Polymers and biopolymers at interfaces

An understanding of the behaviour of biopolymers at interfaces is often considered important for negative reasons. Biopolymers foul interfaces, and the unfeasibly large technological subdomain of PEGylation is designed to prevent proteins reaching interfaces. PEGylation—the attachment of poly(ethylene glycol) (PEG) to surfaces or functional molecules such as drugs—is a standard technique to create biocompatibility. The Oxford English Dictionary defines compatible as meaning 'mutually tolerant' or 'congruous', yet biocompatibilization is the art of making a material inert in a biological environment. The host environment is not to interfere with, or affect, the biocompatibilized foreign object. Creating biocompatibility involves chemistry, if PEGylation is the goal, but understanding polymers at surfaces is key. Proteins are interesting molecules because of the richness of their structure. In a native state, they are folded with large parts of the protein ordered into α-helices and β-sheets. Here they give biological function [39, 40], be it molecular transport, facilitating biochemical reactions, supporting the immune system, or whatever other role is required. In their unfolded (non-native) state, they can be the source of many problems. Once proteins are folded, they are stable, but when misfolded, or unfolded, they can be prone to aggregation, which is thought to be responsible for conditions such as Alzheimer's disease, or bovine spongiform encephalopathy (mad cow disease). Proteins can fold or unfold at surfaces, and so their role is important in understanding proteins themselves. Force spectroscopy studies [33] are a useful means of understanding the conformation of proteins in different environments and thus may play a key role in understanding the causes of misfolding.

The prokaryotic cell wall is a complex environment consisting of many proteins, biosurfactants, and polysaccharides. Their adhesion is controlled by adhesin molecules, and the process of adhesion involves the secretion of these molecules to test the viability of a surface. Adhesins are part of a broader class of molecule known to biologists as virulence factors, but here they can be taken to be either proteins or polysaccharides. If these molecules are compatible with a surface, the cell will adhere to that surface, and a biofilm may form: the surface is not biocompatible. It is common practice for researchers to consider the cell as an inert colloidal particle and its adsorption to be controlled by electrostatic and van der Waals forces. The application of theory based on these ideas—Derjaguin, Landau, Verwey, and Overbeek (DLVO) theory [41]—is commonplace. However, cells are not static inert objects, but adapt to their environment. Their behaviour, and thus their physics is environmentally dependent. This is as true in the bulk as it is on surfaces. In the bulk, chemotaxis depends on the availability of nutrients or presence of toxins and the cell's response to these informs different dynamical properties. On surfaces a dispassionate consideration of cell walls is not in itself enough for a determination of whether or not a cell will adhere to a given surface; for example, bacteria can express many different adhesins and adhesin expression is dependent upon environmental factors [42].

That proteins and polysaccharides are biologically important is clear, but it would be worthwhile to identify whether their function within biological systems is completely driven by their chemistry, or whether the underlying physics inherent in large molecules also affects their behaviour. A protein is a series of amino acids joined by peptide bonds, and a polysaccharide is a series of sugar molecules joined together by glycosidic bonds. These are shown in figure 2. In both cases there are plenty of hydroxyl groups that give rise to the hydrophilicity of the components, as they are sources of hydrogen bonding. Whilst the specific combination of chemical groups or amino acids will interact individually with their environment, the ability of the chain to work as a single entity, as required for its biological function, is driven by the form of the molecule. This is particularly pertinent in the case of proteins, with their naturally folded peptide chains. This type of conformation arises from an energy landscape where one of the most stable low-energy forms is a 'marginally compact tube' [43], which has a propensity for generating α-helices and β-sheets, the most important secondary structures in proteins [44]. Furthermore, proteins have been shown to maintain their native folded structure despite amino acid replacement, suggesting that the overall shape and associated crystalline behaviour is at least partially independent of the chemistry of these large biomolecules [45].

Zoom In Zoom Out Reset image size

Eukaryotes do not have a cell wall, but rather have a membrane. (Plants, however, can have cellulose-based cell walls surrounding the outer membrane.) Their adhesion to surfaces is based on a wider array of adhesion molecules, and these are largely proteins. Their purpose is to ensure that tissue binding is complete, and are not optimized to bind to inorganic surfaces. Since their binding is expected to be to biological tissue, eukaryotic cells are best grown on hydrophilic surfaces in contrast to many prokaryotes, which are best cultured on hydrophobic surfaces.

Surface analysis is a phrase often used by veterans of the field specifically to mean photoelectron spectroscopy and secondary-ion mass spectrometry (SIMS). These two techniques provide chemical information about the surface. Initially, routine high-resolution surface imaging could only be provided by scanning electron microscopy (SEM). Now there is much more choice; photoelectron spectroscopy, SIMS, and SEM are high vacuum techniques and so newer techniques were designed to be somewhat more flexible. The suite of scanning probe techniques allows much greater choice in how samples can be measured, and provides a wide variety of data. Neverthless, if chemical information is needed, photoelectron spectroscopy and SIMS are hard to beat. Infrared techniques, however, can also provide much useful chemical information. There are many reviews concerning surface analysis, but if the reader wishes to see a complete overview, the book edited by Vickerman is to be strongly recommended [46]. Here, with due deference to its position in the history of polymer surfaces, x-ray photoelectron spectroscopy, which is sometimes called electron spectroscopy for chemical analysis, is discussed.

2.1.1. X-ray photoelectron spectroscopy.

The two photoelectron techniques, XPS and its longer wavelength sibling, ultraviolet (UV) photoelectron spectroscopy (UPS), offer the same operating principle although their uses are slightly different. X-rays or UV are used to eject electrons from the sample. The high-energy x-rays are capable of ejecting core electrons and the resultant information (the energy of these electrons) provides information on the chemical composition and bonding in the material under illumination. UV radiation ejects outermost (valence) electrons from the material under study, and this can enable a better understanding of the electronic structure of the material. UPS is often used for measuring density of states in materials. UPS therefore is mostly (but not exclusively) used for understanding general (bulk) properties and so is of less interest than XPS to those whose primary concern is the surface.

Monochromatic x-rays are generated in a metal source, such as a K_α source of magnesium or aluminium, in an ultra-high vacuum (UHV) chamber (figure 3). These x-rays, of energy 1254 eV (Mg $K_\alpha$) or 1487 eV (Al $K_\alpha$) are generated in an anode, which is usually a heated filament. The anode also generates other emission lines as well as the $K_\alpha$ lines that are often used, and so a crystal monochromator can be employed to remove unwanted lines. (Recent developments include the use of synchrotron radiation in XPS measurements which disposes with the need for a metal source. Furthermore these are tuneable in energy, and here crystal monochromators are particularly helpful.) X-rays are then incident on the sample and they will eject core electrons. X-rays, being uncharged and highly energetic, have a very strong penetration of most samples, but the photoelectrons ejected in the analysis do not travel far. Indeed, it is only the photoelectrons generated very near the surface that can escape the sample and this limits the depth of available information from XPS to no more than 10 nm from the surface, although in most cases it will be less than this.

Zoom In Zoom Out Reset image size

The energy of the core electrons is different from one element to another, and furthermore bonding changes this energy slightly. The photoelectron energy is measured, and the count rate plotted accordingly. The greater the photoelectron (kinetic) energy is, the less the electron binding energy must be. Alternatively, the count rate can be plotted as a function of binding energy, and an example of this is shown in figure 4. The binding energy is often plotted in a decreasing energy scale for easy comparison to the kinetic energy. Either way, the binding energy plots are more useful because it is these rather than the kinetic energy plots that can be compared between different x-ray sources. The binding energy is calculated using the simple Einstein energy relation,

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle E_{\rm b} = h \nu - E_{\rm KE}, \label{binding energy} \nonumber \end{align} \tag{ 1 }$$

where E_KE is the kinetic energy of the ejected photoelectron. This relation needs to be modified to account for the chemical shift due to different bonds (figure 4) or if the sample is conducting, but it nonetheless encompasses the basic physics.

Zoom In Zoom Out Reset image size

The data shown in figure 4 provide an example of the utility of XPS. Poly(3,4-ethylene dioxythiophene) (PEDOT) is often used as an anode or a hole-transport layer in contact with the anode. (It is often synthesized as a complex with poly(styrene sulfonate) (PSS), which acts as a dopant and allows processing of the PEDOT in aqueous solution because PEDOT is generally insoluble. It is known that adding high boiling point alcohols such as glycerol [48] or sorbitol [49] to PEDOT/PSS layers can improve performance of polymer devices, but optimizing performance is aided by a detailed knowledge of their structure. The XPS data in figure 4 show that there is a strong contribution of C–OH bonds in the scan, which is commensurate with a significant amount of glycerol at the surface [47]. The glycerol is not conducting (or semiconducting) and so charge transport between this hole-transport layer (or anode) and the semiconducting layer of the device must be though an insulating layer. Whether this insulating layer improves the quality of the device is not clear, but it is difficult to remove it because thermodynamics (surface energy) control its presence at the surface. It is known from XPS [50] and neutron reflectometry [51] measurements that, without the alcohols, the PSS is located at the surface, so even then the PEDOT is not in contact with the semiconducting layer.

2.1.2. Secondary-ion mass spectroscopy.

SIMS is a technique whereby medium-energy ions are incident on the surface and, on collision, send out fragments of the surface of the film which are detected in a mass spectrometer. SIMS complements XPS because, unlike XPS, it can provide molecular information. A SIMS spectrometer uses a beam of ions—argon (Ar⁺) is typical, but other ions (which may be positive or negatively charged) such as caesium, oxygen, gallium, and xenon are common—accelerated to an energy of typically 40 keV. This energy is large enough to eject many ions from the film, but, because they are charged, only material in the near surface region can escape, as is also the case for XPS. The typical set-up (figure 5) is in principle quite similar to an XPS system.

Zoom In Zoom Out Reset image size

The fragments ejected from the surface are detected in a mass spectrometer, and are differentiated by their mass/charge ratio. Most fragments are singly charged and so the mass/charge ratio simply represents the molecular mass in g/mol of the fragment. Example SIMS data are shown in figure 6 in which a mixture of poly(l-lactic acid) (PLLA) and polystyrene is compared with pure films of the two surfaces. (The PLLA is contaminated with polydimethylsiloxane (PDMS), which provides extraneous peaks.) Here, the phase separation of PLLA and polystyrene within the film disrupts the flatness of its surface, creating a film with a topographically structured surface [52]. It has long been known [53] that certain topographically structured surfaces can provide (for reasons not completely understood) a beneficial substrate on which eukaryotic cells can grow and proliferate better than they can on a flat surface [54]. In this work, osteoblast (bone-forming) cells do not grow particularly well on either the polystyrene or PLLA surfaces. PLLA is not particularly hydrophilic; despite its name, it is not an acid: l-lactic acid is polymerized and its carboxylic group is lost in the polymerization reaction. Nevertheless, the addition of structure by mixing the two polymers improves the surface for cell growth, and the SIMS data provide direct evidence showing that both components are at the film surface.

Zoom In Zoom Out Reset image size

SIMS and XPS experiments are complementary, and one can be used in place of the other. In the study of PLLA and polystyrene described here [52], XPS was also performed and a slightly increased polystyrene fraction at the surface was determined than for the SIMS experiments, which helps to add limits to the accuracy of the experiments.

2.1.3. Scanning probe microscopy.

The scanning probe microscopies are a suite of techniques in which a small probe scans across a surface, determining information about the surface topography, strength, chemical structure, or electrical or magnetic properties.

The most common form of scanning probe microscopy is scanning force microscopy (SFM) in which topographical information about a film surface is obtained by scanning the probe across that surface and measuring the probe deflection as it navigates nanoscale surface features. Alternatively the deflection is kept constant and the topography can be monitored from the vertical displacement of the probe [55]. Here the probe is a small cantilever with a tip of  ∼30 nm in radius, typically made of silicon or silicon nitride. (These materials allow for manufacturing accuracy coupled with durability during repeated measurements.) The key components of the SFM are shown in figure 7. As the tip is moved across the surface, the forces that the surfaces exert on the tip will change. If necessary, a feedback loop can be maintained to keep these forces constant. As the tip rises and falls a map of the surface can be obtained with great (sub-nanometre) precision. For polymer surfaces, dragging a SFM tip across a surface can cause damage to the surface and so intermittent contact is often desirable. In this 'tapping' mode the cantilever is driven to a frequency near its resonance and makes contact with the surface only during part of the oscillation. How much of the oscillation is spent in contact with the surface can be controlled. The key requirement is to tap hard enough so that the tip can detect the contours of the surface, but not so hard that it actually indents into soft materials [56].

Zoom In Zoom Out Reset image size

There are many other variants of the technique. Friction (or lateral) force microscopy (FFM) is a contact mode experiment, whereby the tip is dragged along the surface. An increase in friction results in an increase in lateral force on the cantilever, which will cause it to bend. The lateral force on the cantilever is directly proportional to the load, according to Amontons' law [57]. Here the load is the force on the cantilever driven by the electronics of the instrument. FFM is a particularly useful technique for determining interactions with the tip that might not be clear from measurements of topography. A flat surface with hydrophobic and hydrophilic components can be readily studied using FFM [58]. In the case of polymer surfaces, friction force microscopy can provide valuable information about the viscoelastic properties of a polymer surface [59] or material interactions, as schematized in figure 7, through grafting polymers to the tip surface [60].

The resolution of optical microscopy is generally restricted by the Abbe limit of approximately $\lambda /2$. This diffraction limit can be circumvented by the use of scanning near-field optical microscopy (SNOM), which is a scanning probe technique in which the probe is an optical fibre brought very close to the sample surface as schematized in figure 8. The use of SNOM to interrogate polymer surfaces is much less commonly used than SFM, but is nevertheless important, especially in areas such as polymer optoelectronics. Here, UV light is passed through the optical fibre and it illuminates the surface. The light may pass through the film as the probe scans across the sample, allowing a transmission image to be built up. Alternatively, scattered light may be detected, in which case a reflection image would be obtained. SNOM, however, is at its most powerful when the incident light from the fibre is used to excite optoelectronic molecules in the film. Through a judicious choice of illuminating light, the molecules in the material fluoresce. Here, an image may be taken locating chemical species in the film by their optical behaviour. This information may be obtained simultaneously with a topography image. The SNOM tip is small enough to be used in the same manner as an SFM tip. It is not ideal, but it is good enough for correlating optical and morphological properties of the film.

Zoom In Zoom Out Reset image size

Data that exemplify how well SNOM can be used to reveal information about a film are shown in figure 9. Here, a blend of the optoelectronic polymers poly(9,9-dioctylfluorene-alt-benzothiadiazole) (commonly denoted F8BT) and poly(9,9-dioctylfluorene-alt-bis-N, N'-(4-butylphenyl)-bis- N, N'-phenyl-1,4-phenylenediamine) (PFB) is imaged using SNOM [61]. The fluorescence is obtained in transmission in this case after being irradiated with 440 nm light from the SNOM. When mixed with F8BT, PFB emission is suppressed at long wavelengths, allowing a mapping of the the F8BT lifetime as a function of its environment. Here, it is likely that the exciting radiation generates excitons (Coulombically-bound electron–hole pairs) which affect the speed of fluorescence. The more PFB that is present, the slower the luminescence.

Zoom In Zoom Out Reset image size

Of course, there are many other examples of scanning probe methods. STM, for example has an extremely good resolution, but it is seldom used to study polymer surfaces because it requires conducting and flat substrates in order to function. Further developments have revealed the possibility of parallel processing of polymer surfaces using AFM [62] and SNOM [63, 64]. These techniques have been developed with an eye on memory storage and lithography, and are beyond the scope of this review.

2.1.4. Force spectroscopy.

Molecular force spectroscopy and single molecule force spectroscopy are AFM-based techniques used to measure the interaction of an object with a surface, and may or may not be linked to a scanning probe. The object in question may be a polymer, a bacterium or eukaryote, a microparticle, or any other suitably small material [31, 34, 66–70] that can be attached to an AFM cantilever. The cantilever is brought towards the surface and allowed to adhere, if indeed it does adhere, with its deflection monitored at all points along the approach. The cantilever is then removed, and again its deflection is monitored. As a material approaches the surface there is initially no measurable interaction, but very often an accelerated adhesion is observed as the material 'snaps to' the surface. At this point any further approach is repulsive because there is a compressive effect on the cantilever and material. Retraction of the cantilever will be affected by the adhesion of the material to the surface, so the probe will need to be pulled off the surface. For many particulate probes, this is the end of the process. The probe is retracted, and there is no further interaction. However, for polymer-coated probes or single polymer chains, mechanical work is done on the polymers after the initial detachment, assuming that some polymer is left on the surface. Here polymers are straightened as they are stretched away from the surface, with an associated entropic cost. This is reflected as an increasing force on the polymer as the tip is retracted from the surface. Results for different conformations of a polymer are indicated in figure 10. More complicated polymers include proteins, which can be denatured as they are pulled away from the surface, with associated information about the length scale of the folding. In the case where there are multiple points of attachment between the polymer and the surface, mathematical models can be used to extract information from the multiple binding events.

Zoom In Zoom Out Reset image size

Although DLVO theory can be used to describe the interactions measured by force spectroscopy [71–73], in its basic form it does not account for the interplay between all of the different types of force [74], being solely based on a combination of attractive van der Waals interactions and repulsive double-layer forces [41], and discretion must be used when selecting it for analysis [75, 76]. The extended form of DLVO theory includes some extra features, and can be a better model for forces in a more complex system [77]. However, there is currently no comprehensive theory for bacterial or cellular interactions at a colloidal level, as adhesion in these systems involves a wide range of surface molecules and polymers with different physio-chemical natures as well as contributions from cell elasticity and hydrodynamics [78].

Beyond DLVO theory, there are two alternative models that have been extensively used in the interpretation of force spectroscopy data, and model selection is typically based upon the predicted physical properties of the molecules involved in the interaction. In biological systems, data from flexible polysaccharides are more likely to be fitted using the 'freely-jointed chain' (FJC) model, whereas the 'worm-like chain' (WLC) model is generally considered more appropriate for DNA and proteins, which are more rigid [79].

The FJC model considers the polymer as comprising n rigid elements with a length $l_{{\rm K}}$ (the Kuhn length) that are connected through flexible joints which can rotate freely in any direction. At low forces, the polymer formation is that of a Gaussian chain, but as force is increased, orientation becomes less random, with preferential alignment oriented along the direction of the external force. Smaller Kuhn lengths correspond to a more flexible polymer [73, 80]. In the FJC model, the force required to stretch a polymer to a length x is given by

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \label{eq:FJC} F_{{\rm chain}} = \frac{- k_{{\rm B}}T}{l_{{\rm K}}}\boldsymbol{L}^{-1} \left(\frac{x}{l_{{\rm c}}}\right), \nonumber \end{align} \tag{ 2 }$$

where $k_{{\rm B}}$ is the Boltzmann constant, T is the absolute temperature, $l_{{\rm c}}$ is the contour length of the portion of the chain that was stretched, and $\boldsymbol{L}^{-1}$ is the inverse Langevin function, approximated by the first four terms of its series,

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \label{eq:Langevin} \boldsymbol{L}^{-1} \left(\frac{x}{l_{{\rm c}}}\right) = 3\left(\frac{x}{l_{{\rm c}}}\right) + \frac{9}{5}\left(\frac{x}{l_{{\rm c}}}\right)^3 + \frac{297}{175}\left(\frac{x}{l_{{\rm c}}}\right)^5 + \frac{1539}{875}\left(\frac{x}{l_{{\rm c}}}\right)^7 . \nonumber \end{align} \tag{ 3 }$$

An example of the application of the FJC model in a biological context is the investigation of bacterial surface macromolecules on Pseudomonas putida using SMFS. It was found that the biopolymers had segment lengths in the range 0.154–0.45 nm, but that 65% of measurements gave a segment length of 0.154–0.20 nm, suggesting that many of the polymers present on the cell surface were highly flexible [80].

In the WLC model, the polymer is considered as a continuous flexible chain of length $L_{{\rm c}}$ with a bending stiffness, κ, that can be used to evaluate the persistence length, $L_{{\rm p}}$ by [81]

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \label{eq:persistence_WLC} L_{{\rm p}} = \frac{\kappa}{k_{{\rm B}}T}. \nonumber \end{align} \tag{ 4 }$$

There is no analytical solution to the WLC model, but the most common approximation is the interpolated WLC [82], given by

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \label{eq:WLC} F = \frac{k_{{\rm B}}T}{L_{{\rm p}}} \left \lbrack \frac{1}{4} \left(1 - \frac{x}{L_{{\rm c}}} \right)^{-2} - \frac{1}{4} + \frac{x}{L_{{\rm c}}} \right \rbrack, \nonumber \end{align} \tag{ 5 }$$

where F is the elastic restoring force of the chain and x is the end-to-end separation distance. In some cases, the persistence length can be obtained from techniques such as small angle x-ray scattering [83]. The suitability of fixing persistence length can be evaluated by comparing experimental data and the WLC model in a force versus normalized distance plot. Such fixing was appropriate in a study of the swelling of grafted poly(methacrylic acid) (PMAA) layers [84], where $L_{{\rm p}}$ was taken as 0.5 nm (it had been previously determined using small angle x-ray scattering for bulk PMAA [83]), which gave a good fit up to high extension levels, and resulted in greater data fitting accuracy for the contour length, since only one parameter was fitted to the WLC model. This model has also been applied to data obtained when performing force spectroscopy on different bacteria [85–87]. (Hydrophilic silicon nitride tips do not always give the required peaks and on occasion hydrophobic tips were required to obtain data that could be fitted to the WLC model [88].)

Force spectroscopy is usually performed with an uncoated AFM tip, or an AFM tip coated with gold or a suitable self-assembled monolayer. With such tips, the polymers to be investigated may be in contact with the surface, in which case the process is known colloquially as 'fishing'. Molecules that cannot be functionalized, for example by thiolation to react with a gold-coated tip, are generally studied after being picked up on an AFM tip. (It is by no means always necessary to functionalize a molecule in order for it to attach to an AFM tip.) For this to work as a single molecule technique, the polymers must be dilute on the surface, and not entangled with other polymers. The tip can then pick up these molecules and thus measure the interaction with the surface through a retraction curve. This method can be very frustrating as most experiments will not yield a result, because the probe simply makes contact with the substrate. Automation allows for easier experiments, and is becoming increasingly sophisticated [89]. Generally, these fishing expeditions are for biomolecules, which are often very large, making the experiment a little easier. An uncoated AFM tip can be used to characterize dense layers of grafted polymers (brushes), which are tethered by one end to the substrate [90–97]. By retracting the tip, physisorbed polymers are extended and their length and dispersity in lengths can be assessed.

2.1.5. Quartz crystal microbalance with dissipation.

A key part of understanding the interaction of polymers and biopolymers at surfaces relates to adsorption, be it either favourable or unwanted. One method to measure this in both gaseous and liquid environments is to use a quartz crystal microbalance with dissipation (QCM-D). The QCM-D relies on a freely oscillating piezoelectric sensor formed of a thin single crystal quartz (SiO₂) disc with electrodes on the upper and lower face (see figure 11) [98]. Direct piezoelectric materials are well suited to use as sensors because mechanical deformation induces a proportional change in the electric polarization of the material [99]. In the reciprocal effect, application of an external electric field to the piezoelectric material induces mechanical stress in the crystal lattice, resulting in deformation of the crystal. When this crystal is oscillated using a sinusoidal electric field (i.e. an alternating voltage), the fundamental crystal resonance, f₀, is dictated by the thickness of the quartz in the sensor, t_q, the shear modulus of the quartz, $\mu_{\rm q}$, and its density, $\rho_{\rm q}$, such that

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle f_0 = \frac{1}{2t_{\rm q}} \sqrt{\frac{\mu_{\rm q}}{\rho_{\rm q}}}. \label{QCM_resonance} \nonumber \end{align} \tag{ 6 }$$

Zoom In Zoom Out Reset image size

For use in the QCM-D, quartz crystals are typically cut in the AT form, which generates motion lateral to the sensor surface and is relatively immune to temperature-induced fluctuations at room temperature [100]. This mode of operation is referred to as thickness shear mode.

In its simplest form, prior to a dissipation measurement, the quartz crystal is driven at a selected frequency close to its fundamental mode, setting up oscillations with a constant amplitude. At the start of the measurement the driving voltage is withdrawn, allowing the oscillations to decay naturally to zero. As additional viscoelastic material is adsorbed the rate of decay of oscillations will be 'damped' (dissipation) and the resonance frequency of the oscillations is associated not just with the quartz sensor, but with any adsorbed material [101–103]. More commonly, and to allow for much longer measurements to understand the binding of multiple molecule types by sequentially flowing them across the sensor (generally interspersed with flow of a wash buffer to remove any non-adherent molecules in between tests, see figure 11), a continuous resonance mode can be used, where the driving electric field is maintained throughout and relative shifts in the crystal frequency ($\Delta f$) are measured [100]. The QCM-D can therefore be used to track adsorption of molecules and biomolecules to the sensor surface [104]. In addition to monitoring the amount of adsorbed material, it can also provide information about the viscoelasticity of the adsorbed molecules, and is therefore useful for testing for adsorption barriers [105].

The mass of adsorbed material can be evaluated using the Sauerbrey model, which neglects viscoelastic effects, and assumes a rigid adsorbed layer. According to the Sauerbrey equation [106], the frequency change $\Delta f$ in the piezoelectric crystal due to the adsorbed mass $\Delta m$ is given by

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \Delta f = - f_0^{3/2} \sqrt{\frac{\eta_{\rm l} \rho_{\rm l}}{\pi \rho_{\rm q} \mu_{\rm q}}}, \label{Sauerbrey} \nonumber \end{align} \tag{ 7 }$$

where $\rho_{\rm l}$ and $\newcommand{\e}{{\rm e}} \eta_{\rm l}$ are the density and viscosity of the fluid, respectively. The adsorbed mass is determined from the changes in viscosity. Masses given include that of any water that is bound or coupled to the surface. This model is appropriate if changes in energy dissipation are small and the adsorbed layer is relatively rigid. The use of an unmodified Sauerbrey equation for soft or viscoelastic films can lead to an underestimate of adsorbed mass [102].

Dissipation data may also be obtained to allow analysis of viscoelastic effects; if the material is tightly-bound and rigid, then minimal dissipation modification would be expected, so the greater the dissipation change for a unit change in gained mass, the more viscoelastic the adsorbed material is [107]. This increase in dissipation can be caused by two things: a more viscoelastic molecule or a poorly attached layer causing dissipation due to friction associated with moving out of synchronization with the resonating crystal. If there is doubt, the origin of the dissipation can be established from thickness data obtained by methods such as ellipsometry; if the change is related to molecule properties, the layer should be thicker, whereas if it is due to loosely bound material there should be a thinner layer. One might argue that ellipsometry is as versatile a method, but QCM-D offers much higher sensitivity and has been shown to detect single molecule binding events, which is beyond the capability of any ellipsometer. However, the benefits of combining QCM-D with other surface analysis techniques within a single system (such as ellipsometry [105] or localized surface plasmon resonance [108]) is not without merit, as this enables additional information to be obtained simultaneously through the same sample and measurement platform.

Techniques that allow us to understand the variation of a particular quantity as a function of distance from an interface are known as depth-profiling experiments. The ability to provide information away from the exposed surface of a film is very important, and particularly so in the case of buried interfaces. Experiments in aqueous solution are important for biological samples; these are often not of interest in a dry state. To image buried interfaces is not easy. If the interfaces are in a low viscosity liquid environment then scanning probe techniques can be used, assuming the liquid will not damage the apparatus, but other techniques described above such as SIMS and XPS are inappropriate due to their vacuum requirements. For these reasons most buried interfaces can only be effectively studied in one dimension by depth profiling.

2.2.1. Reflectometry.

Perhaps the most powerful depth profiling technique is that of neutron reflectometry [110–113]. Here, a beam of neutrons is incident on the sample to be studied. The neutrons may be monochromatic, or as is more often the case, pulsed with a spectrum of wavelengths in the Ångstrom range. The very short wavelength of the neutrons, and for x-rays when considering the similar technique of x-ray reflectometry [111], allows for an extremely good depth resolution. Neutrons are versatile because they are only weakly scattered by many materials, which means that they can traverse large distances before reaching the interface in question. Water, unfortunately, does scatter neutrons, but it is not necessary to send neutrons through a bath of water in order to reach the interface to be studied; neutrons can interrogate the interface in question by passing through many substrates with little loss of intensity (e.g. silicon).

Contrast in neutron reflectometry experiments is most often achieved by selective deuteration of one of the components in the film. If two polymers are mixed, then one of them would normally be deuterated. In aqueous environments it is generally advantageous to use heavy water, because of the extensive scattering of neutrons by normal water (H₂O). Although it is possible to account for this scattering by careful analysis, it detracts from the quality of the data. The use of heavy water does make some biopolymer experiments difficult to consider, because often H/D exchange occurs in biopolymer solutions, particularly in the case of certain proteins [114]. Nevertheless, neutron reflection has been used in many experiments to study biomacromolecules, and an example is shown in figure 12 for the rather shorter chain molecule, mycolic acid [109].

Many of the advantages of neutron reflectometry also apply to x-ray reflectometry. X-ray scattering techniques work best when there is a heavy element present in the sample, as these provide contrast. This restriction is not prohibitive, however, and x-rays are also useful for the characterization of the surface of thin films. X-ray reflection is, however, not viable for the study of buried liquid interfaces. Nevertheless x-rays are less expensive than neutrons and are available in laboratory apparatuses. Which technique is used depends therefore on a variety of different parameters but the information obtained is broadly similar. By way of example, the use of x-ray reflectometry to study the structure of films to explain adhesive behaviour [115] provides information broadly comparable to studies used for similar purposes with neutrons [116–120].

Zoom In Zoom Out Reset image size

2.2.2. Ion beam analysis.

Other depth profiling techniques require the use of ions to penetrate the film, revealing chemical information as a function of depth. Dynamic SIMS, for example, involves the ion beam etching the sample [121, 122]. By knowing the etching rate, the chemical information yielded can be correlated with the depth. Dynamic SIMS is less often used, partly because of the difficulties in interpreting data when different components in the film have different etching rates. Contrast, such as H/D differentiation, although not always necessary, can help data interpretation.

Techniques falling under the umbrella of ion beam analysis do not include dynamic SIMS, which is more a mass spectrometry technique. Ion beam analysis generally requires a means of monitoring the energy of ions as they exit a film that was interrogated by a beam of ions, which may or may not be the same as those incident on the film. With ion beam analysis, different techniques allow for a very flexible approach. The best known ion beam technique is Rutherford Backscattering (RBS) [123, 124]. Here a beam of ions (usually α-particles) of energy of the order of 3 MeV is incident on the sample. These ions have a small collision cross section (recall the revelations about the interaction of α-particles with gold and other metals in the original experiments of Geiger and Marsden [125]) and so it takes a relatively thick film, up to 10 μm, to stop them all. By measuring the energy of the detected α-particles, one can determine the depth in the film at which the backscattering event took place. (The energy loss of α-particles in different materials is tabulated [126].) With these measurements, a depth profile can be constructed. RBS is less commonly used in soft matter research, because there are relatively few polymeric systems that contain the heavy elements required for good RBS data, although some studies have been performed [127], and other studies have been concerned with the location and behaviour of heavy elements in polymer matrices [128–130]. Figure 13 shows an example of such data. Heavy elements can, however, also be added as markers to define the movement of polymers [131], or to stain an individual component of a mixture [132]. RBS, like SIMS and XPS, is restricted by a vacuum requirement because the α-particles will be scattered in air. Nevertheless, developments have been made to allow their use to study samples in different environments [133, 134], although their application is away from the areas covered in this review.

Zoom In Zoom Out Reset image size

Ion beam techniques more commonly used for the study of thin polymer films include nuclear reaction analysis (NRA) and forward recoil spectrometry (FReS) because they both allow a means of determining the profile of light elements in the sample, and as such are very often complementary to neutron reflectometry. (Neutron reflectometry provides excellent resolution, but because the information obtained is in Fourier space, it is often useful to start with depth profiling information which can provide the initial conditions for fitting the reflectivity data.) FReS, sometimes called elastic recoil detection analysis (ERDA) is a relative of RBS. In RBS the incident α-particle recoils from the nucleus. In FReS the energy of the deflected α-particle is not needed, but rather the energy of protons and deuterons ejected from the nucleus are measured. Their energy is obtained using kinematics, just as in RBS; here however the nucleons are forward scattered, as opposed to the backscattering in RBS. Backscattering means that the incident particles recoil on hitting the nucleus and forward scattering means that the ejected particles are sent away from the initial beam. The forward scattering is a consequence of the protons and deuterons being a lighter mass than the incident α-particles. An advantage of FReS is its ability to determine the volume fraction-depth profile of both deuterated and non-deuterated components of a film.

Nuclear reaction analysis is another ion beam analysis technique that can provide a volume fraction-depth profile of a given component in a film. There are many different variants of NRA, but the most popular for the determination of the structure of polymer films is ³He NRA. Here a mono-energetic beam of ³He⁺ ions is incident on the film. The energy of the ions is typically  ∼1 MeV, and at about this energy ³He⁺⁺ (the other electron is immediately lost on impact with the film) reacts with deuterons to yield a lithium compound nucleus, which quickly decays to create an α-particle and a proton, as well as about 18.5 MeV of energy; the reaction is very exothermic. The ³He nucleus loses energy as it penetrates the sample, and this changes the reaction kinetics, altering the energy of the proton and α-particle produced. Either the proton or α-particle can be detected, and their energy allows a calculation of the position in the sample at which the reaction took place. (How well ³He is stopped by matter is tabulated [126], just as for α-particles, so one can work out the energy that these nuclei must have had at a certain depth in the sample.)

The stability of most polymer films can be characterized by the use of contact angle experiments (figure 14), which typically give a description of the free energy of a sample compared to that of water: hydrophilic surfaces are those where the interfacial energy of the water-solid contact is below the free surface energy of the solid, whereas hydrophobic surfaces exhibit the opposite relationship, with the solid having the smaller surface energy, or tension.

Zoom In Zoom Out Reset image size

In systems where the contact angle does not vary with time and an equilibrium, static contact angle is attained, the surface energies in the system can be related to the contact angle θ, by

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \label{eq:Young1} \gamma_{\rm LV} \cos{\theta} = \gamma_{\rm SV} - \gamma_{\rm SL}, \nonumber \end{align} \tag{ 8 }$$

where $\gamma_{\rm LV}$, $\gamma_{\rm SV}$, and $\gamma_{\rm SL}$ are the liquid–vapour, surface–vapour and surface–liquid interfacial energies, respectively. This equation is only true for an ideal surface (flat, rigid, insoluble, chemically homogeneous, and unreactive) and it takes its name (the Young equation) from Young's work in the early 19th century [135].

The contact angle is particularly relevant for biomaterials, where very hydrophilic surfaces tend to be associated with good biocompatibility. The ability of a material to remain solvated in the presence of other macromolecules means that that surface is less likely to be fouled. Wettability of surfaces is also a strong indicator of adsorption, with profound consequences for cell growth and protein fouling [136]. This can be characterized through instability in the temporal behaviour of the contact angles, with changes in droplet volume (in conditions with minimal evaporation) either due to droplet absorption, spreading, or a mixture of the two [137]. In the case of such dynamic contact angle measurements, wetting of a hydrophilic surface is associated with an advancing water contact angle [138], and dewetting of a hydrophobic surface with a receding contact angle [139]. Absorption is not always relevant because it requires the surface to have some limited solubility in water but can be considered in terms of a decrease in droplet basal area along with a decrease in volume, whereas spreading results in an increase in basal area.

Both static and dynamic contact angle measurements play an important role in the characterization of polymer surfaces for use in various applications, including the medical field, in terms of reducing harmful biofouling or encouraging native cell growth on implants [140, 141], optometry and contact lens anti-fouling properties [142–144], wetting agents [145], in-built drug delivery [146, 147], and the replacement of petrolium-based polymers with biopolymers which are able to replicate their hydrophobic properties [148].

Although the water contact angle is an important characterization tool for biomaterial surfaces and for surfaces that may be placed in contact with water, it is less useful for understanding if a thin film is stable. The Young equation (equation (8)) does allow prediction of film stability, and can be used to define stability through the spreading coefficient,

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle S = \gamma_{\rm SV} - \left( \gamma_{\rm SL} + \gamma_{\rm LV} \right). \nonumber \end{align} \tag{ 9 }$$

When S  >  0, the solid-vapour interface is the high energy surface, and so the liquid will preferentially coat that surface; i.e. the film will wet the surface. Similarly, when S  <  0, the film will be unstable on the surface and will want to dewet the substrate.

The stability of thin polymer films depends predominantly on long-range interactions between the film and its environment, and these generally take the form [41]

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \label{eq:dispersion} W_j(x) = - \frac{A_{ijk}}{12 \pi x^2}, \nonumber \end{align} \tag{ 10 }$$

where W_j represents the interaction energy of two parallel and planar semi-infinite media (i and k) separated a distance x by a medium j. The parameter A_ijk is known as the Hamaker constant and depends on all three materials, although approximations can be made to calculate A_ijk from Hamaker constants of the component materials, A_ii where the two (identical) components are separated by vacuum. A more detailed (Lifshitz) treatment relies on the dielectric properties of the different layers [41].

The different Hamaker constants allow an interplay between the stability of different components in polymer films. In principle a film (B) on a substrate that may be stable in itself can be made unstable by placing a different film (A) on top of it. If the interaction of A with the substrate is stronger than B, even accounting for the fact that A is separated from the substrate by B, then B will dewet. Such a 'phase' diagram for dewetting (figure 15) has been demonstrated for bilayers of poly(methyl methacrylate) (PMMA) on polystyrene on silicon substrates [149].

Zoom In Zoom Out Reset image size

Whilst varying the thickness of the layers allows the interaction between the films to be tailored to control stability, a film may create its own interface potential that can strongly affect film stability. Diblock copolymers with each block containing similar chain sizes will order to form a stacked (lamellar) structure, with the block that lowers the interfacial energy segregating to the substrate and the block that lowers the surface energy segregating to the surface. These blocks may be the same, or different blocks may segregate to the different interfaces. A lamellar structure will only form if the ratio of the chain lengths in each block is close to unity and the blocks are immiscible. A surface will perturb the structure, and even if the two blocks are miscible, they will order close to the surface, with the amount of order decaying with distance from the substrate. This creates an oscillating interface potential the minima and maxima of which become less pronounced with distance from the substrate, which permits a situation where the melt above a residual layer ordered at the substrate may dewet [150]. Here the dewetting of a diblock copolymer containing immiscible blocks (poly(2-vinylpyridine) and polystyrene) occurs to allow a disordered state to exist close to the substrate whilst retaining a lamellar structure at the air interface, and can proceed when the energy driving ordering is weak. However, the differing diblock copolymer film thicknesses at which this dewetting may occur result in discrete contact angles, even though the composition of the film is the same.

Other recent challenges in dewetting concern the effect of the preparation of thin films on their subsequent behaviour. The microscale morphology of films, and the conformation of polymers within the structures produced by common techniques such as spin coating, are not expected to be at equilibrium. Indeed, there are no methods currently available that routinely produce films at equilibrium. Part of the reason for the resultant out-of-equilibrium structures is the rapid quench in many drying processes, but also there will be a contribution due to the presence of solvent within the polymer film after casting. Later evaporation of the solvent leads to residual stresses within the film, which in many cases is already glassy. There is a clear effect on the distribution of holes in such films, as has been determined by simple optical microscopy experiments [151], where it was observed that the longer the films have to relax, the fewer the number of holes which appeared in the dewetted film (figure 16).

Zoom In Zoom Out Reset image size

Relaxation effects are categorized in terms of a response to 'residual stresses' in polymer films. The nature of these residual stresses is open to some debate, and they are likely to be dependent on the method of film preparation. Nevertheless, there are other generic factors that may play a significant role across different preparation routes, albeit to different degrees. One of these is the role of entanglements. In confined films, entanglements are expected to remain in films at equilibrium [152], but it is accepted that this is unlikely to be observed in experiments and that polymers may have perturbed conformations [1, 153–156]. Although a molar mass dependence of polystyrene film stability was observed on topographically structured surfaces [153] or surfaces with a gradient in surface energy [157], these experiments did not perturb stresses within the films. Other experiments have shown that long (polystyrene) chains were capable of relaxing rather quickly, indicating that they were closer to equilibrium than shorter polymers of the same chemical structure [158].

A rather clever means of addressing stresses in thin films was to create them in a controlled manner during dewetting. This was achieved by spin coating polystyrene onto PDMS substrates [159]. Initiation of the dewetting perturbs and deforms the substrate, which arrests the dewetting. The process continues on time scales longer than the reptation time of the polymer, which is its longest viscoelastic relaxation time and is the time it takes to diffuse its own length, which is valid for polymers that are long enough to be considered entangled [8, 9]. On these longer time scales, the dewetting was that of a viscous fluid, with a behaviour dominated by slippage. Slippage in itself has been of some considerable interest in dewetting since systematic SFM experiments were able to follow the shape of the rim during dewetting from a deformable polymer interface [160]. The shape of the rim is important, because it can reveal much about the viscoelastic behaviour of polymers. An initial simple approach to dewetting proposed that the gain in surface energy by the dewetting process is translated into the kinetic energy of the dewetting polymer [161], which is only valid for a low viscosity liquid on a non-deformable substrate. For deformable and liquid substrates, the dewetted polymer accretes into a rim, whose shape depends strongly on the viscosity of that polymer, as well as that of the substrate. Here friction can also be important, and this will increase with the size of the rim, which forces the dewetting speed to decrease. Of course, if friction is limited, then the dewetting polymer may well slip at the interface. Its tendency to slip depends on how much the chains of the two layers can interpenetrate; generally, interpenetration or interdigitation is limited, because otherwise the layers would be compatible and the films stable. Certainly, it has been clearly demonstrated that dewetting proceeds more rapidly on surfaces with less slippage, and hence less friction [162, 163].

Applications of dewetting in various nanotechnologies are still being developed but the basic science must be fully understood before real progress can be obtained. It is expected that dewetting can be advantageously used in a number of new areas, in particular those where patterning is important [164]. For example, the use of selective dewetting to align electrode materials has been developed, a route that may be easily applied using ink-jet printing [165, 166]. Microcontact printing and other patterning techniques can be used to control the structure of films [167–171]. Although the fundamental science of dewetting is still proceeding, the development of technologies based on dewetting probably has slowed somewhat in the past decade and it is perhaps likely that a new breakthrough is needed to reignite the field.

Ever since the observation of electrical conductivity in excess of 10 kS/m in doped polyacetylene was reported in 1977 [172], there has been interest in developing polymers for electronics applications. Although conducting polymers, and PEDOT in particular, have applications in various areas from antistatic films [173], transparent electrodes [173, 174], and bioelectronics [175, 176], it is in the area of polymeric semiconductors where the greatest interest is to be found. Interfaces in polymer electronics are crucial because they influence and direct morphological behaviour, and can also influence the electronic properties directly.

Blends are of primary interest in photovoltaic devices and light-emitting diode (LED) technology. In the case of LEDs, a current is driven through a forward-biased polymer blend, in which one component is a hole-transporting and the other an electron-transporting polymer. Electrons and holes meet, and form a bound state, called an exciton [37]. This exciton may decay back to an electron and hole, or it may radiatively decay, giving off a visible photon. (The band gap is the primary determinant of the wavelength of the light emitted.) Clearly for an optimal LED, the amount of excitons giving off light must be maximized, and the number of holes and electrons making it to the opposite electrode without giving off light must be minimized. A parameter by which a device may be controlled is the ratio of the two components in the polymer layer. Other aspects, such as the contact between the layer and electrodes, and the benefits of injection layers will not be discussed here, but they are important considerations [177], and they also affect polymeric photovoltaic cells and transistors [37, 178].

Solar cells essentially work in the opposite sense to LEDs, with incoming light creating an exciton in the blend, which is required to dissociate at an interface between the two components. This dissociation will create electrons and holes, and these must dominate over the competing process of re-radiation. A small bias potential is placed over the cell in order to optimize current collection. The difference between the morphology of the active layer in photovoltaic devices and LEDs is that photovoltaic cells require sharp interfaces at which excitons can dissociate, whereas LEDs are presumed to benefit from a greater degree of mixing because sharp interfaces can provide charge traps [1, 37], which are certainly not desired. In both cases, a clear path for the charges to the electrodes is necessary. These requirements for photovoltaics and LEDs are largely based on the understanding of the behaviour of charge transport and excitonic behaviour rather than systematic experimental studies linking morphology to optoelectronic behaviour.

Despite the difficulties in creating systematic studies, it is possible to link structure-property relationships, and this will be exemplified considering a blend of poly(9,9-dioctylfluorene) (F8) and F8BT which is commonly used as a model LED. Electroluminescent properties of blends of these two polymers are optimized when films are made with 5% F8BT [180]. Varying the composition of F8BT with respect to F8 changes device properties, but also enables a test of the effect of morphology. For example, casting this blend (F8/F8BT) from different solvents reveals different structures, as can be seen in figure 17 for toluene and chloroform [179]. The thin layer of F8 attracted to the surface in the toluene-cast film has a profound effect on the photoluminescence properties of these blends. In photoluminescence experiments light is incident on the surface and the radiation emission from the surface is detected. When F8 is irradiated with 405 nm wavelength light, it fluoresces at around 430 nm. However, when mixed with F8BT, this fluorescence is quenched and a broad green emission at  ∼500 nm is observed. Here, energy transfer between F8 and F8BT takes place. For blends cast from chloroform, where the F8 and F8BT are much better mixed, there is indeed energy transfer. However, when the film is cast from toluene, the blue emission from F8 is retained remarkably well even at concentrations as low as 50% F8. The F8 layer at the surface is pure enough to fluoresce. The results are shown as a function of composition in figure 18 [181].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Photoluminescence is not usually an important factor in real devices, but the data shown in figures 17 and 18 illustrate important phenomena, namely how the role of the casting solvent for the blend can affect device properties. In real devices, other effects can be important, and the role of crystallinity is particularly important in polymer devices, because of the strong affect crystallinity has on charge transport. For example, in model solar cells, blends of poly(3-hexylthiophene) (P3HT) and the fullerene-based [6,6]-phenyl-C₆₁-butyric acid methyl ester are particularly susceptible to the crystalline behaviour of P3HT [182, 183], which emphasizes the importance of understanding the crystalline structure of P3HT [184, 185]. Structure-property relationships in polymer electronics is an active and ongoing area of research. New materials are continuously being developed and optimized, so new understanding on how to prepare the optimal morphology is important. Isoindigo-based polymers [186, 187] are an example of a recent area of research; their structure-property relationships have already been the focus of different studies [188, 189].

Polymers cannot form single crystals of the quality that is necessary in many industrial processes, such as the polysilicon used in solar cells. Routes to high purity crystals require chemical vapour deposition, which is incompatible with polymers, which have no vapour pressure at room temperature, and decompose at high temperature. Excepting those that are atactic, polymers are inherently semicrystalline. The crystallization starts with the formation of lamellae [190–193], due to chain folding effects. These lamellae grow until spherulites are formed. However, an important question is whether there is an energy barrier to the formation of polymer crystals. Given that lamellae grow at different rates and that the fastest growing rate dominates the crystallization process, crystallization is accepted to be a kinetic phenomenon, and so it is perhaps more likely that an activation energy is required. (There has been some consideration of a spinodal process [194], which does not require an activation energy and would be spontaneous.)

The role of a surface in polymer crystallization is important, particularly if it is accepted that polymer crystallization proceeds by nucleation. Here the surface can act as a nucleating site for crystallization, which would mean that crystallization should occur at a higher temperature on cooling, or at a lower temperature on heating. Experimentally, this has been shown to be the case. Grazing incidence x-ray diffraction has been used to show that poly(ethylene terephthalate) (PET) will start a process of surface crystallization at a temperature close to its glass transition temperature, and below that for which bulk crystallization occurs [195, 196]; SFM has also been used to confirm the lower temperature of surface crystallization in PET [197]. Similar results have also been shown for polymers confined within nanopores [198].

Although the temperature is a significant parameter in polymer crystallization, film thickness also plays an important role, because confinement controls crystallization. This is apparent by the observation that crystallization is faster at the surface than in the bulk [196]. Although x-rays are an important means of observing crystallization phenomena in confined films, they are by no means unique. Dielectric spectroscopy has been used to observe PET crystallization as a function of film thickness [199]. Here the structural relaxation time is observed to increase for films thinner than 20 nm (figure 19). This relaxation time is associated with the glass transition of the films (it is also known as the α-process), and its increase suggests a glassy film. In this case, a layer of PET was observed to be irreversibly adsorbed onto the aluminium electrodes (substrate) required for the broadband relaxation spectroscopy experiment. This strong adsorption is responsible for the increased glass transition for these thinner films. For these very thin films, crystallization was not observed and no crystallization time was measured. The 20 nm thickness limit on the crystallization suppression was approximately twice the thickness of the adsorbed layer. It had been known previously that crystallization was strongly suppressed in ultrathin films [202], but there is nevertheless a mechanism for the formation of crystallized structures in two-dimensional confinement [203, 204]. In these examples, a monolayer-thick film of PEG can dewet from its substrate, forming crystalline structures as it does so. Another route to highly confined crystallized structures was demonstrated with the synthesis of polyethylene nanoparticles [205]. These semi-crystalline layers comprised a mere fourteen polymer chains, and in this case were not found on the surface, but were formed sandwiched between amorphous layers.

Zoom In Zoom Out Reset image size

Much of polymer crystallization research has been dedicated to semiconducting polymers. Semiconducting polymers are generally rather rigid, with a long persistence length, which means that they have a smaller entropic cost in forming lamellae than more loosely bound polymers. Formally, semiconducting polymers behave as 'worm-like chains', whereas more flexible polymers can often be considered as 'freely-jointed chains' (see section 2.1.4) [9]. Furthermore, the ability to form semicrystalline phases has advantages in terms of the charge transfer of these polymers. Semiconducting polymers are often conjugated, which means that they display alternating single and double bonds, along which charges can move. Movement of charges along a conjugated backbone is impeded by 'traps', which are more likely near a kink. Certainly, there is experimental evidence linking the size of crystalline domains and charge transport phenomena [206]. It has also been shown that changing the nature of the material substrate will change the way that the film forms, affecting its crystallization. Polymers are known to align on structures that have a direction associated with them; these are often called alignment layers, and grazing incidence x-ray techniques have been used to determine the behaviour of polymers on these surfaces [207, 208].

The role of crystallization and its impact on charge transport is an underdeveloped subject, however, because there is evidence of charge transport remaining good, even though interfaces may be rough, or amorphous materials have been added that would normally be expected to limit polymer performance [209, 210]. Similar results have been shown to be the case in mixtures of small molecule semiconductors with polymeric additives (binders). The addition of polymers to small-molecule semiconductors is important because polymeric additives can allow greater control of the solution viscosity and the creation of good quality films from a variety of different coating methods, such as dip-coating, and particularly processing routes appropriate for real-world applications, such as ink-jet printing [211]. The first demonstration concerned the addition of different polymers to an important small-molecule semiconductor, 6,13-bis(triisopropylsilylethynyl) pentacene (TIPS-pentacene) with different polymeric additives [212], which was followed by further work using the same semiconductor [213–219]. In all of these, some effort was made to understand the location of the different components, with XPS used to locate the polymer and small molecule near the surface [212]; time-of-flight SIMS [214] (in three dimensions, figure 20); neutron reflectometry [213]; optical absorption and x-ray diffraction coupled with solvent etching [216]; Raman spectroscopy [217]; electron microscopy [218]; and profilometry [219]. As an aside, it is worth noting that similar effects have been observed with a range of small molecules blended with TIPS-pentacene rather than polymers [220], but it is clear that polymeric additives have better film-forming properties.

Zoom In Zoom Out Reset image size

The TIPS-pentacene segregates strongly to only one interface (that of air) when cast from a blend with an amorphous polymer [212, 213], although this is a non-equilibrium structure and annealing ensures [213, 214] that TIPS-pentacene segregates from a blend with poly(α-methylstyrene) (PαMS) to both interfaces. (Large molar mass PαMS also segregates to both interfaces [213].) This is important because segregation of the TIPS-pentacene to the active interface ensures that there is good charge transport where it is needed. Importantly, however, the need for annealing can be removed by using a semi-crystalline polymer instead of PαMS (or atactic polystyrene [212]). It seems that the speed of crystallization (or structure formation in general) is an important parameter, and the use of crystallizable polymers such as isotactic polystyrene and poly(α-vinyl naphthalene) [212] means that the TIPS-pentacene can segregate to the surface forming crystals during spin-coating. A similar result was obtained by substantially increasing the molar mass of the amorphous material [213]. It is probable that the surface initiates the formation of TIPS-pentacene crystals. Of course, the lower surface energy phase (including remaining solvent) is directed towards the surface. The surface energetics of phases containing crystalline and semi-crystalline materials is therefore of some interest. Isotactic polystyrene could be mixed to 90% of the film and the field-effect mobility of holes due to TIPS-pentacene (the other 10%) in a transistor created from the blend was shown to be as good as a transistor created with 90% TIPS-pentacene [212]. (Isotactic polystyrene is semi-crystalline, unlike its amorphous atactic counterpart.) The competition for surface crystallization is therefore critical for the creation of good quality devices. That TIPS-pentacene segregated to the (silicon oxide) substrate in a blend with isotactic polystyrene was determined by XPS measurements of the substrate-segregated side of the film, after it had been lifted from the substrate [221]. Other experiments show indirect evidence of TIPS-pentacene segregating to a photoresist interface [219].

Understanding surface and interfacial properties is important for situations where appropriate surfaces are needed for a given purpose, but the ability of surfaces to actuate and interact with their environment conveys a form of smart behaviour that allows ex situ control of a device. Work on mixed brushes illustrated the possibility of switching their conformation [222]. In this early work, the environment was shown to be capable of switching the surface composition: a toluene environment would bring a polystyrene component to the surface, but in an acid environment, the poly(2-vinylpyridine) brush would coat the surface. This allowed control of the hydrophilicity of the surface. Such switchable properties (hydrophilic to hydrophobic and vice versa were later shown to be capable of being actuated by electric fields for smaller molecules [223]. Temperature responsive brushes were also developed starting with poly(N-isopropyl acrylamide) (PNIPAm) [224], a model temperature responsive polymer.

An environmentally switchable surface is one that requires no user interaction to make the surface change behaviour. Some reports have claimed switchable behaviour, but simply reported environment-dependent behaviour rather than demonstrating true switching [225–227]. Two components may not attach in a certain environment, but it does not follow that they will necessarily detach under those same conditions. An early and rather clever AFM-based experiment to measure in situ reversible behaviour involved the attachment of poly(methacryloyl ethylene phosphate) brushes to an AFM cantilever that was inserted into a liquid flow cell, through which pH or salt concentration could be altered [228]. Environment-induced conformational changes forced a change in the deflection of the AFM cantilever, which was therefore a signature of the environment. In essence, these experiments provided a prototype for how a nanotechnological 'nose' might operate as a chemical sensor.

Currently, research into switchable behaviour is dominated by wettability and adhesion. These two properties are closely related, since the latter is in part a consequence of the former. Switchable wettability is not as intrinsically interesting as switchable adhesion, because it is hard to imagine how a switchable surface would not exhibit a change in contact angle. Nevertheless, there has been substantial interest in switchable wettability with a number of review articles published [229–234]. Surface control is usually achieved by using polymer brushes in different geometries [224, 235–244]. Experimental work has also been augmented by a growing body of theoretical studies [245–247].

There are other means to control solubility and thus brush conformation which have been investigated, such as passing gases through the solution [248]. Nevertheless, the main route to switching polymer brushes is to use environmental factors such as pH and temperature. It is however sometimes desirable to use external electric fields to switch conformation, and there has been some experimental success demonstrating this [249–251] but the absence of a substantial body of work reflects the very real difficulty in producing reliably repeatable experiments. There has however been significant interest in the theory of polyelectrolyte brushes in electric fields [252–257]. There are other possibilities, with different advantages.

Actuation is a rather vague application for what is simply a conformational switch, but there are nonetheless many possible uses for brushes that undergo such changes, whether environmentally triggered or otherwise. The use of hydrogels for pH-control of fluid flow has long since been demonstrated [259], but the pH-switchable nature of polyelectrolyte brushes can be used for electrochemical logic gates. Electrochemistry is, in its most basic form, integral to the switchable nature of polyelectrolyte brushes because it is charge that controls their conformational behaviour [236, 237, 239, 250, 260]. However, for the control of electrochemical reactions not explicitly involving the brush, a mechanism has been shown that involves polyanionic brushes (polycations can also be used in the appropriate circumstances) attracting redox probes to the brush, which in turn allows electron transfer to take place at the substrate surface. If the pH is adjusted (lowered for polyanions), the polyelectrolyte loses its charge and the function is turned off (figure 21). Electrochemical switching was seen to work well using standard silicon substrates [258], eliminating the need for the metal surfaces used in earlier work [260]. It is probable that the ability to replace standard metallic electrodes with silicon is likely to be of greater importance than the switchability aspect, and to this extent polyelectrolytes are not necessary [261].

Zoom In Zoom Out Reset image size

The control of adhesion by environmental change remains a particularly appealing use of switchable surfaces. The demonstration of environmentally switchable adhesion using polymer brush layers has been achieved by a number of groups [117, 262–265]. The simplest example is that where pH controls the adhesion, and here a polyelectrolyte brush has been shown to have its adhesion with hydrogels switched off when the pH is changed in aqueous solution. The hydrogel used need not itself be a polyelectrolyte [263], but oppositely charged polyelectrolytes are certainly very effective [117, 265]. Switchable adhesion between oppositely charged polyelectrolyte brushes has also been demonstrated [262], although in this case the 'switch' was not pH, but rather the ionic strength of the medium. Adding NaCl was observed to cause the adhesion to fail. Since the adhesion was due to the electrostatic interaction between the two brushes, the salt was able to shield these charges and interfere with adhesion. This phenomenon is interesting, because it affects materials that are already adhering, which is important because it presents another means of controlling adhesion. However, it also demonstrates the challenge of applying such switchable adhesion in biological systems, which have considerable salt present. Bonding between two identical brush layers has also been demonstrated, in which polyzwitterions were shown to adhere due to long-range forces between the components [264]. However, in this case a change in temperature caused the brushes to swell, terminating the adhesion. Despite these efforts, application development is still not forthcoming. This is perhaps surprising because there are a number of areas where reversible glues could be useful, with a headline example being in wound dressing; removing dressings from a wound has risks of infection and numerous therapeutic challenges [266]. Recycling is another area because separating parts without damaging individual components decreases waste [267], with perhaps the automotive industry leading the way in separation technologies [268]. Despite these and many other needs, it is an open question as to whether brushes are robust enough to withstand the wear to which they may be subjected, and there is also the problem of ease of use because 'one-pot' reversible adhesives of this nature are currently lacking. Nevertheless, these problems are related to development and it is inconceivable that there are no uses for what is likely to be a relatively inexpensive and versatile technology.

That surface switchability is exhibited more often than not in aqueous environments and that many water-soluble polymers are biocompatible leaves open the possibility of engineering switchable surfaces with biological applications. An early example of this is in the cultivation of eukaryotic cells at surfaces. The idea that cells could be grown on a surface and then ejected on demand was appealing because cells need harvesting when they are to be used. In particular, growing cells on surfaces means that once a monolayer of cells has formed, no more area can be covered and although cells can divide, they cannot expand under their turgor (internal) pressure. Ejecting cells from the surface gives them the freedom to expand. The use of a conformational switch to eject cells has the additional advantage of removing the requirement for enzymes to detach them from tissue culture plates as is used in standard protocols. This has been demonstrated using a PNIPAm brush attached to a hydrogel of poly(2,3 propanediol-1-methacrylate) [269]. The hydrogel allows nutrients to circulate in the culture medium. Above its lower critical solution temperature (LCST) of 32 °C PNIPAm adopts a collapsed conformation caused by intra-chain hydrogen bonding. It is solvated at lower temperatures, which makes it ideal for cell culture. In this particular case chondrocytes were cultured at 37 °C, but when the temperature was allowed to relax to room temperature, the PNIPAm brush expanded and the cells consequently detached (figure 22). This research was quickly followed by experiments showing that another temperature responsive brush, this time a copolymer of oligo(ethylene glycol) methacrylate and 2-(2-methoxyethoxy)ethyl methacrylate is shown to be resistant to L929 mouse [270] and 3T3 [271] fibroblasts below its LCST, which allowed the harvesting of cells cultured at higher temperatures. A different example comprises a biocidal polymer brush on which bacteria are killed, but which can be converted to a non-fouling surface by a simple hydrolysis procedure [272]. This ejects dead bacteria and leaves a bio-inert surface behind. Such a protocol involves an irreversible chemical change, and so is not fully switchable. Another irreversible route involves converting a PEG brush-coated surface, which inhibits cellular adhesion to an adhesive substrate by the removal of the brush layer, achieved by ultraviolet light which acts on a photo-cleavable group connecting the brush to a glass substrate [273].

Zoom In Zoom Out Reset image size

The ability to create 'smart' surfaces that interact with cells is a considerable advance, and similar work has also been done on switchable adsorption of proteins [274, 275]. Specific interactions of biopolymers and cells with surfaces needs to be treated separately, and this is highlighted in section 4 below.

A topical challenge is to understand the behaviour of complex biopolymers which are integrated into the outer wall of eukaryotic and prokaryotic cells. More generally, the interactions of biomacromolecules with living organisms (a means of testing molecular recognition), is a rapidly growing area of research, which is benefitting from the use of the SMFS-based techniques [289–292]. Various forms of force mapping evolved simultaneously, with experiments performed using unmodified AFM tips [293, 294], chemically modified probes [295] (i.e. probes functionalized with a self-assembled monolayer), and probes functionalized with biopolymers [296]. The use of SMFS has revolutionalized the determination of cellular adhesion forces; previously micro-mechanical techniques were used [297], which were crude and did not allow mapping. An additional benefit of SMFS or SFM is that force-distance curves contain much information about the interaction between the tip and the surface; whilst adhesive and repulsive forces are related to the portion of the force-distance curve that is in solution, the portion of the curve where the tip is in contact with and pressing into the sample surface contains information about the mechanical properties of the sample, and thus, with a well characterized probe [298], the viscoelastic properties of the sample can also be profiled. This technique has been applied to a wide range of samples including lipid bilayers [299], polymers and films [300–304], and whole cells [305–312].

To exemplify the use of AFM tips modified with a biopolymer to explore the location of adhesive moieties on a living cell, work examining a mutant strain of Lactococcus lactis bacteria can be considered [313]. This strain lacks the exopolysaccharides that normally coat the wild type of this Gram-positive bacterium and so SFM images and corresponding force maps were obtained using tips modified with a selectively binding peptidoglycan lysin motif: a recombinant AcmA protein cell wall-binding domain [314]. Using this chemically-modified tip, localized distributions of peptidoglycan molecules (which give bacteria structural integrity and are a substantial component of the mass of lactococci) in lines parallel to the septum of the dividing bacterium were identified (figure 23). The lines were still visible upon rotation of the cell by $90^\circ$, illustrating that the lines are a feature of the cell and not caused by imaging artefacts. Selectivity of adhesion was confirmed by adding unbound peptidoglycan to the imaging media and taking another force map, which revealed few adhesion events. Similar specific force mapping has recently been applied in studies of a wide range of substrates with biological relevance, including the presence of microRNA in neuron cells [315], the orientation and arrangement of P1 adhesin on Streptococcus mutans [316], biofilm-forming properties in genetic mutants of Candida albicans [317], clustering of adhesins on pathogenic bacterium Bordetella pertussis [318], and collagen-binding activity on Staphylococcus epidermidis [319].

Zoom In Zoom Out Reset image size

While force mapping provides a wealth of spatially resolved information about the adhesion between the probe and the sample, it is important to consider that the magnitude and distance from the sample surface at which binding events occur is dependent on the selected experimental conditions (including the applied force, and probe speed and spring constant) [320, 321], so they must be selected with care and caution should be used in any direct comparisons of force spectroscopy data.

Dynamic force spectroscopy [323, 324] offers a a means of exploring molecular binding without spatial resolution, but with an appreciation of the impact that the probe settings have upon the binding energetics of the system: measuring molecular interactions as a function of loading rate [325, 326]. This allows a measurement of the strength of molecular interactions and a determination of binding constants. To exemplify the utility of dynamic force spectroscopy, the interaction of mucins with molecules from a bacterial surface will be used. Mucins are glycoproteins with different biological functions involving the immune system, where they attach to pathogens, or in the lubrication of epithelial cells [327]. Their presence can also be used in cancer diagnosis [328]. The important question that follows is with which molecules on the cell surface are the mucins interacting, and what components in the mucin participate in that interaction. Much is known about what molecules are contained on the cell surface, and it is possible to harvest many of these to coat a surface, or an AFM tip, as in the example presented here. In figure 24 data describing the interaction of a mucin with soybean agglutinin (SBA), a carbohydrate-binding protein (lectin) are presented [322]. In this case, the mucin, which will be referred to herein by the authors' abbreviation of TnPSM (which abbreviates a modified porcine submaxillary mucin, which is secreted from the salivary glands of pigs), was attached to an AFM tip and the SBA attached to a mica surface. As the mucin was pulled away from the SBA, multiple binding events were observed over the length of the mucin chain, which supported an earlier model that proposed such 'binding and jumping' events for this system [329].

Zoom In Zoom Out Reset image size

The AFM can pull the polymer at different loading rates, and the force required to break these bonds (i.e. to unbind) is very much dependent upon the speed of the tip [330], since slower speeds correlate with a smaller loading rate. Slower loading rates allow the polymer to explore a broader energy landscape and so there is more chance of the polymer unbinding. Correspondingly, faster loading rates require a larger force to unbind. Dynamic force spectroscopy allows the calculation of thermodynamic energies of binding, and also the size (in nm) of the potential barrier over which the binding takes place.

Analysis of dynamic force spectroscopy data is typically based upon a modified Bell-Evans model [331–333], with the theory described in detail elsewhere [320, 324, 334–337]. In brief, an unbinding force, F, is applied to the bond, reducing the energy of the potential barrier, E by Fx_b, where x_b is the distance between the potential minimum and the potential barrier, along the direction of the applied force. The lifetime of the bond is dependent on the potential barrier height [333], and therefore the probability of the bond breaking under the applied unbinding force is dependent on the size of the applied force (as the barrier energy effectively becomes smaller when larger forces are applied). In the Bell-Evans model, at a constant loading rate, the probability distribution of the unbinding force, P, is described as

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle P = P_{\rm n} \exp{\left({\frac{(F - F^*)x_{\rm b}}{k_{\rm B}T}}\right)}\exp{\left[{1 - \exp{\left({\frac{(F - F^*)x_{\rm b}}{k_{\rm B}T}}\right)}} \right]}, \label{Bell-Evans} \nonumber \end{align} \tag{ 11 }$$

where F^* is the most frequent rupture force and P_n is a normalization constant. F^* can be obtained from the histogram of measured unbinding forces obtained at a given loading rate, and is linearly dependent on the logarithm of the loading rate, ${\rm d}F/{\rm d}t$ by

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle F^* = \frac{k_{\rm B}T}{x_{\rm b}} \left[ \ln{\frac{{\rm d}F}{{\rm d}t}} + {\rm ln} \left( \frac{\tau_{{\rm b}}x_{\rm b}}{k_{\rm B}T} \right) \right], \label{Bell-Evans2} \nonumber \end{align} \tag{ 12 }$$

where $\tau_{{\rm b}}$ is the lifetime of the molecular bond [324, 335, 337]. Following from equation (12), the width of the potential barrier can be obtained from the gradient of the relationship between F^* and ${\rm ln}\, \dot{F}$, since the slope is proportional to $x_{\rm b}^{-1}$. In the example shown in figure 24, the potential energy barrier is of width 0.1 nm, which is quite small, and the lifetime of the interaction between the mucin and the SBA was determined to be more than 1 s, which is long. It was concluded in this case that the SBA was able to move along the TnPSM some distance before detaching. Clearly therefore, dynamic force spectroscopy reveals quite sophisticated biological information from a relatively simple experiment, and has been used in a wide range of biologially-relevant fields, including work with yeast [338], embryonic zebrafish cells [339], lipid bilayers [340] and proteins [341, 342]. The technique is also finding utility in synthetic polymer systems [22, 343–345] and can be undertaken in parallel with other high sensitivity force-measuring tools such as optical tweezers [346].

Whereas force mapping and dynamic force spectroscopy are employed to understand and characterize molecular binding in detail, the use of the QCM-D to interrogate molecular binding is appropriate in cases where the specifics of bond formation are not required, but where adsorption rates of different molecules are of importance, or where the viscoelastic or conformational properties of molecules forming an adsorbed layer are of interest.

4.3.1. The effect of environmental conditions on adsorbed protein layers.

By probing a population of like molecules, the QCM-D can be applied to intricate scientific questions, such as the effect of pH on the orientation of adsorbed molecules. Fibrinogen has been shown to alter its conformation on hydrophilic but not hydrophobic surfaces as a result of cycling the pH of the buffer solution flowed over the sensor (figure 25) [347]. αC domains are largely globular regions of fibrinogen, of which there are two in each protein [348]. The terminus of the molecule is positively charged at physiological pH (7.4), resulting in a 'folded'-type conformation where they interact with a negatively-charged central region (E domain). On a negatively-charged hydrophilic silica test surface, when adsorbed at pH 7.4, fibrinogen presents an 'αC-hidden' orientation due to a combination of the positive charge of the αC domain at pH 7.4 and its hydrophilic nature relative to the other domains on the fibrinogen. In platelet adhesion experiments on this fibrinogen-coated surface, few platelets were found adhered to the surface, and those that were adsorbed had a non-activated appearance. By adsorbing fibrinogen at pH 7.4 then cycling the buffer pH down to 3.2 and back to 7.4, its conformation was modified, becoming 'αC-exposed' (figure 25). Platelet adhesion experiments on this surface showed increased platelet adherence and activation. This conformation change is due to the charge of the E domain at lower pH: it holds a positive charge at pH 3.2 and therefore electrostatic repulsion occurs between it and the two αC domains. The fibrinogen molecule therefore unfolds, adopting an extended conformation which leads to an increase in the rigidity of the adsorbed layer. When the pH is returned to pH 7.4, the E domain regains a negative charge and attracts the αC domains, re-folding the fibrinogen. However, the side nearest the surface is no longer accessible due to the associated unfavourable entropic cost, and so the αC domains re-bind on top of the molecule, resulting in the αC-exposed conformation.

Zoom In Zoom Out Reset image size

On the hydrophobic surface, the fibrinogen tends to adsorb with the αC domains uppermost due to the substantially more hydrophobic nature of the other domains on the molecule, meaning that it is αC-exposed even at pH 7.4. The combined hydrophobic interactions along the molecule are so strong that the fibrinogen forms a very thin layer (1.3 nm compared to 3.3 nm for the layer adsorbed to the hydrophilic surface), suggesting that it might have become denatured as a result of the strong interactions [347]. This type of experiment illustrates that QCM-D can play a role in the interrogation of biomolecule-surface interactions beyond simple mass adsorption, with the mass, thickness and viscoelasticity of the adsorbed material bringing new insights to the understanding of molecule-surface interactions and the effect of environment on the behaviour of adsorbed molecules.

Similarly, detailed QCM-D analysis can be used to explore the interaction of water with molecules at the test surface, and the influence of molecular architecture upon that interaction, such as in the case of synthetic hydrophilic random and block copolymers on different nanocellulose fibril-coated surfaces, which explored the possibility of water expulsion from the interface as a result of copolymer deposition and the driving forces behind that interaction (in this case, predicted to be of an electrostatic nature) [349]. In some cases this too can be linked to pH responsivity, such as in the case of mycolic acid monolayers [109].

4.3.2. Preventing protein adsorption.

Protein adsorption is significant in large part due to the formation of largely unwanted biofilms which can lead to health issues. These areas include sanitation i.e. biofilm formation in water delivery systems; and medicine and dentistry, such as infection caused by bacterial colony formation on the teeth and post-surgical implant-related complications. Interface engineering can help prevent protein adsorption, either by making it energetically unfavourable or by out-competing the unwanted protein with a less damaging alternative. Water-soluble polymers are the basis for protein-resistant surfaces because they can remain solvated, which makes it more difficult for proteins to adsorb. This field is large and there are already substantial reviews [350–353]. Here, examples are limited to those that exemplify the use of the QCM-D.

The fouling of surfaces is of much more than academic interest, and QCM-D can be used to assess unwanted tissue growth on implants or other medical devices. For example, in the assessment of the use of an aqueous fish protein extract (FPE) to prevent non-specific fibrinogen adhesion to a range of surfaces [354]. Experiments were performed both by preadsorbing FPE to the substrates then adding a test protein (figure 26(b)), and adding FPE after the adsorption of test proteins to check for its ability to out-compete previously attached molecules. Of the four substrates used (gold (Au), stainless steel (SS), silicone dioxide (SiO₂) and polystyrene (PS)), FPE was shown to adsorb most effectively to gold, while polystyrene exhibited the least FPE adsorption (figure 26(a)). These surfaces are different and it is difficult to fully explain the adsorption behaviour. For example, the limited adsorption on polystyrene is likely to be due to a combination of the smoothness of the polystyrene compared to the other surfaces, and its hydrophobicity. Increased surface hydrophobicity means that the adsorbed proteins may have a smaller, more collapsed, conformation; the entropic penalty of water being close to the sample surface results in the proteins being less hydrated. (This depends on the relative balance of interactions between polymer, fluid, and surface.) In contrast, more hydrophilic silicon oxide, gold, and stainless steel surfaces encourage protein swelling, enabling a more elongated polymer conformation and allowing easier diffusion towards the surface.

Zoom In Zoom Out Reset image size

Adsorbed FPE was effective at preventing fibrinogen adsorption (figure 26(b)), but less so human serum albumin (HSA). HSA is comparatively small, which may allow it to exploit any gaps in the adsorbed FPE layer (there is also a lower entropic cost to stretch a small chain towards a surface compared to a long one e.g. HSA compared to fibrinogen) [354]. It was also shown that the orientation of FPE in the adsorbed layer did not affect the repellant properties, irrespective of test protein. This independence of adsorption blocking and molecular arrangement in the layer indicates that steric repulsion is the main contributor to the repulsion and so is independent of substrate. The post-fibrinogen/HSA FPE addition experiments showed that FPE is able to replace pre-adsorbed HSA and fibrinogen, indicating that the interfacial energy is comparatively lower for an FPE coating than for fibrinogen or HSA.

Of the four substrates tested, the three hydrophilic surfaces showed more FPE loss upon rinsing (figure 26(a)), suggesting that the more hydrated proteins are less strongly bound than those on the polystyrene. Consideration of the dissipation confirmed this by suggesting that the protein on the hydrophilic surfaces exhibited more viscoelastic (hydrated) behaviour. The observed differences in levels of FPE on the contrasting surfaces could be at least partly explained by the amount of water close to the surface and involved in protein swelling [354]. Here, however, is an example of the compromises involved in such experiments. The higher energy hydrophilic surfaces are not expected to have many contact points with the proteins, but where they exist, it is quite possible that they are very adhesive. The hydrophobic surfaces are unlikely to have strong contacts, but the number of these contacts ensures that the proteins remain bound to the surface. In the case of these experiments, the hydrophobic surfaces retained more protein than the hydrophilic ones after rinsing, but a true test of how strongly the proteins are bound would require direct measurements using, for example, SMFS. The benefit of this combined approach is becoming more widely appreciated, with SMFS being employed increasingly in partnership with QCM-D measurements. This enables bulk behaviour to be linked to the adhesion strength of individual molecules and individual binding sites along the length of a single molecule [355–357]. An example of this is the interaction of 16 amino acid-long random coil peptides with different inorganic surfaces, where a positive trend was observed between the total mass uptake as measured with QCM-D and the single molecule adhesion force on the same test substrate [358].

The creation of protein-resistant surfaces is directed towards retarding adsorption rather than stopping it; ultimately resistance is futile and surfaces will be coated. QCM-D remains an attractive technique for the study of the growth of biofilms in order to help address the mechanism of formation. The formation of biofilms is ubiquitous but different processes occur as the bacteria attach to the surface from an initial deposition to a final virtually immovable film. The early stages of this process are notoriously complex and can be followed in real time using QCM-D [359]. Considerable effort is devoted to the preparation of hydrophilic protein-resistant surfaces [351, 360–362], and it is also possible to convey protein resistance to hydrophobic polymers, by decorating them with hydrophilic side chains [363, 364].