Home Search Collections Journals About Contact us My IOPscience

Crucible corrosion in crystal growth of YBa2Cu3O7-x

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1989 Supercond. Sci. Technol. 2 17

(http://iopscience.iop.org/0953-2048/2/1/004)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 139.184.14.159
This content was downloaded on 03/10/2015 at 17:27

Please note that terms and conditions apply.

Supercond. Sci. Technol. 2 (1989) 17-21. Printed in the UK

H J Scheelt, W Sadowskl and L Schellenberg

Department of Physics DPMC, Universlty of Geneva 24, Quai Ernest-Ansermet,
CH-1211 Geneva 4, Switzerland

Recelved 21 September 1988, in final form 23 February 1989

Abstract. Cruclble corrosion Is one of the major obstacles to obtaining large

crystals of the high-temperature superconductor YBa,Cu,O,-, . The crucible-meit
interfaces of three types of ceramics (alumina, zirconia and tin oxide) have been
studied in order to understand the corrosion mechanism. This may help to reduce
the cruclble problem In crystal growth of YBCO and to indicate alternative crucible
materials of improved corrosion resistance.

1. Introduction After a brief discussion of corrosion effects in

With the advent of superconductivity above the tem-
perature of liquid nitrogen [l] there is a large demand
for single crystals and single-crystalline films of
YBa,Cu,O,-. (YBCO)to study this phenomenon and, in
the case of films, to develop devices for numerous appli-
cations. However, several problems based on the
thermal and chemical instabilities of YBCO [2, 31 have
allowed only modest results with respect to crystal size
and film quality. The first problem consists in finding
mobilised phases (gaseous, liquid) from which crystals
and epitaxial films can be grown. In the case of crystals,
this problem was solved by phase-diagram studies and
by systematic crystal-growth experiments from high-
temperature solutions [4,5].
Another problem is based on the formation of very
thin crystals due to growth instability [2, 31. The low
growth rate required to achieve stable growth and bulk
crystals free from inclusions [2] leads to very low
cooling rates and thus to long crystal growth runs,
which cause severe corrosion of the crucibles and fre-
quently the liquid creeps along the crucible walls and
sometimes even leaves the crucible. The corrosion is
apparent from the roughened crucible wall surface in
the region of liquid contact and by the formation of
undesired crystals containing chemical constituents
from the crucible material.
The purpose of this work is to study the corrosion
mechanism of various crucible materials under typical
mco-growth conditions. This understanding of corro-
sion may lead to crucible materials of improved corro-
sion resistance and may be helpful in modifying the
experimental crystal-growth conditions so as to mini-
mise the contamination from the crucible.
t Present address: Institute of Micro- and Optoelectronics, Physics
Department, Swiss Federal Institute of Technology, CH-1015 Lau-
sanne, Switzerland.
crystal-growth experiments the various crucible-
solidified flux interfaces are discussed. Finally the conse-
quences drawn from the corrosion studies are evaluated
and possible solutions to the crucible corrosion
problem are indicated.
2. Crystelllsatlon of YBCO
Due to the limited thermal and chemical stability of
YBCO there have been only two routes for obtaining
crystals so far: lengthy solid-state reactions and crystal
growth from high-temperature solutions [2, 31. In the
latter approach the so-called cavity method has been
most widely applied: YBCO crystallises from partially
liquid mixtures with BaO-CuO during cooling from
loo0 f 50°C to room temperature. Solidification and
contraction lead to the formation of cavities in which
crystals separated from liquid can be found. The cruci-
bles used for this process, mostly high-density alumina
or gold, are corroded. The contamination of YBCO by
the crucible material varies significantly and depends on
the composition, the temperatureltime programme and
the distance of the YBCO crystals from the crucible wall.
However, the cavity method allowed one to prepare
YBCO crystals of a few mm in size in which no AI con-
tamination could be detected by an electron microprobe
analysis [ 6 ] . Moreover, when using gold crucibles or
gold sheets, small concentrations of gold (about
0.7 at.%) were found [7] which had no appreciable
effect on the superconducting properties.
The cavity method did not allow nucleation control,
and the crystal size was necessarily limited by the size of
the cavities. A chemical separation of the crystals, by
dissolving the flux, has not been possible due to the
chemica1 instabi1ity Of mm c21*Therefore the decant-
ing process was developed, which consists of pouring off

0853-2048/89/010017 + 05 $02.50 (8 1989 IOP Publishing Ltd 17


Flgure 2. (a) Line scan from alumina (left) into flux. The
following phases are successively crossed: AI,O,;
BaAl,O,:Cu; YBa,Cu,O,-, + flux; BaAl,O,:Cu. Separated
concentration changes of the elements are shown (not to
scale). (b) The spectral peak intensities (arbitrary units) for
each element are plotted, with the same scale, as a function
of distance.
The line scan across the crucible-flux interface along
the path indicated in figure 1 is shown in figures 2(a)
and (h). The width of the AI,O,-Ba(AI,_,Cu,)O, inter-
face is of the order of 5 pm. It is somewhat smeared out
as each point actually represents the analysis of a region
of a few pm" The width between the other interfaces in
the flux (BaAl,O,-YRco) is wider due to the surface
roughness as observed in figure 1. The concentration of
0 . 1 4 4 at.'%,(metals)of Ba and Cu in alumina indicates
solid-state diffusion which, according to the slight differ-
ence in darkness visible on the secondary-electron
picture (figure l), reaches a rather uniform depth of
about 3 pm.
Thus one may conclude that corrosion of alumina
crucibles by BaO-CuO melts starts with the diffusion of
Ba and Cu ions into alumina. This process is followed
by the formation of a reaction layer of Ba aluminate
containing Cu, which is irregularly dissolved and
carried away so that aluminate regions, separated from
the crucible wall region, can be found in the solidified
flux. After saturation of the crystal-growth melt the alu-
minate may be deposited in crystalline form at other
solid boundaries and thereby interfere with the forma-
tion of large YBCO crystals.
Crucible corrosion in crystal growth of YBCO
Flgure 3.Zirconia-flux interface (backscattered electrons).
The relatively porous ZrO, ceramic (dark grey) and the
surface-contaminated flux region are separated by the
lighter reaction layers.
5. Zirconia crucibles
The higher melting point of ZrO, compared with
alumina (2700 as opposed to 2040°C) suggested its use
in crystal growth of YRCO despite its higher price and its
poor thermal shock resistance [SI. The latter property
led to cracking of the crucibles when they were removed
from the hot furnace in order to decant the residual flux
at the end of the experiment.
The picture of the zirconia-flux interface from back-
scattered electrons is shown in figure 3. It shows the
dark grey and rather porous region of zirconia with an
interface roughness of about 10 pm. The reaction layer
has a thickness of 10 & 3 pm and consists of a barium
zirconate with the formula Ba(Y, Ca, Cu)Zr(Cu)O,. The
analysis indicates a somewhat higher Zr concentration
than that of BaZrO, but such a Zr-rich compound
could not be found in the phase diagram [16]. The
analysis of the line scans shown in figures 4(a) and (h)
indicates Ca-doped ZrO, on the left-hand side which
gives way to a reaction layer of BaZrO, containing
some Ca, Cu and Y. The Zr content of this layer
decreases to the right-hand side while the Cu concentra-
tion clearly increases. The Cu 'peak' on the right-hand
side of figure 4(b) corresponds to a BaCuOJCuO-rich
region, which can also be identified on figure 3. Solid-
state diffusion of Ba and Cu into the ZrO, ceramics was
indicated by a difference in darkness visible on the SEM
picture (figure 3) but could not be measured quantitat-
ively. The surface of the flux region (not the BaZrO,
region), after polishing, had reacted with the humidity
of air, rendering the analysis unreliable.
6. Tin oxide crucibles
Recently the growth of single-domain YBCO crystals
with a T, of 89 K using SnO, crucibles was reported
19
H J Scheel et a /

Figure 5. Interface of porous tin oxide and flux. The broad

reaction layer and its dissolution are clearly visible.

7. Dlscusslon

The wetting and corrosion of the BaO-rich and CuO-

rich high-temperature solutions for the growth of
YBa,Cu,O,-, is severe, not only in noble metal cruci-
Figure 4. (a) Line scan from zirconia into the flux, passing
through the phases: Zr0,:Ca; Ba(Ca, Y, Cu)Zr(Cu)O,:
decreasing Zr and Increasing Cu concentration ; BaCuO,
+ CuO; BaCu(Zr)O, (separate plots for each element, not
to scale). (b) The spectral peak intensities (arbitrary units)
for each element are plotted with the same scale. as a
function of distance.

[17]. It was also indicated that these crystals were not

contaminated by tin [17]. In order to check whether the
obtaining of contamination-free crystal was due to the
cavity method discussed above or due to the resistance
to corrosion of SnO, , a crystal-growth experiment
using a nearly complete liquid was performed. After the
growth run, the liquid could be only partially removed
because of its high viscosity, and no sizable crystals
were obtained. In comparison with simultaneous
crystal-growth runs using alumina and zirconia crystals,
the tin oxide crucible was severely corroded.
The Sn0,-flux interface is shown in figure 5. The tin
oxide ceramic is recognised by its fine pores of about
1 pm in size. A solid-state diffusion of flux ions into the
ceramic could not be detected, possibly due to the high
dissolution rate of tin oxide. However, the reaction
layer (BaSn, -xCu,O,) was well defined, with a Cu con-
centration increasing up to about x = 0.3. One can also
recognise large sectors of the outer reaction layer which
are penetrated by the CuO-rich flux and then carried
Figure 6. (e) Line scan from SnO, through the
away from the interface. Figure 6(a) shows the thick
BaSn, -xCu,O, reaction layer into the CuO-BaO flux
reaction layer and figure 6(b) gives evidence for the (separate plots for each element, not to scale). (b) The
increasing copper concentration towards the outer spectral peak intensities (arbitrary units) for each element
region of the reaction layer. are plotted with the same scale, as a function of distance.

20

bles but also in ceramic crucibles. The latter form com-
pounds with BaO of the type BaA1,0,, BaZrO, and
BaSnOJ which may contain further constituents from
the crucible or from the flux. These reaction layers are
then partially dissolved and carried away in successive
corrosion. Since these oxide compounds and the base
crucible refractory oxides already have very high
melting points there is not much hope of finding a really
corrosion-resistant oxide material for crucibles. Never-
theless, MgO, BeO, Tho, and other highly refractory
compounds should be tested. Another approach would
be to use yttria or a high-melting compound containing
Y,O, as the crucible material.
If no crucible material of sufficient corrosion resist-
ance can be found, the growth technology could be
modified by using crucible-free or skull-melting tech-
niques. Another possibility would consist of applying
relatively high concentrations of YBCO in the BaO-CuO
melt so as to create a solid YBCO layer along the crucible
walls; by this method the speed of corrosion can be
reduced though not prevented. This problem of crucible
corrosion has somehow to be solved in order to allow
growth of large inclusion-free crystals of
YBa,Cu,O, -r.
Acknowledgments
Polishing of samples by Mr F Liniger and discussions
with Dr T Graf are gratefully acknowledged. The
authors thank Dr J H James for improving the English.
Crucible corrosion in crystal growth of YBCO
This work was supported by the Fonds National Suisse
de la Recherche Scientifique.
References
[l] Wu M K, Ashburn J R, Torng C G, Hor P H, Meng R
L, Gao L, Huang Z J, Wang Y Q and Chu C W 1987
Phys. Rev. Lett. 58 908
[2] Scheel H J 1988 Physica C 153-155 44-9
[3] Scheel H J and Licci F 1988 MRS Bulletin 13 56-60
[4] Scheel H J and Licci F 1988 Mater. Res. Soc. Symp.
Proc. 99 595-9
[5] Licci F, Scheel H J and Besagni T 1988 Physica C
153-155 431-2
[6] Licci F, Scheel H J and Sommerauer J unpublished
[7] Kaiser D L, Holtzberg F, Chisholm M F and
Worthington T K 1987 J. Cryst. Growth 85 593
[SI Sadowski W and Scheel H J 1989 J. Less-Common Met.
150 at press
[9] Scheel H J 1989 J . Less-Common Met. 150 at press
[lo] Elwell D and Scheel H J 1975 Crystal Growthfiom
High-Temperature Solutions (London: Academic)
ch. 7
[ll] Laligant Y, Ferey G, Hervieu M and Raveau B 1987
Europhys. Lett. 4 1023
[12] Scheel H J and Licci F 1987 J . Cryst. Growth 85 607
[13] Calestani G, Riuoli C and Andretti G D 1988 Solid
State Commun. 66 223
[14] Kaiser D L, Holtzberg F, Scott B A and McGuire T R
1987 AppI. Phys. Lett. 51 1040
[l5] Schmid H 1988 Private communication
[16] Jonker G. H and Kwestroo W 1958 J. Am. Ceram. Soc.
41 392
[171 Thomson C, Cardona M, Gegenheimer B, Liu R and
Simon A 1988 Phys. Rev. B 37 9860
21