Surface Glass TransitionChemE 554/Overney
4.5 Surface Glass Transition and Transitions in Thin Films
Index
4.5 Surface Glass Transition and Transitions in Thin Films......
4.5.1 Background......
4.5.2 Interfacial confinement effects and film preparation history
4.5.3 Liquid-like surface models that address Tg depletion in thin films......
Capillary wave induced surface melting
Near surface polymer chain sliding (sliding model)
4.5.4 Shear modulation scanning force microscopy (SM-SFM)......
4.5.5Mobile surface layer theories and preliminary SM-SFM results......
4.5.6 SM-SFM transition measurements of ultrathin supported films......
References......
4.5.1 Background
Extensive literature deals with the determination and the interpretation of the glass transition temperature, Tg, of homopolymer systems. Within bulk systems, many debates have meanwhile been settled. At free surfaces and in thin films however, because of constraints and size effects, the relaxation dynamics is still poorly understood.
It has been recognized theoretically and experimentally that in thin homopolymer films the proximity of a free surface, substrate interactions, and stress induced anisotropy within the film, are responsible in an intricate way for shifts of the glass transition temperature with respect to the bulk.1-20 Glass transition temperatures have been reported to be depressed up to tens of degrees Celsius for films thinner than a few hundred angstroms.11,14,20 It has also be recognized that substrate effects, can cause an opposite shift in Tg from the bulk to higher values.11,17
Models that have been developed over the last few years favor the idea of a liquid-like surface layer that is responsible for a Tg depletion in none-substrate-confined thin films.5,20,21 One of them, inspired by dewetting studies of thin films,22,23 in which morphological changes below the bulk Tg value were reported, proposes coupling of capillary waves with flow properties in thin films responsible for surface melting.21 In another liquid-like surface model (referred to it as thesliding model), the scenario of the glass transition is split into two; i.e., a standard bulk transition that is based on freezing-unfreezing of certain local degrees of freedom), and a near-surface transition that originates from polymer chain sliding motions. Depending on the temperature and thickness of the film, the model suggest either a sandwich structure of bulk and surface phases, or a single semifluid phase.5,24
The two liquid surface models are applied to different regimes of molecular weights: The capillary wave model21 that is based on capillary wave-induced dewetting deals with low molecular weight polymers melts. The sliding model5,20 that is considering in parallel standard bulk transition and near-surface chain sliding motions is proposed to apply to polymers above 100k. Both models do conceptually not depend on the film thickness, i.e., they apply for ultrathin films (< 100 nm) and at surfaces of bulk films.
Liquid-like surface theories have been partially supported by macroscopically averaging techniques, such as, for instance, ellipsometry11,25,26, Brillouin light scattering20,27, and thermal expansion spectroscopy28. The macroscopically determined Tg depletion is often related to the film thickness. However, more local probes used in dielectric measurements 29 and scanning force microscopy, SFM,12,17 led to results, where the glass transition temperature is much less affected by the film thickness. Contrary to the last statement about local probes are some recently published interpretations of SFM friction force measurements30. Possible misinterpretations are addressed below.
After the development of a nanorheological technique that has shown to be quite successful in determining Tg values of polymer films over a wide range of molecular weights, one has recently investigated (a) the possibility of surface melting below the glass transition temperature, and (b) interfacial substrate confinement effects on apparent Tg values.
Glass transition values obtained of monodisperse, homopolymer bulk material or at surfaces of >200nm thick films by macroscopic techniques (i.e., differential scanning calorimetry (DSC) and electron spin resonance (ESR)), and by local techniques (i.e., shear modulation SFM), respectively, are found to correspond over a wide range of molecular weights, Figure 1. The finding that surface and bulk transition properties correspond agrees well with the low surface energy of these systems, and the van der Waals liquid-like description of polymer melts.
/ Figure1. Glass transition of atactic bulk polystyrene as function of the molecular weight determined independently by three different methods:- Differential scanning calorimetry (DSC),3
- Electron spin resonance (ESR)31 and
- Shear modulation scanning force microscopy (SMSFM)1
The solid line represents a Fox-Flory fit.
However, the more complex the polymer system, i.e., the more anisotropic it is, the less correspondence is found between bulk, thin film and surface Tg measurements. Origins for deviations have to be carefully analyzed. Common assumptions have to be revisited and challenged. A common hypothesis is that thermal annealing of a spin coated homopolymer film, with a thickness exceeding the pinning regime of a few nanometers, is sufficient to relax the film back to the bulk phase. It is interesting to note that near surface Tg values determined with shear modulation SFM method correspond well to bulk Tg measurements as illustrated above in Figure 1. Hence, the annealing presumption of ultrathin films seems to be justified.
Over the last few years however, various experimental findings have challenged common hypotheses such as the annealing presumption. It has, for instance, been found that the polymer rheological properties are modified over tens to hundreds of nanometers.1,17,32-34
4.5.2 Interfacial confinement effects and film preparation history
While materials such as ceramics, metals, oxides, exhibit size limitations ('quantum well effects') only noticeable below 10 nm, it was found that in polymer systems, interfacial effects could be noticeable over distances of tens to hundreds of nanometers. Over the last few years, various groups reported bulk-deviating structural and dynamic properties for polymers at interfaces.15,34-39 For instance, increased molecular mobility was observed at the free surface for thick films.15 Reduced molecular mobility at the film surface of ultrathin films was reported based on forward recoil spectroscopy measurements.35 In secondary ion mass spectrometry (SIMS) and scanning force microscopy (SFM) studies of graft-copolymers, it was found that the degree of molecular ordering significantly affects dynamic processes at interfaces.34 Self-organization of graft and block-copolymers at surfaces and interfaces were found with transmission electron microscopy (TEM) and neutron reflectivity (NR).36-39
Application of mean-field theories to interfacially constrained and size-limited polymer systems failed to describe the rather unexpected mesoscale behavior observed experimentally. The extension of the interfacial boundary far into the bulk is unexpected because many amorphous polymer systems are theoretically well treated as van der Waals liquids with an interaction length on the order of the radius of gyration, i.e., the effective molecular size. At solid interfaces the radius of gyration is further compressed, like a pancake, and thus, any memory effects of the solid are expected to be even more reduced to a pinning regime of only 0.5 to 2nm.40 Within the pinning regime, it is commonly accepted that the material is structurally altered and exotic properties (for instance, quantum-well effects) are expected. Outside the pinning regime, the polymer is expected to behave bulk-like. Experiments show however, that such scaling theories, i.e., mean-field theories, fail in describing the observed unique mesoscale properties because they do not consider effects that occur during the film coating process, e.g., rapid solvent evaporation. For instance, recent SFM experiments revealed that the spin coating process altered the structural properties of polyethylene-copropylene (PEP) at silicon interfaces due to anisotropic molecular diffusion that is caused by process-induced structural anisotropy.41 The polymer structure at the interface affects properties such as the shear mechanical properties, the entanglement strength, and dewetting instabilities and velocities, as illustrated in Figure2.
An extensive SFM analysis involving also density measurements by X-ray diffraction and self-diffusion measurements by neutron reflectivity (NR) revealed a three component system after spin coating: (i) An adjacent to the surface immobilized and fully disentangled sublayer (20 nm thick, Fig. 2), (ii) a partially disentangled intermediate layer (100-200 nm thick, Fig. 2), and (iii) beyond the intermediate layer the bulk polymer phase. he strained interfacial sublayer can be pictured as highly disentangled and laterally anisotropic system with a thickness on the order of the polymer's radius of gyration. NR reveals that the polymers adjacent to the surface immobilized sublayer can diffuse through the sublayer's pores.39 X-ray diffraction measurements exhibit a shrinkage of the pores if annealed which immobilizes the diffused polymers 'permanently'.32 Hence, a boundary layer that exceeds by orders of magnitudes the pinning regime is formed between the interfacial sublayer and the polymer bulk phase.
Naturally one could expect that the glass transition temperature be affected within the constrained boundary regime. Indeed, it has been observed that substrate supported spin coated ultrathin films of polystyrene exhibit an increase in Tg (by 5 oC at a thickness of 30nm) compared to the bulk.11,17 Much larger changes in Tg by tens of degrees Celsius and in the opposite direction have been observed for substrate unsupported ultrathin homopolymer monodisperse PS films.11,14,20 These findings have let to theoretical models that go beyond the free volume theory and consider either coupling of capillary waves with flow properties, or near surface polymer chain sliding motions responsible for the observed depletion in Tg in ultrathin films. 5,20,21
Figure 2(a):(Left) SFM rheological lateral force measurements on thermally annealed PEP reveal a comparable qualitative film thickness behavior-as the dewetting velocities in Fig.4(b). A decrease in the measured lateral force is interpreted as an increase in the shear mechanical properties (moduli) of the interfacially confined polymer film.
(Right)The range over which the confinement is recognizable (Thin Film Regime, 100-200 nm,) exceeds by orders of magnitude the surface pinning regime (< 5 nm). Above the thin film regime, the polymer behaves bulk-like. /
Figure 2(b): Lateral (friction) force vs. load SFM experiments provide fundamental insight into the origin for changes of the rheological properties in spin coated thin PEP films. The transition point Px (x = thickness of polymer film), which corresponds to the kink in the friction vs. loading curve is a measure of the entanglement strength of the polymer (PEP). The bulk value is reached for films thicker than 230 nm. Films thinner than 230 nm are partially disentangled due to the spin coating process, and thus, the transition point occurs earlier. Within a 20 nm boundary regime the film is entirely disentangled (gel-like).
Figure 2(c): (left a+b) Illustrative sketch of the dewetting process measured by SFM. (right)5050 m recorded dewetting pattern (topography (left) and friction (right)) of PS/PEP. (top) Large and deep dewetting holes for thick (400 nm) PEP films. (bottom)Small and interfacially bound dewetting holes for thin (4 nm) PEP films.33,42 / Figure 2(d): PS/PEP dewetting velocities measured by optical microscopy reveals a decrease in velocity for PEP films thinner than 100-200 nm. The triangle indicates that the history of the sample preparation is very important ( PEP spin coated on silicon wafer, PEP floated on polyvinylpyridine (PVP)).34
4.5.3 Liquid-like surface models that address Tg depletion in thin films
Capillary wave induced surface melting
In this model, capillary wave coupling on the free surface to the bulk flow of the polymer is responsible for the formation of a melt in the vicinity of the surface. Herminghaus' capillary model21 involves coupling between interfacial eigenmodes of the viscoelastic polymer melt with capillary waves. This model is considering only low molecular weight polymers. (Note: The polymer-chain-sliding model (discussed below) applies for high molecular weights exceeding 100k). The capillary model is not restricted in polymer film thickness, i.e., applies to the free surface. It suggests a continuum scenario responsible for the glass transition, which is largely independent of the molecular weight. The theory considers strain relaxation fluctuations (density fluctuations are neglected) and its viscoelastic eigenmode spectrum. The central assumption is that the physical cause for the melting or freezing of the film (or surface layer) are memory effects in the polymer material (i.e. using as a theoretical treatment: a convolution integral with memory kernel). It is found that only high frequency modes contribute appreciably to the reduction of the glass transition temperature. The model predicts that the highest eigenvalue modes are those of the molten layer alone, which depend inversely on the thickness of this layer. The mean square amplitude of the strain fluctuation of these high modes, a measurable quantity that can be obtained from rheological spectrum analysis, should be appreciable only down to a critical thickness of the melt. The model predicted thickness of the molten layer, d, scales as d (Tgb-T)- with a critical exponent =1. Recently developed other multilayer models predict =0.5.20 Herminghaus' model consequently proposes for ultrathin films that the transition temperature value measured over the entire film is continuously changing from the time surface-melting occurs until the thickness of the molten layer corresponds to the thickness of the film. Note however, in thick films, i.e., in films in which the thickness of the film exceeds any molten surface layer thickness below the bulk Tgb value, two transition values should be obtainable with a surface sensitive tool.
Near surface polymer chain sliding (sliding model)
Experimentally observed extraordinary depleted Tg values for ultrathin freely suspended polystyrene films10,43 have recently been interpreted by deGennes via a two "melting-scenario" 5: (a) a standard bulk transition, related to the freezing-unfreezing of certain local degrees of freedom, and (b) sliding motions of each polymer chain along its own path. In the bulk it is assumed that chain sliding is hindered by the end points. At or near the free surface, however, a thin fluid skin allows the chains to slide. The skin is estimated to be less than a nanometer thick. The two transition processes are illustrated in Figure 3.4
Figure 3: Model of polymer chain in a thin film. Two contributions arise from a segment formed by loop AB and a bridge BC. For films thinner than the coil size, the dominant process maybe the collective motion of a loop which does not involve the chain ends.4The bulk Tg motions are based on short-range rearrangements. These types of motions are contributions of bridge formations shown in the polymer chain touching points A and B in Figure 3. The other type of motion is based on polymer sliding, where a chain advances along a path via mobile kinks and the free volume required for sliding is less than the bulk cooperative motion. These types of motions are contributions of “loop” formations shown in the polymer chain touching points B and C in Figure 3. A loop at the surface should slide easily, however, it is believed that sliding is hindered in the bulk because chain ends would have to invade new territory, thus requiring more free volume. The two transition model lead to a Tg that depends on the distance from the free surface (h-h*), described by
(1)
where Tg, Mw, and h refer to the glass transition temperature, molecular weight, and film thickness respectively. The parameters b, Tg*, Mw*, and h* are fitting parameters from Tg vs. h plots characteristic of the observed deviation from bulk Tg values. The model predicts bulk behavior at the surface for very high molecular weights (> several 1,000k), which is experimentally confirmed in ultrathin films.5,11
Although the mobile surface model has been motivated by ellipsometric ultrathin film studies, which lead to discussions about experimentally observed apparent transition values and theoretically predicted two scenario transitions, it is not restricted to thin films only.44 It could also be applied to free polymer surfaces of thick films. An experimental test of the model demands a surface sensitive tool such as the shear modulation SFM.
4.5.4 Shear modulation scanning force microscopy (SM-SFM)
The working principle of the shear modulation scanning force microscopy, SM-SFM, method is sketched in Figure 4.17 The technique is well suited for any surface rheological study involving thermally activated transitions or relaxations. Over the last two years, the method has shown to be a highly accurate method for determining near surface glass transition temperatures of thin polymer films. The method involves a nanometer sharp SFM cantilever tip that is brought into contact with the sample surface as sketched in Figure 4. A constant load of a few nanonewton is applied, and the probing tip is laterally modulated (with a nanometer amplitude that guarantees no relative probe-sample slippage) while the temperature is stepwise increased by 0.1oC. At each temperature step the system is idle until thermal equilibrium is obtained before any viscoelastic responses are recorded. The recorded response amplitude, which is a measure of the contact stiffness,17 is then plotted versus the temperature. The Tg value is determined from the "kink" in the response curve as documented in Figure 4(a). The figure also illustrates the high accuracy of the method. It allows Tg studies of various parameters such as, for instance, the molecular weight dependence as shown in Figure4(b), (see also Fig. 1 above).
It is important to note that the SM-SFMmethodis a non-scanning method. The reason is briefly describe here: To obtain high accuracy in Tg measurements, it is essential, not to induce by other means than temperature, changes in the contact area. This is to avoid system-driven artifacts in the contact stiffness, kc. To be precise, kc(AL, G*), i.e. the resistance of the contact to deform, is dependent on (a) the laterally projected contact area, AL, (e.g., the side wall of an indentation dip), and (b) the relative shear properties of the two materials, G*. Thus, any local plastic deformation, for instance, the generation of a deformation wave (Schallamach wave)45 that travels ahead of a scanning SFM tip can change kc. Any plastic deformation is intrinsically rate and load dependent. Thus, it is not astonishing that scanning methods such as the lateral (friction) force microscopy revealed scanning velocity dependent apparent transition values for Tg, Fig. 5.8