J. Phys. Chem.

C 2010, 114, 7637–7645 7637

Preparation of Sol-Gel Films by Dip-Coating in Extreme Conditions

Marco Faustini,† Benjamin Louis,† Pierre A. Albouy,‡ Monika Kuemmel,† and David Grosso*,†
Chimie de la Matière Condensée de Paris, UMR UPMC-CNRS 7574, UniVersité Pierre et Marie Curie
(Paris 6), Collège de France, 11 place Marcelin Berthelot, 75231 Paris, France, and
Laboratoire de Physique des Solides, UniVersité Paris-Sud, 91405 Orsay, France
ReceiVed: December 3, 2009; ReVised Manuscript ReceiVed: February 2, 2010

Dip-coating of sol-gel solutions is a complex dynamic process that is difficult to model because it is associated
with time-dependent evaporation-induced concentration and viscosity gradients in the solution. It is, however,
highly used in the coating technology because it is simple and provides excellent reproducibility. Existing
fair models have been proposed some decades ago to describe this method, but they are based on Newtonian
and nonevaporating liquids and require several important assumptions and simplifications. In this work, we
present a simple experimental study of sol-gel film formation by dip-coating, through which we propose a
general semiexperimental model to predict the final film thickness. Spectroscopic ellipsometry was used as
the main technique to obtain the film physical thickness and optical density for various dip-coating processing
conditions (withdrawal speeds from 0.01 to 20 mm · s-1 and temperatures from 25 to 60 °C) and for several
different chemical solutions (TiCl4, TEOS, and MTEOS, all in the presence, or not, of block PEO-b-PPO
copolymer templates in EtOH/H2O, with concentrations from 10-1 to 10-3 mol · L-1). We show that phenomena
that are difficult to assess during deposition, such as viscosity variation, evaporation cooling, chemical reaction,
and thermal Marangoni flow, may not have to be taken into account. The influences of various experimental
parameters are discussed together with the limitations and the full potentiality of the dip-coating technique.
We show that two regimes of film formation independently exist at extreme withdrawal speeds, while they
combine into a third regime at intermediate speeds. Although the first regime is well-known and is governed
by gravity-induced viscous drag at higher speeds, the second one is barely used and is governed by
interdependent evaporation and capillarity rise at lower speeds. We show that both regimes can be selected
to build up films with a tunable thickness and that a minimum thickness exists for each given solution at a
critical speed for which we believe that the capillarity rise effect perfectly counterbalances the viscous drag.
We also show that the capillarity regime is well-suited when one needs to deposit thick films from highly
diluted solutions.

Introduction and is proportional to the withdrawal speed at the power of

2/3. Some studies were then conducted to improve this model,2-4
Liquid deposition processes are very versatile methods to but only a few describe the influence of evaporation using
produce homogeneous coatings through the spreading of a simplified systems.5-7 However, most of the solutions used in
solution onto a substrate and evaporation of volatile compounds dip-coating are non-Newtonian fluids, especially the sol-gel
(solvent). Some related techniques, such as dip-coating or spin- solutions for which the solvent evaporates, simultaneously
coating, are commonly used in electronics and optics industries, inducing a modification of the viscosity, the density, and the
for instance, to prepare inorganic layers from sol-gel solutions. surface tension and also often triggering polycondensation of
Each technique has advantages and drawbacks. As an example, the inorganic species through pH change and spatial gathering
dip-coating (see Figure 1), for which the substrate is successively of the precursors. Kinetics of evaporation of solute/solvent
dipped into the solution and withdrawn at a constant speed, systems at the meniscus has been modeled,6,7 and the various
offers a good control of the thickness and produces no waste. phenomena taking place during dip-coating have been accurately
However, spin-coating is preferred if one wants only one face described and explained by Scherer and Brinker more that 20
to be deposited. Both methods are restricted to flat or simple- years ago.8 In this process, the thickness is tuned not only
shaped substrates. Some models have been proposed to describe through the adjustment of the withdrawal speed but also by the
the formation of a homogeneous fluid layer by dip-coating. The concentration of the solution, knowing that increasing the speed
most referred one is the Landau-Levich model for which the or the solution concentration, or both, leads to thicker films. In
thickness of a Newtonian and nonevaporating fluid on the sur- most cases, the deposition is performed with solutions loaded
face of the substrate is described by the equilibrium between with around 5 to 10% in weight of the nonvolatile species and
the adhesion of the fluid on the substrate and gravity-induced at speeds ranging from 1 to 10 mm · s-1. The thickness typically
viscous drag.1 It predicts that the equilibrium thickness depends is between 50 and 500 nm. Appropriate solvents are alcohols
on the density, the surface tension, and the viscosity of the fluid because they have low surface tension, so as to prevent
dewetting, and are fairly volatile to promote relatively fast
* To whom correspondence should be addressed. E-mail: david.grosso@ evaporation. Both properties are necessary to obtain highly
upmc.fr. Tel: 33 (0) 1 44 27 15 30. Fax: 33 (0) 1 44 27 15 04 (14 43).
†
Université Pierre et Marie Curie. homogeneous films. In sol-gels, the basic chemistry involves
‡
Université Paris-Sud. hydrolysis of the inorganic molecular precursor and polycon-
10.1021/jp9114755  2010 American Chemical Society
Published on Web 04/02/2010
7638 J. Phys. Chem. C, Vol. 114, No. 17, 2010 Faustini et al.

Figure 1. Schemes of the dip-coating method and both capillarity and draining regimes involved at low and fast withdrawal speeds, respectively.

densation into extended solids. The first reaction needs to be speeds (disregarding all evaporation-dependent quantitative
initiated for the second one to start. Both reactions start thus in parameters and examining only the slope). In addition, a second
the solution, run during deposition along the evaporation process, regime of film formation that is governed by capillarity feeding
and are generally completed on the dry film during a final and for which the thickness decreases with speed was evidenced
thermal treatment.8 It is, therefore, important to tune the kinetics at lower speeds. The latter regime was called the capillary/
of both reactions because it governs the degree of condensation, evaporation regime for which the thickness was found to be
the density, and the structure of the species assembling to form inversely proportional to the speed. For intermediate speeds, a
the final film upon evaporation. It is not unusual to end up with mixed draining/capillary/evaporation regime was found, for
porous films where the porosity is controlled by the latter which the thickness was perfectly described by the sum of the
chemical and processing conditions. The Landau-Levich model contribution of both previous regimes. These regimes are
is thus barely verified with these complex systems,7 especially illustrated in Figure 1. Interestingly, when both regimes perfectly
in extreme conditions (ultralow or -fast speeds, high evaporation counterbalance each other, a minimum of thickness is experi-
rate, high solution concentration, highly reactive species...) at mentally found, which corresponds to the critical speed uC found
which additional phenomena can take place. Indeed, at ultralow around 0.2-0.8 mm · s-1, depending on the system. Such uC
speed, the capillary rise of the solution at the substrate surface values are independent of the solution concentration but increase
is strongly opposing the draining, whereas at high temperatures, with the solvent evaporation rate that was modulated by the
the rapid evaporation of the solvent creates a fast increase in environmental temperature. Semiexperimental equations linking
viscosity. The capillary rise of the solution is analogous to the the expected thickness versus the solution composition and the
“pinning” phenomena known as the coffee-ring effect.9,10 It has speed of withdrawal were established without having to take
been recently demonstrated that, for the “coffee-ring” effect, into account involved important phenomena that are difficult
the evaporation rate is linear. The study was performed by in to assess, such as viscosity variation, evaporation cooling, and
situ FTIR analysis during evaporation of an aqueous/methylene thermal Marangoni flow. The equation linking the thickness to
blue sessile droplet deposited onto a surface.11 The so-called the speed was derived to mathematically determine uC, which
“convective deposition” has been used to deposit and self- matches with experimental results. Relations between the critical
arrange colloids12-14 or nonvolatile molecules15 on surfaces. The speed uC, the minimal thickness h0,min, the solution composition,
description of this capillary-driven convective deposition process and the temperature are discussed. The influence of speed and
was made and compared to the Landau-Levich regime of thickness on the refractive index is also presented. The last point
deposition for phospholipidic solutions on horizontal sub- that will be underlined here is that thicker films can be built up
strates.16 However, no study was dedicated to the capillary effect at lower speeds (in the capillary/evaporation regime) than at
during dip-coating of complex reactive sol-gel solutions. In higher speeds with equivalent optical quality, which is useful
the present work, we performed an experimental investigation when species to deposit can be stabilized only in highly diluted
of dip-coating of various sol-gel solutions between extreme solutions.
withdrawal speed conditions (from 0.01 to 20 mm · s-1). The
Experimental Section
thickness and refractive index of the final stabilized films were
measured by ellipsometry and were used to assess the quantity Initial Solutions. Absolute ethanol was purchased from
of material deposited per unit surface. From mass conservation Normapur, and inorganic precursors (TiCl4, TEOS (tetraethyl
law, we deduced the solution consuming rate, associated with orthosilicate) and MTEOS (methyltriethyl orthosilicate)) were
the rate of evaporation, responsible for the formation of the solid purchased from Aldrich. F127 Pluronic (EO106-PO70-EO106) and
films. We also show that several distinct regimes of deposition PB-b-PEO (polybutadiene-b-polyethyleneoxide, P3017-BdEO;
exist for different ranges of speed. The draining/evaporation MWPB ) 5500 g mol-1, MWPEO ) 5000 g mol-1) were
regime, described by the Landau-Levich model and for which purchased from Aldrich and Polymer Source, respectively. Each
the thickness increases with the speed, was verified for higher solution was prepared following a specific protocol. BTi
Sol-Gel Dip-Coating in Extreme Conditions J. Phys. Chem. C, Vol. 114, No. 17, 2010 7639

TABLE 1: Solution Composition and Film Thermal Treatments

solution composition in molar ratio treatment
samples TEOS MTEOS TiCl4 F127 PB-PEO HCl H 2O EtOH 5 min at
F-L 0.006 40
F-H 0.006 20
Si-L 1 0.005 5 40 400 °C
Si-H 1 0.005 5 20 400 °C
FSi-L 1 0.006 0.005 5 40 400 °C
FTi-L 1 0.005 10 40 500 °C
TiSi-L 0.112 1 4.4 145 200 °C
TiSi-H 0.112 1 4.4 72 200 °C
BTi-L 1 0.004 42 320 500 °C
BTi-H 1 0.004 42 160 500 °C
FMSi-L 0.5 0.5 0.006 0.005 5 80 450 °C
FMSi-H 0.5 0.5 0.006 0.005 5 40 450 °C

solutions were obtained by mixing two solutions containing part the name of the solution, followed by L or H for low or high
of EtOH, H2O, and PB-b-PEO (aged 2 h at 70 °C) for the first concentrated solution, respectively, and, finally, with the applied
one and TiCl4 dissolved in the remaining part of EtOH for the temperature of evaporation in °C. In the solution names, F, B,
second one.17 FTi and FSi solutions were obtained by dissolving Si, Ti, and M stand for Pluronic F127, polybutadiene-co-
the precursors (TiCl4 or TEOS) in ethanol, water, and hydro- polyethylene oxide, TEOS, TiCl4, and MTEOS, respectively,
chloric acid for FSi only (HCl 2M), followed by the addition that enter the solution composition in addition to ethanol, water,
of F127. TiSi, Si, and FMS solutions were prepared following and, potentially, hydrochloric acid.
the same protocol with the appropriate precursor composition. Characterization. The thickness (h0) and refractive index
F solutions only contain F127 dissolved in EtOH. Silica- (n0) of optical transparent films were assessed using the
containing solutions were stirred for at least 12 h at room spectroscopic ellipsometry investigation (variable-angle spec-
temperature before deposition to ensure complete dissolution troscopic ellipsometry (VASE), M-2000U Woollam spectro-
of each component and partial hydrolysis of the inorganic scopic ellipsometer) in the visible range and at an incidence
precursors. The molar compositions of the final solutions are angle of 70°. To avoid possible film pollution due to extended
reported in Table 1. exposure to the environment, measurements were quickly
Film Preparation. Films were deposited on silicon wafer operated after thermal treatment and at a relative humidity below
pieces (1 × 7 cm) washed with acetone. Substrates were 5% (variable-humidity flow chamber SOPRA) so as to ensure
deposited by dip-coating at relative low humidity using the that the porosity was not filled up with condensed water. Data
previous solutions. Before deposition, the atmospheric chamber analysis was conducted with the WVase32 software. Ellipso-
was flux with dry air so as to decrease as much as possible the metric Ψ and ∆ plots were fitted using a single Cauchy layer
humidity. Films were deposited at withdrawal speeds ranging model, for which asymmetric optical properties emanating from
from 0.01 to 20 mm · s-1 and at three different temperatures (e.g., unidirectional contraction were not observed. All analyzed films
25, 40, and 60 °C). The temperature was adjusted by heating were of optical quality (no diffusion) and exhibited no density
the outside of the whole stainless steel chamber with hot air. gradients in the thickness profile. Optical characteristics of films,
Both hot and dry flows were stopped before deposition to which were cracked or delaminated because of their too high
prevent convection. The whole dip-coater was isolated from thickness or density, could not be analyzed and are, therefore,
natural vibration and external convection so as to prevent the not reported in the plots. Errors made on the determination of
formation of horizontal strips (thickness nonuniformity) associ- the refractive index and the thickness were measured to be less
ated with the fluctuation of the solution surface that creates than 5%.
meniscus instability. In the present investigation, the same Data Treatment. “Fresh” sol-gel films are highly sensible
container with an aperture gap of 2 mm was used to make sure to atmospheric conditions, and their thickness can instantly and
the meniscus shape was similar for each experiment and also considerably be modified by a slight change in temperature and
to diminish the potential fluctuation of the meniscus height humidity due to equilibration with the external relative humidity.
during deposition due to vibration-induced surface instability. In addition, they are unstable because they undergo drying,
These controls are necessary for ultralow withdrawal speeds. condensation, and shrinkage upon aging. As a result, final
During the whole deposition process, the vapor pressures of thicknesses (h0) and refractive indices (n0) were measured after
EtOH and water inside the chamber were not measured but were stabilization of the films by thermal treatment. An additional
considered to be constant because they self-equilibrate through reason for such a choice is that, after thermal stabilization, the
the constant evaporation of the solution from the reservoir films are composed of metal oxide of known density, molar
combined with the constant exchange with the outer atmosphere weight, and refractive index, required for the modeling of the
through small evacuation holes present on the chamber walls. optical response. Because the final xerogel films also contain a
We estimated that the ethanol vapor pressure in the chamber significant fraction of porosity, the volume fraction of solid
was slightly below saturation; however, the evaporation rate material can be deduced from the refractive index by effective
was increased for higher temperatures due to the associated medium approximation laws using optical constants of both
increase of the saturated vapor pressure. After complete drying, dense materials and voids.18 In our case, the refractive index of
films were directly transferred underneath a curing IR lamp the final films prepared from each independent solution does
(except the polymeric F127 films) to be thermally treated not vary significantly (maximum deviation of 5%) with the
between 200 and 500 °C to decompose the organic parts and to applied processing conditions, two different concentrations, and
complete the inorganic condensation. Conditions for thermal for three different temperatures. We thus decided to use a simple
treatment are given in Table 1. Samples are labeled with, first, linear dependence of the global refractive index with the volume
7640 J. Phys. Chem. C, Vol. 114, No. 17, 2010 Faustini et al.

TABLE 2: Symbols Used in Equations and Their Corresponding Meaning and Units
meaning units meaning units
h0 final film thickness nm hS thickness before evaporation nm
n0 final film refractive index none ηS solution dynamic viscosity Pa · s
ni material refractive index none γS solution surface tension J · m-2
Ri fraction of material in the film % FS solution density g · cm-3
u withdrawal speed mm · s-1 D global constant m1/3 · s2/3
F rate of film formation m3 · s-1 h0,min minimal final thickness nm
ci inorganic precursor concentration mol · cm-3 uC critical speed mm · s-1
Mi inorganic material molar weight g · mol-1 Vf final film volume m3
Fi density of the inorganic material g · cm-3 l film height m
E rate of solution evaporation m3 · s-1 L film width m
ki material proportion constant none t time s

fraction of both pore (n ) 1) and bulky dense reference materials Model for the Capillarity Regime. At low withdrawal
(ni) to deduce the proportion of solid material (Ri) through speeds, solvent evaporation becomes faster than the motion of
relation 1 (see Table 2 for a listing of the symbols used in the drying line (vapor-liquid-solid, three-phases frontier),
equations in this paper and their corresponding meaning and which leads to the continuous feeding of the upper part of the
units). The refractive index n0 and physical thickness h0 of the meniscus by the solution through capillarity rise. Here, as
final film, after thermal stabilization, were measured for each demonstrated for the pinning effect,11 we assume that the
system for withdrawal speeds ranging from 0.01 to 20 mm · s-1 evaporation rate is constant and we can thus apply the mass
for conservation law that tells that the rate of film volume formation
(F) can be related to the rate of solution feeding the upper zone
n0 - 1 of the meniscus, which is directly related to the rate of solution
Ri ) (1)
ni - 1 undergoing evaporation (E).16 If ci is the solution inorganic molar
concentration and Ri the volume fraction of inorganic material
A typical evolution of the physical thickness (h0) with the in the final film, deduced from the refractive index, one can
withdrawal speed (u) is given in Figure 2 for the FMS-H-25 write
systems. The plot exhibits two distinct tendencies. At low
speeds, the thickness decreases with u, whereas the tendency is
inverted at high speeds. At intermediate speeds, a minimum is ciMi
found that corresponds to a critical speed addressed uC. F) E ) kiE (2)
Intuitively, one attributes naturally the low-u regime to a RiFi
combined effect between continuous capillary feeding and
evaporation (capillarity regime), recently reported by Le Berre F and E are in m3 · s-1, whereas Mi and Fi are the molar weight
et al. for the horizontal knife coating of proteins. The high-u and the density of the thermally stabilized inorganic material,
regime corresponds to a combined effect between gravity-driven respectively. ki is the material proportion constant for each
viscous drag and evaporation (draining regime). Both regimes solution. The rate of film formation F corresponds to the
are illustrated in Figure 1. variation of film volume with time, which leads to

Figure 2. (a) Plot of the thickness versus withdrawal speed (log-log scale) for the FMSi-H-25 system experimental points and corresponding
models for both independent (dashed line) and combined (solid line) capillarity and draining regimes of film formation. (b) h0/ki versus 1/u plot
(experimental points and corresponding models). (c) D versus u plot (log-log scale).
Sol-Gel Dip-Coating in Extreme Conditions J. Phys. Chem. C, Vol. 114, No. 17, 2010 7641

dVf h0Ldl ciMi h0 ) kiDu2/3 (6)

) ) E (3)
dt dt RiFi For the Landau-Levich-based model to be verified, D must be
with Vf, h0, l, and L being the volume, the thickness, the height, constant, whatever the value of u. D can be calculated from
and the width of the considered final film, respectively. Because experimental data because one knows ki and h0 for each speed.
L is constant and l varies with the withdrawal speed u ) dl/dt Figure 2c shows the evolution of D with u, and one observes
and thus the time t, eq 3 can be transformed into eq 4 that D is roughly constant for high values of u, where the
draining regime dominates, as expected. However, a slight
ciMi E 1 decrease of D is noticed for the highest values of u. Such
E1
h0 ) ) ki (4) deviation is attributed to the fact that, in these extreme
RiFi L u Lu conditions, the thickness of the deposited solution is too high
because the refractive index of the final films does not vary such that the gravity-induced viscous drag cannot be counterbal-
significantly with the withdrawal speed (see further below). The anced by the adhesion of the layer to the surface.7 The solution
h0/ki versus 1/u plot gives a straight line when the capillarity/ layer is thus thinning more than what can be predicted from eq
evaporation regime is dominating the formation of the film. The 6. Selecting a range of u for which D is rather constant, and
slope corresponds to the solution consuming rate E/L (in averaging on this range, we found a D value of 5.8 × 10-4
m2 · s-1). The latter value is thus deduced experimentally and m1/3 · s2/3 for the FMS-H-25 systems in Figure 2. D can then be
can then be reintroduced in eq 4 to predict the thickness with reintroduced in eq 6 to predict the thickness with respect to the
respect to withdrawal rate in the capillarity regime. For the FMS- withdrawal rate in the draining regime.
H-25 systems in Figure 2, E/L was found to be 2.3 × 10-10 Combining Models to Describe Simultaneously Both
m2 · s-1, when using the physical constant reported in Table 3. Regimes. At intermediate u values, both regimes of film
The latter tendency confirms that the thickness proportionally formation are overlapping and must be taken into account. We
varies with u at the power of -1 in these dip-coating conditions, found that, summing both contributions (eqs 4 and 6 into eq
whereas Le Berre et al. found a power of -1.1 with their 7), allows a fairly good description of the final thickness with
horizontal setup. The reason for such a difference is not yet respect to the speed, suggesting that each regime does not
fully understood. influence significantly the other one when overlapping. As a
Model for the Draining Regime. For high-speed values, the result of the interdependence of the viscosity and the evaporation
formation of film is usually described by the Landau-Levich rate (evaporation induces the increases of viscosity while the
equation (eq 5) that allows one to predict the thickness (hS) of increase of viscosity provokes the decrease of the evaporation
a nonevaporating Newtonian fluid deposited by dip-coating with rate) that are the critical parameters in the viscous drag and
respect to the fluid physical-chemical properties and to the capillarity regimes, respectively, a continuous and progressive
deposition processing conditions, such as speed and temperature. evolution of the thickness with speed for each independent
This equation was established considering that the fluid thickness regime is observed. It is thus likely that both regimes overlap
is homogeneous all over the substrate surface if the speed is spatially in the coexistence region. An additional in situ
constant, which is due to a delicate balance between the viscous investigation has to be done to better understand the phenomena
drag and the wetting property (adhesion) of the fluid with the taking place in this regime of coexistence.
substrate surface.1

0.94ηs2/3
h0 ) ki ( LuE + Du )
2/3
(7)

hs ) u2/3 (5) As observed for the FMS-H-25 system in Figure 2, the global
γs1/6(Fsg)1/2 evolution of the thickness described by this equation perfectly
matches the experimental points, especially at intermediate u
ηS, γS, and FS are the viscosity, the surface tension, and the values. Because the thickness reaches a minimum value h0,min
density of the fluid, respectively, while g is the standard gravity. at the intermediate critical speed uC, the derivative of eq 7 must
In the present case, the fluid is the initial solution. For the then be null for u ) uC. uC can thus be calculated using relation
modeling, we consider that the solution behaves as a Newtonian 8 for which (dh0/du) ) 0.
fluid (which is an approximation because the viscosity and
-3/5
surface tension change with evaporation that simultaneously
induces a concentration increase and condensation). The evapo-
uC ) ( 2DL
3E )
(8)
ration is taken into account by introducing the material For FMS-H-25, uC was calculated to be 0.186 mm · s-1, which is
proportion constant ki in the equation while the physicochemical in good agreement with experimental data (0.15 < uC experimental <
constants of the solution are gathered into a global constant D 0.20 mm · s-1). In addition, calculated and experimental h0,min values
such that eq 5, describing the fluid equilibrium thickness, for uC are also in good agreement (h0,min ≈ 100 nm).
becomes eq 6, describing the final film thickness, disregarding Generalization to Other Sol-Gel Systems. The previous
the mentioned evaporation-dependent parameters. complete analyses for FMS-H-25 films were performed on six
other classical sol-gel systems with solution compositions
TABLE 3: Molecular Weight (Mi/g · mol-1), Density (Gi/ reported in Table 1. These solutions were chosen so as to
g · cm-3), and Bulk Refractive Indices (ni/at λ ) 700 nm) confirm that the latter fitting equations are verified for all types
Used to Calculate ki for Each Solutiona of sol-gel film formation by dip-coating. Pure silica (SiO2),19
or mixed oxides (SiO2/TiO2),20 hydride silica (Me functionalized
solution F Si FSi FTi TiSi BTi FMSi
SiO2),21 mesoporous silica and TiO2 (templated with two dif-
Mi 12 600 60 60 80 78 80 58 ferent block copolymers) films22 were studied in addition to a
Fi 1.5 2.4 2.4 4 3.7 4 2.4 pure polymer non-sol-gel film. They cover a wide range of
ni 1.6 1.5 1.5 2.5 2.2 2.5 1.5
refractive indices (e.g., from 1.15 for FMS to 2.2 for TiSi) and
a
These values correspond to pure bulky oxide final materials. can find applications in optics, miocroelectronics, energy,
7642 J. Phys. Chem. C, Vol. 114, No. 17, 2010 Faustini et al.

Figure 3. Relative refractive indices (ni/ni average) measured by ellipsometry for different solutions and withdrawal speeds.

sensing, protection, and separation domains. The films’ structural

characteristics (refractive index, porosity, and pore organization
for mesoporous films) seem not to change dramatically with
the thickness and the regime they have been constructed with.
All measured relative refractive indices (ni/ni average) versus speed
have been reported in Figure 3 for films obtained for one
concentration (L or H) and at 25 °C. A maximum deviation of
7% was measured, but most of the points are centered around
1 ( 0.02. The very slight deviation from the average refractive
index value suggests that densification of sol-gel films is not
significantly dependent on the deposition regime, even if a slight
tendency to form denser films at low speeds (capillarity regime)
may be noticed. Indeed it would have been expected to have
denser films in the latter regime because more time is given to
the inorganic species to arrange into a compact structure through
capillary-induced Laplace forces (these forces are created during
solvent evaporation from the pores and drives the entities to Figure 4. SEM image of the surface of a BTi-L-60 film, performed
compact through weak interactions). However, in sol-gel in the capillarity regime of u (u ) 0.1 mm · s-1) and after thermal
materials, most of the densification is dominated by the final treatment at 500 °C.
thermal treatment, where strong chemical bonds are formed
between inorganic entities. Denser films may be encountered characteristic independent regimes of formation. Values of ki
when packing hard colloidal entities for which no thermal were calculated from the solution composition, and the values
treatment is required. Figure 4 shows the SEM image of the of E/L, D, uC, and hmin were deduced for each corresponding
surface of a BTi-L-60 film, performed in the capillarity regime system and are reported in Table 4. In Figure 5, one can clearly
of u (u ) 0.1 mm · s-1) and after thermal treatment at 500 °C. see that, for both regimes, the tendency lines are parallel one
One can clearly see the well-organized porosity created by the to the other, suggesting that the model corresponding to eq 7 is
presence of the block copolymer micellar templates and the verified for each system. Only the values of E/L, D, uC, and
small TiO2 nanoparticle composing the solid framework. This hmin change (see Table 4). Values of E/L are found to be between
type of structure is conventionally observed when the same 10-9 and 2 ×10-10 m2 · s-1, whereas values of D vary between
solution is deposited in the draining regime,23 suggesting that 10-4 and 6 ×10-4 m1/3 · s2/3. These differences are related to the
film structure is not significantly modified by the regime of difficulty in assessing parameters, such as viscosity, inorganic
formation. Final film thickness (h0) versus withdrawal speed precursor, concentration, and reactivity, that are different for
(u) plots are reported for all systems formed at 25 °C in Figure each system and evolve differently with evaporation. This is
5 in log-log coordinates. They all show the typical behavior the reason why no evident correlation with ki could be found.
described above for the FMS-H-25 systems, with the two However, these values are of the same order of magnitude,
Sol-Gel Dip-Coating in Extreme Conditions J. Phys. Chem. C, Vol. 114, No. 17, 2010 7643

Figure 5. Evolution of thickness (h0) versus withdrawal speed (u) for all systems formed at 25 °C. Dashed lines correspond to fits associated with
eq 7.

TABLE 4: Solution Concentration in Inorganic Species, ci (mol · cm-3), Proportion of Solid Materials Inside the Film, ri (%),
Material Proportion Constant, ki (unitless), Solution Consuming Rate, E/L (m2 · s-1), Draining Regime Constant, D (m1/3 · s2/3),
Critical Withdrawal Speed, uC (mm · s-1), Theoretical Minimum Thickness, hmin The (nm), and Experimental Minimum
Thickness, hmin Exp (nm) for Each System
samples ci Ri ki E/L D uC hmin The hmin Exp
-6 -2 -10 -4
F H 25 5.1 × 10 91 4.7 × 10 8.3 × 10 4.0 × 10 0.494 201 219
F L 25 2.6 × 10-6 86 2.5 × 10-2 5.2 × 10-10 3.3 × 10-4 0.420 80 77.5
FSi L 25 3.8 × 10-4 42 2.3 × 10-2 5.4 × 10-10 3.8 × 10-4 0.398 79 84
FTi L 25 3.9 × 10-4 34 2.3 × 10-2 2.8 × 10-10 1.9 × 10-4 0.400 41 43.5
FTi L 40 3.9 × 10-4 35 2.3 × 10-2 6.3 × 10-10 1.9 × 10-4 0.648 55 51
FTi L 60 3.9 × 10-4 34 2.3 × 10-2 1.1 × 10-9 1.8 × 10-4 0.930 68 63
TiSi L 25 1.3 × 10-4 88 3.1 × 10-3 1.1 × 10-9 4.3 × 10-4 0.54 15.6 13.9
TiSi H 25 2.5 × 10-4 90 6.0 × 10-3 1.6 × 10-9 3.8 × 10-4 0.71 32 28.5
TiSi H 40 2.5 × 10-4 88 6.1 × 10-3 2.5 × 10-9 3.7 × 10-4 1.01 38 36
TiSi H 60 2.5 × 10-4 87 6.2 × 10-3 4.8 × 10-9 3.6 × 10-4 1.51 50 39
FMS L 25 2.0 × 10-4 29 1.7 × 10-2 1.8 × 10-10 3.5 × 10-4 0.216 37 31
FMS H 25 4.8 × 10-4 37 3.2 × 10-2 2.3 × 10-10 5.8 × 10-4 0.186 100 100
BTi H 25 9.8 × 10-5 55 3.6 × 10-3 7.0 × 10-10 2.6 × 10-4 0.585 11 9.9
BTi L 25 5.1 × 10-5 51 2.0 × 10-3 7.8 × 10-10 2.2 × 10-4 0.680 5.8 4.2
BTi L 40 5.1 × 10-5 56 1.8 × 10-3 1.6 × 10-9 2.2 × 10-4 1.04 7 6.2
BTi L 60 5.1 × 10-5 59 1.7 × 10-3 3.3 × 10-9 2.6 × 10-4 1.470 9.8 8.8
Si L 25 3.7 × 10-4 87 1.1 × 10-2 3.4 × 10-10 1.3 × 10-4 0.561 16.9 15
Si H 25 6.7 × 10-4 87 1.9 × 10-2 3.5 × 10-10 1.4 × 10-4 0.570 31 29

which is due to the common solvent (EtOH/H2O) that dominates of (h0 High conc/h0 Low conc). Knowing that the corresponding ratio
the evaporation rate and the viscous drag. Deviation from the of (ki High conc/ki Low conc) is 1.9, one expects the ratio of thickness
model at highest speeds, as already observed in Figure 2, is to be close to this value, but higher values are found. Concerning
also present for each system with different limits and intensities. this difference, several tendencies are observed: at low u in the
As expected, systems with high values of ki provide thicker capillarity regime, the (h0 High conc/h0 Low conc) ratio is increasing
films. A rather good linear dependence of h0,min with ki is found. from 2.4 to 3, whereas in the draining regime at high u values,
The influence of inorganic concentration has been investigated it increases even more up to 4.5. In addition to the concentration
for F, FMS, BTi, and TiSi systems (see Table 4). Figure 6 shows effect, one must, here, take into account the viscosity that is
the h0 versus u typical plots for two concentrations of the FMS the main factor for viscous drag that governs the thickness of
system at 25 °C (H for high concentration of inorganic (ci ) the solution and thus the thickness of the final film in the
4.8 × 10-4 mol · cm-3) and L for low concentration (ci ) 2 × draining regime only. More concentrated solutions are also more
10-4 mol · cm-3)). As expected, higher concentrations give higher viscous due to higher solute content (quantifying the viscosity
thicknesses, whatever u. The same behavior is found with the of both solutions during evaporation is extremely difficult in
other systems. The inset graphic (Figure 6) represents the ratio such systems due to concentration gradients that are formed
7644 J. Phys. Chem. C, Vol. 114, No. 17, 2010 Faustini et al.

Figure 6. Thickness versus withdrawal speed (log scale) for FMSi systems for two different concentrations. Inset: evolution of the thickness ratio
(h0 High conc/h0 Low conc) with u.

of the thickness with temperature in the draining regime suggests

that evaporation is slower than the time needed for the solution
layer to reach the equilibrium thickness for which viscous drag
has no more effect. It thus has a minor effect on the thickness.
The D value is expected to vary for Newtonian and nonvolatile
fluids. However, in the sol-gel film cases, the layer of solution
is simultaneously submitted to viscous drag and evaporation
that are interdependent. A temperature rise leads to less viscous
fluid and thus a thinner layer. On the other hand, a temperature
rise leads also to higher solvent volatility (lower surface tension)
and faster evaporation, which thus allow less time for the fluid
layer to be submitted to the viscous drag, which, in that case,
leads to thicker films. Both opposite consequences may perfectly
Figure 7. Thickness versus withdrawal speed (log-log scale) for BTi
counterbalance the effect of the other (viscosity and surface
systems for three different temperatures. tension decrease, due to temperature increase, have an opposite
effect on the D value; see eq 5), and this is the reason why one
does not observe a significant change in the D value with
between both interfaces and that are due to evaporation at the
temperature (see Table 4). Once more, such a dependency could
solution/air interface only). The deviation is thus greater for
not be quantified here as a result of the continuous and
high withdrawal speeds for which the viscosity has a more
progressive variation of viscosity and surface tension during
pronounced effect. For the highest values of u, the thickness
evaporation time. In the capillarity regime, the film formation
ratio has an inverse tendency. We already noticed that there
was a deviation from the model for the highest u values for is governed by the solution capillary raising. The faster the
which the solution is too thick for the viscous flow to be evaporation is, the faster the solution feeds the upper part of
counterbalanced by the adhesion. The thickness of the deposited the drying line. This assumption is corroborated by the
solution is thus greater for higher viscosities; the latter deviation experimental value of the rate of solution consuming (E/L) that
must then start at lower u values for more concentrated solutions, is directly related to the rate of evaporation. Indeed, for each
which leads to the observed reverse tendency of the thickness system, E/L significantly increases with temperature (see Table
ratio. 4). The fact that the plots of thickness dependencies with u are
The influence of the chamber temperature during deposition parallel one to the other in the capillarity regime confirms that
was investigated on FTi, SiTi, and BTi systems (see Table 4 the model applies for any evaporation rate. The evolution of
for corresponding data). Figure 7 shows the typical evolution the solution consuming rate (evaporation rate) versus temper-
of thicknesses with withdrawal speed for the BTi system for ature is shown in Figure 8b. Figure 8a,c displays the evolution
25, 40, and 60 °C. One observes that the atmosphere temperature of h0,min and uC with temperature for the three studied systems.
does not influence considerably the thickness in the draining They all show the same tendencies. The minimum thickness
regime, whereas thicknesses are significantly greater for higher increases with temperature, which is expected because the film
temperatures in the capillarity regime. In the draining regime, formation in such critical conditions is partly due to capillarity
viscosity and surface tension are temperature-dependent and are feeding. The critical speed (uC) also increases with temperature,
parameters that govern the film thickness. They, however, seem which suggests that the evaporation dominates the formation
not to be sufficiently modified so as to influence the final of the film. The capillarity regime has thus a greater influence
thickness. In any case, evaporation is faster at high temperature at high temperatures than at low ones, and its limit of domination
as a result of higher saturated vapor pressure. The invariability is shifted toward higher speed. Interestingly uC(T)/uC(25 °C) versus
Sol-Gel Dip-Coating in Extreme Conditions J. Phys. Chem. C, Vol. 114, No. 17, 2010 7645

Figure 8. Evolution of (a) the minimal thickness (hmin), (b) the solution consuming rate (E/L), and (c) the relative critical withdrawal speed
(uC(T)/uC(25 °C)) with temperature for FTi-L, BTi-L, and TiSi-H systems. Dashed lines are added as guides to the eye.

T has a similar slope for the three systems, which suggests that Acknowledgment. The authors would like to thank the
the temperature effect is similar, whatever the system, and that French CNRS, The University Pierre et Marie Curie (Paris VI),
only the evaporation rate is influenced by the temperature, as the European Community (Terramagstore project), and the
already mentioned. French CIFRE for funding. The LPS technical and engineering
In summary, we confirmed that two opposite regimes of film team is greatly acknowledged for its help in dip-coating
formation exist in dip-coating processing of sol-gel solutions, conception and building.
as reported for horizontal knife coating, but with different speed
References and Notes
dependencies. This semiexperimental simple approach will, with
no doubt, be very useful to predict the thickness of sol-gel (1) Landau, L.; Levich, B. Acta Physicochim. URSS 1942, 17, 42.
(2) Lee, C. Y.; Tallmadge, J. A. Ind. Eng. Chem. Fundam. 1976, 15,
films performed by dip-coating. At low withdrawal speeds, the 258.
film thickness is governed by the interdependence of continuous (3) Spiers, R. P.; Subbaraman, C. V.; Wilkinson, W. L. Chem. Eng.
evaporation of the solvent at the meniscus and the capillary rise Sci. 1974, 29, 389.
(4) Higgins, B. G.; Silliman, W. J.; Brown, R. A.; Striven, L. E. Ind.
at the drying line. At high speeds, the final thickness is mainly Eng. Chem. Fundam. 1977, 16, 393.
dependent on the viscous drag. At intermediate speeds, both (5) Schunk, P. R.; Hurd, A. J.; Brinker, C. J. Liquid Film Coating;
phenomena overlap and a critical speed, for which the film has Chapman & Hall: London, 1997; pp 673-708.
a minimal value, is found. A semiexperimental model can (6) Kuznetsov, A. V.; Xiong, M. Int. Commun. Heat Mass Transfer
2002, 29, 35.
accurately describe the latter global tendency. For a given (7) Lee, C. H.; Lu, Y. F.; Shen, A. Q. Phys. Fluids 2006, 18, 052105.
solution, a thickness lower limit exists that cannot be overcome, (8) Brinker, C. J.; Frye, G. C.; Hurd, A. J.; Ashley, C. · S. Thin Solid
except by dilution of the initial solution. On the other hand, Films 1991, 201, 97.
(9) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.;
thicker films can be obtained with highly diluted solutions at Witten, T. A. Nature 1997, 389, 827.
ultralow withdrawal speeds than at high speeds. This is (10) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.;
important when species to deposit cannot be dissolved or Witten, T. A. Phys. ReV. E 2000, 62, 756.
(11) Innocenzi, P.; Malfatti, L.; Piccinini, M.; Grosso, D.; Marcelli, A.
dispersed in high concentrations. In addition, we show that Anal. Chem. 2009, 81, 551.
increasing temperatures allow depositing thicker films in the (12) Conway, J.; Korns, H.; Fisch, M. R. Langmuir 1997, 13, 426.
capillarity regime, which might be useful when the solvent is (13) Kraus, T.; Malaquin, L.; Schmid, H.; Riess, W.; Spencer, N. D.;
poorly volatile, such as water. Finally, in the capillarity regime, Wolf, H. Nat. Nanotechnol. 2007, 2, 570.
(14) Rabani, E.; Reichman, D. R.; Geissler, P. L.; Bruss, L. E. Nature
the film formation rate is a steady state, which means that an 2003, 426, 271.
equilibrium is found between evaporation and capillary feeding. (15) Yabu, H.; Shimomura, M. AdV. Funct. Mater. 2005, 15, 575.
Because any wetting solvent will give a meniscus, a solvent (16) Le Berre, M.; Chen, Y.; Baigl, D. Langmuir 2009, 25, 2554.
(17) Kuemmel, M.; Allouche, J.; Nicole, L.; Boissière, C.; Laberty, C.;
with high surface tension, such as water, again, can be used Amenitsch, H.; Sanchez, C.; Grosso, D. Chem. Mater. 2007, 19, 3717.
without dewetting problems that are always encountered in the (18) Boissiere, C.; Grosso, D.; Lepoutre, S.; Nicole, L.; Brunet Bruneau,
draining regime. Films of optical quality are obtained in this A.; Sanchez, C. Langmuir 2005, 21, 12362.
capillarity regime because ellipsometry investigations could be (19) Brinker, C. J.; Scherer, G. W. Sol-Gel Science: The Physics and
Chemistry of Sol-Gel Processing; Academic Press: San Diego, CA, 1989.
applied to them, but one must take care to avoid any fluctuation (20) Almeida, R. M.; Christensen, E. J. Sol-Gel Sci. Technol. 1997, 8,
of the solution surface level, due to vibration or convection, 409.
for instance, to ensure thickness homogeneity. Preliminary tests (21) Itagaki, Y.; Deki, K.; Nakashima, S.-I.; Sadaoka, Y. Sens. Actuators,
B 2006, 117, 302.
revealed that silica precursors, polymers, and metallic colloids (22) Sanchez, C.; Boissiere, C.; Grosso, D.; Laberty, C.; Nicole, L. Chem.
can be deposited in the capillary regime from pure aqueous Mater. 2008, 20, 682.
solutions or poorly volatile solvents, such as chlorobenzene. The (23) Smarsly, B.; Grosso, D.; Brezesinski, T.; Boissière, C.; Pinna, N.;
Sanchez, C.; Antonietti, M. Chem. Mater. 2004, 16, 2948.
latter results will be published in a separate article because the
concept deserves a full study. JP9114755