Interstitial oxygen in the Ga-based melilite ion conductor: A neutron total scattering study

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Interstitial oxygen in the Ga-based melilite ion conductor: A neutron total scattering study Alessandro Mancini, Cristina Tealdi, Lorenzo Malavasi* Dept. of Chemistry, INSTM and IENI-CNR, University of Pavia, Viale Taramelli 10-16, I-27100 Pavia, Italy

article info

abstract

Article history:

In this paper we report a neutron total scattering experiment coupled to pair distribution

Received 30 June 2011

function analysis (PDF) on a recently discovered Ga-based oxide-ion conductor charac-

Received in revised form

terized by interstitial oxygen migration. In particular, this study focuses on the short-range

20 September 2011

order analysis of the defect structure in La1.50Sr0.50Ga3O7.25 with the aim of shedding light

Accepted 30 September 2011

on the position of the interstitial oxygen and on the distortion induced in the structure by

Available online 26 November 2011

the extra-oxygen. The data analysis reveals that the preferred position, at lowtemperature, for the interstitial oxygen is within the pentagonal ring formed by the Ga1

Keywords:

and Ga2 atoms and shifted with respect to the centre of the pentagon.

Total scattering

Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

Pair distribution function analysis Ionic conductivity Interstitial oxygen

1.

Introduction

Significant efforts of the current research in the field of energy materials for solid oxide fuel cells (SOFC) are directed towards the discovery of new ion-conductors with optimal conductivity and chemical stability properties [1e4]. Over the last few years several new phases have appeared in the literature such as the oxide-ion conductors La2Mo2O9 (LAMOX) [5e8], Gabased oxides with tetrahedral units [9,10], apatites [11e14] and the proton conducting niobates [15e17]. The most recent material for which a high ionic conductivity has been reported is the La1þxSr1xGa3O7þ0.5x system [18]. The parent compound LaSrGa3O7 adopts the melilite structure, which consists of alternating cationic (La/Sr)2 and corner-sharing tetrahedral anionic Ga3O7 layers. The structure is characterized by five-fold tunnels that accommodate the eight-coordinate La/Sr as chains of cations [18,19]. Fig. 1A and B reports a sketch of the structure along two axes.

In the tetragonal unit cell two distinct crystallographic sites are occupied by the Ga atoms (labelled Ga1 and Ga2), both in tetrahedral coordination, and three distinct crystallographic sites are occupied by the O atoms. The peculiarity of the Ga2O4 units is the presence of non-bridging oxygen atoms, that is, oxygens that are not shared between Ga units. One interesting aspect of Ga-based melilite oxides is that the structure can accommodate interstitial oxygen by making the La/Sr ratio >1. This means that interstitial oxygen atoms are created according to: 

00 2Sr Sr þ La2 O3 /2LaSr þ Oint þ 2SrO

(1)

Kuang and colleagues [18] found a very high purely ionic (oxide-ion) conductivity due to interstitial oxygen for the La1.54Sr0.46Ga3O7.27 oxide composition [18]. This resulted in a new class of electrolytes for SOFC where the oxygen does not move by a vacancy mechanism such as in yttria-stabilized zirconia [1] but through interstitial oxygen.

* Corresponding author. Tel.: þ39 382 987921; fax: þ39 382 987575. E-mail address: [email protected] (L. Malavasi). 0360-3199/$ e see front matter Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2011.09.157

8074

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

Fig. 1 e Melilite structure of LaSrGa3O7: A) view along the c axis, showing the Ga tetrahedral units connected through bridging oxygens to form distorted pentagonal rings; B) view along the b axis, showing the layered nature of the structure with non-bridging oxygens pointing towards the La/Sr layer. Key: Ga1 (blue); Ga2 (pink); O (red); La/Sr (green). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

New materials characterized by interstitial defects require a good knowledge of the actual position of the interstitial atoms. Average structure probes such as diffraction are valuable in giving insights into these aspects; however, the possibility of using a local structure probe may further increase the level of knowledge of the defect positions and the local structure relaxations occurring around these defects. To this end we undertook a neutron total scattering investigation and pair distribution function (PDF) analysis on the LaSrGa3O7 and La1.50Sr0.50Ga3O7.25 samples. The use of total scattering methods for the investigation of energy materials is now well established [6,20e24] and has proved to be a successful way of getting information about the local structure and defect distribution in complex oxide systems.

2.

Material and methods

Samples of composition LaSrGa3O7 and La1.50Sr0.50Ga3O7.25 were prepared by solid-state reaction starting from La2O3,

Ga2O3 and SrCO3 (Aldrich). Following the procedure described in Ref. [18], in order to compensate for the Ga volatilization at high temperature, an excess of Ga2O3 has been used (1 wt% for LaSrGa3O7 and 3.05 wt% for La1.50Sr0.50Ga3O7.25). A first thermal treatment was performed at 1200  C for 20 h, followed by a thermal treatment at 1200  C for 15 h and a final treatment at 1400  C for 8 h. Each step was preceded by careful grinding and isostatic pressing of the samples to improve physical contact between particles. After each thermal treatment the samples have been slowly cooled down to room temperature. Neutron powder diffraction (NPD) measurements at 15 K were carried out on the NPDF diffractometer at the Lujan Center at Los Alamos National Laboratory in a cylindrical vanadium tube (diameter 0.6 cm). Total acquisition time was approximately 8 h for each sample. Data from an empty container were also collected to subtract the container scattering. The corrected total scattering structure function, S(Q), was obtained with the program PDFGETN [25]. Finally, the PDF was obtained by Fourier transformation of S(Q). A ˚ 1 was used. Refinement of the experimental PDF Qmax ¼ 35.0 A data was carried out with the aid of PDFGUI and PDFFfit2 software [26]. Refinement of the average structure was carried out with the GSAS package [27]. Three data banks were used for Rietveld and four for PDF analysis.

3.

Results and discussion

On the two samples investigated we first refined, by means of the Rietveld method, the neutron diffraction patterns at 15 K in order to get the average structure parameters that were used as the starting model of the PDF analysis. The starting average structure model used to refine the neutron diffraction patterns were taken from Ref. [18]. The Rietveld-refined patterns are reported in Fig. 2A for LaSrGa3O7 and Fig. 2B for La1.50Sr0.50Ga3O7.25 (one bank reported) while the structural parameters obtained from the refinement are shown in Tables 1 and 2 for LaSrGa3O7 and La1.50Sr0.50Ga3O7.25, respectively. Table 3 reports selected distances for the two compositions as

Fig. 2 e Rietveld refinement of neutron data at 15 K for LaSrGa3O7 (A) and La1.50Sr0.50Ga3O7.25 (B).

8075

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

derived from analysis of the structural data shown in Tables 1 ˚ are considered. and 2. Only the bond lengths lower than 2.5 A We remark here that the use of 15 K as the data collection temperature instead of room temperature was motivated by the resulting increase in the resolution of the PDF data due to the reduction of the thermal vibrations. As a result of the insignificant oxygen diffusion at ambient temperature, these data can be representative of the room temperature situation. The structural study reported here has the aim of giving a picture of the interstitial oxygen location when the defects are not mobile in the structure providing information on the most stable distribution in these conditions. The analysis of the results reported in Table 1 and Table 2 reveals a good agreement with the available data for the average structure of the two systems [18]. The good agreement factors obtained in this study, therefore, support the validity of the model previously proposed in the literature for the average structure of the oxygen over-stoichiometric sample [18]. However, as already pointed out [18], local structural relaxation around the interstitial defect are expected to play a relevant role in the description and interpretation of the positional disorder associated to both cations and anions in the oxygen over-stoichiometric sample (we observe here that the refined isotropic displacement parameters for the La1.50Sr0.50Ga3O7.25 composition are generally higher than those obtained for LaSrGa3O7, where no interstitial oxygen is present). In order to represent the local structure distortion around the interstitial defects, Kuang et al. [18] proposed a simple model, named the “split-site model”, in which the crystallographic positions of several species (cations and anions in the first coordination sphere of the interstitial defect) are split into two distinct sites, whose positions are refined independently. The occupancies of the split sites are consistent with the occupancy of the oxygen interstitial site. The main idea behind the use of this structural model is that only the species in real proximity of the defect will relax to positions different from the average structure in order to accommodate the charged defect; the remaining of the crystal will represent the average structure. Proposed as a simple model to take into account local structure distortions, the “split-site model” still constitutes an average model of the local structure as a result of the intrinsic nature of the Rietveld refinement method

Table 1 e Rietveld refinement results of neutron diffraction data at 15 K for LaSrGa3O7 (space group n. 113, P-421m).

La1 Sr1 Ga1 Ga2 O1 O2 O3

site

x

y

z

˚ 2) Uiso (A

Occ.

4e 4e 2a 4e 2c 4e 8f

0.66251(6) 0.66251(6) 0 0.85647(6) 0.5 0.86101(9) 0.91393(9)

0.83749(6) 0.83749(6) 0 0.64353(6) 0 0.63899(9) 0.83736(9)

0.4910(1) 0.4910(1) 0 0.0345(1) 0.8106(3) 0.6962(2) 0.2029(1)

0.0003(1) 0.0003(1) 0.0016(1) 0.0010(1) 0.0013(2) 0.0033(2) 0.0038(1)

0.5 0.5 1 1 1 1 1

˚) a (A 8.04209(8)

˚) b (A

˚) c (A

Rwp

Rp

8.04209(8)

5.32389(6)

4.07%

2.00%

Table 2 e Rietveld refinement results of neutron diffraction data at 15 K for the La1.50Sr0.50Ga3O7.25 composition (space group n. 113, P-421m).

La1 Sr1 Ga1 Ga2 O1 O2 O3 O4

site

x

y

z

˚ 2) Uiso (A

Occ.

4e 4e 2a 4e 2c 4e 8f 4e

0.66363(6) 0.66363(6) 0 0.85669(6) 0.5 0.86229(8) 0.9164(1) 0.3381(8)

0.83637(6) 0.83637(6) 0 0.64331(6) 0 0.63771(8) 0.8373(1) 0.1619(8)

0.4921(2) 0.4921(2) 0 0.0327(1) 0.8099(3) 0.6913(2) 0.2062(2) 0.001(1)

0.0022(1) 0.0022(1) 0.0005(1) 0.0032(1) 0.0074(3) 0.0044(2) 0.0095(2) 0.018(1)

0.75 0.25 1 1 1 1 1 0.125

˚) a (A 8.0332(1)

˚) b (A

˚) c (A

Rwp

Rp

8.0332(1)

5.2764(1)

4.54%

2.37%

applied. In this study, local structure distortions around the interstitial defect site are analyzed through the PDF method, a tool for the investigation of the local structure. The experimental neutron PDF obtained from the total scattering data of the two systems under investigation are ˚. shown in Fig. 3 up to 10 A As can be appreciated from Fig. 3, the general shape of the PDF of LaSrGa3O7 and La1.50Sr0.50Ga3O7.25 is similar (as expected being the same structure with the same tetragonal symmetry). As the effect of local structure distortion around the interstitial oxygen is expected to greatly affect the positions of cations and anion closest to it (those forming the distorted pentagonal rings shown in Fig. 1), we focus our attention on ˚ (Fig. 3B), where the GaeO and OeO bond the range 1.5e2.5 A pairs should fall according to the average models reported in Table 1 and Table 2. Indeed, one interesting point in the comparison of the PDFs of the two samples is the observation of some extra intensity (in addition to the expected ripples) ˚ (marked with an arrow) in the peaked at approximately 2.08 A PDF of La1.50Sr0.50Ga3O7.25 where the average structure does not predict the presence of any bond pair for the oxygen-

˚ ) for the Table 3 e Selected bond lengths (<2.5 A La1.50Sr0.50Ga3O7.25 and La1.50Sr0.50Ga3O7.25 compositions as derived by Rietveld refinement of neutron diffraction data at 15 K (space group n. 113, P-421m). Atoms pair LaSrGa3O7 Ga2-O2 Ga2-O1 Ga1-O3 Ga2-O3 La1.5Sr0.5Ga3O7.25 Ga2-O2 Ga2-O1 Ga1-O3 Ga2-O3 O1-O4 Ga2-O4 O3-O4 Ga2-O4

˚) Bond length (A 1.8018 1.8289 1.8321 1.8567 1.8025 1.8277 1.8284 1.8699 2.0975 2.2193 2.3135 2.4628

8076

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

˚ range; B) Neutron PDF at 15 K for the Fig. 3 e A) Neutron PDF at 15 K for the LaSrGa3O7 and La1.50Sr0.50Ga3O7.25 in the 1.5e10 A ˚ range. LaSrGa3O7 and La1.50Sr0.50Ga3O7.25 in the 1.5e2.5 A

stoichiometric compound LaSrGa3O7 (see Table 3). Unfortunately, this contribution falls in a region where ripples (due to the truncation of the G(r) at finite Q) are also present, therefore making the analysis of the results more difficult. However, careful generation of PDFs at various values of Qmax all show this feature and combined with the fact that the intensity of ˚ for the the peak whose maximum is at approximately 2.08 A La1.50Sr0.50Ga3O7.25 sample is significantly higher than that of ripples at lower r (it is well known that the intensity of ripples decays with distance [28]) is a clear proof that the feature marked with the arrow in Fig. 3 contains intensity contributions coming from bond pairs for this sample. In contrast, for the LaSrGa3O7 sample the small peaks in that region can be safely associated with the truncation ripples and show progressive intensity decay as a function of r. The experimental PDFs were fitted using as starting models the results obtained from the Rietveld refinement of the average structure (Tables 1 and 2) within the same symmetry (P-421m space group). For the La1.50Sr0.50Ga3O7.25 sample the interstitial oxygen is placed in a general position of the P-421m space group (Wykoff index 4e). The best fit of the PDFs

(according to this average model) for the LaSrGa3O7 and La1.50Sr0.50Ga3O7.25 samples are reported in Fig. 4A and B, respectively (structural data reported in Tables 4 and 5). Fitted parameters were the scale factor, the correlation parameters (which takes into account the correlated motion of atomic pairs at low-r [29]), lattice parameters, atomic positions and isotropic atomic displacement parameters (adp). The agreement factors for the fit of the two samples show that, overall, the average structure can catch most of the features at the local scale. However, the c2 for the fit of the La1.50Sr0.50Ga3O7.25 sample is four times higher than that for LaSrGa3O7. Inspection of Fig. 4B shows that some problems for the ˚ range. To be La1.50Sr0.50Ga3O7.25 sample are found in the 1e5 A noted, for example, the fact that in this range most of the calculated peaks maxima do not correspond to the experimental peaks maxima, that the calculated peak width is in many cases smaller that the experimental one and that the ˚ is only described by a narrow single double peak just below 4 A peak. Finally, it should be noted that the model of Fig. 4A for the stoichiometric sample well reproduce the features at low-r

˚ with the average structure model; Panel B: PDF fit of La1.50Sr0.50Ga3O7.25 up to Fig. 4 e Panel A: PDF fit of LaSrGa3O7 up to 5 A ˚ with the average structure model. Blue circles: experimental data; red line: calculated PDF; horizontal green line: 5A residual. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

8077

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

Table 4 e PDF refinement at 15 K for LaSrGa3O7 (space group n. 113, P-421m). Site

x

y

z

˚ 2) Uiso (A

Occ.

4e 4e 2a 4e 2c 4e 8f

0.662(1) 0.662(1) 0 0.857(1) 0.5 0.860(3) 0.911(1)

0.838(1) 0.838(1) 0 0.643(1) 0 0.640(3) 0.836(1)

0.492(2) 0.492(2) 0 0.035(2) 0.810(4) 0.699(5) 0.201(2)

0.0033(7) 0.0033(7) 0.0018(1) 0.0024(6) 0.005(1) 0.007(3) 0.007(1)

0.5 0.5 1 1 1 1 1

La1 Sr1 Ga1 Ga2 O1 O2 O3

˚) a (A 8.050(6)

˚) b (A

˚) c (A

Rw

c2

8.050(6)

5.324(8)

13.5%

0.056

while in Fig. 4B it is clear that the calculated intensity of the ˚ is not described by the ripple only (as occurs peak at about 2 A on the opposite for the stoichiometric sample). Indeed, as shown in Table 6, the average model doesn’t predict in this ˚ , whose case the existence of bond pairs between 2 and 2.15 A presence should increase the intensity of the peak marked with an arrow in Fig. 4. A first conclusion that can be drawn from the analysis reported in Fig. 4 is that the average structure describes quite well the stoichiometric sample while it lacks in describing the local structure of the La1.50Sr0.50Ga3O7.25 sample, particularly at low-r. This might be expected considering that the average model used for the La1.50Sr0.50Ga3O7.25 sample adds an interstitial oxygen in a symmetric position within the cell without describing the distortion created in the local environment around it. Indeed, Kuang and coworkers noticed that in the average model for the oxygen-excess system the size and shape of the thermal ellipsoids of several atoms, particularly O3 and O4, were consistent with positional disorder, most probably associated with structural relaxation around the interstitial oxygen [18]. As described above, to overcome this problem, the “split-site model”, a simple model for relaxation around the defect site was proposed. This model takes into account the displacement of atoms in the first coordination environment of the oxygen interstitial by using a split-site model for specific crystallographic sites. Its use, although still

Table 5 e PDF refinement at 15 K for La1.50Sr0.50Ga3O7.25 (space group n. 113, P-421m). Site

x

y

z

˚ 2) Uiso (A

Occ.

4e 4e 2a 4e 2c 4e 8f 4e

0.661(1) 0.661(1) 0 0.854(1) 0.5 0.862(1) 0.912(1) 0.366(3)

0.839(1) 0.839(1) 0 0.646(1) 0 0.638(1) 0.834(1) 0.134(3)

0.491(2) 0.491(2) 0 0.034(1) 0.820(4) 0.690(1) 0.209(1) 0.001(1)

0.0058(4) 0.0058(4) 0.0020(5) 0.0055(4) 0.014(2) 0.008(1) 0.010(1) 0.015(4)

0.75 0.25 1 1 1 1 1 0.125

La1 Sr1 Ga1 Ga2 O1 O2 O3 O4

˚) a (A 8.046(4)

˚) b (A

˚) c (A

Rw

c2

8.046(4)

5.275(4)

15.9%

0.241

˚ ) for the Table 6 e Selected bond lengths (<2.5 A La1.50Sr0.50Ga3O7.25 and La1.50Sr0.50Ga3O7.25 compositions as derived by PDF refinement of neutron diffraction data at 15 K (space group n. 113, P-421m). Atoms pair LaSrGa3O7 Ga2-O2 Ga2-O1 Ga2-O3 Ga1-O3 La1.5Sr0.5Ga3O7.25 O1-O4 Ga2-O2 Ga2-O1 Ga2-O3 Ga1-O3 Ga2-O4

˚) Bond length (A 1.7892 1.8252 1.8395 1.8443 1.799 1.8169 1.8311 1.8325 1.871 2.2625

related to an average structure description, significantly improved the fit. In order to consider a representative local structure and its deviation from the average structure we modelled the PDF of the La1.50Sr0.50Ga3O7.25 sample in a P-1 space group and with a 1  1  2 cell in order to respect the stoichiometry and place a full interstitial oxygen in the unit cell. Two starting positions for the interstitial oxygen and for the atoms in the environment around it have been chosen and are reported in Fig. 5. These positions have been chosen because the one at the centre of the pentagon (5a) has been proposed as a possible position by Kuang et al. [18] and it also corresponds to a local energy minimum based on energy minimization techniques [30]. The position reported in Fig. 5B, where the interstitial oxygen is off-centred from the centre of the pentagon, has been suggested as a global energy minimum in the computational work carried out on this material [30]. First we optimized the average structure starting from the data reported in Table 5; then only the structural parameters of the interstitial oxygen and the atoms around it were allowed to vary without symmetry constrains (let us remember that the structure was modelled in the P-1 space group). This allowed us to use a very small set of parameters thus assuring that they were independent of each other. The final results of the fits starting with the local structures reported in Fig. 5 are shown in Fig. 6. The Rwp is ca. 13% for both the fits of Fig. 6. As it can be observed, the introduction of a local distortion due to the interstitial oxygen improved the fit with respect to the model proposed in Fig. 4B (Rwp ¼ 15.9%). One interesting point is to look at the final position of the interstitial oxygen and of the environment around it obtained starting with model 1 (that place the Oint at the centre of the pentagon) and model 2 (which place the Oint off-centred with respect to the pentagon) which are reported in Fig. 7. Table 7 reports some selected interatomic bond lengths as determined from the refinement of the PDF for the two models while Table 8 reports the atomic positions of the Ga atoms around the interstitial oxygen and the position of Oint. Concerning the fit with model 2 the final result is close to the starting atomic positions of the model used (Ref. [28]).

8078

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

Fig. 5 e Starting model for the local structure refinement.

This is quite reasonable since the initial parameters were already obtained from a computational minimization of the energy of the local structure which was shown to be reliable. The fit using model 1 (Ref. [18]) ended with a significant shift of the interstitial oxygen with respect to the centre of the pentagon that was the initial position used according to Ref. [18]. Such a shift introduces a shorter distance with Ga2 (even though longer than for model 2) and also OeO pairs at ˚ as witnessed also by the correct fit of the peak around 2 A around that distance in the final fitted PDFs (see Fig. 6). The ˚ while Ga1-Oint distance in the refined structure is around 2.5 A ˚ . We also remark that in in the starting model it is close to 3 A both the models the final adp for all the atoms are in the range ˚ [2] thus suggesting that the positions of the 0.004 < u < 0.008 A atoms are reliable. As both models provide improved fits, it is not possible, based on the available data, to unequivocally distinguish which one is more representative of the actual local distortion within the sample. However, it is clear that upon relaxation of the local structure, both models suggest a shift of the oxygen interstitial from the centre of the pentagon.

We would like to stress that our analysis, performed at low-temperature, has led to the conclusion that the most preferred position for the interstitial oxygen is off from the centre of the pentagon (that has been identified as a global energy minimum from modelling [30]). However, at higher temperature also the position at the centre of the pentagon may become populated (as predicted computationally), this being a local energy minimum which is part of the oxygen migration path [30]. We also remark that the diffraction data reported in Ref. [18] actually suggests that the oxygen may be displaced from the centre of the pentagon, as can be appreciated in particularly in the density maps from the MEM (maximum entropy method) analysis of neutron diffraction data at 800  C. However, the use of neutron diffraction lead to a high adp value for the “extra” oxygen in the average structure which in fact did not allow to exactly find a possible small displacement of the interstitial defect. Further analysis by means of MEM technique, possibly at even higher temperature, may improve the experimental visualization of the oxygen displacement and migration path in this system.

˚ with starting local structural model of Fig. 5A; Panel B: Final PDF Fig. 6 e Panel A: Final PDF fit of La1.50Sr0.50Ga3O7.25 up to 5 A ˚ with starting local structural model of Fig. 5B. Blue circles: experimental data; red line: fit of La1.50Sr0.50Ga3O7.25 up to 5 A calculated PDF; horizontal green line: residual. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

8079

Fig. 7 e Local structure around oxygen interstitials as obtained from the PDF fits of La1.50Sr0.50Ga3O7.25 starting with model 1 (A) and model 2 (B). See the text for details.

Table 7 e Selected bond lengths around the interstitial oxygen (Fig. 6) for the final fit of the PDF according to model 1 and 2 (see text for details).

˚) Ga2(a)-Oint (A ˚) Ga2(b)-Oint (A ˚) Ga2(c)-Oint (A ˚) Ga1(d)-Oint (A ˚) Ga1(e)-Oint (A ˚) O1-Oint (A ˚) O4-Oint (A

Model 1

Model 2

1.65(1) 2.72(1) 3.14(1) 2.50(1) 3.31(1) 2.21(2) 1.98(2)

2.436(1) 1.748(1) 3.478(1) 2.008(1) 4.194(2) 2.27(2) 2.06(2)

Table 8 e Atomic positions of Ga atoms around the interstitial oxygen and of interstitial oxygen for the PDF refinement at 15 K for La1.50Sr0.50Ga3O7.25 by using the two model described in Fig. 7. x Model 1 Ga2(a) Ga2(b) Ga2(c) Ga1(d) Ga1(e) Oint Model 2 Ga2(a) Ga2(b) Ga2(c) Ga1(d) Ga1(e) Oint

4.

y

z

0.151(3) 0.651(3) 0.362(8) 0.501(7) 0 0.329(6)

0.348(3) 0.150(3) 0.861(7) 0.508(4) 0 0.247(7)

0.524(4) 0.483(4) 0.484(5) 0.500(4) 0.5 0.496(4)

0.149(5) 0.608(2) 0.357(1) 0.500(6) 0 0.436(2)

0.350(5) 0.152(5) 0.857(1) 0.515(2) 0 0.281(3)

0.513(3) 0.484(3) 0.480(1) 0.501(3) 0.5 0.457(4)

Conclusions

In this paper we reported a neutron total scattering experiment coupled to pair distribution function analysis on

a recently discovered Ga-based oxide-ion conductor characterized by interstitial oxygen migration [18]. This study focussed on the short-range order analysis of the defect structure in La1.50Sr0.50Ga3O7.25 with the aim of shedding light on the localization of the interstitial oxygen and on the distortion induced in the structure by the extra-oxygen. The data analysis revealed that the preferred position, at lowtemperature, for the interstitial oxygen is within the pentagonal ring formed by the Ga1 and Ga2 atoms and shifted with respect to the centre of the pentagon. The interstitial oxygen appears to be coordinated to Ga1 and Ga2 atoms within the first coordination shell.

Acknowledgement We acknowledge the financial support of the INSTM-Regione Lombardia Project “PICASSO” and the Cariplo Foundation through project 2009-2623.

references

[1] Malavasi L, Fisher CAJ, Islam MS. Oxide-ion and proton conducting electrolyte materials for clean energy applications: structural and mechanistic features. Chem Soc Rev 2010;39:4370. [2] Orera A, Slater PR. New chemical systems for solid oxide fuel cells. Chem Mater 2010;22:675. [3] Brett DJL, Atkinson A, Brandon NP, Skinner SJ. Intermediate temperature solid oxide fuel cells. Chem Soc Rev 2008;37: 1568. [4] Wang K, Hissel D, Pera MC, Steiner N, Marra D, Sorrentino M, et al. A review on solid oxide fuel cell models. Int J Hydrogen Energy 2011;36:7212. [5] Lacorre P, Goutenoire F, Bohnke O, Retoux R, Laligant Y. Designing fast oxide-ion conductors based on La2Mo2O9. Nature 2000;404:856. [6] Malavasi L, Kim H, Billinge SJL, Proffen Th, Tealdi C, Flor G. Nature of the monoclinic to cubic phase transition in the fast

8080

[7]

[8]

[9]

[10] [11]

[12]

[13]

[14]

[15] [16] [17]

[18]

[19]

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 8 0 7 3 e8 0 8 0

oxygen ion conductor La2Mo2O9 (LAMOX). J Am Chem Soc 2007;129:6903. Tealdi C, Malavasi L, Ritter C, Flor G, Costa G. Lattice effects in cubic La2Mo2O9: effect of vacuum and correlation with transport properties. J Solid State Chem 2008;181:603. Vega-Castillo J, Mogni L, Corbel G, Lacorre P, Caneiro A. On the thermodynamic stability of La2Mo2O9-d oxide ion conductor. Int J Hydrogen Energy 2010;35(11):5890. Kendrick E, Kendrick J, Knight KS, Islam MS, Slater PR. Cooperative mechanism of fast ion conduction in galliumbased oxides with tetrahedral moieties. Nat Mater 2007;6:871. Li S, Scho¨nberger F, Slater PR. La1xBa1+xGaO4x/2: a novel high temperature proton conductor. Chem Comm 2003;21:2694. Nakayama S, Aono H, Sadaoka Y. Ionic Conductivity of Ln10(SiO4)6O3 (Ln ¼ La, Nd, Sm, Gd and Dy). Chem Lett; 1995:431. Kendrick E, Islam MS, Slater PR. Developing apatites for solid oxide fuel cells: insight into structural, transport and doping properties. J Mater Chem 2007;17:3104. Najib A, Sansom JEH, Tolchard JR, Slater PR, Islam MS. Doping strategies to optimise the oxide ion conductivity in apatite-type ionic conductors. Dalton Trans; 2004:3106. Zang L, He HQ, Wu HW, Li CZ, Jang SP. Synthesis and characterization of doped La9ASi6O26.5 (A ¼ Co, Sr, Ba) oxyapatite electrolyte by a water-based gel-casting route. Int J Hydrogen Energy 2011;36(11):6862. Haugsrud R, Norby T. Proton conduction in rare earth orthoniobates and ortho-tantalates. Nat Mater 2006;5:193. Haugsrud R, Norby T. High temperature proton conductivity in acceptor-doped LaNbO4. Solid State Ionics 2006;177:1129. Malavasi L, Ritter C, Chiodelli G. Investigation of the high temperature structural behaviour of La0.99Ca0.01NbO4 proton conducting material. J Alloys Compd 2009;475:L42. Kuang X, Green MA, Niu H, Zajdel P, Dickinson C, Claridge JB, et al. Interstitial oxide ion conductivity in the layered tetrahedral network melilite structure. Nat Mat 2008;7:498. Skakle JMS, Herd R. Crystal chemistry of (Re,A)2M3O7 compounds (Re ¼ Y, Lanthanide; A ¼ Ba, Sr, Ca; M ¼ Al, Ga). Powder Diffr 1999;14:195.

[20] Malavasi L, Orera A, Slater PR, Panchmatia PM, Islam MS, Siewenie J. Local structure investigation of oxide-ion and proton defects in Ga-apatites by pair distribution function analysis. Chem Commun 2011;47:250. [21] Hull S, Norberg ST, Tucker MG, Eriksson SG, Mohn CE, Stolen S. Neutron total scattering study of the delta and beta phase of Bi2O3. Dalton Trans; 2009:8737. [22] Breger J, Dupre N, Chupas PJ, Lee PL, Proffen Th, Parise JB, et al. Short- and long-range order in the positive electrode material Li(NiMn)0.5O2: a joint X-ray and neutron diffraction pair distribution function analysis and NMR study. J Am Chem Soc 2005;127:7529. [23] Malavasi L, Kim H, Proffen Th. New insight into the properties of proton-conducting oxides from neuron total scattering. Chem Phys Chem 2008;16:2309. [24] Malavasi L. Total scattering investigation of materials for clean energy applications: the importance of the local structure. Dalton Trans 2011;40:3777. [25] Peterson PF, Gutmann M, Proffen Th, Billinge SJL. PDFgetN: a user-friendly program to extract the total scattering structure factor and the pair distribution function from neutron powder diffraction data. J Appl Crystallogr 2000;33: 1192. [26] Farrow CL, Juhas P, Liu JW, Bryndin D, Bozin ES, Bloch J, et al. PDFfit2 and PDFGui: computer programs for studying nanostructure in crystals. J Phys. Condens Matter 2007;19: 335219. [27] Toby BH. EXPGUI, a graphical user interface for GSAS. J Appl Cryst 2001;34:210. [28] Egami T, Billinge SJL. Underneath the Bragg peaks: structural analysis of complex materials. Amsterdam; Boston: Pergamon; 2003. in ES, Juhas P, Proffen Th, Billinge Simon JL. [29] Qiu X, Boz Reciprocal-space instrumental effects on the real-space neutron atomic pair distribution function. J Appl Crystallogr 2004;37:110. [30] Tealdi C, Mustarelli P, Islam MS. Layered LaSrGa3O7-based oxide-ion conductors: cooperative transport mechanisms and flexible structures. Adv Funct Mater 2010;20:3874.