Compositional effect of Cr contamination susceptibility of La9.83Si6−x−yAlxFeyO26±δ apatite-type SOFC electrolytes in contact with CROFER 22 APU

Apatite-type lanthanum silicates (ATLS) are attracting great interest as a new class of solid electrolytes possessing high oxide-ion conductivity at relatively low temperatures for solid oxide fuel cells (SOFC). In this study, doped ATLS of the composition La9.83Si6−x−yAlxFeyO26±δ (x: 0, 0.25, 0.75, 1.5 and/or y: 0, 0.25, 0.75, 1.5) were successfully prepared by solid state chemistry. They were brought into direct contact with CROFER-22 interconnector alloy in order to study Cr migration into the electrolyte. Due to inconclusive SEM-EDX results, a depth profile was acquired by Laser Induced Breakdown Spectroscopy (LIBS) and the results showed that the increase of Fe concentration in the apatite oxide's composition enhanced Cr uptake. At the same time, lower conductivity values were measured for the materials after Cr contamination i.e. in Fe containing ATLS. No significant change in conductivity was found for Fe-free ATLS sample.


Introduction
Research effort has been attracted towards the study of apatitetype lanthanum silicates (ATLS) materials as potential electrolytes in solid oxide fuel cell (SOFC) technology mainly due to their high conductivity at moderate temperatures (600-800 C). [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] However drawbacks have also been identied for silicate apatite electrolytes and their practical application. High sintering temperatures are required for the preparation of dense electrolyte layers 1,3,5,8-10 but composition can tune the refractoriness of such materials. 3,16 Generally such silicate apatites show high structural stability but electrode deposition is not straight forward. 17 Furthermore poor cell or half-cell performance is observed due to Si migration towards the electrolyte/ electrode interface. 17,18 Poor electrode performance is found due to solid state reactions between ATLS and certain types of electrodes. [16][17][18] A review on the above advantages, drawbacks including fabrication issues and approaches towards cell and/or suitable electrodes is given by Sadykov et al. 16 In the design of a SOFC stack the interconnector plays a signicant role in its structural and electrical behavior. It is a structural component in contact with both cathode and anode electrode of the SOFC and must be oxidation resistant, impermeable to the diffusion gases, and chemically stable. Another important function of interconnect is the separation of fuel and air as well as the electronic conduction among adjacent cells. [19][20][21][22][23][24][25] A variety of advanced ceramic materials have been tested, with LaCrO 3 as the state of art ceramic material but in recent years metallic interconnectors gained more attention due to their advanced properties and features compared to the ceramics. Metallic interconnects show lower cost of manufacturing, easier processing, higher electrical and thermal conductivity, high toughness, etc. [23][24][25][26][27][28][29][30][31] The fact that the SOFCs are tending to operate at intermediate temperatures (600-800 C) is also enhancing the potential of high temperature oxidation resisting alloys (HTORs) as interconnectors rather than ceramics. The potential use of chromia-forming ferritic stainless steel has been investigated the past years because of the conductive nature of the formed Cr 2 O 3 layer as compared to the traditional Al 2 O 3 and SiO 2 insulators. 23,32 Cr ferritic steels such as CROFER 22 APU are the most widespread SOFC interconnects. [33][34][35][36][37][38] The weakness, though, of such chromia-forming ferritic stainless steel is volatility of Cr 2 O 3 and CrO 2 (OH) 2 at operating temperatures and conditions of the fuel cells. Such chromia species formed on state of art electrodes such as La 1Àx Sr x MnO 3 (LSM) or La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 (LSCF), migrate and deposit on the triple-phase boundary (TPB) in contact with the electrolyte in use causing poisoning, overall cell performance and structural changes and eventually failure. 21,39 Especially under current, chromium is mainly deposited near the TPB along the perimeter of the pores, on the surface of the yttria stabilized zirconia (8YSZ) electrolyte particles, thereby decreasing the number of active sites necessary for oxygen reduction. 39,40 CROFER 22 behavior has been thoroughly studied in contact with electrode materials. 31,35,[37][38][39][40][41][42][43][44] For the case of the state of the art LSM or LSCF cathodes, it has been reported surface segregation and migration of cationic species on the surface of the cathodes from CROFER 22 APU. Cr species are accumulated at the interface of cathode/electrolyte and on the TPB leading to lower power densities and degradation of the cell operation. In order to overcome such issues, the research effort has been focused in the development of Cr-tolerant cathodic electrodes 45 but also in the application of suitable ceramic diffusion barrier coatings. 43 Similar issues of cell degradation due to Cr are recently reported also for solid oxide electrolysis cells (SOEC). 45 Thus Cr species can affect both cathode and anode cermet side of the fuel cells. It is also reported that the extent of Cr poisoning depends not only on cathode materials but also on electrolyte to be used with cathode. 39 However regarding a new group of medium range SOFC electrolytes such as ATLS no data have been reported regarding spatial distribution of Cr ions and their poisoning effect. In the present work we investigate the interaction of ATLS based electrolytes with CROFER 22 APU metallic interconnect focusing on Cr migration. The evaluation of Cr migration into the electrolyte was monitored by SEM-EDX and laser induced breakdown spectroscopy (LIBS) measurements. LIBS was applied as a far more sensitive and versatile method. 46 LIBS, besides its industrial, 47,48 environmental, 49 biological 50 and cultural heritage 51,52 applications, is a 'noninvasive' technique that can be applied in situ to provide a fast tool for depth prole analysis as only a few mg are ablated from the sample. [53][54][55] The samples needed required no preparation so it is a very useful tool as a non-destructive analysis technique. The depth proling of LIBS have found many applications in metals, analyzing multilayer metallic coatings, 56,57 protective paint coatings of the naval sector 58 and the oxidation behavior of metal-based super alloys. 59 On the other hand, in the eld of ceramics, LIBS has been mainly applied for the depth prole analysis of cultural heritage materials such as unglazed ceramics 60,61 and the discrimination between glaze and main body of attic black ceramics, 51 while on technical ceramics, the analysis of yttria stabilized zirconia ceramic layer thermal barrier coating on a super alloy substrate 62 has been reported. A thorough review with recent LIBS applications in various elds was recently published by Carter et al. 63 and references therein. In this study we apply LIBS on fuel cell type materials, proving its exibility and analytical capacity in another eld of technical ceramics. The aim of this study has been the evaluation of ATLS materials on Cr poisoning by Cr species directly generated by the state-of-the-art interconnect CROFER 22 APU. For this reason, CROFER 22 APU and ATLS pellets were brought in contact at temperatures 900-1100 C for 250-1000 h. The results of LIBS, combined also with prolometric measurements showed a compositional dependence on La 9.83 Si 6ÀxÀy Al x Fe y O 26AEd (x: 0, 0.25, 0.75, 1.5 < 1.5 and/or y: 0, 0.25, 0.75, 1.5) ATLS materials regarding their susceptibility to contamination by Cr. AC impedance results showed that Cr poisoning has a negative effect on the electrical properties of the apatite silicates studied with a trend similar to the compositional trend of their Cr susceptibility.

Materials synthesis
The following apatites compounds: La 9.83 with purity of 99.9% have been treated at 700 C for 2 h in order to eliminate humidity, carbonates and impurities. The mixtures were annealed twice at 1400 C for 10 h and ball milled for 24 h. Particle size distribution was followed by laser particle size analysis using a Malvern Mastersizer 2000 (Fig. 1S †). Pellets of 11 mm diameter and 1 mm thickness were uniaxially pressed at 3 MPa and sintered at 1500 C for 1 h. Relative density was measured with Archimedes method and crystal phase formation was followed by powder XRD diffraction.
Samples preparation and thermal treatment CROFER 22 APU square pieces of 12 mm Â 12 mm Â 0.1 mm were kept in contact with the apatite pellets under constant uniaxial weight pressure using dense alumina plates (Fig. 1). These apatites-CROFER 22 APU sets were treated at 900 C for 1000 h, at 1000 C for 500 h and 1100 C for 250 h in atmospheric air with heating rate of 2 K min À1 . Aer thermal treatment, the apatite pellets were diametrically cut, casted and ne polished on the cross cut surfaces for the SEM-EDX study. No particular sample preparation was required for LIBS measurements.
Characterization of tested samples XRD diffraction analysis was performed using CuKa radiation, l ¼ 1.5406Å, 2q range from 15-75 with scan step of 0.01 per min in a Siemens D5000 diffractometer. Both the prepared powders and the tested pellets were measured. ICDD PDF-2 Release 2000 database was used for the identication of crystal structure with support by crystal impact Match! Soware followed by Rietveld renement with FullProf Soware. XRD analysis was performed on the surface of the apatites in contact with the CROFER 22 APU in order to identify any solid state reaction at the interface. The surfaces of the apatites both in contact with the CROFER 22 APU and the open ones were observed by means of Scanning Electron Microscopy (SEM) using a JEOL6380LV, and by Laser Induced Breakdown Spectroscopy (LIBS) using LIBS2500 device, by Ocean Optics. The depth of LIBS ablation was measured by prolometry (Ambios Technology XP-2). The effect of Cr poisoning on the electrical properties of ATLS samples was investigated. AC impedance spectroscopy measurements were performed on ATLS pellets aer contact with CROFER 22 APU using a SP-150 Potensiostat (Biologic Science Instruments).

LIBS specications and parameters
The laser source of LIBS was a Q-switched, Nd:YAG laser (Model: Ultra CFR, Big Sky Laser) delivering 8.5 ns pulses, at 1064 nm, with repetition rate of 15 Hz. Plasma was generated by focusing the laser beam on the sample's surface, through a 70 mm, quartz, plano-convex lens. The light emitted by plasma was then collected by a hepta furcated ber (one-to-seven furcation) and driven into seven HR2000 Spectrometers (Ocean Optics), each equipped with a 2048-element linear CCD array. All spectrometers are triggered to acquire and read out data simultaneously, providing the emission spectrum from UV to near IR region (200-980 nm). 51 Spectra were captured and continuously saved by LIBS soware, providing a quick 'depth prole' investigation. [51][52][53] Therefore it was necessary to calibrate the sputter rate during LIBS in order to be able to evaluate diffusion proles. Laser pulses of 35 mJ energy were applied at the same spot of the apatite's pellets. The successive ablation of material led to the gradual penetration of laser beam from the surface to the inner bulk. Spectra were captured and continuously saved by LIBS soware, providing a 'depth prole' investigation.
In the end of each thermal treatment the apatite samples did not had any visual changes or ndings. Nor shrinkage either color change was observed (Fig. 3). Oxidation in square CROFER 22 APU supports is clearly seen indicating signicant surface formation of Cr 2 O 3 layer aer a typical operational time of 500 h at 1000 C.
Properties of each sample composition aer thermal treatment at 1100 C for 250 h are depicted at Table 1. The XRD patterns aer the thermal treatment at 900, 1000 and 1100 C (Fig. 2) did not reveal any crystal phase changes or formation of any new oxides. Apatite structure is maintained in all cases with no solid state reaction identied. For example at 1100 C the characteristic two major peaks of apatite structure at 2q ¼ 31.1 and 2q ¼ 32.2 showed a small shi indicating the doping effect of Al and Fe into Si site of a La 9.33 Si 6 O 26AEd apatite structure as reported elsewhere. 3,4 This small shi at major peaks was preserved even at the samples' treatment at 1100 C indicating that no phase transition can be deduced by the interaction of apatites and CROFER 22 APU. No further shi of major peaks of the apatite spectra is found and no signicant changes in the lattice parameters were observed aer the thermal treatment. It may be initially concluded that no extended Cr introduction was present in the apatite structure.
This result is in good agreement with the results of McFarlane et al. 64 Through their systematic study of ATLS doping they concluded that no signicant solubility of Cr was possible on   the ATLS phase. As reported, chromium forms LaCrO 3 phase due to its octahedral rather than tetrahedral coordination preference. However their approach was on a synthesis level with stoichiometric amounts of chromium. By the SEM/EDX analysis of CROFER 22 APU specimens (Fig. 4) the visually observed surface oxidation is conrmed as a thermally grown oxide layer predominately consisting of Cr. Namely aer the treatment of CROFER 22 APU at 1100 C for 250 h, a layer of about 3 mm was created providing a Cr-rich area in-contact with apatite samples (Fig. 4). On the other hand, EDX analysis on the ATLS materials in contact with CROFER 22 APU did not clearly revealed the existence of Cr element close to the counter contact interface. In all cases the measured concentration was very low compared to the analytical method limitations and the detectability limits (<1 wt%) of the SEM/EDX unit in use. Furthermore the close positioning of Cr (L) and O (K) peak as well as the lack of Cr (K) peak limited the quantication process towards clear results on Cr content (Fig. 4S-7S and Table 1S †). No Cr was identied on the side not in contact with CROFER 22 APU in all ATLS samples. Thus, LIBS was applied in order to overcome the analytical issue named above and also to investigate its application potential as a more sensitive analytical method. From the earliest studies sensitivity of LIBS when testing solids, was reported in the ppm range with precision <10%, depending of course on various factors. 65,66 LIBS laser pulses prolometry measurements revealed a variation in the mean ablation depth per LIBS pulse among samples (Fig. 5). Aer 30 pulses LASO proved its higher resistance to the mean ablation depth per pulse reaching at 32 mm while LAFSO-1 reached 43 mm, LAFOS-2 reached 51 mm and LFSO 60 mm. A clear trend in DpP with increasing Fe content is observed. This can be attributed to the refractoriness decrease of ATLS caused by Fe content. 67 This observation is in good agreement with Cowpe et al. 68 study that reported a relationship between sample hardness and LIBS plasma properties. Fig. 6 shows typical LIBS spectra of LAFSO-2 pellet, in the 355-368 nm spectral window.
In the contact side the three Cr characteristic lines are present, with descending intensity aer 1, 8 and 30 laser pulses ( Fig. 6 and 7). For comparison purposes the spectrum of the open side of the pellet is shown aer 1 laser pulse, were the Cr lines are absent. LASO pellets do not present any Cr migration into their bulk, while in the LAFSO-1 and LAFSO-2 pellets the Cr migration reaches approximately a depth of 12 mm and 15 mm respectively. The effect is more intense in the case of LFSO pellets were the Cr is still identied aer 30 pulses at approximately 60 mm depth (Fig. 7). As indicated by the qualitative results of SEM/EDX, it is conrmed by LIBS that the higher is the Fe content of ATLS the easier Cr migrates into the ATLS tested. However a remarkable behavior is observed for the non Fe containing ATLS i.e. La 9.83 Si 4.5 Al 1.5 O 26AEd LASO which is not susceptible at all in Cr migration under the tested conditions. No Cr was detected by LIBS. These are interesting results as in early ATLS synthesis studies by XRD, Cr did not show solubility in order to substitute Si in the sublattice of SiO 4 towards the formation of Cr doped apatite structure. 64 It is known that doping ATLS with Fe leads to increasing unit cell volume due to Fe 3+ larger ionic radius compared to Al 3+ or Si 4+ and as the Fe content increases the structure is found hyperstoichiometric in oxygen under oxidative conditions. 67 In the same study Mossbauer measurements identied Fe 4+ presence increasing with   increasing total iron content in La 9.83 Si 6ÀxÀy Al x Fe y O 26AEd caused by the A site deciency. Similar stabilization of Fe in high oxidation states has been observed elsewhere. 69 As explained by Kharton this tendency is probably associated with a Frenkeltype disorder in O sites, induced by lanthanum vacancies. 67 As the Fe content increases lattice distortions increase and O anions displacement too towards interstitial positions. As a result La 3+ cations as well as their vacancies are rearranged and promote stabilization of extra O into formed oxygen vacancies due to Frenkel-like disorder. However in our case the increasing tendency of Cr uptake with Fe content can be attributed to a compensation role of Cr ions on such structural distorted sites. Such an effect may have a negative impact on the conductivity properties that are strongly related to the structural disorder of ATLS structure. 67 In ATLS materials of such composition dopant's spatial distribution inhomogeneity has a strongly negative effect on oxygen mobility too. 4 Detailed studies are needed though to identify the position and coordination of Cr ions in La 9.83 Si 6ÀxÀy Al x Fe y O 26AEd structure.
The effect of Cr poisoning on the electrical properties of apatite silicates was evaluated by impedance spectroscopy. Typical results from the LFSO, LAFSO-1, LAFSO-2 and LASO samples are shown in Fig. 8 and summarized in Table 2. The impedance spectra in all samples consist of well-dened semicircular graphs: a large high frequency (HF) from the bulk component of the impedance and a small low frequency (LF) from grain boundary. Similar results are presented by Cao 70 and Gasparyan et al. 3 as far as the contributions to the overall resistance. The total conductivity of the ATLS samples tested is found lower as the Fe-content increases. Fe rich samples have also higher Cr contamination. The higher the Cr contamination is, lower total conductivity of ATLS sample is found.
In Table 2 a comparison of the conductivity values at 700 C is made, between Cr free ATLS samples from our previous work 3 and the results on ATLS of the same composition and preparation method aer being in contact with CROFER 22 APU. All Fe-containing ATLS compositions aer their contact with CROFER 22 APU suffer a signicant decrease in the conductivity values. LASO, on the other hand, that has no Fe in its composition and no Cr contamination shows no decrease in its conductivity. Thus Cr poisoning is signicantly affecting the electrical properties of Fe containing ATLS samples.
In our previous studies the materials with the best conductivity are proven to be the Fe containing such as LFSO or LAFSO-1 and not LASO. Thus when in contact with Cr species a protective layer should be considered such as a Fe free ATLS   i.e. LASO. However in these previous studies, 3 we also report a strong inuence of the synthesis and the powder processing rather than composition on conductivity properties for ATLS. Furthermore independent studies showed that La 9.83 Si 4.5 Al 1.5 -O 26.8 exhibits optimum conductivity. 71 From a fabrication aspect and considering the above results Fe free ATLS such as La 9.83 Si 4.5 Al 1.5 O 26AEd (LASO) may be used with no constraints regarding Cr contamination. However if a Fe rich material e.g. LFSO, is to be used as electrolyte, a layered conguration may be designed utilizing a protective layer such as LASO type ATLS in order to prevent Cr contamination from electrode/interconnect side.

Conclusions
By SEM/EDX and LIBS analytical techniques, apatites of the La 9.83 Si 6ÀxÀy Al x Fe y O 26AEd general formula show a clear compositional effect on their susceptibility on Cr contamination originated from CROFER 22 APU. By SEM/EDX it was not possible to quantify Cr content and the results were inconclusive. On contrary LIBS due to its much lower detection limit clearly identied Cr in the La 9.83 Si 6ÀxÀy Al x Fe y O 26AEd samples with a depth prole clearly depending on the materials composition. Prolometry was useful for evaluating the etching depth for LIBS method. The results indicated that the increase of Fe in the apatite samples seems to enhance Cr uptake. This was mainly related to the increased lattice distortions with increasing Fe 3+ content. La 9.83 Si 4.5 Al 1.5 O 26AEd LASO was found not susceptible at all in any Cr contamination under the tested conditions. At the same time AC impedance measurements showed that LASO conductivity was also not affected. In contrast the Fe rich ATLS samples resulted in signicantly lower conductivity values aer being in contact with CROFER 22 APU. Total conductivity values of fresh Fe rich ATLS samples can be orders of magnitude higher but suffer a drastic decrease aer Cr contamination. This drastic loss shows compositional effect similar to their susceptibility on Cr contamination. The La 9.83 Si 4.5 Fe 1.5 O 26AEd material with the higher level of Cr contamination showed the lowest conductivity but also the highest conductivity loss compared to a fresh non contaminated composition. For Fe free ATLS no signicant change in conductivity was observed. Thus La 9.83 Si 4.5 Al 1.5 O 26AEd type of materials may be suggested as advantageous ATLS electrolytes due to their chemical inertness to Cr contamination from the triple phase boundary (TPB) Cr accumulation. Alternatively such materials can be applied as a Cr blocking layer between Fe rich ATLS type electrolytes and their respective electrodes. Such a feature may simplify the housing of an ATLS based SOFC and minimize overall cell degradation issues from Cr poisoning.