Comments on the wetting behavior of non-porous substratesfor ceramic coated-conductor applications
Pieter Vermeir • Frank Deruyck • Jonas Feys •
Petra Lommens • Joseph Schaubroeck •
Isabel Van Driessche
Received: 18 July 2011 / Accepted: 5 March 2012 / Published online: 27 March 2012
� Springer Science+Business Media, LLC 2012
Abstract This work gives an overview of the possibilities
to improve the wetting behavior of precursors for coated
conductors on non-porous substrates. Within this work, all
coatings were performed on a metallic Ni–W/La2Zr2O7/
CeO2 substrate using water-based Y, Ba, Cu containing
precursors. The results described in this paper can be used
for different technologies of chemical solution deposition,
as there are ink jet printing, dip coating, spin coating etc.
Starting from the forces involved during wetting, a sepa-
ration between solid and liquid modifications was made.
This study revealed that if a good cleaning procedure of the
substrate, whether or not combined with a targeted modi-
fication of the precursor is applied, water-based solutions
can be used without restriction towards their wetting
behaviour leading to a sustainable technology within the
coating industry. Within this work, special attention is
given to (1) fast determination of the substrate cleaning
procedure quality by the creation of wetting envelopes and
(2) the use of a screening design of experiment to study the
effects of intrinsic solution factors, such as precursor for-
mulation, influencing the coating behavior. All modifica-
tion discussed are expandable to all kinds of precursors and
substrates.
Keywords Wetting � Chemical solution deposition �Coating � Water-based
1 Introduction
The achievement of low cost deposition techniques for high
critical current YBa2Cu3O7-d coated conductors is one of the
major objectives to achieve a widespread use of supercon-
ductivity in power applications. The main advantages of
using chemical solution deposition (CSD) for the develop-
ment of textured oxide coatings are the lower investment
cost, an energy efficient operation mode, the faster deposi-
tion with higher yield and the processing under ambient
pressure enabling a complete continuous process [1–8].
Deposition of thin layers for coated conductors is today
mainly based on physical deposition processes. The most
common techniques are laser ablation, thermal and electron
beam evaporation and sputtering [9, 10]. Various products
with such functional layers have been commercialized in
the past, especially in the field of micro-electronics
[11, 12]. Nevertheless, the established physical deposition
processes are hampered by their complex process tech-
nology. The necessity of the deposition in vacuum cham-
bers, the use of energy consuming evaporation systems and
the low yield and deposition speed are the main factors
restricting the further economic use of these processes.
In this regard, taking into account that, in solution chem-
istry, highly organized materials can be prepared by gelation
followed by heating at moderate temperatures, soft processing
for advanced ceramics has been developed opening the way
for a new generation of low expenditure of energy in chemical
engineering. The possibility to fabricate a variety of materials
from solutions has been demonstrated already, but for the
design of innovative materials processing, particularly when
P. Vermeir � J. Schaubroeck
Faculty of Applied Engineering Sciences, University College
Ghent, Schoonmeersstraat 52 (C), 9000 Ghent, Belgium
P. Vermeir � J. Feys � P. Lommens � I. Van Driessche (&)
Department of Inorganic and Physical Chemistry, Ghent
University, Krijgslaan 281 (S3), 9000 Ghent, Belgium
e-mail: [email protected]
F. Deruyck
Faculty of Science and Technology, University College Ghent,
Valentin Vaerwyckweg 1B, 9000 Ghent, Belgium
123
J Sol-Gel Sci Technol (2012) 62:378–388
DOI 10.1007/s10971-012-2737-3
rapid and environmentally friendly routes are preferred, sig-
nificant industrial progress is still needed.
Therefore, within the CSD-methods, there is an
increasing tendency to prepare films starting from envi-
ronmental friendly water-based precursor solutions,
resulting in a sustainable process for film deposition
[13–17]. The main challenge of this approach is the poor
coating behavior of these water-based precursors on a
variety of substrates. Polarity and cohesive hydrogen
bonding between water molecules lead to a very high
surface tension causing this poor coating performance. In
addition, the low surface tension of several substrates such
as metals, polyesters etc. makes things even more
complicated.
In this work, a summary is given of several possibilities to
improve wettability. Starting from Young’s equation, a
subdivision of two categories is made: (1) substrate or solid
and (2) solution or liquid modifications. Within the last
category, this paper gives new insights in the coating
behavior of water-based CSD precursors, more specifically
on the influence of precursor composition. This study was
performed using a design of experiment (DOE). In this work,
CSD precursors containing Y, Ba and Cu are used to evaluate
the coating behavior on Ni–W/La2Zr2O7/CeO2-substrates
applicable in coated-conductor applications. The results are
expandable to different precursor systems and substrates.
2 Experimental
2.1 Precursor solutions
Preparing clear and stable metal (M) containing precursor
solutions is a primary step in producing good quality
coatings via CSD. In order to realize this, a general strategy
was followed. To obtain a suitable layer thickness, all
precursor solutions were set to obtain a total M-concen-
tration of 0.6 M. Therefore, as all metals are introduced via
M-salts into the solvent, the most common precipitation is
in the form of the M-salt itself as the solubility product is
mostly exceeded. When water is used as a solvent, one
has to take into account the formation of insoluble
M-hydroxides [18]. The latter implies the importance of pH
control in every step of the precursor preparation. In order
to avoid precipitation, complexing agents are added to
form complexes with the free metals in the solution. Again
pH-control is a very important issue as complex formation
is pH dependent [19–22]. In this work, nitrates and acetates
are used as M-sources, ethylenediaminetetraacetic acid
(EDTA) and citric acid (CA) are used as complexing
agents, ammonia and ethanolamine are used as a base
while formic acid is used as an acid. As every precursor is
based on different M-sources, complexing agents and
bases, specific details for the preparation of different pre-
cursor solutions are given in Table 3.
2.1.1 Preparation of the precursor solutions containing
M-acetates, EDTA and NH4OH as base
EDTA (Sigma Aldrich [ 99 %) is mixed with 40vol% of
H2O. NH4OH (Sigma Aldrich 28 % in H2O) is added
until a stable pH of 8.5 is reached. After dissolution
of all components, Y(CH3COO)3�xH2O (Sigma Aldrich
99.9 %) and Ba(CH3COO)2 (Sigma Aldrich 99 %) were
added successively. Before Cu(CH3COO)2�H2O (Sigma
Aldrich C 98 %) was added, the pH was set neutral using
NH4OH. Afterwards, NH4OH or HCOOH (Sigma
Aldrich C 96 %) was used until the exact pH, conform
Table 1 or 2, was reached. Finally, H2O was used to dilute.
2.1.2 Preparation of the precursor solutions containing
M-acetates, EDTA and ethanolamine as base
EDTA is mixed with 40 vol% of H2O and ethanolamine
(Sigma Aldrich C 98 %) is added until a stable pH of 8.5
is reached. After dissolution of all components, Ba(CH3-
COO)2 was added. Another 20 vol% of H2O was added to
dissolve the Ba(CH3COO)2. Then, Y(CH3COO)3.xH2O
and Cu(CH3COO)2�H2O are added followed by 10 vol% of
additional H2O. Finally, the pH was reached using etha-
nolamine or HCOOH until the exact pH, conform Table 1
or 2, was reached. H2O was added to dilute.
2.1.3 Preparation of the precursor solutions containing
M-acetates, CA and NH4OH as base
After CA (Sigma Aldrich 99 %) is mixed with 20 vol% of
H2O, the pH is stabilized at 8 using NH4OH. Once everything
is dissolved Y(CH3COO)3�xH2O, Ba(CH3COO)2 and
Cu(CH3COO)2�H2O are added separately. Finally, NH4OH
or HCOOH was used until the exact pH, conform Table 1 or
2, was reached and H2O was added to dilute.
2.1.4 Preparation of the precursor solutions containing
M-acetates, CA and ethanolamine as base
After CA is mixed with 40vol% of H2O, the pH is stabi-
lized at 4 using ethanolamine. Once everything is dissolved
Y(CH3COO)3�xH2O, Ba(CH3COO)2 and Cu(CH3COO)2�H2O are added separately. Finally, ethanolamine or
HCOOH was used until the exact pH, conform Table 1 or
2, was reached and H2O was added to dilute.
J Sol-Gel Sci Technol (2012) 62:378–388 379
123
2.1.5 Preparation of the precursor solutions containing
M-nitrates, EDTA and NH4OH as base
EDTA is mixed with 40 vol% of H2O and NH4OH is added
until a stable pH of 8.5 is reached. When everything
is dissolved, Y(NO3)3�6H2O (Alfa Aesar 99.9 %) and
Ba(NO3)2 (Alfa Aesar 99 %) were added successively. For
a low M/EDTA-ratio, the pH needs to be adapted to 8.3
using NH4OH before continuing. Then Cu(NO3)2�2.5H2O
(Alfa Aesar 98 %) was added very slowly, adding 5 wt%
each time till dissolved in the mixture. Afterwards,
NH4OH or HCOOH was used until the exact pH,
conform Table 1 or 2, was reached. Finally, H2O was
used to dilute.
2.1.6 Preparation of the precursor solutions containing
M-nitrates, EDTA and ethanolamine as base
EDTA is mixed with 40vol% of H2O and ethanolamine is
added until a stable pH of 8.5 is reached. When everything is
dissolved, Y(NO3)3�6H2O and Ba(NO3)2 were added suc-
cessively. Before Cu(NO3)2 was added, the pH was changed
to 8.5 using ethanolamine. Then Cu(NO3)2�2.5H2O was added
Table 2 Experimental runs
DOE High pH rangeRun M-source Complexant pH high [Complexant] Alkalinity
control
Surface
treatment
Coverage
score
1 Acetate CA 9.5 Low Ammonia Heat 10
2 Acetate EDTA 8.5 High Ethanolamine Heat 10
3 Nitrate EDTA 9.5 Low Ethanolamine Ultrasonic 7
4 Nitrate CA 8.5 High Ammonia Ultrasonic 2
5 Nitrate EDTA 8.5 Low Ammonia Heat 10
6 Acetate EDTA 9.5 High Ammonia Ultrasonic 4
7 Nitrate CA 9.5 High Ethanolamine Heat 10
8 Acetate CA 8.5 Low Ethanolamine Ultrasonic 3
9 Nitrate EDTA 8.5 High Ethanolamine Ultrasonic 2
10 Acetate EDTA 9.5 Low Ethanolamine Heat 7
11 Acetate CA 8.5 High Ammonia Heat 6
12 Nitrate CA 9.5 Low Ammonia Ultrasonic 4
13 Nitrate EDTA 9.5 High Ammonia Heat 10
14 Acetate CA 9.5 High Ethanolamine Ultrasonic 10
15 Acetate EDTA 8.5 Low Ammonia Ultrasonic 1
16 Nitrate CA 8.5 Low Ethanolamine Heat 10
Table 1 Experimental runs
DOE Low pH rangeRun M-source Complexant pH low [Complexant] Alkalinity
control
Surface
treatment
Coverage
score
1 Acetate CA 7.5 Low Ammonia Heat 9
2 Acetate EDTA 6.5 High Ethanolamine Heat 6
3 Nitrate EDTA 7.5 Low Ethanolamine Ultrasonic 2
4 Nitrate CA 6.5 High Ammonia Ultrasonic 3
5 Nitrate EDTA 6.5 Low Ammonia Heat 10
6 Acetate EDTA 7.5 High Ammonia Ultrasonic 4
7 Nitrate CA 7.5 High Ethanolamine Heat 10
8 Acetate CA 6.5 Low Ethanolamine Ultrasonic 1
9 Nitrate EDTA 6.5 High Ethanolamine Ultrasonic 1
10 Acetate EDTA 7.5 Low Ethanolamine Heat 10
11 Acetate CA 6.5 High Ammonia Heat 10
12 Nitrate CA 7.5 Low Ammonia Ultrasonic 2
13 Nitrate EDTA 7.5 High Ammonia Heat 10
14 Acetate CA 7.5 High Ethanolamine Ultrasonic 5
15 Acetate EDTA 6.5 Low Ammonia Ultrasonic 7
16 Nitrate CA 6.5 Low Ethanolamine Heat 10
380 J Sol-Gel Sci Technol (2012) 62:378–388
123
followed by a pH adjustment, conform Table 1 or 2, using
ethanolamine or HCOOH. Finally, H2O was used to dilute.
2.1.7 Preparation of the precursor solutions containing
M-nitrates, CA and NH4OH as base
After CA is mixed with 40 vol% respectively 20 vol% of
H2O for a low respectively high M/CA-ratio, the pH is
stabilized at 4.0 using NH4OH. Once everything is dis-
solved Y(NO3)3�6H2O and Ba(NO3)2 are added separately.
Again, the pH was adjusted using NH4OH to 8.3 respec-
tively 6.0 for respectively a low and high M/CA-ratio.
Then Cu(NO3)2�2.5H2O was added very slowly, adding
5 w% each time till dissolved in the mixture. Afterwards,
NH4OH or HCOOH was used until the exact pH, conform
Table 1 or 2, was reached. Finally, H2O was used to dilute.
2.1.8 Preparation of the precursor solutions containing
M-nitrates, CA and ethanolamine as base
After CA is mixed with 30 vol% of H2O, the pH is stabi-
lized at 4 using ethanolamine. For a high M/CA-ratio no
pH adjustment is needed. After adding Y(NO3)3.6H2O and
Ba(NO3)2 separately, the pH was changed using ethanol-
amine to 6.0 respectively 3.0 for respectively a low and
high M/CA-ratio. Then Cu(NO3)2�2.5H2O was added very
slowly, adding 5 w% each time till dissolved in the mix-
ture. Afterwards, ethanolamine or HCOOH was used until
the exact pH, conform Table 1 or 2, was reached. Finally,
H2O was used to dilute.
2.2 Substrates and coatings
The substrates used are Ni–W/La2Zr2O7/CeO2-substrates.
The Ni–W was performed by Evico using the RABiTS
procedure [23], while the La2Zr2O7 and CeO2 layers were
obtained by Zenergy power GmbH via a CSD method [24].
In order to avoid dust contamination during thin film
deposition, cleaning procedures are performed in a dedi-
cated class 10,000 clean room. Coating procedures are
performed using a computer-controlled precision dip coater
in a laminar flow box at class 10.
Dust particles are removed from all solutions by filtering
the solution with 0.45 lm hydrophilic polypropene (GHP)
membrane filters. All films were dip coated at a withdrawal
speed of 50 mm min-1. Again, this principle is expandable
to other coating technologies such as spin coating and ink
jet printing [25–27].
2.3 Techniques
Contact angle measurements and surface tension calcula-
tions were obtained using the DSA30 module from Kruss.
For screening the possible effect of the precursor compo-
sition parameters the fractional factorial DOEs approach is
used. Using the experimental data, a regression model was
constructed. This model was validated using (1) the cor-
relation coefficient Radjusted2 which is a modified R2 that
adjusts for the number of explanatory terms in the model,
(2) the F-test statistic determining if the regression model is
statistically significant and (3) the t test of every parameter
estimate. The validation of the regression model using the
F-test, lies in the rejection or acceptance of the so-called
‘Null’ hypothesis H0, which states that the linear model
regression is not significant. As F is defined as the ratio of
the mean squares of the regression model to residuals, large
F-values occur with models that explain a significant
fraction of the variation in the response. In that case, the H0
is rejected. Together with the F-ratio, a p value is obtained,
defined as the probability of the actual outcome of the test
statistic, assuming H0 is true. If p \ 0.05 we take \5 %
risk of rejecting H0. The 5 % significance level was set in
advance. The validation of parameter estimates with the
t test shows a similar interpretation as the F-test, where H0
states that the parameter of the regression model is not
significant.
In order to comprehend the analysis of variance
(ANOVA) table, several definitions are given: (1) df stands
for degrees of freedom, specified as the number of obser-
vations minus the number of constraints, (2) mean squares
can be calculated as the sum of squares divided by the
respectively degrees of freedom.
3 Results and discussion
The use of water as primary solvent involves a high surface
tension of the obtained solutions, mostly resulting in poor
wetting behavior when used for coatings. According to the
extended Young’s equation for non-ideal surfaces [28–30],
one can anticipate substrate and/or precursor manipulations
to increase the wetting behaviour. Therefore, a subdivision
between solid and liquid modifications is made.
3.1 Substrate
In order to improve the wettability for a certain precursor
solution on a specific substrate, a good cleaning procedure
will generally be very important [31–33]. Normally, when
the extended Young equation is used, a measurement of
many different parameters, which is time consuming
knowing that maybe not all parameters are included, is
necessary. As an alternative, this work proposes to incor-
porate all extending variables correlating to a non-ideal
surface into a semi interfacial tension of the solid versus
vapour csv’ in order to have a quick overview of the
J Sol-Gel Sci Technol (2012) 62:378–388 381
123
different substrate pre-treatments. To expand this theory
from a static to a dynamic environment, the evaluation
criterion is changed towards the coating quality using a
coverage score going from 0 to 10 which is in turn
reflecting the coverage percentage of respectively 0 to
100 % of the substrate. Figure 1 gives an overview of the
coverage scores that are generally seen in practice.
In this work the wettability of Ni–W/La2Zr2O7/CeO2
substrates using three different cleaning procedures was
evaluated: (1) uncleaned, (2) degreased with isopropanol
followed by an ultrasonic (US) treatment for 10 min in
isopropanol and (3) degreased with isopropanol followed
by a heat-treatment at 400 �C for 5 min on a hot-plate.
Precursor 5 with a high amount of complexant and a final
pH of 9.5 was used to obtain the coverage scores after the
different cleaning procedures. For an uncleaned Ni–W/
La2Zr2O7/CeO2 substrate a coverage score of 2 was
obtained, for the US-cleaned one the coverage score
increased to 4, while for a heat-treated Ni–W/La2Zr2O7/
CeO2 substrate a coverage score of 10 was obtained.
When csv
0was determined using the Owens–Wendt
method [34], a csv
0of 37.19 ± 0.16, 40.24 ± 0.23 and
52.76 ± 0.17 mN/m was obtained for an uncleaned, US
cleaned and respectively heat-treated substrate (Table 3).
These values were calculated based on the contact angle
measurement of water and diiodomethane. Remark that
these results are in good agreement with the experimental
coverage scores, as an increase in csv
0implies an increase in
wettability. With the Owens–Wendt, not only the total csv
0,
but also the polar and dispersive part is calculated
(Table 3). This subdivision is based on different interaction
forces: the dispersive component contains Van der Waals
forces while the polar component contains polar interaction
forces and lewis acid–base interactions.
The introduction of wetting envelopes is an extra tool to
predict wetting of different precursors on a specific sub-
strate. When the plot of the polar and dispersive part of the
surface tension of a liquid is enclosed in the area within the
curve, the liquid meets the requirements of the character-
istics of the specific wetting envelope. The wetting enve-
lope of a solid can be calculated using a reverse Owens–
Wendt method. Moreover, different wetting envelopes
can be obtained for a certain substrate corresponding to a
certain wetting angle: complete wetting for a h = 0�,
complete dewetting when h = 180�.
When combining the introduction of csv
0with wetting
envelopes, we can set up different envelopes (f.e. h = 0�)
for different cleaning procedures of a certain substrate.
This makes it possible to verify each cleaning procedures
rapidly. In Fig. 2, three wetting envelopes for complete
wetting (h = 0�) for Ni–W/La2Zr2O7/CeO2 substrates with
the different cleaning procedures described before are
presented. It clearly shows that the area within the curve is
the smallest for an uncleaned Ni–W/La2Zr2O7/CeO2 sub-
strate, while it is the biggest for a heat-treated substrate. As
a final proof, we plotted the surface tension subdivision of
precursor 5 onto Fig. 2. This subdivision can be made by
measuring the total clv in air and clv, polar in heptane,
Fig. 1 Different coverage scores using a dip-coating technique (gray line is upper coating limit)
Table 3 Contact angle measurements for the determination of csv
0
Treatment Contact angle versus
water (�)
Contact angle versus
diiodomethane (�)
csv
0(mN/m) csv, polar (mN/m) csv, dispersive (mN/m)
Uncleaned 37.19 ± 0.16 1.24 ± 0.05 35.95 ± 0.11
93.3 ± 2.07 44.7 ± 0.41
US cleaned 40.24 ± 0.24 4.22 ± 0.04 36.02 ± 0.18
82.3 ± 0.52 41.0 ± 0.78
Heat treated 52.76 ± 0.17 22.47 ± 0.12 30.29 ± 0.05
41.8 ± 1.07 42.5 ± 0.20
382 J Sol-Gel Sci Technol (2012) 62:378–388
123
respectively 50.0 and 32.5 mN/m, via the pendant drop
method. As clv is the sum of the polar and dispersive part,
clv, dispersive is 17.5 mN/m. It is noticed that only the heat
treated substrate give rise to complete wetting, as obtained
by the experiment. Remark that this method can be used in
an opposite approach as well, f.e. when arrays or patterns
are needed [35].
3.2 Solution
Secondly, an improvement of wettability can be obtained
by adapting the CSD precursor. A common way to lower
the surface tension is the introduction of wetting agents
such as surfactants or alcohols [36, 37]. A less understood
way of improving the wettability is located within the
precursor composition.
3.2.1 Surfactants
The influence of Rokanol IT9, C13H27O(CH2CH2O)9H,
addition to the surface tension of precursor 5, is observed in
Fig. 3. For a 10-4 M surfactant containing precursor, a
coverage score of 9 was obtained on uncleaned Ni–W/
La2Zr2O7/CeO2 and a score of 10 was obtained for both US
cleaned and heat-treated Ni–W/La2Zr2O7/CeO2. A clear
improvement in comparison with the use of precursor 5
without surfactant, where a coverage score of 2, 4
respectively 10 was obtained.
Nevertheless good coatings are observed, the amorphous
layer need to be transformed to a crystalline phase by a
specific heat-treatment profile. It was found that the addi-
tion of surfactant resulted in inhomogeneous and porous
layers. If this inhomogeneity does not affect the coating
application, surfactants can be used [38–40], if not, an
alternative needs to be found. In this work, YBCO
formation at high temperatures is affected by this inho-
mogeneity resulting in bad layers.
3.2.2 Alcohol
Firstly, ethanol, the most simple and sustainable water
soluble alcohol, was used to verify the influence on coating
quality. As can be seen from Fig. 4, the decrease of clv of
precursor 5 is observed in function of an increase in vol%
of ethanol. Nevertheless, the decrease is much slower than
in the case of a surfactant. A coating resulting from pre-
cursor 5 with the addition of 10 vol% of ethanol, resulted
in a coverage score of 6, 8 and 10 for an uncleaned, US
cleaned and respectively heat-treated Ni–W/La2Zr2O7/
CeO2 substrate. Nevertheless, during drying at 80�C for
several minutes, layer shrinkage was observed resulting in
a lower coverage score of respectively 3, 5 and 10, which is
close to the initial coverage score for precursor solutions
without the addition of alcohol. This phenomenon can be
explained by the rapid evaporation of the alcohol during
drying. The disappearance of the alcohol increases the
surface tension of the precursor again resulting in the
contraction of the liquid. It was observed that an increasing
Fig. 2 Wetting envelopes for complete wetting of different cleaning
procedures and (filled circle) surface tension subdivision plot of
precursor 5
Fig. 3 Influence of clv on the concentration of surfactant
Fig. 4 Influence of clv on the vol% ethanol
J Sol-Gel Sci Technol (2012) 62:378–388 383
123
amount of ethanol had a positive effect on the shrinkage.
Nevertheless, an amount higher than 30 vol% of ethanol
was necessary to overcome shrinkage on both uncleaned
and US cleaned Ni–W/La2Zr2O7/CeO2. Higher boiling
alcohols such as isopropanol, showed a similar effect not-
withstanding the fact that less alcohol is needed if com-
pared to lower boiling alcohols to overcome shrinkage.
3.2.3 Intrinsic factors of precursor chemistry
A third and new way to improve wettability is the adaption
of the precursors chemical formulation. Preliminary
experiments have shown that by changing the precursor
formulation a strong variability of the coating behavior
from very poor to excellent was observed. A screening
DOE approach is chosen to identify the influencing factors
out of five potential formulation composition effects: (1)
nature of the M-source, (2) type of complexing agent, (3)
pH of the solution in the range 6.5–9.5, (4) complexant
concentration and (5) base for alkalinity adjustment.
Moreover, the substrate surface treatment (US cleaning and
heat-treatment) was included in the design.
In a screening DOE each factor is tested for a linear
effect at two levels, however in this broad pH range non-
linear effects can be expected. To overcome this, the pH
range is split up in two sub-ranges: a lower pH sub-range
between 6.5 and 7.5 and a higher pH sub-range between 8.5
and 9.5. For each pH sub-range a screening DOE is
launched.
A. Input variables The factor levels in the two screening
DOE’s are the following:
1. M-source: M-nitrates/M-acetates
2. Complexant: EDTA/CA
3. pH: screening DOE 1: Lower pH range : 6.5–7.5;
screening DOE 2: Higher pH range: 8.5–9.5
4. Complexant concentration [complexant]: High/Low
(Dependent on the complexant type)
5. Alkalinity control: Ammonia/Ethanolamine
6. Surface treatment: Heat/Ultrasonic (to verify previous
results)
Note that there is only one continuous variable (pH); the
others are categorical factors.
B. Response variables The influence of the 6 factors will
be tested on the coverage of the liquid after deposition (dip
coating) on the substrate (Score 0–10).
C. Experimental design For each pH sub-range a frac-
tional factorial 26-2 Resolution IV DOE with 16 experi-
mental runs is performed. This screening approach allows
to determine (1) the significant main effects non-con-
founded with possible interactions and (2) second order
interactions which are confounded (aliased). The aliasing
of effects is presented in Table 4. The experimental runs of
each pH-range are presented in Table 1 and 3.
D. Analysis low pH range First, a screening analysis is
performed to obtain a first idea of influencing parameters.
The results are given in Table 5. One can observe a very
significant (p value \ 0.05) effect of the factor ‘‘surface
treatment’’. The results also show two significant interac-
tion effects: ‘‘alkalinity control 9 pH’’ and ‘‘surface
treatment 9 M-source’’, however they are confounded
with other second order interactions.
Using the significant factors from the screening analysis
above a screening DOE (Resolution IV) model for Lower
pH range can be constructed; the model is presented in
Table 6 and visualized in Fig. 5.
These results indicate that (1) a very good screening
regression model is obtained: R2 = 93 %, R2adj = 91 %
and a p value \ 0.0001, (2) a heat treatment cleaning
procedure of the substrate has a strong positive effect on
surface coverage (p value \ 0.0001), as already observed
in paragraph 3.1, (3) the choice of alkalinity controlling
agent should be taken into account as it is borderline sig-
nificant (p value = 0.0490) and (4) two interactions, which
are confounded with other interactions, are significant
(p value \ 0.05). Taking into account the alias structure of
Table 4 Confounding effectsEffects Aliases
M-source 9 complexant =Alkalinity control 9 surface treatment
M-source 9 pH low =[Complexant] 9 surface treatment
M-source 9 [complexant] =pH low 9 surface treatment
M-source 9 alkalinity control =Complexant 9 surface treatment
M-source 9 surface treatment =Complexant 9 alkalinity control = pH low 9 [complexant]
Complexant 9 pH low =[Complexant] 9 alkalinity control
Complexant 9 [complexant] =pH low 9 alkalinity control
384 J Sol-Gel Sci Technol (2012) 62:378–388
123
the interaction effects (Table 4), we can conclude that there
are 6 possible models left:
y¼6:25þ3:125x1þ0:625x2�0:875x1 �x3�1:75x2 �x4 ð1Þy¼6:25þ3:125x1þ0:625x2�0:875x1 �x3�1:75x5 �x6 ð2Þy¼6:25þ3:125x1þ0:625x2�0:875x2 �x5�1:75x2 �x4 ð3Þy¼6:25þ3:125x1þ0:625x2�0:875x2 �x5�1:75x5 �x6 ð4Þy¼6:25þ3:125x1þ0:625x2�0:875x4 �x6�1:75x2 �x4 ð5Þy¼6:25þ3:125x1þ0:625x2�0:875x4 �x6�1:75x5 �x6 ð6Þ
With y = coverage score, x1 = surface treatment, x2 =
alkalinity control, x3 = M-source, x4 = pH, x5 = complexant,
x6 = [complexant].
It is expected that the interaction surface treat-
ment 9 M-source has a significant higher probability than
his two aliases (Table 4) because of the high significance
and effect of the factor ‘‘surface treatment’’. The aliased
interactions of surface treatment 9 M-source do not have a
significant main effect. Therefore, we can state with a very
high certainty that this specific second order interaction
surface treatment 9 M-source can be included in the
model’s equation, leaving out Eqs. 3 till 6. A visual rep-
resentation of this interaction is given in Fig. 6. It is clearly
observed that the nitrate salt precursor compositions (1)
yield a significantly higher coverage consistency and (2)
coverage scores are significantly higher than acetate pre-
cursor compositions for heat treated substrates and vice
Table 5 Screening analysis of
lower pH range
* Significant effect
Table 6 Model lower pH range
* Significant factors
Summary of fit
R2 0.93
R2 Adj 0.91
Root mean square error 1.13
Mean of response 6.25
Observations 16
Source df Sum of squares Mean square F p [ F
Analysis of variance
Model 4 187 46.8 36.732 \0.0001*
Residuals 11 14 1.27
C. Total 15 201
Term Estimate SE t ratio p [ |t|
Parameter estimates
Intercept 6.25 0.28 22.16 \0.0001*
Alkalinity control 0.625 0.28 2.22 0.0490
Surface treatment 3.125 0.28 11.08 \0.0001*
M-source 9 surface treatment -0.875 0.28 -3.10 0.0101*
pH low 9 alkalinity control -1.75 0.56 -3.10 0.0101*
J Sol-Gel Sci Technol (2012) 62:378–388 385
123
versa for US cleaned substrates. This difference in wetting
behavior of M-nitrates versus M-acetates as a function of
cleaning procedure, can be explained by starting from the
subdivided csv’ . From Table 3, it is observed that a heat
treated cleaning procedure shows a much higher csv,polar
than an ultrasonic cleaned substrate, 22.47 versus 4.22 mN/
m, while a slightly lower csv,dispersive is obtained, 30.29
versus 36.02 mN/m.
To explain the coating behavior two interactions are
taking into account: (1) polar interaction forces and (2) Van
der Waals forces as those can be linked to csv,polar and
csv,dispersive respectively. For the polar interactions one can
have the interaction between substrate and (1) salts in the
bulk of the solution as well as (2) M-salt coordination
complexes. Nitrates show a higher polar interaction
because (1) in the bulk, acetates are positioned with the
apolar methyl-group focused towards the substrate instead
of a polar O atom of the nitrate and (2) acetates show a
potential for coordination with Y, Ba and Cu resulting with
a non coordinating methyl-group at the outside of the
complex [41]. For the Van der Waals forces, acetates are
slightly higher than nitrates.
As for the heat treated substrate, a high csv,polar part is
observed so the polar interaction takes the upper hand of
the weak Van der Waals forces. As a conclusion, nitrate-
based precursors show better wetting behavior on cleaned
substrates using heat treatment because (1) the polar
interactions are much larger for nitrates when compared to
acetates and (2) weak Van der Waals forces are negligible
when compared to the polar interactions. Moreover, for an
US cleaned substrate, a very low csv,polar part is observed so
the Van der Waals forces become relevant. As the Van der
Waals forces are higher for acetates than for nitrates,
acetates coat better on ultrasonic cleaned substrates. Both
examples clearly show that precursor composition in
combination with substrate preparation is a very important
issue in obtaining good coatings.
For the two other confounded interactions, alkalinity
control 9 pH and [complexant] 9 complexant, only alka-
linity control is a borderline significant main effect. Conse-
quently, these aliased interaction effect have a nearly equal
probability to occur making further interpretation of these
interaction effects impossible. Therefore a separation of both
aliases is necessary by performing a resolution V DOE.
E. Analysis high pH range A screening analysis (Table 7)
shows, similar as for the lower pH range, a very significant
(p value of 0.0092) effect of the factor ‘‘surface treatment’’.
For the higher pH-range, also pH with a p value of 0.1027
needs to be taken up in the further high pH range modeling.
From the screening there is no indication for significant
interaction effects.
With the factors surface treatment and pH, a screening
model for Higher pH range can be constructed; the model
is presented in Table 8 and visualized in Fig. 7. It is
observed that (1) a significant (p value = 0.0008)
screening regression model with a slightly lower R2 of
67 % and Radj2 of 62 % is obtained when compared to the
lower pH range, (2) as for the lower pH range again heat
treatment has a strong positive effect on the surface
coverage score (p value = 0.0007) and (3) pH has a sig-
nificant effect on the coverage score (p value = 0.0260).
Again, an intrinsic factor of the precursor chemistry, pH
for the higher pH range, shows a significant effect on the
wettability.
As no significant second order interactions are present, a
linear model with only main factors is obtained:
Fig. 5 Predicted lower pH Model plot
Fig. 6 Coverage score versus M-source by surface treatment for the
low pH range
386 J Sol-Gel Sci Technol (2012) 62:378–388
123
y ¼ �17:19þ 2:312x1 þ 2:625x4 ð7Þ
With y = coverage score, x1 = surface treatment, x2 =
alkalinity control, x3 = M-source, x4 = pH, x5 = com-
plexant, x6 = [complexant].
Since the acquisition of a quantitative model was not the
main focus, no further optimization of the design was
performed.
In order to produce high quality films, in this specific
case YBCO as final compound, the realization of a good
coating is a primary step in the optimization process.
Therefore, a good cleaning procedure whether or not
combined with an adjusted precursor composition seems to
be very effective. For YBCO production this can be
achieved by using a heat treatment cleaning procedure
for the Ni–W/La2Zr2O7/CeO2 substrate combined with
preferably metal-nitrate containing precursors. This
approach results in numerous combinations that could give
rise to high quality YBCO films, after optimizing the
subsequent thermal process.
4 Conclusion
This work gives an overview of the opportunities to
improve the coating behavior of CSD precursors. The main
focus was on water-based solutions, but is expandable to all
types of CSD precursors. Starting from the extended
Young’s equation, a separation between solid and liquid
modifications was made. Substrate cleaning procedures
form the base of solid modifications. We showed that
wetting envelopes can be used as a fast determination of
the cleaning procedure quality by the introduction of csv
0.
Liquid modifications include: (1) the addition of wetting
agents like surfactants or alcohols as commonly done and
(2) intrinsic factors, such as precursor formulation. The
Table 7 Screening analysis of
higher pH range
* Significant factors
Table 8 Model higher pH
range
* Significant factors
Summary of fit
R2 0.67
R2 Adj 0.62
Root mean square error 2.09
Mean of response 6.44
Observations 16
Source df Sum of squares Mean square F p [ F
Analysis of variance
Model 2 113 56.6 12.94 0.0008*
Residuals 13 57 4.37
C. Total 15 170
Term Estimate SE t ratio p [ |t|
Parameter estimates
Intercept -17.19 9.42 -1.82 0.0912
Surface treatment 2.312 0.52 4.42 0.0007*
pH 2.625 1.05 2.51 0.0260*
J Sol-Gel Sci Technol (2012) 62:378–388 387
123
latter was performed using a screening DOE. More spe-
cifically, it was found that intrinsic factors of precursor
composition, such as the origin of the M-source in com-
bination with the cleaning procedure of the substrate within
the low pH range and the pH itself within the high pH
range, are very important for the coating behaviour. Since
intrinsic factors show an effect on the coverage score, a
broad range of precursor compositions can give rise to
equivalent good coatings, showing that water-based
designs and CSD in general are promising technologies for
coated conductor development and coating of ceramics
layers in general.
As a final conclusion, this study revealed that the nature
of sustainable water-based CSD precursors does not ham-
per their coating ability, as long as a good cleaning pro-
cedure of the substrate, whether or not combined with a
targeted modification of the precursor is used.
References
1. Shimoda T, Matsuki Y, Furusawa M, Aoki T, Yudasaka I, Tanaka
H, Iwasawa H, Wang DH, Miyasaka M, Takeuchi Y (2006)
Nature 440:783–786
2. Schwartz RW, Schneller T, Waser R (2004) C R Chim 7:433–461
3. Hardy A, Mondelaers D, Vanhoyland G, Van Bael MK, Mullens
J, Van Poucke LC (2003) J Sol Gel Sci Technol 26:1103–1107
4. Vermeir P, Cardinael I, Schaubroeck J, Verbeken K, Baecker M,
Lommens P, Knaepen W, D’haen J, De Buysser K, Van Dries-
sche I (2010) Inorg Chem 49:4471–4477
5. Llordes A, Zalamova K, Ricart S, Palau A, Pomar A, Puig T,
Hardy A, Van Bael MK, Obradors X (2010) Chem Mater 22:
1686–1694
6. Van Driessche I, Penneman G, Bruneel E, Hoste S (2002) Pure
Appl Chem 74:2101–2109
7. Zhong XL, Wang JB, Liao M, Huang GJ, Xie SH, Zhou YC, Qiao
Y, He JP (2007) Appl Phys Lett 90:152903
8. Sakamoto W, Iwata A, Yogo T (2008) J Appl Phys 104:104106
9. Frumar M, Frumarova B, Nemec P, Wagner T, Jedelsky J,
Hrdlicka M (2006) J Non-Cryst Solids 352:544–561
10. Zheng J, Yang R, Xie L, Qu JL, Liu Y, Li XG (2010) Adv Mater
22:1451–1473
11. Yang Z, Ko C, Ramanathan S (2010) J Appl Phys 108:073708
12. Peng XL, Matthews A, Xue S (2011) J Mater Sci 46:1–37
13. Vermeir P, Cardinael I, Baecker M, Schaubroeck J, Schacht E,
Hoste S, Van Driessche I (2009) Supercond Sci Technol 22:
075009
14. Van de Velde N, Van de Vyver D, Brunkahl O, Hoste S, Bruneel
E, Van Driessche I (2010) Eur J Inorg Chem 2:233–241
15. Berber H, Sarac A, Yildirim H (2011) Prog Org Coat 71:225–233
16. Chang YM, Hu WH, Fang WB, Chen SS, Chang CT, Ching HW
(2011) J Air Waste Manage 61:35–45
17. Arin M, Lommens P, Avci N, Hopkins SC, De Buysser K,
Arabatzis IM, Fasaki I, Poelman D, Van Driessche I (2011) J Eur
Ceram Soc 31:1067–1074
18. Schoofs B, Van de Vyver D, Vermeir P, Schaubroeck J, Hoste S,
Herman G, Van Driessche I (2007) J Mater Chem 17:1714–1724
19. Schoofs B, Cloet V, Vermeir P, Schaubroeck J, Hoste S, Van
Driessche I (2006) Supercond Sci Technol 19:1178–1184
20. Van Driessche I, Penneman G, De Meyer C, Stambolova I,
Bruneel E, Hoste S (2002) Key Eng Mater 206–2:479–482
21. De Buysser K, Lommens P, De Meyer C, Bruneel E, Hoste S,
Van Driessche I (2004) Ceram. Silikaty 48:139
22. Penneman G, Van Driessche I, Bruneel E, Hoste S (2004) Key
Eng Mater 264–268:501–504
23. Goyal A, Paranthaman MP, Schoop U (2004) MRS Bull 29:
552–561
24. Cordero-Cabrera MC, Mouganie T, Glowacki BA, Baecker M,
Falter M, Holzapfel B, Engell J (2007) J Mater Sci 42:7129–7134
25. Krebs FC (2009) Sol. Energ. Mat. Sol. C. 93:394–412
26. Le HP (1998) J Imaging Sci Techn 42:49–62
27. Wang YH, Zhou WD (2010) J. Nanosci. Nanotechno. 10:
1563–1583
28. Young T (1805) Phil Trans R Lond 95:65–87
29. Degennes PG (1985) Rev Mod Phys 57:827–863
30. Chow TS (1998) J. Phys.-Condens. Mat. 10:L445–L451
31. Lubna N, Auner G, Patwa R, Herfurth H, Newaz G (2011) Appl
Surf Sci 257:4749–4753
32. Yaghoubi H, Taghavinia N, Alamdari EK (2010) Surf Coat Tech
204:1562–1568
33. Petersson L, Meier P, Kornmann X, Hillborg H (2011) J Phys D
Appl Phys 44:034011
34. Owens DK, Wendt RC (1969) J Appl Polym Sci 13:1741–1747
35. Lejeune M, Chartier T, Dossou-Yovo C, Noguera R (2009) J Eur
Ceram Soc 29:905–911
36. Zdziennicka A, Janczuk B (2010) J Colloid Interf Sci 343:
594–601
37. He G, Cai JH, Ni G (2008) Mater Chem Phys 110:110–114
38. Yang JX, Peterlik H, Lomoschitz M, Schubert U (2010) J Non-
Cryst Solids 356:25–27
39. Ban T, Tanaka Y, Ohya Y (2011) Thin Solid Films 519:
3468–3474
40. Han C, Pelaez M, Likodimos V, Kontos AG, Falaras P, O’Shea
K, Dionysiou DD (2011) Appl Catal B Environ 107:77–87
41. Martell AE, Smith RM (1977) Critical stability constants—vol-
ume 3: other organic ligands. Plenum Press, New York
Fig. 7 Predicted higher pH Model plot
388 J Sol-Gel Sci Technol (2012) 62:378–388
123