- Email: [email protected]
The structure of BaCu3O4 particles occurring on thin HoBa2Cu3O7 films prepared by MOCVD
Physica C 328 Ž1999. 211–220
The structure of BaCu 3 O4 particles occurring on thin HoBa 2 Cu 3 O 7 films prepared by MOCVD H.W. Zandbergen a,) , J. Jansen a , V.L. Svetchnikov a , I.E. Graboy b, S. Samoylenkov b, O. Gorbenko b, A.R. Kaul b a
National Centre for HREM, Laboratory of Materials Science, Delft UniÕersity of Technology, Rotterdamseweg 137, Delft 2628 AL, The Netherlands b Moscow State UniÕersity, Department of Chemistry, Moscow 119899, Russia Received 9 February 1999; received in revised form 2 October 1999; accepted 5 October 1999
Abstract The structure of BaCu 3 O4 phase occurring as particles on the surface of Ž001. RBa 2 Cu 3 O 7 epitaxial films prepared by metalorganic chemical vapor deposition ŽMOCVD. has been investigated with quantitative electron diffraction and HREM. The orthorhombic unit cell is a s 1.097Ž9. nm, b s 0.554Ž3. nm, c s 0.394Ž2. nm with space group Cmmm, the values being in agreement with X-ray diffraction ŽXRD. study. The structure consists of alternating Cu 3 O4 and Ba layers along the c-axis. The compound is stabilised due to the formation of low-energy coherent boundaries with RBa 2 Cu 3 O 7 andror perovskite substrate. q 1999 Elsevier Science B.V. All rights reserved. Keywords: BaCu 3 O4 ; HoBa 2 Cu 3 O 7 ; MOCVD
1. Introduction The intensive study of complex oxides over the last decades brought up many new individual compounds and structural families. One of the most intriguing research followed the discovery of A ny 1Cu nq1O 2 n family Žwhere A s Ba, Sr, Ca., which includes so-called ‘‘infinite layer’’ IL Ž n s infinity. and ‘‘spin ladder’’ SL compounds Ž n G 3. w1,2x. Most of these phases are metastable under ambient conditions and have been, at first, prepared )
Corresponding author. Tel.: q31-15-278-2266; fax: q31-15278-6730. E-mail address: [email protected] ŽH.W. Zandbergen.
by means of high-pressure synthesis w3x. More lately, MBE and PLD w4,5x, among other deposition techniques, were adopted to prepare them also as epitaxial films. The IL structure is very simple and consists of 2D CuO 2 planes separated by anion-free A layers. These cuprates are superconducting with Tc values up to 100 K, provided charge carriers are doped in CuO 2 planes, e.g., due to the presence of vacancies in A sites w6x. The building block of the copper– oxygen layer of SL compounds could be presented as a quasi-one-dimensional fragment of CuO 2 layer, joining together two or more Cu–O chains, or ‘‘legs’’, into a ‘‘ladder’’. The ladders are connected with each other with a shift by one Cu–O distance along the chain direction to form Cu nq 1O 2 n layers. Like in the IL structure, two adjacent Cu nq 1O 2 n
0921-4534r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 4 5 3 4 Ž 9 9 . 0 0 5 3 7 - 7
212
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
layers in ladder compounds are separated by anionfree A layers. The spin gap, giving a name for SL, is either opened or closed, for ladders with an even or odd number of legs, respectively w7x. There was also a theoretical prediction of superconductivity in slightly doped A ny 1Cu nq1O 2 n ladders with an even number of legs w7x. Eventually, Sr14yx Ca x Cu 24 O41 cuprate, containing two-leg Cu 2 O 3 ladder, was shown to be a superconductor under high pressure with Tc of 12 K w8x. Until now, this is the only known example of a superconducting cuprate without CuO 2 layer. The search for new phases in A– Cu–O systems proceeds with the use of high-pressure synthesis and film growth techniques. Recently, superconductivity with Tc above 120 K has been found in Ba 2 Ca ny1Cu nO x family synthesised under high pressure w9x, and a new high Tc superconductor in thin films of Ba–Cu–O system has been observed w10x. BaCu 3 O4 phase was first reported in Ref. w11x as an oriented impurity phase appearing in superconducting YBa 2 Cu 3 O 7y d single crystals. Surprisingly, in spite of the numerous research in the field over the last decade, BaCu 3 O4 was never reported, to the best of our knowledge, as a single-phase material. Like the IL and SL compounds, BaCu 3 O4 is believed to be a metastable phase. The thermodynamical stability under various conditions has been proved so far only for four barium cuprates: Ba 2 CuO 3qx , BaCuO 2q y , Ba 2 Cu 3 O5qz and BaCu 2 O 2qw . A very detailed study on the thermal stabilities of Ba–Cu–O system is reported in Refs. w12–14x. Recently, we observed the presence of oriented BaCu 3 O4 in epitaxial RBa 2 Cu 3 O 7 thin films ŽR — rare earth element. grown by metalorganic chemical vapor deposition ŽMOCVD.. Also, we have prepared single-phase BaCu 3 O4 films on LaAlO 3 and SrTiO 3 substrates. The present paper deals with the structural analysis of BaCu 3 O4 performed by high-resolution electron microscopy and electron diffraction. The study of the stability of BaCu 3 O4 will be reported elsewhere w15x.
2. Experimental Epitaxial Ž001. RBa 2 Cu 3 O 7y d ŽR s Lu,Ho,Y,Gd. and BaCu 3 O4 thin films were grown on Ž001. LaAlO 3
Žhere and below, the indices for LaAlO 3 are given for a pseudo-cubic perovskite unit cell. and Ž001. SrTiO 3 single-crystal substrates by the single-source flash evaporation MOCVD w16x. The deposition runs were carried out at temperature 8008C–8358C and oxygen partial pressure pŽO 2 . in the range of 1.7–2.5 mbar. The growth rate was about 0.4 mmrh. After the deposition, all the films were cooled down under oxygen flow with an intermediate annealing at 4508C for 1 h. The surface morphologies and cation compositions of the films were studied with the use of CAMSCAN scanning electron microscope with EDX element analysis system. X-ray diffraction ŽXRD. study was carried out using four-circle SIEMENS D5000 diffractometer with CuK a irradiation. The lattice parameters of BaCu 3 O4 crystal lattice were derived from d hkl values for in-plane and out-of-plane reflections using the standard least-square procedure. The peaks positions for a given reflection were measured for all possible values of w with a fixed x angle, then averaged and corrected in relation to the nearest peaks from a single-crystal substrate. Electron transparent areas in cross-section were obtained by ion milling with the film facing away from the ion guns. Electron microscopy was performed with a Philips CM30UT electron microscope with a field emission gun operated at 300 kV and a Link EDX element analysis system. Electron diffraction patterns were recorded with a 1024 = 1024 pixel Photometrix CCD camera with a dynamic range of 12 bits. Electron diffraction was performed with spot sizes of about 10 nm, using a 15 mm condensor lens aperture. Exposure times ranged from 0.5 to 2 s. To reduce the electron-microscope-induced contamination, the specimen was cooled to about 100 K. A small spot size for electron diffraction is used to have a relatively small variation in the diffraction area of thickness and misorientation since most crystals are wedge-shaped and show often local variations in misorientation. Calculations have shown w17x that a variation in thickness of the illuminated area leads to an increase in the R-value and a less reliable structure determination. The refinements were performed with the recently developed software package MSLS w17x, in which multislice calculation software is combined with least-squares refinement software used in X-ray crystallography. With multislice
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
213
algorithm, which is routinely used for image calculations of HREM images, dynamic diffraction is taken into account explicitly. The MSLS package is described in Ref. w17x. The R value used in the refinements is RŽ I 2 . s SŽ I Žobs. y I Žcalc.. 2rSI Žobs. 2 . For com pleteness, R Ž I . s S Ž < I Ž obs . 1 r 2 y I Žcalc.1r2 <. 2rSI Žobs. is also listed in the table. The standard deviations given between brackets in the tables in this paper are based on the statistics in the refinement and consequently, experimental errors are not included. As input for this procedure, diffraction patterns were obtained by using an incident electron beam with only a small convergence — yielding rather sharp dots — where the integrated intensity of each reflection is used. Apart from the parameters related to the crystal structure, parameters like crystal thickness and crystal orientation can also be refined. Since simultaneously a number of diffraction data sets can be refined, each set has its own scaling parameter and crystal thickness and crystal orientation parameters.
3. Results 3.1. XRD of BaCu 3 O4 Ž00l.-oriented BaCu 3 O4 has been observed in epitaxial Ž001. RBa 2 Cu 3 O 7 films for any R, i.e., Lu, Ho, Y or Gd. The value of c lattice parameter measured, 0.3923Ž2. nm, did not vary with R in RBa 2 Cu 3 O 7 . It is a strong evidence, that there is no rare earth doping in BaCu 3 O4 . Noteworthy, BaCu 3 O4 phase in RBa 2 Cu 3 O 7 films on LaAlO 3 and SrTiO 3 substrates possessed the same value of the lattice parameter. The Ba:Cu cation ratio of single-phase BaCu 3 O4rLaAlO 3 film was proved to be 1:3 with the use of EDX. The lattice parameters were derived from d hkl values for in-plane Ž00l. and out-of-plane
Fig. 1. XRD of BaCu 3 O4 film on Ž001.LaAlO 3 substrate showing the perfect orientation of the phase. Ža. u –2 u scan, Žb. w-scans for two out-of-plane reflections of BaCu 3 O4 and Ž111. peak of the substrate.
Ž111., Ž422., Ž201., Ž802. q Ž042. reflections using the standard least-square procedure. The last two
Table 1 Lattice parameters of BaCu 3 O4 ED for BaCu 3 O4 particles on HoBa 2 Cu 3 O 7 film Žnm.
XRD for BaCu 3 O4 singlephase film on LaAlO 3 Žnm.
XRD BaCu 3 O4 inclusions in YBa 2 Cu 3 O 7 single crystals w11x Žnm.
a s 1.097Ž9. b s 0.554Ž3. c s 0.394Ž2.
a s 1.101Ž2. b s 0.550Ž1. c s 0.3923Ž2.
a s 1.0986Ž3. b s 0.5503Ž1. c s 0.3923Ž1.
214
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
Fig. 2. Low resolution TEM image of two BaCu 3 O4 particles on a film of HoBa 2 Cu 3 O 7 on a SrTiO 3 substrate.
peaks could not be measured separately because of the presence of twinning and because b f ar2. The intensities of other reflections were too low to be measured accurately. The orthorhombic unit cell parameters of BaCu 3 O4 are given in Table 1. Our unit cell parameters are in agreement with those of Bertinotti et al. w11x Ž a s 1.098Ž6. nm, b s 0.550Ž3.
nm, c s 0.392Ž3. nm. . The lattice mismatch in the interface plane is 3% and 0.5% for LaAlO 3 and SrTiO 3 substrates, respectively. u –2 u- and w-scanning showed perfect out-ofplane and in-plane orientations of BaCu 3 O4 ŽFig. 1.. The pole figures taken for Ž111. reflection of BaCu 3 O4 films on LaAlO 3 and SrTiO 3 substrates
Fig. 3. HREM image of the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface along w110x HoBa 2 Cu 3O 7 , whereas BaCu 3 O4 is in w010x orientation. The interface shows no additional phases. The c axes of BaCu 3 O4 and HoBa 2 Cu 3 O 7 are parallel. The HoBa 2 Cu 3 O 7 lattice image is wavy and shows a spacing along the c axis of 1.4 nm, which indicates partial degradation of the lattice by intercalation during initial storage, ion milling and subsequent storage.
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
revealed an admixture of Ž210. orientation. The latter cannot be observed in the symmetrical u –2 u scans since the reflections with h q k / 2 n are of zero intensity. Epitaxial in-plane alignment was found for both orientations:
Ž 001. BaCu 3 O4 I Ž 001. ABO 3 , w 210x BaCu 3 O4 I w 100x ABO 3 and w 210 x BaCu 3 O4 I w 010 x ABO 3 Ž 210. BaCu 3 O4 I Ž 001. ABO 3 , w 001x BaCu 3 O4 I w 100x ABO 3 and w 001 x BaCu 3 O4 I w 010 x ABO 3 . 3.2. TEM obserÕations Fig. 2 shows a low-resolution image of two particles on a film of HoBa 2 Cu 3 O 7 on a LaAlO 3 substrate. The particles were observed to have two possible orientations, in agreement with the X-ray
215
results. Twinning within one particle was rather rare. EDX element analysis indicated the Ba:Cu ratio of the droplets to be close to 1:3, leading to a composition BaCu 3 O4q d . The structure refinements using electron diffraction data reported further in this paper show that d is 0 and therefore, the notation BaCu 3 O4 is used throughout this paper. The unit cell was determined by rotating the diffraction pattern. The lattice parameters of LaAlO 3 were used as standard. The BaCu 3 O4 phase was observed only on the surface of the films. The thickness of the HoBa 2 Cu 3 O 7 film underneath the BaCu 3 O4 particles is the same as that in areas having no BaCu 3 O4 particles on the top. This indicates that the presence of these particles on the surface does not hinder the growth of the HoBa 2 Cu 3 O 7 film. This requires a rapid diffusion along the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface during the film growth. The particles have a strict epitaxial relation with the underlying HoBa 2 Cu 3 O 7 film. The a and b axes
Fig. 4. HREM image of the BaCu 3 O4 rHoBa 2 Cu 3 O 7 interface along w110x of HoBa 2 Cu 3 O 7 with BaCu 3 O4 in w010x orientation, showing the presence of an additional phase. The main lattice spacing of the additional phase along the interface normal is 0.38 nm. The wavy HoBa 2 Cu 3 O 7 lattice image is explained in the caption of Fig. 2.
216
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
of the BaCu 3 O4 particles are parallel to the ²110: directions of the HoBa 2 Cu 3 O 7 film, giving the relationship:
w 001x BaCu 3 O4 I w 001x HoBa 2 Cu 3 O 7y d and w 100 x BaCu 3 O4 I w 110 x HoBa 2 Cu 3 O 7y d . Thus, viewing along w110x of HoBa 2 Cu 3 O 7 , one can observe either w100x-oriented or w010x-oriented BaCu 3 O4 . Fig. 3 shows an HREM image of the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface. As in this example, the interface is mostly sharp without another phase. The epitaxy of the two phases is evident from this figure. Sometimes, another phase occurs at the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface, as is shown in Fig. 4. This extra phase at the interface is thinner than 3 nm and is also epitaxial. The lattice spacing Ž0.38 nm. along the interface normal suggests a perovskite-like structure. In Fig. 3, as well as Fig. 4, the HoBa 2 Cu 3 O 7 lattice shows contrast differences. These contrast variations are mainly due to a degradation of the 123 film upon exposure to air. No role of the additional phase in this degradation was observed. Annealing of specimens containing BaCu 3 O4 in argon flow Ž pŽO 2 . ; 0.1 mbar. at 8008C for 1 h, nor annealing in oxygen, with a gradual decrease from 5508C to 1508C in 4 h, showed no change of the c lattice parameter.1 This is a strong indication that under these experimental conditions, no oxygen is removed from or inserted into the Ba layer. 3.3. Structure determination from electron diffraction data A number of diffraction patterns were recorded of BaCu 3 O4 in w100x and w010x orientations. Fourteen data sets were selected, yielding 826 reflections with Iobs ) 2 s Ž Iobs . and 588 ones with Iobs - 2 s Ž Iobs .. Two typical diffraction patterns are shown in Fig. 5. One is almost perfectly oriented, whereas the other one shows asymmetry due to a misorientation of the crystal. A larger misorientation results in less diffrac-
1 Changes of the a and b unit cell parameters could not be determined, because XRD of the epitaxial thin films of BaCu 3 O4 does not allow this determination with a high enough accuracy.
Fig. 5. Examples of recorded electron diffraction pattern of w010xoriented BaCu 3 O4 . Ža. and Žb. are data sets 9 and 14 of the list given in Table 3. The 000 spots are the strongly overexposed spot. Due to the overexposure and the way the CCD is read out, a streak is created through this 000 spot. Asterisks indicate the calculated centres of the Laue circles. Several observed spots are indicated. In Žb., the 007 reflection is visible, corresponding to a d-spacing of 0.055 nm. Due to the misorientation of the crystal in Žb., the visibility of the reflections continues much further on the topside of the 000 spot.
tion spots on the side where they are already weak and an increase in the high-order reflections on the other side, such that even reflections with g-values
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
217
Table 2 Comparison of the atomic positions of BaCu 3 O4 , obtained refinement from ‘single’ X-ray data of peaks of small inclusions of BaCu 3 O4 in single crystals of YBa 2 Cu 3 O 7 w11x, and electron single crystal diffraction along w100x and w010x zones Atom
Position
Ba CuŽ1. CuŽ2. O
0,0,0 1r2, 1r2, 1r2 x, 0, 1r2 x,y, 1r2
XRD on inclusions in single crystal
Electron diffraction
x
x
0.2479 0.1264
y
B
0.2543
0.60 0.45 0.60 0.91
well above 20 nmy1 Žequivalent to a d-spacing of 0.05 nm. can be significant. The 14 data sets used for the refinement were taken from areas with a range of thickness, 4–16 nm, to increase the accuracy of the determination of the atomic parameters. The refinement was started assuming the spacegroup to be Cmmm. Since only w100x and w010x diffraction patterns are used, there is a strong correlation of the x parameter of CuŽ2. with its B parameter and the y parameter of O with its B parameter. Because of this correlation, the refinement was first done with an overall B-value. Next, the individual B parameters were refined with fixed x ŽCuŽ2.. and y ŽO.. Finally, a single cycle was done with a refinement in which all parameters were refined. Refinements of possible O positions in the
0.248Ž9. 0.127Ž1.
y
B
0.252Ž9.
0.8Ž1. 0.2Ž1. 1.1Ž1. 1.6Ž1.
z s 0 plane Žthe Ba plane. indicated that no significant amount of O is present in this plane. Refinements with space groups of lower symmetry than Cmmm did not result in significantly better fits. The results of the MSLS refinement with an overall RŽ I 2 . value of 3.6% Ž3.8% for all reflections. are listed in Table 2. The data sets used for the refinements are listed in Table 3. For completeness, RŽ I . s SŽ< I Žobs.1r2 y I Žcalc.1r2 <. 2rSI Žobs. is also listed in Table 3. We have not added calculated HREM images based on this model to the HREM image of Figs. 3 and 4 because the projection of the structure in this viewing direction is so simple that one can easily obtain a good match with a wrong model. Kinematic refinements of all data sets resulted in R-values ranging from 3% to 20% Žsee Table 3..
Table 3 Data of the electron diffraction sets used for the structure refinement The overall R value is 3.2% Ž3.6% using only the reflections Iobs ) 2 s Ž Iobs ... The overall RŽ I 2 . is 12.1% for a kinematic refinement. Zone
w100x w100x w100x w100x w100x w100x w100x w010x w010x w010x w010x w010x w010x w010x
Number observed Žrefl.. 25 44 43 48 38 33 34 79 75 72 73 94 87 81
Thickness Žnm.
3.9Ž3. 5.0Ž3. 5.9Ž3. 6.8Ž5. 6.9Ž2. 7.4Ž3. 8.5Ž3. 3.9Ž3. 5.3Ž3. 5.8Ž4. 6.8Ž5. 7.5Ž3. 9.9Ž3. 16.8Ž6.
Crystal misorientation h
0 0 0 0 0 0 0 3.6Ž9. y1.1Ž5. y1.0Ž3. 3.3Ž5. 1.8Ž5. 0.6Ž2. 2.1Ž1.
k
0.2Ž3. y1.1Ž2. y2.9Ž3. y2.6Ž3. y2.9Ž1. y2.5Ž1. y2.4Ž1. 0 0 0 0 0 0 0
R-value Ž%. l
1.7Ž4. 0.8Ž2. 1.0Ž1. 0.4Ž1. 3.0Ž1. 2.0Ž1. 2.1Ž1. 3.5Ž3. y0.5Ž2. y0.1Ž1. y4.2Ž3. 2.6Ž1. 0.8Ž1. 2.8Ž1.
MSLS
Kinematic
RŽ I 2 .
RŽ I .
RŽ I 2 .
2.7 4.2 1.0 1.1 7.7 4.5 5.1 1.7 3.2 1.6 4.6 3.0 6.8 5.5
2.4 1.9 1.9 2.3 7.1 2.5 3.2 1.8 1.6 2.0 3.0 2.6 6.3 6.4
3.3 8.1 3.3 8.1 13.3 15.2 19.4 3.2 8.8 11.6 5.7 20.4 5.9 19.3
218
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
These refinements were carried out using the MSLS program and only refining the scale using exactly the same parameters as used for the dynamic refinement except for occupancies of 1% Žinstead of 100%., yielding an almost kinematic diffraction condition.
4. Discussion The unit cell of BaCu 3 O4 is orthorhombic with space group Cmmm. The structure consists of alternating Cu 3 O4 and Ba layers along the c axis, as is shown in Fig. 6. The Cu ions have a square coordination of oxygens and the Ba ions in a cube oxygen coordination. The Cu 3 O4 layers are densely packed whereas the Ba layers contain several possible positions for oxygen atoms with a maximum content of BaO 3 . If an oxygen atom would occupy such a possible position, the Cu–O distance would be 0.195 nm for a non-buckled Cu 3 O4 planes. BaCu 3 O4 is reported by Bertinotti et al. w11x as a secondary phase inside YBa 2 Cu 3 O 7 single crystals. They determined the structure and composition of the cuprate using single crystal XRD and taking the reflections, which can only be attributed to BaCu 3 O4 . Comparison Žsee Table 2. of the atomic parameters
obtained from the MSLS refinement and those obtained by Bertinotti et al. w11x from ‘single’ X-ray data from BaCu 3 O4 inclusions in a single crystal of YBa 2 Cu 3 O 7 shows a good agreement in the atomic positions. The structure of BaCu 3 O4 compound is related to that of infinite layer and spin ladder compounds w1,2x. Indeed, like for IL and SL compounds, BaCu 3 O4 consists of alternating Cu nq1O 2 n and oxygen-free A layers. BaCu 3 O4 can be considered as the member of A ny 1Cu nq1O 2 n structural family with A s Ba and n s 2. Structure determination from electron diffraction data is by no means a standard procedure yet, in contrast with X-ray and neutron diffraction. This difference is largely due to the dynamic diffraction, which cannot be neglected for electron diffraction. With kinematic diffraction, which is valid for X-ray and neutron diffraction, the intensities of the reflections increase linearly with thickness. Dynamic scattering, which occurs in particular with electron diffraction ŽED., will change the intensities of all reflections with respect to each other as a function of the specimen thickness. Therefore, the kinematic refinement software can only be used for electrons in the regime where the dynamic scattering is negligi-
Fig. 6. Schematic representation of the structure of BaCu 3 O4 and that of SrTiO 3 ; the latter one serves as example of the simple perovskite structure. The z s 0 and z s 1r2 planes are given. Thick black lines indicate the unit cells.
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
ble, which is for specimen thicknesses smaller than 4 nm for BaCu 3 O4 . We have developed a software package w17–19x that takes dynamic diffraction fully into account. Comparison of the R values of the dynamic and kinematic refinements of all data sets shows the R values of the kinematic refinement to be larger, in particular for the thicker areas. This indicates the importance of the use of a dynamic diffraction procedure in the refinement. Since the diffraction patterns were obtained from cross-section specimens, no diffraction patterns within a cone of 658 from the c axis were obtainable because the maximum specimen tilt is 258. Of the unobtainable diffraction patterns, the w001x zone pattern is the most important one, because this has the highest information content. Plan view specimen preparation, yielding a specimen, which is thin along the substrate normal, was not considered because when not all HoBa 2 Cu 3 O 7 films are removed, wrong diffraction intensities will be obtained. Since the absence of the w001x zone patterns in the refinements resulted only in higher standard deviations in the x and y positions, but not in a significant loss in reliability of the structure refinement and since w001x zone patterns are unreliable because of the reason given above, the refinement was only done with w100x and w010x zone patterns. The lattice parameters of BaCu 3 O4 phase allow an epitaxial match with the structure of HoBa 2 Cu 3 O 7 and perovskite substrates. While c lattice parameter of BaCu 3 O4 is close to that of perovskites, a and b lattice parameters are close to the double or single face diagonal of the perovskite cube, respectively. These relations agree well with the twinning of BaCu 3 O4 on Ž001. LaAlO 3 and Ž001.SrTiO 3 substrates. Bertinotti et al. w11x also report that BaCu 3 O4 phase occurs exclusively in a fixed orientation relations with the surrounding YBa 2 Cu 3 O 7 matrix, where the c axis of BaCu 3 O4 is parallel to the a, b or c axis of YBa 2 Cu 3 O 7 . The orientation of BaCu 3 O4 phase seems to play a key role in its stabilisation, since bulk BaCu 3 O4 is unstable under equilibrium thermodynamical conditions. Another substrate orientation, e.g., Ž110.SrTiO 3 , makes nucleation of BaCu 3 O4 impossible, indicating that a coherent low energy interface cannot be formed. Samoylenkov et al. w15x have reported that BaCu 3 O4 can be prepared as single film on
219
Ž100. substrates of LaAlO 3 up to a thickness of about 0.3 mm. Above this thickness, BaCu 3 O4 decomposes in BaCuO 3 and CuO. Since the thickness of the HoBa 2 Cu 3 O 7 film is not affected by the BaCu 3 O4 particles, the presence of these particles on the surface does not hinder the growth of the HoBa 2 Cu 3 O 7 film. This requires a rapid diffusion along the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface during the film growth. Samoylenkov et al. w15x have reported that the BaCu 3 O4 particles on the surface of HoBa 2 Cu 3 O 7 films do not affect the superconducting properties, provided the deviation from the perfect 1:2:3 stoichiometry was not too large. The film with Ho:Ba:Cu cation ratio of 13:33:54 corresponding to 20% of BaCu 3 O4 phase demonstrated a sharp superconducting transition with Tc of 90 K and rather high critical current density at 77 K, 2 = 10 6 Arcm2 .
Acknowledgements The work was partly supported by INTAS Žgrant N93-0936., INTAS-RFBR Žgrant IR 97-1954., the Copernicus program Žgrant ERBIC15CT-960735. and the Dutch Organization for Fundamental Physical Research ŽFOM..
References w1x M. Takano, Y. Takeda, H. Okada, M. Miyamoto, K. Kusaka, Physica C 159 Ž1989. 375. w2x Z. Hiroi, M. Azuma, M. Takano, Y. Bando, J. Solid State Chem. 95 Ž1991. 230. w3x Z. Hiroi, M. Takano, Physica C 235–240 Ž1994. 29. w4x M. Lagues, J.Y. Laval, C.F. Bueran, C. Deville Cavellin, B. Eustache, J.B. Moussy, C. Partiot, X.Z. Xu, Physica C 282–287 Ž1997. 162. w5x G. Balestrino, S. Martellucci, P.G. Medaglia, A. Paoletti, G. Petrocelli, Physica C 302 Ž1998. 78. w6x M. Takano, M. Azuma, Z. Hiroi, Y. Bando, Y. Takeda, Physica C 176 Ž1991. 441. w7x E. Dagotto, T.M. Rice, Science 271 Ž1996. 618. w8x M. Uehara, T. Nagata, J. Akimutsu, H. Takahashi, N. Mori, K. Kinoshita, J. Phys. Soc. Jpn. 65 Ž1996. 2764. w9x C.W. Chu, Y.Y. Xue, Z.L. Du, Y.Y. Sun, L. Gao, N.L. Wu, Y. Cao, I. Rusakova, K. Ross, Science 277 Ž1997. 1081. w10x H. Yamamoto, M. Naito, H. Sato, Jpn. J. Appl. Phys. 36 Ž1997. L341. w11x A. Bertinotti, J. Hamman, D. Luzet, E. Vincent, Physica C 160 Ž1989. 227.
220
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
w12x G.F. Voronin, S.A. Degterov, J. Solid State Chem. 110 Ž1994. 50. w13x E.L. Brosha, F.H. Garzon, I.D. Raistrick, J. Am. Ceram. Soc. 78 Ž1995. 1745. w14x T.B. Lindemer, E.D. Specht, Physica C 255 Ž1995. 81, and references therein. w15x S. Samoylenkov, O. Gorbenko, I. Graboy, A. Kaul, H. Zandbergen, E. Connolly, Chem. Mater. 11 Ž1999. in press.
w16x S.V. Samoylenkov, O.Yu. Gorbenko, I.E. Graboy, A.R. Kaul, Yu.D. Tretyakov, J. Mater. Chem. 6 Ž1996. 623. w17x J. Jansen, D. Tang, H.W. Zandbergen, H. Schenk, Acta Cryst. A 54 Ž1998. 91. w18x H.W. Zandbergen, S.L. Anderson, J. Jansen, Science 227 Ž1997. 1221. w19x H.W. Zandbergen, J. Jansen, J. Microsc. 190 Ž1998. 222.
The structure of BaCu 3 O4 particles occurring on thin HoBa 2 Cu 3 O 7 films prepared by MOCVD H.W. Zandbergen a,) , J. Jansen a , V.L. Svetchnikov a , I.E. Graboy b, S. Samoylenkov b, O. Gorbenko b, A.R. Kaul b a
National Centre for HREM, Laboratory of Materials Science, Delft UniÕersity of Technology, Rotterdamseweg 137, Delft 2628 AL, The Netherlands b Moscow State UniÕersity, Department of Chemistry, Moscow 119899, Russia Received 9 February 1999; received in revised form 2 October 1999; accepted 5 October 1999
Abstract The structure of BaCu 3 O4 phase occurring as particles on the surface of Ž001. RBa 2 Cu 3 O 7 epitaxial films prepared by metalorganic chemical vapor deposition ŽMOCVD. has been investigated with quantitative electron diffraction and HREM. The orthorhombic unit cell is a s 1.097Ž9. nm, b s 0.554Ž3. nm, c s 0.394Ž2. nm with space group Cmmm, the values being in agreement with X-ray diffraction ŽXRD. study. The structure consists of alternating Cu 3 O4 and Ba layers along the c-axis. The compound is stabilised due to the formation of low-energy coherent boundaries with RBa 2 Cu 3 O 7 andror perovskite substrate. q 1999 Elsevier Science B.V. All rights reserved. Keywords: BaCu 3 O4 ; HoBa 2 Cu 3 O 7 ; MOCVD
1. Introduction The intensive study of complex oxides over the last decades brought up many new individual compounds and structural families. One of the most intriguing research followed the discovery of A ny 1Cu nq1O 2 n family Žwhere A s Ba, Sr, Ca., which includes so-called ‘‘infinite layer’’ IL Ž n s infinity. and ‘‘spin ladder’’ SL compounds Ž n G 3. w1,2x. Most of these phases are metastable under ambient conditions and have been, at first, prepared )
Corresponding author. Tel.: q31-15-278-2266; fax: q31-15278-6730. E-mail address: [email protected] ŽH.W. Zandbergen.
by means of high-pressure synthesis w3x. More lately, MBE and PLD w4,5x, among other deposition techniques, were adopted to prepare them also as epitaxial films. The IL structure is very simple and consists of 2D CuO 2 planes separated by anion-free A layers. These cuprates are superconducting with Tc values up to 100 K, provided charge carriers are doped in CuO 2 planes, e.g., due to the presence of vacancies in A sites w6x. The building block of the copper– oxygen layer of SL compounds could be presented as a quasi-one-dimensional fragment of CuO 2 layer, joining together two or more Cu–O chains, or ‘‘legs’’, into a ‘‘ladder’’. The ladders are connected with each other with a shift by one Cu–O distance along the chain direction to form Cu nq 1O 2 n layers. Like in the IL structure, two adjacent Cu nq 1O 2 n
0921-4534r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 4 5 3 4 Ž 9 9 . 0 0 5 3 7 - 7
212
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
layers in ladder compounds are separated by anionfree A layers. The spin gap, giving a name for SL, is either opened or closed, for ladders with an even or odd number of legs, respectively w7x. There was also a theoretical prediction of superconductivity in slightly doped A ny 1Cu nq1O 2 n ladders with an even number of legs w7x. Eventually, Sr14yx Ca x Cu 24 O41 cuprate, containing two-leg Cu 2 O 3 ladder, was shown to be a superconductor under high pressure with Tc of 12 K w8x. Until now, this is the only known example of a superconducting cuprate without CuO 2 layer. The search for new phases in A– Cu–O systems proceeds with the use of high-pressure synthesis and film growth techniques. Recently, superconductivity with Tc above 120 K has been found in Ba 2 Ca ny1Cu nO x family synthesised under high pressure w9x, and a new high Tc superconductor in thin films of Ba–Cu–O system has been observed w10x. BaCu 3 O4 phase was first reported in Ref. w11x as an oriented impurity phase appearing in superconducting YBa 2 Cu 3 O 7y d single crystals. Surprisingly, in spite of the numerous research in the field over the last decade, BaCu 3 O4 was never reported, to the best of our knowledge, as a single-phase material. Like the IL and SL compounds, BaCu 3 O4 is believed to be a metastable phase. The thermodynamical stability under various conditions has been proved so far only for four barium cuprates: Ba 2 CuO 3qx , BaCuO 2q y , Ba 2 Cu 3 O5qz and BaCu 2 O 2qw . A very detailed study on the thermal stabilities of Ba–Cu–O system is reported in Refs. w12–14x. Recently, we observed the presence of oriented BaCu 3 O4 in epitaxial RBa 2 Cu 3 O 7 thin films ŽR — rare earth element. grown by metalorganic chemical vapor deposition ŽMOCVD.. Also, we have prepared single-phase BaCu 3 O4 films on LaAlO 3 and SrTiO 3 substrates. The present paper deals with the structural analysis of BaCu 3 O4 performed by high-resolution electron microscopy and electron diffraction. The study of the stability of BaCu 3 O4 will be reported elsewhere w15x.
2. Experimental Epitaxial Ž001. RBa 2 Cu 3 O 7y d ŽR s Lu,Ho,Y,Gd. and BaCu 3 O4 thin films were grown on Ž001. LaAlO 3
Žhere and below, the indices for LaAlO 3 are given for a pseudo-cubic perovskite unit cell. and Ž001. SrTiO 3 single-crystal substrates by the single-source flash evaporation MOCVD w16x. The deposition runs were carried out at temperature 8008C–8358C and oxygen partial pressure pŽO 2 . in the range of 1.7–2.5 mbar. The growth rate was about 0.4 mmrh. After the deposition, all the films were cooled down under oxygen flow with an intermediate annealing at 4508C for 1 h. The surface morphologies and cation compositions of the films were studied with the use of CAMSCAN scanning electron microscope with EDX element analysis system. X-ray diffraction ŽXRD. study was carried out using four-circle SIEMENS D5000 diffractometer with CuK a irradiation. The lattice parameters of BaCu 3 O4 crystal lattice were derived from d hkl values for in-plane and out-of-plane reflections using the standard least-square procedure. The peaks positions for a given reflection were measured for all possible values of w with a fixed x angle, then averaged and corrected in relation to the nearest peaks from a single-crystal substrate. Electron transparent areas in cross-section were obtained by ion milling with the film facing away from the ion guns. Electron microscopy was performed with a Philips CM30UT electron microscope with a field emission gun operated at 300 kV and a Link EDX element analysis system. Electron diffraction patterns were recorded with a 1024 = 1024 pixel Photometrix CCD camera with a dynamic range of 12 bits. Electron diffraction was performed with spot sizes of about 10 nm, using a 15 mm condensor lens aperture. Exposure times ranged from 0.5 to 2 s. To reduce the electron-microscope-induced contamination, the specimen was cooled to about 100 K. A small spot size for electron diffraction is used to have a relatively small variation in the diffraction area of thickness and misorientation since most crystals are wedge-shaped and show often local variations in misorientation. Calculations have shown w17x that a variation in thickness of the illuminated area leads to an increase in the R-value and a less reliable structure determination. The refinements were performed with the recently developed software package MSLS w17x, in which multislice calculation software is combined with least-squares refinement software used in X-ray crystallography. With multislice
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
213
algorithm, which is routinely used for image calculations of HREM images, dynamic diffraction is taken into account explicitly. The MSLS package is described in Ref. w17x. The R value used in the refinements is RŽ I 2 . s SŽ I Žobs. y I Žcalc.. 2rSI Žobs. 2 . For com pleteness, R Ž I . s S Ž < I Ž obs . 1 r 2 y I Žcalc.1r2 <. 2rSI Žobs. is also listed in the table. The standard deviations given between brackets in the tables in this paper are based on the statistics in the refinement and consequently, experimental errors are not included. As input for this procedure, diffraction patterns were obtained by using an incident electron beam with only a small convergence — yielding rather sharp dots — where the integrated intensity of each reflection is used. Apart from the parameters related to the crystal structure, parameters like crystal thickness and crystal orientation can also be refined. Since simultaneously a number of diffraction data sets can be refined, each set has its own scaling parameter and crystal thickness and crystal orientation parameters.
3. Results 3.1. XRD of BaCu 3 O4 Ž00l.-oriented BaCu 3 O4 has been observed in epitaxial Ž001. RBa 2 Cu 3 O 7 films for any R, i.e., Lu, Ho, Y or Gd. The value of c lattice parameter measured, 0.3923Ž2. nm, did not vary with R in RBa 2 Cu 3 O 7 . It is a strong evidence, that there is no rare earth doping in BaCu 3 O4 . Noteworthy, BaCu 3 O4 phase in RBa 2 Cu 3 O 7 films on LaAlO 3 and SrTiO 3 substrates possessed the same value of the lattice parameter. The Ba:Cu cation ratio of single-phase BaCu 3 O4rLaAlO 3 film was proved to be 1:3 with the use of EDX. The lattice parameters were derived from d hkl values for in-plane Ž00l. and out-of-plane
Fig. 1. XRD of BaCu 3 O4 film on Ž001.LaAlO 3 substrate showing the perfect orientation of the phase. Ža. u –2 u scan, Žb. w-scans for two out-of-plane reflections of BaCu 3 O4 and Ž111. peak of the substrate.
Ž111., Ž422., Ž201., Ž802. q Ž042. reflections using the standard least-square procedure. The last two
Table 1 Lattice parameters of BaCu 3 O4 ED for BaCu 3 O4 particles on HoBa 2 Cu 3 O 7 film Žnm.
XRD for BaCu 3 O4 singlephase film on LaAlO 3 Žnm.
XRD BaCu 3 O4 inclusions in YBa 2 Cu 3 O 7 single crystals w11x Žnm.
a s 1.097Ž9. b s 0.554Ž3. c s 0.394Ž2.
a s 1.101Ž2. b s 0.550Ž1. c s 0.3923Ž2.
a s 1.0986Ž3. b s 0.5503Ž1. c s 0.3923Ž1.
214
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
Fig. 2. Low resolution TEM image of two BaCu 3 O4 particles on a film of HoBa 2 Cu 3 O 7 on a SrTiO 3 substrate.
peaks could not be measured separately because of the presence of twinning and because b f ar2. The intensities of other reflections were too low to be measured accurately. The orthorhombic unit cell parameters of BaCu 3 O4 are given in Table 1. Our unit cell parameters are in agreement with those of Bertinotti et al. w11x Ž a s 1.098Ž6. nm, b s 0.550Ž3.
nm, c s 0.392Ž3. nm. . The lattice mismatch in the interface plane is 3% and 0.5% for LaAlO 3 and SrTiO 3 substrates, respectively. u –2 u- and w-scanning showed perfect out-ofplane and in-plane orientations of BaCu 3 O4 ŽFig. 1.. The pole figures taken for Ž111. reflection of BaCu 3 O4 films on LaAlO 3 and SrTiO 3 substrates
Fig. 3. HREM image of the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface along w110x HoBa 2 Cu 3O 7 , whereas BaCu 3 O4 is in w010x orientation. The interface shows no additional phases. The c axes of BaCu 3 O4 and HoBa 2 Cu 3 O 7 are parallel. The HoBa 2 Cu 3 O 7 lattice image is wavy and shows a spacing along the c axis of 1.4 nm, which indicates partial degradation of the lattice by intercalation during initial storage, ion milling and subsequent storage.
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
revealed an admixture of Ž210. orientation. The latter cannot be observed in the symmetrical u –2 u scans since the reflections with h q k / 2 n are of zero intensity. Epitaxial in-plane alignment was found for both orientations:
Ž 001. BaCu 3 O4 I Ž 001. ABO 3 , w 210x BaCu 3 O4 I w 100x ABO 3 and w 210 x BaCu 3 O4 I w 010 x ABO 3 Ž 210. BaCu 3 O4 I Ž 001. ABO 3 , w 001x BaCu 3 O4 I w 100x ABO 3 and w 001 x BaCu 3 O4 I w 010 x ABO 3 . 3.2. TEM obserÕations Fig. 2 shows a low-resolution image of two particles on a film of HoBa 2 Cu 3 O 7 on a LaAlO 3 substrate. The particles were observed to have two possible orientations, in agreement with the X-ray
215
results. Twinning within one particle was rather rare. EDX element analysis indicated the Ba:Cu ratio of the droplets to be close to 1:3, leading to a composition BaCu 3 O4q d . The structure refinements using electron diffraction data reported further in this paper show that d is 0 and therefore, the notation BaCu 3 O4 is used throughout this paper. The unit cell was determined by rotating the diffraction pattern. The lattice parameters of LaAlO 3 were used as standard. The BaCu 3 O4 phase was observed only on the surface of the films. The thickness of the HoBa 2 Cu 3 O 7 film underneath the BaCu 3 O4 particles is the same as that in areas having no BaCu 3 O4 particles on the top. This indicates that the presence of these particles on the surface does not hinder the growth of the HoBa 2 Cu 3 O 7 film. This requires a rapid diffusion along the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface during the film growth. The particles have a strict epitaxial relation with the underlying HoBa 2 Cu 3 O 7 film. The a and b axes
Fig. 4. HREM image of the BaCu 3 O4 rHoBa 2 Cu 3 O 7 interface along w110x of HoBa 2 Cu 3 O 7 with BaCu 3 O4 in w010x orientation, showing the presence of an additional phase. The main lattice spacing of the additional phase along the interface normal is 0.38 nm. The wavy HoBa 2 Cu 3 O 7 lattice image is explained in the caption of Fig. 2.
216
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
of the BaCu 3 O4 particles are parallel to the ²110: directions of the HoBa 2 Cu 3 O 7 film, giving the relationship:
w 001x BaCu 3 O4 I w 001x HoBa 2 Cu 3 O 7y d and w 100 x BaCu 3 O4 I w 110 x HoBa 2 Cu 3 O 7y d . Thus, viewing along w110x of HoBa 2 Cu 3 O 7 , one can observe either w100x-oriented or w010x-oriented BaCu 3 O4 . Fig. 3 shows an HREM image of the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface. As in this example, the interface is mostly sharp without another phase. The epitaxy of the two phases is evident from this figure. Sometimes, another phase occurs at the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface, as is shown in Fig. 4. This extra phase at the interface is thinner than 3 nm and is also epitaxial. The lattice spacing Ž0.38 nm. along the interface normal suggests a perovskite-like structure. In Fig. 3, as well as Fig. 4, the HoBa 2 Cu 3 O 7 lattice shows contrast differences. These contrast variations are mainly due to a degradation of the 123 film upon exposure to air. No role of the additional phase in this degradation was observed. Annealing of specimens containing BaCu 3 O4 in argon flow Ž pŽO 2 . ; 0.1 mbar. at 8008C for 1 h, nor annealing in oxygen, with a gradual decrease from 5508C to 1508C in 4 h, showed no change of the c lattice parameter.1 This is a strong indication that under these experimental conditions, no oxygen is removed from or inserted into the Ba layer. 3.3. Structure determination from electron diffraction data A number of diffraction patterns were recorded of BaCu 3 O4 in w100x and w010x orientations. Fourteen data sets were selected, yielding 826 reflections with Iobs ) 2 s Ž Iobs . and 588 ones with Iobs - 2 s Ž Iobs .. Two typical diffraction patterns are shown in Fig. 5. One is almost perfectly oriented, whereas the other one shows asymmetry due to a misorientation of the crystal. A larger misorientation results in less diffrac-
1 Changes of the a and b unit cell parameters could not be determined, because XRD of the epitaxial thin films of BaCu 3 O4 does not allow this determination with a high enough accuracy.
Fig. 5. Examples of recorded electron diffraction pattern of w010xoriented BaCu 3 O4 . Ža. and Žb. are data sets 9 and 14 of the list given in Table 3. The 000 spots are the strongly overexposed spot. Due to the overexposure and the way the CCD is read out, a streak is created through this 000 spot. Asterisks indicate the calculated centres of the Laue circles. Several observed spots are indicated. In Žb., the 007 reflection is visible, corresponding to a d-spacing of 0.055 nm. Due to the misorientation of the crystal in Žb., the visibility of the reflections continues much further on the topside of the 000 spot.
tion spots on the side where they are already weak and an increase in the high-order reflections on the other side, such that even reflections with g-values
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
217
Table 2 Comparison of the atomic positions of BaCu 3 O4 , obtained refinement from ‘single’ X-ray data of peaks of small inclusions of BaCu 3 O4 in single crystals of YBa 2 Cu 3 O 7 w11x, and electron single crystal diffraction along w100x and w010x zones Atom
Position
Ba CuŽ1. CuŽ2. O
0,0,0 1r2, 1r2, 1r2 x, 0, 1r2 x,y, 1r2
XRD on inclusions in single crystal
Electron diffraction
x
x
0.2479 0.1264
y
B
0.2543
0.60 0.45 0.60 0.91
well above 20 nmy1 Žequivalent to a d-spacing of 0.05 nm. can be significant. The 14 data sets used for the refinement were taken from areas with a range of thickness, 4–16 nm, to increase the accuracy of the determination of the atomic parameters. The refinement was started assuming the spacegroup to be Cmmm. Since only w100x and w010x diffraction patterns are used, there is a strong correlation of the x parameter of CuŽ2. with its B parameter and the y parameter of O with its B parameter. Because of this correlation, the refinement was first done with an overall B-value. Next, the individual B parameters were refined with fixed x ŽCuŽ2.. and y ŽO.. Finally, a single cycle was done with a refinement in which all parameters were refined. Refinements of possible O positions in the
0.248Ž9. 0.127Ž1.
y
B
0.252Ž9.
0.8Ž1. 0.2Ž1. 1.1Ž1. 1.6Ž1.
z s 0 plane Žthe Ba plane. indicated that no significant amount of O is present in this plane. Refinements with space groups of lower symmetry than Cmmm did not result in significantly better fits. The results of the MSLS refinement with an overall RŽ I 2 . value of 3.6% Ž3.8% for all reflections. are listed in Table 2. The data sets used for the refinements are listed in Table 3. For completeness, RŽ I . s SŽ< I Žobs.1r2 y I Žcalc.1r2 <. 2rSI Žobs. is also listed in Table 3. We have not added calculated HREM images based on this model to the HREM image of Figs. 3 and 4 because the projection of the structure in this viewing direction is so simple that one can easily obtain a good match with a wrong model. Kinematic refinements of all data sets resulted in R-values ranging from 3% to 20% Žsee Table 3..
Table 3 Data of the electron diffraction sets used for the structure refinement The overall R value is 3.2% Ž3.6% using only the reflections Iobs ) 2 s Ž Iobs ... The overall RŽ I 2 . is 12.1% for a kinematic refinement. Zone
w100x w100x w100x w100x w100x w100x w100x w010x w010x w010x w010x w010x w010x w010x
Number observed Žrefl.. 25 44 43 48 38 33 34 79 75 72 73 94 87 81
Thickness Žnm.
3.9Ž3. 5.0Ž3. 5.9Ž3. 6.8Ž5. 6.9Ž2. 7.4Ž3. 8.5Ž3. 3.9Ž3. 5.3Ž3. 5.8Ž4. 6.8Ž5. 7.5Ž3. 9.9Ž3. 16.8Ž6.
Crystal misorientation h
0 0 0 0 0 0 0 3.6Ž9. y1.1Ž5. y1.0Ž3. 3.3Ž5. 1.8Ž5. 0.6Ž2. 2.1Ž1.
k
0.2Ž3. y1.1Ž2. y2.9Ž3. y2.6Ž3. y2.9Ž1. y2.5Ž1. y2.4Ž1. 0 0 0 0 0 0 0
R-value Ž%. l
1.7Ž4. 0.8Ž2. 1.0Ž1. 0.4Ž1. 3.0Ž1. 2.0Ž1. 2.1Ž1. 3.5Ž3. y0.5Ž2. y0.1Ž1. y4.2Ž3. 2.6Ž1. 0.8Ž1. 2.8Ž1.
MSLS
Kinematic
RŽ I 2 .
RŽ I .
RŽ I 2 .
2.7 4.2 1.0 1.1 7.7 4.5 5.1 1.7 3.2 1.6 4.6 3.0 6.8 5.5
2.4 1.9 1.9 2.3 7.1 2.5 3.2 1.8 1.6 2.0 3.0 2.6 6.3 6.4
3.3 8.1 3.3 8.1 13.3 15.2 19.4 3.2 8.8 11.6 5.7 20.4 5.9 19.3
218
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
These refinements were carried out using the MSLS program and only refining the scale using exactly the same parameters as used for the dynamic refinement except for occupancies of 1% Žinstead of 100%., yielding an almost kinematic diffraction condition.
4. Discussion The unit cell of BaCu 3 O4 is orthorhombic with space group Cmmm. The structure consists of alternating Cu 3 O4 and Ba layers along the c axis, as is shown in Fig. 6. The Cu ions have a square coordination of oxygens and the Ba ions in a cube oxygen coordination. The Cu 3 O4 layers are densely packed whereas the Ba layers contain several possible positions for oxygen atoms with a maximum content of BaO 3 . If an oxygen atom would occupy such a possible position, the Cu–O distance would be 0.195 nm for a non-buckled Cu 3 O4 planes. BaCu 3 O4 is reported by Bertinotti et al. w11x as a secondary phase inside YBa 2 Cu 3 O 7 single crystals. They determined the structure and composition of the cuprate using single crystal XRD and taking the reflections, which can only be attributed to BaCu 3 O4 . Comparison Žsee Table 2. of the atomic parameters
obtained from the MSLS refinement and those obtained by Bertinotti et al. w11x from ‘single’ X-ray data from BaCu 3 O4 inclusions in a single crystal of YBa 2 Cu 3 O 7 shows a good agreement in the atomic positions. The structure of BaCu 3 O4 compound is related to that of infinite layer and spin ladder compounds w1,2x. Indeed, like for IL and SL compounds, BaCu 3 O4 consists of alternating Cu nq1O 2 n and oxygen-free A layers. BaCu 3 O4 can be considered as the member of A ny 1Cu nq1O 2 n structural family with A s Ba and n s 2. Structure determination from electron diffraction data is by no means a standard procedure yet, in contrast with X-ray and neutron diffraction. This difference is largely due to the dynamic diffraction, which cannot be neglected for electron diffraction. With kinematic diffraction, which is valid for X-ray and neutron diffraction, the intensities of the reflections increase linearly with thickness. Dynamic scattering, which occurs in particular with electron diffraction ŽED., will change the intensities of all reflections with respect to each other as a function of the specimen thickness. Therefore, the kinematic refinement software can only be used for electrons in the regime where the dynamic scattering is negligi-
Fig. 6. Schematic representation of the structure of BaCu 3 O4 and that of SrTiO 3 ; the latter one serves as example of the simple perovskite structure. The z s 0 and z s 1r2 planes are given. Thick black lines indicate the unit cells.
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
ble, which is for specimen thicknesses smaller than 4 nm for BaCu 3 O4 . We have developed a software package w17–19x that takes dynamic diffraction fully into account. Comparison of the R values of the dynamic and kinematic refinements of all data sets shows the R values of the kinematic refinement to be larger, in particular for the thicker areas. This indicates the importance of the use of a dynamic diffraction procedure in the refinement. Since the diffraction patterns were obtained from cross-section specimens, no diffraction patterns within a cone of 658 from the c axis were obtainable because the maximum specimen tilt is 258. Of the unobtainable diffraction patterns, the w001x zone pattern is the most important one, because this has the highest information content. Plan view specimen preparation, yielding a specimen, which is thin along the substrate normal, was not considered because when not all HoBa 2 Cu 3 O 7 films are removed, wrong diffraction intensities will be obtained. Since the absence of the w001x zone patterns in the refinements resulted only in higher standard deviations in the x and y positions, but not in a significant loss in reliability of the structure refinement and since w001x zone patterns are unreliable because of the reason given above, the refinement was only done with w100x and w010x zone patterns. The lattice parameters of BaCu 3 O4 phase allow an epitaxial match with the structure of HoBa 2 Cu 3 O 7 and perovskite substrates. While c lattice parameter of BaCu 3 O4 is close to that of perovskites, a and b lattice parameters are close to the double or single face diagonal of the perovskite cube, respectively. These relations agree well with the twinning of BaCu 3 O4 on Ž001. LaAlO 3 and Ž001.SrTiO 3 substrates. Bertinotti et al. w11x also report that BaCu 3 O4 phase occurs exclusively in a fixed orientation relations with the surrounding YBa 2 Cu 3 O 7 matrix, where the c axis of BaCu 3 O4 is parallel to the a, b or c axis of YBa 2 Cu 3 O 7 . The orientation of BaCu 3 O4 phase seems to play a key role in its stabilisation, since bulk BaCu 3 O4 is unstable under equilibrium thermodynamical conditions. Another substrate orientation, e.g., Ž110.SrTiO 3 , makes nucleation of BaCu 3 O4 impossible, indicating that a coherent low energy interface cannot be formed. Samoylenkov et al. w15x have reported that BaCu 3 O4 can be prepared as single film on
219
Ž100. substrates of LaAlO 3 up to a thickness of about 0.3 mm. Above this thickness, BaCu 3 O4 decomposes in BaCuO 3 and CuO. Since the thickness of the HoBa 2 Cu 3 O 7 film is not affected by the BaCu 3 O4 particles, the presence of these particles on the surface does not hinder the growth of the HoBa 2 Cu 3 O 7 film. This requires a rapid diffusion along the BaCu 3 O4rHoBa 2 Cu 3 O 7 interface during the film growth. Samoylenkov et al. w15x have reported that the BaCu 3 O4 particles on the surface of HoBa 2 Cu 3 O 7 films do not affect the superconducting properties, provided the deviation from the perfect 1:2:3 stoichiometry was not too large. The film with Ho:Ba:Cu cation ratio of 13:33:54 corresponding to 20% of BaCu 3 O4 phase demonstrated a sharp superconducting transition with Tc of 90 K and rather high critical current density at 77 K, 2 = 10 6 Arcm2 .
Acknowledgements The work was partly supported by INTAS Žgrant N93-0936., INTAS-RFBR Žgrant IR 97-1954., the Copernicus program Žgrant ERBIC15CT-960735. and the Dutch Organization for Fundamental Physical Research ŽFOM..
References w1x M. Takano, Y. Takeda, H. Okada, M. Miyamoto, K. Kusaka, Physica C 159 Ž1989. 375. w2x Z. Hiroi, M. Azuma, M. Takano, Y. Bando, J. Solid State Chem. 95 Ž1991. 230. w3x Z. Hiroi, M. Takano, Physica C 235–240 Ž1994. 29. w4x M. Lagues, J.Y. Laval, C.F. Bueran, C. Deville Cavellin, B. Eustache, J.B. Moussy, C. Partiot, X.Z. Xu, Physica C 282–287 Ž1997. 162. w5x G. Balestrino, S. Martellucci, P.G. Medaglia, A. Paoletti, G. Petrocelli, Physica C 302 Ž1998. 78. w6x M. Takano, M. Azuma, Z. Hiroi, Y. Bando, Y. Takeda, Physica C 176 Ž1991. 441. w7x E. Dagotto, T.M. Rice, Science 271 Ž1996. 618. w8x M. Uehara, T. Nagata, J. Akimutsu, H. Takahashi, N. Mori, K. Kinoshita, J. Phys. Soc. Jpn. 65 Ž1996. 2764. w9x C.W. Chu, Y.Y. Xue, Z.L. Du, Y.Y. Sun, L. Gao, N.L. Wu, Y. Cao, I. Rusakova, K. Ross, Science 277 Ž1997. 1081. w10x H. Yamamoto, M. Naito, H. Sato, Jpn. J. Appl. Phys. 36 Ž1997. L341. w11x A. Bertinotti, J. Hamman, D. Luzet, E. Vincent, Physica C 160 Ž1989. 227.
220
H.W. Zandbergen et al.r Physica C 328 (1999) 211–220
w12x G.F. Voronin, S.A. Degterov, J. Solid State Chem. 110 Ž1994. 50. w13x E.L. Brosha, F.H. Garzon, I.D. Raistrick, J. Am. Ceram. Soc. 78 Ž1995. 1745. w14x T.B. Lindemer, E.D. Specht, Physica C 255 Ž1995. 81, and references therein. w15x S. Samoylenkov, O. Gorbenko, I. Graboy, A. Kaul, H. Zandbergen, E. Connolly, Chem. Mater. 11 Ž1999. in press.
w16x S.V. Samoylenkov, O.Yu. Gorbenko, I.E. Graboy, A.R. Kaul, Yu.D. Tretyakov, J. Mater. Chem. 6 Ž1996. 623. w17x J. Jansen, D. Tang, H.W. Zandbergen, H. Schenk, Acta Cryst. A 54 Ž1998. 91. w18x H.W. Zandbergen, S.L. Anderson, J. Jansen, Science 227 Ž1997. 1221. w19x H.W. Zandbergen, J. Jansen, J. Microsc. 190 Ž1998. 222.