Qiu X.G. (Ed.) High Temperature Superconductors
Подождите немного. Документ загружается.
Deposition technologies, growth and properties of high-T
c
films 21
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
heavy-metal elements restricted MOCVD of HTS to the use of solid precursors.
These solid precursors turned out to be unstable in the vapour-phase composition,
e.g. the commonly used precursors for YBCO deposition (the so-called thd
precursors: 2,2,6,6,-tetramethyl-3,5-heptanedionate for Y, Ba, and Cu, respectively)
exhibit a high melting point and a low vapour pressure. This instability is strongest
of the Ba component, as demonstrated for Ba(thd)
2
(Becht, 1996; Richards et al.,
1995; Nagai et al., 1997; Burtman et al., 1996; Busch et al., 1993). In order to
stabilise the vapour-phase composition of YBCO, alternative precursors or/and
novel concepts had to be developed.
For example, the successful use of a fluorinated barium precursor
(Ba(tdfnd)
2
·tetraglyme, where tdfnd is tetradecafluorononanedione) has been
reported that is stable at its operating vaporisation temperature of ~145°C
(Richards et al., 1995). Using the precursors Ba(tdfnd)
2
·tetraglyme,
Y(thd)
3
·4tBuPyNO, and Cu(tdfnd)
2
·H
2
O, which are all evaporated from the
liquid phase, a high reproducibility of the film composition was reported (Busch
et al., 1993). However, these fluorinated precursors required the addition of water
vapour in the deposition process to avoid the formation of BaF
2
, which in turn led
to problems associated with the removal of HF. Nagai et al. (1997) and Yoshida
et al. (1996) reported that high-crystalline quality MOCVD YBCO films
1.8 Pressure–temperature phase diagram for YBCO according to
Bormann et al. (1989) and Hammond et al. (1989). The 1–2–3 phase
is stable in the regime indicated by the shaded area, which is divided
into the tetragonal and orthorhombic phase. The deposition typically
takes place in the tetragonal regime. In-situ post-deposition treatments
typically follow routes (A) or (D), here indicated for high-pressure
sputtering.
22 High-temperature superconductors
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
are obtained by liquid-state nonfluorinated sources of Y(thd)
3
·4tBuPyNO,
Ba(thd)
2
·2tetraene, and Cu(thd)
2
. Other adducts that have been used to stabilise
the Ba-thd precursor include triglyme and phenanthroline.
Another approach of improved stability is given by the transition from solid-
source precursor delivery to liquid-source precursor delivery. These concepts
include: (i) liquid-source precursor delivery systems, (ii) direct injection of the
precursor in a flash vaporiser (Onabe et al., 2001) or (iii) a two-step evaporation
process in a belt-driven system, where the solvent is evaporated in the first
step and the precursor is subsequently vaporised in the second step (Stadel
et al., 2000).
Irrespective of whether solid- or liquid-phase precursors were used, the
traditional method of vapour delivery, with each precursor delivered from a
separate vessel, had limited the application of MOCVD to short time runs for
YBCO wafer fabrication. Single-source liquid delivery systems (Zhang et al.,
1992) are more attractive, especially for the continuous deposition of YBCO
films, for instance for coated conductors (Stadel et al., 2003; Selvamanickam
et al., 2003). Recent developments in MOCVD precursors and their delivery
schemes have been responsible for the dramatic progress in using the MOCVD
technique for short time runs and continuous deposition of HTS material.
1.2.3 CSD techniques
Chemical solution deposition (CSD) of HTS material represents all chemical
deposition methods ranging from sol-gel to metal-organic decomposition (MOD).
Since CSD is no vacuum technology, the major advantage of this process lies in
the low investment costs and low energy consumption. Among the different
ex-situ techniques, CSD is nowadays one of the most promising candidates for
deposition of HTS coated conductors including both buffer layer and YBCO
(Pomar et al., 2006). On the one hand, it has been demonstrated that high-quality
epitaxial YBCO can be grown by PVD onto MOD buffer layers (Rupich et al.,
1999; Jarzina et al., 2005). On the other hand, it has been shown that high critical
currents can be achieved in YBCO thin films grown by the trifluoroacetates
(TFA, (OCOCF
3
)) route on single crystals or on vacuum deposited buffer layers
(Araki and Hirabayashi, 2003; Castano et al., 2003; Rupich et al., 2004; Obradors
et al., 2004).
Although different chemical solutions can be chosen (e.g. TFA route, fluorine-
free route, water or water-free routes) the basic steps of the CSD processes are
fairly similar (Fig. 1.9). Starting with the different educts, the precursor is
synthesised. The resulting solution is deposited onto the substrate via spray, spin
or dip coating. Depending on the process a consolidation starts due to chemical
reaction (sol-gel) or evaporation of the solvent (MOD). Drying steps and pyrolysis
(typically at 150–400°C) lead to an amorphous layer. Finally this layer is
crystallised at elevated temperature (typically 600–1000°C).
Deposition technologies, growth and properties of high-T
c
films 23
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
1.3 HTS film growth characterisation
Growth of cuprate films has remained at best a complicated matter due to its
rather complex multicomponent crystal structures of the layered cuprates, which
are prone to a variety of defects and growth morphologies, and the extreme
deposition parameters (high process pressure and temperature), which furthermore
have to be controlled very accurately. A brief sketch of growth mechanisms, in
general, and HTS thin film growth, in particular, is given in this section. The
general outline of the description is given in Fig. 1.10, detailed information is
given in Wördenweber (1999). The emphasis of the following brief description
is on PVD processes and the specific problems of adsorption of particles,
oversaturation, epitaxy, lattice mismatch, defect generation, stress and strain that
represent (a) important problems and (b) options for engineering the properties of
oxide films, especially HTS thin films.
1.3.1 Nucleation and phase formation
After transport, particles impinge on the substrate. Chemical adsorption is only
possible if the chemical adsorption potential E
a
> 0, i.e., if the system is chemically
(600–1000°C)
1.9 Schematic sketch of a CSD technology according to Bäcker (2008).
(a) Diagram of the AWAT process based on acetates (C
2
H
4
O
2
) and
(b) continuous dip-coating process for the production of HTS coated
conductors.
24 High-temperature superconductors
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
not inert (Fig. 1.11). The adsorption rate is defined by the number of adatoms per
area and time:
[1.5]
and the adatom average migration is given by:
[1.6]
respectively. D = D
o
exp (– E
d
/kT
s
) describes the diffusion, a
o
= (D
o
/v)
1/2
is the
hopping rate,
τ
= (1/v) exp (E
a
/kT
s
) is the average resident time,
υ
≈ 10
12
–10
14
Hz is
the attempt frequency, and T
s
, D
o
, E
a
and E
d
represent substrate temperature, diffusion
constant, chemical adsorption potential and desorption potential, respectively. During
growth from melt (e.g. liquid phase epitaxy) or solution the adatom can easily be
desorbed from the surface, i.e. the residence time is relatively short. During growth
from vapour (PVD and CVD techniques) the activation energy for desorption is
typically larger, resulting in a larger migration length and a larger sticking coefficient.
Nevertheless, inserting typical values for E
a
, E
d
and D
o
for oxides and PVD
yields extremely small values for the resident time and migration of
τ
<< 1ps and
λ
< 1 nm. Experimentally much larger values of
τ
≈ 0.2–0.5 s and
λ
≈ 100–500 nm
are reported (Koster et al., 1998). This difference is explained by the particle
transport via the gas phase (Wördenweber, 1999). For this mechanism, a
1.10 Schematic sketch of the adsorption, nucleation, and thin film
growth on ideal substrates (left route) and dominated by defects of the
substrate’s surface (right route). The nucleation and growth process of
films can be considered to be a transition from a ‘1D’ particle to a 3D
layer (arrow left side).
Deposition technologies, growth and properties of high-T
c
films 25
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
supersaturated gas phase has to be established at the substrate. Desorbed atoms
will be re-adsorbed and, therefore, they can migrate larger distances than
theoretically expected (Fig. 1.12). As a consequence supersaturated gas phases
are vital for the phase formation and deposition. This represents a problem for a
number of deposition devices, especially for large-area PVD.
During growth of lattice-matched layers, a layer-by-layer (Frank-van der Merwe)
growth mechanism is usually observed. However, even in a layer-by-layer growth
the resulting heterojunction interfaces are not perfectly planar. This is especially
important in devices (e.g. Josephson junctions) where a monolayer change or defects
can strongly affect the electronic properties. Layer-by-layer growth is generally
limited by the surface morphology of the underlying substrate (see Fig. 1.13):
• As long as the surface migration length
λ
of the oncoming adatoms is clearly
smaller than substrate features (e.g. l
s
defined by terraces separated by step
edges in vicinally cut single crystal substrates) growth occurs due to the
nucleation of two-dimensional (2D) islands.
• If the migration length is larger than surface terrace size (i.e.
λ
> l
s
), growth
is dominated by attachment of adatoms to the step edges of the terraces.
The step edges then propagate at a velocity depending on the step density
and growth rate. This limit is called step-flow growth. The difference
between these two regimes can be observed by RHEED oscillations (see
Fig. 1.13(a)).
1.11 Schematic sketch of the adsorption potentials the impinging atom
experiences at the film surface.
26 High-temperature superconductors
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
1.12 Schematic sketch of the phase formation without (a) and with (b)
particle transport via the supersaturated gas phase.
1.13 (a) Schematic sketch of step flow growth and nucleation of 2D
islands, (b) sketch of various topographical structures of real single
crystalline substrates, and (c) atomic force microscopy micrographs
of YBCO step flow growth on a vicinal cut SrTiO
3
substrate and spiral
growth (Matijasevic et al., 1996).
Deposition technologies, growth and properties of high-T
c
films 27
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
• Finally, since ‘real substrates’ are usually cut in slight vicinal ways in both
in-plane directions, kinks are usually present at the steps (Fig. 1.13(b)). These
kinks can be the starting point for spiral growth mode (Fig. 1.13(c)).
Thus, the growth (heteroepitaxial nucleation and homoepitaxial growth) are
strongly determined by the competition between the mobility of the oncoming
particles, which can be affected by the preparation method and conditions, and the
surface morphology of the substrate, which is among others defined by the cut of
the substrate. By modifying the preparation process or the miscut (vicinal) angle
of the substrate, the growth mode can be changed from 2D island growth to step
flow growth or spiral growth.
1.3.2 Heteroepitaxial growth, stress and defects
in HTS films
Heteroepitaxial layer growth represents a major technology for advanced thin film
electronic – e.g. for semiconductor and oxide electronics – and for basic solid
state research. The most fundamental questions in this automatically strained-
layer growth nevertheless are:
• Up to what thickness are heteroepitaxial layers stable?
• In which type of misfit do defects develop?
• What happens upon modification of the misfit for instance due to cooling of
the film?
Generally, it is believed that below a critical thickness the strained state is the
thermodynamic equilibrium state, and above a critical thickness a strained layer
may be metastable or it may relax (Downes et al., 1994). Different critical
thicknesses might be associated to different types of misfit defects, e.g.
dislocations, misalignments, or even cracks (Zaitsev et al., 1997a). Stress
σ
in
heteroepitaxially grown films results from both an intrinsic and temperature-
dependent component. The main reason for the development of intrinsic stress in
heteroepitaxially grown films is given by the nominal lattice mismatch
[1.7]
at growth condition, where ‘a’ represents the different in-plane lattice parameters.
The thermal contribution is due to the difference in thermal expansion coefficient
between the film and the underlying substrate resulting in a temperature modified
strain
ε
o
(T) for instance during cooling of the sample (Ohring, 1991; Hoffman,
1976; Taylor et al., 2002). Due to these two contributions, the resulting generation
of defects in our films is discussed in two steps:
(1) generation of lattice misfit dislocations during the growth of HTS films at
elevated temperature (typically 700–900°C for YBCO), and
28 High-temperature superconductors
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
(2) generation of cracks that are most likely generated during the cooling down
after deposition.
During deposition, defect-free films grow up to a critical thickness d
c
from whereon
defects develop in the layer. This situation is most generally described initially by van
der Merwe (1962) and later by Matthews and Blakeslee (1974). In these theories, the
line tension of a misfit dislocation of finite length is balanced by the force due to the
strain in the layer on the termination of the misfit dislocation. Alternatively and
equivalently, the energy of the system with and without the misfit dislocation may be
considered (Hu, 1991; Fitzgerald, 1991; Freund, 1990). Although in either approaches
several approximations of uncertain effect are made, reasonable values for the critical
thickness d
c
are obtained that are comparable to experimental values (People and
Bean, 1985) and provides the basis for the discussion of defect development in
semiconducting thin films. For instance the Matthews theory predicts a critical
thickness for the development of dislocation lines (Matthews and Blakeslee, 1974):
[1.8]
where C contains the details of the crystal and the dislocation. Different versions
of the term O(d
c
) have been discussed by different authors (Downes et al., 1994;
Matthews and Blakeslee, 1974; Hu, 1991; People and Bean, 1985), the resulting
predictions of critical thickness can vary by at least a factor of two. For instance
the development of misfit dislocation lines has been described with this approach
by People and Bean, (1985):
[1.9]
where b and
ν
represent the extension of the dislocation line, and Poisson’s ratio,
respectively.
Typical predictions for the critical thickness describing the development of
misfit dislocations in YBCO are given in Fig. 1.14. Inserting reasonable values
and taking into account that the theories are estimates of the critical thicknesses
within a factor of about 2, for YBCO on SrTiO
3
a critical thickness d
c
< 10 nm is
predicted (Fig. 1.14(c)). Thus, misfit dislocations will be generated already for
extremely thin films.
After deposition of the HTS layer at elevated temperature (e.g. 600–900°C for
different PVD technologies) the sample is cooled down to room temperature. Due
to the mismatch of the thermal expansion coefficient, additional stress is imposed
on the HTS layer (see eq. [1.7]). Taking into account the presence of misfit
dislocations generated during the growth process, the resulting strain imposed on
the film can be described (Zaitsev et al., 1997a; Fitzgerald, 1991) by:
[1.10]
Deposition technologies, growth and properties of high-T
c
films 29
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
1.14 (a) Experimentally determined critical thickness d
crack
for YBCO/CeO
2
on r-cut sapphire, (b) images of cracks in YBCO on sapphire (Kästner et
al., 1995), and (c) theoretical prediction of different critical thicknesses as
function of lattice misfit. In (c), the critical thickness for the generation of
misfit dislocations according to van der Merwe (1962) (black line) and
Matthews theory (dotted line, eq. [1.9]), and the critical thickness for crack
generation according to eq. [1.10] and different values of
δ
is plotted. The
lines indicate the lattice misfit of YBCO with respect to various substrates.
30 High-temperature superconductors
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
20
1
2
3
4
5
6
7
8
9
30
1
2
3
4
5
6
7
8
9
40
1
2
43X
© Woodhead Publishing Limited, 2011
where
δ
describes the release of stain due to the presence of defects or
misalignments. While defects and misalignments have developed during growth,
cracks are expected to develop during cooling down if this is energetically
favourable. According to the theory of fracture of solids, the amount of strain
energy that is released per unit length of a two-dimensional crack (Zaitsev et al.,
1997a; Cottrell, 1975) is given by:
[1.11]
where a represents the height of the crack (a = d, if the crack is normal to the film
surface) and Y is the Young’s modulus of the material. The Young’s modulus of
bulk YBCO amounts to Y ≈ 300 (Fitzgerald, 1991). The formation of a crack is
energetically favourable, if E
crack
exceeds the energy E
surface
(energy per unit crack
length) required for the (initial) formation of the two new surfaces of the crack.
Therefore, the critical film thickness for crack formation is given for E
crack
= 2E
surface
:
[1.12]
with E
'
surface
= E
surface
/d. The energy of the crack’s surfaces is estimated by
summation of the binding energies E
B
of all atoms at these surfaces. Generally
binding energies of oxides are of the order of 15 eV. The resulting surface energy
for crack along [100], [010] or [001] axes of YBCO results in a surface formation
energy E
'
surface
≈ 31–34 J/m
2
(Zaitsev et al, 1997a; Hollmann et al., 2009).
Figure 1.14 shows the resulting critical thickness d
crack
for different values of
δ
(in units of
ε
o
). The theoretical values of d
crack
are comparable with experimental
observation for epitaxial HTS layers in general. Ideal, defect-free films show the
smallest critical thickness. For a misfit of ~10–11% (e.g. YBCO on sapphire)
crack-free layers are limited by d
crack
< 30 nm. With increasing
δ
the critical
thickness increases.
All heteroepitaxially grown HTS films have the following.
• They possess misfit defects that are generated during the growth unless they
are quite thin (d < d
c
). As a result,
δ
> 0 and the critical thickness for crack
formation is enhanced with respect to the critical thickness of defect-free HTS
films.
• The critical thickness for crack formation d
crack
is large for small lattice
mismatch, it decreases strongly for larger mismatch, e.g. YBCO on YSZ,
MgO or sapphire.
• The critical thickness d
crack
can be enhanced by artificially inducing adequate
defects that take up the stress in the layer. This has been demonstrated for
YBCO on sapphire used for microwave applications (Einfeld et al, 2001).
The generation of defects (e.g. Y
2
O
3
precipitates in YBCO) represents one
way to engineer the mechanical and electronic properties of HTS films.