Growth mechanisms of YBa2Cu3O7−δ single crystals grown by the flux method using alumina crucibles

Growth mechanisms of YBa2Cu3O7−δ single crystals grown by the flux method using alumina crucibles

j. . . . . . . . C R Y S T A L G R O W T H ELSEVIER Journal of Crystal Growth 165 (1996) 42-49 Growth mechanisms of YBa2Cu307_ ~ single crystals g...

2MB Sizes 0 Downloads 11 Views

j. . . . . . . .



Journal of Crystal Growth 165 (1996) 42-49

Growth mechanisms of YBa2Cu307_ ~ single crystals grown by the flux method using alumina crucibles B. Z h o u , G . J . R u s s e l l * Adranced Electronic Materials Group, School of Physics, The Unit,ersiO, of New South Wales, Sydney, NSW 2052, Australia Received 22 August 1995; accepted 6 November 1995


The growth behaviour of large, thin, plate-like and almost cubic YBa2Cu307 ~ single crystals grown from high temperature solution (BaO-CuO flux) using alumina crucibles was studied by optical microscopy. Growth spirals of various shapes were observed on both {001} and {100} surfaces while surface dendrites and hopper morphology were also observed on {100}/{010} surfaces. Both the dislocation growth and two-dimensional growth mechanisms have been used to interpret the observed features, with the dislocation growth mechanism being found responsible for the formation of the {001} surfaces, and the two-dimensional growth mechanism being responsible for the formation of the lateral {100}/{010} surfaces. The reason for this difference in the growth mechanisms for the two types of surfaces is considered to arise from the anisotropic crystal structure.

1. Introduction

Since the discovery of high temperature superconductivity in YBa2Cu30 7 ~ polycrystalline material [1], single crystals of this oxide have been successfully grown using the flux method [2,3]. The single crystals grown by this method display very distinct growth behaviour and have shiny, nearly flux-free surfaces, which make it possible to study the growth mechanism for this particular material. Studies of the morphology and the growth mechanism are both important for fundamental studies of the growth mechanism of such single crystals from high temperature solution and also for the successful growth of

* Corresponding author. Fax: +61 2 385 6060.

different shapes and sizes of these single crystals, which are required for characterisation studies of this interesting new HTSC cuprate. Sun et al. [4-6] have studied the (001) face growth features of flux-grown YBa2Cu307_ ~ single crystals and discussed the formation of the observed octagonal spirals in terms of the periodic bond chain (PBC) theory [7]. They also pointed out that the difference of the growth mechanisms between the basal {001} and the lateral {100}/{010} faces arises from the anisotropic bonds between the {001} and {100}/{010} faces. In this report, we present further detailed observations of the growth features on the {001} and lateral {100}/{010} faces. We also propose that the anisotropic crystal structure of YBa2Cu30 7 ~ single crystals is responsible for the difference of growth mechanisms observed for the basal and lateral faces.

0022-0248/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved Pll S 0 0 2 2 - 0 2 4 8 ( 9 6 ) 0 0 1 6 3 - 7

B. Zhou, G.J. Russell/Journal of Crystal Growth 165 (1996) 42-49

2. Experimental procedure All the crystals studied in this work were grown from a B a O - C u O flux using alumina crucibles [3]. The eutectic composition of BaO-CuO chosen had a molar ratio of 28:72. A stoichiometric mixture of Y203, BaCO 3, CuO to produce 30 g of YBa2Cu307 ~ was thoroughly mixed, packed into an alumina crucible, placed in a furnace, heated to 930°C at 250°C/h, soak for 24 h at this temperature, then cooled at 120°C/h to room temperature. The sintered material was then placed in a ring mill and ground for 10 rain. This fine YBa2Cu307_ ~ powder was then used for the crystal growth experiments. The starting material for the crystal growth process was the sintered YBa2Cu307 6 powder mixed with the flux material of BaCO 3 and CuO in the ratio YBazCu307_~ : BaCO 3 :CuO of 6.8:50.0:51.83 by weight. The mixture was transferred to a glass bottle and using a rotating mixer, mixed for 6 h. The resulting well-mixed powder was then packed tightly into a dense, flat-bottomed, alumina crucible. The typical procedure for the growth of crystals consisted in heating the furnace from room temperature to 1020°C at a rate of 200°C/h, soaked at 1020°C for 24 h, then cooled slowly at a rate of l ° C / h to 930°C. The decant method was used to separate the crystals from the molten flux at this temperature. This particular process will be referred to as the regular method to distinguish it from another, slightly different, crystal growth method which is described below. The second method is called the powder method with the step of preparing the sintered YBa2Cu307 powder being omitted. The flux ratio was kept the same as before. The sintered YBazCu307 ~ powder was replaced by the same quantity of YOI. 5, BaCO 3, CuO powder mixture. A soaking stage was added to the crystal growth cycle at 800°C to convert the carbonate. The rest of the procedure was kept the same. These two methods for the growth of YBa2Cu307_ ~ crystals lead to a difference in the shape of the resultant crystals. The regular method resulted in thin, plate-like single crystals of an average size 4 × 4 × 0.2 mm 3, while the powder method resulted in almost cubic single crystals of average size 3 × 3 × 2 mm 3. The aluminium content in the many crystals grown was found to vary from 0.18 to 0.24 at% [3]. These AI impurities influence the crys-


tal growth process and it is the aim of this study to obtain the observed growth mechanism for the surfaces of crystals grown with this impurity. Except

Fig. 1. (a) Typical polygonal growth spiral observed on the (001) surface of a YBa2Cu3OT_ 6 single crystal (magn. 300×); (b) Octagonal growth spiral observed on the (001) surface of a YBa2Cu307 ~ single crystal (magn. 300×); (c) Circular growth spiral observed on the (001) surface of a YBa2Cu307 ~ single crystal (magn. 300 × ).


B. Zhou, G.J. Russell/Journal qf Crystal Growth 165 (1996) 42-49

where specified, all the crystal surfaces studied with an optical microscope were in an as-grown state. The crystal surfaces were highly reflecting and of good quality, which allowed the growth features involving large growth steps to be readily visible. All measurements were preformed at room temperature.

tallographic directions are decided by the periodic bond chain (PBC) directions. When o~ is high during the growth process of a crystal, growth steps along the stronger bond directions contain a lower kink density, and thus grow slower than for the weaker bond directions with higher kink density. As a result, the growth spiral will be defined by the strong PBC

3. Growth features on the {001} surfaces of Y B a 2 C u 3 0 7_ ~ single crystals

For all of our flux grown YBa2Cu307 ~ single crystals, the screw dislocation growth mechanism was dominant. Different shapes of growth spirals were observed on the majority of the crystals. In most cases, the growth step starting from a group of screw dislocations spreads out on a crystal face into a regular shape, as shown in Fig. la, which shows a typical polygonal growth spiral. In this figure, hilltype spirals can be seen with square morphology bounded by (100) and (010) planes. Other growth steps similar to polygonal and circular shapes were also observed, see Figs. lb and lc. These results are consistent with those previously reported in Refs. [4-6]. The morphology of a growth spiral is principally decided by the lateral roughness of the spiral step, that is, the kink density [8]. Usually the kink density can be described by Jackson's ~x factor [9]. If the step is rough (low o~), which means that the kink density is high, the step can advance independently of the crystallographic directions and a circular spiral will result. When the spiral step is smooth (high a), the growth anisotropy of the different crystallographic directions becomes pronounced and the growth spiral will be polygonal. Usually these crys-

Fig. 2. (a) Photomicrograph showing a square growth spiral on the (001) surface of a YBa2Cu307 ~ single crystal. The spiral was generated by two dislocation groups with opposite signs. The dislocation source, " S " in the figure, consists of several different signed dislocations (magn. 300 × ); (b) Photomicrograph showing a square growth spiral on the (001) surface of a YBa2CuaO 7 single crystal. The spiral was generated by two dislocation groups of opposite sign. The growth steps close upon themselves during the growth, except for the new born step (magn. 300×); (c) Detailed structure between two neighbouring growth steps. Points A and B in the figure show obstacles which stop the atomic steps moving further (magn. 300 × ).


B. Zhou, G.J. Russell~Journal qf Crystal Growth 165 (1996)42-49

directions in the surface. According to Sun et al. [5] for YBa2Cu307 ~ single crystals, the (001) plane contains two primitive PBCs. The stronger one is the Cu 2-O3-Cu 2-03 chain that lies parallel to the (100) direction, the other one is the B a - O z - B a - O 2 chain, which is parallel to the (110) direction. So when ct is high, the kink density is low. The growth spiral will be controlled by strong PBCs, that is, (100) directions on the (001) plane, resulting in a square growth spiral similar to that shown in Fig. la. When oL is low, the next strongest PBCs will take effect. The growth spiral will then be bound by the (100), (010) and (110), <110), <110), <110) directions on the (001) plane. An octagonal spiral is thus predicated, see Fig. lb. For an even lower c~, weaker PBCs will influence the shape of the growth spiral and a circular growth spiral is expected, as shown in Fig. lc. According to Jackson [9] ~, = ~ L/kT",


where ~ is a factor depending on the crystallographic anisotropy, L is the latent heat of melting or dissolution, k is Boltzmann's constant and T the absolute temperature. Note that at low temperatures, where c~ is high, the growth spiral will be more polygonal. Thus, most of the square growth spirals should be formed at the later stage of crystal growth. For all the crystals examined, the majority of the growth spirals were either square or polygonal. Usually, in the case of crystals with a clean shiny surface, only a single growth centre dominates an entire face of the crystal. This indicates very slow growth with supersaturation sufficient to propagate only the most active growth centre and is further evidence that the growth spiral occurred at a later growth stage during the cooling cycle, which is consistent with Jackson's equation. Under the present crystal growth conditions, the growth spirals were not generated by a single screw dislocation, but by a group of screw dislocations, see Fig. 2a. The growth steps emitted by two sources of opposite sign screw dislocations annihilate each other where they meet. In this figure, the distance between the two opposite-sign screw dislocation groups was very small. The result of this type of dislocation interaction was perfectly closed growth steps. When the distance of the two group dislocations reach


some critical value, the growth steps cannot close, see Fig. 2b. The critical distance for the two screw dislocation groups is defined by Y0 (where Y0 is the distance between the steps of the spiral). Following the crystal growth theory of Burton, Cabrera and Frank (BCF) [10], when the distance between the two dislocation groups d < yo/2d, see Fig. 2a, the two opposite-sign dislocation groups work together and generate closed growth steps continuously. When d = yo/2, the growth steps cannot close at the instant at which they just formed, see Fig. 2b. When they have grown large, they can still close upon themselves. Finally, for the case d > yo/2, the two screw dislocation groups will operate independently, see Fig. lc. In fact, the growth steps shown in the optical micrographs were macro-steps, consisting of many atomic steps. Fig. 2c shows the detailed structure between two neighbouring growth steps. The formation of the macro-steps can be explained by the concept of step bunching. The motion of atomic steps in a bunching process is not visible, but the motion of the macro-steps resulting from the microsteps is visible after the steps have reached a certain size. At the points A and B in Fig. 2c, an obstacle occurs that stops the atomic steps moving further. The later atomic steps generated by the screw dislocation group will catch up to the steps stopped by the obstacle, accumulate, and become a visible step bunching. At the last stage of our crystal growth experiments, the flux was decanted onto the porous alu-

Fig. 3. Dendritic growth flux observed on the (001) surface of a Y B a 2 C u 3 0 ~ _ ~ single crystal (magn. 300 X ).


B. Zhou, G.J. Russell/Journal of Cr3'stal Growth 165 (1996) 42-49

Fig. 4. Two-dimensional growth of the (001) surface observed on flux-grown YBa2Cu~O 7 ~ single crystals (magn. 300×).

mina firebricks inside the furnace. Sometimes, liquid flux would stay or splash onto the crystal surface. The liquid would continue to provide material for further growth and formed dendrite structures due to the fast cooling rate. Fig. 3 displays some of these dendritic growths. After studying a very large number of

YBa2Cu307 ~ single crystals over a long time period, it was found that the screw dislocation growth mechanism was not the only mechanism involved in the growth of YBa2Cu307_ a crystals. On a number of occasions, the two-dimensional growth morphology shown in Fig. 4 was observed. There is some flux in the centre region and at the edge of the crystal. When observed, the surfaces on the two sides of the crystal, showed no sign of a growth spiral. This type of growth morphology can be explained in terms of two-dimensional nucleation followed by layer spreading, which was established by Kossel [11] and Stranski [12]. This concept will be discussed in the following section.

4. Growth features on the lateral crystal surfaces

For the thin plate-like crystals grown by the regular method, the features on the { 100} and {010} faces

Fig. 5. Micrographs showing the four lateral {100} faces of a YBa2Cu307_ ~ grown by the flux method. A growth spiral is observed in the upper right comer of (b) and (c). Flux is found at the centre for all four sides (magn. 300 × ).

B. Zhou, G.J. Russell/Journal of Co,stal Growth 165 (1996) 42-49

are too small to be studied by optical microscopy. Crystals grown by the powder method were found to be either rectangular or cubic and this made it possible to investigate the growth morphology on the (100} and {010} faces. Fig. 5 shows the four lateral faces of a YBa2CU3OT_ ~ single crystal with typical growth behaviour. Only two growth spirals were found and they seemed to be located in a small area and not spread over the whole face. These spirals provide evidence for the existence of screw dislocation outcropping on the {100} and {010} faces. The other two faces were smooth with no growth spirals detected. At the centre of all the faces, there was either flux or a crater. It can be concluded that the growth mechanism for the lateral {100} and {010} faces is via two-dimensional nucleation growth, which started at the edges or corners of the crystal, driving the impurities and low melting point materials to the centre of the face. The impurities and low melting point materials would accumulate at the centre which was the lowest surface area of the face. At the end of the growth process, a crater or flux accumulation would form. Scheel and Niedermann [13] studied this type of YBa2Cu307_ ~ single crystal surface by a scanning tunnelling microscope combined with a scanning electron microscope (SEM) and found that for the {100} faces of the crystal, the edges along the (010) directions were a little higher than the centre region. It was suggested that steps initially formed at the crystal edges were propagated towards the centre of the crystal over the {100} planes. Steps from opposite directions meet in the lowest surface region and annihilate each other. Thus the edges play the role of two-dimensional nucleation centres. Restricted by the limitation of our microscopy, we cannot detect the possible height differences that exist between the edges and the centre of the {100} planes of our crystals. But the above morphological observation for the lateral faces of our YBazCu307_ ~ crystals is consistent with the results of Scheel and Niedermann [13]. Usually, two-dimensional growth will occur at high temperature and supersaturation [14], which is contrary to the screw dislocation growth mechanism that needs a lower supersaturation. After studying the present crystal growth conditions in detail [3], it is evident that the growth of the lateral faces is via the two-dimensional nucleation mechanism. Crystals like the


one shown in Fig. 5 are usually in the shape of rectangular or cubic blocks and are found at the bottom of the crucible, while the thin plate-like crystals were normally suspended in the flux. The later crystals would come out with the flux during the decanting process and collect on the porous firebrick. Now, the temperature is higher at the bottom of the crucible, which is not a favourable condition for the two-dimensional nucleation growth mechanism, but considering the fact that the y3+ and Ba 2+ ions are heavier than Cu 2+, there may exist an ion concentration gradient due to the absence of stirring. Supersaturation at the bottom of the crucible may be higher, at least within a few small regions than at other parts of the crucible and this would make two-dimensional nucleation growth possible. Further evidence that supports the two-dimensional nucleation growth mechanism is a hopper-like

Fig. 6. Hopper morphology observed on (010) or (100) faces of YBa2Cu307 ~ single crystals (magn. 300×).


B. Zhou, G.J. Russell/Journal of C~stal Growth 165 (1996) 42-49

morphology seen on a number of crystal surfaces. Fig. 6 presents two hopper-like morphologies observed on the {100} or {010} faces of YBa 2Cu 307 - 8 single crystals. Amelinckx [15] suggested a hopperlike morphology formation mechanism based on the Frank screw dislocation theory [16]. Such a growth mechanism is not applicable to our crystals as no spiral terraces were observed. A more reasonable explanation may be found in the tip effect of Fredriksson [17], which causes a higher driving force for crystal growth at the comers. It has been shown that the two-dimensional nucleation growth barrier at the apex is ~ 1 / 4 smaller than it is for the flat face [18]. According to this mechanism, the advancement velocity of the step will be faster along the edges than towards the centre of the crystal face. When the crystal has reached a sufficient size, this discrepancy between the growth speeds for the edge and centre steps will result in a hopper-like morphology. From the morphological observations on the {100} and {010} faces of YBa2Cu307_ ~ single crystals, the following conclusions can be reached. The growth mechanism for the lateral faces is primarily that of two-dimensional nucleation growth. Screw dislocation growth does exist, but it is not found to be involved in the whole crystal face. Observed hopper-like morphology is further direct proof of the two-dimensional growth mechanism for the lateral crystal faces.

5. Origin of the different growth mechanisms on the {001} and {100}/{010} faces From the above observations and discussions, it is clear that the dominant growth mechanism for the {001} faces is screw dislocation growth with its characteristic growth morphology of different shapes of growth spirals. Two-dimensional nucleation growth was found to occur in only rare cases. The growth of the lateral {100} and {010} faces is, on the contrary, primarily by the two-dimensional nucleation growth mechanism. The difference in the growth mechanisms, according to Sun and Schmid [4], may be related to the degree of anisotropy for the PBC bonds within the surface. The (001) face is a flat face (F face), containing four non-parallel PBCs. The relative bond strengths within the (001) face are

considered to be isotropic. For YBa2Cu306, the {100}/{010} faces have to be considered as a stepped form (S face), no connections existing along the [001] direction. However, for YBa2Cu307, PBCs [100]/[010] bonds are anisotropic being determined by the ratio of the strength of the Cu2-O 3 bonds and Y - O 3 bonds, the latter being much weaker. Therefore, the {100} and {010} faces of YBazCU3OT_ ~ have a much higher degree of bond anisotropy than that of the {001} face. The more isotropic faces grow by the screw dislocation mechanism. The higher anisotropic faces grow according to the two-dimensional nucleation growth mechanism. From our experiments, it is believed that the important factors that decide which growth mechanism is dominant is influenced by the crystal structure. The formation of screw dislocations, which outcrop on the (001) plane, is vital for the BCF theory of crystal growth. If the screw dislocation is more stable for the (001) face than the dislocations on the {100} and {010} faces, it is obvious that the dislocation growth mechanism will be active on the (001} faces. The stable structure of YBa2Cu307 ~ is tetragohal at temperatures above 900°C. From the point of view of the dislocation concept, the tetragonal structure of YBa2Cu307_ ~ has the following characteristics: (1) The characteristic dislocations of the YBa2Cu307_ ~ unit cell are (100), (010), (001) with the corresponding Burgers vectors b equalling a, b, c, respectively. (2) The glide plane of the YBa2Cu307_ ~ crystal is the (001) plane. In the growth process, if there is a screw dislocation outcropping on any of the (001), (010), (100) planes, it will be a growth centre. The crystal will grow on that plane in the form of a dislocation growth. The problem is that when a screw dislocation has the Burgers vector a or b, it will lay in the (001) plane and outcrop on the (100) or (010) plane. Because the (001) plane is a glide plane, then according to dislocation theory, a screw dislocation has higher energy and is not stable. It will move out of the crystal. Apparently such screw dislocations will not act as the growth centre in the screw dislocation growth mechanism. While a screw dislocation with Burgers vector c is perpendicular to the glide plane

B. Zhou, G.J. Russell~Journal of Co'stal Growth 165 (1996) 42-49

and outcrops on the (001) plane, such a screw dislocation will be pinned to the (001) plane and act as the dislocation source for the growth spiral. This is the reason for the growth spirals being normally found on the (001) face. Under some circumstances, the crystal plane may incorporate impurities or inclusions and these will hinder and slow down the movement of the dislocation or even pin the dislocation in the glide plane. If this dislocation happens to outcrop on the {100} and {010} planes, it will become a fixed centre for a growth spiral, see Fig. 5.

6. Conclusion From the above discussion, it can be concluded that for YBa2CU3OT_a single crystals grown from a B a O - C u O flux using alumina crucibles, there are two types of growth mechanisms involved, namely BCF screw dislocation growth and two-dimensional nucleation growth. The morphology of the {001} faces and the lateral {100}, {010} faces is basically different. The screw dislocation growth mechanism dominates growth on the {001} faces with the characteristic feature of various shapes of growth spirals. In the case of the {100} and {010} faces, two-dimensional nucleation growth prevails, resulting in a generally flat surface and in some cases, hopper-like morphology. The reason which causes this difference is related to the particular crystal structure of YBa2Cu3OT_ a. If the screw dislocation outcrops on the {100} and {010} planes, but is laying in the (001) plane, the glide plane, then the screw dislocation in this case is in a higher energy state and will glide out


of the crystal to release itself; whereas, screw dislocations that end on {001} planes are stable and act as the growth centre for a growth spiral.

References [1] W.K. Wu, J.R. Ashburn, C.J. Tomg, P.H. Hor, R.L. Meng, L. Gao, ZJ. Huang, Y.Q. Wang and C.W. Chu, Phys. Rev. Lett. 58 (1987) 908. [2] B.A. Hunter, S.L. Town, D.N. Matthews, G.J. Russell and K.N.R. Taylor, Aust. J. Phys. 44 (1991) 421. [3] B. Zhou, Growth and Properties of Large YBa2Cu3OT_ a Single Crystals, PhD Thesis, University of New South Wales, 1995. [4] B.N. Sun and H. Schmid, J. Crystal Growth 100 (1990) 297. [5] B.N. Sun, R. Boutellier and H. Schmid, Physica C 157 (1989) 189. [6] B.N. Sun, K.N.R. Taylor, B. Hunter, D.N. Matthews, S. Ashby and K. Sealey, J. Crystal Growth 108 (1991) 473. [7] P. Hartman and W.G. Perdok, Acta Cryst. 8 (1955) 49. [8] I. Sunagawa and P. Bennema, in: Preparation and Properties of Solid State Materials, Vol. 7, Ed. W.R. Wilcox (1982) pp. 1-129. [9] K.A. Jackson, in: Liquid Metals and Solidification (American Society for Metals, Cleveland, OH, 1958) p. 147. [10] W.K. Burton, N. Cabrera and F.C. Frank, Philos. Trans. R. Soc. London A 243 (1951) 299. [11] W. Kossel, in: Falkenhagen, Quantentheoric und Chemie, Leipzig (1928) p. 46. [12] I.N. Stranski, Z. Physik. Chem. 136 (1928) 259. [13] H.J. Scheel and P. Niedermann, J. Crystal Growth 94 (1989) 281. [14] H.J. Scheel and D. Elwell, J. Crystal Growth 20 (1973) 259. [15] S. Amelinckx, Philos. Mag. 44 (1953) 337. [16] F.C. Frank, Disc. Faraday Soc. 5 (1949) 48. [17] H. Fredriksson, Jernkont. Ann. (1971) 571. [18] A.A. Chemov, in: Crystal Growth and Characterization, Eds. R. Ueda and J.B. Mullin (North-Holland, Amsterdam, 1975) p. 33.