1996 atomic force microscopy study of (001) sr tio3 surfaces

s u r f a c e science

ELSEVIER Surface Science 369 (1996) 313-320

Atomic force microscopy study of (001) SrTiO3 surfaces
B. St/iuble-POmpin, B. Ilge, V.C. Matijasevic 1 P.M.L.O. Scholte, A.J. Steinfort,
F. Tuinstra
Department of Applied Physics, Solid State Physics, TU Delft, Lorentzweg 1, 2628 CJ, Delft, The Netherlands
Received 17 March 1996; accepted for publication 27 June 1996

Abstract

By means of atomic force microscopy, we have investigated the surfaces of SrTiO3(001) substrates (used for the growth of
high-T~ superconducting films) after different heat treatments. These treatments were (i) annealing in Oa, (ii) annealing in 02, (iii)
annealing in Oa followed by deposition of SrTiO3, and (iv) annealing in UHV. Our results suggest that, independently of the type of
annealing, the surfaces of the polished SrTiO3 substrates rearrange, either by diffusion or desorption of atoms, until reaching the
energetically most favourable surface structure. For vicinal (001) substrates, this equilibrium structure corresponds to atomically
fiat terraces separated by smooth steps of one unit-cell in height, with the orientation and width of the terraces being deter-
mined by the miscut of the substrate. However, the kinetics involved in the reconstruction of the SrTiOa surface were observed
to be strongly dependent on the annealing conditions. In particular, the use of O 3 instead of 02 or UHV was found to accelerate
the reconstruction of the SrTiOa surface. Finally, this study provides an accurate characterisation of the defects present on SrTiO3
substrates.

Keywords: Atomic force microscopy; Strontium titanate; Surface defects; Surface structure, morphology, roughness, and topography;
Vicinal single crystal surfaces

1. Introduction T h e surface of single-crystal SrTiO3 (which is
used extensively as a s u b s t r a t e for the epitaxial
W i t h recent years, the u n d e r s t a n d i n g of the g r o w t h of YBa2Cu3Or a n d r e l a t e d high-T~
m e c h a n i s m s i n v o l v e d in the n u c l e a t i o n a n d g r o w t h s u p e r c o n d u c t o r s ) has recently been the subject of
of thin films of c u p r a t e s u p e r c o n d u c t o r s has p r o - b o t h t h e o r e t i c a l a n d e x p e r i m e n t a l studies I - 1 - 7 ] .
gressed rapidly. I n p a r t i c u l a r , it is n o w a c c e p t e d H o w e v e r , due to the v a r i e t y of different surface
t h a t the surface s t r u c t u r e of a s u b s t r a t e c a n t r e a t m e n t s a p p l i e d to the i n v e s t i g a t e d SrTiOa crys-
s t r o n g l y influence the m o r p h o l o g y of the subse- tals, the p u b l i s h e d results are difficult to c o m p a r e
q u e n t l y d e p o s i t e d high-T~ film. Therefore, a n accu- a n d often seem conflicting.
r a t e c h a r a c t e r i s a t i o n a n d u n d e r s t a n d i n g of the T h e a i m of the a t o m i c force m i c r o s c o p y ( A F M )
surface o f s u b s t r a t e s is essential to g r o w high- s t u d y p r e s e n t e d here was to investigate a n d c o m -
q u a l i t y t h i n a n d u l t r a t h i n films.
p a r e the surface m o r p h o l o g y of SrTiO3(001) after
1 Present address: Conductus, 969 West Maude Avenue, different t r e a t m e n t s . These t r e a t m e n t s were: (i)
Sunnyvale, CA 94086, USA. a n n e a l i n g in 03, (ii) a n n e a l i n g in 02, (iii) a n n e a l i n g

0039-6028/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved
PI1 S0039-6028 (96) 00897-7

314 B. Stdiuble-Piimpinet al./Surface Science 369 (1996) 313-320

in 03 followed by deposition of SrTiO3, and (iv) operating in the tapping mode with a silicon-type
annealing in UHV. cantilever, under ambient conditions.

3. Results
2. Experimental

SrTiO3 is a widely used substrate to grow epitax- 3.1. Annealing in 03
ial RBa2Cu30 r ( R = Y or lanthanide) films. The
substrates investigated in our study are commer- The first type of heat treatment used was annea-
dally available (Crystal GmbH, Berlin) polished ling of the substrate for 15-25 min in 03 at a
single-crystal substrates with a nominal (001) ori- temperature of 700, 750 or 900°C. The first two
entation. In order to simplify the comparison, most temperatures are typical for the growth of cuprate
of the A F M results shown here (Figs. 1-5) were films in our reactive MBE system. Within the
obtained from the same lcm x lcm x l m m sub- investigated range, the treated SrTiO3 surface was
strate which was cut into four parts. Comparable not found to depend significantly on the annealing
measurements were performed on other SrTiO3 time or temperature. Typically, the substrates were
substrates and gave consistent results. warmed to the desired annealing temperature
Annealing in 03 (ozone) was performed in an within 5-10 min and cooled down within approxi-
U H V - M B E system adapted for reactive coevapor- mately 30 min.
ation, with a background pressure of 5 x 10 .6 As shown in Fig. 1; the resulting surface mor-
mbar. The ozone was generated with a commercial phology of the SrTiO3 is characterised by a series
generator and then purified in a home-made still. of parallel steps indicating the existence of a slight
The ozone concentration was between 50 and miscut of the substrate with respect to its nominal
100%. The incidence rate of the ozone beam was (001) orientation. The height of the steps is
1016 molecules cm -2 S-1. 0.39 __0.01 nm (i.e. one lattice constant of SrTiO3)
Annealing in 0 2 was performed in flowing and the terrace width is equal to 85_+ 5 nm. From
oxygen, in a furnace at ambient pressure. independent X-ray diffraction measurements, the
The deposition of SrTiO3 was done by coevapor- vicinal angle e of the same substrate was found to
ating Sr and Ti at 750°C in an ozone beam in the be equal to 0.27___0.05°, which corresponds to a
reactive MBE system. By adequately timing the terrace width of 85__15nm, in good agreement
deposition process, a total of 10 monolayers of with the above A F M results. The step ledges are
SrTiO 3 were grown on top of the previously perpendicular to the projection of the surface
annealed substrates. This was confirmed by the normal into the (001) plane. Their orientation with
analysis (after film growth) of the R H E E D oscilla- respect to the crystallographic lattice is character-
tions of the specular spot. ised by the miscut angle fi (see definition in
The miscut angles of the substrates were deter- Section 2). As this angle fi can take on any value
mined by optically aligning the sample with a laser (for the sample shown in Fig. 1, fi=17__5°), we
and then measuring the position of two different would expect that on a microscopic scale, due to
crystal reflections using a standard four-circle the orientation dependence of the step energy, the
diffractometer. The direction of the surface normal step ledges should zigzag along well-defined crys-
with respect to the crystal lattice was characterised tallographic directions. However, within the reso-
by the angles e (the vicinal angle between the lution of the measurements performed here, the
surface normal and the (001) direction) and fl (the observed step ledges are in general smooth, i.e.
angle between the ( 1 0 0 ) direction and the pro- only a limited amount of kink bunching is
jection of the surface normal into the (001) plane). observedl
All images were obtained on a Nanoscope III In addition, the surface of SrTiO3 was character-
A F M (Digital Instruments, Santa Barbara, CA) ised by different types of defects. Dislocations(see

B. Stgiuble-Pfimpinet al./Surface Science369 (1996) 313-320 315

(a) 3.0 r~ were observed with a density of approximately
2 + 1 x 109 cm -2.

,00 i
1.5

O.01~
r'm 3.2. Annealing in Oe

Figs. 3 and 4 display an SrTiO3 surface after
annealing for 1 h at 750°C in 02. In order to
minimise thermal strain resulting from a possible
temperature gradient over the sample, the substrate
was warmed up to the desired temperature within
8 h and cooled down again to room temperature
within approximately 10 h.
The surface morphology shown in Fig. 3a is
1.(iO
characterised by irregularly shaped terraces.
Although the step ledges appear to be wavy due
(b) to step bunching, they are locally (on a scale of
10-100 nm) parallel to crystallographic directions,
such as for example, (100), (110), ( 1 2 0 ) and
(130). The height of the steps typically varies
between 0.5, 1, 1.5, 2 and 2.5 unit cells of SrTiO3
(see, e.g., Fig. 3b), indicating the presence of
different termination layers. To exclude calibration
~o o.zo o.~0 o.~o 03,0
IIN errors, the z-calibration of the A F M was checked
A A' immediately before and after the recordings of the
Fig. 1. (a) AFM image of an SrTiOs(001)substrate after annea- A F M images shown in Figs. 3 and 4.
ling in Oa at 750°C for 15 rain. The black arrow points to a As in the case of 03 annealing, line-shaped
dislocation. The vertical profile taken along the line A-A' is defects and holes were also observed after heat
shown in (b). The profilewas averaged over a width of 200 nm. treatment in 02 (see Fig. 4). Typically the holes
The observed step height is equal to one unit cell of SrTiO3. have the form of inverted pyramids with the sides
( 1 0 - 2 0 n m long) aligned along the (110) direc-
tions. Larger holes with the sides aligned along the
arrow in Fig. 1) were observed with a density of (100) directions were also observed. Comparable
6 + 1 x 107 cm -2. Such dislocations could, by inher- square holes have been observed by Jiang et al.
itance, lead to the formation of spiral'shaped [6] after annealing in hydrogen as well as by
islands in the subsequently deposited high-T~ Kawasaki et al. [3] after etching the surface of
superconductor film. Furthermore, line-shaped their samples. While some of t h e holes on our
substrates were distributed randomly on the
defects running along the (100) and ( 0 1 0 )
SrTiO3 surface (approximately 2 x 10s cm-2), most
directions of the substrate were found. The length
of them were found to be grouped into chains
of these one-dimensional arrays of defects was
parallel to the (100) and (010) directions (see
observed to vary between a few hundred nm and
Fig. 4).
a few #m. Typically, a height difference of one
Finally, annealing a substrate first in O2 and
lattice constant of SrTiO3 is observed across such then in 03, or vice-versa, resulted in a surface
a defect. In some cases, two linear defects cross comparable to that described in Section 3.1.
each other, resulting in the cross-shaped defect
displayed in Fig. 2. Similar cross-shaped defects 3.3. Annealing in 03followed by the deposition o f
have been reported by Sum et al. [5]. As shown SrTi03
in Fig. 2, step ledges are occasionally found to
have a very wavy, almost dendritic structure. We attempted to examine how the SrTiO3
Finally, holes with a diameter of 1 0 - 2 0 n m deposition affected the surface morphology and in

316 B. Stgiuble-Pfimpin et al./Surface Science 369 (1996) 313-320

~onN

.z,oo

0 nM

0 nM

~! .00

1.00 2.00

Fig. 2. A F M image of the same sample as displayed in Fig. 1, showing two line-shaped defects crossing each other. The linear defects
are parallel to the (100) and (010) directions.

particular whether it would improve its smooth- its surface can generally be recognised. However,
ness. Therefore, one of the treatments investigated unlike the atomically fiat substrates annealed in
was annealing of the substrate in 03 as described 03, this UHV annealed substrate displays rough
in Section 3.1 followed by the deposition of 10 step ledges and small clusters are clearly seen on
layers of SrTiO3 in the reactive MBE system. The top of its terraces. This is consistent with results
resulting surface is shown in Fig. 5. by Jiang et al. 1-7] showing that higher temper-
As for all SrTiO3 substrates with 03 annealing, atures and longer annealing times are required to
no step bunching was observed. Although line- obtain an atomically fiat SrTiO3 surface by
shaped defects were still present, they were annealing in UHV.
observed to heal as additional material was depos-
ited. No holes could be resolved. Furthermore, the
surface roughness was found to have increased due 4. Discussion
to the formation of islands nucleating on the
terraces, in between two steps. These islands are Two possible models could explain the pro-
typically 0.4 nm high, which suggests that they are nounced differences in the surface morphology of
SrTiO3 unit cells. SrTiO3 substrates after annealing at 750°C in 03
(Fig. 1) or in O2 (Fig. 3).
3.4. Annealing in U H V Model I. Each observed surface reflects a thermo-
dynamically stable state for a given annealing
Fig. 6 displays the surface of a SrTiO3 substrate pressure and temperature. When cooling down to
which was annealed for 20 min at 750°C in UHV room temperature, this state is frozen in.
(background pressure of 10 -7 mbar). The warming Model H. Before annealing, the surface is disor-
and cooling procedures for this sample are compa- dered, probably due to the polishing of the sub-
rable to those used in 03 annealing described in strate. While annealing the sample, adatoms are
Section 3.1. desorbed and/or diffuse on the surface. Different
When imaging this sample with AFM, steps on annealing conditions (e.g. pressure and temper-

B. Stgiuble-Pfimpin et al./Surface Science 369 (1996) 313-320 317

(a) Model II, however, provides a consistent expla-
nation of the results discussed so far. It should be
pointed out that the idea of a disordered SrWiO 3
surface before annealing is in agreement with the
results of Ref. [3]. Using i o n scattering spectro-
scopy, Kawasaki et al. [3] showed that for com-
mercially available SrTiO3, the terminating atomic
layer was 5-25% SrO and 95-75% TiO2. Annealing
the substrates under appropriate conditions allows
the SrTiO3 surfaces to reconstruct and reach the
A energetically most favourable surface structure. For
a vicinal (001) SrTiO3 substrate, this corresponds
0 0.25 0.50 0.75 1.00 m, to atomically flat terraces separated by smooth
steps of one unit-cell in height, running parallel to
(b) ~ ] ~ v.. ,',mnc. each other. Our results are consistent with those
of Sum et al. [2,5], who showed that after 30 min
annealing in 0 2 at 800°C, the SrTiO3 surface was
characterised by parallel steps with rather curved
step ledges, while after 30 min at 900°C, these step
ledges were much straighter.
0 0.~ o,~ 0.~ A comparison between Figs. 1 and 3 and the
A ~ A' results of Refs. [2,5] clearly shows that the time
Fig. 3. (a) AFM image of an SrTiO3(001)substrate after annea- scale for surface reconstruction is much slower for
ling in 02 at 750°C for 1 h. The open arrows show step ledges the annealing process in Oz than for the annealing
which are parallel to crystallographic directions (see text for process in 03. In particular, the observation of
more details). The vertical profile taken along the (110) direc- steps of half a unit-cell in height (indicative for
tion (line A-A') is shown in (b). The markers show steps with different termination layers) strongly suggests that
heights of 1 and 2.5 unit cells of SrTiO3, respectively.
either the annealing time is too short or the
annealing temperature too low.
ature) correspond to different kinetics all leading It is likely that the lower background pressure
to the same end state, the surface structure with used for the 03 annealing and the reactiveness of
lowest (surface) energy. the 03 are responsible for a higher desorption rate
In order to test which of these two models can and a stronger interaction with the SrTiO3 surface,
best explain our experimental results, we checked respectively. In order to investigate the importance
whether the obtained surface morphologies were of the low background pressure and the role of the
reversible under subsequent annealings. A set of ozone, a substrate was heated to 750°C in U H V
substrates was first annealed in 03 (as described (see Section 3.4). A comparison between Fig. 1
in Section 3.1) and then in 02 (as described in (annealing in 03) and Fig. 6 (annealing in UHV)
Section 3.2). Another sample was first annealed in clearly shows that 03 plays an important role in
O2 (as described in Section 3.2) and then in 03 (as accelerating the reconstruction of the SrTiO 3
described in Section 3.1). In both cases, the surface surface.
structure after two consecutive annealings was Because of the reactiveness of the 03 molecules,
always characterised by parallel steps with a step the existence of a stronger interaction between
height of one unit-cell of SrTiO3. The surface SrTiO3 and 03 than between SrTiO3 and 02 is in
displayed in Fig. 3, which was obtained after a principle not surprising: it is well known that much
single annealing in 02 at 750°C, is therefore not a lower temperatures are required for an 03 molecule
reversible structure. As a consequence, model I to give up an O atom than for an 02 mole-
cannot explain our experimental data. cule. Furthermore, high-resolution transmission

318 B. Stgiuble-Piimpin et al./Surface Science 369 (1996) 313-320

5.0 nN

2.5 nH

0,0 nM
.50

.2S

0 0.25 0o 50 0,75
IJN

Fig. 4. A F M image of the same sample as displayed in Fig. 3, showing two chains of holes both parallel to the (100) direction. The
black arrow points to a square-shaped hole. The open arrows show step ledges which are parallel to crystallographic directions.

,0 n K
.3.1111]

.0 n M

.2.(11)

,0 nM

-1.011

o 1 .oo z.bo 3.bo °

Fig. 5. A F M image of an SrTiO3(00l) substrate which was first annealed in 03 followed by the deposition of Ten unit-cell layers of
SrTiO3. Note that the linear defect shown in this image is healing. Furthermore, the surface roughness has increased due to the
nucleation of SrTiO3 islands.

electron microscopy results have shown that the therefore speculate that during the annealing,
terminating layer of substrates having the equilib- oxygen atoms react with the uppermost SrO
rium surface structure is TiO2 [8]. One could atomic layer, leading to its dissolution. For com-

B. Stgiuble-Pf#npin et al./Surfaee Science 369 (1996) 313-320 319

1,0 rim

.~,0 nH

1.0 nN
~250

....~ 0
0 250 500
nl<

Fig. 6. A F M image of an SrTiO3(001 ) substrate after annealing in U H V at 750°C for 20 min. The terraces are not atomically flat.
The open arrows show step ledges which are parallel to crystallographic directions.

pleteness, it is interesting to mention that SrTiO3 dislocations during the annealing of the substrate.
substrates with a n equilibrium surface structure Such a relaxation can occur by glide of the disloca-
were obtained by Kawasaki et al. [3] without tion. However, the chains of holes described in
annealing the substrates, but by etching them with Section 3.2 indicate that another relaxation mecha-
a pH-controlled N H 4 F - H F (BHF) solution. Using nism might also play a role in the case of SrTiO3
ion scattering spectroscopy, Kawasaki et al. [3] substrates. Due to the stress field induced by
observed that the terminating atomic layer is dislocations below the surface of the substrates,
TiOz with a coverage factor of 100%, and con- favourable desorption sites for adatoms can align
cluded that the BHF solution selectively dissolves along crystallographic axes. During the annealing,
the SrO atomic plane. material preferentially evaporates from these
The remaining part of this discussion will con- desorption sites, leading to the formation of chains
centrate on the origin of the defects observed on of holes. As soon as enough material is evaporated,
most substrates. As described in Section 3, the the dislocations within the SrTiO3 could relax,
most striking features are one-dimensional arrays forming the one-dimensional arrays of defects
of defects parallel to the (100) and (010~ direc- described in Section 3.1.
tions. Because such defects were also observed after An attempt to smoothen the surface of the
the very slow annealing in O2 (see Section 3.2 it is substrates by first annealing them in 03 and then
unlikely that they are the result of thermal strain depositing a few layers of SrTiO3 was only partially
induced by a temperature gradient over the successful. It was found that the SrTiO3 was prefer-
samples. The fact that these linear defects follow a entially deposited along steps, such as, for example,
crystallographic orientation indicates that they can those of the line-shaped defects. As a consequence,
be related to dislocation lines already present in these defects were observed to heal. However,
the single-crystalline SrTiO3 before it was cut and SrTiO3 islands were also found to nucleate on the
polished. In Ref. [5], Sum et al. suggest that line- existing terraces, resulting in an increased rough-
shaped defects result from the relaxation of such ness of the surface of the substrates.

320 B. Stiiuble-Piimpin et al./Surface Science 369 (1996) 313-320

5. Conclusions found to heal after an O3 annealing followed by
the deposition of a few layers of SrTiO3. However,
This A F M study emphasises the importance of after the deposition of SrTiO3, the surface rough-
an appropriate pretreatment of SrTiO3 substrates ness increased due to the nucleation of SrTiO3
in order to obtain smooth and well-defined sur- islands.
faces. The results presented here strongly suggest
that, at high temperatures, the surface of the pol-
ished SrTiO3 substrate rearranges itself, either by Acknowledgements
diffusion or desorption of atoms, until reaching
the energetically most favourable surface structure. We are very grateful to J. van Wingerden for
helpful and stimulating discussions leading to this
For vicinal (001) substrates, this equilibrium struc-
work. One of us (B.S.-P.) acknowledges financial
ture corresponds to series of smooth, one unit-cell
support from DIMES. This work was supported
high steps running parallel to each other, with the
by F O M and the Dutch National Research
terrace width being defined by the magnitude of
P r o g r a m ( N O P ) for High-T~ Superconductors.
the vicinal angle.
The time needed to reach this equifibrium struc-
ture was found to depend strongly on the annealing References
conditions. For instance, it was observed that, for
the same temperature (750°C), annealing a sub- [ l'l N. Bickel, G. Schmidt, K. Heinz and K. Mtiller, Phys. Rev.
strate in 03 resulted in a smoother, better defined Lett. 62 (1989) 2009.
surface than annealing (even for longer times) in [2] R. Sum, R. LtRhi, H.P. Lang and H.-J. GOntherodt, Phys-
O2 or in UHV. In terms of rearrangement of the ica C 235-240 (1994) 621.
['3] M. Kawasaki, K. Takahashi, T. Maeda, R. Tsuchiya, M.
SrTiO3 surface, annealing in 03 is more efficient Shinohara, O. Ishiyama, T. Yonezawa, M. Yoshimoto and
than annealing in Oz or in UHV. Based on our H. Koinuma, Science 266 (1994) 1540.
results, it is likely that the reactiveness of 03 plays ['4] V. Ravikumar, D. Wolf and V.P. Dravid, Phys. Rev. Lett.
an important role in accelerating the reconstruc- 74 (1995) 960.
[5"] R. Sum, H.P. Lang and H.-J. Gtintherodt, Physica C 242
tion of the SrTiO3 surface. (1995) 174.
Finally, several types of defects were observed [6] Q.D. Jiang, D.-M. Smilgies, R. Feidenhans'l, M. Cardona
on the substrates investigated, such as holes (diam- and J. Zegenhagen,Proc. EUCAS 95, Edinburgh, July 3-6,
1995, to be published.
eter 10-20nm), (screw-) dislocations (density [7] Q.D. Jiang and J. Zegenhagen, Surf. Sci., in press.
6-t-1 x 10 7 c m - 2 ) and line-shaped defects parallel [8] J.G. Wen, C. Traeholt and H.W. Zandbergen, Physica C
to crystallographic orientations. The later were 205 (1993) 354.

1996 atomic force microscopy study of (001) sr tio3 surfaces

More Related Content

Viewers also liked

Similar to 1996 atomic force microscopy study of (001) sr tio3 surfaces

More from pmloscholte

Recently uploaded

1996 atomic force microscopy study of (001) sr tio3 surfaces