*3.2. Optimization of Thickness and Post-Deposition Annealing*

When polymer films are exposed to humidity they change their thicknesses and refractive indices. This results in change of the measured reflectance or transmittance spectra. We have recently shown

that the dimensional change in polymer films with nanometer thickness in the range 100–400 nm depends on the initial thickness and increases with increasing thickness [25]. Figure 2a presents the change in thickness for studied samples (80 and 200 nm) at their exposure from low to high humidity. As expected, the degree of swelling of thicker films (200 nm) is substantially higher as compared to thinner films especially for 19% modified PVA films where the relative increase of thickness (Δ*d*/*d*) is 120%. Furthermore, the post-deposition thermal treatment of samples at higher temperature (180 ◦C) does not lead to an improvement of swelling, as we have already shown in [21]. On the contrary, the dimensional changes of films pre-annealed at 180 ◦C are smaller as compared to those treated at 60 ◦C, especially for the neat PVA films which degree of swelling is almost 7 times lowered.

**Figure 2.** (**a**) Thickness change upon humidity exposure ranging from 5% to 95% RH of polymer films with thickness of 80 and 200 nm and of different acetal content pre-annealed at 60 and 180 ◦C; (**b**) Percentage of hysteresis, *H* calculated with eq. 3 for polymer films (80 and 200 nm) and of different acetal content pre-annealed at 60 and 180 ◦C.

It is well known that the hysteresis, *H*, is another very important parameter that determines the suitability of the material for sensor applications. The existence of *H* means measuring of different signal (reflectance or transmittance in our case) for the same humidity values depending whether humidity increases or decreases. It is obvious that *H* is unwanted parameter and the main goal is to keep its value as low as possible.

The hysteresis values *H* of all samples studied is summarized in Figure 2b and Table 2. A substantial decrease of hysteresis due to annealing at 180 ◦C is observed for neat PVA films. The smallest *H*-values (4.3% and 3.8%) are achieved for PVA-modified samples (80 nm) with acetal content of 18% and 24%, pre-annealed at 180 and 60 ◦C, respectively.

**Figure 3.** Reflectance versus relative humidity curves for films with 24% and 18% acetal content (80 nm thick), pre-annealed at 60 and 180 ◦C, respectively, measured for increasing (solid black symbols) and decreasing (open blue symbols) humidity.


**Table 2.** Post-annealing temperature (*T*post), wavelength at which *R* (*T*) measurements were conducted (λmax), percentage of hysteresis (*H*), dynamic range, sensitivity, and accuracy for studied samples.

One can conclude that the most suitable samples are Ac24 and Ac18 annealed at 60 and 180 ◦C, respectively. The dependence of reflectance on relative humidity in the range 5%–95% RH (*R*-vs-RH curves) for both samples of the smallest H-values are presented in Figure 3.

It is seen that reflectance for Ac18 is almost the same in wide humidity range (5%–60% RH) and starts to increase exponentially at RH > 60%. On the contrary, for sample Ac24 two linear dependences of *R*-vs-RH plot with different slopes are well distinguished. The sensitivity is 0.03 at RH = 5%–65% and increases to 0.14 at RH > 65%. Considering that the measurement error in reflectance is 0.3% the accuracy of humidity measurement is 10% and 2% RH, respectively (Table 2). We should note that although the swelling is the strongest for thick films (200 nm) they are not suitable for sensing because exhibit high values of hysteresis (Figure 2b). Additional measurements of reflectance as a function of relative humidity (not shown) demonstrated that thicker samples have narrow dynamic range. Generally, for thicker samples *R* changes only for humidity higher than 80% RH. For sample Ac19 the case is even worse because there is unambiguity—one and the same reflectance values are measured for different humidity. The reason is the periodicity of the dependence of R on d. When the change of thickness due to humidity is higher than the period of *R*-vs-*d* dependence then a periodicity in *R*-vs-RH curve could be observed. Usually this unambiguous behavior appears for thicker films (*d* higher than 150 nm) where swelling is stronger as compared to thinner ones [25].

Considering all results presented above we concluded that the most appropriate sample for our purposes is Ac24 (24% acetal content PVA) with approximate thickness of 80 nm pre-annealed at 60 ◦C. Further efforts are concentrated on optimization of sensing properties of this material using two approaches: (i) humidity sensing through transmittance measurements; and (ii) doping with SiO2 particles.
