Acta Materialia 51 (2003) 975–981

www.actamat-journals.com

Temperature dependence of faceting in ⌺5(310)[001] grain

boundary of SrTiO3
Sung Bo Lee ∗, Wilfried Sigle, Wolfgang Kurtz, Manfred Rühle
Max-Planck-Institut für Metallforschung, Heisenbergstraße 3, D-70569 Stuttgart, Germany

Received 20 September 2002; received in revised form 18 October 2002; accepted 24 October 2002

Abstract

Using high-resolution electron microscopy (HREM), temperature dependence of faceting of an asymmetric ⌺5 grain
boundary (GB) in SrTiO3 is observed. The bicrystal samples have been fabricated by ultra-high vacuum diffusion
bonding and heat-treated between 1100 and 1600 °C. Below 1300 °C, this GB facets into symmetric (310) and asymmet-
ric (100)//(430) GB planes. At 1300 °C, in addition to the asymmetric facet, the symmetric {210} facet appears: three
different facets are thus observed at this temperature. At 1400 and 1500 °C, the asymmetric facet disappears and the
two kinds of symmetric facets remain. At 1600 °C, faceting disappears and the GB becomes defaceted. The
faceting/defaceting transition behavior of the GB is interpreted in terms of the Wulff plot and its corresponding equilib-
rium crystal shape.
 2002 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved.

Keywords: Diffusion bonding; High-resolution electron microscopy (HREM); Grain boundary faceting; Strontium titanate bicrystal

1. Introduction were found to facet, but when reannealed in hydro-

Faceted general and special grain boundaries
(GBs) with hill-and-valley shapes have been
observed in a number of metals [1–4] and oxides
[5–7]. In polycrystals such impurities and additives
as Cu in Bi [8,9], Te in Fe [10,11], and O in Ni
[12,13] have been observed to induce GB faceting.
For Al2O3, SiO2 and CaO were observed to induce
faceting, but MgO made GBs defaceted [14]. For
BaTiO3, when annealed in air, most of the GBs
∗
Corresponding author. Tel.: +49-711-6893650; fax: +49-
711-6893522.
E-mail address: bolee@hrem.mpi-stuttgart.mpg.de (S.B.
Lee).
gen atmosphere the faceted GBs became defaceted
[15]. In a Cu bicrystal, Bi induced GB faceting,
but when Bi was removed, the GB became
smoothly curved [16]. Reintroduction of Bi caused
the GB to facet reversibly.
With temperature increase, faceted GBs undergo
defaceting transition. In pure polycrystalline Ni,
Lee et al. [13] and Henry et al. [12] observed that
the number fraction of faceted GBs decreased with
temperature increase. Lee and coworkers [13] sug-
gested that abnormal or normal grain growth was
related to the faceting/defaceting phase transition,
because faceted GBs have a different migration
mechanism from defaceted GBs. Similar corre-
lations between GB structure and grain growth

1359-6454/03/$30.00  2002 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved.
doi:10.1016/S1359-6454(02)00500-1
976 S.B. Lee et al. / Acta Materialia 51 (2003) 975–981
have been observed in a Ni-base superalloy [17],
Ag [18], Cu [19], and 316L stainless steel [20,21].
However, there have been only a few observations
of faceting/defaceting transition with temperature
in a single GB of a bicrystal.
The purpose of this work is to observe effects
of temperature on faceting of a GB, using a SrTiO3
bicrystal as a model system. In 1989 the first obser-
vation of a faceting/defaceting transition with tem-
perature of a GB in a bicrystal was reported by
Hsieh and Balluffi in a ⌺3 asymmetric 具111典 tilt
grain boundary in Al and Au, and a ⌺11 asymmet-
ric boundaries in Al [22]. (The quantity ⌺ is
defined as the reciprocal density of common sites
of the adjoining grains associated with CSL
points.) They observed that the faceting/defaceting
transitions were reversible, but did not determine
the crystallographic indices of the facets at each
temperature. Furthermore, Hsieh and Balluffi [22]
pointed out that none of the observed faceted GBs
might be at thermodynamic equilibrium, because
the annealing time at each temperature was only
30–40 min which may not have been not enough
for the GBs to reach equilibrium.
In this study, the change of facet orientations of
a single GB with temperature is crystallograph-
ically determined using high-resolution electron
microscopy (HREM) of an asymmetric ⌺5 GB in
a SrTiO3 bicrystal. Defaceting transition of the GB
is observed. The specimens were heat-treated
between 1100 and 1600 °C for up to 7 days in
order for the grain boundaries to reach their equi-
librium shape. This is the first report of thermally
induced faceting/defaceting transition of GBs in
ceramics. We interpret the observed faceting/
defaceting phase transition in terms of the equilib-
rium shape of a grain embedded in another grain
(the GB Wulff shape).
2. Experimental
The ⌺5 (310)[001] GB in a SrTiO3 bicrystal was
fabricated by joining the (310) surfaces of two sin-
gle crystals of 15 × 10 × 2 mm3 by ultra high-vac-
uum (UHV) diffusion-bonding. The purity of the
single crystal was measured by inductively coupled
plasma-optical emission spectroscopy (ICP-OES)
to be 99.99 wt%. Small amounts of Ca, Ba, Cu,
Ni and Fe were detected. The [13̄0] directions of
the two crystals were aligned antiparallel to each
other and hence to the GB. The miscuts were 0.14°
from the [310] direction towards the [13̄0] direc-
tion for one surface and 0.12° towards [13̄0] for
the other surface, which were measured by a laser-
Laue apparatus. As a result, the nominal inclination
of the GB in the bicrystal was about 0.1° with
respect to the exact symmetric (310) plane. This
leads to a slight asymmetric component of the
resulting ⌺5 GB. Details of the diffusion bonding
and bicrystal geometry were described by Kurtz
[23] and Fischmeister et al. [24].
Specimens of 2 × 5 × 4 mm3 were cut out of the
bicrystal, and heat-treated in air at 1100 °C for 7
days, at 1200 and 1300 °C for 5 days, at 1400 and
1500 °C for 3 days and at 1600 °C for 1 day. Prior
to these heat-treatments, all specimens were
annealed at 1500 °C for 8 h in air. The cooling
and heating rates were 10 °C/min. The heat-treated
bicrystals were ion-milled at 5 kV for observation
by HREM. The HREM work was carried out at
1.25 MeV using the Stuttgart JEOL JEM-
ARM1250 with 0.12 nm point-to-point resolution
[25].
3. Results
The HRTEM observation of the SrTiO3 ⌺5
(310) bicrystal GB with an asymmetric component
(designated as “initial GB”, henceforth) were per-
formed in the [001] projection. Fig. 1 shows that
the initial GB is faceted into the symmetric (310)
and the asymmetric (100)//(430) facets in the
bicrystal annealed at 1100 and 1200 °C. Annealing
at 1300 °C produces an additional (120) / / (21̄0)
facet (for brevity, the symmetric {210} facet,
henceforth) (Fig. 2). The symmetric {210} and the
asymmetric (100)//(430) facets alternate, separated
from each other by the symmetric (310) facet. The
symmetric (310) and the asymmetric (100)//(430)
facets remain stable up to a temperature between
1200 and 1300 °C. At 1400 and 1500 °C, the
asymmetric facet disappears and the two sym-
metric facets are observed (Fig. 3). For the GB
heat-treated at 1600 °C, no faceting is observed,
 S.B. Lee et al. / Acta Materialia 51 (2003) 975–981
Fig. 1. HRTEM image of the GB heat-treated at (a) 1100 °C
for 7 days and (b) 1200 °C for 5 days. In both cases the GB
does not remain planar, but breaks up into the symmetric (310)
and (100)//(430) facets.
Fig. 2. HRTEM image of the GB heat-treated at 1300 °C for
5 days. In addition to the two pre-existing facets, the symmetric
{210} facet appears.
977
Fig. 3. HRTEM image of the GB heat-treated at (a) 1400 °C
and (b) 1500 °C for 3 days. The GB exhibits symmetric (310)
and {210} facets.
and all grain boundaries are defaceted as shown
in Fig. 4(a), indicating that the defaceting occurs
between 1500 and 1600 °C. No regular array of a
structural unit is found at 1600 °C as shown in
Fig. 4(b).
4. Discussion
Grain boundary faceting/defaceting transition
was observed in which the faceted structures
between 1100 and 1500 °C were replaced by the
defaceted one at 1600 °C. It is likely that the GB
normal varies slightly along its length, because the
initial single crystal surfaces will not be perfectly
flat. An observed GB at each temperature may not
be reproducible and the change of the GB struc-
tures with temperature may show inconsistency. To
exclude such a possibility, in this study, we selec-
978 S.B. Lee et al. / Acta Materialia 51 (2003) 975–981
Fig. 4. (a) HRTEM image of the GB heat-treated at 1600 °C
for 1 day showing a completely defaceted structure and (b) an
enlarged view exhibiting a rough structure.
ted two specimens randomly from the sectioned
samples for heat-treatment at each temperature.
The two specimens heat-treated at each tempera-
ture revealed the same types of facets, and we con-
clude the GB facet type appearing at each tempera-
ture as typical.
As shown in Figs. 1–3, the average plane orien-
tation of the observed faceted grain boundaries
appears to deviate from the symmetric (310) plane
by larger angles than the stated inclination angle
of 0.1°. We believe that after diffusion-bonding the
GB is not perfectly planar across the whole bicrys-
tal. Since we are observing very small specimen
areas at high magnification, we may find locally a
larger inclination than nominally expected.
The faceted and defaceted equilibrium interface
structure should be considered in terms of the plot
of interfacial free energy against GB plane normal
orientation (g-plot) and a corresponding equilib-
rium crystal shape [26–28]. A cusp appearing in
the g-plot corresponds to a singular plane in the
equilibrium shape [26,27]. If a surface normal
orientation does not appear in the equilibrium
shape, a faceted structure composed of planes of
allowed orientation will replace that orientation
[26–30]. At high temperatures singular planes cor-
responding to cusps in the g-plot undergo thermal
roughening [31–35], and above thermal roughen-
ing transition temperature (TR) the equilibrium
crystal shape will have no slope discontinuity. In
this case, all orientations appear in the equilibrium
shape and no faceting occurs. The temperature
evolution of the equilibrium crystal shapes has
been observed in Pb crystals [36,37], an NaCl crys-
tal [38] and 4He crystals [39,40]. Surface
faceting/defaceting transition has been well ana-
lyzed in view of the equilibrium crystal shape and
its temperate evolution [41–47]. Faceting/
defaceting transition is a first-order phase tran-
sition, while thermal roughening is an infinite-
order transition with an extremely weak singularity
in the surface free energy [32–34].
In contrast to the case of the equilibrium shapes
of crystalline solids, observations of equilibrium
shape of island grains will be difficult because they
will shrink during heat-treatment. However,
Westmacott and Dahmen [48] showed the quasi-
equilibrium shape of an island grain in Al. The
island grain embedded into another grain had a
polyhedral shape at low temperatures and with
temperature increase it assumed a nearly spherical
shape. The change with temperature was revers-
ible. The equilibrium structure of a GB can be
determined if it is moving sufficiently slowly. The
observed GB structures, thus, will correspond to
those for the grains with equilibrium shapes.
Therefore, the faceting/defaceting transition of the
initial GB can be understood with the help of
description of the g-plot and a corresponding equi-
librium shape of a grain embedded in another grain
(the GB Wulff shape).
As for the crystalline surfaces [26,27], flat
planes of faceted GBs are represented as singular
planes corresponding to cusps in the g-plot. Junc-
tions between facets represent ranges of missing
 S.B. Lee et al. / Acta Materialia 51 (2003) 975–981
orientations. The symmetric (310) and the asym-
metric (100)//(430) facets observed at 1100 and
1200 °C (Fig. 1) will thus correspond to cusps in
the g-plot, and appear as singular planes in the GB
Wulff shape, meeting each other at an angle of
18.4°, as shown in Fig. 5(a). Fig. 5(a) shows that
the average inclination of ~0.1° of the initial GB
with respect to the symmetric (310) plane belongs
to the band of forbidden orientations between the
Fig. 5. Schematic diagrams for the equilibrium grain shape at
different temperatures: (a) and (b) 1100–1300 °C, (c) 1300–
1500 °C, and (d) 1600 °C.
979
GB normal orientations of the two planes (ranging
from 0 to 18.4°), indicating that the initial GB
facets into the two neighboring stable planes at
equilibrium. (In Fig. 5, the average inclination
(~0.1°) of the initial GB is exaggerated to make it
distinct.) Asymmetric GBs can correspond to cusps
in the g-plot, existing as singular planes in the GB
Wulff shape, as calculated by Merkle and Wolf [3].
At 1300 °C, three kinds of facets appear (Fig. 2).
This structure is likely to be in non-equilibrium
state, which will be considered in detail below.
The faceting into the symmetric (310) and {210}
planes at 1400 and 1500 °C (Fig. 3) is represented
as the GB Wulff shape in Fig. 5(c). The two sym-
metric planes meet each other at an angle of 45°
in the shape. The average inclination of the initial
GB with respect to the symmetric (310) plane lies
between the normal orientations of the two sym-
metric planes, as shown in Fig. 5(c), and at equilib-
rium, the initial GB will facet into the two sym-
metric planes. It appears from Figs. 1–3 that the
asymmetric facet disappears from the faceted GB
structure above 1300 °C. It is represented by the
shrinkage of the asymmetric facet with temperature
increase and its disappearance above 1300 °C in
the GB Wulff shape, as shown in Fig. 5(a)–(c). The
disappearance of the asymmetric facet from the
shape prior to those of the symmetric ones indi-
cates that the cusp depth of the asymmetric facet
can be shallow compared with those of the sym-
metric facets. However, we do not have any pre-
vious calculation results showing the relative
depths of these cusps and their changes with tem-
perature in ⌺5 GB of SrTiO3. At 1600 °C, no facet-
ing occurs (Fig. 4), which is represented in Fig.
5(d). As temperature increases from 1500 to 1600
°C, these two symmetric facets will undergo ther-
mal roughening and eventually the shape will not
have any slope discontinuity (e.g. junction), as
exhibited in Fig. 5(d). In this case, the normal
orientation of the initial GB will appear in the
shape and the GB will be defaceted (Fig. 4).
At 1300 °C, in addition to the asymmetric
(100)//(430) facet, the symmetric {210} facet
appears: Three kinds of facets are found at this
temperature. Viewed along the [001] zone axis, all
these three facets are found to be edge-on and an
edge at which two nonequivalent facets meet is
980 S.B. Lee et al. / Acta Materialia 51 (2003) 975–981
observed as a facet junction point, as shown in Fig.
2. This means that two kinds of edges between the
three nonequivalent facets (one between the sym-
metric facets and the other between the symmetric
(310) facet and the asymmetric one) are parallel to
each other, going along the [001] zone axis direc-
tion. Cahn [28] pointed out through the consider-
ation of the GB phase rule that when faceting
occurs in an initially flat GB, a maximum of three
kinds of nonequivalent facet phases are possible,
which coexist at a facet corner where three edges
between these facets meet [28]. Fig. 2 reveals that
different kinds of facet edges are parallel to each
other, which is in contradiction to the GB phase
rule. Therefore, it can be concluded that the struc-
ture observed at 1300 °C is likely to be in nonequi-
librium state. It is not clear why the GB heat-
treated at 1300 °C for 5 days shows three kinds of
facets. It can, however, be assumed that while the
cusp corresponding to the asymmetric facet is
becoming shallower with increasing temperature,
at an intermediate temperature such as 1300 °C,
the asymmetric facet can appear as a metastable
phase. In addition, the symmetric {210} facet
begins to be observed at 1300 °C and the asymmet-
ric facet disappears at higher temperatures such as
1400 and 1500 °C. Therefore we expect that after
prolonged annealing for more than 5 days at 1300
°C the GB will reach thermal equilibrium, faceting
into the two symmetric facets.
In this study we have concentrated on tempera-
ture dependence of the faceted structure of the
initial GB, because the bicrystal studied is nomin-
ally pure. However, many results have shown that
faceting is induced or removed by addition of
impurities [8–16]. Therefore, the presence of
impurities will affect faceting/defaceting phase
transition. Investigations on impurity effects are
planned for a SrTiO3 ⌺5 GB with well-controlled
impurities and will be compared with the experi-
mental results of this study.
5. Concluding remarks
The faceted GBs observed in the ⌺5 bicrystal of
SrTiO3 have shown the change of facet orientations
with temperature, undergoing defaceting transition
between 1500 and 1600 °C. The geometry of facet
orientations at each temperature has been charac-
terized crystallographically by HREM. The
faceting/defaceting transition of the GB has been
interpreted in view of the GB Wulff shape. Our
thermodynamic description of faceting/defaceting
transition using the GB Wulff shape can be further
extended to general GBs without being limited to
CSL GBs, because general GBs can also undergo
faceting/defaceting transition with temperature
change, which has been indirectly observed in
polycrystals [12,13,17–21]. This study motivates
further investigation by computer simulation
(e.g. molecular dynamics method) for GB
faceting/defaceting phase transition which has not
been done so far.
As noted above in Section 1, abnormal or nor-
mal grain growth in polycrystals is related to GB
faceting/defaceting phase transitions [13]. As in the
case of grain growth (i.e. GB migration), GB
faceting/defaceting transition will be critical for
other GB properties such as sliding, corrosion,
diffusivity, segregation, and precipitation. How-
ever, investigations of such effects have hardly
ever been carried out and are yet to be exploited.
Acknowledgements
The help of Dr Gunther Richter and Dipl.-Ing.
(FH) Rainer Höschen at the ARM microscope and
the TEM sample preparation by Mrs Ute Salz-
berger are appreciated. The ICP-OES analysis at
the Analysis Department of the Pulvermetallurg-
isches Laboratorium (PML) in the MPI für Met-
allforschung is acknowledged. The Alexander von
Humboldt Foundation supported one of the authors
(SBL). We thank Prof. Duk Yong Yoon for critical
reading and valuable comments.
References
[1] Wagner WR, Tan TY, Balluffi RW. Phil. Mag.
1974;29:895.
[2] Goodhew PJ, Tan TY, Balluffi RW. Acta metall.
1978;26:557.
[3] Merkle KL, Wolf D. Phil. Mag. A 1992;65:513.
[4] Hall EL, Walter JL, Briant CL. Phil. Mag. A 1982;45:753.
 S.B. Lee et al. / Acta Materialia 51 (2003) 975–981
[5] Vaudin MD, Rühle M, Sass SL. Acta metall.
1983;31:1109.
[6] Carter CB, Kolstedt DJ, Sass SL. J. Am. Ceram. Soc.
1980;63:623.
[7] Barber J, Tighe NJ. Phil. Mag. 1966;14:531.
[8] Donald AM, Brown LM. Acta metall. 1979;27:59.
[9] Menyhard M, Blum B, McMahon Jr CJ. Acta metall.
1989;37:549.
[10] Rellick JR, McMahon Jr CJ, Markus HL, Palmberg PW.
Metall. Trans. 1971;2:1492.
[11] Pichard C, Rieu J, Goux C. Mém. scient. Revue Métall.
1973;70:13.
[12] Henry G, Plateau J, Waché X, Gerber M, Behar I, Crus-
sard C. Mém. scient. Revue Métall. 1959;56:417.
[13] Lee SB, Hwang NM, Yoon DY, Henry MF. Metall. Mater.
Trans. A 2000;31:985.
[14] Park CW, Yoon DY. J. Am. Ceram. Soc. 2000;83:2605.
[15] Lee BK, Chung SY, Kang SJL. Acta mater. 2000;48:1575.
[16] Ference TG, Balluffi RW. Scripta metall. 1988;22:1929.
[17] Lee SB, Yoon DY, Henry MF. Acta mater. 2000;48:3071.
[18] Koo JB, Yoon DY. Metall. Mater. Trans. A
2001;32A:469.
[19] Koo JB, Yoon DY. Metall. Mater. Trans. A
2001;32A:1911.
[20] Lee SH, Choi JS, Yoon DY. Z. Metallkd. 2001;92:655.
[21] Choi JS, Yoon DY. ISIJ Int. 2001;41:478.
[22] Hsieh TE, Balluffi RW. Acta metall. 1989;37:2133.
[23] Kurtz W. Z. Metallkd. 2002;93:432.
[24] Fischmeister HF, Elssner G, Gibbesch B, Kadow KH,
Kawa F, Korn D, Mader W, Turwitt M. Rev. Sci.
Instrum. 1993;64:234.
[25] Phillipp F, Höschen R, Osaki M, Möbus G, Rühle M.
Ultramicroscopy 1994;56:1.
[26] Herring C. Phys. Rev. 1951;82:87.
[27] Frank FC. In: Robertson WD, Gjostein NA, editors. Metal
surfaces: structure, energetics and kinetics. Metals Park
(OH): ASM; 1963. p. 1.
981
[28] Cahn JW. J. Physique 1982;43:199.
[29] Rottman C, Wortis M. Phys. Rep. 1984;103:59.
[30] Wortis M. In: Vanselow R, Howe R, editors. Chemistry
and physics of solid surfaces, vol. 7. Berlin: Springer-Ver-
lag; 1988. p. 367.
[31] Burton WK, Cabrera N, Frank FC. Phil. Trans. Roy. Soc.
London 1951;A243:299.
[32] Kosterlitz JM, Thouless DJ. J. Phys. C 1973;6:1181.
[33] Chui ST, Weeks JD. Phys Rev B 1976;14:4978.
[34] van Beijeren H. Phys. Rev. Lett. 1977;38:993.
[35] van Beijeren H, Nolden I. In: Schommers W, von Blanck-
enhagen P, editors. Structures and dynamics of surfaces,
vol. 2: phenomena, models, and methods. Berlin:
Springer-Verlag; 1987. p. 259.
[36] Heyraud JC, Métois JJ. Surf. Sci. 1983;128:334.
[37] Métois JJ, Spiller GDT, Venables JA. Phil. Mag. A
1982;46:1015.
[38] Heyraud JC, Métois JJ. J. Cryst. Growth 1987;84:503.
[39] Balibar S, Castaing B. Surf. Sci. Rep. 1985;5:87.
[40] Avron JE, Balfour LS, Kuper CG, Landau J, Lipson SG,
Schulman LS. Phys. Rev. Lett. 1980;45:814.
[41] Robinson IK, Vlieg E, Hornis H, Conrad EH. Phys. Rev.
Lett. 1991;67:1890.
[42] Ocko BM, Mochrie SGJ. Phys. Rev. B 1988;38:7378.
[43] Phaneuf RJ, Williams ED. Phys. Rev. Lett. 1987;58:2563.
[44] Phaneuf RJ, Williams ED, Bartelt NC. Phys. Rev. B
1988;38:1984.
[45] Hwang RQ, Williams ED, Park RL. Phys. Rev. B
1989;40:11716.
[46] Noh DY, Blum KI, Ramstad MJ, Birgeneau RJ. Phys.
Rev. B 1993;48:1612.
[47] Yoon M, Mochrie SGJ, Zehner DM, Watson GM, Gibbs
D. Phys. Rev. B 1994;49:16702.
[48] Westmacott KH, Dahmen U. In: Ranganathan S, Pande
CS, Rath BB, Smith DA, editors. Interfaces: structure and
properties. Switzerland: Trans Tech Publication; 1993.
p. 133.