A systematic approach for the accurate and rapid measurement of water vapor transmission through ultra-high barrier films

Recent developments in flexible organic electronic devices require equally flexible encapsulation materials with extremely high barrier properties against moisture and oxygen to protect the sensitive active and electrode materials. A current approach for the production of such encapsulation films is to coat polymeric substrate films such as polyethylene terephthalate (PET) or polyethylene naphthalate (PEN) with alternating inorganic and organic layers.^1,2 These materials have a long lag time before water penetration, in addition to a low steady-state water vapor transmission rate (WVTR), to ensure the long-term protection of the sensitive device layers. The WVTR requirements for the encapsulation material of state-of-the-art organic solar cells and organic light emitting diode structures are ∼10⁻³ g m⁻² d⁻¹ and ∼10⁻⁶ g m⁻² d⁻¹, respectively, at 38 °C and 90% relative humidity (RH), to maintain reliable performance for more than 10 000 h.³ 

In order to detect such low quantities of permeants, and to improve the barrier properties of such materials, the metrology must also evolve. Several approaches can be used for the measurement of water vapor permeation through films, including calcium corrosion tests, coulometric electrochemical devices, radioactive methods, mass spectrometry, and infrared light spectroscopy.^4–14 

Two isostatic carrier gas systems are commercially available for the measurement of moisture permeation: the Aquatran^® devices from MOCON®, Inc., with a detection limit of (5 × 10⁻⁵ ± 5 × 10⁻⁵) g m⁻² d⁻¹ and the HiBarSens® from SEMPA Systems GmbH, with a detection limit of 1 × 10⁻⁶ g m⁻² d⁻¹.^15,16

The measurement of ultra-high barrier materials is challenging because of the long conditioning and measurement times. Depending on the barrier performance and the layer structure of a multilayer barrier material, WVTR measurements may take several months to reach steady-state transmission rates, and it can be difficult to determine the onset of steady-state conditions. In some cases, the WVTR values recorded before the water vapor permeation process reaches its steady-state are too low, thus producing inaccurate data that overestimate the performance of barrier materials.

We therefore developed an efficient and systematic approach for the measurement of water vapor permeation based on the accumulation of water vapor and an appropriate evaluation method. This approach has the following benefits:

• Larger measurement areas may be used (188 cm²) compared to the commercially available measurement devices.

• Precise temperature control is possible within broad ranges of temperature (23–80 °C) and RH (15%–90%).

• The steady-state WVTR values can be determined more rapidly because measurements can be taken at high temperatures, with 30 measurement cells operating at the same time independent of the barrier performance of the samples.

• The combination of measurements and advanced theoretical calculations based on the finite element method (FEM) allows us to predict steady-state WVTR values accurately, long before steady-state conditions have been reached.

The permeation process involves mass transport over a certain time period through a solid, which has at least two surfaces acting as boundaries to the surrounding gas. It is characterized by four stages.

The adsorption of gas molecules on the surface of the solid (polymeric) sample is followed by the absorption in the near-surface volume area. At sufficiently low partial pressures of permeating gas molecules, the amount of gas absorbed per volume of the solid sample can be described by Henry’s law (1), which is applicable if the substance sorbed in the polymer does not dissociate. Here, c is the equilibrium concentration in the polymer, S is the solubility coefficient, and p is the partial pressure of the gas,¹⁷ 

The third step is the diffusion of the gas molecules through the solid. Diffusion was described quantitatively for the first time by Adolf Fick¹⁸ using the diffusion coefficient D and the locally varying concentration c of the permeating substance. This yields Fick’s first law (2) for the (molar) flux density j of a permeating substance (shown here in its one-dimensional form),^19,20

The concentration c of the diffusing substance remains constant over time at every point in the material, when the permeation process has reached its steady state. The time dependence of c is given by Fick’s second law (3a), also shown in its one-dimensional form,

Finally, the gas molecules escape from the solid on the opposite film surface and desorb into the surrounding atmosphere. This process is the counterpart to the absorption step and usually also follows Henry’s law.

Typically, WVTRs are recorded as a function of time as shown in Fig. 1 (In this paper, the term transmission rate is used for the (mass) flux density). After the start of water vapor absorption in the sample, the WVTR value approaches asymptotically to a stationary permeation value (dotted line).

We used the simulation program COMSOL Multiphysics Version 4.4, which is based on FEM and can solve physical problems described by partial differential equations.^21,22

The time-dependent change of the concentration c of water vapor in a multilayer structure (Fig. 2) can be described within each layer L_i of thickness h_i by Fick’s second law,

assuming that the diffusion coefficients D_i and solubility coefficients S_i are concentration independent.

The solutions of the diffusion equation in two adjacent layers i and i$+1$ are coupled by the following conditions Eqs. (4) and (5), must be fulfilled at their interface on the x-axis $x=h1+⋯+hi$⁠, and are implemented in COMSOL. The chemical potential and, therefore, the partial pressure $p=cS$ of the water vapor are continuous at interfaces,

The conservation of the amount of permeating water vapor results in the continuity of the normal component of its flux density $j=−D∂c∂x$⁠,

During the drying period, the initial concentration of water vapor within each layer i at time t = 0 is defined as $ci(0,x)=Sip∼$ with $p∼=1.4$ kPa, which is equal to the ambient water vapor partial pressure (23 °C and 50% RH). The concentrations at each boundary x = 0 and x = h are set to zero. During the permeation period, the concentrations in the layered structure at the boundaries x = 0 and x = h correspond to the partial pressures according to Henry’s law (1), e.g., 2.4 kPa for the measurement condition of 38°C/36% RH, and 0 kPa, respectively.

In order to find the lower detection limit and determine the accuracy of the ultra-permeation accumulation (UPA) measurement device, we selected laminate films from Toyo Seikan, Co., Ltd, consisting of polyamide, polyester, Al foil (6 $μm$ thick), and polyethylene. These films are typically used for control measurements in the commercially available MOCON® Aquatran® devices. The structure of these Al foil laminate (AFL) films is shown in Table I.

ACLAR® polychlorotrifluoroethylene (PCTFE) is a well-characterized material. We used ACLAR® 3000 films of 76 $μm$ thickness from Honeywell International, Inc., to validate the repeatability of the measurement results.

Furthermore, a high-barrier film (HBF) was produced according to the Fraunhofer POLO® concept,²³ which is based on the coating of alternating inorganic and hybrid-organic barrier layers on the top of a polymeric substrate. The substrate was a polyethylene terephthalate (PET) film from DuPont Teijin Films (a newly developed version of PET Melinex®). This has greater outdoor stability than standard PET, without any effect on the final water vapor barrier performance of the multilayered structures.

The inorganic barrier layers (transparent silicon oxide, SiO_x) were deposited using physical vapor deposition by electron beam evaporation (Amcor Flexibles Kreuzlingen AG). An ORMOCER® (ORM) lacquer was applied as an intermediate hybrid-polymeric layer by a reverse-gravure process at Fraunhofer IVV, Freising. These lacquers are organically modified ceramics, which were synthesized in a sol-gel process from organo-alkoxysilane precursors, by Fraunhofer ISC, Würzburg.²⁴ 

The HBF was produced by the face-to-face adhesive (adh) lamination of the two barrier films. Each of these films consisted of the new outdoor stable PET Melinex® substrate (50 $μm$ thick) coated with SiO_x (100 nm), ORM (1 $μm$⁠), and a second layer of SiO_x (100 nm). The final structure of the HBF is shown in Table I.

Fig. 3 shows a schematic diagram of the ultra-permeation accumulation (UPA) measurement device developed at Fraunhofer IVV. A previous version of this unit has already been described.²⁶ 

Compared to the original version, this tailor-made unit has been modified to include new options allowing the transient and steady-state WVTR to be measured in barrier films under different conditions. Sample films with a diameter of 155 mm are mounted inside 30 identical measurement cells, which are placed in a temperature-controlled chamber. This allows the measurement of 30 different films at the same time. The measurement cells include dry He-flushed elastomeric double seals to minimize the possible leakage of water vapor through the samples’ edges or the inner seal. The large sample size improves the measurement sensitivity by increasing the area for water vapor to permeate into the dry chamber.

The system includes two 16-way selector valves each covering 15 measurement cells. The permeated water vapor from each of the measurement cells is transported via He gas through the valves to the pre-trap. One He pipe is not connected to a measurement cell but directly connects the gas-inlet and the pre-trap allowing water leakages to be detected and controlled during the measurement.

The RH of the nitrogen flow can be varied between 15% and 90%, by mixing dry and humid nitrogen. A sensor (HMP238 from Vaisala) placed within the temperature-controlled chamber measures the RH of the gas mixture with an accuracy of 0.5% and controls the flow ratio via proportional-integral-derivative (PID) controlled flow meters. Furthermore, the drying and measurement temperatures can be varied from 23 °C to 80 °C with an accuracy of ±0.1 °C.

The entire system, including the barrier films, is conditioned properly before measurement to ensure well-defined initial conditions. During this period, the whole unit is heated to 50 °C and flushed with nitrogen that has been dried in a cooling trap, to remove residual water within the samples and the apparatus. The zero value is measured individually for each measurement cell providing an intrinsic indicator for the leak rate of the system.

After this drying period (Fig. 4), humid nitrogen is admitted to the measurement cells and comes in contact with the upper side of the films within the measurement cells. Due to the partial-pressure difference, water vapor starts to permeate towards the dry side and is then carried to the pre-trap by He, where the water is adsorbed and accumulates for 10 min. The pre-trap is filled with Chromosorb^® 104 (Imerys Minerals California, Inc., USA) material and operates at room temperature. The accumulation of water vapor in the pre-trap allows us to detect very low WVTR values.

In the next step, the adsorbed water is thermally desorbed at 240 °C, transferred into the main-trap, which is filled with Tenax® polymer (Buchem B.V., Netherlands), and there, it is frozen at −50 °C. After the second thermal desorption step at 200 °C, the water is transferred to and quantified by a thermal conductivity detector with a detection limit of 5 × 10⁻⁷ g m⁻² d⁻¹. This detector measures the thermal conductivity of the surrounding atmosphere, which can be related by a calibration factor to the mass of water that permeated per area and time (e.g., g m⁻² d⁻¹). Calibration was carried out as previously described.²⁶ The measurement cycle of each cell took ∼22 min, so each cell was measured every 12 h. The measurements were stopped as soon as a steady-state permeation value was reached.

The conventional approach to define steady-state conditions is strongly influenced by the resolution of the measurement device and the experimental methodology. Numerical simulation of the time-dependent initial phase of the measured permeation rate with the COMSOL simulation program allows us to calculate the steady-state WVTR. The measurement can be stopped as soon as the mean of the last five measurement values is within 1.5% of the calculated steady-state WVTR value. For barrier films with very long lag times, the measurement can be stopped earlier and the extrapolated steady-state value can be used. This enables us to reduce the measurement times for high-barrier films significantly.

The AFL film described above was used to determine the detection limit of the UPA device. The initial measurements were taken under dry conditions of 38 °C and 0% RH for 7 days (drying period). When the values were within a standard deviation of 1.5% within 3 days, a steady-state zero value was achieved. The mean of the last five measurement values was 2.5 × 10⁻⁵ g m⁻² d⁻¹ with a standard deviation σ_WVTRzero of 1.0 × 10⁻⁵ g m⁻² d⁻¹. After this zeroing period, the relative humidity was set to 36%. The WVTR was measured as a function of time and is shown in Fig. 5. There was a slight increase in the values compared to the drying period due to the pinholes in the Al foil. The mean and standard deviation σ_WVTRsteady of the last five values, calculated using the same procedure applied during the zeroing period, were 6 × 10⁻⁵ and 1.6 × 10⁻⁵ g m⁻² d⁻¹, respectively. Therefore, the WVTR for this AFL sample, calculated by subtracting the mean of the zeroing period from the mean of the measurement period, was 3.1 × 10⁻⁵ g m⁻² d⁻¹ with a standard deviation $σWVTRtotal=(σWVTRzero)2+(σWVTRsteady)2$ of 1.9 × 10⁻⁵ g m⁻² d⁻¹.

If the difference between the two mean values is significant, it is at least twice the standard deviation σ_WVTRzero, i.e., 2 × 10⁻⁵ g m⁻² d⁻¹. This value is the smallest significant amount of change that can be detected by the UPA device at an accumulation time of 10 min, which is the measurement sensitivity.

Fig. 6 provides one example of the two measurements of a PCTFE film in one measurement cell to determine whether the results are reproducible. Therefore the film was measured twice consecutively at 40 °C and 75% RH.

The WVTR curves showed close agreement, with calculated mean values of 69.5 mg m⁻² d⁻¹ for the first measurement and 70.5 mg m⁻² d⁻¹ for the second. The deviation between consecutive measurements in one measurement cell was 1.5%, which is similar to the changes in the atmospheric pressure observed during the measurement.

It is important to ensure that the barrier film does not undergo any structural changes due to water during measurement because this would influence the permeation rate.

Table II shows the average WVTR values and 95% confidence intervals $Δx¯$ obtained for PCTFE films in g m⁻² d⁻¹. Fifteen samples were measured simultaneously in individual measurement cells. To validate the results, the WVTRs were also measured using the commercially available coulometric method (MOCON® Aquatran®) in accordance with DIN ENISO 15106-3 for plastic films. The specified measurement limit of the MOCON® Aquatran® Model 2 is 5 × 10⁻⁵ g m⁻² d⁻¹.^15,27,28 These measurements were taken twice so both values are presented rather than the mean. One of the advantages of the UPA device is its ability to measure up to 30 samples simultaneously, providing more reliable data more rapidly.

The WVTR values obtained by the UPA device agree closely with those obtained using the MOCON® Aquatran® and with the manufacturers’ data sheets. The largest deviation was detected at 40 °C and 75% RH. Here, the UPA values lie between the Aquatran® values and those provided by the manufacturer.

The WVTR values of the HBF (structure as shown in Table I) measured with the UPA and the Aquatran® devices at 38 °C and two different RH (36% and 90%) are shown in Table III.

The WVTR values determined using the UPA device agreed again closely with the values determined using the Aquatran® device. The small deviations in the measurements mainly reflect the processing and handling of the multilayered films.

The simulation tools described in Section II B were used to predict the steady-state WVTR based on the values measured as a function of time before the steady state was achieved. This helps to reduce the experimental measurement times from several months to a few weeks. Fig. 7 shows the mean WVTRs and standard deviations of four specimens, calculated at each measurement time, through the HBF as a function of time. The WVTR values at the steady state are shown in Table III.

During this measurement, the film was initially dried for 24 days at 38 °C, then humidified for 42 d at 36% RH, and finally (starting at day 66) dried again. The solid line in Fig. 7 is the calculated water permeation flux curve for the structure shown in Fig. 8. The thickness of each layer is shown in Table IV. By varying the simulation parameters until there was a close match between the numerical and experimental data, we were able to determine the D and S values of SiO_x, ORM, and adhesive (adh) layers. The values for PET were taken from the literature.²⁹ The variation of the measurement conditions in the simulation allows the zero value (dotted line) and the steady-state value (broken line) to be calculated.

There may be several combinations of all D and S values, all of which give water vapor permeation flux density curves similar to the one shown in Fig. 7. Nevertheless, the D and S values are physically reasonable because the D and S values of PET, ORM, and adhesive are in the same order of magnitude. The main parameters affecting the water vapor permeation flux density and the time lag are the diffusion (D) and the solubility (S) of SiO_x because their values are much lower than those of the polymeric and hybrid-polymeric materials involved. The calculated curve using the D and S values in Table IV represents the average measurement curve with high precision and allows us to calculate the mean steady-state WVTR. In this paper, the D and S values we determined are called as effective diffusion (D_eff) and solubility coefficients (S_eff).

In the second measurement (Fig. 9), the measurement time was reduced from 42 d to 7 d. Nevertheless, the calcul-ated solid curve can still be adapted to the experimental data, leading to the same steady-state WVTR of 6.7 × 10⁻⁴ g m⁻² d⁻¹. Importantly, the simulation was performed using the same parameters as described above. This indicates that the characteristic time constants of the system, which are mainly a function of the materials D and S, are well reproduced by the model under changing boundary conditions. Although it is difficult at this stage to determine if the parameters represent a global minimum for the least square fit, this test supports the determined effective values in Table IV. The whole measurement time, including both the drying and measurement periods, was reduced from ∼90 d to ∼40 d.

The ultra-permeation accumulation (UPA) measurement device described herein makes it possible to measure the water vapor permeation time resolved at different temperatures (23 °C–80 °C) and relative humidities (15% RH–90% RH) for 30 samples simultaneously. The sensitivity of the measurement device is 2.0 × 10⁻⁵ g m⁻² d⁻¹. The validation of WVTR values with PCTFE films under different conditions agreed well with values measured using the MOCON® Aquatran® devices and values reported by the manufacturer of the films. The WVTR values measured for a multilayer high-barrier film are in close agreement with the values measured using commercial devices. Numerical simulation of the diffusion equation allows us to calculate the D and S values of each layer of a multi-layered structure. This helps to predict the steady-state WVTR values of ultra-high barrier structures in a much shorter measurement time, i.e., before the steady-state value is reached. In this manner, it becomes possible to reduce the long permeation measurement times for multilayer high barrier structures from several months to a few weeks.

The authors want to thank Tobias Schmidt for his support in the development of the adjustable gas humidification. Furthermore we thank DuPont Teijin Films, Amcor Flexibles Kreuzlingen AG, and Fraunhofer ISC for the supply of barrier materials and films. This work was partly financially supported by the European Union 7th Framework Program under Grant Agreement Nos. 260086 (NANOINSULATE project), 280581 (NANOMEND project), and 287594 (SUNFLOWER project).