Effect of Molecular Architecture on Single Polymer Adhesion

Article pubs.acs.org/Langmuir

Effect of Molecular Architecture on Single Polymer Adhesion Sandra Kienle,† Markus Gallei,‡ Hao Yu,§ Baozhong Zhang,§ Stefanie Krysiak,† Bizan N. Balzer,† Matthias Rehahn,‡ A. Dieter Schlüter,§ and Thorsten Hugel*,† †

Physik Department E22 and IMETUM, Technische Universität München, 85748 Garching, Germany Ernst-Berl Institut für Makromolekulare Chemie, TU Darmstadt, Alarich-Weiss-Str. 4, 64287 Darmstadt, Germany § Department of Materials, ETH Zurich, Vladimir Prelog Weg 5, HCI J 541, Zurich, Switzerland ‡

S Supporting Information *

ABSTRACT: Several applications require strong noncovalent adhesion of polymers to substrates. Graft and branched polymers have proven superior to linear polymers, but the molecular mechanism is still unclear. Here, this question is addressed on the single molecule level with an atomic force microscopy (AFM) based method. It is determined how the presence of side chains and their molecular architecture influence the adhesion and the mobility of polymers on solid substrates. Surprisingly, the adhesion of mobile polymers cannot significantly be improved by side chains or their architecture. Only for immobile polymers a significantly higher maximum rupture force for graft, bottle-brush, and branched polymers compared to linear chains is measured. Our results suggest that a combination of polymer architecture and strong molecular bonds is necessary to increase the polymer−surface contact area. An increased contact area together with intrachain cohesion (e.g., by entanglements) leads to improved polymer adhesion. These findings may prove useful for the design of stable polymer coatings.

1. INTRODUCTION

The interaction and mobility of linear chains on solid substrates have been extensively studied on the single molecule level.10−13 Examples are hydrophilic and hydrophobic polymers like poly(L-lysine) and poly(D-tyrosine)14 but also spider silk motifs15 and polystyrene,16 which were probed on solid surfaces with different hydrophobicities15 and on polymer films.16 The effects of salt concentration, pH,15 and temperature14 were analyzed. It is noteworthy that most of the effects observed for the adhesion strength were smallusually less than a factor of 2. Here we show that force−distance curves with mobile (in the surface plane) polymers result in plateaus of constant force (Figure 1B), independent of polymer architecture. For polymers that are rather immobile on the surface, but still unspecific bound, mainly rupture events were observed (Figure 1C). In this case the maximal rupture force significantly depends on polymer architecture. Please note that mobile here is defined from the view of a desorption experiment; this might be sliding or desorption stick in the view of lateral pulling.13 The two main contributions to the desorption force are the solvation free energy (hydrophobic effect) and the surface interaction free energy (dispersion), which largely compensate each other. For more details see ref 17.

Biofouling is the (unfavorable) process of biological material adhering to different surfaces. It has a huge impact, for example, not only on the shipping industry1 but also on the design of biomedical implants.2 To minimize these unwanted interactions, polymer coatings with antifouling properties were and are still developed.3 Noncovalent polymer−surface interactions could be advantageous because of easier fabrication and better long-term stability. The latter is possible because noncovalent bonds can constantly re-form, while covalent bonds will basically not re-form once broken. As a single noncovalent bond is weaker than a covalent bond, the number of contacts has to be increased for noncovalent interactions. One approach toward that is to use not linear but graft or brush polymers. The antifouling properties of PEG could for example be combined with poly(L-lysine), which enables physisorption of the graft polymer.4 Another possibility is to use dendritic architectures like dendrimers5−7 or dendronized polymers.8,9 Because of the branched nature of their side chains, the functional end groups are denser packed compared to those of polymers with linear side chains. This means that more possible interaction sites with the surface are present, which could lead to a stronger adsorption. To our knowledge, there is no systematic study on how polymer architecture impacts noncovalent binding of polymers onto solid substrates. Here we compare the adhesion properties and the mobility of various linear, graft, bottle-brush, and branched polymers on a solid substrate by AFM-based single molecule force spectroscopy (Figure 1A). © 2014 American Chemical Society

Received: March 3, 2014 Revised: March 27, 2014 Published: March 29, 2014 4351

dx.doi.org/10.1021/la500783n | Langmuir 2014, 30, 4351−4357

Langmuir

Article

Figure 1. Scheme of an AFM tip with a covalently attached single, linear polymer, and schematics of the different polymer architectures used (A). Schematics for a force−distance curve for mobile polymers (plateaus of constant force) (B) and for immobile polymers (rupture events) (C). calibrated with the thermal noise method.19 In short, the frequencydependent motion of the cantilever was recorded and the resonance peak was fitted. Together with the inverse optical lever sensitivity (InvOLS), which was determined from the contact regime of 10 force−distance curves, the normal spring constant could be determined. The spring constants ranged from 10 to 40 pN/nm. The error for spring constant determination is around 15%. All force− distance curves were performed in a similar way. The surface was approached with a constant velocity (0.1−10 μm/s). After a waiting time (0.01−3 s) at the surface, the cantilever was retracted with the same constant velocity. This was repeated for at least 100 times at different surface positions for each experiment. If the force−distance curve showed a plateau of constant force, the height of the plateau (plateau force) and the length (detachment length) were fitted by a sigmoidal function. If there was no plateau but one or more rupture events, the force of the highest rupture peak was determined. The first peak, which reflects the unspecific interaction with the surface, was (if present) neglected. We chose to focus on the maximal rupture force and not the average rupture force because rupture peaks are not always clearly defined, while the maximal rupture force is unique. Afterward, the plateau force, the detachment length, or the maximal rupture force was plotted in a histogram. Please note that for the nonequilibrium rupture events force cannot directly be related to free energy, but as similar experimental conditions were used throughout the experiments, a relative comparison based on rupture forces is possible.

2. EXPERIMENTAL SECTION For all measurements we used the atomic force microscope (AFM) MFP-3D SA (Asylum Research, an Oxford Instruments Company, Santa Barbara, CA) with a temperature-controlled fluid cell. Different polymers were covalently coupled to the tip of an AFM cantilever (MLCT, Bruker AFM probes, Camarillo, CA). The polymers used include polyisoprene (PI), polyisoprene grafted with polystyrene side chains of different molecular weight (PI-g-PS, synthesis is described in the Supporting Information), aggrecan (from bovine articular cartilage, Sigma), chondroitin sulfate (from shark cartilage, Sigma), and a dendronized polymer (synthesis is described in the Supporting Information). The first step of the coupling process was the activation of the cantilever in a plasma chamber filled with oxygen (Diener, Germany). Afterward, the cantilevers were rinsed with acetone (anhydrous, ≥99.8%, VWR) and incubated for 10 min in Vectabond (50 μL was dissolved in 2.5 mL of acetone) to obtain a covering of the tip with amino groups. The next step of the protocol depended on the solubility and the functional group of the polymer used. For polymers with a thiol end group (PI and PI-g-PS) the cantilevers were again rinsed in acetone and chloroform (anhydrous, ≥99%, Sigma-Aldrich) before they were incubated for 45 min in a 1:1500 mixture of CH3OPEG-NHS (5 kDa, Rapp Polymere GmbH, Tübingen, Germany) and α-maleinimidohexanoic-ω-NHS-PEG (5 kDa, Rapp Polymere GmbH, Tübingen, Germany). Afterward, the cantilevers were rinsed in chloroform and incubated for at least 1 h in the polymer solution (about 2 mg/mL dissolved in chloroform) before they were rinsed again in chloroform and stored in chloroform at room temperature until the measurements were performed. Polymers with an amino group (aggrecan) were coupled with a similar protocol. The difference was that instead of α-maleinimidohexanoic-ω-NHS-PEG, we used PEG with two NHS esters (PEG-α-ω-DiNHS, 6 kDa, Rapp Polymere GmbH, Tübingen, Germany), and we rinsed not only in chloroform after the incubation in the PEG mixture but also in ethanol (absolute, >99.9%, Merck) and borate buffer (pH 8−8.5). The borate buffer was also used to dissolve the polymer and to store the cantilever chips until the measurement. For polymers with a carboxyl group as a functional group (chondroitin sulfate and dendronized polymers, synthesis is described in the Supporting Information) the coupling protocol had to be adjusted in a different way. After the activation of the cantilever tip and the incubation in Vectabond, the cantilever chips were rinsed in acetone and either phosphate buffered saline (PBS, Sigma, chondroitin sulfate) or methanol (anhydrous, 99.8%, Sigma-Aldrich, dendronized polymers) and directly incubated in the corresponding polymer solution. The polymers were either dissolved in PBS (chondroitin sulfate) or in methanol (dendronized polymers) with a concentration of 0.5−2 mg/mL. Included in the polymer solution was also 50 μM 1ethyl-3-[3-(dimethylamino)propyl]carbodiimide hydrochloride (EDC, Thermo Scientific, Rockford, IL) and 150 μM N-hydroxysulfosuccinimide (Sulfo-NHS, Thermo Scientific, Rockford, IL) to allow a coupling of the carboxyl groups to the amino groups on the cantilever tip. After an incubation time of about 1 h, the cantilever chips were rinsed and stored in PBS or methanol. All measurements were performed in solution on a hydrogenterminated diamond. The termination was conducted as described before.14,18 The normal spring constant of each cantilever was

3. RESULTS AND DISCUSSION To investigate the effect of side chains and their architecture on the adhesion force, we used single molecule force spectroscopy. We coupled polymers without and with side chains of different architecture covalently attached to our AFM tip (see Experimental Section) and measured the force between a single polymer and a solid substrate in solution. The results shown were measured on a hydrophobic, hydrogenated diamond surface. We also measured on other surfaces like silicon oxide or PTFE, but diamond is the best defined surface and shows all motifs we have observed on any surface. Additionally, the interactions with the hydrogenated diamond are only unspecific as none of the functional groups present on the polymers can interact specifically with this substrate (e.g., by forming hydrogen bonds, complex bonds, or covalent bonds). Please note that in this study single polymers are covalently coupled to the AFM tip and desorbed from plain, uncoated, and therefore well-defined substrates. 3.1. PI-Based Graft Polymers. In a first set of experiments, we compared a simple polyisoprene backbone (PI, 119 kDa) to polyisoprene PI which has been grafted with polystyrene PS side chains of different molecular weight, namely PI-g-PS(3) with 3 kDa, PI-g-PS(14) with 14 kDa, and PI-g-PS(88) with 88 kDa polystyrene side chains (synthesis is described in the Supporting Information). Well-defined PI-g-PS chains featuring a thiol end group were obtained by combining living anionic polymerization and hydrosilylation protocols. These polymers 4352

dx.doi.org/10.1021/la500783n | Langmuir 2014, 30, 4351−4357

Langmuir

Article

Figure 2. Typical force−distance curves for polyisoprene PI (A, B) and for polyisoprene with 88 kDa polystyrene side chains (PI-g-PS(88)) (C, D) in water on hydrogen-terminated diamond.

were feasible for AFM-tip functionalization. The entanglement length of polystyrene is on the order of 18 kDa.20 Therefore, we had side chains that should not entangle with one another (3 kDa), side chains that are in the region of the entanglement length (14 kDa), and side chains where entanglements are likely to occur (88 kDa). Because of the gain in molecular weight, which was measured after addition of the side chains, the polyisoprene backbone should be on average grafted with 270 side chains of 3 kDa or 60 side chains of 14 kDa or 10 side chains of 88 kDa molecular weight (due to the low grafting density to the flexible backbone the entanglement length should not change significantly). The measurements were conducted in water. Water is a nonsolvent for polyisoprene and polystyrene and therefore enhances the adhesion to the substrate. Additionally, we performed measurements in a good solvent (chloroform) with similar general behavior. To determine the adhesion force for the four different polymers, we performed force−distance curves at room temperature with a constant velocity of 1 μm/s. First, the surface was approached until the cantilever tip was in contact with the surface. After a waiting time (typically 1 s), to allow (physical) polymer adsorption, the cantilever was retracted and the polymer was desorbed. Two different motifs were observed: first, a plateau of constant force and second rupture events. As shown before, plateaus of constant force are visible if the polymer is mobile on the surface or could at least be desorbed in the z-direction and rupture events if the polymer is rather immobile. If a constant force plateau was observed, then we used a sigmoidal fit to determine the height of the plateau (plateau force) and the length where the polymer detaches (detachment length) (black curves in Figure 2A,C). In the case of rupture events, the maximal rupture force was determined. If present, the first peak in force−distance curve, which corresponds to the unspecific adhesion with the surface, was neglected for the evaluation. Exemplary force−distance curves for PI and PI-gPS(88) are shown in Figures 2B and 2D. Multiple peaks correspond to the rupture of multiple interaction points. This could result not only from the interaction of a single polymer with several side chains with the surface, but also from the

simultaneous interaction of several polymers with the surface. Multiple peaks are more often observed than single peaks. Most of the curves showed plateaus of constant force (Figure 2A,C and Table 1). The same holds true for measurements Table 1. Occurrence of Plateaus or Rupture Events in Force−Distance Curves with Polyisoprene-Based Polymers Together with the Total Number of Force−Distance Curves PI PI-g-PS(3) PI-g-PS(14) PI-g-PS(88)

plateau events (%)

rupture events (%)

no. of curves

79 88 87 66

21 12 13 34

789 1528 1676 1803

with PI-g-PS(3) and PI-g-PS(14) (Figure S1 and Table 1). In Figure 2A, a single plateau of constant force with only one step is shown. This single plateau reflects the desorption of a single PI chain. Multiple steps of plateaus of constant force like those shown in Figure 2C have been assigned to the simultaneous desorption of several polymer chains.21 In the case shown here, two polymers with different lengths were desorbed simultaneously until the first, shorter polymer detached from the surface (first step in the force−distance curve at around 0.35 μm). The second, longer polymer was still in contact with the surface. This resulted in an ongoing plateau of constant force but now with a lower force. After detachment of this longer polymer (around 0.65 μm) the force dropped to zero. This means that in the case of multiple steps the last step reflected the detachment of a single polymer from a plain substrate. For polymer chains with grafted side chains we expected in analogy to the simultaneous desorption of several individual chains an accumulation of this stepwise desorption process because the backbone as well as side chains should interact simultaneously with the surface, therefore increasing the probability to desorb several chains at the same time and to observe more than one step. Surprisingly, our measurements did not support this expectation. We usually observed plateaus of constant force for PI-g-PS with a single step (similar to the curve in Figure 2A for PI). 4353

dx.doi.org/10.1021/la500783n | Langmuir 2014, 30, 4351−4357

Langmuir

Article

Even the plateau forces (Figure 3A) and the detachment lengths (Figure S2) for PI and PI-g-PS did not show significant

differences between graft polymers and linear chains. The variations in force in between measurements with the same sort of polymer are comparable to the differences which were measured with different polymers. So the presence of side chains was not visible in the characteristics of plateaus of constant force. We also confirmed that for the dependence on the dwell time at the surface (Figure S3A) the velocity dependence (Figure S3B) and the influence of the temperature on the plateau force (Figure S3C). Neither the dwell time nor the velocity or the temperature affected the plateau force of a linear polymer chain or a graft polymer. For a linear polymer, this has been shown before.14 Finally, AFM images of graft polymers show molten globule-like structures (see Figure S4). Altogether our results suggest that the adhesion is not significantly changed until many polymers interact not only with the surface but also with one another. Such cohesion might then result in a stronger adhesion to the substrates. This might explain why cross-linking helps improve polymer films (by increasing cohesion).22 Another possibility is that for PI-gPS(3) and PI-g-PS(14) the side chains are folded on the backbone, forming a cylinder with similar contact area compared to the linear chain. Not all force−distance curves showed plateaus of constant force. In 12−34% (dependent on the side chain weight) rupture events or a mixture of plateau and rupture events in one force−distance curve were measured (Figure 2B,D and Table 1). The maximal rupture force (Figure 3B) increased slightly with increasing molecular weight of the side chains. The same could be observed for the occurrence of rupture events (Table 1). The difference was especially visible for PI-g-PS(88). A possible explanation for this is the entanglement length. 88 kDa is much larger than the entanglement length of 18 kDa. The entanglement of side chains results in a polymer bundle with stronger intrachain interactions, which is not as mobile on the surface as a polymer bundle without entanglements. This reduced mobility could lead to a stronger interaction with the surface (higher force). Intrachain interactions could also lead to a deviation from the cylindrical shape and therefore an

Figure 3. Plateau force (A) and maximal rupture force (B) obtained from measurements with polyisoprene PI without and with 3 kDa (PIg-PS(3)), 14 kDa (PI-g-PS(14)), and 88 kDa (PI-g-PS(88)) polystyrene side chains. Each color represents the result of a different experiment with a different cantilever. Altogether 1243 rupture events and 2552 plateaus were evaluated.

Figure 4. Force−distance curves for rupture events in PBS on a hydrogen-terminated diamond with chondroitin sulfate (A) and with aggrecan (B). The maximal rupture force for each curve was determined for chondroitin (black) and aggrecan (blue) and plotted as a histogram in dependence of the velocity (C). We evaluated 4201 force−distance curves measured with chondroitin sulfate and 1927 curves for aggrecan. 4354

dx.doi.org/10.1021/la500783n | Langmuir 2014, 30, 4351−4357

Langmuir

Article

Figure 5. Two exemplary force−distance curves obtained with a dendronized polymer on a hydrogenated diamond in water. In 88% of the cases rupture events were observed (A) and in 12% plateaus of constant force (B). Histograms of the maximal rupture force (C) and of the plateau force (D). Each color represents a measurement with a different cantilever. The total number of curves was 4043 and in 69% of the curves events were observed. For all force−distance curves, a velocity of 1 μm/s was used.

curves showed interactions with the surface. Because of the fact that aggrecan consists of about 100−150 chondroitin sulfate side chains, we even expected a much higher rupture force not only around a factor of 2. One possible explanation could be that not all bonds with the surface are loaded at the same time to the exact same extent. This could for example lead to a logarithmic dependence of the rupture force f on the number of parallel bonds N (f ∼ ln(N)).28 If we translate this to our experiment, where the force between chondroitin sulfate and aggrecan differs by a factor of 1.5−2, we obtained about 8 times more bonds that were ruptured in parallel when measuring with aggrecan compared to chondroitin sulfate. This seems reasonable because during the desorption of aggrecan not all side chains will desorb simultaneously. In summary, measuring with an immobile bottle-brush polypeptide resulted in significantly more interaction with the surface and in a higher maximal rupture force than measurements with a linear sugar chain. 3.3. Dendronized Polymer. The next step was to investigate the adhesion behavior of a dendronized polymer (synthesis is described in the Supporting Information). A dendronized polymer is a linear polymer with highly branched side chains and consists in our case of about 500 individual dendrons.8 Dendrons are wedge-shaped with a single starting point and several functional end groups. The great advantage of this geometry is the high density of functional end groups compared to linear side chains. In our case, the dendrons were of generation four with 16 carboxyl end groups each. This means that we had a linear, almost cylindrically shaped polymer with around 8000 end groups. Direct AFM imaging of the negatively charged dendronized polymers on uncoated substrates is nontrivial, but neutral and positively charged dendronized polymers have been imaged by AFM previously.29 Measuring the adhesion to a solid substrate resulted in 12% of the force−distance curves in constant force plateaus (Figure 5B). This means that this polymer with highly branched side chains can be mobile on the surface. Surprisingly, the height of the force plateaus was in the same range (∼60 pN) as for polyisoprene-based polymers (Figure 5D). So when it comes to plateaus of constant force (i.e., high mobility and unspecific

increased contact area. Both scenarios would explain the higher occurrence of rupture events with PI-g-PS(88). In summary, we can say that side chains had no significant effect on plateaus of constant force. Only measurements with PI-g-PS(88), which is considerably above the entanglement limit, resulted in an increase in the occurrence of rupture events and also in an increase in the maximal rupture force. This can be explained with the formation of stronger interactions in a polymer bundle caused by entanglements and consequently a lower mobility on the surface. Please note again that we investigate single polymers on plain substrates. 3.2. Aggrecan and Chondroitin Sulfate. To further investigate the effect of side chains on the adhesion behavior, we chose a naturally occurring linear sugar chain (chondroitin sulfate) and compared it to aggrecan a bottle-brush polypeptide which can be found in cartilage (see schemes in Figure 4).23 Chondroitin sulfate is a negatively charged glycosaminoglycan chain with alternating sugars. Aggrecan consists of an extended protein core with mostly chondroitin sulfate chains as side chains.24 Each aggrecan molecule is composed of up to 150 side chains with a length of about 40 nm and a dense spacing of 2−3 nm. The architecture of aggrecan can nicely be seen in AFM images made by Ng et al.25 Again, we performed force− distance curves. Typical examples for a force−distance curve with chondroitin sulfate and one with aggrecan are shown in Figure 4A,B. For both polypeptides rupture events appeared prevalently in the force−distance curves. In contrast to PI and PI-g-PS, almost no plateaus of constant force were observed (