Effect of Surface Chemistry on the Glass-Transition Dynamics of Poly(phenyl methyl siloxane) Confined in Alumina Nanopores

Broadband dielectric spectroscopy (BDS) and differential scanning calorimetry (DSC) are combined to study the effect of changes in the surface chemistry on the segmental dynamics of glass-forming polymer, poly(methylphenylsiloxane) (PMPS), confined in anodized aluminum oxide (AAO) nanopores. Measurements were carried for native and silanized nanopores of the same pore sizes. Nanopore surfaces are modified with the use of two silanizing agents, chlorotrimethylsilane (ClTMS) and (3-aminopropyl)trimethoxysilane (APTMOS), of much different properties. The results of the dielectric studies have demonstrated that for the studied polymer located in 55 nm pores, changes in the surface chemistry and thermal treatment allows the confinement effect seen in temperature evolution of the segmental relaxation time, τα(T) to be removed. The bulk-like evolution of the segmental relaxation time can also be restored upon long-time annealing. Interestingly, the time scale of such equilibration process was found to be independent of the surface conditions. The calorimetric measurements reveal the presence of two glass-transition events in DSC thermograms of all considered systems, implying that the changes in the interfacial interactions introduced by silanization are not strong enough to inhibit the formation of the interfacial layer. Although DSC traces confirmed the two-glass-transition scenario, there is no clear evidence that vitrification of the interfacial layer affects τα(T) for nanopore-confined polymer.


■ INTRODUCTION
When going down with the size of a soft matter to a nanometre size, its physical and chemical properties drastically change. For that reason, scientists are incessantly interested in obtaining novel nanomaterials with remarkable morphologies or structures that, in many cases, cannot be achieved by any other means at the macroscale. Such nanomaterials can find numerous promising applications in drug delivery systems, coatings, sensors, electronic devices, and many others. 1−6 One of the very useful strategies allowing for the study of the properties of polymers and molecular liquids at the nanoscale level is by constraining them within solid interfaces of nanometer size in either one-(thin films), two-(nanochannels/cylindrical pores), or three-dimensions (suspended nanoparticles or droplets). Depending on the boundary conditions, one can also discriminate between soft and hard confinement. In the case of soft confinement, the investigated material is more viscous than the confining environment. Some examples are polymer nanoparticles prepared in aqueous suspensions or microemulsion droplets. On the other hand, in the case of hard confinement, the sample is located within the rigid porous matrix/nanochannels or supported on hard substrate (e.g., aluminum or silicon). 7−9 Irrespectively of the given geometry, it is generally established that the changes in the dynamics and glass transition behavior under nanoscale confinement are controlled by two factors: decreasing length scale and the increasing importance of the surface effects. Studies on thin polymer films as well as molecular systems embedded within nanopores have demonstrated that the interplay between both factors might result in either decrease, increase, or no change on the glass-transition temperature compared with the bulk material. 7,10−22 In addition to that, theoretical and experimental results leave little doubt that the dynamics of confined material is strongly heterogeneous and subdivided into different fractions. In nanopore confinement, this includes a core fraction with enhanced mobility and a surface layer with dynamics slowed down by the interactions with the solid interface. In turn, in thin films, the presence of air/polymer interface can enhance the overall dynamics, while at the same time the polymer/substrate interface strongly retards it. 16,17,23,24 The manifestation of such two coexisting layers with much distinct mobility is the double glass transition event, detected in the temperature region above and below the bulk T g . 25,26 The one located above bulk T g comes from the nearsubstrate dynamics, while that one below bulk T g , from the enhanced dynamics in the middle of the nanopore or either free-surface interface. The origin of the gradient of the mobility is related to frustration in the molecular packing arising from the interactions with the confining surface. For glass-forming liquids embedded within nanometer pores, it has been demonstrated that vitrification of the molecules at the interfacial layer introduces changes in the packing density (more free volume), which in turn is responsible for enhanced mobility and deviation of τ α (T) from the bulk behavior at lower temperatures. 27−31 Perturbation of the density near the confining surface and interfacial interactions also plays an important role in understanding the thin film dynamics. 32−34 The strength and the type of interactions that occur between the confined sample and constraining environment play a key role in determining various properties of such materials, including glass transition and crystallization behavior, so it can also lead to promising applications in nanotechnology. This is one of the main reasons for numerous studies aimed to modify the confining surface conditions. 35−42 One of the strategies that can be used to tune the interfacial effects involves chemical or physical modification of the pore walls. Silanization, which replaces the hydroxyl groups at the surface by various silane agents, is a common strategy employed to render the surface more hydrophobic. As a result of such substitution, the effects of confinement on the glass transition dynamics may be partially or even completely lost. 19,21,31,43,44 Such feature is typically related to the lack or either difficulty in forming the interfacial layer/or weaker interfacial interactions due to more hydrophobic surface of the pore walls. For example, Zhang et al. demonstrated that silanization of anodic aluminum oxide (AAO) nanopores with the use of hexamethyl disiloxane weaken interactions with confined material but still leads to the formation of the interfacial layer. 45 On the other hand, the coating of the nanopore walls with more hydrophobic trichloroperfluorooctylsilane introduces repulsive interactions and prevents the formation of the interfacial layer. As a result, only one glass transition event is detected. 45,46 Modification of the surface conditions can also be introduced using ODPA (octadecylphosphonic acid), which may affect the crystallization tendencies of a confined substance. 37 Likewise, changes in the crystallization were observed for syndiotactic polystyrene in anodized aluminum oxide when the nanopore surface was coated with an alkyl monolayer. 36 Another promising strategy for the modification of the physical and chemical properties of the nanoscale environment is atomic layer deposition (ALD) which enables covering of the solid substrate, such as thin films or either inner pore surface, with nanometer-thick layers of metals, oxides, or nitrides. 47 In this work, we investigate the effect of changes in the surface chemistry on the glassy dynamics of polymethylphenylsiloxane (PMPS) confined in anodic aluminum oxide nanopores. To do that, we have employed dielectric spectroscopy and differential scanning calorimetry. The surface of the alumina nanopores was modified using chlorotrimethylsilane (ClTMS) and (3-aminopropyl)trimethoxysilane (APTMOS). The difference between both silanizing agents is that the former replaces the hydroxyl groups with trimethylsilane unit, while the latter with aminopropylsilane species. As reported in the literature, APTMOS molecules can form cyclic structures or 3-dimensional networks, making the surface morphology more complex. Such a network is relatively loose under ambient conditions but tightens with thermal processing. Thus, improving the stability of the polymer thin films. 48 Our interest in APTMOS, as a silanizing agent for PMPS confined within AAO nanopores, also stems from its specific interfacial interactions. As reported in the literature, in some cases, APTMOS is able to embrace the polymer chains within its complex network or even form the covalent bonding with some polymers, like polydimethylsiloxane. 48,49 Such complex surface characteristics were observed for silicon substrates with APTMOS based coating that contains only two or three monolayers. 48 Our research focuses on the effect of changes in the surface conditions in 55 nm pores on the temperature evolution of the segmental relaxation time and its distribution, as well as the equilibration kinetics. The results show that the τ α (T) characteristic for the bulk material is restored upon annealing with the time constant that does not depend on the specific surface conditions. We also demonstrate the presence of two glass-transition events in DSC thermograms of all considered systems, which means that the changes in the interactions between the polymer and silanized pore walls are not strong enough to inhibit in any case the formation of the interfacial layer.
■ EXPERIMENTAL SECTION Materials. The tested polymer is poly(phenyl-methyl-siloxane) labeled in the texted as PMPS 2.5k, with M n = 1800, and polydispersity index (PDI) = 1.40. We demonstrate the molecular structure of PMPS in Figure 1. The sample was purchased from Polymer Source Inc. (Canada) as a clear, viscous, transparent liquid and used without further purification. The glass-transition temperature of bulk PMPS 2.5k determined from the broadband dielectric spectroscopy (BDS) is T g = 230 K (T g = T at which τ α = 1 s). While from the differential scanning calorimetry (DSC) we get T g = 230 K. Numerous studies report T g value for PMPS which increases with the molecular weight. 50 AAO Templates and Method of Infiltration. Native AAO Nanopores. We have used commercially available anodized aluminum oxide membranes (Synkera) composed of uniform arrays of unidirectional and non-cross-linking nanopores (pore diameter of 55 ± 6 nm, pore depth 53 ± 1 μm). The diameter of the alumina membrane is 13 ± 0.2 mm, and its porosity is 13%. In this study, we also use AAO membranes purchased from InRedox (pore diameter of 80 ± 9 nm, pore depth 100 ± 5 μm). The diameter of InRedox AAO membranes is 13 ± 0.1 mm, and its porosity is 15%. Before filling, AAO membranes were dried at 473 K in a vacuum oven for 24 h to remove any volatile impurities from the nanochannels. In the next step, PMPS was placed on the top of the AAO membranes, and then the entire system was kept at T = 353 K under vacuum for 2 days. This allows the liquid to flow inside the nanopores by capillary forces. The membranes were weighed before and after infiltration. We have Langmuir pubs.acs.org/Langmuir Article assumed that the filling is completed once the mass of the membrane ceased to increase. Then, the surface of the membranes was dried using delicate dust-free tissues. Silanized AAO Membranes. For silanization, we have used AAO membranes (Synkera) characterized by the same paraments as described above: diameter 13 ± 0.2 mm, thickness: 53 ± 1 μm, pore size: 55 ± 6 nm, pore density: 6 × 10 9 cm −2 , and porosity: 13%. Silanization of the AAO matrices was carried out under argon atmosphere using a vacuum line. The AAO matrices were silanized using two surface modifiers with the highest purity available: chlorotrimethylsilan (ClTMS) and (3-aminopropyl)trimethoxysilane (APTMOS). These reagents were purchased from Sigma-Aldrich and used as supplied. Just before the silanization process, we have dried AAO matrices under vacuum in the temperature of 423 K for 24 h to remove any excess of water from the pores. Then, the completely dry matrices were immersed in the solution of ClTMS or APTMOS in toluene (3% of volume), in the Teflon chamber of hydrothermal synthesis autoclave reactor. Subsequently, we have placed the dish containing the samples in a vacuum chamber and degassed under the pressure of 0.2 Pa for 1 min. Next, the pressure was gradually increased. This process assures complete filling of the AAO channels by the solution. Then, we have closed the dish in the autoclave and placed it into the thermal chamber for 24 h at the temperature of 343 K. After this, the samples were rinsed with toluene, immersed in the pure toluene again, and put into an ultrasonic bath for 1 h to remove any excess of the ClTMS and APTMOS and to avoid polymerization.
We have repeated the procedure twice. Afterward, the samples were dried in a vacuum overnight. A very simple scheme demonstrating the silanization procedure is illustrated in Figure 2. The difference between APTMOS and ClTMS is that the former was reported to form cyclic structures or 3-dimensional networks which might be anchored to the substrate, while the latter replaces only the hydroxyl groups with trimethylsilane units. The estimated value of the dipole moment for ClTMS is 1.29 D, while for APTMOS it is 1.09 D. Before infiltration, silanized membranes were dried at 373 K in a vacuum oven for 7 h. Subsequent stages of filling the silanized nanopores were the same as for native AAO nanopores.
The porous aluminum oxide matrices are functionalized very precisely by individual molecules of aminopropyl silane. Practically, we should not mention this as a layer. However, even in the case of a very dense covering of the surface by aminopropyl silane molecules, their length is about 6 Å. Taking under consideration the diameter of alumina pores of 55 nm, the influence of the functionalization on the pores volume is negligible. As we do not know the grafting layer density, we assume the pore size and porosity as determined for the native membranes.
Methods. Water Contact Angle. The wettability of native and silanized membranes (same bath and pores size) was evaluated by contact angle (CA) measurements using a drop shape analysis instrument (JC2000D Contact Angle Tester) under ambient humidity and temperature. Figure 3 demonstrates the shapes of a water droplet on (a) a native AAO membrane surface and modified  Langmuir pubs.acs.org/Langmuir Article with either (b) APTMOS or (c) ClTMS. As can be seen, the hydrophilic character of AAO membranes (44.68°) can be significantly affected by the choice of the modifying agent. APTMOS treatment increase hydrophilicity of the nanopore surface, resulting in a decrease of the contact angle value (28.16°). On the other hand, the nanopore surface treated with ClTMS becomes more hydrophobic. In this case, the water contact angle reaches almost ∼90°. Thus, the change of the nanopore surface character from more hydrophilic to hydrophobic can be arranged in the following order APTMOS < native < ClTMS. Increased adhesion of APTMOS coating relies on the specific structure of the molecule, which contains active −NH 2 terminal groups that interact with the surface hydroxyl groups and the hydrolyzed hydroxyl groups (from the methoxy group of the APTMOS molecules). In this way, APTMOS molecules form complex cyclic structures or the disordered polymerized 3d-network on the inorganic substrate. 48 Broadband Dielectric Spectroscopy (BDS). Dielectric spectroscopy measurements for bulk and nanopore-confined PMPS were made with Novocontrol Alpha frequency analyzer. For bulk sample, we use standard plate−plate electrodes of 20 mm in diameter separated by a 50 μm Teflon spacer. Native and silanized alumina membranes AAO filled with the investigated polymer were placed between two round electrodes with a 10 mm diameter. Bulk and nanopore-confined samples were measured as a function of temperature in the frequency range from 10 −2 Hz to 10 6 Hz. The temperature was controlled with stability better than 0.1 K by the Quatro system. The complex dielectric permittivity ε* = ε′ − iε″, where ε′ is the real and ε″ is the imaginary part, were collected on (i) slow cooling with 0.2 K/min from 293 to 217 K and (ii) slow heating from 217 to 293 K with 0.2 K/min followed by 10 K/min cooling step. The time-dependent measurements were also carried out using Novocontrol Alpha Analyzer for a period of 35 h at T = 231 K. Thermal protocol for such experiments involved fast cooling from the room temperature to a selected annealing temperature.
Analysis of the Dielectric Permittivity for the Polymer Confined in Alumina Nanopores. Herein, it should be noted that the nanopore-confined system under consideration is an inhomogeneous dielectric that includes tested polymer located inside the alumina matrix. Because the electric field runs along the nanopore channels, the entire heterogeneous dielectric response problem can be modeled using the equivalent circuit composed of the two capacitors connected in parallel. In such a case, the dielectric permittivity of a composite material (the raw data that we measure using impedance analyzer) is the sum of the dielectric permittivity of the individual components, confined polymer and alumina matrix, weighted by the respective volume fractions where φ is the porosity of the alumina membrane, ε AAO is the dielectric permittivity of the alumina membrane, and ε polymer is the dielectric permittivity of the confined polymer. Thus, for the real and imaginary parts one gets To characterize the dielectric permittivity of the matrix, we have measured the dielectric signal of the empty AAO membranes. The data presented in Figure 4a demonstrates that the self-ordered native and silanized AAO membranes with the pore diameters of 55 nm show frequency invariance of ε′ and can be considered as loss-free. In addition to that, the values of ε′ are almost temperature-independent, which is illustrated in Figure 4b for alumina templates silanized with ClTMS.
Because empty membranes are filled with air (ε′ air = 1), finding the permittivity of the bare alumina matrix, ε AAO , will also employ eq 2. The porosity of AAO membranes with 55 nm pore sizes is 13%, which makes φ = 0.13. The obtained values are as follows ε′ AAO = 4.1, 4.5, and 4.8 for native, APTMOS, and ClTMS-treated alumina membranes. In the next step, the correction of the dielectric data for PMPS 2.5k confined in such nanoporous templates was carried out accordingly with eq 2. Figure 5 compares the pure polymer contribution with that of the polymer-matrix composite material. The representative dielectric curve for the tested polymer confined in 55 nm alumina nanopores with ClTMS coating is only shifted toward higher values of ε″, as it depends only on the porosity of the membrane. The position of the maximum and the breadth of the αloss peak remains the same. This is in line with the work by Alexandris  Langmuir pubs.acs.org/Langmuir Article and co-workers (see supporting materials therein) demonstrating that the only variable affected in such geometry is the absolute value of the dielectric permittivity. 50 It should be noted that the scenario described above refers to the ideal situation when the nanopores are completely filled with the investigated sample. However, in many cases, especially due to high viscosity, the polymer imbibition within AAO nanopores is greatly impeded. In such a case, it is impossible to avid air-gaps inside the nanochannels, and additional corrections for some insulating blockage within the pore are needed. This can be done according to the procedure described in our recent paper. 55 However, assuming that the nanopores are filled with PMPS 2.5k, only up to 90%, we also expect no shift of the α-peak and spectral broadening (see Figure 5, "pure polymer, porosity, and air gaps corrections").
Differential Scanning Calorimetry (DSC). Calorimetric measurements were carried out by using a Mettler-Toledo DSC apparatus equipped with a liquid nitrogen cooling accessory and an HSS8 ceramic sensor (heat flux sensor with 120 thermocouples). Temperature and enthalpy calibrations were performed by using indium and zinc standards. Crucibles with prepared samples (bulk or either crushed alumina membranes containing confined PMPS) were sealed and cooled down to 183 K with the rate of either 0.3 or 10 K/min inside the DSC apparatus. Then, DSC thermograms were recorded on heating with a rate of 10 K/min in the temperature range from 183 to 293 K. T g values were determined from heat flow data as the point corresponding to the midpoint inflection of the extrapolated onset and end of the transition curve.

■ RESULTS AND DISCUSSION
We start our investigation by characterizing the glass-transition dynamics of bulk PMPS 2.5 k. For that, dielectric and calorimetric measurements were performed. The correspond-ing dielectric loss spectra measured above and below the glass transition temperature are presented in Figure 6a. As can be seen, except for the dominant α-relaxation associated with the glass-transition, the two other relaxation processes can be detected in the loss spectra of the tested polymer. The slower one, labeled here as the α′-relaxation, appears at higher temperatures, and it is interpreted as the sub-Rouse mode. The faster one, seen at lower temperatures in the supercooled liquid and glassy state, is the β-relaxation. The properties of both processes found on the low-and high-frequency flanks of the segmental (α) relaxation of bulk PMPS are discussed elsewhere. 56 In Figure 6b−d, we present the dielectric loss spectra of the composite materials measured at different temperatures for the polymer embedded within 55 nm native and silanized AAO nanopores. As can be seen, under confinement, we observe only αand α′-relaxation process. The α′-relaxation is very broad and hardly perceivable in the loss spectra of the confined system. In turn, the secondary β-relaxation process is no longer resolved as a pronounced loss peak in nanopores. For PMPS 27.7k (M w ) confined in AAO nanopores, we have also reported the same feature. 54 Suppression of the β-relaxation under confinement was also seen in the literature, in vapor-deposited glasses or epoxy resins confined in alumina nanopores. 57,58 This effect can be connected to the changes in density or additional interactions with the host material which may affect the geometry and conformation of the molecules, as well as the variation in the angle of dipole moment librations, among others. In addition to that, the results depicted in Figure 6 where ε ∞ is the high frequency limit of the permittivity, Δε is the dielectric strength, a and b are the shape parameters, σ 0 is the dc-conductivity, τ HN denotes the relaxation time, and ω is the angular frequency (ω = 2πf). In HN function the characteristic time constant, τ HN , is related to the maximum of loss peak frequency f max by the following relation 60 The contribution of αand α′-processes were single out from the fitting procedure, which has included superposition of 2 HN functions. From that, the relaxation times were obtained by calculating τ max = 1/(2πf max ). The temperature dependences of the α-relaxation time obtained in this way are presented in Figure 7a,b. Due to very weak intensity and broad distribution of the α′-loss peak in nanopores obtained dependences were excluded from further consideration. Figure 7a demonstrates τ α (T) for PMPS 2.5k in bulk and nanopores that were determined after heating with 0.2 K/min following cooling step with 10 K/min from 293 to 217 K. The segmental relaxation for bulk polymer exhibits Vogel− Fulcher−Tammann (VFT) behavior and be described using the following relation 61 where τ ∞ is limiting relaxation time at very high temperatures, T 0 is "ideal" glass temperature often termed as Vogel temperature, while B is the activation parameter. The results show that the segmental process in nanopores follows bulk behavior at higher temperatures. As the temperature decrease, τ α (T) starts to deviate from the bulk VFT dependence. The dynamics associated with the glass transition systematically speed up with the increasing degree of confinement, i.e., lowering the pore size (see results for PMPS 2.5k confined in 80 and 55 nm native pores). This is a typical feature reported for glass-forming substances in confined geometry. 21,27,31,38,44,64−67 However, the origin of such enhanced dynamics in nanopores, as compared to the bulk, is often interpreted in different ways. This includes, for example, approaching the length scale of cooperative dynamics, 44 dynamic exchange between the surface layer and free molecules, 68 frustration in the density, 38,69,70 or crossing a spinodal temperature. 71 By combining together dielectric relaxation and calorimetric studies, it has been demonstrated that the temperature at which vitrification of the interfacial layer takes place is signified by the departure of τ α (T) from the bulk dependence. 27,28,30 For PMPS 2.5k confined in 80 and 55 nm native pores, such deviation is observed below 239 and 245 K, respectively. Thus, with decreasing the pore size, confinement effect seen in temperature evolution of the segmental relaxation is more pronounced. Subsequently, we have repeated the same experiment using 55 nm pores in which their inner surface was silanized with ClTMS and APTMOS. The results are also shown in Figure  7a. In contrast to the case of PMPS 2.5k confined in native The a(T) dependence for the tested polymer as obtained in 55 nm native, APTMOS-and ClTMS-treated pores depending on the thermal protocol. The two considered protocols involve (i) slow cooling with 0.2 K/min from 293 to 217 K and (ii) slow heating from 217 to 293 K with 0.2 K/min followed by cooling step with 10 K/min. When error bars are not displayed, the standard error is smaller than the symbol size. Langmuir pubs.acs.org/Langmuir Article pores, the segmental process in ClTMS-treated membranes is virtually unaffected by confinement and shows no systematic crossover of τ α (T). On the other hand, in APTMOS-treated nanopores, the temperature dependence of the segmental relaxation is enhanced compared to that in bulk. However, the results are also different from those reported for native pore surfaces, as the departure from the bulk-like dynamics is now shifted toward lower temperatures. By comparing τ α (T) obtained for PMPS 2.5k confined within native and silanized pores, we conclude that in the case of pristine alumina templates, the change in the pore size from 55 to 80 nm produces virtually the same result as when modifying solely the inner surface of 55 nm pores with the use of APTMOS molecules. Our observation is in line with the literature data demonstrating that silanization of silica pore surfaces allows us to remove partially or even completely the confinement effect seen in temperature evolution of α-relaxation time. 19,31,43,44 Such a feature is typically related to the lack or either difficulty in forming the interfacial layer/or weaker interfacial interactions due to more hydrophobic surface of the pore walls. Experimental evidence suggests that the glass-transition dynamics in nanopores depends on the thermal treatment. 40,45,57,70−72 The results demonstrated in Figure 7b show that segmental mobility for PMPS 2.5k confined in native and silanized nanopores relies on the thermal protocol. Except for the τ α (T) recorded upon heating with 0.2 K/min following cooling with 10 K/min from 293 to 217 K, we have also conducted dielectric relaxation measurements in a reverse direction, i.e., when decreasing temperature from 293 to 217 K with the rate of 0.2 K/min. As can be seen, on slow cooling τ α (T) in 55 nm pores approach the bulk behavior, irrespectively of the surface conditions. Thus, when provided enough time, confinement effect seen in temperature evolution of the segmental relaxation can be eliminated entirely. The importance of the cooling rate and nonequilibrium phenomena that occur when the interfacial layer is immobilized on the time scale of the viscous flow of the core fraction is also supported by combined dielectric relaxation and solvation dynamics studies. 73,74 We will explore this issue more in the following part of the paper.
Besides the enhanced dynamics, another meaningful change in the behavior of PMPS 2.5k confined in alumina nanopores involves broadening the distribution of the α-relaxation time relative to the bulk (see Figure 8). Such spectral broadening is widely reported for the confined systems and indicates that in a spatially restricted environment, the heterogeneity of the relaxation dynamics increases. 60,67,75−77 The literature data show that the broadening of the α-relaxation peak decreases or can be even eliminated by the silanization of the inner pore walls. 19,67 However, the results present in Figure 8 shows the opposite trend. The distribution of the α-relaxation time in ClTMS-treated nanopores is practically the same as for the pristine pore surface, while it is evident that the hydrophilicity of both membranes is much different. Stronger interactions between the surface of the pores and confined molecules are expected in the former case. In turn, in the presence of more hydrophilic APTMOS coating, the spectral broadening increases even more. To describe such broadening of the αloss peak more quantitatively, we use of the fractional exponent β KWW from the Kohlrausch and Williams and Watts function 78,79 where β KWW changes from 0 to 1. The value of β KWW decreases with increasing the width of the relaxation spectrum. At 245 K, the stretching exponent for PMPS 2.5k confined in 55 nm native and ClTMS-treated pores is 0.36, while for more hydrophilic APTMOS surface, we get 0.28. Obtained values are much lower, compared to the bulk polymer of approximately the same segmental relaxation time (β KWW = 0.44 at 243 K). As we suppose, the broadest distribution of the relaxation time observed for APTMOS functionalization might arise from the complex properties and specific interfacial interactions between the polymer and APTMOS molecules. 48,49 At this point, we also wish to remark that the distributions of the segmental relaxation time in 55 nm native and ClTMS-treated pores are practically the same, while their τ α (T) does not necessarily show the same features. From that, it can be concluded that the changes in the interaction of the polymer segments with the confining surface might affect various aspects of the relaxation dynamics in entirely different ways.
As a next step, we have investigated the effect of changes in the surface chemistry on the out-of-equilibrium dynamics of nanopore-confined polymers. This is an important aspect as polymers often fail to equilibrate on the time scale of a typical processing experiment, and exhibits nonequilibrium behavior. 80 To do that, we have performed annealing experiments carried out for up to 35 h at 231 K, i.e., in the temperature region at which deviation from the bulk dynamics is expected. The samples were rapidly cooled from room temperature to selected annealing temperature and measured as a function of time employing the BDS technique. In Figure 9a, we have plotted changes in the dielectric loss spectra of the studied polymer confined to 55 nm silanized CITMS pores with time.
The collected results demonstrate that upon annealing, the αrelaxation peak shifts toward lower frequencies. This effect reflects the slowing down of segmental mobility with time. Additionally, in Figure 9b, we have compared the shapes of the α-relaxation peak at the initial and final stages of the annealing experiment carried out at 231 K. Note that the distribution of the relaxation times slightly narrows with time, but does not Figure 8. Comparison of the shape of α-loss peak for PMPS 2.5k confined in 55 nm native and silanized pores. The bulk spectrum collected at T = 243 K was used as a reference. For the studied polymer confined in 55 nm pores spectra were chosen to match approximately the same τ α . Dashed lines represent KWW fits to the data.
Langmuir pubs.acs.org/Langmuir Article reach that characteristic for the bulk polymer. Qualitatively, the same scenario was also observed for PMPS 2.5k confined in native and APTMOS functionalized membranes. However, at this point, it is essential to realize whether the changes in the surface conditions affect the equilibration kinetics, that is, the corresponding time constant and the final equilibrium state that the confined polymer can reach.
To study the effect of annealing on the segmental dynamics, we have analyzed time-dependent changes in the α-relaxation time at 231 K, as presented in Figure 9c. The evolution of the α-relaxation time upon annealing of confined PMPS 2.5k was parametrized using the stretched exponential function (eq 6) from which the annealing time constant, τ ANN , was obtained. The values of the fitting parameters are collected in Table 1. As can be seen, the characteristic equilibration constants are almost the same for PMPS 2.5k confined in native and silanized nanopores. Through observation, the starting point, in Table 1, and the following recovering behavior from nonequilibrium state to the bulk state, suggests that the mechanism which governs the equilibrium kinetics related to exceedingly slow viscous is practically independent of the surface chemical treatment.
By analyzing τ α (T), we also found that nanopore-confined polymer recovers its bulk α-relaxation time characteristic for a given temperature, see Figure 9d. These results agree with the recent studies carried out in 2D confinement, 57,70 but the same effect was observed in polymer thin films. 81−85 Thus, the confinement effect seen in τ α (T) can be eliminated with time. Upon such an equilibration process, it is expected that the polymer segments rearrange to attain more dense packing. By the same token, they lose the excess of the free volume being the source of the enhanced dynamics. Noteworthy, after subsequent heating of the equilibrated samples, τ α are slightly higher than the corresponding bulk values (see Figure 9d).
In order to complement the results derived from the dielectric relaxation studies, we have also performed calorimetric measurements. The corresponding DSC thermograms recorded for PMPS 2.5k in bulk, and 55 nm native and silanized nanopores are presented in Figure 10. For bulk sample, it only appears one endothermic event related to the vitrification process (T g = 230 K). On the other hand, for PMPS 2.5k embedded in 55 nm native and silanized pores, the two glass-transition events are observed. The first one that occurs at lower temperatures corresponds to the glass transition of the molecules at the center of the pores. While the other process, located at higher temperatures, is associated with the vitrification of the interfacial layer. The presence of  Langmuir pubs.acs.org/Langmuir Article two glass-transition events in DSC thermograms of molecular systems and polymers confined in nanoporous alumina has been reported many times in the literature as the evidence for the two-layer model. 26,27,40,45,57,71,72,86 On the other hand, one may expect not to see two T g 's in silanized nanopores. The interfacial interactions in such cases are expected to be much weaker compared to native alumina surface, and the concentration of surface hydroxyl groups, responsible for the hydrogen bonding with confined molecules, will also be greatly reduced.
To further explore this issue, we have investigated the cooling rate dependence of the DSC traces and compare obtained results with the dielectric data. We have performed DSC scans using two different cooling rates: 0.3 and 10 K/min followed by heating with 10 K/min. The results collected in Table 2 indicate that slower cooling rates shift the glass transition temperature of the core fraction toward higher temperature irrespectively of the particular surface conditions. As slow cooling is an equivalent of the annealing process, we can imagine that the polymer chains have enough time to locally rearrange and form a more densely packed structure with an increased value of T g . From the calorimetric results, we also found that the changes in the surface chemistry practically does not affect the values of both T g 's.
The presence of the two-glass transition events in DSC thermograms of PMPS 2.5k confined in native and silanized nanopores is quite the opposite to the results of the dielectric studies. As noted before, a characteristic kink in τ α (T) is often ascribed to the vitrification temperature of the interfacial layer. Thus, the bulk-like behavior seen in the temperature evolution of the segmental relaxation time for PMPS 2.5k embedded within 55 nm ClTMS-treated pores might indicate that silanization impedes stable interfacial layer. On the other hand, we had also observed that the segmental dynamics in 55 nm alumina templates show bulk-like features when slowly cooled or annealed below the crossover temperature. This would imply that the evolution of the α-relaxation time for the nanoconfined polymer is, in some way, insensitive to immobilization of interfacial layer at T g_high . Discrepancies between the glass-transition behavior observed in the dielectric and calorimetric studies might probably arise from the decoupling between molecular mobility and glass transition. As reported in the literature, at large confinement length scales, T g and τ α are not equivocally related and could be profoundly decoupled. In such a case, dynamic glass transition temper-  Langmuir pubs.acs.org/Langmuir Article ature determined from the τ α (T) may not necessarily match with that obtained from the conventional DSC. 87−89 The most striking indications decoupling between vitrification kinetics and the α-relaxation were reported for the ultrathin polymer films with markedly depressed T g that exhibit the dominant bulk-like dynamics. Lastly, to gain some insight into changes in the content of the interfacial layer, in particular dependence on the cooling rate and surface conditions, we have estimated it thickness using the heat capacity changes in the two-T g 's region 26 Ä where d is the pore diameter. The results of the calculations are summarized in Table 3 and serve as the evidence that the changes in the interactions between nanopore walls and the confined polymer introduced by silanization are, in any case, not strong enough to prevent the formation of the interfacial layer. We also found that the interfacial layer gradually thickens with the decrease of cooling rate for native and APTMOSfunctionalized nanopores. On the other hand, for PMPS 2.5k located in ClTMS-treated alumina templates, the growth of the interfacial layer is practically independent of the cooling rate. This would mean that there is no exchange process between the polymer chains located in the center of the pores and the interfacial region.

■ CONCLUSIONS
In this work, by employing dielectric spectroscopy and differential scanning calorimetry, we have investigated the influence of surface modification on the segmental dynamics of PMPS 2.5k confined in nanoporous alumina templates. For that, we have used two quite different silanizing agents, ClTMS and APTMOS. The former, more hydrophobic, replaces the hydroxyl groups attached to the pore walls with trimethylsilane units. While the latter one was expected to form more complex cyclic structures or three-dimensional networks, as reported for thin films prepared on silicon substrates. However, based on collected data, we instead exclude this situation for the ATMOS-based coating of the alumina pore walls. The results of the dielectric studies reveal that for the studied polymer located in 55 nm pristine and silanized nanopores, the τ α (T) shows a strong dependence on the thermal treatment. The bulk-like evolution of the segmental relaxation time can be restored upon decreasing the cooling rate or long-time annealing. Interestingly, the time scale of such an equilibration process was found to be independent of the particular surface chemistry. In contrast to temperature evolution of the segmental relaxation time, the confinement effect seen in broadening the α-loss peak is not eliminated by chemical modification, the surface of the pore, nor prolonged annealing experiments. In contrast to dielectric relaxation studies, the results of the calorimetric measurements reveal the presence of two glasstransition events in DSC thermograms of all considered systems, meaning that the changes in the interfacial interactions introduced by both silanizing agents are not strong enough to inhibit the formation of the interfacial layer. Although the two-glass-transition scenario was confined by DSC traces, there is no clear evidence that the immobilization of the interfacial layer at T g_high has any impact on the τ α (T) in confined geometry.
It should be noted that the silanization is, in general, much less effective in terms of hydrophobic functionalization than ODPA-treatment of AAO. In this regard, the distinct quality of the pore-wall functionalization may also explain the partially contradicting results found for the glassy dynamics reported here. Just as reported by Sentker et al. 90 in the case of the molecular anchoring and the phase behavior of the liquid crystals in ODPA-treated AAO nanopores. Specifically, the chemical pore-surface grafting with silanes never produces edge-on anchoring of the discotic liquid crystals. On the other hand, more robust and homogeneous hydrophobic surfaces obtained by the phosphonic functionalization enforces such edge-on anchoring in AAO nanopores. Some experiments using surface-sensitive X-ray diffraction show that in a planar silica surface 91,92 and silica nanopores 93 the silanization treatment works better on the hydroxylated silica surface. In opposite, on planar aluminum oxide surfaces, the phosphonic acid linking gives more ordered self-assembled monolayers than silanization treatment. 92,94 Our results might contribute to a better understanding of the glass-transition dynamics of nanopore confined polymers, especially showing that the behavior of nanopore-confined systems is strongly affected by the thermal treatment, surface chemistry or interfacial interactions. Knowing the effect these factors will help to design polymer nanomaterials with controlled physical properties and stability required for a number of applications.