Oxygen transport and surface exchange mechanisms in LSCrF-ScCeSZ dual-phase ceramics

Zonghao Shen*, Stephen J. Skinner, John A. Kilner

Department of Materials, Imperial College London, Exhibition Road, London SW7 2AZ, UK

Corresponding Author Email: zonghao.shen12@imperial.ac.uk

Abstract

For the mechanisms by which the oxygen corporates in a dual-phase composite system, three hypotheses i.e. cation inter-diffusion, spillover type and self-cleaning of the perovskite-structured phase, have been provided in literature. However, experimentally consensus on the most likely mechanism has yet to be reached. In this work, a specially fused sample of the Lanthanum strontium chromium ferrite (LSCrF)-Scandia/Ceria-stabilised zirconia (ScCeSZ) dual-phase materials was investigated. Among the three potential mechanisms, no obvious cation inter-diffusion was firstly observed. A cleaner surface of the ScCeSZ phase was confirmed in the fused sample compared to the isolated ScCeSZ single-phase sample while impurity layers were clearly observed on the LSCrF surface, suggesting the cleaning effect from the perovskite. However, more evidence implies the cleaning effect is not the only reason for the synergistic effects between these two phases. Observations via SIMS analysis lend strong support to the ‘spillover-type’ mechanism as the oxygen isotopic fraction on the surface of the ScCeSZ increased compared to the isolated single-phase and as the distance to the heterojunction increases, the oxygen isotopic fraction decreases. Moreover, oxygen depleted layers were clearly seen on the top layers of the LSCrF surface which may be associated with the higher oxygen diffusivity in the surface/sub-surface layers, oxygen grain boundary fast diffusion and the impurities on the perovskite phase. For this sample, a combination of ‘spillover’ and ‘self-cleaning’ type mechanisms is suggested to be the potential possibilities while the contributions from the cation inter-diffusion for this specific sample is proven to be low.

Keywords: dual-phase composite, oxygen diffusion, cleaning effect, spillover, cation interdiffusion

Introduction

An understanding of the transport mechanisms of oxygen in ceramic oxide materials is fundamental to many physical and chemical phenomena, relevant to technological applications as diverse as Solid Oxide Fuel Cells (SOFCs), Oxygen Transport Membranes (OTMs), -sensors etc. This understanding can be difficult to obtain for many of the complex ceramic oxide systems of technological interest and is dependent upon the accurate measurement of the kinetics of oxygen exchange by a variety of methods. Oxygen exchange has been intensively studied over the past 40 years by the oxygen Isotopic Exchange Depth Profiling (IEDP) technique, for example, in the literature, oxygen diffusion kinetic parameters of the isolated single phase materials e.g. (La,Sr)MnO3 (LSM)1 (La,Sr)(Mn,Cr)O32, (La,Sr)(Mn,Co)O3 (LSMC)3,4, Yittra-Stabilised-Zirconia (YSZ)5–8, (La,Sr)(Co,Fe)O3 (LSCF)9, La2-xSrxNiO4+δ 10 Ceria-doped-Gallium (CGO)11,12 etc. have been reported. This data has allowed a deeper understanding of the mechanisms of oxygen transport in these single phase ceramic materials.

Unfortunately, this is not the case for dual phase ceramic composites, commonly used as OTMs or SOFC cathodes, where many complex effects can take place between electronically conducting perovskite structured component and the fluorite structured ionic conducting component. The volume fraction of each component in the composite and their distribution have important extra effects. Investigations of the oxygen surface exchange and bulk diffusion in dual-phase composites are relatively scarce because of the added complexity of the systems. Two different types of kinetic data can be obtained from ceramic composites using the IEDP technique. By analysing over large areas (e.g. 2-300 μm), much larger than the grain size of the component oxides, ‘effective’ diffusion and exchange parameters can be obtained. If the microstructural parameters are known, for example for a composite electrode, then these ‘effective’ parameters can be used to compare with electrochemical measurement, as the electrochemical measurements also integrate over large length scales.13 More information can be gained by using high lateral resolution techniques such as Focused Ion Beam SIMS (FIB-SIMS) capable of analysis within the component grains, and in this case the kinetic parameters for each phase, when embedded in a composite can be obtained for comparison with the ‘effective’ parameters and for those of the isolated single phase materials. 14

Oxygen diffusion data has been reported in the literature for the LSM-YSZ 15–17 and LSCF-CGO 13,14,18 systems. There are some similarities in the reported behaviour and some important differences. For example, an enhancement in the ‘effective’ surface exchange coefficients has been observed in the LSM-YSZ dual phase system as a function of the volume fraction of the components, suggesting a synergistic effect. On the contrary the LSCF-CGO system showed a scattered variation of the oxygen surface exchange with the volume fraction of the LSCF phase. Evidence of a synergistic effect was apparent for both systems when the microscopic kinetic parameters for the individual grains was examined. At a given temperature and under identical conditions, the oxygen surface exchange rate decreased for the electronically conducting perovskite phase but was enhanced for the ionic conductor. In all cases the values obtained for the composite were compared to the values for the isolated single phases.

Based on the previous studies on the LSM-YSZ and LSCF-CGO systems, three hypotheses of the potential mechanisms have been suggested for the observed microscopic behaviour 14,16,18, namely a ‘spillover’ type mechanism, a ‘cleaning’ effect of the perovskite and transition metal inter-diffusion.

1. ‘Spillover’ type mechanism: the conventional understanding for isolated single-phase material is that the whole surface exchange reaction is assumed to proceed by several intermediate steps and each step can be rate-limiting. For an isolated electronically conducting phase with adequate surface adsorption sites for oxygen reduction and dissociation, oxygen is easily dissociated and reduced on the surface but a low oxygen vacancy concentration limits the bulk diffusion process (e.g. LSM). On the contrary, for an isolated ionic conducting phase (e.g. YSZ), though the high oxygen vacancy concentration provides a high oxygen diffusivity, the low concentration of electronic species and surface adsorption sites limit the reduction and dissociation of oxygen molecules. However, in the dual-phase composite material, the reduced and dissociated oxygen species are able to migrate across the triple-phase boundaries (TPBs) on the surface from the perovskite phase onto the ionic conducting phase surface and then diffuse through the ionic conducting phase. Fig. 1 is a schematic presenting the ‘spillover’ process where the electronic conductor functions as a source of reduced and dissociated oxygen species and the ionic conductor functions as a sink for oxygen bulk diffusion.

Figure 1 Schematic of the ‘spillover’ type mechanism (k is the oxygen surface exchange coefficient)

1. ‘Cleaning’ effect of the perovskite phase: For the isolated single-phase fluorite-structured ionic conducting materials, the surface is likely to be covered by a passivation layer of the common impurities: e.g. silica or calcia and this impurity layer further depresses the oxygen exchange rate on the surface of the ionic conductor 19. However, for a mixed conductor with the perovskite structure, it normally presents a ‘clean’ surface because its structure is capable of tolerating many impurity elements 14,20. Thus, in dual-phase composite materials, the ionic conductor is continuously cleaned by the dissolution of impurities into the perovskite lattice which is believed to be beneficial for the oxygen surface exchange process of the ionic conducting phase21.

1. Transition metal inter-diffusion: during the sintering process transition metal elements, e.g. Fe/Co/Cr, are likely to migrate onto the surface of the ionic conductor. Therefore, the reduced transition metal concentration on the surface of the mixed conductor is considered to be responsible for the decreased k while the enhancement of the surface exchange kinetics of the ionic conductor is activated by the presence of the transition metal species 22.

However, with limited investigations there is no consensus on the most likely mechanism whereby the oxygen diffuses through the dual-phase system and the validity of these hypotheses has not been visually confirmed or excluded except that the possibility of the third hypothesis is suggested to be low for the LSCF-CGO dual-phase system 14.

In the recently reported LSCrF-ScCeSZ dual-phase composite system 23, the basic understanding of the diffusion mechanism is that the surface exchange process is predominantly performed by the LSCrF perovskite phase while the dominant phase for the bulk diffusion is the pure ionic conducting phase ScCeSZ. Moreover, similar to the LSCF-CGO and LSM-YSZ systems, an increased surface exchange rate for the ScCeSZ phase and a decreased surface exchange rate for the LSCrF were also observed. In this present work, the (La0.8Sr0.2)0.95Cr0.5Fe0.5O3-δ (LSCrF) and 10 mol%Sc2O3-1 mol%CeO2-89 mol%ZrO2 (ScCeSZ) powders were used as the starting materials to investigate the possible mechanisms by which the oxygen transports in this dual-phase composite material. To obtain further understanding of the diffusion mechanisms, a specially designed pellet was fabricated and a combination of Secondary Ion Mass Spectrometry (SIMS) and Low Energy Ion Scattering (LEIS) as well as Energy Dispersive X-ray (EDX) spectroscopy was performed on this sample.

Experimental

To obtain an understanding of the potential diffusion mechanisms in the (La0.8Sr0.2)0.95Cr0.5Fe0.5O3-δ (LSCrF)-10 mol% Sc2O3-1 mol% CeO2-89 mol% ZrO2 (ScCeSZ) dual-phase composite system, studies have been performed on a thermally fused specimen composed of two polished ceramic pellets (LSCrF and ScCeSZ). Fig. 2 presents a schematic of the fabrication procedures used to produce the samples for the following experiments.

The powders of the LSCrF and ScCeSZ were both directly supplied by Praxair Inc., USA. The green bodies of the two pellets were achieved by uniaxial pressing and successive isostatic pressing. The individual pressed pellets of each material were sintered at 1450 for 6 hours with a heating and cooling rate of 5 min-1. The single phase sintering results in a larger grain size than is typically found when sintering the composite at the same temperature because of the heterogeneous nature of the grain to grain interfaces. Before thermally fusing the two pellets together, they were both ground using successive grades of SiC papers and finally polished down to ¼ μm finish by using diamond suspensions. This preparation process followed the same procedure as reported earlier to prepare samples for isotopic exchange measurements 24. The ScCeSZ pellet was then placed upon the LSCrF pellet with the polished surfaces in contact with each other. In order to obtain a better contact between the two pellets, an alumina crucible was placed on top of the ScCeSZ piece in order to introduce pressure during the heat treatment at 1400 for 4 hours. After fusing the two pellets into one, a small piece was cut from this thermally fused pellet and the cross section of this piece was subsequently ground and polished down to a mirror finish (1/4 μm final polish, as above) for the following oxygen isotopic exchange experiment. The dry oxygen isotopic exchange (<1 vppm H2O) was performed at 900 in the enriched 18O2 atmosphere for 0.5 hr after a 15 hr pre-anneal in a dry, research grade (>99.999%) 16O2 atmosphere. The detailed information on the oxygen isotopic exchange procedure can be found in 24.

Figure 2 Schematic of the procedures used to obtain the fused sample for subsequent transport property measurement

Transition metal diffusion was firstly investigated by Scanning Electron Microscopy (SEM)-Energy Dispersive X-ray (EDX) spectroscopy. The instrument used was a JSM-6400 SEM (JEOL Ltd.) equipped with an Oxford Instruments INCA EDS. A combination of the Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS, ION-TOF GmbH, Germany) and Qtac100 Low Energy Ion Scattering (LEIS, ION-TOF GmbH, Germany) was further used to investigate the oxygen diffusion behaviour and surface composition of each phase in both the thermally fused sample and the corresponding isolated single-phase pellets.

Additionally, in this work a different sample was also prepared which will be presented in the section 3.2.2 Figure 7. The ScCeSZ green body was placed upon the LSCrF green body after isostatic pressing and the sample was sintered at 1450 for 6 hours. Then a small piece was cut from this thermally fused pellet and the cross section of this piece was subsequently ground and polished before the SEM-EDX analysis.

In this work, three different environments for the two phases will be mentioned with the oxygen diffusion measurements. In order to avoid any misunderstanding, the three different samples will be explained here: LSCrF or ScCeSZ isolated sample is the single phase sample; the fused pellet is the specially prepared sample in this work (Figure 2) and the LSCrF-ScCeSZ composite sample is a mixed dual-phase sample with details in 23.

Results and DiscussionsTransition Metal Inter-Diffusion

Fig. 3(a) is the secondary electron image of the cross section of the fused pellet with the left of the image corresponding to the ScCeSZ phase while the right of the image corresponds to the LSCrF phase, showing the EDX acquisition positions. 11 points were analysed with a gap of 3 μm between adjacent points, to allow for the low lateral resolution of EDX which is limited to around 1μm 25. The longest distance between the analysed points and the heterojunction was 15 μm for both phases. Fig. 3 (b) presents the EDX results of the sample reflecting the distribution of the metal elements across the interface of the two phases. The standard deviation was calculated from the parallel acquisition points and the absence of some error bars is because they are smaller than the data symbols. The atomic fraction of cerium, which is around the detection limit of the EDX technique, has not been presented here. However, a previous study 23 on the homogenously mixed the dual-phase composite samples reveals the possibility of Ce diffusion into the LSCrF perovskite lattice during sintering and the effects of this Ce diffusion process on the oxygen diffusion in the dual-phase composite pellet remains unknown and will not be further discussed.

Figure 3 EDX analysis across the interface of the two phases: (a) secondary electron image showing the acquisition positions (green cross) (b) calculated atomic fraction of each metallic element

An obvious and relatively sharp interface can be clearly observed in Fig. 3 (b) and in the bulk area the cation ratios are in good agreement with the theoretical values (Cr:Fe = 1:1, Zr:Sc = 8.9:2) with slightly lower Sr content (La:Sr>4:1). At the positions close to the heterojunction, the contents of Sr, Fe and Zr varied from the theoretical values, reflecting the phenomenon of cation intermixing at an annealing temperature lower than the normal sintering temperature for dense ceramic membrane. In the later case, where the dual-phase materials are sintered at 1450 for 6 hrs, the influence of cation inter-diffusion should be more significant because of the smaller grain size found for the sintered composite, however the magnitude of this effect remains to be explored. For the samples in this work where the distance from the interface was greater than 15 m, cation inter-diffusion was negligible. Since the following SIMS and LEIS measurements were carried out in areas with a much longer distance to the heterojunction, the influence of the cation intermixing on the oxygen diffusion in the following studies is expected to be low.

SIMS Analysis

Fig. 4 displays the positions of the SIMS analysis areas (analysis area: 100 x 100 μm2; sputtering area: 300 x 300 μm2) in the two phases and the distance from the centre of each acquisition area to the heterojunction. In the ScCeSZ part four areas were analysed with the distance from 120 μm to >1000 μm and in the LSCrF part, one area was measured with a distance of 140 μm from the heterojunction. The distances were estimated using the ToF-SIMS Surfacelab software and the depth profiles were subsequently recorded in these areas by using ToF-SIMS.

Figure 4 Schematic showing the analysis areas in both phases and the distance from the centre of the acquisition area to the interface. The sample was isotopically exchanged at 900 for 0.5 h before SIMS analysis.

ScCeSZ Phase

Fig. 5 shows the normalised 18O fractions from area No.1 to area No.4 obtained using the depth profiling mode at the surface of the ScCeSZ phase in the fused pellet as a function of sputtering time (Fig. 5a) and as a function of distance from the heterojunction (Fig. 5b). The normalised 18O fraction obtained from the isolated ScCeSZ phase sample, which is plotted as the gold shaded area, is also presented in Fig. 5(b). Though the isotopic fractions display some scatter, with the low counting statistics, it still clearly shows that as the distance to the interface increased, the normalised 18O surface fraction of the ScCeSZ phase decreased. Moreover, if the acquisition area was significantly far away from the heterojunction (~1500 μm, area 4), the difference between area No.4 and the isolated ScCeSZ pellet was negligible. Based on the observations in Fig. 5, compared with the isolated pellet, when the ionic conductor ScCeSZ is adjacent to the LSCrF phase, an apparent enhancement in the surface exchange process of the ionic conductor was confirmed, and this enhancement was observed to be related to the distance to the heterojunction.

If this isotopic distribution is thought of as a lateral diffusion profile away from the hetero-junction, then a solution to the diffusion equation for a semi-infinite medium 26 can be applied to the four points and a diffusion coefficient and an ‘exchange coefficient’ can be obtained. This yields values of D* = 2.410-6 cm2 s-1 and k = 8.810-7 cm s-1, respectively. The estimated ‘exchange coefficient’ for the ScCeSZ in the fused pellet is of the same order of magnitude as the surface exchange coefficient of the isolated LSCrF pellet (k = 2.110-7 cm s-1 24). Due to the sparsity of data, the lateral D* and k* of the ScCeSZ are only estimated values with a high uncertainty, hence the exchange coefficients here can be considered as in agreement with each other. However, if the lateral oxygen diffusion coefficient, in the ScCeSZ from the heterojunction of the fused sample is compared to the value of the isolated ScCeSZ phase from surface to bulk (D*=4.710-7 cm2 s-123), this lateral diffusion coefficient is almost one order of magnitude higher. Based on this result, it is probable that what is observed here is not caused by lateral diffusion from the interface but is caused by another mechanism. This could be spillover of oxygen species from the LSCrF phase or exchange between the gas phase and the ScCeSZ where the surface of the ScCeSZ has lower content of blocking impurities on the surface as the interface is approached, and hence a higher local value of k*, i.e. the so called cleaning effect.

Figure 5 Normalised 18O fraction of the ScCeSZ surface at areas 1-4 with the increased distance from the interface (a) as a function of sputtering time (the straight lines are the averaged values of the corresponding area) (b) as a function of the distance from the heterojunction. Golden line represents the averaged result obtained from an isolated ScCeSZ single phase sample isotopically exchanged at the same temperature for the same duration (900, 0.5h) and the golden area includes the error bars. The red line is the fitted oxygen diffusion profile determined from the four depth profile data displayed in Fig. 5(a).

LSCrF Phase

In contrast to the ScCeSZ phase where an increased oxygen isotopic fraction has been observed compared to the isolated single-phase material, the LSCrF phase in the fused sample presents a more complex distribution of the oxygen isotopic fraction in the outermost layers. Fig. 6 displays the diffusion profiles of the LSCrF phase as a component in the fused pellet (Fig. 6a) and the isolated LSCrF phase 24. Clearly, the diffusion profiles of the isolated phase and the LSCrF phase in the composite pellet present different behaviour:

1. For the isolated LSCrF single phase, the normalised 18O surface fraction is around 90% which falls in the region of high where the Isotope Exchange Depth Profile (IEDP) method starts to lose accuracy in the measurement of the surface exchange coefficient 27 because the associated and reduced oxygen species accumulate on the sample surface and wait to diffuse into the bulk due to its low bulk diffusion. For this kind of material, a bulk diffusion limited mechanism was confirmed and, realistically, only a lower bound of the surface exchange coefficient can be achieved. 24

1. However, for the LSCrF phase in the fused pellet, the normalised 18O fraction of the acquisition area with a distance of 140 m to the heterojunction dramatically drops to 47% on the surface compared with the isolated single phase LSCrF where 90% was detected. Moreover, within the first 10 nm beneath the sample surface, very clear oxygen depletion layers were observed. (Fig. 6c highlights the near surface area in Fig. 6a.)

Figure 6 Diffusion profiles of the LSCrF8255 (a) as a component in the fused pellet isotopically exchanged at 900 for 0.5 h; (b) isolated single phase pellet isotopically exchanged 900 for 0.43 h (from 24); Figure (c) magnifies the near surface area of Fig. (a).

Combined with the data obtained from the ScCeSZ phase where the surface exchange coefficient increased as the distance to the interface decreased, this significant depletion of the 18O fraction in the LSCrF phase lends support to the ‘spillover’ type mechanism. For the isolated LSCrF phase, low oxygen vacancy concentration limits the bulk diffusion process, and therefore the reduced oxygen species accumulate on the surface. The normalised 18O concentration on the isolated single-phase LSCrF sample surface is close to the gas phase composition, with further detailed discussion available in the previous work24. For the isolated ScCeSZ sample without an active surface catalyst layer, oxygen exchange is severely limited. For the fused pellet, a much higher surface 18O fraction was seen in the ScCeSZ material while a much lower 18O fraction was observed in the surface regions of the LSCrF material. This implies that the reduced and dissociated oxygen species on the LSCrF surface can easily transport across the heterojunction and then diffuse into the bulk through the ionic conductor, ScCeSZ. More interestingly, an oxygen depleted surface layer or an uphill-like behaviour was observed in the LSCrF component of the fused pellet. In the literature, more obvious oxygen uphill diffusion behaviour was reported by Huber et al 28 on an LSM thin film on a YSZ substrate by applying a cathodic bias. This behaviour was suggested to be attributed to a combination of fast grain boundary diffusion, a cathodic bias and Wagner-Hebb-type stoichiometry polarization of the LSM bulk. In the current work, no cathodic bias was applied but other possible explanations may contribute to the low surface concentration of 18O on the LSCrF phase.

1) Previous studies by Staykov et al 29 on SrTiO3 revealed interesting findings on oxygen dissociation on Sr-O terminated perovskite surfaces, as is the case for LSCrF, (see section 3.3.2) where oxygen vacancies were considered as the active sites and catalysts for O-O bond cleavage. They also found that the energy required for oxygen diffusion between the surface vacancy and sub-surface vacancy was proven to be low compared to migration of these species across the surface. From the information here, a possible picture may be painted that the dissociated oxygen species on the LSCrF surface can either come over the TPBs on the surface and jump to the vacant site in the ScCeSZ phase or jump to the vacant site in the sub-surface of the LSCrF phase and then transport across the heterointerface in the sub-surface layers into the ScCeSZ phase. Hence oxygen species diffuse from the LSCrF phase to the ScCeSZ phase not only on the surface but also in the sub-surface. This is different from the conventional understanding of a ‘spillover’ type mechanism that the reduced and dissociated oxygen species diffuse only from the LSCrF surface onto the ScCeSZ surface.

2) The grain boundary fast diffusion behaviour in the isolated (La0.8Sr0.2)0.95Ce0.7Fe0.3O3-δ (LSCrF73) phase has already been reported [24]. Though in the (La0.8Sr0.2)0.95Ce0.5Fe0.5O3-δ (LSCrF55 or LSCrF in this work) the grain boundary diffusion is not prominent at high temperatures, combining the diffusion coefficient of the isolated LSCrF at 900 with the critical diffusion coefficient for the obvious grain boundary fast diffusion 30, it is likely that at 900 the LSCrF phase is in the transition from Harrison type B to type A preferential diffusion 31. The reduced and dissociated oxygen species may still tend to diffuse along the grain boundaries directly into the bulk. Combined with the above potential reason, a schematic is provided in Figure 7. Because the oxygen diffusivity in the outermost layers is higher than that in the bulk and oxygen species tend to diffuse along grain boundaries into the bulk of the LSCrF phase, the 18O concentration was diluted in the surface and sub-surface layers of the LSCrF.

Figure 7 Schematic of the potential mechanism of the up-hill like behaviour in the LSCrF phase in the fused pellet where Ds, Dgb and Db are the diffusion coefficients of surface/sub-surface, grain boundaries and bulk. The green curves in each phase are the normalised 18O fraction in each phase.

3) The 18O depleted layers of the LSCrF phase could be due to compositional change on the surface. The compositional change can be divided into two different categories. Firstly, despite the EDX results, Fig. 3, suggesting that transition metal inter-diffusion is negligible in this sample, taking the depth resolution of EDX into account, the EDX results reflect the bulk behaviour instead of the behaviour of the sample surface, given that the region showing the depleted isotopic fraction is only ~ 10 nm thick. Moreover, the 18O depleted layers could be related to the impurity layers on the LSCrF phase in the fused pellet, and the impurity layers may also be within the range affected by SIMS ion beam mixing. Thus, further studies using Low Energy Ion Scattering (LEIS) may provide compositional information on the sample surface to a depth of only 1-2 nm, with atomic layer resolution, of the LSCrF material but Zr and Sr may be too close to be resolved in the LEIS spectrum. Details will be provided in the next section: 3.3 - Analysis of Surface Composition by LEIS.

The kinetic parameters for diffusion and exchange were further extracted from the diffusion profiles using TraceX 32,33. Table 1 lists the oxygen diffusion data for the LSCrF phase in 3 different environments: the two described above in Fig. 6, and the diffusivity data from an intragranular depth profile in the LSCrF grains in a LSCrF-ScCeSZ dual-phase pellet detailed in an earlier publication 23.

Table 1 Oxygen diffusion and surface exchange coefficients of the LSCrF phase in different sample configurations

Samples

D* (cm2 s-1)

k (cm s-1)

LSCrF55 (isolated bulk sample)

3.710-12

2.210-7

LSCrF55 (fused pellet)

4.410-12

1.510-7

LSCrF55 (in dual-phase composite pellet)23

1.110-12

1.210-8

In order to achieve a fit to the diffusion profile of the LSCrF phase in the fused pellet, the top layers with severe oxygen depletion were excluded from the fit, which may result in an overestimation of the surface exchange coefficient. Clearly, among the three samples, the surface exchange coefficient of the LSCrF in the dual-phase pellet is one order of magnitude lower than the isolated bulk pellet and is more affected by the ScCeSZ phase. Excluding the systematic errors, the different surface exchange coefficients between LSCrF in the fused pellet and LSCrF in the dual-phase composite pellet (obtained from an intragranular depth profile) 23 can be attributed to the following possibilities:

1. The lateral length scale in the two experiments are different. The grain size of the mixed dual-phase pellet (~20 μm) is much smaller than the distance between the acquisition area and the heterojunction in the fused pellet (>100 μm). As a result, the influence of the ionic conductor phase is limited in the fused pellet compared with the dual-phase pellet while in the dual-phase pellet the reduced and dissociated oxygen species can more easily overcome the heterojunction and then be diffused by the pure ionic conductor into the bulk resulting in the lower 18O surface fraction and surface exchange coefficient of the LSCrF phase.

1. The sample for this study was thermally fused at a temperature lower than the normal sintering temperature and for a shorter duration. The dual-phase pellet was produced by the normal sintering routine where more cation inter-diffusion should occur due to the increased temperature and time. To illustrate this difference, a separate experiment was performed where LSCrF and ScCeSZ green bodies were sintered together, to compare with the earlier experiment where pre-sintered single-phase bodies were fused together. Similar to the first experiment, a small piece was cut from the sintered pellet and the cross-section of the pellet was ground and polished. Fig. 8 presents the elemental maps across the interface of the sintered pellet where clear Sr segregation can be seen at the interface and substantial Fe, Sr and Sc inter-diffusion is clearly observed. Hence for the dual-phase pellet the effects from the cation diffusion were more prominent.

Figure 8 EDX elemental maps across the interface of LSCrF and ScCeSZ in a sintered pellet. The preparation of the sample is similar to the sample shown in Fig.1 except that the two starting pellets are green bodies of the two materials rather than polished sintered pellets.

Analysis of Surface Composition by LEIS

From the SIMS data, where it was observed that the surface exchange of the ScCeSZ phase increased while the 18O concentration on the LSCrF surface decreased, a synergistic effect between these two phases was clearly observed and the contributions from the ‘spillover’-like type mechanism was suggested. To investigate the cleaning effects from the perovskite phase, LEIS was applied in order to observe the variations in the surface chemistry of each phase.

LEIS Spectra of the ScCeSZ Phase

LEIS analysis was performed on both the isolated ScCeSZ single phase pellets and the ScCeSZ phase as a component in the fused pellet. Firstly, LEIS spectra of the isolated ScCeSZ single-phase pellets with different treatment histories: (a) an as-sintered surface, (b) an as-polished surface and (c) the surface after oxygen isotopic exchange annealing at 900 , are presented in Fig. 9.

For the as-sintered pellet (Fig.9(a)), a clear peak which corresponds to Si in the LEIS spectrum indicates the existence of an impurity layer on the ScCeSZ surface. Thus, the scattering intensities of Si and Sc as a function of the dose density of the primary beam/the estimated depth for the as-sintered pellet are plotted in Fig. 9(d). The depth presented is an estimation based on the dose density and is therefore only an approximation of the actual depth. Fig.9(d) shows that from surface to bulk, the Si content decreased reflecting that the Si impurity tends to accumulate on the surface after sintering at high temperature. Moreover, for its analogous material YSZ, it was also reported that a thin impurity film covered the as-sintered sample surface as well as impurities segregating to grain boundaries 34 and this silica passivation surface layer is believed to be detrimental for the oxygen surface exchange process 19,35. The as-polished surface (Fig. 9(b)) presenting bulk information shows a clean surface and no clear extra peaks representing impurities were observed in the LEIS spectra. After oxygen isotopic exchange at 900, the LEIS spectrum of the surface is presented in Fig. 9(c). Na/Mg, Si and K/Ca impurities are clearly observed on the outermost surface of the oxygen exchanged sample, and Zr and Sc have been reduced to subsurface edges compared to the polished sample. Hence, a message obtained here is that impurities have strongly segregated onto the sample surface during annealing at 900 . The origins of the impurities may be due to the impurities at grain boundaries or exsolved from the bulk and during high temperature annealing, the impurities move to the surface along grain boundaries 34. Additionally, impurities from the oxygen isotopic exchange apparatus could also contribute to the contaminants on the sample surface. Hence, for the isolated single phase ScCeSZ pellet, except the as-polished surface which presents the bulk behaviour exhibiting a clean surface, the segregated impurities were observed on both surfaces of the samples after the following heat treatments: sintering at 1450 and isotopic exchange annealing at a temperature of 900 .

Figure 9 LEIS spectra of the isolated ScCeSZ single phase pellets with different treatment histories by using 3 KeV He+ primary beam (a) as-sintered ScCeSZ surface; (b) as-polished ScCeSZ surface; and (c) the surface of the polished ScCeSZ after oxygen isotopic exchange annealing at 900 for 0.5h. Figure (d) presents the scattering intensities of Si and Sc as a function of dose density, i.e. depth profiling spectra of Sc and Si for the (a) as-sintered sample

Further LEIS analysis was performed on two areas of the ScCeSZ phase in the fused pellet and Fig. 10 presents the positions of the analysed areas and the LEIS spectra of the corresponding areas 1 and 2. Due to the lower lateral resolution of LEIS, the analysis area in LEIS is 0.5 mm x 0.5 mm which is large when compared to the analysed area of the ToF-SIMS (0.1 mm x 0.1 mm). This sample was annealed in a dry oxygen atmosphere at 900 with the same thermal treatment history as the sample depicted in Fig. 9(c).

In comparison with the isolated single-phase pellet where clear impurity peaks were observed in the LEIS spectrum, the ScCeSZ exhibited a very clean surface for both analysed areas when the ScCeSZ was fused with the perovskite phase LSCrF. This clean surface of the ScCeSZ phase is direct evidence for the ‘cleaning’ hypothesis i.e. that the perovskite structure which is capable of tolerating impurity species, continuously dissolves the impurities exsolved from the ScCeSZ phase.

Figure 10 LEIS analysis on the ScCeSZ phase as a component of the fused pellet after oxygen isotopice exchanged at 900 for 0.5h, with the same heat treatment history as Figure 9(c), by using 3 KeV He+ primary beam (a) schematic presenting the positions of the LEIS analysis areas relative to the SIMS analysis area; (b) and (c) are the LEIS spectra of the areas 1 and 2, respectively, showing a clean surface with no detectable impurities

LEIS Spectra of the LSCrF Phase

Additionally, LEIS analysis was performed on the LSCrF phase, Fig. 11, in the fused pellet in order to obtain information on the atomic composition on the outermost layers of the LSCrF phase. It is clear that the He+ analysis beam is incapable of distinguishing Cr and Fe. In order to separate the Cr and Fe, a heavier noble gas analysis beam, e.g. Ne+ was required to resolve these features. From the previous investigations, the analogous perovskite single-phase materials present very clean surfaces and no impurities were observed 20,36. However, for the LSCrF phase in the fused pellet, in the first spectrum with zero dose density where the outermost surface elemental information is provided, clear contaminants were observed while the ScCeSZ phase in the same sample (Fig. 10) presents a very clean surface, suggesting the cleaning effects from the perovskite LSCrF phase. Moreover, from the spectra additional interesting information has been obtained:

1. For the B-site elements Cr and Fe, only a small feature is present in the spectrum of the outermost surface while both Sr and La can clearly be observed suggesting the surface of the perovskite material is predominantly A-O layer terminated.

1. Comparing Sr with La, obviously on the sample surface Sr shows a much stronger intensity than the La indicating that there was Sr segregation on the sample surface.

1. On the surface layer, 18O is nearly invisible and after sputtering off several layers 18O appears reflecting the low fraction of 18O on the top layers of the LSCrF phase in the fused pellet. Since no oxygen plasma cleaning was performed before the measurement, an estimation of the normalised 18O fraction can be performed. Based on the previous investigation by Tellez et al 37 that 18O presents 18% higher sensitivity than the 16O, the normalised 18O fraction at the layer 4.3 nm beneath the outermost surface was determined to be 0.70. However, as mentioned above, the depth scale for the LEIS spectra was estimated based on the dose density which may vary between materials and thus introduce uncertainty. The normalised 18O fraction at the layer 4.3 nm beneath the surface measured by SIMS was 0.76. The estimated depth by SIMS is based on the assumption that the sputtering process is even, thus considering the depth resolution of SIMS and the ion beam inter mixing, the normalised 18O fraction at the depth of 4.3 nm is only for reference.

Figure 11 LEIS spectra of the LSCrF phase in the fused pellet by using 3 KeV He+ primary beam

Among these three points, the first two phenomena agree well with the findings in other reported perovskite materials, e.g. La1-xSrxCo1-yFeyO3-δ etc. 20 that the perovskite material tends to terminate with an A-O layer and Sr containing materials tend towards Sr surface segregation. Fig. 12 is the depth profiling spectra of the La, Sr and Cr/Fe elements where the relative intensities of each element as a function of depth can be clearly observed. From Fig. 12, although Sr tends to segregate on the surface of the perovskite LSCrF, the intensity of Sr on the surface is very low, which may be attributed to the impurity layer on the sample surface. Thus, combined with the SIMS depth profile into the LSCrF phase (Fig. 6), the 18O depleted surface layer may also be related to the impurities on the surface of the perovskite phase. To clarify, if the cleaning effect only operates at high temperatures and during the quenching process the impurities in the ScCeSZ phase do not have time to incorporated into the surface of ScCeSZ and dissolve in the perovskite phase, the impurity layers with a lower isotopic fraction will further dilute the isotopic fraction on the surface of the LSCrF phase. Moreover, for the third point, considering the inaccuracy in determination of depth in LEIS, the low fraction of 18O in the top layers obtained by LEIS can be considered as being in good agreement with the SIMS depth profiling results where clear 18O depletion layers were observed confirming a synergistic effect between the two phases.

Figure 12 Depth profiling LEIS spectra of the Sr, Fe/Cr and La elements highlighting the Sr segregation at the surface

Overall discussions

The rather interesting results obtained via SIMS and LEIS analysis along with EDX spectra, have revealed several key points to aid in understanding the mechanisms of surface exchange and diffusion in the LSCrF-ScCeSZ dual-phase system:

Firstly, we will discuss one possible origin of the increase of the surface isotope fraction on the ScCeSZ close to the heterointerface. Previous work had suggested that transition metal impurities could lead to an enhancement of the surface exchange coefficient in the related fluorite ionic conductor ceria-gadolinia, CexGd1-xO2-δ (CGO) 22,35. The SEM-EDX spectra have shown that at a distance of 15 μm from the heterointerface cation inter-diffusion became negligible, which is rather short compared to the distance to the heterojunction of the areas used for the SIMS and LEIS measurements. In other words, for the SIMS depth profiling analysis on the ScCeSZ surface inter-mixing of the transition metal cations has a negligible effect on the final results, in contrast to the case of CGO, and the reasons for the increased 18O fraction are mainly due to other mechanisms.

LEIS spectra of the ScCeSZ surface were acquired from both the isolated single phase pellets with different treatment histories and the ScCeSZ part as a component of the fused pellet after oxygen isotopic exchange, and presented several interesting observations, of which the expanded description is provided in the section above. It was shown that the polished surface was covered by impurities after oxygen isotopic exchange annealing, i.e. in the dry oxygen atmosphere at 900 , while the ScCeSZ phase in the fused pellet with the same thermal history presented a much cleaner surface, i.e. no obvious observed impurities in the LEIS spectra of both acquisition areas. Thus, the conclusion here is that the surface of the ionic conductor has been cleaned due to the proximity of the perovskite phase.

Combining the result of LEIS analysis with the SIMS data, it is unlikely that self-cleaning from the perovskite phase is the only mechanism active in these dual-phase materials. The LEIS spectra revealed that the entire ScCeSZ part in the fused pellet displayed a clean surface and in the event that the cleaning effect was the only valid interpretation, the four SIMS analysis areas should have identical 18O fractions on the surface. However, according to the SIMS data the 18O fraction decreased with increasing distance to the heterojunction. Moreover, another point to be presented here is that compared to its analogous YSZ system 19,38 this isolated ScCeSZ single phase itself exhibits a cleaner surface with only very weak impurity peaks observed in the LEIS spectra. A clean surface is believed to be beneficial for the surface exchange process.

Fig. 5 demonstrated that the normalised 18O surface fraction decreased as the distance to the interface increased. Considering the provided error bars for the normalised 18O fraction, the 18O surface fractions of area 4 on the ScCeSZ side of the fused pellet can be recognised as identical to the 18O fraction of the isolated ScCeSZ single-phase pellet. For area 4, due to its distance to the interface, the influence from the perovskite phase became negligible. Thus, for material fabrication, theoretically a homogenously distributed microstructure with small grain size should be beneficial for oxygen diffusion in the dual-phase system because of the improved accessibility to the ionic phase for incorporation of oxygen diffusion. Furthermore, combined with the lateral diffusion coefficient obtained from Fig. 5, i.e. the diffusion coefficient estimated on the ScCeSZ surface of the fused sample from the heterojunction to the unilateral side, it is one order of magnitude higher than the bulk diffusion coefficient of the ScCeSZ, suggesting that it is not caused by the lateral diffusion but another mechanism. In other words, spillover-type mechanism or the surface cleaning surface make significant contributions.

The lower 18O fraction on the surface of the LSCrF phase in the fused pellet suggested a ‘spillover’-like effect. However, on top of the conventional understanding that the reduced and dissolved oxygen species diffuse onto the pure ionic phase on the surface, the top surface layers, sub-surface layers, i.e. around ten nanometers beneath the outermost surface of the LSCrF phase, present more active behaviour towards oxygen diffusion than the inner (bulk) parts. In the top layers, oxygen species can easily move across the heterojunction and diffuse through the ionic conducting phase. Moreover, the dissociated and reduced oxygen species may tend to diffuse into the bulk of the LSCrF via the grain boundaries. Hence in the top layers the oxygen isotopic fraction was diluted due to the activated surface/subsurface layers and grain boundary fast diffusion, and the 18O depleted layers were observed. LEIS spectra of the LSCrF phase surface also provided considerable useful information. Firstly, the A-O terminated surface and Sr surface segregation were consistent with previous reports on other perovskite materials in the literature 20. According to the intensities of the separated 16O and 18O peaks, the 18O depleted surface layers were also observed in the LEIS spectra which agrees well with the SIMS data. Additionally, impurity layers were also observed on the LSCrF surface which is not only complementary evidence of a cleaning effect but also a potential reason for the 18O depleted layer on the LSCrF surface. To summarise, the 18O depleted layer in the LSCrF phase of the fused pellets may be associated with grain boundary fast diffusion, higher oxygen diffusivity in subsurface/surface layers and the impurity layers on the surface.

Thus, from the above points, the SIMS and LEIS results confirmed the synergistic effect between the pure ionic conductor ScCeSZ and the MIEC phase LSCrF. Transition metal inter-diffusion has a minor influence on the oxygen surface exchange and diffusion behaviour in this fused pellet. However, for the dual-phase composite sintered at higher temperature for longer time, the diffused cations may also have effects on the oxygen surface exchange and diffusion behaviour. A cleaner surface of the ScCeSZ phase has been observed when the ScCeSZ pellets are annealed adhering with the perovskite phase. The detailed mechanism and significance of the cleaning effects by perovskites on oxygen diffusion and surface exchange process is yet to be fully understood but it is believed that this cleaner surface is advantageous for the oxygen surface exchange. Furthermore, the ‘spillover’-like mechanism makes significant contributions to the oxygen diffusion behaviour in the dual-phase system but it is not the only reason responsible for the synergistic effects between these two phases. Therefore, a combination of the ‘spillover’-like mechanism and perovskite self-cleaning mechanism is suggested.

Conclusions

In literature, three potential theories have been provided to explain the synergy of dual-phase composite systems, namely ‘spillover’ type mechanism, self-cleaning mechanism of the perovskite phase and a transition metal inter-diffusion type mechanism. In this present work, by using SEM-EDX, SIMS and LEIS techniques, interesting behaviour has been observed.

1. No obvious cation inter-diffusion was detected by using EDX, suggesting the minor contributions from the catalyst surface layers of the transition metals for this fused sample.

2. Moreover, a cleaner surface of the ScCeSZ was observed due to the presence of the perovskite phase LSCrF despite the distance to the heterojunction, confirming the cleaning theory which is believed to be beneficial for the oxygen surface exchange process.

3. Additionally, as the distance to the heterojunction increased, the normalised 18O fraction decreased on the ScCeSZ surface. However, based on the obtained oxygen lateral diffusion coefficient and bulk diffusion coefficient of the ScCeSZ phase, exclusive the oxygen lateral diffusion, other mechanisms, i.e. both spillover-type mechanism and the cleaning effect from the perovskite phase, make the main contributions.

4. Combined with previous simulation results, further information obtained from the LSCrF side is that the surface and sub-surface area both present high oxygen diffusivity, slightly different from the conversional wise of the ‘spillover’ type mechanism. Besides, on the surface of the LSCrF phase in the fused pellet, the 18O depleted layers may be associated with grain boundary fast diffusion, higher oxygen diffusivity in subsurface/surface layers and the impurity layers on the surface.

To conclude, a combination of the ‘spillover’-like and perovskite self-cleaning is suggested to be the potential mechanisms for the oxygen diffusion in the LSCrF-ScCeSZ dual-phase composite system while the effect from the cation inter-diffusion is proved to be low.

Conflicts of interest

There are no conflicts to declare.

Acknowledgement

The authors would like to thank Praxair Inc., USA for research funding.

Reference

1R. . De Souza, J. . Kilner and J. . Walker, A SIMS study of oxygen tracer diffusion and surface exchange in La0.8Sr0.2MnO3+δ, Mater. Lett., 2000, 43, 43–52.

2E. S. Raj, J. A. Kilner and J. T. S. Irvine, Oxygen diffusion and surface exchange studies on (La0.75Sr0.25)0.95Cr0.5Mn0.5O3−δ, Solid State Ionics, 2006, 177, 1747–1752.

3R. A. De Souza and J. A. Kilner, Oxygen transport in La1−xSrxMn1−yCoyO3±δ perovskites: Part I. Oxygen tracer diffusion, Solid State Ionics, 1998, 106, 175–187.

4R. . De Souza and J. . Kilner, Oxygen transport in La1−xSrxMn1−yCoyO3±δ perovskites: Part II. Oxygen surface exchange, Solid State Ionics, 1999, 126, 153–161.

5P. S. Manning, J. D. Sirman and J. A. Kilner, Oxygen self-diffusion and surface exchange studies of oxide electrolytes having the fluorite structure, Solid State Ionics, 1996, 93, 125–132.

6P. S. Manning, J. D. Sirman, R. A. De Souza and J. A. Kilner, The kinetics of oxygen transport in 9.5 mol % single crystal yttria stabilised zirconia, Solid State Ionics, 1997, 100, 1–10.

7R. A. De Souza, M. J. Pietrowski, U. Anselmi-Tamburini, S. Kim, Z. A. Munir and M. Martin, Oxygen diffusion in nanocrystalline yttria-stabilized zirconia: The effect of grain boundaries, Phys. Chem. Chem. Phys., 2008, 10, 2067–2072.

8M. Gerstl, T. Frömling, A. Schintlmeister, H. Hutter and J. Fleig, Measurement of 18O tracer diffusion coefficients in thin yttria stabilized zirconia films, Solid State Ionics, 2011, 184, 23–26.

9S. Benson, Imperial College London, PhD Thesis, 1998.

10S. Skinner and J. Kilner, Oxygen diffusion and surface exchange in La2−xSrxNiO4+δ, Solid State Ionics, 2000, 135, 709–712.

11E. Ruiz-Trejo, J. D. Sirman, Y. M. Baikov and J. A. Kilner, Oxygen ion diffusivity, surface exchange and ionic conductivity in single crystal Gadolinia doped Ceria, Solid State Ionics, 1998, 113–115, 565–569.

12J. A. Lane and J. A. Kilner, Oxygen surface exchange on gadolinia doped ceria, Solid State Ionics, 2000, 136–137, 927–932.

13A. Esquirol, J. Kilner and N. Brandon, Oxygen transport in La0.6Sr0.4Co0.2Fe0.8O3−δ/Ce0.8Ge0.2O2−x composite cathode for IT-SOFCs, Solid State Ionics, 2004, 175, 63–67.

14J. Druce, H. Tellez, T. Ishihara and J. A. Kilner, Oxygen exchange and transport in dual phase ceramic composite electrodes, Faraday Discuss., 2015, 182, 271–288.

15Y. Ji, J. A. Kilner and M. F. Carolan, Electrical properties and oxygen diffusion in yttria-stabilised zirconia (YSZ)–La0.8Sr0.2MnO3±δ (LSM) composites, Solid State Ionics, 2005, 176, 937–943.

16M. Dhallu and J. A. Kilner, Oxygen transport in YSZ/LSM composite materials, J. Fuel Cell Sci. Technol., 2005, 2, 29–33.

17M. Dhallu, Imperial College London, PhD Thesis, 2006.

18J. Druce, Imperial College London, PhD Thesis, 2010.

19M. de Ridder, R. G. van Welzenis, H. H. Brongersma, S. Wulff, W.-F. Chu and W. Weppner, Discovery of the rate limiting step in solid oxide fuel cells by LEIS, Nucl. Instruments Methods Phys. Res. Sect. B Beam Interact. with Mater. Atoms, 2002, 190, 732–735.

20J. Druce, T. Ishihara and J. Kilner, Surface composition of perovskite-type materials studied by Low Energy Ion Scattering (LEIS), Solid State Ionics, 2014, 262, 893–896.

21M. De Ridder, R. G. Van Welzenis and H. H. Brongersma, Surface cleaning and characterization of yttria-stabilized zirconia, Surf. Interface Anal., 2002, 33, 309–317.

22J. Druce and J. A. Kilner, Improvement of Oxygen Surface Exchange Kinetics for CGO with Surface Treatment, J. Electrochem. Soc., 2013, 161, F99–F104.

23Z. Shen, J. A. Kilner and S. J. Skinner, Mass Transport in (La0.8Sr0.2)0.95CrxFe1-xO3-δ –Scandia-Stabilised Zirconia Dualphase Composite as a Dense Layer in Oxygen Transport Membranes, J. Phys. Chem. C, 2018, acs.jpcc.8b06302.

24Z. Shen, J. A. Kilner and S. J. Skinner, Electrical conductivity and oxygen diffusion behaviour of the (La0.8Sr0.2)0.95CrxFe1-xO3-δ(x = 0.3, 0.5 and 0.7) A-site deficient perovskites, Phys. Chem. Chem. Phys., 2018, 20, 18279–18290.

25P. J. Goodhew, J. Humphreys and R. Beanland, Electron Microscopy and Analysis, Taylor & Francis, London and New York, 3rd Ed., 2001.

26J. Crank, The Mathematics of Diffusion, Second Edition, Oxford University Press, 1979.

27J. A. Kilner and R. A. De Souza, Measurement of oxygen transport in ceramics by SIMS, High Temp. Electrochem. Ceram. Met., 1996, 41–54.

28T. M. Huber, E. Navickas, G. Friedbacher, H. Hutter and J. Fleig, Apparent Oxygen Uphill Diffusion in La0.8Sr0.2MnO3Thin Films upon Cathodic Polarization, ChemElectroChem, 2015, 2, 1487–1494.

29A. Staykov, H. Téllez, T. Akbay, J. Druce, T. Ishihara and J. Kilner, Oxygen Activation and Dissociation on Transition Metal Free Perovskite Surfaces, Chem. Mater., 2015, 27, 8273–8281.

30J. A. Kilner, Optimisation of oxygen ion transport in materials for ceramic membrane devices, Faraday Discuss., 2007, 134, 9–15.

31L. G. Harrison, Influence of Dislocations on Diffusion Kinetics in Solids With Particular Reference To the Alkali Halides, 1960, 1191–1199.

32S. J. Cooper, Imperial College London, PhD Thesis, 2015.

33S. J. Cooper, M. Niania, F. Hoffmann and J. A. Kilner, Back-exchange: A novel approach to quantifying oxygen diffusion and surface exchange in ambient atmospheres, Phys. Chem. Chem. Phys., 2017, 19, 12199–12205.

34K. V. Hansen, K. Norrman and M. Mogensen, TOF-SIMS studies of yttria-stabilised zirconia, Surf. Interface Anal., 2006, 38, 911–916.

35M. De Ridder, R. G. Van Welzenis, H. H. Brongersma and U. Kreissig, Oxygen exchange and diffusion in the near surface of pure and modified yttria-stabilised zirconia, Solid State Ionics, 2003, 158, 67–77.

36J. Druce, H. Téllez and J. Hyodo, Surface segregation and poisoning in materials for low-temperature SOFCs, MRS Bull., 2014, 39, 810–815.

37H. Téllez, R. J. Chater, S. Fearn, E. Symianakis, H. H. Brongersma and J. A. Kilner, Determination of 16 O and 18 O sensitivity factors and charge-exchange processes in low-energy ion scattering, Appl. Phys. Lett., 2012, 101, 151602.

38M. de Ridder, A. G. . Vervoort, R. . van Welzenis and H. . Brongersma, The limiting factor for oxygen exchange at the surface of fuel cell electrolytes, Solid State Ionics, 2003, 156, 255–262.

23