Thickness effects of transparent thin coatings often seems to be a critical factor in the design of functional devices. These coatings should be thick enough to protect the underlying materials. On the other hand, losses in transparency can be noticed with an increased thickness, which is often detrimental for their application.
This paper looks into the field of insulating window films. These window films are designed in such a way that heat is kept outside of buildings during summer times and vice versa. This functionality is achieved by the deposition of a far infrared reflective Fabry-Pérot stack with low emissivity (ε) on polymer films. The Fabry-Pérot stack consists of transparent layers, deposited typically by subsequent sputtering. An example is a TiO2
multi-layer stack. This stack shows excellent low-ε properties [1
The aforementioned stack is often used outside the field of window films and placed in between two glazing panels, where no moisture can reach the stack. This is crucial, as it is well known that the inner metallic silver layers are not stable and that they are especially sensitive to moisture. Several reports have been published on the moisture-induced degradation of sputtered low-ε coatings containing silver, concluding that high concentrations of chlorides could lead to the corrosion of the stack [2
]. Therefore, contact with moisture and chlorides is detrimental for the low-ε properties of these stacks and should be protected with an additional coating.
When looking outside the scope of flexible polymeric window films, many coatings have been reported to effectively protect this kind of stacks on solid substrates. One way of protecting stacks is to apply ceramic materials, such as ZnO, Al-doped ZnO [8
], or Al2
material, as the final hard coat [6
]. These ceramic coatings are often deposited by sputtering [9
], which is not very cost-effective. Another approach is to use organic coatings to protect underlying layers and materials which are prone to mechanical damage [10
]. These organic coatings, however, absorb far infrared radiation, which renders the reflecting stacks on window films useless, as the films will heat up and loss of heat reflection will occur. For the same reason, no hybrid materials containing scratch-resistant metal oxide particles, such as silica nanoparticles, embedded in an organic matrix [12
] could be used for this application.
To overcome the drawbacks of all known protective coatings and thus preserve the low-ε characteristics of mentioned window films, a silica thin layer was deposited through a wet chemical deposition method. This protective coating should not only meet the requirements for industrial applications, but also should be cost-effective. Chemical solution deposition (CSD) is a cost-effective deposition method for thin films. CSD starts from molecules in solution or suspension, which are deposited on a surface, which results in the growth of a new phase. The method has proven to be effective to deposit a wide range of materials [8
Industrial requirements for the protective coating include a class 0 adhesion and 3H pencil hardness, according to ASTM D3359 [20
] and ASTM D3363 [21
], respectively. Because of these goals, the possibility to deposit this coating by using roll-to-roll deposition processes was further explored.
The aim of this work is to develop a thickness determination toolbox for transparent protective coatings on transparent substrates and to assess the minimal thickness to pass the predefined application tests of a protective silica coating for low-ε window films. Ellipsometry, scanning, and transmission electron microscopy (SEM and TEM) are often the characterization techniques of choice. However, ellipsometry is not suitable for fully transparent devices containing a relatively thick, transparent substrate. This is because ellipsometry is not able to distinguish thin layers from the underlying substrate in transparent devices, as the difference in refractive indices between the layers is too low. Ellipsometry fails because the signal coming from the thin layers is shadowed by the signal of the substrate. Cross-sectional view SEM and TEM measurements, on the other hand, are very good techniques to determine the thickness of top coatings, but are not suitable if polymer substrates are present, as the electron beam burns through the organic substrates [22
]. To circumvent these drawbacks, we propose to combine secondary ion mass spectrometry (SIMS) and profilometry as an effective thickness determination toolbox.
The idea behind this proposed toolbox is to use the SIMS apparatus to dig a crater in the film while checking the chemical composition of the surface (by analyzing the ions that are emitted). In this way, the sudden appearance of the ions attributed to the support is a direct indication that the upmost film has been eroded. The sputtering is then interrupted and the crater is analyzed by profilometry to determine its depth, which strictly corresponds to the thickness of the film.
To develop this toolbox, model samples were designed (Figure 1
a,b), comparable to window films (Figure 1
c). The model samples were characterized with ellipsometry, SEM/TEM, and the proposed SIMS-profilometry toolbox. An additional gold layer was sputtered to make ellipsometry and SEM/TEM measurements possible. On the one hand, the gold layer is needed to make a distinction between the individual layers with ellipsometry. On the other hand, this conducting gold layer is necessary to make Focused ion beam (FIB)-SEM measurements possible. Otherwise, too much drifting of the sample would occur as the samples become charged during measurement. In order to make TEM measurements possible, the polymer substrate was changed to silicon.
In this work, it is shown that the proposed toolbox delivers similar thicknesses to those observed in cross-sectional view SEM/TEM images, while ellipsometry shows different values of thickness. Electronic microscopy (EM) methods are considered to be accurate in the determination of film thickness. In order to check this hypothesis, both results obtained by the proposed toolbox and by ellipsometry are compared to EM. The conclusions drawn from model sample characterizations are transferred to comparable functional window films, which cannot be characterized by EM. It is shown that the proposed toolbox can be used to determine the thicknesses of thin coatings on transparent polymeric substrates.
2. Materials and Methods
The chemicals were used as received. Absolute ethanol was obtained from Applichem Panreac (Darmstadt, Germany). Hydrochloric acid (36 %) and glacial acetic acid (100 m%) were obtained from Roth (Karlsruhe, Germany). The silane precursor tetraethyl orthosilicate (TEOS, 99% pure) was supplied by ABCR (Karlsruhe, Germany). Deuterated water was obtained from Sigma Aldrich (Overijse, Belgium).
Sample fabrication: A graphical representation of sample compositions can be seen in Figure 1
. The substrate used as model sample was either silicon (1 × 1 cm2
) (Figure 1
a) or plain PET (1 × 1 cm2
) (Figure 1
b), provided with a gold layer and a titania buffer layer, both deposited by physical vapor deposition. The 100-nm thick gold layer was deposited using electron beam operation with a base pressure of 1.0 × 10−6
mbar. Also, the 5-nm thick TiOx
layer was deposited using the electron beam operation of Ti, under an oxygen partial pressure of 4.0 × 10−5
Window films (Figure 1
c) were provided by Group Michiels Advanced Materials (M.A.M.) and consisted of a 20-cm wide and 50-µm thick PET film. This PET film was functionalized with a single sputtered Fabry-Pérot stack (Zele, Belgium). On top of this stack, a titania buffer layer was sputtered.
Coating precursor synthesis: A sol-gel synthesis was adapted from previous work [15
]. In a typical synthesis procedure, tetraethyl orthosilicate was added to ethanol. In a separate container, a solution of glacial acetic acid in distilled water was prepared. Acetic acid is needed in order to catalyze the condensation reaction. All reagents were mixed while stirring. An additional pre-condensation step was conducted by heating the total mixture for 3 h under refluxing conditions at 60 °C. As a final step, the reaction was quenched with ethanol. Four hundred milliliters of a transparent sol was obtained in accordance with patent EP15199592.5 [26
Coating deposition: TiO2/Au/silicon and TiO2/Au/PET model substrates were coated using spin-coating by dropping 90 µL of the transparent sol on the substrate. The sample was then spun for 60 s at 2000 rpm with a spin acceleration of 40 rpm/s2. After spin-coating, it was annealed at 80 °C for 60 s on a hot plate in an ambient atmosphere.
Window film substrates were coated via roll-to-roll coating (Figure 2
), using the coating conditions summarized in Table 1
. Characterized window films were coated once and coated twice using these coating settings.
Raman characterization: The kinetics of the pre-condensation step were monitored with a RamanRxn system from Kaiser optical systems (Ann Arbor, MI, USA), model Rxn1-532. The measurements were performed in situ by placing a laser probe inside the reaction mixture. The mixture was put in a three-neck flask, equipped with a reflux cooler and a dropping funnel to add reagents, together with the laser probe. The flask, the dropping funnel, and the reflux cooler were covered in aluminum foil to prevent the detection of radiation coming from the outside of the mixture. The wavelength of the laser was 532 nm with a maximum power output of 450 mW. A resampling interval of 1 cm−1 was used while recording 13 accumulations with an exposure time of 10 s.
Nuclear Magnetic Resonance (NMR): To evaluate the degree of condensation and the correlated shelf life of the sols, 29Si-NMR measurements were performed using a Bruker (Billerica, MA, USA) Avance III spectrometer with a 1 h frequency of 500 MHz, equipped with a 5-mm BBI Z probe. Samples were prepared by adding 100 µL of D2O to 900 µL of the sol before transferring to an NMR test tube.
Spectroscopic characterization: Optical transmission and reflection spectra were recorded with an Agilent Cary 5000 UV-Vis-NIR spectrophotometer (Santa Clara, CA, USA). The spectra were recorded from 300 to 2000 nm.
Application Tests: Crosshatch tests were performed according to ASTM D3359 standard [16
]. A cutter size of 1 mm was used with six cutting edges. The adhesion test tape has an adhesive strength of 9.5 N per 25-mm width. Both were supplied by Paint Test Equipment (Congleton, UK). The sample was prepared by a first cut in the coating with the cutter. A second cut, perpendicular to the first cut was made. The adhesive tape was placed over the damaged area and, finally, the tape was removed from the surface. Using a magnifying glass, the coating was checked for damage.
The hardness of the coating was evaluated with an Elcometer (Utrecht, The Netherlands) 501 Pencil Hardness Tester. The method used was the Wolff-Wilborn method. The equipment complies with the ASTM D3363 standard [17
The emissivity ε was measured with a TIR 100-2, supplied by Inglas (Friedrichshafen, Germany). The samples were cut to a size of 10 by 10 cm.
Layer thickness: The proposed toolbox should be compared to other layer thickness characterization techniques to verify its effectiveness. The cross-sectional view images and lamella were prepared by a FEI Nova (Hillsboro, OR, USA) 600 Nanolab Dual-Beam Focused Ion Beam system and associated SEM.
For TEM analysis, a cross-sectional lamella was obtained using ion milling techniques via the FIB in situ lift-out procedure with an Omniprobe (FEI, Hillsboro, OR, USA) extraction needle and top cleaning. High-angle annular dark-field (HAADF) and bright-field (BF) scanning TEM images were taken on a JEOL JEM-2200FS TEM (Tokyo, Japan) with a Cs corrector, operated at 200 kV. Chemical information was obtained via the phase analysis via energy dispersive X-ray spectroscopy.
Spectroscopic ellipsometry measurements were performed using a J.A. Woollam M-2000 ellipsometer (J.A. Woollam Co., Lincoln, NE, USA). The spectral range of the ellipsometer ranged from 250 to 1680 nm and the COMPLETEEASE software (version 6.34) was used for fitting and data analysis.
Ellipsometry mapping measurements were performed on a homemade mapping stage. The nominal angle of incidence for all measurements was fixed at 70°. The acquisition time for one spectrum was set at 1.5 s.
The measured data was modeled using a B-spline layer for the Au film, a Cauchy layer for the TiO2 substrate, and another Cauchy layer for the silica film. The thicknesses of the various layers were determined using fixed parameters for the Cauchy parameters.
Depth profiles SIMS were made using a TOF.SIMS5 (IONTOF GmbH, Münster, Germany) time-of-flight secondary ion mass spectrometer [27
]. This instrument is equipped with a Cs+ ion beam as a sputtering source and a liquid metal ion gun (Bi) as an analytical source, both mounted at 45° with respect to the sample surface. The time-of-flight mass analyzer is perpendicular to the sample surface. Depth profiles were carried out in the ‘interlaced’ mode in a cycle time of 100 µs. During this cycle time, the analytical beam (pulsed Bi5+
for SIMS analysis) was followed by periods of continuous Cs+
ion sputtering. Low energy electrons were also sent to the sample during this cycle in order to recover the initial surface potential. The Cs+
ion source was operated at 500 eV with a direct current of 40 nA. For depth profiling, the focused Cs+
beam of primary ions was rastered over an area that typically measured 450 × 450 μm2
. A pulsed beam of 30 keV Bi5+
(alternating current of 0.11 pA) ions was employed to provide mass spectra from a 150 × 150 μm2
area in the center of the sputter crater. Charge compensation was conducted using an electron flood gun (Ek
= 20 eV). All data analyses were carried out using the software supplied by the instrument manufacturer, SurfaceLab (version 6.5). Depth profiling was stopped when the Ti+
signal reached 50% of its maximum intensity.
Stylus profilometer was used to measure the craters obtained by SIMS measurements (DektakXT, Bruker Nano Surfaces Division, Tucson, AZ, USA). The stylus had a radius of 0.7 µm and the applied force was 0.1 mg. Four line scans were conducted over the measured crater in order to obtain average and standard deviation measurements.
The main conclusion of this work is that the proposed thickness characterization toolbox for transparent protective coatings on transparent devices is proven to be a valuable method. The results obtained by this combination of SIMS and ellipsometry are in agreement with those achieved using electron microscopy for the designed model samples. The toolbox was able to determine the thicknesses of thin layers deposited on functional window films, where other techniques failed.
This research also showed synthesis and deposition methods for the large-scale application of scratch-resistant coatings on polymeric, heat-reflecting window films. Both Raman and 29Si-NMR spectroscopy were used to assess precursor synthesis and the following shelf-life. The results showed that an optimal coating precursor was obtained after 3 h of condensation and that a shelf-life up to five days was guaranteed. After simple and straightforward coating deposition with roll-to-roll coating, application tests were conducted. It was shown that a coating thickness of 274 ± 12 nm leads to class 0 adhesion, 3H pencil hardness resistance, optimal visual transparency, and an even improved infrared reflectivity, compared to uncoated samples. It is possible that a lower coating thickness would be sufficient, keeping in mind the lower threshold of 126 ± 8 nm, but this was not studied further. By avoiding use of chloride containing reagents, further scaling up towards industrial settings is possible. It is proven that this chloride-free synthesis has no negative effects on the drying and curing of the coating, as the coating passed all application tests. A low temperature heat-treatment program was shown to be sufficient to obtain these scratch-resistant coatings. A maximum temperature of 80 °C was needed to fully dry and cure the coating. It is thus possible to transfer these results to other temperature-sensitive substrates.
In short, this research demonstrated an effective thickness characterization toolbox. Additionally, during this research an easily up-scalable coating was developed. This coating is intended to protect scratch-sensitive substrates, which cannot be coated with organic or hybrid state-of-the-art materials, due to their optical properties.