavior in CaxSr1  xRu þ 4O3 system (x ¼0.75) [4]. In general, the more extended nature of the 4d orbitals relative to 3d electrons is expected to considerably enhance the electron-lattice interaction, which may be the reason for the formation of a variety of structures for ruthenates ranging from perovskites to pyrochlores [5]. Also, the facility to stabilize ruthenates with valence up to Ru þ 6 in solids using oxidizing conditions such as electrocrystallization in molten hydroxydes can lead to high valent hexagonal perovskites, combining M þ alkali and Ru5/6 þ in the B sites, such as in Ba6Ru52 þ Na2  2O17 (X¼tetrahedral P5 þ , V5 þ , As5 þ …) [6], þ Ba5Ru3þ 5.33Na2O14 [7] or Ba3Ru5.5 NaO9 [8]. It is interesting in 2 the latter that, beyond the possibility to tune metal valence by carbonate incorporation [9], the existence of a low-temperature Ru5 þ /Ru6 þ charge ordering with abrupt magnetic and electric

n

Corresponding author. Tel.: þ 33 3 20 33 64 34; fax: þ 33 3 20 43 68 14. E-mail address: [email protected] (P. Roussel).

0022-4596/$ - see front matter & 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jssc.2014.01.012

anomalies [8] has been reported. In view of similar interesting properties, we have prepared a ruthenate-KSbO3 derivative namely, (Ru6þ 4.85O18)(ClSr4)0.89(OSr4)0.11 prepared via high pressure in KClO3 media. Several related polytypes were already reported, e.g. (O2La3)2(Ru6O18) and (OLa4)(Ru6O18) in which metal–metal bond length in a dimer of edge-shared RuO6 octahedra is decisive on the peculiar behavior for the latter (non-Fermi liquid state accompanied by a spin-gap) against liquid-Fermi behavior for the former [10]. Also, in the title compound the incorporation of ClSr4 units is rather original since, to the best of our knowledge only Chloride-centered tetrahedra were found in the isomorph 5d osmium oxide of formula (Os6þ 4.83O18) (ClBa4) in absence of any physical characterizations [11]. Beside the characterization of these kinds of exciting physical properties, in our group, we are also interested in the structural characterization of new compounds (that we usually synthesize using different approaches), and more especially in the present manuscript, to the now offered possibility of solving structures using electron diffraction data. This rather new approach has already been described in details elsewhere (see Part 1.1 for a complete overview), but, in the present case, instead of working on an already reported compound, we have chosen to optimize our methodology on this new material that offers several particularities in terms of symmetry and partial disorder, well suited to highlight the advantages and limits of individual methods. The paper is constructed as follows: after this introduction and a long reminder on

100

P. Roussel et al. / Journal of Solid State Chemistry 212 (2014) 99–106

the structure determination by electron diffraction, the crystal structure established by single crystal XRD is described and commented. Then it is compared to the models of Sr4Ru6ClO18 structure obtained by various methods (data acquisition and structure solution) from electron diffraction. Finally exotic physical properties of this new mixed-valence (IV–V) Ruthenium oxide are presented and discussed. 1.1. State of the art of electron-based structure elucidation Structure analysis of new materials by electron diffraction (ED) in a transmission electron microscope (TEM) presents the advantage that, compared to conventional single crystal X-ray diffraction (XRD), very small crystals (up to few nanometers) are observable. This is particularly interesting in the case of multiphase samples or if only nano-crystals are available. However, if on one hand it is useful to have a strong interaction of electron with matter (several thousand times stronger than X-rays), on the other hand, the diffracted intensity can be strongly affected by dynamical effects, thus complicating accurate structure determination from ED data. In 1994, Precession Electron Diffraction (PED) was proposed by Vincent and Midgley [12] (commercialized since 2004 by NanoMEGAS [13]); in this approach, while the crystal is oriented along a zone axis, the beam is tilted and precessed on a conical surface (with opening semi-angle typically between 1 and 31). The incident beam is never oriented exactly along the zone axis, and the recorded intensities are an integral over all orientations of the incident beam. Consequently, the precession patterns are less dynamical [12,14]. Moreover, they display a larger number of reflections than both the conventional microdiffraction and Selected Area Electron Diffraction patterns (SAED). In particular, more First Order Laue-zone (FOLZ) spots are observable, [15], making it easier to determine the “net” symmetry (which is based on the observation of the periodicity-difference between reflections located in the FOLZ with respect to the ones located in the Zero Order Laue Zone, ZOLZ). Precession intensities are also less sensitive to slight crystal misorientation. This means that the PED intensities are better suited for the determination of the point group and symmetry of the crystal than non-precessed SAED data [16–18]. The PED technique is thus of great benefit for identification of the crystal symmetry. With the quite recent development of commercial electron precession systems [13] there is a regain of interest in the utilization of PED for crystal structure determination. Indeed, as already pointed out, even if the intensities are not completely kinematic, the dynamic contribution is strongly reduced and they can be used to solve structures using the kinematical approximation [19–33]. Prior to any structure determination, it is necessary to collect a high quality dataset, that is, as little dynamic as possible with the maximal possible completeness. The sample thickness should also be as low as possible to minimize the dynamical effects. Different methods of collecting three-dimensional electron diffraction data have been developed: zone axis [A] – automated diffraction tomography (ADT) [22,27–31,34–37] – rotation electron diffraction [38]. The zone axes PED data needs to satisfy some essential criteria: (i) the orientation has to be perfect (this can be checked using the ELD software [39]) (ii) reflections below d ¼0.9 A should be checked for their signal to noise ratio and eventually eliminated (iii) zones used have to be as much as possible not main zones. This approach is time consuming and because it has been proved that the off-zone diffraction data have significantly reduced dynamical effects [40,41] an intuitive solution is to collect off zone patterns, what is applied in ADT collecting mode. It consists to cover a large portion of reciprocal space by recording diffraction patterns at small tilt steps (for instance, 11), starting from a random (i.e. not aligned) orientation of the nano-crystal. This

acquisition method is reminiscent of “standard” single crystal XRD data collection using area detectors. We used this method in manual mode like in [22]. All the more it allows a large coverage of the reciprocal space. It has been established that a large precession angle is necessary to minimize multi-beam conditions and so to reduce the dynamical effects [21]. However, if on one hand the maximum precession angle is desired for achieving the maximal reduction of dynamical effects, on the other hand, this angle must be low enough to avoid encroachment of the zero- and first- order Laue zones (ZOLZ, FOLZ). The acquired intensities will then be used to (i) find the actual point group symmetry (ii) find the extinction symbol (and thus the space group or a set of space-group candidates) (iii) solve and refine the crystal structure. Nowadays, the standard method for structure solution from XRD are the direct methods, with, for instance, SIR2008 [42] among other software. The same methods and software can be used also for electron diffraction data. In particular, the version Sir2011 contains special algorithms adapted to PED [22,43]. Another alternative to solve the structure is the charge flipping algorithm [44] which has also been demonstrated several times to work with electron diffraction data [30,45,46]. The last tool available for crystal structure determination that we consider in this work is the “direct-space” approach, namely a combined global optimization of the difference between calculated and observed diffraction data and of the potential energy of the system [47]. Due to the additional usage of the potential energy, the method is usually less sensitive to low-quality diffraction data than e.g. direct methods, and can thus be useful for PED data. In summary, if one needs to solve structures from PED data, two main points are topical: (i) the collection method for a good quality dataset (classical oriented tilt series vs. diffraction tomography on non-oriented patterns) (ii) the structure determination method (direct methods vs. charge flipping vs. direct-space method). In the present paper, these different aspects will be presented and compared.

2. Experimental 2.1. Synthesis The Sr4Ru6ClO18 crystals were synthesized under high pressure hydrothermal conditions in supercritical water media. 0.5 mmol of strontium hydroxide Sr(OH)2, 8H2O (99% Prolabo) and 0.25 mmol of ruthenium oxide RuO2 (99.5% Acros) were used as raw materials, with 0.5 mmol of KClO3 in order to obtain a good oxidizing medium. The reactants were sealed in a gold capsule half full with water and put down in a Novaswiss autoclave. The pressure equilibrium was established with supercritical water at 600 1C, under 2000 bars, during 24 h. Then, heating was turned off and the autoclave was cooled down at room temperature. To eliminate the excess of strontium carbonate, the resulting product was washed with a solution 0.1 M HCl, filtered and dried. The preparation is not homogeneous, containing at least two distinguishable crystalline phases. The major one is formed of black hexagonal platelets, identified as the already reported oxide Sr4Ru3.05O12 [48], the second minor phase, Sr4Ru6ClO18, is formed of small black cubic crystals. Various synthesis conditions have been tested (pressure, temperature, KCl addition…) and when this phase is present, it is always in minority and with a small crystals size. One must notice that it can also be present as an impurity, from the synthesis at high pressure and temperature of the monoclinic phase Sr2Ru3O10 [49] which is, as well, always mixed with Sr4Ru3.05O12 This shows a strong competition between those three phases and Sr4Ru6ClO18 seems to be the least stable in these synthesis conditions.

P. Roussel et al. / Journal of Solid State Chemistry 212 (2014) 99–106

101

susceptibility versus temperature measurement was carried out on a 0.11 mg sample (0.11 mg) under 0.2 T.

3. Results 3.1. Structure determination by single crystal X-ray diffraction

Fig. 1. Presentation of the ClSr4 and OSr4 tetrahedra, localized in the center of the Ru6O18 cages.

2.2. Transmission electron diffraction The “classical” oriented PED patterns were acquired on a Philips CM30 microscope in selected area diffraction mode with a large precession angle of 3.21. The samples were ground and spread out on a grid with holey carbon membrane. The data were acquired on a Gatan 1k CCD camera. A NanoMEGAS Spinningstar device was used for precession electron diffraction. The diffraction patterns were originally recorded in DM3 (GATAN) format and later converted into 16 bit unsigned TIF format for further processing. Intensities were extracted from images using CRISP-ELD software [39]. Chemical analysis was performed by EDS method on approximately fifty nanocrystals. Twenty of them yielded an average formulary of SrRu1.5Cl0.25, corresponding to Sr4Ru6ClO18, while other tested crystals did not contain chlorine atoms and correspond to the Sr2Ru3O10 formula. The electron diffraction tomography data were collected in steps of 11, on a FEI Tecnai G2 20 twin microscope operating at 200 kV (λ¼0.0251 Å) and intensities were recorded on an Orius CCD camera (Gatan, 14 bit dynamic range). 112 non oriented PED patterns were collected (from a tilt angle of  601 to þ511), with a precession angle of 1.61, not too high to minimize the superposition of spots, but large enough to reduce dynamical effects. After conversion in 16 bit unsigned format, this series of diffraction images was processed using the computer program PETS [50]. The indexing and integration procedure is analogous to commonly used procedures in X-ray diffraction. Its particular implementation in PETS is described in [46]. 2.3. Single crystal X-ray diffraction XRD datasets of 5 suitable black crystals were recorded at room temperature on a Bruker SMART CCD1K diffractometer equipped with a Mo monochromatized sealed tube as source. XRD data were collected at room temperature. The diffracted intensities were collected at a completeness of 99% up to 2θ¼109.761 (redundancy¼4.16) following a strategy based on ϕ and ω scans. Intensities were corrected for absorption effects using a semiempirical method based on redundancy.

The structure was solved both by Direct Methods using the program SIR2011 [51] and by the Charge Flipping method using the program Superflip [50]. By analyzing the symmetry of the charge flipping solution [52], Superflip also determined the correct space group (I23) among the three space groups I213, I23 and Im  3 compatible with the observed Laue class Im  3 (Rint ¼0.0214 for m  3 against Rint ¼0.497 for m  3m). This specific space group represents only E 5% of the cubic space groups for inorganic compounds according to the ICSD database. Its identification among other space groups by ED will be the first challenge to overcome. The crystal structure belongs to the KSbO3-type. It involves a 3D inorganic Ru6O18 framework with cavities filled by ClSr4 cationic groups similarly to OLa4 oxo-centered tetrahedra in (OLa4)(Ru6O18) [53]. We note that in this latter compound the interplay between the cage content and the framework is able to drive very particular properties, such as non-Fermi liquid transport behaviour [10]. The refinement of the ideal (ClSr4)(Ru6O18) model lead to R1 ¼0.0492 for 1522 observed reflections and 25 parameters. However, relatively large residual density on Fourier difference maps evidenced intra-cage disorder near Sr1 sites. The disorder corresponds to occupancy of the cages by  89% of ClSr4 and  11% of OSr4 tetrahedra. The two types of tetrahedra differ in size (Fig. 1). After taking into account this additional disorder, the refinement R-value decreased to R1 ¼0.0389 for 27 parameters and the same number of reflections. Thus, the exact formula of the compound is (ClSr4)0.89(OSr4)0.11(Ru6O18). The strontium atom is split into two positions (hereafter Sr1 and Sr2, whose occupancies were refined), while the Cl1 site is occupied by chlorine when Sr1 is present, and by oxygen when Sr2 is present. The corresponding distances are reported Table 2. We note the excellent reproducibility of the refined formula; the same disorder has been refined from the 5 collected data sets of several samples prepared in different batches prepared by similar high-pressure procedure. The detection or not of the static disorder is a challenge for the structure solving using PED data. 3.2. Structure determination by precession electron diffraction In a first step, oriented patterns were used to determine the symmetry as well as the space group because as we will see latter the non-oriented patterns used in tomography were not well adapted to determine them. Traditionally three-dimensional diffraction data are collected by manually tilting a crystal around a selected crystallographic axis and recording a set of oriented diffraction patterns (called a tilt series) at various crystallographic zones. In a second step, diffraction data from these zones are combined into a 3D data set and analyzed to yield structural information, e.g. the “net” symmetry and/or extinction symbol [19,21,23,24,27,37]. In a second step non-oriented PEDP in diffraction tomography mode were used to reach the structure.

2.4. Transport and magnetic measurements Conductivity measurements were carried out on a homemade cell four-probe cell in the 10–330 K temperature range. Due to the crystal size limit (o200 μm), an agglomerate of 4–5 crystals was used. Gold wires were fixed using graphite paste. Magnetic measurements were carried out on a MPMS SQUID (Quantum Design) magnetometer between 1.8 and 300 K at 0 to 7 T. The

3.2.1. Zone-axis patterns method 3.2.1.1. Cell parameters determination. The 3D reciprocal space was reconstructed using nine oriented zone axis patterns (ZAP) from the same crystallite (approximate size 50 nm, tilt angle varying from  401 to þ251). By combining the diffraction patterns of this tilted series, the cubic unit cell is verified: a EbE cE9.2 Ǻ and αEβE γE901. Note that for the determination of the unit cell

102

P. Roussel et al. / Journal of Solid State Chemistry 212 (2014) 99–106

Fig. 2. (a) ED tilted series allowing the reconstruction of the reciprocal space. (b) [0 0 1] and [1 1 0] ZAP. The comparison of the ZOLZ and FOLZ leads to a partial extinction symbol I - - - that is body centered lattice and no glide plane. (c) simulated [1 1 1] CBED ZAP of Sr4Ru6ClO18 with the I23 symmetry.

Fig. 3. Ideal model for Sr4Ru6ClO18, as obtained by Endeavour software [30].

parameters the use of the precession is not required, but as it facilitates the orientation of the patterns and increases the number of reflections in the diffraction patterns, it was used in the present experiment. Note also that, using this approach, errors on cell parameters are quite high and thus it is very difficult to make conclusions about the crystal system before the analysis of the symmetry of intensities.

3.2.1.2. Symmetry analysis 3.2.1.2.1. Point group determination. As already mentioned, the extraction of the intensity from the PED diffraction patterns has

been completed with the software ELD [39] in integration mode. The symmetry of intensities can be used to determine the Laue class of the compound. To estimate the Laue group, the merging agreement factor was calculated. It measures the mean deviation of equivalent intensities compared to the average intensity of the equivalent reflexions, i.e. Rint ¼ ∑jF 2obs  〈F 2obs 〉j=∑jF 2obs j, the sum being calculated over all measured reflections, where 〈F 2obs 〉 ¼ ∑wjF 2obs j=∑w and w ¼ 1=s2 ðF 2obs Þ, with the sum running over all symmetry-equivalent reflections. This merging factor is then calculated for all possible Laue groups, and the lowest factor for the highest symmetry is chosen. It is possible to use a single zone axis pattern but, to have a significant sampling, it is better to merge all zone axis patterns with the software Triple [39]. The results for all the Laue groups are gathered in Table 3. The cubic symmetry with Laue group m  3 (Rint ¼ 21.8%) is probable, but the m  3m Laue class could not completely be refuted. 3.2.1.2.2. Space group determination. It is well known that SAED patterns of the Zero Order Laue Zone (ZOLZ) reciprocal planes at several orientations of the crystal give information about the symmetry elements, but this method, tedious and long, is not always sufficient. Indeed, it has been shown [15] that the comparison between the ZOLZ and the FOLZ is a more efficient manner to determine the extinction symbol. According to the tables given in [15], for a cubic symmetry, the [0 0 1] and [1 1 0] ZAPs are required. The two ZAPs are shown in Fig. 2b, which demonstrates that there is no difference of periodicity between ZOLZ and FOLZ, but a shift in two directions, which matches well the partial extinction symbol I - - -, i.e. the body centered lattice and no glide plane perpendicularly to 〈1 0 0〉, 〈1 1 1〉 and 〈1 1 0〉. Only three space groups are compatible with the Laue group m  3 and the extinction symbol I - - -: I213, I23 and Im  3. It is not possible to distinguish between I213 and I23 space group from spot diffraction patterns, because they contain the same symmetry operators, only differently arranged in the unit cell. But distinction between these space groups has been reached using coherent-beam electron diffraction [54]. From spot diffraction patterns, it is also practically

P. Roussel et al. / Journal of Solid State Chemistry 212 (2014) 99–106

103

Table 1 Structural parameters and atomic coordinates for “ideal” Sr4Ru6ClO18, refined with the different strategies given in the text (first line: PED oriented patterns, second line: diffraction tomographyþ PED, third line: X-ray single crystal). Δ gives the difference in Å of the positions between the selected model and the reference structure refined again the XRD data. Crystal system

Cubic

Space-group Cell parameters

I 23 (no. 197) a¼ 9.20(5) Å a¼ 9.185(2) a¼ 9.2167(3) 0.2183 (averaged from 967 recorded reflections) 0.3115 (averaged from 2419 recorded reflections) 0.0214 (averaged from 6629 recorded reflections) 0.3270 on 343 independent reflections 0.2512 on 562 independent reflections 0.0507 on 1593 independent reflections

Rint

R(all)

Atom

Wyck

Site

x/a

y/b

z/c

Δ (Å)

Ru1

12e

2..

Sr1

8c

.3.

O1

24f

1

O2

12d

2..

Cl1

2a

23.

1/2 1/2 1/2 0.69041 0.69120 0.68129 0.47613 0.47285 0.47652 1/2 1/2 1/2 1/2 1/2 1/2

0.36878 0.36261 0.33914 0.30959 0.30880 0.31871 0.21877 0.21019 0.21229 1/2 1/2 1/2 1/2 1/2 1/2

0 0 0 0.30959 0.30880 0.31871 0.14055 0.14841 0.16500 0.16010 0.16131 0.13457 1/2 1/2 1/2

0.27 0.22 – 0.15 0.16 – 0.23 0.16 – 0.24 0.25 – 0 0 –

Table 2 X-ray single crystal atomic parameters of the actual, disordered, compound with chemical formula Sr4Ru6Cl0.892O18.108 and characteristic distances of the ruthenium and strontium environment. The strontium atom is split into two positions, Sr1 and Sr2, corresponding to the 2 types of ClSr4 and OSr4 tetrahedra. Atom1

Wyck.

Site

Ruthenium dimers Ru1 12e

2..

O2 O1

1 2..

12d 24f

S.O.F.

Atom2

x/a

y/b

z/c

d 1,2 [Å]

O1 O2 O1 Ru1 O2(short edge) O1(long edge)

0.28800 1/2 0.16550 0.66102 1/2 0.52360

0.33450 1/2 0.47640 1/2 1/2 –0.21200

–0.02360 0.13370 0.21200 0 –0.13370 0.16550

1.6118(28) 1.9301(24) 2.5356(28) 2.9698(4) 2.4660(52) 3.9343(39)

ClSr4/OSr4 Tetrahedra O3 2a Cl1 2a

23. 23.

0.108 0.892(4)

Sr2 Sr1

0.65140 0.31774

0.65140 0.68226

0.65140 0.31774

2.4183(74) 2.9112(8)

Sr–O polyhedra Sr1

8c

.3.

0.892(4)

O1

Sr2

8c

.3.

0.108

O2 O1

0.47640 0.78800 1/2 0.78800 0.97640 1/2

0.21200 0.16550 1/2 0.16550 0.28800 1/2

0.16550 0.52360 0.13370 0.52360 0.33450 0.13370

2.5546(29) 2.5546(29) 2.9207(23) 2.6538(79) 3.0516(79) 2.7976(78)

O2

impossible to distinguish between I23/I213 and Im  3. However, in convergent beam mode, it is possible to differentiate I23/I213 from Im  3 by analyzing the [1 1 1] ZAP. The Im 3 space group presents a six-fold symmetry while only a 3-fold symmetry exists in I23, what is observed in the [1 1 1] simulated CBED pattern (Fig. 2c) of Sr4Ru6ClO1. Unfortunately, the quality of the crystals did not allow the acquisition of experimental CBED pattern.

3.2.1.3. Solution of the structure. The same 3D dataset comprising 9 ZAP was used for the structure determination. Attempts to reach the structure with either SIR2011 (direct methods) or Superflip

(charge flipping) were unsuccessful. No chemically meaningful structure model could be obtained. We then focused on Endeavour software [47], which combines the potential energy and the experimental data. This approach is much less sensitive to lowquality diffraction data. A chemically reasonable structure is obtained after 8100 cycles with a 50–50 relative weighting for diffraction data and potential energy (see Fig. 3). The final atomic parameters are given in Table 1. Note that the same structural model was established even if only potential energy is taken into account without any PED data (weighting 0–100%), and that no reasonable model was established with 100–0% weighting. This means that the acquired data from oriented patterns indeed have

104

P. Roussel et al. / Journal of Solid State Chemistry 212 (2014) 99–106

very low quality. That is not surprising because the essential criteria of the oriented ZAP cited in the state of art have not been respected, and therefore the intensity did not contribute significantly to the structure solution—the structure model was obtained purely by the optimization of potential energy. It is thus very difficult to draw a general conclusion from this particular experiment, but from our experience, obtain a good dataset from this approach is possible but rather tricky and time consuming compared to the tomography approach that we present in the next part. 3.2.2. Method by tomography on non-oriented patterns To overcome the problems with structure solution encountered with oriented diffraction patterns, we tried to use the method of electron diffraction tomography (EDT), i.e. to sample the whole reciprocal space, but with non-oriented patterns. As already discussed, the principle of the method is similar to single crystal XRD acquisition using 2D detectors, the reciprocal space is recorded every 11 without any prior alignment of the crystal. Therefore, a fine sampling of the complete reciprocal space can be obtained. If, in addition, the tomographic data acquisition is combined with PED, the result is a complete data set (within the limits of the goniometer tilt) of quasi-kinematical intensities. The denomination “electron diffractometer” summarized well this approach. 3.2.2.1. Unit cell determination. Similarly to the first method, the pertinent crystals were sorted using EDS analysis. After peak hunting using the PETS software, the JANA2006 indexing plug-in [55] was used. To improve the statistics, two nano-crystals were used: on the first one, after refinement on 1406 clusters, the cell parameters converged to a¼ 9.1627(9) Å, b¼9.185(1) Å, c¼ 9.196 (5) Å, α ¼89.77(2)1, β ¼89.56(2)1, γ¼ 90.0(1)1 while for the second nano-crystal, refinement on 1576 clusters converged to a ¼9.179 (2) Å, b¼9.1848(9) Å, c¼ 9.203(1) Å, α¼ 89.96(1)1, β¼89.56(2)1, γ¼ 89.99(2)1. This leads, after averaging and symmetrisation, to the cubic parameter a¼ 9.185(2) Å α¼ β¼γ ¼901. It follows that the tomographic approach allows for a more precise (smaller e.s.d’s) estimation of the lattice parameters. Note however, that this does not necessarily imply a more accurate value. The accuracy can be compromised by several instrumental effects, mainly by possible inaccuracy of the calibration of the microscope… 3.2.2.2. Point group determination. As for the previous method, after integration of intensities with PETS, the determination of the Laue symmetry was attempted from the merging factors, Rint, see Table 3. However, it is difficult to unambiguously conclude on the actual Laue class, since, unlike in the case of oriented patterns, the Rint factors are relatively high in all cases, and no clear choice can be made based solely on Rint. In this respect, the standard “oriented-ZAP” approach appears more useful than the use EDT data. 3.2.2.3. Space group determination. The methodology to determine the space group using the “electron-diffractometer” is performed by the examination of systematic absences due to translational symmetry operators. In most cases, this will allow inference of the possible space groups (the true space group being only a hypothesis until the structure has been solved, for PED data as well as for X-ray ones). In the “non-oriented” approach, the space group determination is less easy than with the “oriented” ones, since automatic space group determination algorithms (as for instance in JANA2006) are based on the examination of structural reliability indicators, which are biased by the rather bad estimation of the errors on the measured intensities. In addition, systematic absences,

along the principal axes for screws, are less reliable indicators since there are relatively few axial reflections in a full three-dimensional data set and some of these may be unrecorded using the nonoriented approach. On the other hand, as already discussed, the combination of SAED and CBED data on oriented patterns enable a clear determination of the space group. 3.2.2.4. Solution of the structure. The three previously introduced methods (charge flipping/Superflip, direct methods/SIR2011 and combination of potential energy with diffraction data in direct space/Endeavour) were tested with intensities extracted from the non-oriented patterns. All of them were successful, and provided complete structure models with all five independent atoms. It is notable, however, that attempts to solve the structure by the standard charge flipping algorithm [56] failed, and the iteration always ended by a false solution with all phases equal to zero, the so-called “uranium-atom” solution, i.e. with one dominant peak in the corresponding Fourier syntheses. This problem could be avoided by employing the iteration scheme called averaged alternating reflections (AAR) [56,57]. Using AAR, most attempts converged to a correct solution. The three solutions of Sr4Ru6ClO18 from data collected on nonoriented ZAPs (i.e. obtained by Superflip, SIR2011 and Endeavour), are very similar to the solution from single crystal XRD data. All three solutions were refined using JANA2006; the three refinements converged to the same result given in Table 1. Note that the errors in the atomic positions and distances, derived from the estimated standard deviations, are obviously underestimated. With R values as high as E0.30 we can conclude that this final model is only approximate. Indeed, attempts to distinguish the interstitial SrO4 vs. SrCl4 disorder in difference Fourier maps were

Table 3 Merging R factor calculated for all possible Laue groups (in different settings where necessary) for the three different datasets (oriented PED, non oriented PED, X-ray single crystal). Crystal system

Point group

Oriented PED

NonOriented PED

Single crystal X-ray diffraction

Triclinic Monoclinic—setting “a” Monoclinic—setting “b” Monoclinic—setting “c” Orthorhombic Tetragonal—setting “a” Tetragonal—setting “a”

0.143 0.165 0.162 0.149 0.174 0.192 0.215

0.109 0.195 0.195 0.195 0.236 0.287 0.339

0.019 0.020 0.020 0.020 0.021 0.432 0.474

0.183 0.216

0.278 0.310

0.449 0.473

0.218 0.249

0.337 0.397

0.465 0.489

“a,  b,

1 2/m 2/m 2/m mmm 4/m 4/ mmm 4/m 4/ mmm 4/m 4/ mmm 3

0.197

0.243

0.021

“a,  b,

 3m

0.244

0.391

0.495

“  a, b,

3

0.188

0.200

0.021

“  a, b,

 3m

0.255

0.399

0.494

“  a,

3

0.187

0.211

0.021

“  a,

 3m

0.241

0.397

0.471

“a, b, c” “a, b, c”

3  3m m3 m  3m

0.218 0.271 0.218 0.271

0.258 0.410 0.324 0.458

0.021 0.473 0.023 0.497

Tetragonal—setting “b” Tetragonal—setting “b” Tetragonal—setting “c” Tetragonal—setting “c” Rhombic—setting  c” Rhombic—setting  c” Rhombic—setting  c” Rhombic—setting  c” Rhombic—setting  b, c” Rhombic—setting  b, c” Rhombic—setting Rhombic—setting Cubic Cubic

P. Roussel et al. / Journal of Solid State Chemistry 212 (2014) 99–106

not successful, showing the limits of the method compared to single crystal X-ray structure determination.

105

3.5x10-5 3.0x10-5

3.3. Transport and magnetic properties

4. Conclusion In this paper, beside the description of a new Ru4 þ /Ru5 þ compound with atypical low-temperature behavior, we have illustrated in detail the possibility now offered to crystallographers to solve crystal structures ab initio from precession electron diffraction data. We believe that this topic is of interest for solid state chemists because even though some experts in crystallography already know that it is now possible to determine structures from electron diffraction data (in the same way as single crystal X-ray diffraction), many chemists are unaware of this new possibility of high interest for nanometric crystals. We have described this rather new approach, following a pedagogic route,

( .cm)

2.0x10-5 1.5x10-5 1.0x10-5 5.0x10-6 0.0

0

50

100

150

200

250

300

350

250

300

350

T(K)

0.0025

M/H (emu/Oe.mol)

Due to the rather limited size of available single crystals (cubicshaped with typical dimensions smaller than 200 μm), the resistivity was collected on an agglomerate of 4–5 single crystals along undetermined directions. However the cubic symmetry is expected to limit the anisotropic behavior, and one should consider the ρ(T) plot shown on Fig. 4a, as representative in average. It was collected on heating and cooling between room temperature and 20 K and shows a nearly reproducible behavior. Between room temperature and 120 K the resistivity is almost temperature independent assorted with a small decrease of less than Δρ/ρ¼  10% of its absolute value on cooling. This behavior could be related to impurity levels or non-stoichiometry in a metallic matrix. Here, only defect-electrons close to the Fermi level can be scattered distorting considerably the metallic behavior from its ideal metallicity. We recall that both (O2La3)2(Ru6þ 4.33O18) and (OLa4) (Ru6þ 4.33O18) are metallic at room temperature, while the incorporation of 10% of [OSr4]6þ tetrahedra for [ClSr4]7þ in the KSbO3 cages in the title compound deviates the valence from the ideal Ru þ 4.833 to Ru þ 4.85. This non-stoichiometry could be responsible for the creation of shallow donors above the conduction bands [58,59]. Below 120 K, suggestions of electric events occur. The resistivity suddenly increases until a maximum at 48 K and drops again on further cooling. At this point of our study this phenomenon appears atypical and is not understood yet. At least in the related structures (OLa4)(Ru6þ 4.33O18), metal non Fermi-liquid behavior is associated to the anomalously short (2.5 Å) RuRu bond, while in the title compound the Ru–Ru distance of 2.95 Å is similar to those (2.9 Å) observed in (O2La3)2(Ru6O18), too large for a direct Ru–Ru bonding [60]. Then, the observed transition seems out of the scope of weakly interacting particules that are sometimes associated with 4d electrons in ruthenates [10,61–63]. At the moment at best could we associate the observed behavior to Ru þ 4/Ru þ 5 charge ordering, as already observed between Ru þ 5 and Ru þ 6 in Ba3Ru2NaO9 below 210 K with segregation of pentavalent and octavalent Ru ions in ordered distinct Ru2O9 pairs. In this compound both ρ(T) and χ(T) show abrupt variations at the transition. Fig. 4b shows the magnetic susceptibility collected on a small amount of selected crystals of the title compounds. The rather noisy signal occurs from the small amount of collected sample. At least, between room temperature and 120 K the low value of M/H slowly decreasing with T (between 3.3  10  4 and 2.5  10  4 emu/mol Ru) is mainly consistent with Pauli paramagnetism associated to the metal band character of the d electrons. The behavior on further cooling mainly emulates those of the thermal conductivity, while the maxima of ρ(T) is marked by an anomaly of χ(T) at 48 K. A Néel ordering at this temperature cannot be excluded but the full interpretation of electric and magnetic data deserve additional experiments under consideration.

2.5x10-5

0.0020 0.0015 0.0010 0.0005 0.0000

0

50

100

150

200

T(K) Fig. 4. (a) Resistivity versus temperature upon cooling (black) and heating (red) with low temperature anomaly maximal around 50 K. (b) Correlation of ρ(T) with the magnetic susceptibility. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

according to systematic comparison between the several available ED collection-data methods and in regard to the standard single crystal XRD technique. Different approaches have been used and compared. Even if this manuscript is not intended as a complete and comprehensive review, some conclusions can be drawn, both on the acquisition method and on the structure solution methods: The “classical” approach of electron microscopists, i.e. the use of oriented patterns (SAEDþCBED) turns to be advantageous to identify the point group and the space group of the structure. However, its use is not the easiest and fastest approach for the recording of intensities for structure solution, while the tomography on non-oriented patterns combined with PED is recommended to obtain the best pseudokinematical intensities. Concerning the different algorithms available for the structure solution, Charge Flipping in the newer AAR variant and Direct Methods produced equivalent results. The best results are of course obtained with single crystal X-ray data, but the current work clearly showed that if sizeable crystals are not available, or in case of a small impurity in a chemical preparation, the data obtained by an experiment with Precession Electron Diffraction Tomography on nonoriented patterns can now yield valuable structural information0 s that could be afterward used for example in a powder Rietveld refinement, or could form the basis of new syntheses. Concerning physical properties, the preliminary measurements show correlated magnetic and electronic properties, that may be related to Ru4 þ /Ru5þ charge ordering. However in the disordered context of the title compound where 10% of the Cl  are replaced by O2 ions, the plausible influence of impurity electronic levels should be kept in mind and further experiments are required to conclude. References [1] Y. Maeno, H. Hashimoto, K. Yoshida, S. Nishizaki, T. Fujita, J.G. Bednorz, F. Lichtenberg, Nature 372 (1994) 532–534.

106

P. Roussel et al. / Journal of Solid State Chemistry 212 (2014) 99–106

[2] M.K. Crawford, R.L. Harlow, W. Marshall, Z. Li, G. Cao, R.L. Lindstrom, Q. Huang, J.W. Lynn, Phys. Rev. B: Condens. Matter 65 (2002) 5. [3] T. Kiyama, K. Yoshimura, K. Kosuge, H. Michor, G. Hilscher, J. Phys. Soc. Jpn. 67 (1998) 307–311. [4] P. Khalifah, I. Ohkubo, H.M. Christen, D.G. Mandrus, Phys. Rev. B: Condens. Matter 70 (2004) 6. [5] G. Ehora, S. Daviero-Minaud, M.C. Steil, L. Gengembre, M. Frere, S. Bellayer, O. Mentre, Chem. Mater. 20 (2008) 7425–7433. [6] E. Quarez, F. Abraham, O. Mentre, J. Solid State Chem. 176 (2003) 137–150. [7] E. Quarez, O. Mentre, Solid State Sci. 5 (2003) 1105–1116. [8] K.E. Stitzer, M.D. Smith, W.R. Gemmill, H.C. zur Loye, J. Am. Chem. Soc. 124 (2002) 13877–13885. [9] E. Quarez, M. Huve, F. Abraham, O. Mentre, Solid State Sci. 5 (2003) 951–963. [10] P. Khalifah, K.D. Nelson, R. Jin, Z.Q. Mao, Y. Liu, Q. Huang, X.P.A. Gao, A.P. Ramirez, R.J. Cava, Nature 411 (2001) 669–671. [11] J.R. Plaisier, R.A.G. deGraaff, D.J.W. Ijdo, Mater. Res. Bull. 31 (1996) 279–282. [12] R. Vincent, P.A. Midgley, Ultramicroscopy 53 (1994) 271–282. [13] NanoMEGAS SPRL, Blvd Edmond Machtens 79, B-1080, Brussels Belgium. [14] T.A. White, A.S. Eggeman, P.A. Midgley, Ultramicroscopy 110 (2010) 763–770. [15] J.P. Morniroli, J.W. Steeds, Ultramicroscopy 45 (1992) 219–239. [16] J.P. Morniroli, A. Redjaimia, S. Nicolopoulos, Ultramicroscopy 107 (2007) 514–522. [17] J.P. Morniroli, A. Redjaimia, J. Micros.-Oxford 227 (2007) 157–171. [18] J.P. Morniroli, P. Stadelmann, G. Ji, S. Nicolopoulos, J. Micros.-Oxford 237 (2010) 511–515. [19] P. Boullay, V. Dorcet, O. Perez, C. Grygiel, W. Prellier, B. Mercey, M. Hervieu, Phys. Rev. B: Condens. Matter 79 (2009) 8. [20] P. Boullay, L. Palatinus, N. Barrier, Inorg. Chem. 52 (2013) 6127–6135. [21] A.S. Eggeman, T.A. White, P.A. Midgley, Ultramicroscopy 110 (2010) 771–777. [22] M. Gemmi, I. Campostrini, F. Demartin, T.E. Gorelik, C.M. Gramaccioli, Acta Crystallogr., Sect. B: Struct. Sci. 68 (2012) 15–23. [23] J. Hadermann, A.M. Abakumov, A.A. Tsirlin, V.P. Filonenko, J. Gonnissen, H.Y. Tan, J. Verbeeck, M. Gemmi, E.V. Antipov, H. Rosner, Ultramicroscopy 110 (2010) 881–890. [24] J. Hadermann, A. Abakumov, S. Van Rompaey, T. Perkisas, Y. Filinchuk, G. Van Tendeloo, Chem. Mater. 24 (2012) 3401–3405. [25] J. Hadermann, A.M. Abakumov, A.A. Tsirlin, M.G. Rozova, E. Sarakinou, E.V. Antipov, Inorg. Chem. 51 (2012) 11487–11492. [26] D. Jacob, L. Palatinus, P. Cuvillier, H. Leroux, C. Domeneghetti, F. Camara, Am. Mineral. 98 (2013) 1526–1534. [27] H. Klein, J. David, Acta Crystallogr., Sect. A: Found. Crystallogr. 67 (2011) 297–302. [28] H. Klein, Z. Kristallogr. 228 (2013) 35–42. [29] U. Kolb, T. Gorelik, C. Kubel, M.T. Otten, D. Hubert, Ultramicroscopy 107 (2007) 507–513. [30] E. Mugnaioli, T. Gorelik, U. Kolb, Ultramicroscopy 109 (2009) 758–765. [31] I. Rozhdestvenskaya, E. Mugnaioli, M. Czank, W. Depmeier, U. Kolb, A. Reinholdt, T. Weirich, Mineral. Mag. 74 (2010) 159–177. [32] Q. Simon, V. Dorcet, P. Boullay, V. Demange, S. Deputier, V. Bouquet, M. Guilloux-Viry, Chem. Mater. 25 (2013) 2793–2802. [33] S. Van Rompaey, W. Dachraoui, S. Turner, O.Y. Podyacheva, H.Y. Tan, J. Verbeeck, A. Abakumov, J. Hadermann, Z. Kristallogr. 228 (2013) 28–34. [34] I. Andrusenko, E. Mugnaioli, T.E. Gorelik, D. Koll, M. Panthofer, W. Tremel, U. Kolb, Acta Crystallogr., Sect. B: Struct. Sci. 67 (2011) 218–225.

[35] U. Kolb, T. Gorelik, M.T. Otten, Ultramicroscopy 108 (2008) 763–772. [36] U. Kolb, T. Gorelik, E. Mugnaioli, in: P. Moeck, S. Hovmoller, S. Nicolopoulos, S. Rouvimov, V. Petkov, M. Gateshki, P. Fraundorf (Eds.), Electron Crystallography for Materials Research and Quantitative Characterization of Nanostructured Materials, Materials Research Society, Warrendale, 2009, pp. 11–23. [37] J.G. Kim, K. Song, K. Kwon, K. Hong, Y.J. Kim, J. Electron Microsc. 59 (2010) 273–283. [38] D.L. Zhang, D. Gruner, P. Oleynikov, W. Wan, S. Hovmoller, X.D. Zou, Ultramicroscopy 111 (2010) 47–55. [39] X.D. Zou, Y. Sukharev, S. Hovmoller, Ultramicroscopy 49 (1993) 147–158. [40] J.P. Morniroli, G. Ji. In: Symposium on Electron Crystallography for Materials Research/Quantitative Characterization of Nanostructured Materials, San Francisco, CA; 2009. pp. 37-48. [41] P. Oleynikov, S. Hovmoller, X.D. Zou, Ultramicroscopy 107 (2007) 523–533. [42] M.C. Burla, R. Caliandro, M. Camalli, B. Carrozzini, G.L. Cascarano, L. De Caro, C. Giacovazzo, G. Polidori, D. Siliqi, R. Spagna, J. Appl. Crystallogr. 40 (2007) 609–613. [43] G.L. Cascarano, C. Giacovazzo, B. Carrozzini, Ultramicroscopy 111 (2010) 56–61. [44] G. Oszlanyi, A. Suto, Acta Crystallogr., Sect. A: Found. Crystallogr. 60 (2004) 134–141. [45] A. Eggeman, T. White, P. Midgley, Acta Crystallogr., Sect. A: Found. Crystallogr. 65 (2009) 120–127. [46] L. Palatinus, M. Klementova, V. Drinek, M. Jarosova, V. Petricek, Inorg. Chem. 50 (2011) 3743–3751. [47] H. Putz, J.C. Schon, M. Jansen, J. Appl. Crystallogr. 32 (1999) 864–870. [48] C. Renard, S. Daviero-Minaud, M. Huve, F. Abraham, J. Solid State Chem. 144 (1999) 125–135. [49] C. Renard, S. Daviero-Minaud, F. Abraham, J. Solid State Chem. 143 (1999) 266–272. [50] L. Palatinus, PETS a Computer Program for Analysis of Electron Diffraction Data, Institute of Physics of the AS CR, Prague, Czechia, 2011. [51] M. Camalli, B. Carrozzini, G.L. Cascarano, C. Giacovazzo, Cryst. Res. Technol. 46 (2011) 555–560. [52] L. Palatinus, A. van der Lee, J. Appl. Crystallogr. 41 (2008) 975–984. [53] F. Abraham, J. Trehoux, D. Thomas, Mater. Res. Bull. 12 (1977) 43–51. [54] K. Tsuda, K. Saitoh, M. Terauchi, M. Tanaka, P. Goodman, Acta Crystallogr., Sect. A: Found. Crystallogr. 56 (2000) 359–369. [55] V. Petricek, M. Dusek, L. Palatinus, Jana2006 The Cristallographic Computing System., Institute of Physics, Praha, Czech Republic, 2013. [56] G. Oszlanyi, A. Suto, Acta Crystallogr., Sect. A: Found. Crystallogr. 67 (2011) 284–291. [57] H.H. Bauschke, P.L. Combettes, D.R. Luke, J. Approx. Theory 127 (2004) 178–192. [58] M. Iorgulescu, P. Roussel, N. Tancret, N. Renaut, N. Tiercelin, O. Mentre, J. Solid State Chem. 198 (2013) 210–217. [59] K. Bapna, R.J. Choudhary, D.M. Phase, Mater. Res. Bull. 47 (2012) 2001–2007. [60] F.A. Cotton, C.E. Rice, J. Solid State Chem. 25 (1978) 137–142. [61] L. Klein, L. Antognazza, T.H. Geballe, M.R. Beasley, A. Kapitulnik, Physica B 259-61 (1999) 431–432. [62] R.S. Perry, L.M. Galvin, S.A. Grigera, L. Capogna, A.J. Schofield, A.P. Mackenzie, M. Chiao, S.R. Julian, S.I. Ikeda, S. Nakatsuji, Y. Maeno, C. Pfleiderer, Phys. Rev. Lett. 86 (2001) 2661–2664. [63] G.R. Stewart, Rev. Mod. Phys. 73 (2001) 797–855.