Impact of Interface Modification on the Behavior of Phenyl Alcohols within Silica Templates

: Herein, thermal and dynamical properties, as well as host − guest intermolecular interactions, and the wettability of a series of monohydroxy phenyl-substituted alcohols (PhAs) infiltrated into native and silanized silica mesopores (of pore diameter, d ∼ 5 nm) were investigated by means of dielectric and infrared (IR) spectroscopy, differential scanning calorimetry, and contact angle measurements. Calorimetric data showed the occurrence of the two glass transition temperatures, T g . Importantly, around the one detected at higher temperatures ( T g,interfacial ), a strong deviation in the temperature evolution of the relaxation time of the main process was observed for all systems. Moreover, an additional mode unrelated to the mobility of the interfacial layer and core molecules was revealed. One can suppose that it could be connected to either the slow Arrhenius process (SAP, previously reported for the polymer thin films) or the mobility of hydrogen-bonded structures. Further, IR investigations showed that the applied nanoconfinement had little impact on the strength of the hydrogen bonds (HBs), but it influenced the HBs’ distribution (including the ’new’ population of HBs) and the degree of association. Additionally, for the first time, we calculated the activation energy values of the dissociation process for PhAs in mesopores, which turned out to be lower with respect to those estimated for bulk samples. Thus, our research clearly showed the impact of the spatial geometrical restriction on the association process in alcohols having significant steric hindrance.


INTRODUCTION
Geometric nanoconfinement is a versatile platform for modifying the structural and dynamic properties of soft matter systems, making them highly relevant for nanoscale applications.Previous investigations have revealed that materials adjacent to a solid and near-surface can exhibit interfacial regions of reduced density, 1−3 interfacial freezing, 4−6 interfacial melting, 7−10 molecular layering, 11−14 molecular orientation, 15 or specific lateral molecular arrangements. 16−28 Due to these facts, confined systems hold promise for diverse industrial applications, including nanolithography, lubrication, paints, surface treatments, and elastic membranes.
Among many types of nanorestrictions, silica mesoporous materials, such as MCM-41 or SBA-15, possess the O−H groups (silanol groups (�Si−OH)) inside pores that can easily interact through H-bonds with lattice oxygen.To successfully tailor these materials for nanotechnology applications, the membrane characteristic is often adjusted by altering the pore diameter or surface properties.The latter includes postsynthesis modifications, typically involving silanizing agents like methoxytrimethylsilane, trimethyl chloride, 29 mercaptopropyltrimethoxysilane, 30 hexamethyldisilazane, 31 or chlorotrimethylsilane, 32 which change the surface of the silica from hydrophilic to hydrophobic by replacing the O−H groups with organic groups via O−Si−C covalent bonds.This alteration affects the pore structure, which determines the nature of host−guest interactions.To illustrate that, one can refer to the studies on water in hydrophilic and hydrophobic nanochannels.In the former case, IR spectroscopy revealed a predominant icelike structure because of comparable strong interactions between the water molecules and the pore wall (interfacial interactions), 29 while for the latter one, water behaved more liquid-like (bulk-like), as water molecules interacted less with the silica surface (fewer interaction sites at the pore wall) and more with each other. 33Similar trends were also observed for aliphatic mono-and polyhydroxy alcohols 31,34 infiltrated into nanopores, where hydrophobization of the silica surface hinders H-bonds between the pore wall and the guest molecules, making the dynamics of the confined liquid more bulk-like. 31The effects related to the reduction of surface interactions were also seen for hydroxylterminated polymers infiltrated into modulated anodic alumina (AAO) templates. 35In this case, the factor responsible for the disturbance of the interaction at the polymer−membrane interface was the roughness of applied nanotemplates.Importantly, similar results were also reported for PhAs infiltrated within mesoporous anodic aluminum oxide (AAO) membranes having constant and varying pore diameters (d = 10−40 nm).Samples within ordinary AAO templates revealed a pronounced confinement effect on the glass transition temperature, T g , and the structural relaxation process, whereas alcohols in the latter templates showed a bulklike behavior in the whole range of studied temperatures. 36−40 In this paper, we investigated the effect of surface interactions on the behavior of a series of phenyl-substituted monohydroxy alcohols (PhAs) from ethanol to pentanol (see their chemical structure in Scheme 1(a)) incorporated into mesoporous (un)treated silica templates by means of broadband dielectric spectroscopy (BDS), differential scanning calorimetry (DSC), and Fourier transform infrared (FTIR) spectroscopy.The applied silica mesopores are composed of uniaxial nanochannels (open from both sides) with a defined pore size (see Scheme 1(b,c)).The characterization of the produced membranes was made by using scanning electron microscopy (SEM, see Figure S1).The mean diameter of the mesopores, d, was determined as λ ∼ 5 nm.

Sample Preparations.
Prior to filling, the silica membrane was dried in an oven at T = 373 K under vacuum in order to remove any volatile impurities from the nanochannels, and alcohols were subjected to freezing in liquid nitrogen to remove water.After cooling, membranes were infiltrated with alcohols.Then, the alcohol-membrane systems were maintained under vacuum (10 −2 bar) to allow the liquid to flow inside via capillary forces.The complete filling was obtained by weighing the templates before and after each infiltration to a constant mass.After the infiltration process, the excess of the sample on the surface was removed.The pore diameter of silica nanopores was confirmed by SEM measurements (Figure S1).The filling degree reaches ∼90%.Note that taking into account the porosity of applied membranes (∼10%) together with the cylindrical shape of porous nanochannels (characterized by a pore diameter of d ∼ 5 nm and length, l ∼ 50 μm), we calculated the maximum amount of alcohol, which can be infiltrated within the molecules, and we compared this value with the weight of the infiltrated substance.

Scanning Electron Microscopy (SEM).
The pore distribution images were taken by using a JEOL JSM-7100F TTLs LV/EDS field emission scanning electron microscope (SEM).For the SEM observations, samples in the form of small flakes cut from the received sheet were stuck to the sample holder by a carbon conductive tape.Secondary electron images (SEIs) obtained at magnifications ranging from 35 to 500,000 times were collected by an E-T detector at beam accelerating voltages between 2 and 15 kV.The mean diameter of the mesopores, d, was determined as 4.8(±1.1)nm.The mesopores were measured by using the ImageJ package.The mean diameter of the mesopores was determined by fitting a log-normal distribution to the experimental data (the histogram in Figure S1(c)).Note that in the case of the modified templates, the silanization process (carried out on the native mesopores) does not affect the mean diameter of the applied mesoporous templates, resulting in a similar SEI as in the case of native membranes.

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 (a heat flux sensor with 120 thermocouples).Temperature and enthalpy calibrations were performed using indium and zinc standards.The sample was prepared in an open aluminum crucible (40 μL) outside of the DSC apparatus.Samples were scanned at various temperatures at a constant heating rate of 10 K/min.

Surface Tension and Contact
Angle Measurements.The surface tension of liquids, γ L , and the contact angle, θ, were measured by the drop shape analysis 100S Kruss tensiometer, GmbH Germany, with Advance software.−45 The γ L was measured in the temperature range, T = 293−333 K, with a step of 5 K.For each temperature, several drops of a similar volume (∼2.5 μL) were equilibrated in an argon atmosphere.The temperature measurement uncertainty was ±0.1 K, whereas the uncertainty of the surface tension was ±0.1 mN/m.The contact angle measurements were performed at nine temperatures in the wide temperature range, T = 263−298 K. Thanks to the measuring chamber adopted to work in the wide temperature range with humidity The Journal of Physical Chemistry C control, it was possible to measure the contact angle below the freezing point of water.The contact angle measurements on the solid smooth surfaces have been repeated a dozen or more times.The precision of contact angle measurements was 0.01°, and the estimated uncertainty was not more than ±2°.For the surface energy (γ S ) estimation of alumina and (native and silanized) silica, some of the following liquids were considered: water, ethylene glycol, diiodomethane, and glycerol.The dispersive and nondispersive parts in the surface tension for these substances were taken from ref 40.The calculated surface energy for native silica was γ S = 67.6 mJ/m 2 with the dominant nondispersive part equaling 66.6 mJ/m 2 .For silanized silica, the respective value was γ S = 25.3 mJ/m 2 with the dominant dispersive part equaling 22.8 mJ/m 2 . 46Details are presented in the SI file.
2.6.Broadband Dielectric Spectroscopy (BDS).Measurements were carried out on heating after a fast quenching of the liquid state in a wide range of temperatures (175−243 K) and frequencies (10 −1 to 10 6 Hz) using a Novocontrol spectrometer equipped with an α impedance analyzer with an active sample cell and a Quatro Cryosystem.Dielectric measurements of bulk samples were performed in a parallelplate cell (diameter: 15 mm, gap: 0.1 mm), as described in ref 47.Silica membranes filled with studied alcohols were also placed in a similar capacitor (diameter: 10 mm, membrane thickness: 0.05 mm). 41,42Nevertheless, the confined samples are heterogeneous dielectrics consisting of a matrix and an investigated compound.Because the applied electric field is parallel to the long pore axes, the equivalent circuit consists of two capacitors composed of ε compound * and ε templates * .Thus, the measured total impedance is related to the individual values through 1/Z c * = 1/Z compound * + 1/Z templates * , where the contribution of the matrix is marginal.The measured dielectric spectra were corrected according to the method presented in ref 48.
The normalized α-loss peak for the bulk was fitted to the one-sided Fourier transform of the Kohlrausch−Williams− Watts (KWW) function (the dotted line in Figure 2(c,f)) We found that for bulk PhAs, the fractional exponent, β KWW , which describes the α-relaxation time distribution's nonexponentiality, is equal to 0.9.
In order to obtain relaxation times, τ, the all dielectric data were fitted using the Havrilak−Negami (HN) function 49 where σ DC is the dc-conductivity term, ε f is the permittivity of the free space, Δε is the dielectric strength, ω is the angular frequency (ω = 2πf), τ HN describes the HN relaxation time, and α HN and β HN are the shape parameters.The subscript i refers to each process.Note that the examples of dielectric spectra collected for the bulk 2Ph1E as well as confined within native silica templates fitted to either a single HN function or two HN functions with the conductivity term, respectively, are shown in Figure S6.The temperature dependences of the relaxation times (determined for the bulk Debye-like process) were fitted using the Vogel−Fulcher−Tamman (VFT) function 50−52 = i k j j j j j y where τ 0 is the relaxation time in the limit of very high temperatures, T 0 is the so-called Vogel temperature, and B is The Journal of Physical Chemistry C the activation parameter.T g s of the bulk samples was determined as the temperature at which τ = 100 s by extrapolating the VFT fits.To determine T g,core for the confined systems, their τ Dom (T) dependences below T g,interfacial (after the kink) were fitted with either the VFT function or the Arrhenius equation (dependent on the sample) where k b is the Boltzmann constant, and E a is the activation energy.T g,core values of the confined PhAs were determined as the temperature at which τ = 100 s by extrapolating either VFT or Arrhenius fits.
2.7.Fourier Transform Infrared (FTIR) Spectroscopy.The FTIR spectra of bulk and confined samples were recorded on a Nicolet iS50 FTIR spectrometer (Thermo Scientific, Madison, WI) with continuous dry air purging to eliminate carbon dioxide and water vapor.The measurement parameters were set in OMNIC software (Version: 9.6.251,Thermo Scientific) as follows: the spectral resolution of 2 cm −1 , 16 scans, a gain of 8, and the spectral region of the collected spectra 4000−1300 cm −1 .Due to the saturation of silica membranes, the region below 1300 cm −1 was not analyzed.To monitor the temperature, the Linkam THMS 600 stage was used, which allowed for measurements in the temperature range of T = 297−153 K (with a temperature rate of 4 K/min).Additional purging of the apparatus was applied for 30 min before each measurement.In the first step, the background spectra of an empty pore (Figure S2) were recorded in a wide range of temperatures (T = 297−153 K) to eliminate any influence of the template on the spectra of the incorporated alcohol.Infiltrated alcohols were measured under the same conditions as for empty pores, with matching temperature ranges and spectral parameters.The background spectrum of the "empty" pore was automatically subtracted from the measured spectrum.Initial spectral preprocessing, including baseline correction, was performed with the OMNIC program.A simplified scheme for the IR measurements is presented in Scheme S1.

RESULTS AND DISCUSSION
Representative DSC curves recorded for bulk 4Ph1B and samples infiltrated into various silica membranes are presented in Figure 1(a−c).Note that representative DSC curves recorded for 2Ph1E and 3Ph1P are shown in Figure S4.All bulk samples exhibit the presence of one glass transition (manifested by a characteristic endothermic jump of the heat capacity).Moreover, T g s reveal a slight 'odd−even' effect with the change in the molecular weight of alcohol. 47Values of the glass transition temperatures determined for nonconfined systems are listed in Table S1.On the other hand, for PhAs incorporated within various silica membranes characterized by a defined pore diameter, d ∼ 5 nm (see Figure S1), two (double) glass transitions (DGTs, manifested as two endothermic signals) located above and below the bulk T g were observed.One can stress that herein, the DGT phenomenon was recorded for all examined confined samples, regardless of the length of the alkyl chain of PhAs and the type of templates (native/silanized).−58 The presence of the DGT is assumed to originate from the confinementinduced distribution of interactions within the systems due to the presence of the solid interface, leading to the heterogeneity in terms of molecular dynamics as well as the packing density.Moreover, it can be described by the simple 'two-layer' (or 'core−shell') model, 55,59 which implies that one can distinguish previously mentioned 'interfacial' molecules of reduced mobility with the glass transition occurring at higher temperatures (T g,interfacial ), and the particles located at some distance from the interface (more in the middle of the nanochannels) labeled as the 'core' fraction characterized by lower values of the glass transition temperatures (T g,core ).Importantly, until now, it was assumed that the lack of specific interactions between confined molecules and the interface would result in the absence of a defined interfacial layer.However, recent studies of polar van der Waals liquid, (S)-(−)-4-methoxymethyl-1,3-dioxolan-2-one (labeled as S-methoxy-PC), 60 incorporated within AAO membranes of different pore sizes and modified surfaces clearly show that it is not the case, and even for the silanized hydrophobic surface, the DGT phenomenon (especially the presence of T g,interfacial ) can be seen.A similar situation was found in our study, but it seems that the silanization process of the mesoporous templates affects the DSC signal.Values of determined T g s are presented in Figure 1(d) and Table S1.As shown, T g,interfacial and T g,core decrease slightly (ΔT g ∼ 5 K) with the elongation of the alkyl chain for the samples infiltrated into both types of applied templates.Nevertheless, there is a pronounced difference in T g s between materials infiltrated within either native or silanized silica templates, where the difference between both T g s (defined as ΔT g = T g,interfacial − T g,core ) is higher for alcohols within the latter pores.This possibly originates from the specific H-bonding interactions between native silica and alcohol, which are strongly reduced in the silanized pores.
Next, in order to quantify the volume of the interfacial layer dependently on the surface interactions, we calculated the length scale of the interfacial layer, ξ, as follows: 58 = + where ΔC p,i s are heat capacity changes at the corresponding T g,i (determined from the differences between C p after and before the detected heat capacity jump).Note that eq 5 was applied under the following conditions: the volume of the material in the surface layer is proportional to the step change of its heat capacity; the density of the confined sample is constant along the pore radius; and the pore is cylindrical.Additionally, it should be highlighted that eq 5 is a simple mathematical model, assuming direct proportionality between the heat capacity and the number of molecules, and it does not take into account any variation in the density, roughness, or curvature of applied templates.This might lead to some overestimation of ξ.The determined values of ξ for all examined samples plotted as a function of the alkyl chain are shown in Figure 1(f).As illustrated, the interfacial layer reaches the length scale of approximately ξ ∼ 0.5 nm for all samples incorporated into the native or silanized porous template.For comparison, one can mention that for PhAs infiltrated within the AAO membrane of a significantly higher pore size, d = 10− 40 nm, ξ ∼ 2 to 7 nm. 36Such a high value of ξ seems to indicate up to 10 molecular layers, which might be achieved by The Journal of Physical Chemistry C the formation of some associating structure (most likely loosely packed) near or at the interface.Moreover, one can add that for other MAs, i.e., 2-ethyl-1-hexanol, 2-ethyl-1-butanol, and 5-methyl-3-heptanol, infiltrated into native silica templates of comparable pore sizes (d = 4−8 nm), the length scale of the interfacial layer reaches ξ ≈ 0.5 to 1.4 nm. 40n order to understand why we observe two prominent glass transition temperatures in recorded thermograms and a rather comparable length scale of the interfacial layer in alcohols infiltrated within native and modified silica, we carried out additional measurements of the contact angle, θ, 46,62,63 which can be used to quantify wettability.As one can recall, the smaller θ indicates the better spreading out of a liquid drop on the examined surface (and thus, better wettability).Values of determined θ values for all examined PhAs are shown in Table 1 and Figure S3.All of these measurements were carried out on the planar surfaces, which most likely have different properties than the curved surface of the pore walls (generated during electrochemical etching).Nevertheless, such an approach is well accepted in literature and provides approximated information about the effect of wettability on the dynamics and shift of the glass transition temperature of the confined materials. 64,65As can be seen, all materials are characterized by similar contact angles independent of the examined surface; however, in the case of silanized substrates, values of θ are higher, as can be expected due to differences in the surface interactions.Nevertheless, contrary to the expectations, both surfaces can be considered hydrophilic since for all of them, θ < 90°.Therefore, one can state that there is some difference in the wettability of the material located on native and modified silica, but it is not significant.Moreover, it should be pointed out that the contact angle decreases with a lowering temperature, reaching comparable values θ ∼ 22°and θ ∼ 28°for PhAs spread on the respectively native and silanized silica surfaces at T = 266 K (see Table 1 and Figure S3).That means that at low temperatures, wettability and intermolecular interactions between materials and pore walls are comparable for all systems independent of the applied type of template.Therefore, one might assume that specific H-bonding interactions with pore walls do not play a significant role in the enhancement of the wettability of the PhA in native silica with respect to the samples infiltrated into the silanized template.Nevertheless, a noticeable shift in the T g,interfacial to the higher temperature for the samples infiltrated into former membranes with respect to the modified ones is most likely due to the formation of the strong H-bonds between host and guest molecules, which are not present in the latter system.
Furthermore, we performed comprehensive dielectric measurements.Figure 2(a,b,d,e) illustrate the representative dielectric loss (ε″) spectra obtained for 2-phenyl-1-ethanol (2Ph1E) and 4-phenyl-1-butanol (4Ph1B) incorporated within native and silanized silica templates of d ∼ 5 nm.Data for the nonconfined compound are presented as insets in Figure 2(a,d).Dielectric loss spectra recorded for 3Ph1P and 5Ph1P are shown in Figure S5.As shown, the spectra obtained for bulk samples are composed of two relaxation processes, the dcconductivity at lower frequencies (related to the ion transport) and a prominent relaxation process at higher frequencies that dominates the recorded spectra (characterized by a Kohlrausch−Williams−Watts (KWW) stretched exponent, β KWW ∼ 0.90, suggesting the Debye-like response, see Figure 2(c,f)). 61herefore, in this paper, we label this process as a Debye-like (D) one for the macroscale systems.Note that the loss spectra of various MA often exhibit the presence of a prominent Debye-like peak (originating from the formation of associating structures 66 ), whereas the structural (α) process (cooperative motions of molecules) often manifests itself as an excess wing on the high-frequency flank of the Debye mode. 40,42,67,68erein, we observed only one prominent symmetric mode, with no sign of any additional process.Therefore, at first, this single dominant relaxation observed in the dielectric response of PhAs was interpreted as a genuine α mode. 69However, recent studies on phenyl-substituted alcohols by a combination of BDS with different experimental techniques (i.e., either photon correlation spectroscopy, PCS, or mechanical measurements 67,70−72 ) clearly show that this relaxation observed in the case of PhAs is, in fact, the superposition of the two modes (a slow Debye-like and an α one, resulting from both the crosscorrelation between dipole−dipole and self-dipole correlation, respectively) but is observed as only one relaxation peak due to their similar time scale. 67nterestingly, in the case of PhAs infiltrated in various silica membranes, one can observe a different scenario.First, for alcohols within native (untreated) templates, the presence of three relaxation processes can be detected (instead of the two modes observed in the bulk).An additional relaxation process emerges in the middle-frequency range on the low-frequency side of the main relaxation peak.Importantly, it becomes more prominent with an increase in the alkyl chain of examined compounds (see Figure 2(c,f)).On the other hand, it is not resolved for alcohols within silanized (treated) templates.Taking those observations into consideration, one can assume that the additional process might be, in fact, related to the motions of the molecules in close proximity to the interface, which strongly interacts with the pore walls (so-called interfacial molecules). 40,73Therefore, often this process is denoted as the interfacial one.It is assumed that this mobility is no longer present within materials confined within treated silica templates due to a lack of specific interactions between PhA molecules and the interface as a result of the silanization.However, taking into account the above-mentioned calorimetric data, one can also expect to observe the interfacial process in the case of silanized systems.In this context, one can briefly remind that this process was not previously reported for alcohols incorporated in AAO templates of various pore sizes, 36 which was assigned to the difference in the time scale of the mass exchange between interfacial and core molecules and the experiment time. 74It is assumed that when the exchange between both fractions is faster than the time of the experiments, the interfacial process can be detected (and absent if not).Thus, one can assume that the lack of a wellresolved additional process of PhAs within silanized mesopores might be related to the differences between the time scale of

The Journal of Physical Chemistry C
the mass exchange between interfacial and core molecules and the experiment time when compared to the MAs within native templates. 74,75igure 2(c,f), representing the shape of the main relaxation process for all examined systems, is shown in comparison to that of the bulk.The presented spectra were shifted to superposed at the same relaxation time as the main relaxation process.As illustrated, the dominant relaxation peak observed for all examined confined systems is broader (especially in the low-frequency region) than those of the bulk sample.This agrees with the previous data reported for PhAs infiltrated within various AAO templates (data for alcohols within AAO membranes of d = 10 nm were added).−7778 This broadening of the distribution of the relaxation times is often discussed as a result of an increase in the heterogeneity of the molecular mobility observed in the confined systems, most likely induced by the presence of the solid interface (in terms of the introduction of additional interactions within the examined systems). 73,80One can add that for associating materials, changes in the hydrogen bond population under confinement might also be taken into consideration. 67,81At this point, two issues should be highlighted.First, the observed loss peak under confinement is no longer a Debye-like relaxation; therefore, in the case of confined samples, this mode would be referred to as a dominant or main relaxation process.Second, the significant broadening of the low-frequency side of the main relaxation peak for PhAs within native silica templates suggests the emerging of a new mode, which might be assigned to the interfacial process as discussed above.However, when examining the spectra collected for PhAs within silanized (treated) templates, one can also see that the interfacial process can be observed as well, but it has a significantly lower amplitude when compared to the materials within native membranes. 41,42This might imply that even if the specific interactions at the interface are suppressed due to the performed surface modification, some van der Waals interactions near the surface can still be present, contributing to the formation of the molecule fraction characterized by reduced mobility in the proximity of the more hydrophobic surface.Note that the interfacial process is also affected by the magnitude of the dipole moment libration of immobilized molecules as well as the long-distance correlation between dipoles.Nevertheless, it should be mentioned that up to now, no presence of the interfacial process was reported for materials infiltrated within modified silica templates. 41,42,82,83herefore, the question arises as to the origin of the observed additional relaxation process.
To provide additional information, in the next step, we determined the temperature dependence of the relaxation times, τ, of observed relaxation peaks, the dominant (τ Dom ) and additional (τ Add ) ones located at high-and middle-frequency The Journal of Physical Chemistry C regions, respectively.For that purpose, the dielectric data were fitted using (one or two) Havrilak−Negami (HN) function(s) with an additional dc-conductivity term (see Section 2 and Figure S6). 84Determined in this way, relaxation times were plotted versus the inverse temperature and are shown in Figure 3. Data for 3Ph1P and 5Ph1P are shown in Figure S7.As illustrated, there are no differences between the τ Dom (T) dependences of examined systems (either bulk or infiltrated) at the 'high-temperature' region, while further cooling revealed a systematic change in the slope of τ Dom (T) of infiltrated materials from VFT-like of the bulk characteristic to the more Arrhenius-like at some specific temperature.It should be highlighted that this characteristic 'change in the slope' is a common feature reported for the various systems infiltrated into porous media independent of the applied pore size and the materials the templates are made of (i.e., AAO or silica). 85,86Moreover, it should be mentioned that the specific temperature (at which the change in the slope occurs) is reported to strongly depend on the pore size.In fact, the lower the d, the higher is the specific temperature.This deviation of dynamics observed for infiltrated systems was recently assigned to two major effects: (1) changes in the dynamical heterogeneities, ζ, within the glass formers under confinement conditions 74,86 and (2) the vitrification of the interfacial molecules. 85,87,88The former approach implies that the pore size of the applied porous membranes suppresses ζ upon cooling, resulting in the deviation of τ(T).However, this issue seems to be only valid for the systems characterized by d ∼ ζ, where ζ was calculated to approximately ζ ∼ 2 to 3 nm. 74,86,89,90,91Nevertheless, one can recall that the change of τ(T) was also reported for the cases where the pore diameter was higher than 18 nm (significantly higher than ζ).Therefore, this phenomenon is often described by considering the vitrification of the interfacial layer (located near the interface).Consequently, below that temperature, the investigated system might be regarded as pseudoisochoric, 91,92 which results in a change in the temperature dependence of relaxation times of core molecules, which follow τ α (T) at constant volume.Hence, the specific temperature (at which the change in the slope occurs) is usually assigned as the glass transition temperature of the interfacial molecules (located in close proximity to the interface and characterized by the 'higher' interfacial interactions and strongly reduced mobility), T g,interfacial .Note that we also observed a pronounced reduction of the dielectric strength, Δε, of the dominant process observed under confinement in comparison to the bulk samples (see Figure S8).One can recall that in the case of associating materials, this behavior might indicate the reduction in the cluster size, association under confinement. 38owever, alternatively, this might also suggest that some molecules are "adsorbed" at the interface; therefore, they do not contribute to this mode, leading to a reduction of Δε. 93 Taking into account the data shown in Figure 3, we draw the attention to three major observations.First, the deviation of τ Dom (T) can be observed for all applied types of membranes (native and modified), although the change in τ Dom (T) occurs at different conditions (see Table S1).Second, the specific temperature (at which the change in the slope of τ Dom (T) occurs) agrees well with T g,interfacial determined form the calorimetric measurements (see Figure 1(e)).This implies that the vitrification of the interfacial molecules might be responsible for the observed deviation of τ Dom (T).Third, it is worthwhile to stress that the temperature at which τ Dom (T) deviates from the bulklike behavior does not correspond in any way to the behavior of the slower dielectric relaxation process appearing in the dielectric response of alcohols infiltrated within native silica templates (see Figure 3).Note that the relaxation times of this additional dielectric process reach τ Add ∼ 0.01−0.001s when a change in the slope of the relaxation times of the main dielectric mode occurs.Therefore, taking into account also the results obtained from the calorimetric measurements, i.e., the presence of DGT and the fact that interfacial molecules form an interfacial layer, irrespective of the character of the pore walls, one can indeed assume that the deviation of τ Dom (T) originates from the vitrification of the adsorbed molecules in pores.However, the additional dielectric response observed in the loss spectra collected for PhAs within native silica templates of d ∼ 5 nm (see Figure 1) is most likely not connected to the mobility of the interfacial layer.
Therefore, the question arises about the origin of the observed additional relaxation process.One can mention that the τ Add (T) dependence resembles the characteristic of the slow Arrhenius process (SAP) recently observed in the loss spectra collected for the (unequilibrated) polymeric thin films. 94This molecular origin seems to be valid, especially

The Journal of Physical Chemistry C
taking into account that the activation energies, E a , of this process determined for all examined samples reach E a,Add ∼ 70.7−91.1 kJ/mol (see Table 2), whereas those of the SAP determined for a series of various polymeric thin films are of the order of E a,SAP = 100 kJ/mol.Nevertheless, it should be pointed out that the present data are insufficient to clearly distinguish the origin of the additional dielectric mobility observed for the examined series of PhAs within native silica templates.Therefore, this issue will be explored further within the next projects.
Alternatively, one can assume that this additional dielectric mode can originate from the mobility of "the confinementinduced" associating structures.Note that in the case of the bulk materials, we do not observe any separate mobility related to the structural relaxation but only the presence of a Debyelike process (related to the formation of nanoassociates).In fact, recent studies confirmed that the Debye-like process detected for PhAs is indeed related to the association− dissociation process according to the transient chain model (TCM). 66Thus, this additional mobility observed in pores might be related to the formation of 'new' confinementinduced associating structures.To shed new light on this problem, we performed infrared measurements to monitor the population of hydrogen bonds (HBs) of the studied compounds under confinement.Herein, we focused mostly on monitoring the changes in the H-bonding properties of PhAs, which can be well reflected and visible in the stretching vibration bands of hydroxyl groups located in the wavenumber range from 3700 to 2600 cm −1 .In this region, some peaks occur for both bulk and infiltrated PhA samples, namely, the band of weak intensity at ∼3550 cm −1 assigned to the stretching vibrations of the free, nonbonded O−H groups (ν OH free ), and a broad signal centered at ∼3330 cm −1 associated with the stretching vibrations of H-bonded O−H moieties (ν OH assoc ).At lower wavenumbers, the peaks related to the stretching vibrations of aromatic (3100−3000 cm −1 ) and aliphatic (3000−2800 cm −1 ) C−H groups are observed.
In Figure 4, the IR spectra of bulk and infiltrated samples in the 3700−2600 cm −1 spectral range over a wide temperature interval (T = 297−153 K) are shown.One should mention that the H-bonding properties of bulk PhAs were previously detailed in ref 47; hence, in this paper, we mainly analyzed the associating behavior of infiltrated PhAs compared to the bulk ones.As shown in Figure 4, the most significant changes were observed in the ν OH band region, as the position of this band was red-shifted (shifted to lower wavenumbers) with decreasing temperature in both bulk and confined PhAs.Such a spectral behavior clearly indicates the strengthening of H-bonds in these systems upon cooling.The same effect was detected for other primary and secondary monohydroxy  The Journal of Physical Chemistry C alcohols incorporated in silica and alumina mesopores 40 as well as water confined in periodic mesoporous (organo)silicas. 95urthermore, in both groups of systems, the intensity of the ν OH free band decreased as the temperature was lowered/ reduced, indicating the higher association degree (larger number of H-bonded MA molecules).Furthermore, as shown in Figure 5, we compared the FTIR spectra of bulk and confined samples at three temperatures in the 3700−3000 cm −1 spectral range.One can see that at RT, the ν OH bands of bulk and confined PhAs exhibit nearly identical shapes, and their positions at the maximum do not differ significantly (see Table S2).Interestingly, the spectroscopic studies conducted for water under confinement showed that the ν OH band was almost indistinguishable from bulk water, 17,96 or it was only slightly blue-shifted. 97The reported blue shift phenomenon was associated with the poorer H-bond acceptor ability of silica oxygen atoms and thus with the presence of interfacial water molecules. 97Moreover, the intensity of the ν OH free band was higher for confined PhA samples compared with the bulk.The percentage values of the number/amount of free hydroxyl groups for both bulk and spatially restricted PhAs were calculated and are presented in Figure S9.Details are given in the SI file.It can be noticed that approximately twice as many nonassociated OH groups for MAs under confinement compared to the nonconfined samples is observed, indicating that the association process Comparison of FTIR spectra in the frequency region 3720−3000 cm −1 for bulk and confined samples of examined PhAs at various chosen temperature conditions: (a) room temperature (RT) and (b, c) calorimetric glass transition temperature of interfacial (T g,interfacial ) and core (T g,core ) molecules.For the sake of clarity, the spectra were normalized to the OH stretching vibration (ν OH ) band.

I
of PhAs in geometrical restriction is partly suppressed.These parameter values are similar for both native and silanized silica templates, with slightly higher values for silanized ones, suggesting that functionalizing silica membranes (more hydrophobic) and the overall nanogeometrical restriction led to the presence of more amount of nonbonded OH species in nanopores.What is more, considerable changes in the ν OH bandwidths are detected, i.e., the ν OH bands of MAs in both native and silanized templates are broadened compared to their bulk counterparts.This indicates greater heterogeneity in the distribution of HB aggregates for alcohols under nanoconfinement.At RT, the broadening of this band is more distinct in the lower wavenumber region, suggesting that stronger H-bonds are more affected than weaker ones.
As the temperature decreases, the behavior of the ν OH band resembles that at RT, i.e., the subtle structure of the ν OH band remains, with a weak peak originating from the free hydroxyl groups, suggesting partial association at lower temperatures.As shown, the ν OH peak for MAs in confined samples is shifted to higher wavenumbers, indicating weaker H-bonding interactions in spatial restriction (see Figure 5(b)).Moreover, significant discrepancies between confinement and bulk are observed in the ν OH bandwidth with confined samples having broader bands than bulk samples, especially at lower temperatures.The most prominent difference occurred on the left shoulder of the ν OH band, which might indicate the presence of an additional contribution within the 'weaker Hbonded' OH associate range, much more resolved when compared to the spectra at RT.Note that native silica nanopores exhibit slightly broader ν OH bands compared to silanized membranes.Overall, the widening of the ν OH band follows this order: bulk > silanized pore > native pore, implying a more substantial impact of the hydrophilic environment on the H-bond network of the incorporated PhAs.It should also be mentioned that similar observations were reported for other MAs under confinement, demonstrating that the incorporation of MAs into silica nanomaterials is manifested by the changes in the ν OH peak frequencies and the broadening of the band in the same order as described herein. 40o explore the presence of an additional shoulder in the higher frequency region observed for incorporated 4Ph1B (the blue box in Figure 6(a)), we carried out the deconvolution of these spectra recorded at T g,core in the range of 3600−3000 cm −1 .As found, the proper description of the ν OH band of infiltrated 4Ph1B requires the application of three Gaussian curves (see Figure 6(c)), whereas the use of two Gaussian functions is enough to fit the ν OH band of the bulk sample (Figure 6(b)).Interestingly, a new component occurring for the former sample in the higher wavenumber range suggests the existence of additional weak H-bond interactions in the studied system.A similar deconvolution of the ν OH band was presented for water confined in MCM-41 pores.In this case, the region ∼3350 to 3500 cm −1 was assigned to small aggregates of water molecules characterized by different dynamics. 98One can also add that the mobility of these 'new' confinement-induced HB interactions, resolved as an additional shoulder in the IR spectra of incorporated PhAs, might contribute to the appearance of the additional relaxation process observed in the loss spectra shown in Figure 2(c,f).
Further, we performed the calculations of the activation energy, E a , of the dissociation process, using the van't Hoff equation: where E a indicates the activation enthalpy, S is the entropy of the dissociation process, and R is the gas constant.Details of the calculations are shown in the SI file together with the van't Hoff plots for PhAs in native and silanized mesopores, which are presented in Figures S10−14 and Table S3.As can be seen in Figure 7, the estimated values of E a for PhAs in nanorestriction are lower (E a ∼ 7 to 9 kJ/mol) than those in bulk materials (from 10.49 to 13.65 kJ/mol), indicating the significant contribution of weak H-bond interactions occurring in spatially confined geometries compared to the bulk ones in which the stronger HBs dominate.As a result, less energy is required to break these H-bonding interactions in confined samples than in bulk ones.This fact corresponds well with the lower association degree of infiltrated samples relative to that in bulk samples (a slightly smaller number of H-bonded MA molecules relative to free ones under confinement compared to that in bulk samples).Moreover, as the aliphatic chain length increases in bulk alcohols, the dissociation energy tends to rise.However, this trend is not observed in confined samples, where E a values are comparable (refer to Table 3 for values with errors).It should be mentioned that similar values of this parameter for bulk samples were reported for various aliphatic alcohols, differing in the chain length and the localization of

The Journal of Physical Chemistry C
the OH groups, with the activation energy of the dissociation process ranging from E a = 9−14 kJ/mol. 99Additionally, the literature shows that the enthalpy required to break HBs in pure water equals E a ∼ 8 kJ/mol, and a slightly higher value was calculated for HOD in the D 2 O solution and is equal to E a ∼ 10 kJ/mol. 100It should be stressed that, to the best of our knowledge, until now, the values of the activation energy of the dissociation process for systems confined in nanopores have not been determined.Therefore, the values presented herein for PhAs in native and silanized nanomembranes make an important contribution to investigating the impact of nanorestrical confinement on the behavior of associating materials.

CONCLUSIONS
In this paper, we investigated the influence of surface interactions on the associative behavior of phenyl-substituted monohydroxy alcohols.Interestingly, we observed a pronounced deviation of the dominant relaxation process (corresponding to the bulk Debye-like mode) for all examined confined systems occurring at T g,interfacial , independent of the applied type of porous templates.Nevertheless, the dielectric response of PhAs infiltrated within native silica mesopores revealed the presence of an additional relaxation process, particularly pronounced for longer aliphatic chains.Interestingly, it was observed that this additional mobility is most likely not related to vitrification of the interfacial layer, T g,interfacial .Its molecular origin is yet to be clarified.Calorimetric data revealed a double glass transition phenomenon for systems infiltrated in native and modified silica.Moreover, it was found that for both types of membranes, the length of the interfacial layer, ξ, reaches approximately ξ ∼ 0.5 nm for all PhAs.This suggests that an interfacial layer is formed, irrespective of the character of the pore walls.This observation was explained considering contact angle measurements, which revealed that at low temperatures, all examined PhAs have a similar wettability on both surfaces, θ ∼ 22 to 28°.IR measurements showed that the incorporation of PhAs into silica membranes inhibits complete association over the entire temperature range.Moreover, nanogeometrical restriction has a relatively small impact on the H-bonds' strength of infiltrated PhAs, as seen in the ν OH peak position.However, it alters the ν OH bandwidths; i.e., the confined samples are characterized by broader OH bands than those in bulk, indicating greater heterogeneity in the distribution of H-bonded systems in nanoconfinement.Notably, for the first time, we calculated the activation energy values of the dissociation process for confined PhAs, which were lower than those of bulk ones.This result correlates well with the lower association degree of infiltrated PhAs compared to their bulk counterparts, resulting from the spatial restriction.Thus, all experimental methods used consistently confirmed the formation of an additional interfacial layer in infiltrated PhAs in which the alcohol molecules strongly interact with the pore walls.We believe that the presented results offer a better understanding of the processes occurring for associating liquids in nanoconfinement.
It contains additional figures and tables, including figures of SEM pictures of the native silica nanopore, the IR spectra of 'empty' silica templates, the contact angles measured for examined materials at three different temperatures, and the DSC thermograms for bulk and confined samples.Moreover, it includes the dielectric loss spectra, comparison of the dielectric strength, temperature dependences of relaxation times, the comparison of glass transition temperatures obtained from BDS and DSC measurements, the IR spectra of infiltrated MAs and the percentages of free OH groups, the determination of the degree of association and activation energy of the dissociation process, values of integrated intensities of the ν OH bands, the van't Hoff plots used to calculate the activation energies of dissociation processes, and tables containing glass transition temperatures and wavenumbers of the OH peaks (PDF) ■ AUTHOR INFORMATION Scheme 1.(a) Chemical Structures of Investigated Phenyl Alcohols and the Schematic Structure of (b) the Cross-Section and (c) the Top of Applied Silica Mesopore Templates (d Denotes the Pore Diameter)

Figure 1 .
Figure 1.(a−c) DSC thermograms obtained for bulk and confined samples of 4Ph1B.(d) Glass transition temperatures (determined from calorimetric measurements) of all confined samples vs the length of the alkyl chain of the examined PhA.(e) Comparison of the glass transition temperatures obtained from BDS and DSC measurements for all studied infiltrated samples.The uncertainty of T g determination is ±2 K for all presented values.(f) Thickness of the interfacial layer (ξ) plotted as a function of the length of the alkyl chain.

Figure 2 .
Figure 2. Dielectric loss spectra of (a, b) 2Ph1E and (d, e) 4Ph1B infiltrated within (a, d) native and (b, e) silanized silica templates.The insets in panels (a) and (d) show the dielectric data for the bulk systems.(c, f) Comparison of dielectric loss peaks obtained for all measured samples at f = 10 5 Hz.

Figure 3 .
Figure 3. Temperature dependences of the relaxation time (τ) obtained for bulk and confined (a) 2Ph1E and (b) 4Ph1B.For comparison, data for PhAs within AAO templates characterized by d = 10 nm are shown (taken from ref 36).

Figure 5 .
Figure5.Comparison of FTIR spectra in the frequency region 3720−3000 cm −1 for bulk and confined samples of examined PhAs at various chosen temperature conditions: (a) room temperature (RT) and (b, c) calorimetric glass transition temperature of interfacial (T g,interfacial ) and core (T g,core ) molecules.For the sake of clarity, the spectra were normalized to the OH stretching vibration (ν OH ) band.

Figure 6 .
Figure 6.(a) Comparison of the FTIR spectra of bulk 4Ph1B and the sample within native silica mesoporous templates at T g,core .The spectra were normalized to the intensity of the OH band.Deconvolution of the ν OH band of (b) bulk and (c) infiltrated 4Ph1B into the native silica pore at T g,core using Gauss functions.

Figure 7 .
Figure 7. Activation energy of the dissociation, E a , calculated for bulk PhAs and samples infiltrated into native and silanized silica templates of d ∼ 5 nm.

Table 1 .
Contact Angle, θ, at 298 and 258 K, as well as the Surface Tension, γ L a Values of γ L were taken from ref 61.

Table 2 .
Activation Energies, E a,Add , Calculated from the Arrhenius Equation (Eq 4) for the Additional (Slower) Relaxation Mode Observed in the Dielectric Response of PhAs Incorporated into Native Silica Membranes of d ∼ 5 nm

Table 3 .
Calculated Activation Energies of Dissociation, E a , for PhAs Incorporated into Native and Silanized Silica Membranes of d ∼ 5 nm a Values of E a determined for the bulk samples were taken from ref 61.