Fabrication and Characterization of Optically Transparent PVDF Films Embedded with Gold Nanoparticles

Polyvinylidene uoride is a piezoelectric polymer that can be cast into transparent thin lms. New properties can be introduced by embedding nanoparticles in this polymer, making it an excellent platform for exible and tunable electronic and optoelectronic devices. We develop a recipe for embedding plasmonic gold nanoparticles into these lms while maintaining their transparency as an initial step to activate optical response in the lm. We characterize lms made under different poling conditions with and without nanoparticle inclusions using X-ray diffraction. We nd that the inclusion of gold nanoparticles screens the poling eld and has a sizable effect on the phase of the produced lms.


Introduction
Polyvinylidene uoride (PVDF) is a well-established ferroelectric polymer with highly stable chemical, thermal, mechanical and UV properties. [1][2][3][4][5] Research in recent decades have shown that embedding PVDF with various nanoparticles can enhance its existing properties and potentially promote new functionalities. 1,2,6−8 For example, if a nanoparticle upon optical excitation triggers an intrinsic effect (piezoelectricity or pyroelectricity) in a PVDF thin lm to generate a detectable electrical signal, then a new response to optical stimulus is introduced. Such a hybrid system could then be exploited to make exible and transparent optical sensors in wearable devices. In practice, for such a system to be viable, it is necessary to produce a transparent and ferroelectric PVDF lm with the right nanoparticle inclusion. In addition, the density of the particles needs to be substantial enough to produce a measurable effect without affecting the transparency of the lm.
In this report, we have taken the initial steps in producing and testing the viability of PVDF lms embedded with plasmonic nanoparticles. Plasmonic nanoparticles are well-known for exhibiting size dependent optical resonant properties. At resonance, the optical energy is e ciently absorbed leading to ultra-high enhancement of near-eld intensity as well as localized heating. 9 These effects could stimulate the piezoelectric and pyroelectric responses in the PVDF lm. While we do not measure for an optical response in the lms for this study, we are able to synthesize a transparent PVDF lm with plasmonic gold nanoparticles (AuNPs) and con rm that the system exhibits ferroelectricity using X-ray diffraction (XRD). More importantly, we nd that the AuNPs have a statistically signi cant effect on the properties of the PVDF lms. This makes the lms a promising candidate for further optical studies.

Casting Method
A popular method for creating PVDF structures is the technique of electrospinning. Electrospun plastic sheets have applications in membranes for batteries and electronics. [1][2][3][4][5] However, this method leads to opaque structures. Another method of forming piezoelectric PVDF lms is by spin-coating the polymer onto a substrate in the presence of a high electric eld. 1 This creates thin PVDF lms that are clear and thus allow optical access to the nanoparticles embedded within to activate potential optical properties. This method of producing piezoelectric PVDF devices is cost-effective and relatively straightforward. The key is to overcome the need for a high electric eld utilized to induce a ferroelectric polymer phase (500-700 kV/cm) and to consistently create a ferroelectric lm with a high degree of transparency. 10 Past works have shown that mechanical stress, high annealing temperature, and high electric elds all contribute to encouraging polarized phase formation and transparency during lm production. [10][11][12] Taking these parameters into account, we provide a detailed documentation of a straightforward method for producing transparent PVDF lms with and without AuNPs under poled and unpoled conditions. We create multiple lms under each condition to facilitate statistical assessment of the consistency of this method and analyze the effect AuNPs and poling have on the nal phase composition of the lms.

PVDF Film Production
The process for producing bare PVDF thin lm is summarized in Figure 1. First, 2 mL of acetone and .25 g of PVDF (in powder form) are mixed in a beaker at 85 ℃ for ten minutes with a magnetic stirrer. The purpose of the acetone is to suspend and disperse the PVDF powder uniformly in a liquid mixture. This mixture is white and cloudy because PVDF does not dissolve in acetone as shown in Figure 2a. After ten minutes, 1 mL of dimethyl sulfoxide (DMSO) is introduced to the mixture to dissolve the PVDF. The solution is stirred at a temperature of 85 ℃ for 45 minutes to completely boil off the acetone, leaving only the dissolved PVDF in DMSO. The solution will gradually go from a white and cloudy mixture to a clear and viscous solution during the 45 minutes of mixing as shown in Figure 2b.
After 45 minutes, the solution is ready to be cast. A drop of the solution is placed on a 1cm x 1cm silicon substrate in the spin-coater. The spin-coater chuck is grounded. The solution is spun for 25 seconds at 4,000 rpm. To electrically pole the lm, a potential is applied via an aluminum plate placed above the ground plate. We pole our lms using either a weaker eld at 1.31 kV/cm or a stronger eld of 3.25 kV/cm. After spinning, the PVDF coated silicon substrate is transferred to a hot plate and baked for 75 minutes at 85 ℃. After baking, the PVDF lm is dry enough to be peeled off and stored for XRD measurements.

PVDF-Au Film Production
The 60 nm AuNPs from Sigma Aldrich are suspended in an aqueous solution as shown in Figure 3b. The reddish color is due to the wavelength selective enhanced re ection as a result of the surface plasmon polariton resonance. We nd that adding the AuNPs directly to the PVDF and acetone mixture along with the aqueous solution leads to failure in transparent lm production. The main reason is that while water cannot dissolve PVDF, PVDF is miscible in DMSO which dissolves PVDF. This combination leads to the formation of a gelatinous substance which is neither transparent nor spreadable into a thin lm. To solve this problem, we modify the previous procedure to eliminate the water while keeping the AuNPs.
The process for producing AuNP embedded PVDF thin lm is summarized in Figure 3. We start with 1.5 mL of AuNP solution. This amount can vary depending on the desired nal concentration of AuNP. The nanoparticle solution is placed in a centrifuge to separate the AuNPs from the water. After centrifuging, most of the water is removed using a pipet while leaving the AuNPs at the bottom of the test tube. The AuNPs are immediately resuspended in 4 mL of acetone with a vortex mixer to avoid permanent aggregation of the AuNPs which can destroy its plasmonic property.
The 4 mL acetone/AuNP solution is mixed with 0.25 g of PVDF powder in a beaker at 105 ℃ for 10 minutes. This increased temperature, in contrast to 85 ℃ in the bare PVDF lm production, ensures that any remaining water is boiled off from the mixture. After 10 minutes, 1 mL of DMSO is introduced to the mixture. The solution is stirred continuously at 105 ℃ for another 50 minutes. As the acetone (and any remaining water) is boiled off, the PVDF dissolves into the DMSO. The solution will go from an opaque, pale pink mixture to a clear, viscous red-pink hue shown in Figure 3c. The fact that this coloration remains is a sign that the plasmonic property of the AuNPs is maintained. At this point, the solution is viscous enough to be cast. The spin coating step is the same as that in the bare lm production. After the spincoating step, the AuNP PVDF lm with substrate is placed on a hot plate for 75 minutes at 85 ℃ and peeled off afterwards to be characterized in XRD. Examples of PVDF lms produced without and with AuNPs are presented in Figure 4a and 4b, respectively.

Results
XRD is used to characterize the phase compositions of the PVDF lms. We expect to observe peaks associated with the three most common phases in PVDF (α, β, andγ). The non-ferroelectric α phase has two dominant double peaks around 2θ = 18.4 ∘ and 19.9 ∘ . For the ferroelectric phases, the dominant γ phase peak is at 2θ = 20.2 ∘ whereas the dominant β phase peak is at 2θ = 20.8 ∘ . 13,14 These peaks are all near each other making it di cult to identify the phases of the lms using this group alone. It is then helpful to look at minor peaks at larger angles for more clues on the phase composition of our lms. At larger 2θ values, the α phase has a wide, short peak at 26.5 ∘ . Similarly, the γ phase has a wide, short peak at 39 ∘ . It is often noted that a small peak at 36.3 ∘ is associated with the β phase. However, we found it unhelpful for phase identi cation because there are similarly sized peaks from the α and γ phases nearby. 8,14−16 Nevertheless, we have highlighted this peak in Figure 5 to distinguish it from the others. Using both peak information from small and large 2θ, we then assigned the appropriate phaseor combination of phases -for each lm. Figure 5a highlights three examples of pure phase assignments using the guide described above. Figure 5b gives examples of the mixed phase assignments. The α-β mix is assigned to any lms whose XRD spectra have a combination of the α phase peaks at 18.4 ∘ and 26.5 ∘ and a shift of the tallest peak towards 19.9 ∘ . The β-γ mix is assigned to any lms which exhibited no signs of being in the α phase and with its tallest peak directly in the middle of 20.2 ∘ and 20.6 ∘ . For example, in Figure 5b we see the β-γ mix has its tallest peak at 20.4 ∘ . We note that our assignment of pure phase only means that the particular phase has the majority fraction and does not necessarily mean 100% purity.
The phase categorization results for the 93 successful lms produced is presented in Figure 6. The table in Figure 6a provides a summary of the number of lms produced under each category. For unpoled lms (no eld), there is no clear indication of a preferred phase for both lms with and without AuNP inclusions, except maybe a small tendency towards the γ phase as seen in Figure 6b. We attribute this to the high annealing temperature during lm production which has been observed in previous works to produce ferroelectric lms even without a poling eld being applied. 11,12 An important point to note is that the AuNPs present no signi cant effect on the lms under the unpoled condition. Conversely, the difference in poled lms is apparent as shown in Figure 6c and 6d. Poled lms containing no AuNPs are only found in the ferroelectric phases. On the other hand, the poled lms embedded with AuNPs exhibit a trend more like that of unpoled lms and still support the non-ferroelectric α phase. This difference between lms made with and without AuNPs is more apparent at the weaker eld of 1.31 kV/cm eld as shown in Figure 6c. At this eld strength, poling completely removes the α phase from the lms without AuNPs but does not do so for lms with AuNPs. In the case of the stronger eld of 3.25 kV/cm, a pure α phase no longer forms in lms with AuNPs. Rather, a combination α-β phase is still found as seen in Figure 6d. This implies that the embedded AuNPs are having a notable effect on the PVDF during the production process. Since these particles are metal, it is likely that they are screening the electric eld in their immediate vicinity and hence reducing the poling effect for the PVDF nearby. Even though the introduction of AuNPs is reducing the formation of pure β phase in the lms, if the poling eld is strong enough, as seen in Figure 6d, the non-ferroelectric α can still be reduced. This result suggests that the quantity of AuNPs present is enough to produce a measurable modi cation of the PVDF lms on a macroscopic scale while still maintaining lm transparency. This also means that these hybrid lms will be viable for future investigations to explore localized optical modi cation of PVDF using laser light at the resonant plasmonic frequency of the AuNPs.

Conclusion
We have detailed a straightforward and cost-effective method for synthesizing poled/unpoled optically transparent PVDF lms with and without AuNPs. Using XRD, we have characterized our lms and nd that AuNP inclusion in PVDF has a considerable effect on the phase properties of the lm under poling condition. This effect is likely due to the screening effect of the metallic nanoparticles. This result presents a promising outlook for these lms as transparent optical sensors and optically tunable optoelectronic devices. Figure 1 The method for casting PVDF lms with no added AuNP.   Examples of spin-coated a) bare PVDF lm and b) PVDF lm embedded with AuNP. Both lms are transparent as desired. While the bare lm is colorless, the AuNP lm retains a pinkish hue indicative of the intact plasmonic property of the 60nm AuNPs. a) Examples of XRD spectra for lms with the three pure phase assignments. b) Examples of XRD spectra for lms with the two mixed phase assignments. The α-β mix represents lms with signatures of both the α and β phases. The β-γ mix represents lms whose main peak lay in the middle of 20.2° and 20.6° and show no α features. a) A table summary of the numbers of lm produced in each category. The different number of lms in each category is due to lm destruction during removal from the silicon substrates. b-d) Bar graphs detailing the phase composition for a set of lms with a speci c constraint (i.e., poled, gold-embedded, etc.). b) A comparison of all lms made under unpoled condition. c) A comparison of all lms poled with a weaker 1.31 kV/cm electric eld. d) A comparison of all lms poled with a stronger 3.25 kV/cm electric eld.