Supporting Information for the Manuscript: Dramatic Increase In Polymer Glass Transition Temperature Under Extreme Nanoconfinement In Weakly-Interacting Nanoparticle Films Haonan Wang, Jyo Lyn Hor, Yue Zhang, Tianyi Liu, Daeyeon Lee,, and Zahra Fakhraai, Department of Chemistry, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States, and Department of Chemical and Biomolecular Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States E-mail: daeyeon@seas.upenn.edu; fakhraai@sas.upenn.edu UCaRI films: There are two mechanisms of forming UCaRI films. 1 In the NP-top geometry, the process is dominated by fast infiltration. When there is no residual PS layer left to infiltrate, the diffusion becomes the main driving force to form a uniform composite film. As such, if the annealing time is not long enough, there can be a gradient in the concentration of the polymer inside the NP film. A gradient in the fill fractions has been previously reported in PS/TiO 2 UCaRI films with a 10% gradient for 50% fill fraction. 1 Since the diffusion of PS is much slower in SiO 2 compared to TiO 2 NP films, To whom correspondence should be addressed University of Pennsylvania University of Pennsylvania 1
for reasonable annealing times a uniform UCaRI film can only be obtained with 70% PS fillfraction or higher. The fill fraction can be calculated based on the initial thickness of the PS layer. However, this value, as well as the thickness of the NP layers measured before and after infiltration have relatively large errors due to the formation of cracks during infiltration for the NP-top geometry (Figure S6). Figure S4 shows that there is strong correlation between index of refraction of the NP after the completion of the infiltration process and the PS fill-fraction calculated based on the film thickness. As such, in this study the index of refraction is used to calculate the PS fill fraction after the completion of the infiltration process, to account for the crack formation. Linear fit is used in the range between 0.7 and 1. Since the index of refraction can be measured more accurately, the error of PS fill fraction can be significantly reduced using the fitted equation. Characterization of PS-top CaRI films In order to avoid crack formation during thermal infiltration of CaRI films, the method of making PS-top samples is introduced in this study. In this method, the PS is spin-coated on top of the NP film and subsequently infiltrated. Figure S3 shows the index of refraction of SiO 2 (11nm)/PS(8K) and SiO 2 (11nm)/PS(2M) films immediately after spin-coating of the PS layer and after the thermal infiltration. For PS(8K), PS can fill part of the pores during the spin-coating process. The solventinduced infiltration leaves more PS closer to the top of the NP layer resulting in a gradient of index of refraction. For PS(2M), because of the high molecular weight and viscosity of the solution, much less filling of the pores occurs during spin-coating. After annealing above T g, both molecular weight chains can fully fill the pores of the NP layer, forming uniform CaRI films. Additional ellipsometry parameters: In addition to the fitting parameters mentioned in the main text, the offset of angle of incidence was also fit when necessary to ensure that any uncertainty in the angle does not affect the data. The beam size was reduced to 30 microns using focusing optics. Before mounting the focusing 2
optics, the ellipsometer was aligned to ensure a direct light path with a 70 reflection off of the sample. Calibrations were performed after mounting the optics using Si wafers with a thermally grown oxide layer of a known thickness to correct for any changes due to the focusing optics. The spectroscopic wavelength range during the measurement was set to 600-1600 nm to maintain imaging in the transparent region of the spectrum and avoid scattering by NPs. To ensure good thermal contact between the samples and the temperature controller stage (Linkam THMS600), the heating element was coated with Arctic Alumina thermal paste (Arctic Silver, Inc.), and the samples were clamped tightly to the heating element. Due to the use of the ellipsometer s high-accuracy setting (zone averaging), the actual temperature was the average of the listed temperature and the previously listed temperature. This was corrected post acquisition. A small temperature lag (<1 K for 10 K/min cooling rate) was observed in both indium melting tests, and in the thickness data. This temperature shift, which was the same for all samples, was taken into account post acquisition. The effect of plasma treatment on properties: In order to spin coat a good NP film on PS for the NP-top geometry, the solvent of the SiO 2 suspension must wet the surface of PS. Hence, for aqueous suspension, 2s room air plasma treatment on the PS film surface were performed to increase the surface energy. To rule out the influence of the surface plasma treatment on the final properties of the films, SiO 2 (11 nm) in isopropanol (IPA) was used as a control because IPA wets the PS surface well. PS(8K) can fully infiltrate the SiO 2 (11 nm) NPs in 15 s when annealed at 433 K. For spin coated films, it takes hours for the irreversibly adsorbed layer to grow at that temperature. 2 Thus, by controlling the annealing time, we are able to examine the influence of the adsorbed layer. Figure S10 shows T g of PS(8K) with different annealing time at 433 K. Clearly plasma treatment and the growth of adsorbed layer has no significant influence on T g after infiltration. 3
The width of the glass transition: In order to calculate the width of the glass transition, the slope of the refractive index change vs. temperature (dn/dt ) was calculated. The refractive index data was first smoothed with the adjacent average method. Then, the 1st order derivative was calculated followed by another adjacent average smoothing. T + and T, the start and end point of transition, respectively, were determined to be the intersection of the linear fits of the regions before, during and after transition as shown in Figure S11. References 1. Hor, J. L.; Jiang, Y.; Ring, D. J.; Riggleman, R. A.; Turner, K. T.; Lee, D. Nanoporous Polymer- Infiltrated Nanoparticle Films with Uniform or Graded Porosity via Undersaturated Capillary Rise Infiltration. ACS Nano 2017, 11, 3229 3236. 2. Housmans, C.; Sferrazza, M.; Napolitano, S. Kinetics of Irreversible Chain Adsorption. Macromolecules 2014, 47, 3390 3393. Table S1: Thickness of PS and NP layers before and after infiltration. Porosity of the NP film was calculated to be h PS /h NP. 4
Figure S1: (a) Typical ellipsometry model used for CaRI films, including all the fitting parameters (bold blue fonts). (b) Fitting parameter uniqueness test of incident light angle offset shows that angle offset is a necessary parameter in the fitting. (c) Comparing different optical models for each layer confirms that the layers are uniform and ellipsometry is not sensitive to the surface roughness for over-filled CaRI films. 5
Figure S2: Ellipsometric angles Ψ (red) and (green) vs. wavelength for an over-filled PS(8K)/SiO 2 (25 nm) CaRI film measured at T = 303 K. The dashed line shows the fit to the two-layer model used. The film thickness of the NP layer and the residual PS layer were determined to be h PS = 106.1 ± 0.5 nm and h NP = 206.1 ± 0.5 nm, respectively. The spectroscopic wavelength range of 600 1700 nm was used to maintain imaging in the transparent region of the spectrum and avoid scattering by NPs. Incident light angle offset was also fit to correct the tilt angle of the sample. 6
Figure S3: Index of refraction of the film as a function of the distance from the substrate before and after thermal infiltration for PS-top CaRI films with different PS molecular weights. Figure S4: PS fill fraction calculated from initial PS thickness vs. refractive index at 632.8 nm at 423 K for different PS(8K)/SiO 2 (25 nm), NP-top UCaRI films with 200 nm thickness. Error bars are calculated from error of NP layer thickness before and after infiltration. The straight line is the linear fit of the data points. 7
Figure S5: (a) Ellipsometry model for PS (8K)/SiO 2 (25nm) CaRI film. Layers 2 and 3 are the residual PS layer and the NP layer, respectively. (b) During heating and cooling ramps, the thickness of layer 3 (NP film) was not sensitive to temperature. To avoid over-fitting, the thickness of layer 3 was kept constant for data presented in the manuscript. (c) The mean square error (MSE) of the fitting with fitting layer 3 (black) and without fitting layer 3 (red). (d) The calculated refractive index of the NP layer with (black) and without (red) fitting layer 3. It shows that both MSE and Index values are insensitive to the fitting of layer 3. 8
Figure S6: Cracks after the infiltration from the bottom-up for (a) PS(8K)/SiO 2 (11 nm), (b) PS(8K)/SiO 2 (25 nm), and (c) PS(8K)/SiO 2 (100 nm0 NP-top films. All samples were CaRI films with 300 nm NP layer and 250 nm residual PS layer. Cracks start to appear at the beginning of the infiltration at 378K. (d) SiO 2 (11 nm)/ps(2m) PS-top film does not show evidence of crack upon top-down infiltration. Scale bars are 2 mm in all figures. 9
Figure S7: The "Try Alternate Models" feature in the ellipsometry software CompleteEASE for (a) 86% PS fill fraction and (b) 50% PS fill fraction CaRI films. The green columns are the best model for the samples. For Samples with small PS fill fractions, there is a large negative refractive index gradient in the film, which indicates that more PS chains stay on the bottom of the NP packings and cannot spread uniformly within the annealing time. Figure S8: SEM images of (a) UCaRI film ( 50% PS fill fraction) and (b) CaRI film of PS (8K)/SiO 2 (100nm) CaRI films. The amount of PS was controlled by changing the rate of spincoating. 10
Figure S9: (a) Temperature profile and change of index of refraction during infiltration, annealing and T g measurement of PS(2M)/SiO 2 (25nm) PS-top CaRI sample. Index of refraction is stable after 20 min annealing at 433 K. (b) 10 K/min cooling ramps after annealing at 433 K for 20 150 min. Good overlap of the curves indicates no influence of the adsorbed layer. 11
Figure S10: T g of PS(8K)/SiO 2 (11nm) NP-top CaRI film and PS(2M)/SiO 2 (25nm) PS-top CaRI film at a cooling rate of 10 K/min for different annealing times at 433 K with and without room air plasma treatment of the PS surface prior to infiltration. Error bars of PS(8K) samples represent average standard deviation of more than three samples. Error bars of PS(2M) samples represent 95% confidence level of linear fit. 12
Figure S11: (a) The slope of refractive index change (dn/dt ) of bulk PS(8K) and PS in PS (8K)/SiO 2 films of different NP sizes. Dashed lines show linear fit of the regions before, during and after glass transition. (b) Calculated T + (black) and T (red) values for PS(8K) NP-top CaRI samples with various NP diameters and selected UCaRI PS(8K)/SiO 2 (25nm) samples. Dashed lines show the corresponding values of bulk PS. There is no significant change in the width of the glass transition width in the confined state and both T + and T shift with a similar relative value inside the NP films. 13