pubs.acs.org/Langmuir © 2009 American Chemical Society
Chain Dynamics of a Weakly Adsorbing Polymer in Thin Films S. Ayalur-Karunakaran,*,†,‡ B. Bl€umich,† and Siegfried Stapf‡ †
Department of Macromolecular Chemistry, ITMC, RWTH Aachen University, Worringerweg 1, 52074 Aachen, Germany, and ‡Department of Technical Physics II/Polymer Physics, TU Ilmenau, PO Box 100565, 98684 Ilmenau, Germany Received May 16, 2009. Revised Manuscript Received June 29, 2009 Thin films of weakly adsorbing poly(dimethyl siloxane) (PDMS) on porous alumina are examined with NMR fast field cycling (FFC) relaxometry and NMR transverse relaxometry. The longitudinal relaxation dispersion of polymer amounts corresponding to approximate monolayer coverage shows substantial deviation from the bulk and is characterized by a particularly weak temperature dependence. Thicker films, however, show relaxation behavior and temperature dependence more similar to the bulk polymer. Transverse relaxation times were found to cover a range of several orders of magnitudes for any sample investigated; their dependence on temperature is a function of the total amount of adsorbed polymer. While thick films see an overall increase of molecular mobility at higher temperatures, monolayer films are best characterized by the decreasing fraction of a short, i.e. relatively rigid, component. These effects are consistent with the concept of two regions, one in which chain dynamics deviate from bulk and another where chain dynamics are reduced but bulk-like, although chains inside each region may also experience motional heterogeneity.
Introduction The structure of chain molecules at an interface has been a subject of intense research for decades for general understanding as well as its practical importance in applications such as in paints, adhesives, or industrial additives. With the advent of nanoscience and engineering, research on interfacial chains in nanoconfined and ultrathin films has gained interest for similar reasons and has resulted in several important new phenomena in films with thicknesses ranging from a few radii of gyration (Rg) down to dimensions corresponding to the chain diameter.1,2 A subsequent central question is how these surface chains affect dynamical properties such as diffusion,3 tracer diffusivity,4 local segmental mobility,5 chain modes,6 glass transition,7 and so forth, not only immediately at the surface but at distances of several Rg. In fact, the deviation from bulk behavior is already known to originate from the interfacial layers, giving rise to concepts such as “dead layer”4,8 and “reduced mobility layer”9 at the interface in such systems. With many of these properties depending on various factors such as the flexibility of the polymer, attractive forces at the surface, or temperature, it is clear that a thorough understanding can be obtained only by studying different combinations *Corresponding author. E-mail:
[email protected], santhosh.
[email protected]. (1) Mukhopadhyay, M. K.; Jiao, X.; Lurio, L. B.; Jiang, Z.; Stark, J.; Sprung, M.; Narayanan, S.; Sandy, A. R.; Sinha, S. K. Phys. Rev. Lett. 2008, 101, 115501. (2) Jones, R. L.; Kumar, S. K.; Ho, D. L.; Briber, R. M.; Russell, T. P. Nature 1999, 400, 146–149. (3) Zheng, X.; Rafailovich, M. H.; Sokolov, J.; Strzhemechny, Y.; Schwarz, S. A.; Sauer, B. B.; Rubinstein, M. Phys. Rev. Lett. 1997, 79, 241–244. (4) Zheng, X.; Sauer, B. B.; Vanalsten, J. G.; Schwarz, S. A.; Rafailovich, M. H.; Sokolov, J.; Rubinstein, M. Phys. Rev. Lett. 1995, 74, 407–410. (5) Napolitano, S.; Prevosto, D.; Lucchesi, M.; Pingue, P.; D’Acunto, M.; Rolla, P. Langmuir 2007, 23, 2103–2109. (6) Kraus, J.; M€uller-Buschbaum, P.; Kuhlmann, T.; Schubert, D. W.; Stamm, M. Europhys. Lett. 2000, 49, 210–216. (7) Rittigstein, P.; Priestley, R. D.; Broadbelt, L. J.; Torkelson, J. M. Nat. Mater. 2007, 6, 278–282. (8) DeMaggio, G. B.; Frieze, W. E.; Gidley, D. W.; Zhu, M.; Hristov, H. A.; Yee, A. F. Phys. Rev. Lett. 1997, 78, 1524–1527. (9) Napoletano, S.; Wubbenhorst, M. J. Phys. Chem. B 2007, 111, 5775–5780. (10) vanZanten, J. H.; Wallace, W. E.; Wu, W. L. Phys. Rev. E 1996, 53, R2053– R2056.
12208 DOI: 10.1021/la901738q
varying the parameters, one of which would be the adsorption strength.10 Most of the studies in this area have concentrated on polymers on an attractive substrate, and a few on noninteracting substrates such as another polymer.11 Nevertheless, the findings are equally important to contribute to the understanding, and so are thin films of weakly adsorbing polymers, which, so far, remain largely unexplored. Weakly adsorbing polymers are characterized by weak bonding forces such as van der Waals forces, different equilibrium kinetics, short residence time of the bonding segments, and, in general, a broad distribution of the fraction of adsorbed segments per chain.12 The main aim of this work is to give a qualitative picture of the dynamical aspects of chain motion under weak adsorption, and to suggest a concept for the relation between polymer structure and dynamics in such thin films. Methods that have been commonly used to investigate thin film systems include atomic force microscopy, scattering techniques, rheological and other mechanical methods. Nuclear magnetic resonance (NMR) as a potentially powerful technique has so far only seen limited use because of sensitivity issues due to the small number of nuclei in thin films. While new developments such as dedicated microcoils and hyperpolarization aim to overcome this problem, using a nanoporous host with a large surface-to-volume ratio has proven to be a viable alternative,13 where thin films form on the pore surface. A thin film formed inside porous media can be seen as a partially filled confined polymer system. At the same time, the effect of geometrical confinement and surface curvature must be considered and poses more fundamental questions. While NMR is most suitable to investigate porous media, the number of alternative techniques for obtaining supporting information remains limited to those that do not require direct optical access to the interface. Thus independent information on layer properties (11) Rivillon, S.; Auroy, P.; Deloche, B. Phys. Rev. Lett. 2004, 84, 499–502. (12) O’Shaughnessy, B.; Vavylonis, D. J. Phys.: Condens. Matter 2005, 17, R63– R99. (13) Primak, S. V.; Jin, T.; Dagger, A. C.; Finotello, D.; Mann, E. K. Phys. Rev. E 2002, 65, 031804.
Published on Web 07/15/2009
Langmuir 2009, 25(20), 12208–12216
Ayalur-Karunakaran et al.
such as average thickness and smoothness remains difficult to extract. It has previously been discussed that interfacial forces affect the dynamics of polymer chains in a wide range of time scales, for instance, from long-time scale processes such as diffusion3,4 to medium time-scale processes such as Rouse modes.14 With polymer dynamics spanning many orders of magnitude in frequency, it becomes important to investigate the motions of polymer chains on time scales as wide as possible. One NMR technique that serves exactly this purpose is fast field cycling (FFC) relaxometry.15 Studying longitudinal relaxation (T1) at a certain applied field, being proportional to the Larmor frequency, is related to chain motions occurring at that particular frequency. With this method, in contrast to conventional NMR methods, the applied magnetic field itself is varied, and T1 at each field is measured; thus information on dynamical processes can be probed over the whole accessible range, typically from a few kilohertz to a few tens of megahertz, which often corresponds to motion ranging from chain modes to segmental reorientation motion in a polymer. This method has been used earlier to study bulk and confined polymer dynamics (a review of the results can be seen in ref 16) and has been instrumental in verifying and discriminating the various theoretical predictions of polymer dynamics.17-20 In terms of confined systems, mostly polymers completely filling the pore space were studied, while field cycling relaxometry investigations of partially filled pores remain scarce.14 This study is also motivated by probing polymer chain dynamics in such incompletely filled systems. The system in this study, poly(dimethyl siloxane) (PDMS) thin films adsorbed on alumina, was earlier investigated in a recent study14 by the same authors. Thin films of PDMS formed inside porous alumina with nominal thickness varying from a few monolayers to submonolayers were investigated in pores of sizes 20 and 200 nm. Relaxation behavior of the thicker layers (>3 monolayers) in 200 nm pores showed mainly a reduction in the relaxation times, and the dispersion, i.e., the frequency dependence, is similar to that of the bulk. However, for significantly thinner films ( Mc, the slope of the region observed in this frequency regime for the bulk polymer is independent of molecular weight. In Figure 3, the profiles for the 1.13 nm layer shows a significant deviation from bulk with an increased slope of 0.45, but one finds a tendency toward a weaker frequency dependence at increasing temperatures. For this higher molecular weight, even the 3.3 nm layer shows a slight change in the slope, which is less pronounced at lower temperatures. B. Transverse Relaxometry. CPMG and SE pulse sequences were used to study the decaying transverse relaxation. Figure 4 shows the transverse relaxation times of the PDMS 30k samples obtained with a CPMG pulse sequence. T2 decays of all the samples were clearly non monoexponential, and could be best fitted by a biexponential function, while a fit to a higher number of exponentials did not reduce the residuals significantly. The T2 times as seen from Figure 4 have a strong dependence on layer thickness, indicating decreasing mobility in thinner obtained from the fit also films. The weight fraction of Tlong 2 increases with larger layer thickness corresponding to increasing fraction of more mobile chain segments with larger layer thickness. (31) Weber, H. W.; Kimmich, R. Macromolecules 1993, 26, 2597–2606.
12212 DOI: 10.1021/la901738q
The temperature dependence of the transverse (CPMG) relaxation times is shown for the different samples in Figure 5. The most significant dependence on temperature is observed for the sample with the smallest polymer amount, while the remaining samples show only a very weak temperature dependence at most. At the same time, the relative weight fraction of the short component (Ashort) steadily increases for all the samples (cf. Figure 5). In order to visualize this dependence, the weight fractions of two of the samples with increasing temperature are plotted separately in Figure 6. For an interpretation in terms of the relative contributions of more and less mobile fractions of polymer segments, see Discussion. Segments adsorbed on pore walls due to surface interactions have reduced mobility, thereby rendering intermolecular dipolar interactions to be only partially averaged. In order to obtain information about such partially averaged dipolar couplings, the study of SE decay is a well-known method30 since strong incomplete bilinear dipolar interactions are effectively refocused by the SE. The transverse relaxation components of such systems are characterized by times in the range from a few microseconds up to a few milliseconds. T2 fitting results from SE decays as a function of layer thickness for PDMS 30k samples in A200 are shown in Figure 7. The best fit to the experimental data was obtained with a biexponential function. Such a model curve along with the fit is shown in Figure 8. The shortest T2 observed here for any sample is above 100 μs. This should be compared, for instance, to PDMS chains grafted on silica, for which a T2 of around 70-90 μs has been observed.30 Nevertheless, the shape of the decay curves was exponential throughout, and a Gaussian component could not be identified as would be expected for a solid with unaveraged dipolar interactions. In general, The T2 times show a strong dependence on the component (cf. Figure 7). layer thickness, in particular the Tlong 2 The weight fraction of the short component also decreases with increasing layer thickness, implying the presence of a higher fraction of motionally restricted chains in thinner layers. From the T2 values obtained for the 1.25 nm layer (112 μs at 298 K), it can be seen that this sample possesses chain segments that are more restricted in motion than those in the other, thicker layers. In Figure 9, the SE relaxation decay times as a function of temperature are shown. The short and the long components of T2 are almost an order of magnitude different for all samples. The long component shows only a weak dependence on temperature, if at all. But the short components increase more notably, especially for the thick layer. At the same time, the amplitude of the short component decreases with increasing temperature. Langmuir 2009, 25(20), 12208–12216
Ayalur-Karunakaran et al.
Article
Figure 5. T2 relaxation data of (a) 4.50 nm, (b) 2.10 nm, and (c) 1.25 nm layers of PDMS 30k samples at different temperatures. The numbers below the data points denote the weight fraction of each data point obtained from a biexponential fit to CPMG measurements. For comparison, T2 of bulk PDMS 30k is 109 ms at 293 K.
Figure 6. Weight fractions of short and long transverse relaxation component for samples of PDMS 30k obtained from a biexponential fit to CPMG measurements. (b) and (O) correspond to the weight fractions of long and short components, respectively.
As a general feature, one finds that a temperature dependence in the range 298 to 333 K is almost absent for the thinnest layer but gradually becomes stronger with increasing polymer amount. This is similar to the trend found in the results from field-cycling relaxometry.
Discussion For PDMS chains strongly adsorbed on different substrates, chains in monolayer thin films have been observed to assume flat chain conformations on the surface.23 As the layer thickness increases, similar to an increase in surface coverage in particulate systems, more chains progressively form loops and adsorbed trains having different dynamics, decreasing the fraction of chain segments that are adsorbed in a chain with increasing layer thickness. This results in a motional gradient that induces the heterogeneity observed in these films. However, the fraction of adsorbed segments per chain is not a constant for all chains, but Langmuir 2009, 25(20), 12208–12216
rather follows a statistical distribution that can be rather broad as well as bimodal as has been observed in other systems.32,33 This property, together with the exchange dynamics between nonadsorbed and adsorbed chain segments, induces additional complexities in the system. However, a simple three-phase model, each phase with different mobilities as a function of distance from the surface, has proven sufficient to describe the dynamics of polymer chains strongly or even chemically adsorbed onto the surface.30,34 In the present case, such a model may be sufficient for layers with thickness of several monolayers, although the chain flexibility of PDMS results in a wide interface region for such a model. The relatively weak adsorption also supports the existence of a wider (32) Schneider, H. M.; Frantz, P.; Granick, S. Langmuir 1996, 12, 994–966. (33) Douglas, J. F.; Schneider, H. M.; Frantz, P.; Lipman, R.; Granick, S. J. Phys.: Condens. Matter 1997, 9, 7699–718. (34) Cosgrove, T.; Roberts, C.; Garasanin, T.; Schmidt, R. G.; Gordon, G. V. Langmuir 2002, 18, 10080–10085.
DOI: 10.1021/la901738q
12213
Article
Ayalur-Karunakaran et al.
Figure 7. Transverse relaxation times, obtained from a biexponential fit to the decay of the SE (T2,SE) for different samples. Data were measured at T=298 K. The numbers below the data points denote the weight fraction of the corresponding component.
Figure 9. Temperature-dependent plots of (a) T2 and (b) weight fractions obtained from fits of SE decay curves for PDMS 30k thin layers. (b), (1), (2) correspond to the 1.25, 2.1, and 4.5 nm layers, respectively. The closed and the open symbols correspond to long and short components, respectively. The solid lines are guidance to the eye. The errors for the amplitudes were smaller than the symbols themselves. Figure 8. Decay of the SE as a function of echo time for the 1.25 nm layer sample at 298 K. The solid line represents a biexponential fit to the curve.
interface region since it results in a shorter residence time for the chains, compared to strongly adsorbing systems, so that restrictions to the molecular mobility are less pronounced. At very low layer thicknesses, however, the change in the distribution and the conformation of the chains might lead to different dynamical effects. From the relaxation dispersion results (cf. Figures 1 and 2), the bulk melt shows a dispersion slope of 0.25 in accordance with limit II of the renormalized Rouse theory.18 The chains in the 7.1 and 4.5 nm thick layers do not undergo a qualitative change in behavior relative to the bulk at all studied temperatures following the same power law as the bulk. However, in the case of the thinner films, the 2.1 nm layer shows a deviation from bulk in the form of a slight change in the slope of the relaxation dispersion at lower temperatures while showing a tendency toward bulk behavior at higher temperature. In the case of the 1.25 nm layer, the dispersion shows deviation from bulk at all temperatures studied. The temperature dependence of the profiles of the different samples decreases with decreasing layer thickness, as can be seen from Figure 2. The profiles for the 7.1 and 4.5 nm layers show similar temperature dependence as compared to the bulk, while the 2.1 and 1.25 nm samples, the thinner samples in the series, apparently show only a weak temperature dependence. The change of slope from the theoretically predicted value of 0.25 indicates a deviation from the conditions of the renormalized Rouse model, which have been shown experimentally to be fulfilled for three-dimensional (3D) melts.16,31 Obviously, adsorbed polymers cannot be treated as 3D melts in general, and this study gives hints to when this deviation becomes noticeable. 12214 DOI: 10.1021/la901738q
One possible explanation can be sought in the relative importance of entanglements: while PDMS 30k is only marginally above the critical molecular weight (Mc=25 000) for the onset of entanglements,26 i.e., the crossover from Rouse to renormalized Rouse dynamics, PDMS 200k has a much larger number of entanglements per chain. The lower molecular weight might thus more likely be affected by a change in topology, and entanglement effects could become inefficient near the surface. This would not be the case for PDMS 200k. However, the behavior of relaxation dispersion of both batches of PDMS under study show a similar dependence on polymer amount, with the thinnest layers being significantly affected (cf. Figures 2 and 3). The deviation from bulk dynamics is thus not expected to be a simple consequence of varying entanglement density. On the other hand, the entanglement effect can also deviate from bulk when the chain conformations change (hence the chain density and the excluded volume) and differ from the bulk conformations (Gaussian). Non-Gaussian conformations have been described earlier2 in thin films with thicknesses approaching Rg. However, the density itself has been observed to deviate from bulk at lengths several Rg’s away from the surface.1 From relaxation dispersion of the samples studied here, a deviation from bulk is seen for thinner films (d < 0.5 Rg), while thicker films show only a reduction in relaxation times (Rg of PDMS 30k ∼ 5 nm and Rg of PDMS 200k ∼ 12.5 nm). In adsorbed systems such as supported thin films, a part of the whole chain exists in the adsorbed state, while the other part exists in the nonadsorbed state. The longitudinal relaxation times (T1) of a chain are an unresolved weighted average of the two components due to chain connectivity, i.e., the same chain where one part is adsorbed and the other is nonadsorbed, as seen from a single exponential relaxation curve. The effective observed T1 of the system thus depends on the contributions from the region close to Langmuir 2009, 25(20), 12208–12216
Ayalur-Karunakaran et al.
the solid surface where chains deviate from bulk due to restrictions arising from adsorption (henceforth referred to as the “adsorption restricted layer” or “ARL”) and contributions from the region where chains are restricted simply due to connectivity while following reduced bulk-like dynamics. The above results can be discussed as interplay between these contributions from the different regions. Contributions from the ARL dominates the behavior of thinner films, while, in the case of thicker films, contributions from mobile regions dominate and overwhelm the relaxation. The temperature dependence of the relaxation profiles can be explained with a similar argument. The chains inside the ARL show a weak temperature dependence, while the mobile chains have a temperature dependence similar to that in bulk. In the case of 2.10 nm layer, it could be that, with increasing temperature, chains inside this region gain mobility, becoming free to show bulk-like dynamics, letting the relaxation profiles be dominated by the free chains. This is also supported by transverse relaxation results, as will be discussed next. From the transverse relaxation measurements using CPMG pulse sequence on the samples in Figure 3, we see that both the relaxation times and the amplitude of long component (Along) increases with increasing layer thickness. The long component originates from very mobile segments possibly consisting of dangling tails and other long loop segments close to the air interface and far from the solid surface. Clearly the samples with the highest layer thickness show the existence of a large fraction of such free chain segments. However, with decreasing layer thickness, the presence of such chains diminishes as can be seen from the decreasing Along. Interestingly, still no real bulklike mobility in absolute terms (T2) is observed in any of the samples studied. This is indicative of the chains that are pseudofree resulting from chain connectivity, corroborating the reduction in relaxation times seen even for the thicker films from relaxation dispersion. The temperature dependence as obtained from CPMG transverse relaxation data was found to be relatively weak for all samples. At a closer look, an increase by a factor of about 2 is seen only for the thinnest sample (1.25 nm), while the 2.1 nm layer sample shows a slight decrease in the values. The 4.5 nm layer shows only a slight increase compared to the 1.25 nm layer. Despite this, the differences in the absolute values of the relaxation times remain starkly dependent on the layer thickness at all temperatures. A further interesting effect is seen from the amplitudes of the components for all samples. The Ashort CPMG of all the samples increases with increasing temperature, indicating an increasing fraction of less mobile chains with rising temperature. The existence of transverse relaxation components shorter than those observed by CPMG can be identified by means of SE decay curves. After subtracting the background, the shortest components were still longer than 100 μs at ambient temperature, indicating that motional restrictions are not as strong as immobilized PDMS segments grafted onto silica,30 a fact that might be expected for weakly physisorbed systems. For the 4.5 nm layer, from the SE was around 200 μs, hinting at chains the Tshort 2 retaining a certain amount of overall mobility, but in the case of 1.25 nm layer, the value was close to 100 μs. Thus chains in this sample are significantly more restricted compared to the 4.5 nm layer. For the SE, the weight fraction of the short component decreases with increasing layer thickness. The shortest component of the transverse relaxometry spectrum corresponds to the distribution of most motionally restricted chain segments. Thus, with decreasing layer thickness, an increase of the proportion of adsorbed chains segments is witnessed. Langmuir 2009, 25(20), 12208–12216
Article
The temperature dependence of SE relaxation times for the samples from Figure 9 shows a gradual increase in relaxation times for all samples. The temperature dependence of both the long and short components is strongest for the 4.5 nm layer in absolute terms, and nearly absent for the 1.25 nm films. This coincides with the observations made for the longitudinal relaxation dispersion. Before proceeding further, from the discussed transverse relaxation data, it may appear that four distinctly different relaxation components exist. However, this is only possible if two conditions are fulfilled: (i) There exist regions of distinct dynamics with a sharp interface for which the fraction of adsorbed segments per chain (fads) is the same for all chains in each sample. But it is easily conceivable that only a mean value of fads can be defined and that a broad distribution of values must exist in adsorbed polymer systems, as has been found in both experimental and theoretical studies of such adsorbed polymer systems.12 Given the weak interactions of PDMS with alumina, the high flexibility of the backbone, and the dynamic exchange processes between adsorbed and nonadsorbed segments, this distribution will be rather broad, resulting in a distribution of relaxation times. (ii) The effect of the pulse sequence used is the same but acts on different time scales. However, CPMG refocuses isolated spin interactions thus contributions from mobile regions are refocused, while the SE also refocuses bilinear (two-spin) spin couplings perfectly along with weak higher order couplings, where such couplings exist only in motionally restricted regions. Hence, different contributions are refocused by the two-pulse sequences. It is thus important to point out that the relaxation time from one method cannot be related numerically to the other method. Rather, each method generates a distribution of relaxation times that is represented by two components since these generate optimum fitting results within the experimental errors. Keeping this distinction in mind, the key to the quantification of slow polymer dynamics must be to first relate the shortest T2 component with the strongly adsorbed segments, and then to describe the dependence of the longer transverse relaxation components. The shorter of the two components decreases in relative weight; it remains the dominant one for the thinnest layer and the smaller fraction for the thickest layer. However, the relative fractions cannot be explained by a constant amount of polymer strongly adsorbed. The decreasing amplitude with increasing temperature would also correspond to the decreasing thickness of this region due to “thawing” of adsorbed segments and increasing mobility of restricted segments in the region. An interesting feature is the relation between the relative amplitudes of short and long components (Ashort and Along) of the SE and the T1 relaxation dispersion behavior. Samples that show significant deviation from bulk also have Ashort from the SE as the major component, e.g., the 1.25 nm layer, where the sample shows deviation from bulk at all temperatures and Ashort > Along also at all temperatures. In cases where Ashort is less than Along, either with increasing layer thickness (like in the case of the 4.5 nm layer) or with increasing temperature (like in the case of the 2.1 nm layer at 333 K), the relaxation dispersion becomes more bulk-like. Thus it is tempting to assign the relation of the relative amplitude of the short component from the SE to the magnitude of the contribution of the ARL that shows deviation from bulk dynamics. From this, it can also be deduced that the contribution of ARL not only progressively decreases with increasing layer thickness but also with increasing temperature. It is also of interest to see the transition region as to the thickness at which the relaxation behavior deviates from the bulk DOI: 10.1021/la901738q
12215
Article
Ayalur-Karunakaran et al.
and the molecular parameter to which this is related. For instance, for polystyrene (PS) on thin silica wafers, it has been observed that the chains deviate from the bulk (Gaussian) conformation for thicknesses below 0.5 Rg, where Rg is the radius of gyration of the molecule. Here as well, we see a deviation for films with thicknesses below this range (for this sample, Rg ∼ 5 nm). This will be addressed in more detail in a subsequent communication. The deviation or the nonfollowing of the bulk behavior could be due to (dominating) contributions coming from such non-Gaussian chains close to the surface, while for the majority of the theories, the Gaussian chain is a fundamental assumption.
Conclusion The dynamical behavior of thin films of a weakly adsorbed PDMS thin layer on porous alumina with layer thicknesses ranging from about 1.5 to about 10 monolayers was studied. The NMR relaxation behavior of a given sample at a given temperature depends on the width of the so-called “ARL”, a region at the solid surface where the mobility is restricted as a result of chain adsorption, resulting in nonbulk dynamics behavior. This behavior is found to deviate qualitatively (i.e., show a different power-law dependence of the longitudinal relaxation time on the Larmor frequency with respect to the bulk PDMS melt in the range of 104 to several 107 Hz, and thus statistics of the modal distribution of chain motions different from the bulk) when the thickness is large enough to dominate the relaxation behavior and overwhelm contributions from the remainder of the film, where chains follow reduced bulk-like dynamics. The
12216 DOI: 10.1021/la901738q
fact that restricted dynamics does actually exist even for thicker layers is identified from CPMG measurements of the transverse relaxation time being considerably shorter than bulk values, an indication that very slow dynamics outside the window accessible by T1 relaxometry are affected by the confinement and do not follow the Arrhenius temperature dependence of the bulk. Corroborating the above concept, the short component amplitude of the SE that gives information about the magnitude of strongly restricted segmental motions remains the major component when deviation from bulk occurs. In addition to the attractive interaction of chain segments with the surface, the known deviation of the chain conformation from Gaussian statistics in the immediate vicinity of the surface is suggested as a reason for the qualitative change in the frequency distribution of molecular dynamics, where this non-Gaussian conformation can be attributed to an ARL. For PDMS on alumina substrate, the thickness of the ARL is estimated to be less than the chain size (considered to be the radius of gyration), as was observed for other systems. At the same time, an increase in mobility and “thawing” of chains with increasing temperature renders the thickness of the ARL temperature dependent, which has its consequence in a crossover between bulk-like and nonbulk molecular dynamics as a function of temperature. Acknowledgment. S.A.K. wishes to express thanks to Prof. Dan Demco and Dr. Rudra Choudhury for helpful discussions. Financial support from the Deutsche Forschungsgemeinschaft DFG (Project Sta 511/5-1) is gratefully acknowledged.
Langmuir 2009, 25(20), 12208–12216
