Microstructure of superconducting copper oxycarbonate thin films of the Ba-Ca-Cu-C-0 system Maryvonne Hervieu, Bernard Mercey," Wilfred Prellier, Jan L. Allen, Jean-Franqois Hamet and Bernard Raveau Laboratoire de Cristallographie et Sciences des Matkriaux, ISMRA, Universitk de Caen, Boulevard du Markchal Juin, 14050 Caen Cedex, France The microstructural state of a-axis-oriented superconducting oxycarbonate films, grown by pulsed laser ablation, has been studied by high resolution electron microscopy (HREM). The member rn =3 of the structural family (CaCuO,),( Ba2Cu02C03), is preferentially stabilized, in agreement with the nominal composition of the target, but rn' members ranging between rn' =1 and 11 have been identified. Different types of domain boundaries, mainly parallel to {loo}, or { 1 lo},, involve local distortions of the framework.These non-stoichiometry features were analysed, and were found to involve complex structural mechanisms such as variations of the cationic distribution within the different types of layers, intergrowths perpendicular to the layers, and intercalations of [CaCu,O,] and [Ca,CuO,]-type layers. Laser ablation has been revealed as a very promising method for the stabilization of new phases which cannot be pre- pared according to the usual solid-state chemical processes. Sm, -xSrxCu02.5 -x/2+d and La,BaCu,O,,+, perovskite-related phases have been synthesized recently using this The deposition of unusual species is an interesting challenge for conceiving new materials and for understanding the con- ditions of their stabilization.Recently, superconducting thin films of copper oxycarbon- ates have been synthesized for the first time, using laser ablati~n.~The film consists of an intergrowth of several members of an original structural family with the general composition (CaCuO,),( Ba,CuO,CO,),; the main member of the family which forms the matrix has rn= 3 and n= 1, which corresponds to Ba,Ca3Cu408C0,. This species can be described as a double layer of Cu04 groups sandwiched by pyramidal CuO, layers and interconnected with carbonate layers (Fig. 1). The characterization of these new materials, especially their microstructural states compared to those of the bulk samples prepared by solid-state reactions, is important for understanding the structural mechanisms governing the formation of copper oxycarbonates, which are as yet ill-known.This paper presents the microstructural features which have been observed in superconducting oxycarbonate thin films. Experimental The target preparation and the conditions of the film deposition have been previously described., The substrate was a (001) single crystal of LaAlO,; the film was deposited at 625"C, at a pressure of 0.2mbar with a gas mixture of 6% co2-94% 0, followed by slow cooling in 500 mbar of 02.The high resolution electron microscopy (HREM) was performed with a TOPCON 002B instrument, with a point resolution of 1.8 A; the thin film was observed along a direction perpendicular to the substrate. Energy dispersive scattering (EDS) analyses were systematically performed. The resistance measurements showed a broad transition with the T,onset ranging from 80 to 100 K and a T,(R=0)of 58 K.Results A first investigation3 of the thin film showed that the material is characterized by a structure directly related to an oxygen- deficient perovskite: slices of variable thicknesses of the so-called 'infinite layer structure' CaCuO, are stacked along c; in the electron diffraction (ED) patterns, streaky lines are observed along c* due to the variations of the slice thickness, but nodes are observed, generated by the frequent stabilization of one of the members of the structural family. For a nominal composition Ba2Ca,Cu,0g of the target, the mean rn value is 3.The formation of 90" oriented domains is also a characteristic of the material. This is clearly evidenced in the ED patterns recorded along the direction perpendicular to the substrate (Fig. 2); the patterns systematically exhibit a perovskite-type subcell with two perpendicular streaky lines, corresponding to the c* axes of the two 90" oriented components. The reconstruc- tion of the reciprocal space as well as the XRD patterns confirm that the parameter along tFe direction perpendicular to the substrate, a, is close to 3.86 A. The contrast of the enlarged HREM images was inter-pretated, in a first step, on the basis of experimental images previously reported for copper oxycarbonate~~ and infinite layer structure^;^ it allowed a structural model to be proposed3 based on the intergrowth of CaCu0,-type slices separated by one layer of carbonate groups sandwiched between two barium layers.The formula (CaCuO,),( Ba,Cu02C03), has been pro- posed to express the structure of these oxycarbonates. The idealized structure of the [m=3, n =11 member, which is the most frequently observed, is given in Fig. 1. Image interpretation In order to interpret the numerous features which were observed for this film, simulated images were calculated, using the MacTempas program and varying the focus values and the thickness of the crystal. The positional parameters which were used were deduced from the theoretical model proposed in Fig. 1; they are given in Table 1.The microscope parameters are V=200 kV and the coefficient of spherical aberration is Cs =0.4 mm. The focus series calculated for a crystal thickness of 31 A ds given in Fig. 3 (a)for focus v!lues ranging from 200 to -950 A. For a focus value of -250 A, close to the Scherzer value, the cations positions are imaged as dark spots and the rows of brightest spots correspond to the positions of the oxygen vacancies within the [Ca], layers. For a focus value close to -550 A, the positions of the cations appear as bright spots, the two brightest ones being correlated to the barium positions. Note that, as previously rep~rted,~ the carbonate layers exhibit a typical contrast !or different focus values,f, close to 100, 0, -150 and -850 A; it consists mainly of single rows of very J.Muter. Chern., 1996, 6(2), 165-173 165 suhst rat e 'substratef+substrate Fig. 1 Idealized drawing of the Ba,Ca,Cu,O,CO, structure, deposited on an (001) single crystal of LaAlO,. The film is a-axis oricnted. Fig.2 Enlargcd [OOl] ED pattern of the oxycarbonate film; the intense reflections are those of the perovskite subcell. The streaked lincs are correlated to the existence of different periodicities along c; the nodes correspond to the 001 reflections of the m=3 member. Only one set of reflections has been indexed; the other set is oricntccl at 90 to this set. bright spots and is maintained for higher thickness values [Fig. 3(h)]. Two typical images are given as examples in Fig.3(c) and (d), with the m values indicated by white numbers. In Fig. 3(c), the cation: are imaged as white spots for a focus value,j, close to -550 A: for m= 3, double barium layers (large white spots) Table 1 Ba,Ca,Cu,O,CO, : pptional prameters of the idealid tetragonal structure," a=3.86 A, c= 17.5 A atom x J' 0.0 0.0 0.389 0.0 0.0 0.0 0.0 0.0 0.189 0.5 0.5 0.094 0.5 0.5 0.289 0.5 0.0 0.094 0.5 0.0 0.289 0.5 0.5 0.389 0.747 0.5 0.5 0.5 0.5 0.5 Space group P4,mmm, oxygen atoms located in ideal positions. Occupancy factors have been fixed to 1, except for O(4) which is supposed to be in a split position (z=0.25). are intercalated between quadruple copper layers (small white spots). In Fig. 3(d),the cations are imaged as dark spots for a focus value close to -250 A.so that the C03groups are easily identified as rows of small grey spots surrounded by four bright spots [see Fig. 3(a)]. Thus, the good fit between the experimental and calculated images confirms the structural model (Fig. 1) according to the following sequence of layers 166 J. Murer-. C~W/H..1996, 6(2), 165-173 Fig. 3 Cdlculdted imdges of Ba,Ca,Cu,O,CO, The positional parameters are taken from Table 1 (a) focuo seriec for d thickness, t. of 31 4 (the projected potential is indicated beneath each image, (h) thickness oeries for AfzO. Experimentdl imagex recorded for (0A/= -550 A and (d)A/= -250 A stacked along c: Ba0-CO-Ba0-Cu0,-Ca-Cu0,-Ca-CuO,-Ca-CuO,-BaO. The [CuO,] layers with adjacent [BaO] layers form pyrami- dal CuO, layers, whereas the [CuO,] layers sandwiched by calcium layers consist of CuO, square planar groups.The CO, groups are intercalated between two successive [BaO], layers. The parameters of the different members can, therefore, be easily estimated in a tetragonal cell, according to the fcllowing relations: 11 2h =up;c' 2 +rn x cCcz(7.9+y12 x 3.3)A where(aBCC cBCCis the c parameter of Ba,CuO,CO,, which is close to 2 x ap, and ccc is that of the infinite layer structure CaCuO,, which is close to 3.3 A. Film morphology a-Axis orientation of the film. The a axis orientation of the film implies that the carbonate layers and the copper layers are perpendicular to the substrate (Fig.1). Referring to the infinite layer structure films" where the copper layers have been observed parallel to the substrate (in agreement with the good fit of the a and b parameters of the film and the substrate), this can be considered a rather surprising point. Such an orientation may simply originate in the deposition temperature. which has been observed to be a highly critical factor; such behaviour could be compared to that previously observed for the 123 superc~nductor;~ however, deposition experiments show that they differ at higher temperatures since complete destruction of the oxycarbonate film is observed. A second factor could play an important role; i.e., the CO, partial pressure, which would favour the presence of carbonate groups in every structural level, i.t..in every perovskite-type layer parallel to the substrate plane. Mosaic-like morphology. The overall images show a mosaic-like film morphology which results from the existence of 90" oriented domains. An example is given in Fig. 4. The formation of 90' oriented domains is a frequent feature in ordered perovskites, resulting from the establishment of the structural mechanism along one of the equivalent directions J. Meter. Clienz., 1996, 6(2), 165 173 167 Fig.4 Overall [OOl] image of the film. The boundaries are mainly parallel to {loo}, or { llO},. Three types of boundary are observed, depending on the orientation and of the way the junction between the different layers is formed; they are marked by black numbers on white arrows.of the perovskite subcell, provided that the structure does not involve a strong distortion of the framework. Two ex- amples are well known in the layered cuprate compounds i.e. the twinning domains in YB~,CU~O~-~and Bi,Sr,Ca, -1C~,02m+ ,. This phenomenon is rarely observed in the thallium- and mercury-based copper oxycarbonates [1201][S2CC], but it is a systematic feature for the 123-type oxycarbonates, YCaBa,Cu,(N0,),(C03), -,Oil and (Y, -xCax)(Ba,Sr)2,Cu3,-107n-3C03.4'8In both compounds, the orientation of the domain boundaries was observed to be mainly parallel either to the { 110}, or to the {loo}, planes of the perovskite subcell. In the present film, three types of boundaries are observed (Fig.4). Boundaries 1 and 3 are parallel to {loo},, i.e. parallel or perpendicular to the carbon- ate layers, while boundary 2 is roughly parallel to { 1 i.e. oriented at approximately 45". The atomic arrangement at the level of {loo), junction is rather simple. For boundary 1, the 90" oriented copper layers are directly connected to one another as shown schematically in Fig. 5(u),whereas for boundary 3 the junction consists of a triple [BaO-CO-BaO], layer built up from one CO layer sandwiched between two BaO layers [Fig. 5(b)]. In fact, such junctions involve local distortions due to the mismatch between the layer spacings of the two adjacent domains. This is illustrated in the enlarged image of the type 1boundary (Fig. 6) which evidences a significant buckling of the layers at the junctions; this allcws the mismatch between the Cu-Cu distance: (ca.3.3 A along c,) and Cu-0-Cu distances (ca. 3.9 A along b,) to be compensated. For the { 110}, junctions (labelled 3 in Fig. 4), the situation is more complex and depends on the members which are connected on both sides of the boundary. In contrast to the {loo}, junctions, the { 1lo}, are not really planar; their nature depends on the thickness of the copper layers (m value) which may differ from one domain to the other. An example is shown in Fig. 6(u), where the { 1lo}, boundary (indicated by the dashed line) separates two domains, A and B. Starting from the upper left part of the image, the sequences of the different members, marked by black numbers, are: domain A, 5-6-6-2; domain B, 3-2-2-2-3-5.The [BaO-CO-BaO] triple layers (marked by white arrow- heads) intersect at the level of the boundary; no strong local distortion is needed for such a configuration. A model can thus be proposed to illustrate the way the different layers are connected through the boundary [Fig. 6(b)]: the [Ba0lm layers are 90" oriented and the junction between the 90" 168 J. Muter. Chem., 1996, 6(2), 165-173 Fig. 5 Idealized drawing of the layers junctions between {loo}, boundaries: (a) boundary marked 1: copper layers are connected to copper layers; (b) boundary marked 3: one [Ba0-CO-BaO] triple layer borders on the boundary oriented square-planar CuO, group layers can be effected through copper-oxygen octahedra and pyramids.Note that for this ideal {110}, boundary, the interlayer distances of domains A and B coincide so that one does not observe any mismatch at the junction, in contrast to that observed at the {100}, boundary. The only distortion would result from the existence of one CuO, pyramid or one CuO, octahedron to effect the change of the direction of the copper layers. When the [Ba0-CO-BaO] triple layers intersect closely, but not exactly at the same level, one observes shifting and bending of the layers in order to ensure a direct intersection. Two examples are indicated in Fig. 6(a). In the bottom part of the image, one m =2 and one rn =6 member of domain A must connect with one m=2 and one rn=3 member of domain B.In the upper part, one m =5 and one m =6 member of A have to be connected to two rn =2 members and one m =3 member of B. The differences between the m values of the two domains imply that the [BaO-CO-BaO] triple layers cannot intersect at the level of the boundary. In such a configuration, complex phenomena like shifting, bending and interruption of the layers facilitate a direct connection. In the upper part, the [BaO-CO-BaO] triple layers are shifted so that the rn=2 and rn= 3 members of B are transformed into m= 1 and m=4 members, whereas in the bottom part of the image, the layers are bent (curved arrow). Lastly, when the m values of the A and B domains are very different, e.g. m=6 with m=l and m=2, the [BaO-CO-BaO], triple layer is simply interrupted at the level of the boundary (marked by a black arrowhead).The distortion of the matrix due to the translations and bending of the [BaO-CO-BaO] triple layers, results very often in small mis-orientations of the adjacent domains. In this example, the c axes of domains A and B are no longer perpendicular, the angle is close to 94" and area B is slightly tilted. Fig. 6 (u)[OOl] image illustrating the connection of the layers through one { 1l0lp boundary (type 1, marked in a white arrow) and one (loo), boundary (type 2, marked in a white arrow). The {110}, boundary is drawn as a dashed line and the 111 values of the different members are indicated as black numbers. The domains are labelled A (left) and B (right).(h)Idealized drawing of the layer junctions through a { 1lo), boundary.Different members of the Ba-Ca-Cu-C-0 system As observed in Fig. 3 and 4, for example, the origin of the streaks observed along the c* axis is undoubtedly correlated to the existence of coherent intergrowths along c, of different m and ii members. The average /nand n values of the matrix, determined from the images of different areas of the film, are 3 and 1, respectively [Fig. 3(c) and (d)]. Members with /TI ranging from 1 to 12 and ti= 1 have been identified; the two limiting members, /?I= 1 and nz= 11, are shown in Fig. 7(u) (the white numbers in the black circles correspond to the IZI values). Note that the high-m members often arise from a structural mechanism which simply corre- sponds to the replacement of one [BaO-CO-BaO] triple layer with a [Ca-Cu0,-Ca] layer.This is illustrated in Fig. 7(u) where the carbonate layers are imaged as rows of very bright spots. It can be seen (white arrowheads) that most of the larger members result from the replacement of three adjacent [BaO-CO -BaO] layers by three [Ca-Cu02-Ca] layers; in this way, two adjacent members, wzl and m2,are replaced by a larger member, 111 =IZZ~+tn2+2. ((I) 1Fig. 7 Overall image showing the existence of I and I?? 111 members (Lvhite numbers). The white arrowheads indicate the formation of large 111 values as ;I result of the substitution of a [BaO CO BaO] triple layer by a [Ca-CuOz-Ca] triple layer. (h) Enlarged image of such a substitution, with one HI=1 and one 111=3 mcmber repliiced bj one IH =6 member.((,) Idealized drawing of the mechanism. Such a replacement poses the problem of the carbonatei copper layer interconnection because two [BaO] layers ate 3.9 A apart whereas two [Ca] rows are separated by 3.3 A. The way this replacement is effected is shown in the enlarged image of Fig. 7(h).where the cation positions are imaged as bright spots: one IH =1 and one rn=3 member are transformed into an nz=6 member. It can be seen that the rather important difference of interlayer distances is easily accommodated within the matrix by a smooth buckling of the adjacent layers. The idealized arrangement is shown in Fig. 7(c).Such an accommo- dation illustrates the great flexibility of the framework.This remarkable adaptability of the framework originates other types of non-stoichiometry phenomena. Tu7o such phen- omena are shown in Fig. X(rr), indicated by white arrowheads and small arrows, where carbonate layers are subjected to 1ocal interruption and/or trans la t ion mec h a11isms. First. .I. ,‘Lltrter. Chm., 1996, 6(2), 165-I73 169 Fig. 8 ((I) [OOl] image where the carbonate layers are interrupted (white arrowheads) andlor translatcd (small whitc arrows). Idealized drawing of the two types of defects: (h)interruption; and (c) shifting. [BaO-CO-BaO] triple layers are replaced, along b, by [Ca-Cu0,-Ca] layers according to the mechanism described above, but the event extends over a segment which is only a few polyhedra long [see white arrowheads in Fig.X(u)]. It involves the local formation of a mixed carbonkopper layer [Fig. 9(h)]; this point will be discussed further. For the second mechanism [small white arrows in Fig. 8(a)]. the [BaO-CO-BaO] triple layers are shifted by upalong c so that one barium layer out of two remains unchanged whereas the other is connected to a calcium layer and the carbon layer is connected to a copper layer [Fig. 8(c)]. In a number of places. the two barium laycrs surrounding a carbon layer, are translated simultaneously in opposite directions; an example is observed in the bottom part of the micrograph [curved arrow in Fig. 8(u)];such shirtings and interconnections of the layers often involves the formation of ill-ordered slices.Cationic distribution Owing to the nature of the different layers. as seen above, cationic exchanges between Ba and Ca on one hand, or between Cu and C in the layer located between the two [BaO] layers, on the other hand. can be expected. Such points are important if one takes into account the fact that supercon- ducting properties could be relatcd to copper-for-carbon substi- tution or carbon deficiency. as was discussed previously.3 Two examples of variations of the cationic distribution are shown in Fig. ~(LI)and (h)which correspond to the same area recorded with two difl'erent focus values (-550 and -250 A, respect-ively); in the first one, the cations are imaged as bright spots [Fig.9(u)] whereas in thc second one.they appear as dark spots [Fig. 9(h)]. In the bottom part of the micrographs, the contrast observed Fig. 9 [OOl] images of-an areit where cation suhctitutions are observed: ((I) Af? -550 A and (h)Af. -250 A. (c) Calculated images comparing a carhonatc layer [CO] with a copper layer [CumO] intercalated between the two [BaO] layers (crystal thickness =31 A). at the level of the [CO] layer varies locally (white arrowheads). In Fig. 9(u), the small grey spots sandwiched between the barium rows (brightest spots) corresponds to carbon positions; at the level of the defect. two grey spots are replaced by brighter spots (white arrowheads). This suggests that carbon is locally replaced by a heavier atom, which is supposed to be copper.In Fig. 9(b), the phenomenon is reversed and two darker spots (white arrowheads) are observed at this level. Such an observation is consistent with the existence of mixed 'C-Cu' layers. 'Theoretical images. calculated on the basis of a full copper and a full carbon occupancy, keeping the BaO-BaO interlayer distance constant indeed allows the two species to be differentiated as shown for two focus values Af= -250 and -550 A [Fig. 9(c)],in agreement with previous studies of 123-oxycarbonates.' Therefore. this defect can be interpreted as the replacement of two adjacent rows of carbonate groups by two rows of CuO, pyramids [Fig. 10(tr)]. Fig. 10 ((11 Tdealrmi model of the defect correspondiilg to 'I local Cu for C submtution 01) ~ded~l/Kd model of the [RaZCuO,CO,], [Bn('uo,]2 defect Fig.11 ([I) [OOl] images. The curved arrow indicates an area where a regular Cu for C substitution is observed; the new periodicity, 3 x tip, is observed between the small white arrows. (h)Idealized model. In the areas indicated by curved arrows in Fig. 9(a) and (h), one observes a variation of contrast at the level of the calcium atoms; the small dots correlated to the Ca positions are replaced by very bright dots in Fig. 9(a) and by very dark dots in Fig. 9(b) (both similar to those observed for barium positions); simultpeously, the interlayer distance is increased from 3.3 to 3.8 A and the contrast at the level of copper atoms is not modified. These observations suggest that the sequence Ba0-CuO2-Ca-CuO,-Ca-CuO2-Ba0-CO is re-placed by Ba0-Cu0,-Ba-CuO2-Ba-CuO2-BaO-CO i.e., the [Ba,CuO,CO,], [BaCuO,], member of the structural family is formed.Thus. at this level, the double oxygen-deficient perovskite layer [CaCuO,], built up from one [CuO,], layer of square-planar groups sandwiched by two pyramidal copper layers is interrupted. and replaced by a double perovskite layer [BaCuO,],, which exhibits an infinite layer type-structure,8 as schematized in Fig. lO(h). The local substitution of copper for carbonate groups is also evidenced in Fig. 11(a). In this figure, where the cations are imaged as dark spots and the carbonate rows as rows of bright spots, one can observe a zone where the variation of contrast at the level of the carbonate row is more frequent, and consists of the disappearance of the bright spots which are replaced by grey spots.At the level of the curved arrow. this substitution is periodic and corresponds to a tripling of the h parameter, i.e. 3 x ap along b (between the two small white arrows): two grey spots alternate with one bright spot, so that locally two carbonate rows out of three are replaced by two rows of CuO, groups. The corresponding idealized model is shown in Fig. 11(h).The junction between one carbonate group and one copper square-planar group involves the formation of one CuOs pyramid without any variation of the oxygen content, as previously shown in the 123 Such a feature corresponds to the local stabilization, ordered or not, of [Ba,Cu, +x02(C03)1 ,][CaCuO,],,, members, which, from the ~ charge-balance point of view, involves an increase of the copper valency.Intergrowth with other structural units The forlnation of extended defects resulting from the insertion of a structural unit which is different from that of the mother structure, has been observed in the matrix. ‘CaCu,O,’ layer. The first example is shown in Fig. 12(a), in which the position: of the cations are imaged as bright spots (Af’close to -550 A). In the regular matrix, the rows of bright spots running alorg c (corresponding to copper layers) are separated by 3.86 A, and the two successive rows of grey spots Fig. 12 (ti) [OOl] HREM image: the formation of double copper layers appears as elongated bright spots (indicated by curved arrows).(h)Projection of CalCuO,, perpendicular to the square-planar CuO, groups. (c)Idealized model of the defect. viewed along a. (corresponding to calcium layers) are 3.86 A apart. At the level of the defect (curved arrows), one observes the formation of large bright rows running along c; the variation of contrast arises at the level of ;i [CuO,] layer. with elongated bright spots (correlated to copper positions) coupled with a significant increase of the interlayer distances: two successive rows of grey spots (Ca layers) whith sandwich a brighto row (copper) are now separated by 5.7 A. instead of by 3.86 A. Such geometrical characteristics suggest the existence of a double [(CuO)-] layer built up from edge-sharing CuO, square-planar groups; this structural unit corresponds to a CaCu,O, layer;” the idealized model of the defect viewed perpendicularly to the square-planar groups is shown in Fig.12(h). Taking into account the orientation of the infinite J. hlater. C’hcw., 1996. 6(3). 165-173 171 layer structure in the film, one [C~CU,O,] layer will b: observed along that projection as two adjacent Cu rows, 1.9 A apart [Fig. 12(c)]. In order to confirm the interpretation of the contrast at the level of the Ca and Cu layers, images calculations have been performed with positional parameters deduced from the theoretical model proposed in Fig. 12(c);the projection of the cations is shown in Fig.13(u). This structure is built up from the intergrowth of one CaCu,O,-type layer with two [CaCuO,] layers ('infinite layer' structure). The image calculated for similar focus (close to -550 A) and thickness values [Fig. 13(b)] is in agreement with the exper- imental contrast. The formation of such double copper layers within the oxygen-deficient perovskite matrix can be compared to what happens in the 123 compounds, where similar features were originally observed in the form of defects" and, until they were characterized in the form of a single phase, the '124' and related compounds." One can observe that shiftings or interruptions of the [BaO -CO-BaO] triple units, similar to those described in the previous section. are also observed through the [CaCu,O,] triple layer.[Ca,CuO,] layers. A second example of additional layers is shown in Fig. 14(u), where the cations are imaged as dark spots; the carbonate rows appear as rows of very bright spots and the smaller bright dots are correlated to the zones of weak potential located between the calcium atoms. At the level of the defects (curved arrows), one observes the formaiion of double rows of bright sppts, with spacings of about 2.2 A along band translated by 1.9 A along c,which suggests the existence of double calcium rows; such a contrast is usually observed in Fig. 13 ((I) Projection of the structure built up from the intergrowth of one [CaCu,O,] unit with [CaCuO,], units. The doublc copper layers areo indicated byo curved arrows.(h) Calculated image (thick- ness = 31 A, Afz~ 550 A). 172 J. Muter. Chenz., 1996, 6(2), 165-173 Fig. 14 (a) [OOl] HREM image: the forination of double calcium layers appears as two rows of staggered bright dots. (h) Idealized model of the defect. thin films of perovskite-related phases' and is easily interpret- able. It is correlated to the local formation of rock-salt-type layers within a perovskite-type matrix, as shown in the idealized model of Fig. 14(h). Thus, locally the structure may consist either of Ca,CuO,-type ribbons built up from CuO, groups intergrown with CaO rock-salt layers, or of La,CuO,-type ribbons built up from octahedral [CuO,] layers intergrown with CaO rock-salt 1a~ers.l~ To effect the formation of rock- salt-type layers, one can assume that oxygen is located in the two adjacent calcium layers [(CaO),] and, therefore, that the adjacent square groups are tilted 90" with respect to the CaCuO, matrix.Here again, the defect is associated with shiftings or interruptions of the [BaO-CO-BaO] triple layers. The formation of rock-salt-type layers as extended defects has been previously observed in perovskite-like matrices. In bulk material^,'^ the defective rock-salt layers form generally infinite straight or broken (with 90'' angles) lines. In thin films, the rock-salt units are generally limited to segments of a few nanometers in length; such limited lengths imply that one perovskite layer is directly connected to one rock-salt layer, in spite of the structural differences (interatomic distances and atomic coordination); however, the stabilization of copper- based 'collapsed' phases, which have been recently i~olated,'~ l7 shows that such connections are accommodated without any problem, owing to the great flexibility of the perovskite framework. Concluding Remarks This study of a superconducting oxycarbonate thin film deposited on an LaAlO, single crystal has allowed the structure and microstructure of the new material to be understood; it shows that the tetragonal phase, whose parameters can be easily related to the perovskite and infinite-layer struc-ture accprding to a z h z up and c 2cBCC+m x ccc z (7.9+ m x 3.3) A, is deposited with an a-axis orientation and with b and c parallel to the [1001 direction of the LaAlO, perovskite.The alignment of the [loo] and [OOl] directions of the oxycarbonate with the [loo] direction of the substrate is ensured in spite of the mismatch between the parameters (3.78!A for aLaMO3compared to am=,=3.86A and c,,,=~= 17.8A for the film); however, as shown in the case of the a-axis 123-type films deposited on an MgO substrate," the relative orientations of the film and the substrate (bfil,parallel to bsubs) cannot be considered as the result of an epitaxy mechanism, but only as a way to minimize the energy. The a- axis orientation of the oxycarbonate film could be correlated to several factors: the thermal conditions of deposition, the C02 partial pressure or, possibly, a problem of ptting of the parametys of the substrate and of the film, 3.86 A and 17.8 5 x 3.56 A being, respectively, longer and shorter than 3.788 A.The easy formation of oriented domains could be also a consequence of this problem of mismatch between the latt- ices. The oxycarbonates such as S~,CUO~CO~'~ or the 123- related oxycarbonate~~*~and oxycarbonates of the system Ba-Ca-Cu-C-0 prepared under high as bulk materials, do not exhibit such behaviour. Further investigations of thin-film oxycarbonates deposited on various substrates will be necessary to answer this question. The numerous structural features which occur in the super- conducting oxycarbonate thin film have been analysed. They involve various structural mechanisms, such as coherent intergrowths of different members of the (CaCuO,),( Ba,CuO,CO,), family or of different structural units parallel or perpendicular to the carbonate planes, and variations of the cationic distributions within the different types of layers.The formation of intergrowths is a classical mechanism observed in the layered cuprates and is a direct consequence of the flexibility of the framework which allows, in other respects, the stabilization of such phases. The vari- ations in the cationic distribution are favoured by the ability of copper and carbonate groups to form mixed layers on one hand, and by the ability of Ca and Ba to be located between copper layers on the other hand. The defects observed in such films may significantly influence the superconducting properties of these materials.There is no References A. Gupta, B. Mercey, M. Hervieu and B. Raveau, Chem. Muter., 1994,6, 101 1. R. Desfeux, J. F. Hamet, B. Mercey, C. Simon, M. Hervieu and B. Raveau, Physica C, 1994,221,205. J. L. Allen, B. Mercey, W. Prellier, J. F. Hamet, M. Hervieu and B. Raveau, Physicu C, 1995,241,158. M. Hervieu, C. Michel, M. Huve, C. Martin, A. Maignan and B. Raveau, Microsc. Microanal. Microstruct., 1993,4,41. B. Mercey, A. Gupta, M. Hervieu and B. Raveau, J. Solid State Chem., 1995,116,37. N. Sugii, M. Ichiharu, K. Hayashi, K. Kubo, K. Yamamoto and H. Yamauchi, Physica C, 1993,213,345. J. B. Barner, C. T. Rogers, A. Inam, R. Ramesch and S.Bersey, Appl. Phys. Lett., 1991,59,742;J. F. Hamet, B. Mercey, M. Hervieu and B. Raveau, Physica C, 1992,193,465. 8 M. Yoshimoto, M. Nagata, J. P. Gong, H. Ohkubo and H. Koinuma, Physicu C, 1991,185-189,2085. 9 Ph. Boullay, B. Domenges, M. Hervieu and B. Raveau, Chem. Muter., 1993,5, 1683. 10 C. L. Teske and H. K. Muller-Buschbaum, 2.Anorg. Allg. Chem., 1969,370,134. 11 B. Domenges, M. Hervieu, C. Michel and B. Raveau, Europhysics Lett., 1987,4, 211. 12 P. Bordet, C. Chaillout, J. Chenavas, J. L. Hodeau, M. Marezio, J. Karpinski and E. Kaldis, Nature, 1988,334,796. 13 C. L. Teske, H. K. Muller-Buschbaum, 2. Anorg. Allg. Chem., 1969, 371, 325. 14 V. Caignaert, M. Hervieu, N. Nguyen and B. Raveau, J. Solid State. Chem., 1986,62, 281.15 M. Hervieu, C. Michel, A. Q. Pham and B. Raveau, J. Solid State Chem., 1993,104,289. 16 Y. Ieda, H. Ito, S. Shimomura, Y. Oue, Inaba, Z. Hiroi and M. Takano, Physicu C, 1989,159,93. 17 M. Hervieu, M. T. Caldes, D. Pelloquin, C. Michel, S. Cabrera and B. Raveau, J. Solid State Chem., in press. 18 J. F. Hamet, B. Mercey, M. Hervieu, G. Poullain and B. Raveau, Physicu C, 1992,198,293. 19 D. V. Fomichev, A. L. Kharlanov, E. V. Antipov and L. M. Kovba, Superconductivity, 1990,3,216. 20 T. Kawashima, Y. Matsui and E. Takayama-Muromachi, Physicudoubt that the ideal formula (B~,CUO~CO~),(C~CUO,)~ C, 1993,224,69. should not lead to superconductivity owing to the absence of mixed valence for copper and of hole reservoirs in the [BaO-CO-BaO] triple layers. Thus, the replacement of C03 groups by copper polyhedra, in the form of isolated defects or ordered domains as evidenced in this study, could explain the superconducting properties of these films. 21 M. Alario Franco, P. Bordet, J. J. Capponi, C. Chaillout, J. Chenavas, T. Fournier, M. Marezio, B. Souletie, A. Sulpice, J. L. Tholence, C. Colliex, R. Argoud, J. L. Baldenedo, M. F. Gorius and M. Perroux, Physica C, 1994,231,103. Paper 5/040461; Received 22nd June, 1995 J. Mater. Chem., 1996,6(2), 165-173 173