An Investigation into the Influence of Interfacial Effects in a Three Layer Solid Phase Sensing Platform

With modern technological advances, we are seeing an increasing amount of electronic devices and materials going the route of smaller and faster. It has become of great importance to understand how these devices and materials work on the nanometer scale. Many of these new materials are coming to fruition in the form of thin films, and the topic of this dissertation focuses on the application of these materials to the field of chemical sensing. Chapter 1 of this dissertation investigates the photophysics of a thin film of the fluorescent dye, rhodamine 6G (Rh6G), on the surface of poly(vinylidene fluoride) (PVDF) coated glass, to determine the origin of a previously reported fluorescence enhancement. Three factors were identified that contribute to the increase in fluorescence seen in this system when compared to a thin film of Rh6G on a bare glass surface. First, the surface roughness of the underlying PVDF films provides a larger area for an excitation beam to interact with, which leads to the excitation of more Rh6G molecules. Second, the PVDF layer reduces the amount of aggregation between Rh6G molecules as the film thickness of the dye layer increase, which preserves the emissive monomeric form of Rh6G. Last, the PVDF thin films acts as a light trapping layer which leads to more Rh6G absorption events and more efficient use of the incident light. Chapter 2 of this dissertation shows how the surface morphology of polystyrene (PS) thin films on a glass substrate evolves as the molecular weight (Mw) of the PS used to cast the film is changed. Optical profilometry was used to collect images of the different films which revealed that PS pillars are formed in the center which transition into wrinkles that extend out from the center. The periodicity and amplitudes of these features changed with the Mw, which was found to be a result of the direct relationship between the glass transition temperature of PS and its Mw. Furthermore, this study showed a method of producing micrometer sized wrinkled interfaces spontaneously, which before required a more complicated process to fabricate. The last chapter of this dissertation, chapter 3, contains a spectroscopic study on thin films of differently charged xanthene dyes on the surface of PS coated glass to investigate potential ion-π interactions. Coupling deconvolution of absorbance and emission spectra with measurement of excited state lifetimes of the dyes on the PS surface revealed that a large portion of monomer emission from the cationic dye (Rh6G) and the anionic dye (disodium fluorescein, DSF) was quenched by formation of a weakly emissive exciplex. This is in stark contrast to the neutral dye (fluorescein 27, F27), which showed no sign of exciplex formation and had the opposite emission behavior with respect to increasing the dye layer film thickness when compared to the charged dyes. The overall findings of these studies show that there are complicated dynamics that can occur in solid phase, layered thin film systems. Interfacial interactions, whether they originate in physical or chemical nature, can have a vast effect on the performance of these devices. This dissertation demonstrates how having a fundamental understanding of such systems can reveal information necessary for the complete optimization of sensors and other nano-sized devices.

amplitudes of these features changed with the M w , which was found to be a result of the direct relationship between the glass transition temperature of PS and its M w . Furthermore, this study showed a method of producing micrometer sized wrinkled interfaces spontaneously, which before required a more complicated process to fabricate.
The last chapter of this dissertation, chapter 3, contains a spectroscopic study on thin films of differently charged xanthene dyes on the surface of PS coated glass to investigate potential ion-π interactions. Coupling deconvolution of absorbance and emission spectra with measurement of excited state lifetimes of the dyes on the PS surface revealed that a large portion of monomer emission from the cationic dye (Rh6G) and the anionic dye (disodium fluorescein, DSF) was quenched by formation of a weakly emissive exciplex. This is in stark contrast to the neutral dye (fluorescein 27, F27), which showed no sign of exciplex formation and had the opposite emission behavior with respect to increasing the dye layer film thickness when compared to the charged dyes.
The overall findings of these studies show that there are complicated dynamics that can occur in solid phase, layered thin film systems. Interfacial interactions, whether they originate in physical or chemical nature, can have a vast effect on the performance of these devices. This dissertation demonstrates how having a fundamental understanding of such systems can reveal information necessary for the complete optimization of sensors and other nano-sized devices.
iv ACKNOWLEDGMENTS Without these people none of this would have been possible: Bill Euler, for taking the time and having the patience to guide and help me grow as a scientist. Your wisdom, knowledge, and understanding of Chemistry and attention to detail will forever have an impact on the way I think about not only science, but life in general. You have been an outstanding mentor and advisor throughout this process and for that I will forever be grateful.
Rebecca Levine, my fiancée and forever the love of my life. The way that we have grown together throughout this time is something I will treasure for eternity.
From day one, you have always been there for me emotionally and your intellect has always helped me find the right path. I simply could not have done this without you.  Table 3. Component contribution of the global multi-exponential fit for fluorescence decay curves their associated decay times as a function of t Rh6G , t PVDF = 760 nm. α i is a pre-exponential term denoting its relative contribution of the i th exponential to the overall decay. τ i is the time constant for the excited state, or the value of 1/e for the i th exponential. χ 2 is an indication of how good the fit is and should be as close to a value

Introduction
The use of fluorescence to study the polarity of the surrounding environment is textbook material. 1 This methodology is typically applied to solutions, either liquid or solid. Application to interfaces to determine surface polarity has also been reported. 2−8 However, the structure of many fluorophores is dependent on the thickness or surface concentration, 9,10 [27][28][29][30][31][32][33][34] Depending upon the relative orientation of the C-F and C-H bonds, the polymer can crystallize in the nonpolar α-phase, the polar γ-phase, or the ferroelectric β-phase.
The purpose of this study is to investigate the role that the PVDF layer has on the previously reported emission enhancement. 26

Results and Discussion
PVDF films were all produced by spin-casting a 4% PVDF (w/ v) onto a glass slide. Different thicknesses were obtained by changing the rotation rate, and for this study the thickness range of the PVDF layer is 380−770 nm. This thickness range was chosen to optimize light trapping of visible light. The measured thicknesses of the PVDF thin films as a function of rotation rate can be seen in Figure 1 determined from both the center and the edges of the films. The edges of the films tend to be thicker than the centers of the films for lower rotation rates. At higher rotation rates, the films are more uniform across the substrate. As noted by the small error bars in Figure 1, the reproducibility of the spin-coating is excellent. Atomic force microscope scans were performed on some of the neat PVDF films, and the thicknesses at the edge matched those measured by reflection spectroscopy. Figure 2 shows an AFM image of a 760 nm thick PVDF film. The surface roughness is notable so we measured this parameter as a function of PVDF film thickness. The results are shown in Figure 2. For thinner PVDF films, the surface roughness is a significant fraction of the average thickness, which suggests that there may be pinholes on the surface that expose the glass substrate. When the film thickness exceeds ∼450 nm, the surface roughness is fairly constant at ∼175 nm.  FTIR spectroscopy has been shown to be a convenient and accurate method to determine the phase composition of PVDF, correlating directly with XRD. 29,37−40 FTIR spectra of PVDF coated on glass and PVDF scraped from a glass slide can be seen in Figure 3. The peaks are poorly resolved for the coated glass because the glass substrate interferes, but for the PVDF scraped from a glass slide the phases can be determined. The peak at 1286 cm −1 is indicative of β-phase and is assigned as a combination of ν s (CF2 ) , ν s (CC), and δ(CCC) modes. 29,33,34,37−40 The 840 cm −1 peak is characteristic of both the β-and the γ-phases and is assigned to the r(CH2) and ν a (CF2) modes. 29,33,34,37−40 The other unique γ-phase peak is located at 1234 cm −1 , which corresponds to the ν a (CF2), r(CF2), and r(CH2) modes. 29,33,34,37−40 The peak at 760 cm −1 is present only in the α-phase and is assigned to δ(CF2) and δ(CCC) vibrational modes. 29,33,34,37−40 All of the phase sensitive peaks are observed, indicating that all three phases are present in the samples used in this study. Rh6G films were spin-cast onto the PVDF films using EtOH as a solvent.   On the glass substrate, the maximum absorbance wavelength changes gradually as a function of t Rh6G , with the ∼520 nm component decreasing and the ∼560 nm feature increasing as t Rh6G increases. On the PVDF surface, the peak at 518−521 nm remains the most intense until the Rh6G film thickness changes from 1.4 to 1.8 nm. At this point, the most intense peak changes to ∼560 nm, but the intensity of the ∼520 nm peak does not decrease. As the Rh6G film thickness grows from 0.9 to 2.3 nm, both the populations of monomer Rh6G and aggregated/crystalline Rh6G grow concurrently until eventually the aggregated/ crystalline Rh6G species dominates at multilayer Rh6G thicknesses (t ≥ 7 nm). This implies that there are regions on the PVDF surface that prevent aggregation. Whether this is imposed by the surface morphology or electronic effects by regions of different phase is not clear. When the thickness of the Rh6G layer is changed on thin PVDF (380 nm), the resulting spectral changes are similar, as shown in Figure 4C. Deconvolution of the Rh6G spectra on t PVDF = 380 nm leads to the same peak positions and shapes used to deconvolute the spectra on t PVDF = 760 nm. Therefore, the PVDF thickness has little effect on the electronic properties of Rh6G and that the composition of the surface phases of the PVDF is similar for all thicknesses.
Deconvolution of the absorbance spectra (shown in Figure S1 and Table 1) shows the strong dependence that the Rh6G film thickness has on the amount and type of Rh6G species present on the PVDF surface. At submonolayer thicknesses, the absorbance spectrum consists of a main peak at 518 ± 2 nm with a small shoulder to the high energy side at 495 ± 2 nm, and a very small shoulder on the low energy side at 542 ± 2 nm. These features are assigned to monomer Rh6G (518 nm) and an oblique exciton dimer (495 and 542 nm) formed between two Rh6G molecules in the excited state, respectively. Two peaks arise due to the structure of oblique exciton dimers as described by Kasha. 12 Increasing the Rh6G film thickness to 0.9 nm causes a change in the intensities of the peaks mentioned above, but does not require any additional peaks to fit. It is not until t Rh6G = 1.1 nm that an addition peak at 560 ± 2 nm is needed to fit the spectrum. The new peak is assigned to a higher order aggregate of Rh6G that can be formed due to the amount of neighboring Rh6G molecules in confined areas on the PVDF surface. Last, when t Rh6G ≥ 7.0 nm, a fifth peak at 452 ± 2 nm is needed to achieve a good fit. An isolated molecule of Rh6G has approximately C2 symmetry about the center of the middle aromatic ring in the xanthene plane, which means that the only transition moment that is allowed is the one across the long axis of the xanthene plane. However, as Rh6G molecules aggregate, the C2 symmetry is reduced and the transitions forbidden in the isolated molecule become allowed.
Therefore, it is likely that the peak at 452 ± 2 nm is originating from a transition moment perpendicular to the strongly allowed transition of the isolated molecule.
Each of the peak maxima found for Rh6G on PVDF are blue shifted as compared to the parameters reported for Rh6G on glass. 10 This indicates that the surface polarity of the PVDF is less than the glass surface, probably being dominated by the nonpolar α-phase, which has a dielectric environment different from glass. Table 1. Deconvoluted Gaussian peaks for the absorbance spectra. Peak position (λ max ) and FWHM (Γ) have an uncertainty of ±2 nm. All spectra with fits and component peaks can be found in Figure S1. The effect of the PVDF film thickness at constant Rh6G thickness was studied by preparing a set of PVDF films with thicknesses between 380 and 760 nm, which were all then spun-cast with the same concentration of Rh6G, giving all samples the same average t Rh6G , ∼2.3 nm. As shown in Figure 5, the general shape of the Rh6G absorbance spectrum does not change much as a function of the PVDF film thickness.

Peak λ max (nm) Γ (nm) t Rh6G (nm)
As the PVDF layer gets thicker, the 560 and 518 nm peaks fluctuate a small amount.
As stated earlier, the PVDF film can act as a trapping mechanism for light via internal reflection between the Rh6G/PVDF and PVDF/Glass interfaces.   The emission spectra of several thicknesses of Rh6G on thick (760 nm) PVDF are shown in Figure 7A. As the Rh6G becomes thicker, two effects are observed: the spectra become more intense, and the peak maximum shifts to lower energy. At low Rh6G film thicknesses on PVDF, the emission maximum is narrower and positioned at ∼537 nm, but as the thickness increases the maximum shifts to 555 nm, as shown in Figure 7B. Moreover, as shown in Figure 7C, the spectra broaden so that the total intensity of the emission increases as the Rh6G thickness increases up to about 2 nm and then slowly decreases. This is in contrast to a glass substrate where the emission intensity increases up to a nominal thickness of about 2.3 nm and then drops precipitously, 10 also shown in Figure 7C. Deconvolution of the emission spectra as a function of Rh6G film thickness (given in Figure S2 and Table 2) showed complementary data in agreement with the deconvoluted absorbance spectra. At the lowest Rh6G film thickness of 0.7 nm, the absorbance spectrum shows two absorbing species, but deconvolution of the emission spectrum revealed that the spectra require three Gaussian curves centered at 535 ± 3, 561 ± 3, and 592 ± 3 nm to obtain satisfactory fits. These peaks are assigned to monomer, excimer, and exciton emission, respectively. When the Rh6G film thickness is increased to 0.9 nm, the only thing that changes is the relative intensities of the previous three peaks. It is not until tRh6G ≥ 2.3 nm that a fourth Gaussian curve is required, which is located at 644 ± 3 nm and is assigned to the emission from larger Rh6G aggregated species. The reason why the trends in the Rh6G emission differ on the PVDF and glass substrates is presumed to be due to both the surface roughness of the underlying PVDF layer and the lower polarity of the PVDF surface. The smaller surface polarity is responsible for the blue-shifted emission peak maxima relative to the glass substrate, consistent with the absorption spectra. Some of the increased intensity of the emission spectra is due to the increased surface area of the PVDF, because more Rh6G molecules will be available in the area of the excitation area. However, at large thicknesses where aggregates and/or crystallization of Rh6G dominate the surface coverage, the emission of these aggregates is quenched significantly on glass but much less so on PVDF. To help confirm the assignments responsible for the emission, the excitation spectra were measured as shown in Figure   8. The only contribution to the emission is from the species that corresponds to the 518 and 495 nm absorption peaks, which we have identified as monomer and oblique exciton dimer, respectively. To be sure there is no contribution from the species that absorbs at 560 nm and to rule out J-type aggregation, the excitation spectra were recorded at different monitoring wavelengths, shown in Figure 8C. At no point in the emission curve does the structure that absorbs at 560 nm contribute to the emission, and deconvolution of the excitation spectra at any wavelength shows only monomeric and exciton features. The lack of any red-shifted peak in the excitation spectra shows that J-type aggregation is not a contributing factor in the perceived emission enhancement. In Figure 9 can be seen the effect of changing the PVDF film thickness.
The intensity of the emission is dependent on the PVDF film thickness but the shape of the spectra is not, which was the same case with the excitation spectra as a function of PVDF thickness as shown in Figure 9C. Because excitation spectra show only the transitions that lead to emission, it makes sense the excitation spectra follow a similar trend. This intensity effect is caused by the way the Rh6G molecules organize more randomly on thin PVDF, potentially due to the existence of pinholes exposing the underlying glass substrate or a change in the relative amounts of PVDF phase. The normalized plot shows that there is minimal change in the emission maximum when changing the PVDF film thickness. Deconvolution of these emission spectra shows the same peaks as above with just a slight variation in their intensities.
To investigate the changes in the steady-state fluorescence and absorption spectra more in depth, a time-resolved emission spectroscopy (TRES) experiment was performed. Figure 10 shows the decay profiles of different Rh6G thicknesses on a constant thickness of PVDF. At each tRh6G, the decay is multiexponential and strongly dependent on decay collection wavelength. The surface roughness of the PVDF film causes incident light to scatter as well, which has been accounted for in the analysis. As the decays move further away from the source wavelength, the scattering component decreases and the decays are more influenced by the combination of fast and slow fluorescing components. The global multiexponential fitting model used to analyze the wavelength dependent decays was composed of three components for t Rh6G = 7.0, 2.3, and 0.9 nm, and two components for t Rh6G = 0.7 nm. Because of the roughness of the PVDF film, there is a scattering component present in all models that has been assigned to subnanosecond decay, α 1 , and τ 1 . The results are given in Table 3 36 The decrease in lifetime in aqueous solution was attributed to energy transfer from monomers to nonemissive aggregates, which also likely happens here as there is a sufficient overlap between the emission of monomer Rh6G and the absorbance of aggregated Rh6G as shown in Figure S4. The assignment of the short lifetime to the excited-state monomer is also supported by the TRES. For the 0.7 nm sample, the spectrum is the same for all times, with a peak maximum at 540 nm.
Likewise, the short time spectra for all thicknesses are dominated by the higher energy peak. Figure 11. Fluorescence spectra extracted from TRES data at selected times and t Rh6G with PVDF thickness = 760 nm.
The longer lifetime excited state is assigned to an excimer because the excitation spectra for this feature arise solely from the monomer absorption. The TRES spectra show that for tRh6G ≥ 0.9 nm, there is a shift in the emission maximum with respect to time. For the t Rh6G = 0.9 nm data in Figure 11, the emission spectra at t = 1 ns show a maximum at ∼545 nm. As time passes, the dominant emission changes from the fast monomers to the slower excimers, and at t = 20 ns the emission maximum is at ∼560 nm. Table 2. Component contribution of the global multi-exponential fit for fluorescence decay curves their associated decay times as a function of t Rh6G , t PVDF = 760 nm. α i is a pre-exponential term denoting its relative contribution of the i th exponential to the overall decay. τ i is the time constant for the excited state, or the value of 1/e for the i th exponential. χ 2 is an indication of how good the fit is and should be as close to a value of 1 as possible. The standard deviation in the lifetime is determined from two measurements. The behavior for t Rh6G ≥ 2.3 nm is the same as that for the 0.9 nm film except the wavelength shift is slightly further to the red (∼570 nm), which can be seen in Figure   11. As thickness increases, the relative ratio of excimer to monomer is increased due to the increased number of neighboring Rh6G molecules. The excimer lifetime also decreases with increasing Rh6G thickness for the same reason as above, energy transfer to nonemissive aggregates.

Conclusion
When Rh6G is coated onto a PVDF surface, the emission intensity is significantly higher as compared to a similar coating on a glass surface. Both the absorbance and the emission of the Rh6G species are blue-shifted as compared to the glass surface, which is attributed to changes in surface polarity and the dielectric environment between PVDF and glass. This indicates that the surface of the PVDF

Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.7b00144. Absorption spectra with fits, emission spectra with fits, and excitation spectra with fits.       The following is in preparation to be submitted to Macromolecules and is presented here in manuscript format

Introduction
In the present state of materials science, where there is a push for devices to be made smaller, it is important to understand a material's physical and chemical properties from both the perspective of the surface as well as the bulk. It is well documented that a material can behave differently when comparing these two regions of the same substance, 1-5 which is mostly attributed to confinement effects.
Furthermore, an additional area becomes of importance when the surfaces of two different materials interact to form an interface. In devices comprised of layered thin films this interfacial region can have a large effect on the overall behavior of the device and therefore brings about the importance to understand the interactions that can occur at the interface. Understanding and controlling the chemical and physical interfacial properties of a layered thin film system would allow for more control over the devices behavior and may be used in the fabrication of such devices to tailor them for a specific purpose.
A defining feature of an interface in a layered thin film system is the surface morphology of the underlying layer. Surface morphology is a broad term and can be considered on many different scales. In modern thin film systems one must consider the nano/micrometer regions of surface morphology as well as the macro features.
Changing and controlling surface morphology through the use of etching and lithography in hard materials such as silicon wafers has been well studied and understood for some time now, 6-8 however with polymers playing a larger role in devices new methods are needed to control surface morphology. The push for understanding and controlling surface morphologies arises mostly from the field of adhesion, 9,10 but certain patterning in materials has been shown to be promising for use in photonic devices [11][12][13][14][15][16][17] and stretchable microelectronics. 18

Experimental Section
Glass microscope slides were cut into 3. The film thickness of the PS was determined using a Filmetrics F40 microscope via reflection spectroscopy. The reflection spectrum was recorded with a tungsten-halogen light source over the range of 400-900 nm. The resulting interference pattern was then fit with a calculated spectrum of given thickness and refractive index. After the film thickness was determined, the surface morphology of the PS films was analyzed using the Filmetrics Profilm 3D Optical Profiler. The images of the surface are obtained vertical scanning interferometry (VSI) and phase shifting interferometry (PSI).

Results and Discussion
The spin coating method for fabricating thin films is often used because of its ability to create uniform films of a desired thickness with good reproducibility.
However, the actual dynamics of spin casting are complicated, as the final product and quality of the film depend on many factors. As Scriven et al 38 have demonstrated, the final thickness of a dry film can be estimated through the following equation, where is the initial viscosity of the solution, is the mass fraction of polymer in the initial solution, ∞ is the mass fraction of the pure solvent in the gas phase at equilibrium, is the kinematic viscosity of the overlying gas, is the diffusivity of the overlying gas, is the density of the solution, is the vapor pressure of the pure solvent, is the molecular weight of the polymer, R and T represent the ideal gas constant and temperature, and is the angular speed. Equation [1] demonstrates that changing the molecular weight of the polymer will have an effect on the thickness of the film that is cast both explicitly through the term and implicitly through the initial viscosity, since ∝ , where n depends on how the polymer chains coil. Figure 1 shows the determined PS film thickness as a function of PS molecular weight, while all other variables are kept constant. As predicted by equation [1], the film thickness scales to a power of 1/6 with respect to the molecular weight of the polystyrene. In order to make a proper comparison between the films of different PS molecular weight, it is important to ensure they are all of the same thickness, as material properties in thin film systems have been observed with changes in film thickness. [39][40][41][42] Therefore, in order to achieve a uniform film thickness for all samples, the parameters of the spin coating process were analyzed. First, equation [1] implies that the desired average film thickness could be achieved by changing the spin speed. Figure 2 shows the relationship between average PS film thickness and spin speed for M w = 350000 g/mol. The film thickness and spin speed are inversely related, but as the fit in figure 2 shows, equation [1] is not sufficient to describe film thickness as a function of angular speed.
Equation [1] trends down for ω > 4000 RPM, however there is a divergence from the fit when ω > 2600 RPM, which indicates additional forces acting on the film that are not considered in equation [1]. Therefore, the effect of the initial concentration of the PS casting solutions on the film thickness was analyzed and is shown in figure 3. M w = 350000, 105000, 53000, and 13000 g/mol the PS concentration was made to be 2.0%, 2.5%, 3.0%, and 3.5%, respectively. With these initial concentrations, the resulting film thickness was in the region of 300 nm and the surface morphologies of all M w 's could now be compared to each other.    Wrinkling in homogenous films on rigid substrates has been discussed 43,44 and is mostly attributed to expansion from swelling of the polymer or from drying from a swollen state. However, this method usually leads to unstable wrinkle patterns that eventually give way to creases and folds instead of wrinkles. M. Ramanathan et.al. 45  what is reported herein, but do not make mention of the film thickness regime they worked in.
Although film thickness was not studied they did conclude that the wrinkling observed in the PVDMA films was caused solely by the film and the substrate the film was supported on had no effect on the wrinkling wavelength or amplitude.
In the present system being discussed here, the film is in a completely swollen state as it is deposited onto the substrate and dried throughout the spinning process. In a study done by Allain and Pauchard 46 on polysaccharide drops, they observed surface instabilities as the polymer drops dried. The buckling in the dried state of the drops is attributed to a glassy layer forming at the air/liquid interface which then experiences stress as the liquid like layer underneath decreases in volume as evaporation proceeds. The combination of these two previous studies suggests that a similar phenomenon is happening in the PS films, where a more rigid, glassy surface layer is produced as the film is being formed. As the underlying liquid like layer decreases in volume from solvent evaporation, buckling occurs in the glassy layer and causes wrinkle formation. This process is described through a schematic in figure 8. Wrinkle wavelength can be predicted through use of the formula 43 , where λ is the wrinkle wavelength, t s is the skin thickness, and E surface and the bulk is the liquid-like layer that is still swollen with solvent. The difficulty that arises from using eq. 3 to describe the wrinkles in this system is that the skin thickness is unknown and the elastic moduli of the different layers of PS are difficult to consider without direct measurement. With that being said, it is implied that the wrinkling only occurs in the skin layer, or some portion of it, and therefore the amplitude of the wrinkles should be proportional to the skin thickness. Additionally, because the skin is formed by a glassy PS layer at the surface the glass transition temperature, T g , must also be considered as a function of PS molecular weight. Figure 9. Bulk glass transition temperature as a function of polystyrene molecular weight. T g was determined using differential scanning calorimetry with a scan rate of 10 °C/min. Taken from Ref. 44.
It is expected that wrinkle amplitude, and therefore skin thickness, of different molecular weight PS films should scale directly with the glass transition temperature. As figure 9 shows, there is a large change in T g as PS molecular weight goes from 1 -10 kg/mol. For low PS molecular weight, M w = 1 kg/mol, the T g is at or just above room temperature and increases to about 90-100 °C for M w ≥ 10 kg/mol. This means that during the spin casting, the low molecular weight PS films are very close to or slightly above their glass transition temperatures while high molecular weight films are ~60 °C below their glass transition temperatures. Therefore, it makes sense that the low molecular weight PS films show very small wrinkle amplitudes because during the spinning process the PS stays in a rubbery state and no glassy skin layer is formed. Conversely, for high molecular weight PS films, a skin layer forms and its thickness grows with increasing M w due to the glass transition temperatures. Studies have shown that the T g of polymer can vary with film thickness, but this is only significant when the thickness is on the order of the molecular dimensions of the polymer. 48,49 The model above is sufficient to describe the behavior of the PS films up until a molecular weight threshold is crossed, which is demonstrated in figure 7 for both the wrinkles and pillars. The amplitude and wavelength should both scale directly with molecular weight, however when M w > 105 kg/mol there is a steady decrease in the wavelength and amplitude for both the pillars and wrinkles. The reason for this behavior stems from the fact that when M w > 105 kg/mol, formation of the skin layer becomes dependent on a kinetic component. The high molecular weight chains are hindered by their weight and therefore cannot reorganize as fast as the low molecular weight films. This traps the high molecular weight films in a state where the skin layer hasn't had a chance to completely form since all films are spun for the same amount of time.
Above it was mentioned that the concentration of the casting solutions was changed in order to achieve a constant film thickness for all PS molecular weights. Initially, it might appear that this would affect the surface morphology, but as figure 10A shows, changing the casting solution concentration over the range of 1-4% has no effect on either the wavelength or amplitude of the wrinkle pattern formed. However, when the spin speed is altered there is a change in both the wrinkle wavelength and amplitude, as shown in figure 10B. This data supports the fact that there is a kinetic component involved with forming the wrinkles. When the spin speed is fast, the 350 kg/mol PS film has even less time to reorganize and form a skin layer which leads to small wrinkles with a short wavelength. As the spin speed slows down, the heavy polymer chains have more time to move and a larger skin layer can form, leaving the surface with wrinkles of longer wavelengths with more amplitude.

Conclusions
The surface morphology of 300 nm spun cast polystyrene thin films as a function of PS molecular weight has been studied using optical profilometry. It was found the surface morphology changed a function of the molecular weight of polystyrene used to cast the film.
For all PS films, the surface spontaneously forms two different structures depending on the location on the sample. On the extremities of the sample, radial wrinkle patterns are observed which transition into micron sized pillars at the center. The pillars are formed from interference of the wrinkle patterns. Due to the modulation in the glass transition temperature of polystyrene as a function of molecular weight, the characteristics of the wrinkle patterns, such as wrinkle wavelength and amplitude, can be controlled by choosing a particular molecular weight to cast the film with or by changing the spin speed used to cast the film.
Control over these features is attributed to a rigid glassy PS layer forming over a still solvated, liquid-like PS layer underneath. When the liquid-like layer dries completely, the rigid layer buckles giving way to a wrinkle surface. However, when the PS molecular weight passes a critical limit, kinetic effects in the forming of the rigid PS layer must be considered. Changing the spin speed in heavy PS films also has been shown to change the wrinkle characteristics and adds another layer of control over the surface morphology.
ASSOCIATED CONTENT

Supporting Information.
The Supporting Information is available free of charge on the ACS Publications website. Line scans of wrinkle profile with sinusoidal fits for wavelength and amplitude values.

Notes
The authors declare no competing financial interest.   The following is in preparation for submission to the Journal of Physical Chemistry C and is presented here in manuscript format.

Introduction
Since 1981, when Kebarle and co-workers 1 showed that K + ions in the gas phase bind stronger to benzene than to water, cation-π interactions have been studied extensively both from a theoretical standpoint and experimentally. [2][3][4] Investigations into the cation-π interaction have shown that it is not restricted only to atomic ions, but molecular ions like ammonium and larger molecules as well. Dougherty et.al. 5 went on to show using quaternary amines that cation-π interactions are still energetically significant in the solution phase and are responsible for a number of biological interactions in proteins. In more recent studies, the potential for cation-π interactions in the solid phase has been investigated, however most of these studies have been on protein crystal structures. 6,7 In the field of adhesion, the mechanism of mussels adhering to underwater substrates is being studied and it has been proposed that cation-π interactions play a large role in that interaction. 8,9 The discovery of cation-π interactions led researchers to look for evidence of the opposite case of anion-π interactions and a comprehensive review of anion-π interactions has been given by Giese et.al.. 10 It is textbook material that the characteristic emission of fluorophores is strongly dependent on its interactions with its surrounding media. 11   Polystyrene (PS) is a common polymer used in testing cation-π interactions due to its abundant source of aromatic rings. 8,17 Typically, a thin film of PS is fabricated then used as a surface for probing cation-π interactions. Additionally, Bertrand et.al. 18 showed using time of flight secondary ion mass spec. that in spin coated thin films of PS the phenol group constitutes a large majority of the surface.
Herein we report on a layered thin film system that consist of a PS coated substrate which then has a thin film of an organic dye coated on top. The three dyes used in this study: rhodamine 6G (Rh6G), fluorescein 27 (F27), and disodium fluorescein (DSF), are chosen due to their similar chemical structure with a central xanthene plane. Of the three, Rh6G and DSF exist as molecular ions in solution. Rh6G is a monovalent cation, DSF is a divalent anion, and F27 serves as a neutral dye. By studying the changes in the UV/Vis and fluorescence spectra of these dyes as a function of their film thickness, the potential for solid phase ion-π interactions can be probed.

Result and Discussion
To investigate the state of the dyes on the polystyrene surface, the UV/Vis absorbance spectrum was recorded as a function of fluorophore casting concentration and thus the film thickness for each dye, which can be seen in figure 2. The deconvoluted spectra for all dyes can be found in figures S1-S3 and the deconvolution parameters for the absorbance spectra of all dyes can be found in table S1.  To try and differentiate whether the additional absorption peaks that result from the thin films of the fluorophores was due to dye/dye or dye/polymer interactions, the dilute solution (EtOH) absorption and emission spectrum of each dye was collected.
The spectral profiles can be seen in figure 3  For the case of Rh6G and F27, the lowest casting concentration for the thin films results in a markedly different absorbance spectrum when compared to the dilute solution spectra. The solution spectra for both dyes have a main narrow absorption peak that is associated with a higher energy shoulder. However, when the dyes are cast onto the PS surface their spectra broaden appreciably in both directions. Figure 2B shows the height normalized absorption spectra of Rh6G which demonstrates the biggest changes in the shape of the absorption curve as the film thickness increases.
Deconvolution of the dilute solution spectrum of Rh6G showed two peaks, the main absorption at 532 nm and shoulder at 505 nm. Comparatively, deconvolution of the thinnest Rh6G film required three peaks to fit. The main absorption peak and shoulder from the dilute spectrum shift ~20 nm each to higher energies with the peaks being centered at 511 nm and 480 nm. On an energy scale, these peaks are shifted by 772 cm -1 and 1031 cm -1 , respectively, which makes sense as the PS surface is much less polar than EtOH. Furthermore, the deconvolution showed a third absorbing species that is centered at 548 nm.
When the dilute solution of F27 was deconvoluted it required two peaks, one centered at 516 nm and a shoulder at 491 nm, both of which are narrow and well defined. However, when F27 is cast as a thin film on the PS surface its absorption spectrum loses definition and the deconvolution of the spectrum for the thinnest film required three peaks to fit. Like the Rh6G, the peaks from the dilute spectrum shift ~15 nm to higher wavelengths and are centered at 500 nm and 480 nm, which correspond to energy shifts of 620 cm -1 and 467 cm -1 , respectively. The third peak required to fit the thinnest film of F27 appears at 520 nm.
Lastly, deconvolution of the thinnest DSF film required the use of four Gaussian peaks, which has one additional peak when compared to the dilute solution spectrum. The dilute solution spectrum consists of a small shoulder at 430 nm and two main absorption peaks at 451 nm and 479 nm. Interestingly, DSF displays little to no spectral shift when cast into a thin film. The shoulder peak appears at 430 nm while two other peaks appear at 455 nm and 475 nm. The fourth peak in the deconvolution of the thin film absorption spectrum appears at 505 nm. The parameters for the deconvolution of the absorbance spectra for all casting concentrations and dyes can be seen in Table S1. As the film thickness increases the spectra for all dyes broaden and shift, which is most likely the result of dimerization and self-aggregation of the dyes, which xanthene dyes are prone to. 20 However, the extent of aggregation can be used as another tool to probe ion-π interactions.
When comparing the spectrum of the dilute solutions to the thinnest films for all dyes, the thin films required an additional peak, shifted into lower energy wavelengths with respect to the main absorption peak, for all dyes. It is important to note that the thinnest films are cast from dilute solutions that show no sign of higherorder aggregation before casting. Also due to the fact that surface coverage is low and neighboring dye molecules are spread out for the thinnest films, these additional absorption peaks are likely due to an interaction between a monomer dye molecule and the PS surface, but due to the surface morphology of the PS film there could be confined areas that trap dye molecules where exciton dimers can be formed.
Regardless of whether the peak is a result of a dye/polymer or a dye/dye interaction, an indication of the strength of that interaction is the energy difference between the additional peak and the monomer absorption peak for each dye, which is summarized in Table 1. For the case of DSF, a much stronger shift is observed when cast onto the PS surface. Due to the fact that DSF is a divalent anion, the case for ion-π interactions must be considered. Anion-π interactions occur in systems where the aromatic component is experiencing electron withdrawing effects from substituent groups.
Polystyrene consist of an alkyl backbone, which results in little to no inductive effect on the electron density on the aromatic ring. In fact, the only electron deficient aromatic ring in the system of DSF on PS is the phenyl ring coming off the xanthene plane in DSF. This phenyl ring has a carboxylic acid substituent which draws electron density away from the ring. However, this effect is negligible at this casting concentration for two reasons. For one, some of the DSF forms the lactone ring with the electron withdrawing carboxylic acid which changes the properties of that ring system, and two at this thickness neighboring dye molecules are spread out. Therefore, it is unlikely that any anion-π interaction is contributing to the energy shift in the DSF/PS or DSF/DSF complex. This means that π-π interaction and H-bonding are responsible for the DSF/PS or DSF/DSF complex, similar to the F27. The difference between DSF and F27 is the magnitude of the shift and this can be explained by the substituent groups on the two dyes. DSF has an Obonded to the xanthene plane which is a very strong electron donating group. Because of this, the π system in the xanthene plane is more activated which could lead to a stronger π-π interaction.
Additionally, the Omay partake in H-bonding, which can help stabilize a complex.
Lastly, the largest energy shift observed belongs to the complex formed by Rh6G and PS. This is expected because not only can Rh6G interact in all the ways that the previous dyes can in terms of π-π interactions and H-bonding, but it also has the potential to form cation-π bonds with the positively charged ethylamine on the Rh6G and the aromatic ring on the PS surface or the xanthene plane on another Rh6G molecule.
Additional evidence for the potential of cation-π interactions comes from examination of the emission spectra. Like the absorbance spectra, the emission was recorded as a function of casting concentration, and effectively film thickness for each dye. The emission spectra can be seen in figure 4 and the deconvoluted spectra in figure S4-S6. The deconvolution parameters for the emission spectra of all dyes can be found in table S1. Comparing the emission of the dyes in the thin film form to the dilute solution, there are some drastic changes. Specifically, the spectra for Rh6G and DSF change in shape the most, with Rh6G gaining more emission peaks and DSF losing some. For Rh6G, the emission in EtOH consists of a narrow, intense peak centered at 551 nm with a small shoulder at 567 nm, similar to the absorbance spectrum. In contrast, when Rh6G is cast as a film the emission spectrum broadens and multiple peaks arise.
Deconvolution of the Rh6G thin films revealed three emission peaks for all thicknesses, one at 560 nm, another at 608 nm, and lastly a peak at 648 nm.
What is more telling about the interactions on the surface of the PS is the    (Figure 2A and 2B), the main absorption peak is at 511 nm with a small shoulder at 480 nm and a lower energy peak at 548 nm. This suggest that the surface is covered primarily by monomeric Rh6G which is also supported by the emission spectrum for [Rh6G] = 1.0x10 -5 M, as the monomer peak is the most intense in that spectrum as well.
Considering these spectra along with the data from the lifetime decay, it indicates that the short lifetime component, τ 1 = 0.05 ns, must be originating from the monomer Rh6G interacting with another molecule. As the casting concentration increase the decays require three exponentials to fit which is due to the excimer, τ 3 = 4.2 ns, forming in concentrations large enough to be detected by this method.
It is interesting to note that only three casting concentrations of Rh6G and one casting concentration of the DSF were able to be fit with χ 2 < 2.This is due to the fact that the dynamics of absorption and emission of the dyes on the surface of the PS are a For the case of F27 no decay could be fit to give a χ 2 < 2, and therefore no information about the excited state lifetime could be obtained. This likely originates from the fact that the F27 films give absorbance intensities that are an order of magnitude less than the films of Rh6G and DSF, which leads to low emission that could not be detected. However, looking at the steady-state spectra for F27 in figure   4C, it is clear that the intensity trend is different from that of Rh6G and DSF. Instead of decreasing emission intensity as the casting concentration increases, F27 displays the opposite behavior where the film cast from 1.0x10 -3 M has the highest intensity.
From an aggregation perspective this would suggest that F27 forms emissive aggregates, however looking at the normalized emission spectra in figure 4D, it is clear that the shape of the spectrum does not change as the casting concentration increase. This shows that aggregation is not what is responsible for the increase in emission. Deconvolution of the emission spectra show two peaks for all concentrations. Considering all the data, it implies that F27 still has complicated dynamics on the PS surface, but the monomers are not affected by an interaction with the PS itself, instead they free to emit and as the film thickness increases the emission grows from both the monomer and eventually the formation of excimers.
The emission data for all dyes suggest that there are interactions occurring for the charged dyes that do not happen with a neutral dye. For both Rh6G and DSF, the phenomena of aggregation was mentioned, however it is unlikely that the short lifetime originates from the aggregated species due to the excitation wavelength of the laser used to record the decays. The laser emits 452 nm light and in that wavelength region the aggregated species of both dyes has little to no absorption and by effect no emission either. Furthermore, there is very little aggregate formed on the surface at this low of a casting concentration. Excimers have been shown to have longer lifetimes than their respective monomer components due to a mismatch in franck-condon states, which rules out that interaction being assigned to the short lifetime. 25,26 Therefore, by process of elimination the only other source of interaction for a monomer DSF or Rh6G molecule must be with the PS surface itself, which is produced by the coordination from ion-π interactions.

Conclusion
This report investigates the potential for ion-π interactions between charged organic dyes by comparing the spectral response of a cationic dye (Rh6G), an anionic dye (DSF), and a neutral dye (F27) on the surface of a polystyrene film. Absorbance spectroscopy was used to investigate the state of the dyes on the PS surface. It was found for all the dyes that a lower energy absorbing species is formed at the lowest casting concentration, which is attributed to either dye/PS or dye/dye interactions. The interaction energy of the dye complex, whether it is from an interaction between the dye and PS or two dye molecules, is proportional to the spectral shift of the aggregated peak from the monomer peak and can be used as an indication of ion-π interactions.
The spectral shifts, from most intense to least intense, follow the trend of: Rh6G > DSF > F27. Due to the fact that all three dyes share similar chemical structure that includes a π rich xanthene plane, the increase in interaction energy can be attributed to stabilization by ion-π interactions. Furthermore, the use of fluorescence spectroscopy can be used as an additional tool to identify potential ion-π interactions between the charged dyes and the π rich PS surface. By analyzing the trends in emission intensity as a function of dye film casting concentration and correlating these trends with the excited state lifetimes it was found that a large portion of monomer emission from the charged dye films is quenched by the formation of a weakly emissive exciplex. No evidence of the exciplex is found for the F27, which suggest that ion-π interactions are responsible for the formation of the exciplex between an excited Rh6G or DSF molecule and the PS surface.

Supporting Information
The Supporting Information is available free of charge on the ACS Publications website. Deconvoluted absorbance and emission spectra for the three different dyes.

Notes
The authors declare no competing financial interest.           Table S1. Deconvolution parameters for the absorbance and emission spectra of dye thin films on PS. λ max represent the peak position and Γ is the FWHM. Concentration range indicates the range in which the corresponding peak was required to fit the spectrum of the film cast from that concentration.