Control of Surface Morphology, Adhesion and Friction of Colloidal Gels with Lamellar Surface Interactions

Despite recent advances in polyelectrolyte systems, designing responsive hydrogel interfaces to meet application requirements still proves challenging. Here, semicrystalline colloidal gels composed of poly(methacrylamide‐co‐methacrylic acid) are investigated in water with storage moduli in the MPa range. A combination of SEM, X‐ray scattering, and NMR reveals the evolution of the colloidal microstructure, crystallinity, and hydrogen bonding with varying monomer ratio. The gels with the finest colloidal microstructure exhibit the most dissipative rheological behavior and are selected for the study of their interfacial characteristics and underlying interactions. Microstructure stabilization and dynamics results from short‐range (attractive) hydrogen bonding and hydrophobic forces, and long‐range (repulsive) electrostatic interactions—the “SALR” pair potential. Further, the gel's surface exhibits a submicron colloidal topography that greatly determines (colloidal‐like) friction as a result of the viscoelastic deformation of the colloidal network, while electrostatic near‐surface interactions propagate in lamellar adhesion. The dynamic and reversible nature of the involved interactions introduces a stimulus responsive behavior that enables the electrotunability of adhesion and friction. This study advances the knowledge necessary to design complex hydrogel interfaces that enable spatial and dynamic control of surface properties, which is of relevance for applications in biomedical devices, soft tissue design, soft robotics, and other engineered tribosystems.


Introduction
The advances in biomimetic robots, [1] biomedical implants and devices, [2] and human-machine interfaces, [3] is triggering the discovery of new soft materials. The ideal soft interface should be biocompatible, strong, and wear-resistant, and enable the tions are reversible and dynamic, allowing responsive behaviors to pH, temperature, or ionic environmentson short timescales. The stimulus-responsive behavior of these gels has also been demonstrated in drug delivery. [6] While many methods have recently been investigated for creating physically-crosslinked gels, the number of systems available is still limited. [5,7] An alternative, yet less explored, approach consists of colloidal gels, which are composed of a disordered 3D network of interconnected colloidal particles displaying a rich morphological behavior that depends on the underlying type of interparticle interactions and their strength. [8] Colloidal systems with attractive inter-particle interactions are naturally out of equilibrium as the thermodynamic state of the system consists of the macroscopic phase separation between a particulate-rich and a particulate-poor phase. [9] When introducing an additional long-range repulsion, the phase separation can be suppressed. [10] The attraction drives particle aggregation to form a heterogeneous colloidal distribution that forms fractal clusters that span throughout the gel. With the additional long-range repulsion, the growth of the heterogeneous clusters can be hindered and results in a relatively uniform particle distribution at larger length scales. The colloidal network is thus held together in a kinetically arrested state by means of short-range attractive and long-range repulsive interactions, which is referred as the SALR pair potential. [11] Resulting structures thus range from dilute fractal-like percolating networks, to dense cluster-based gels and more homogeneous attractive glasses at higher particle volume fractions. [12] It is often assumed that gels form if the water content is larger than 42 wt.%, and glasses form at smaller water contents. [13] Colloidal polymer-based gels can possess excellent mechanical properties. For example, poly(methacrylamide-co-methacrylic acid) hydrogels exhibit a Young's modulus up to 200 MPa in the glassy state. [14] In a separate work, colloidal gels with anisotropic polymer colloidal particles were prepared with particles of different aspect ratios and rigidity. This work showed that, while the aspect ratio influences the sol-gel transition, the rigidity of the colloids influences the mechanical properties, and higher moduli were achieved with more rigid colloids. [15] The physical nature of the interparticle interactions in colloidal gels is not only fundamentally interesting but also promising in the search of stimulus-responsive strong gel interfaces. Here, we have investigated the interfacial properties of a colloidal gel based on poly(methacrylamide-co-methacrylic acid). Studies of the bulk properties of these gels (proton environment, crystallinity, microstructure, and rheology) enabled us to identify a strong gel stabilized by dynamic and reversible interactions for studies of its interfacial behavior. To the best of the authors' knowledge, this is the first time that in situ interfacial properties of colloidal gels, including surface topography, stiffness, adhesion, and friction, and the involved interactions, have been investigated. We also demonstrate their stimulus-responsive behavior in proof-of-concept experiments, in which an electric potential is applied to modulate adhesion and friction.

Results
The colloidal gel is formed through simultaneous free radical copolymerization of methacrylic acid (mAAc) and methacrylamide (mAAm) ( Figure 1A) using a synthesis protocol inspired by Zhang et al. [14a] Full experimental details are given in the Experimental Section. The synthesis pH (≈2.0) is well below the pKa of methacrylic acid (≈4.6), and hence, very few carboxylic groups are expected to be dissociated during synthesis, which suggests that hydrophobic interactions may play a role during gelation. [16] The gels are then equilibrated in deionized (DI) water (pH 5.5) and heat treated for 20 h at 45 °C to increase homogeneity and reproducibility across measurements and reduce ageing effects during characterization. Mole ratios of 0.60-0.95 methacrylic acid to total monomer (60-95% mAAc) were selected in this work. As-prepared gel samples have a thickness of ≈2 mm. Optical images of the gels are also shown in Figure 1A, displaying a gradient in opacity, with the highest opacity for the smallest mole percentage of methacrylic acid in the synthesis solution, i.e., for 60% mAAc. This is consistent with the opacity reported by Zhang et al., [14a] but the modified protocol leads to different microstructures. The water content of the synthesized gels is 57, 56, 51, and 62 wt.% for gels with 60%, 70%, 80%, and 95% mAAc, respectively. Unless otherwise stated, characterization of the gels occurred one day after synthesis.
Attenuated Total Reflectance Fourier transform Infrared (ATR-FTIR) spectroscopy was used to confirm the successful synthesis of the gels ( Figure 1B). FTIR peaks at 1180 and 1260 cm −1 correspond to the methyl backbone and are used as reference peaks. Peaks at 1585 and 1655 cm −1 are associated with symmetric and asymmetric amide bands, respectively, and the peak at 1690 cm −1 is attributed to the methacrylic acid carboxyl group. The calibration of the carboxyl to amide peak intensity ratio (1655 cm −1 /1690 cm −1 ) was used to determine the amount of mAAc incorporated in the gel network; see Experimental Section. Figure 1C shows the negative deviation of the mAAc% in the gel relative to the monomer solution. This suggests that there is unreacted mAAc that is not incorporated into the gel network. Although speculative at this point, this is likely due to methacrylamide having a higher reactivity compared to methacrylic acid at the synthesis conditions. The synthesis route may introduce structural gradients due to the polymerization and sedimentation. Hence, throughout the work only the surface gelated against the polycarbonate mold (denoted as "top" surface) is considered for consistency.

Hydrogen Bonding Between Methacrylic Acid and Methacrylamide
Solid State 1D proton Nuclear Magnetic Resonance (HNMR) spectra shown in Figure 1D displays how the proton environment changes with mAAc% using pure DI water as a reference peak. Gels with 60-80% mAAc show multiple distinct proton environments that are associated with different types of complexation between methacrylamide and methacrylic acid in an aqueous environment. [17] This is seen primarily as the decrease in relative intensity of the free carboxyl group peak at 6.1 ppm and simultaneous appearance of the peak at 5.9 ppm, which supports the formation of hydrogen bonds. Methylene multiplets are detected ≈4.5 ppm but are too weak to be seen at the selected intensity scale. For gels with 95% mAAc, in contrast, distinct environments are shown for the amide, carboxyl, and water protons but there is little to no presence of peaks associated with complexation and hydrogen bonding. Hence, these measurements support the presence of hydrogen bonding and complexation between methacrylamide and methacrylic acid only in gels with 60-80% mAAc. The presence of hydrogen bonding was also concluded by Zhang et al. for other poly(methacrylamide-co-methacrylic acid) hydrogels by adding urea to destroy the hydrogen bonds. [14a] It is reasonable to expect a maximum hydrogen bonding at 50% mAAc. However, Figure 1D shows that decreasing the mAAc content from 80% to 70 and 60% leads to a less prominent peak at 5.9 ppm, which implies that the hydrogen bonding between carboxylic and amide groups is reduced. The fact that an excess of carboxylic groups is needed to maximize hydrogen bonding suggests that not all carboxylic groups participate in this interaction, due to either incomplete dissociation or to the higher reactivity ratio of methylacrylamide compared to methyl acrylic acid at the synthesis conditions. The latter is also supported by the calibrated FTIR results, as discussed earlier ( Figure 1C). www.afm-journal.de www.advancedsciencenews.com X-Ray Diffraction (XRD) was used to examine the shortrange order of the gels. The contribution of water was identified at peaks ≈30 and 40° in reference measurements. Hence, the diffractogram of water was subtracted (see Experimental Section), such that the remaining diffraction was due to the polymer network only. Figure 1E shows 2θ scattering peaks at ≈16°, 30°, and 41° for gels with 60-80% mAAc, corresponding with d-spacings of 5.5, 3, and 2.2 Å, respectively. This indicates the presence of crystalline domains in the gels. The largest difference among the four gels is the absence of the lowest 2θ peak in the diffractogram of 95% mAAc gels. Additionally, the two peaks present in this gel are shifted to lower 2θ, indicating that the crystal lattice has larger spacings compared to the gels with lower mAAc. Considering the NMR analysis, these results suggests that hydrogen bonding could be an important contributor to the gel (semi)crystallinity. [18]

Refinement of the Colloidal Microstructure with Increase in mAAc
Scanning Electron Microscopy (SEM) was employed to visualize the gel microstructure. The surface of the freeze-dried gels showed some signs of network collapse, and hence, only the cross-sections are discussed here. SEM images are shown in Figure 2 and their particle analysis is displayed in Figure S1 (Supporting Information). The gels with 60% mAAc display clusters of aggregated microsized particles, with an average radius of 2.0 ± 0.5 µm ( Figure 1A). As the mAAc content increases to 70%, a heterogeneous microstructure becomes evident, i.e., there are domains with a fused, close-packed particulate network ( Figure 2B) and domains composed of a finer network with voids 360 ± 140 nm in diameter ( Figure 2C). The further increase of mAAc to 80% leads to homogeneous gels with only a fine network-type microstructure ( Figure 2D), Figure 1. A) Reaction scheme and optical images of the four compositions investigated in this work; 60, 70, 80 and 95% refers to the mol% of mAAc with respect to total monomer content. Grid size corresponds to 1 cm. The water content is given for each gel in each image (black background). B) Representative ATR-FTIR spectra and C) mol% of mAAc in the gel network obtained from calibrated FTIR spectra; details are given in the Experimental Section. D) Solid State NMR spectra with labeled peak assignments showing the existence of hydrogen bonding complexation in gels with mAAc% between 60 and 80%. Methylene peaks are too weak to be seen at this intensity scale. E) X-ray diffractograms of the gels after subtraction of the water background. All gels were characterized one day after synthesis.
www.afm-journal.de www.advancedsciencenews.com but with smaller void diameters than gels with 70% mAAc (170±50 nm vs 360±140 nm). A closer look (see inset), as well as the AFM images discussed later, supports that the fine network is composed of small (submicron) colloidal particles. Figure 2E illustrates the described microstructures. Gels with 95% mAAc, in contrast, undergo a prominent collapse upon drying, so that their microstructure is not well represented in the SEM images, and they are not shown here. The microstructure inferred from SEM images is consistent with the fractal dimensions of the gels that were obtained from Small-Angle X-Ray Scattering (SAXS) measurements; see Figure S2 (Supporting Information) and description in Supporting Infomation.

Dynamic Nature of Interparticle Interactions in Colloidal
Gels with 80% mAAc Figure 3 summarizes the results of rheological experiments. Frequency sweeps in the linear regime show that gels with 60%, 70%, and 80% mAAc achieve storage moduli on the order of ≈1 MPa at 25 °C, while the 95% mAAc gels exhibit moduli two orders of magnitude smaller ( Figure 3A). The magnitude of tan(δ) is much greater for gels with 70% and 80% mAAc than 60% and 95% mAAc gels and strongly frequency-dependent, indicating that 70% and 80% mAAc gels are more dissipative (viscous) and the energy dissipation depends on the timescale ( Figure 3B). This points toward increased rearrangements in this gel's network, presumably resulting from the dynamic nature of the physical interparticle cross-links. [19] 95% mAAc gels exhibit a linear elastic behavior up to strain amplitudes close to 30%, which is consistent with the lack of phase boundaries. In contrast, the gels with smaller mAAc% behave nonlinear elastic already at very low strains (≈0.1%). The storage modulus at 10 rad s −1 follows a slightly increasing trend from 60% to 80% mAAc, followed by an abrupt drop at 95% mAAc ( Figure 3D). This coincides with the change of the microstructure from aggregated clusters (at 60% mAAc) to a network of submicron particles (at 80% mAAc), to a non-colloidal (transparent) gel with 95% mAAc.
Dynamic oscillatory rheology was used to investigate the temperature-dependent response of 80% mAAc gels. Each temperature was equilibrated for 45 min. Drying of the gel samples was avoided by continuously adding water throughout the duration of the measurements. Figure 3E shows the strong decrease in storage modulus with increasing temperature from 25 to 45 °C. This is concurrent with the significant increase in tan(δ), which indicates that the system may be evolving toward a transition; see Figure 3F. To understand better this behavior, van-Gurp Palmen plots were examined ( Figure S3, Supporting Information). Although this plot indicates that the time-temperature-superposition is not fully valid, several characteristics can be inferred from this analysis. At 25 °C, there is a subtle minimum of δ ≈ 13° and |G*| ≈3 MPa, in the glassy state. At higher temperatures, this minimum is not resolved, but a maximum becomes evident in the temperature range 40-45 °C (δ ≈ 40-45°), presumably related to a glass transition. [14b] At even higher temperatures, δ achieves a minimum at ≈13°, with |G*| ≈ 0.03 MPa, presumably the rubbery complex modulus. However, it is difficult to determine this value precisely due to the lack of superposition at high temperature. It is to be noted that drying of the gels did not happen during these experiments, and hence, this is not the origin for the lack of superposition. Nevertheless, microstructural changes are possible since colloidal gels may be trapped in a kinetic state, and hence, a temperature increase , and B) tan(δ) from frequency sweeps at a strain of 0.05%. C) G′ and G″ from amplitude sweeps at 10 rad/s measured by parallel plate oscillatory rheology for poly(methacrylic acid-co-methacrylamide) gels with 60% (green diamonds), 70 (blue circles), 80 (red squares) and 95% (purple triangles) mAAc at 25°C. Storage and loss moduli at 10 rad/s and 0.05% strain are summarized in D). E) Storage (G′), loss moduli (G″) and F) tan(δ) for gels with 80% mAAc from frequency sweeps at 25, 30, 35, 40, and 45 °C (increasing temperature).
www.afm-journal.de www.advancedsciencenews.com could bring the system closer to equilibrium. It is to be noted that G′ and G″ exhibit hysteresis upon increase and decrease of temperature, which supports the occurrence of structural changes that are not fully reversible during the duration of this experiment. These studies are still ongoing.
A potential explanation for the temperature-dependent behavior is based on the dynamic nature of the interparticle forces, as enthalpy-dominated (physical) crosslinking interactions become more dynamic at elevated temperatures, leading to network softening. [20] That is, assuming an energy barrier is associated with particle dynamics, increasing the temperature of the colloidal gels a) decreases the lifetime of the interparticle bonds (e.g., hydrogen bonds) and thereby the storage modulus and b) increases the free volume of the system. [21] Consistent with this, a slight swelling of the gels was observed with increase in temperature. On the other hand, the increasing number of interparticle interactions per colloid with temperature, interactions that are dynamic in nature, leads to a more dissipative system, justifying the increase in tan(δ).
An alternative or complementary explanation is based on the behavior of semicrystalline thermoreversible gels. [22] Here, small crystallites constitute multifunctional physical cross-links in the gel network. At sufficiently high temperatures, the crystallites melt. While we do not have any evidence that supports the crystallinity of the crosslinks, the gels do exhibit certain crystallinity, and hence, this possibility cannot be discarded. Hence, the softening of the gels (and the increase in dissipative behavior) could be also justified by the partial melting of the crystallite crosslinks with increase in temperature. Temperature-dependent crystallinity studies would be needed to provide evidence for this transition, which is out of the scope of this work.

Colloidal Surface Morphology and Lamellar Adhesion
Quantitative Imaging (QI) via Atomic Force Microscopy (AFM) allows the visualization of the topography (height), stiffness, and adhesion (pull-off force) of the gel surfaces in a fluid environment, without introducing any artifact due to drying. QI is performed by applying a small force with a sharp AFM tip so that the deformation of the surface is minimized. Figure 4 shows QI images of the surface of the gels with 60%, 70%, and 80% mAAc. Complementary AFM images are shown in Figures S4-S6 (Supporting Information). The AFM tip cannot detect any structural features on the surface of the gels synthesized with 95% mAAc.
The surface of 60% mAAc gels consists of clusters of µm-sized particles yielding a coarse colloidal structure; the visualized µm-large clusters are composed of aggregated particles, as also shown in the SEM images. A much finer colloidal microstructure with distinct features is visualized on the surface of the gels with 80% mAAc. The low-magnification height and stiffness display the presence of dispersed particles, while the high magnification image reveals a network of nanoparticles. Analysis of the cross-sections (taken with a Si tip of radius 10 nm) shows a particle size of 20-50 nm ( Figure S4, Supporting Information). The surface of 70% mAAc is heterogeneous displaying two different domains: one domain with clusters of particles smaller than that of 60% mAAc and a more homogeneous domain with a colloidal structure closer to that observed for 80% mAAc gels. It thus appears to exhibit a topography intermediate between that of 60% and 80% mAAc gels, but the variation across samples is significant. For the three compositions, the small adhesion of the particles compared to the surrounding matrix is consistent with their larger stiffness.
Interestingly, lamellar patterns can be primarily seen in the adhesion of gels with 70%, and 80% mAAc, although they only appear uniformly and persistently in gels with 80% mAAc. Note that such lamellar phase is not seen in the topography of the colloidal gels and only faintly in the stiffness, where the colloidal microstructure is still prominent. Hence, these lamellae are not topographic features, but result from a spatially varying interaction with the AFM tip. Examination of the force-distance curves extracted from QI images reveal that the approach (noncontact) interaction between the tip and the gel ( Figure S5, Supporting Information) is slightly attractive (within light lamellae) and repulsive (within the dark lamellae), justifying the observed variation of the pull-off force (positive for adhesive interaction and zero for repulsive interaction). The origin of this non-contact interaction force is likely electrostatic, which is supported by other experimental results described later. A lamellae-like charge distribution in the near-surface region of the gels would yield a lamellar electrostatic and pull-off force. The tip used to image the gel surfaces is made of silicon but due to exposure to air it is expected to have an oxide surface layer of a couple of nanometers (silicon oxide). Silicon oxide surfaces in water exhibit a negative charge, and hence, it is typically considered that unmodified Si tips are negatively charged. A spatially varying charge on both the surface and underneath the surface would be sensed by these AFM tips while imaging.
The origin of the lamellar pattern in these colloidal gels is intriguing. A colloidal gel forms via particulate aggregation, as the colloidal particles aggregate in order to minimize their surface energy. [23] Aggregation happens at the cost of i) an increase in electrostatic repulsion as the charged groups approach each other and ii) a decrease in entropy. [11] One potential avenue to reduce the free energy is microphase separation. This introduces more order in the system (and a further decrease in entropy) but reduces the overall repulsive interaction; in our case, presumably electrostatic. Here, charged groups are able to arrange themselves in lamellar patterns that minimize the overall repulsion at the hydrogel surface. [24] Rheological and SAXS results do not support the presence of a lamellar structure in the bulk of the hydrogel. One potential explanation is that the interface has a higher free energy due to the additional contribution of the interfacial free energy (absent in the bulk), and hence, the value of the free energy required to induce the microphase separation is only achieved at the interface. Several system parameters could theoretically impact the period spacing, but it is likely related to the magnitude of the entropic cost relative to the enthalpy reduction. Lamellar adhesion is also seen prior to annealing the sample, although it is not as prominent nor as ordered as after heat treatment ( Figure 4). However, this indicates that the lamellar pattern likely forms during synthesis and becomes more ordered as the system evolves toward equilibrium.
Note that the differences between AFM (top surface) and SEM (cross sections) images reflect structural gradients (surface vs bulk) due to artifacts induced by drying before SEM www.afm-journal.de www.advancedsciencenews.com imaging, the polymerization reaction, and sedimentation, which is expected for colloidal gels.

Hydrogen Bonding Stabilizes Colloidal Particles
A series of complimentary experiments were carried out to investigate the type of interparticle interactions that are relevant for the stabilization of the colloidal surface morphology and lamellar adhesion of 80% mAAc gels, which are the focus in the following.
Using QI imaging in a fluid cell, the surface structure of 80% mAAc gels was monitored over time when urea was added to the solution. Figure 5 shows the initial structure equilibrated in DI water and at selected times after introducing urea up to a final concentration of 3 m. The surface exhibits . Representative QI images of the surfaces of gels with 60%, 70%, and 80% mAAc in equilibrium with DI water. Particulate clusters decrease in size with increasing mAAc content, as shown by the higher magnification topologies, which allow better visualization of the nanostructure; note that the tip cannot fully track large height variations between particles due to its finite size.A lamellar pattern, which primarily appears in the adhesion maps and non-contact interaction with the tip (≈10 nm from the surface; see Figure S5, Supporting Information), is observed for 80% mAAc gels. 0 nN adhesion (pull-off force) happens when the force upon approach of the tip to the surface is repulsive.
www.afm-journal.de www.advancedsciencenews.com a uniform colloidal topography (only the largest particles are visualized at the selected magnification) and surface stiffness, along with lamellar adhesion. Urea leads to the dissolution of the coarse particles (see pits in height images), while it also leads to pits in stiffness, an increase in adhesion and weaker contrast between the two lamellar phases. It is well known that hydrogen bonding is nullified in the presence of urea.
Hence, our results indicate that hydrogen bonding between methacrylic acid and methacrylamide plays a critical role in the stabilization of, at least, the largest colloidal particles. A longer equilibration time could further influence the microstructure, but this was not investigated because the destabilization of the surface hindered AFM imaging. The images were confirmed to be representative of the material by imaging multiple Figure 5. Destabilization of interactions in 80% mAAc gels. A) QI images of the colloidal gels in the presence of urea. The decrease in height and the inversion of the surface stiffness of the particles versus the surrounding matrix indicates that particles are dissolving. B,C) Long-term and short-term influence of the immersion of the 80% mAAc gels in 5 mm NaOH (pH 11.7). B) Optical images showing the increasing transparency of the gels over time and C) reverse phase contact angle measurements. www.advancedsciencenews.com locations on the same sample and multiple samples from different syntheses.

Enhanced Electrostatic Interactions Destabilize the Colloidal Structure
A higher pH is expected to lead to the dissociation of the carboxylic groups and thereby to an increase of the charge density and the electrostatic repulsion. The gel is stable within a pH range of ≈4-9 for at least 1 h, but it responds immediately to pH changes in the range 10-12. Figure 5B shows representative optical images of 80% mAAc gels in 5 mm NaOH solution (pH = 11.7). The response to the increased pH is immediate but slow, perhaps due to the diffusion-limited dissociation of the methacrylic acid. It is expected that the electrostatic repulsion increases as more carboxylic groups dissociate. After days, gels significantly swell and soften ( Figure S7, Supporting Information), and their transparency increases, reflecting the decrease in number density of the colloidal particles as the negative charge increases. Overall, these results support that enhancement of electrostatic interactions destabilizes the colloidal gels.
The response of the gel surface to the increase in pH is much faster than the bulk response ( Figure 5C), likely because there is no diffusion limitation to the surface. Reverse-phase contact angle measurements were conducted by placing a drop of dimethyl carbonate on the gel while immersed in an aqueous solution ( Figure 5C). The reverse contact angle is 140° at pH of 5.5, while it increases to 177° after 20 min at pH 11.7, indicating the water contact angle decreases. Hence, the surface becomes more hydrophilic, as the methacrylic acid near the surface of the gel dissociates with the increase of the pH.

Colloidal Friction
Friction maps were generated using contact mode imaging by AFM to visualize the frictional characteristics of the heterogenous surfaces of 80% mAAc gels. During contact mode AFM imaging of the top surface, the tip slides in forward and reverse direction along a scan line at constant load and velocity before moving to the next scan line. Contact mode imaging produces height images during trace and retrace of the tip. Due to the friction force between tip and gel surface, the tip deflects in the lateral direction, and with the lateral spring constant of the cantilever, the lateral force during trace and retrace can be determined. The friction image results from half of difference of the lateral deflection in forward and reverse directions. [25] An example to illustrate this method is shown in Figure S8 (Supporting Information). A blunt silicon tip was used to avoid puncturing the relatively soft hydrogel surface during contact mode imaging. Friction studies were conducted with the gels immersed in DI water at velocities ranging from 5 to 200 µm s −1 and applied loads of 5 and 10 nN. Details are described in the Experimental Section. Figure 6A,B shows friction maps, while the corresponding height and lateral deflection images generated in contact mode imaging are shown in Figures S8,S9 (Supporting Information) and the QI image is displayed in Figure S10 (Supporting Infor-mation), as reference. Because the tip is blunt, the resolution of the topographic images in contact mode is not as good as in QI imaging, but sufficient to resolve the colloidal microstructure.
At both loads, friction is mainly dominated by the colloidal topography of the gel surface -labeled as colloidal friction. At 5 µm s −1 (Figure 6A), the highest friction force is measured at the largest particles (purple). This high friction region extends to more particles at 10 µm s −1 , while the region of intermediate friction (yellow) becomes more prevalent than the region of low friction (turquoise and dark blue). The particles consistently reflect the highest friction (purple), while the matrix surrounding the particles exhibit lower friction (gold, blue). The differences in height images during retrace and trace ( Figure S8, Supporting Information) point toward the viscoelastic deformation of the particles, which should contribute to the frictional dissipation. Note that the rheological data discussed earlier support the significant viscoelastic behavior of 80% mAAc gels at room temperature (tan(δ)→0.45).
Interestingly, lamellar-like friction appears at 20 µm s −1 at both loads, but more significantly at 5 nN, overall yielding a decrease in friction (turquoise). This lamellar pattern coincides with the lamellar adhesion discussed previously. At higher velocities (≥50 µm s −1 ), the lamellar friction vanishes, and instead, the friction image at both loads reflects the colloidal structure again. This indicates the existence of a turning point at ≈20 µm s −1 , and hence, the action of (at least) two different mechanisms determining friction above and below this transition velocity. The existence of this transition was confirmed on friction measurements on five different samples and experimental set ups, and hence, it is a universal characteristic of this system.
Trace and retrace height images significantly differ from each other at high velocities (≥ 75 µm s −1 , Figure S9, Supporting Information). For example, the trace height image at 10 nN and 100 µm s −1 does not display the surface topography, but instead the lamellae, indicating that the tip does not probe the surface well. This implies the tip lifts toward the right side of the friction maps, but it is in contact with the surface during retrace, and hence, there is a sliding asymmetry at high sliding velocities. At 5 nN, the tip partially lifts during retrace, as well. The (hydrodynamic) lift of the tip causes a prominent decrease in friction since a water film provides more efficient lubrication than the gel surface; this is more significant toward the right side of the friction maps (dark blue) and at lower loads.
Pixel values of all friction images were exported to histograms ( Figure 6B,D). To account for the heterogeneities in friction maps, we determined the contribution of each region (or phase) by deconvoluting the histogram into five gaussian distributions; each gaussian distribution has the same color as the corresponding region in the friction map ( Figure 6A-C). Figure 6E,F shows bubble diagrams, where each bubble gives the average friction of each single gaussian distribution (with color matching). The error bars give the corresponding standard deviation, and the bubble size gives the contribution of each single gaussian distribution to the histogram. The minimum in friction at 20 µm s −1 corresponding to lamellar friction is shown at both loads (see box). The partial hydrodynamic lift (or partial contact of the tip and the surface) appears as a decrease in friction (see arrows).

www.afm-journal.de www.advancedsciencenews.com
The turning point at 20 µm s −1 deserves further discussion. The viscous dissipation originating from the shear of the nearsurface polymer network justifies the steady increase of friction with sliding velocity. As described earlier, the difference between trace and retrace height images indicates that viscoelastic deformation (instead of a simply elastic behavior) of the gels occurs. The viscous component of the deformation leads to an increase in dissipated energy with shear rate. However, the contact area between tip and gel surface is also a function of the sliding velocity. This is because an increase in velocity decreases the contact time. As the contact time between tip and gel decreases, the viscoelastic deformation of the gel also does, and this yields a decrease in the contact area with increasing velocity. For example, the velocity-dependent contact area was estimated for polyacrylamide gels as a function of the loading time. [26] The minimum at 20 µm s −1 is associated with these two opposite contributions to the energy dissipation-the increasing viscous dissipation and the decreasing contact area with velocity-which justify the increase and decrease of friction, respectively, and add together leading to the observed friction minimum.
Adhesion hysteresis can influence the frictional dissipation. [27] For polymeric surfaces, adhesion hysteresis is caused by the interdigitation of the polymer across the interface, disentanglement, and reorientation of polymer chains and plastic deformation. We infer from these results that the contribution of adhesion hysteresis to the frictional dissipation of these colloidal gels must be small, as the lamellar friction (matching the lamellar adhesion) is only observed when the colloidal friction becomes sufficiently small, i.e., at the friction minimum. This was confirmed on multiple experiments with different samples.

Spatial Control of Adhesion via Changes of Electric Potential
To investigate the gel's responsive interfacial behavior, electrochemical studies were conducted on the 80% mAAc gels using a 3-electrode cell in 10 mm KCl, with the working potential applied to the bottom gel surface. The open circuit potential (OCP) of the system was determined to be −0.7 V (∆U = 0) via cyclic voltammetry ( Figure S11, Supporting Information). It is worth noting that the tilted shape of the Cyclic Voltammogram (CV) implies a potential drop due to low conductivity, which indicates that the colloidal microstructure does not allow fast ion diffusion.
Images of the structure of the top surface were taken in situ via Quantitative Imaging AFM to examine the response of the topography, surface stiffness, adhesion and non-contact interactions between gel surface and tip to the applied bias. It was found that the structural response and stability are dependent on the rate at which the potential bias is varied. If the potential is increased slowly from the OCP (1 mV s −1 ), the electrochemical window can be as wide as ∆U = ±2.0 V without evidence of electrochemical degradation. Representative QI images taken during the electrochemical experiments with a slow potential ramp are shown in Figure 7. Positive potentials lead first to gel swelling and to a minor collapse at the largest applied bias. Concurrently, the lamellar adhesion and non-contact interaction decrease in width (from ≈1.4 to ≈0.9 µm) and adhesion contrast between the two lamellar phases (from ≈200 to ≈125 pN) with increasing bias and return to nearly the same initial values upon potential reversal. Note that there are nine bright lamellae at the OCP, but they evolve into 11 bright lamellae at +2 V and return to 9 bright lamellae when the potential is reversed to the OCP. An increased roughness and stiffness are observed upon reversing the potential back to ∆U = 0 V, which could be due to particle aggregation over time. FTIR shows the appearance of stronger hydrogen bonding and decrease of the carboxyl peak ratio upon the application of the positive potential (see Figure S12, Supporting Information), which is consistent with the change of the colloidal structure.
It is reasonable to expect that the increased positive potential leads to a counterion enrichment (anions) in the hydrogel, which increases the osmotic pressure, justifying the swelling of the gel. Concurrently, the cation depletion can reduce screening of the copolymer charge, thereby leading to a more stretched copolymer conformation, promoting the swelling of the gel even further. On the other hand, the negatively charged poly(methacrylic acid) is attracted to the positive electrode, yielding a partial collapse of the gel. The influence of the potential on this balance justifies the intricate (non-monotonic) change of the swelling ratio of the gel, depending on the dominating mechanism. The variation of the lamellae characteristics with potential thus reflects that they respond to the change of the balance between electrostatic interactions.

Electrotunable Gel Friction
Friction images were taken at varying bias applied to the bottom surface of 80% mAAc gels equilibrated in 10 mm KCl to demonstrate the electrotunability of friction ( Figure 8A). The bias was increased at 1 mV s −1 from the OCP (−0.76 V, ∆U = 0 V) to selected potentials (0.4, 0.7, 1.2 and 1.7 V), at which the gel was allowed to equilibrate for ≈1 h. The applied potential caused the gel initially to swell, and then to collapse at 1.7 V. This collapse was concurrent with an increase of RMS roughness from ≈9 to 11.5 nm. The friction measurements were performed at 10 nN and 50 µm s −1 . Slightly different areas were imaged due to the drift happening during the equilibration time at each potential.
The tip tracks the surface well at all potentials, as inferred from the height images ( Figure S13, Supporting Information). Friction images reveal colloidal friction and display the maximum dissipation (purple) at ∆U = 0 and 1.7 V and a minimum at ∆U = 0.4 V (blue). The friction maxima occur with the collapse of the gel and the increase of roughness. The friction www.afm-journal.de www.advancedsciencenews.com histograms ( Figure 8B) widen away from the friction minimum, indicating that the response to the potential is heterogenous as is the surface topography. The friction minimum is also clearly seen in the friction histograms, as they shift to the left from 0 to 0.4 V and to higher friction values with increasing potential above 0.4 V.
The non-monotonic changes in friction ( Figure 8C) suggest the action of competing mechanisms, and the swelling of the gels help to explain this result. On one hand, volume expansion results from the osmotic pressure-induced swelling. That is, when a positive potential is applied, it drives the diffusion of cations out of the hydrogel and of anions into the gel. While there might be variations in osmotic pressure before equilibrium is achieved at each potential, [28] an equilibrium volume expansion is expected from the reduction of screening of the charged acrylic acid groups by K + ions and the more stretched conformation of the charged polymer chains. [28] This is combined with the attractive force acting on negatively charged (non-screened) acrylic acid groups toward the working electrode. At potentials from 0 to 0.4 V, the osmotic pressure and the more stretched chain configuration dictates the swelling of the gel with potential; as more water is present in the near-surface gel, friction decreases. Above 0.4 V, electrostatic attraction takes over, and the gel collapses with potential; and friction increases as the volume fraction of polymer, and the viscous dissipation increase. Because counterion diffusion and swelling/collapse (water uptake/release) is very slow, equilibrium might have not been achieved at the selected potentials. Despite this, these results are a proof of concept for the electrotunability of colloidal friction of 80% mAAc gels.

Discussion
In this work, we investigated the interactions involved in the formation of the microstructure of poly(methacrylamide-comethacrylic acid) gels, and more extensively for gels with 80% mAAc, which exhibit a (submicron) colloidal microstructure, semicrystallinity, and a dissipative rheological behavior. NMR proton environment along with the particles dissolution and gel swelling in urea indicates that hydrogen bonding between carboxylic and amide groups is a main contributor to the stabilization of the colloidal microstructure. Other gel systems involving mAAc have been shown to form hydrogen bond complexes percolating into a stretchable network, [29] but they lack the increased elasticity that the copolymer introduces in this gel system. By comparing mAAc with AAc, it was reported that the hydrophobicity of the methyl group in mAAc greatly increases the stability of hydrogen bonding and improve mechanical properties. [30] In our system, it is likely that hydrogen bonds are also protected from water molecules by the methyl groups, which serves to stabilize them in the aqueous solution and maintain their dynamic characteristic. The high stability upon changes of pH (<10) further supports the protection of the carboxyl groups from water, [31] effectively increasing the pKa of mAAc. [30] Similar reversible dynamic behavior has been shown in systems where the mAAm was replaced with www.afm-journal.de www.advancedsciencenews.com 1-vinylimidazole, leading to different hydrogen bond dynamics but high mechanical strength despite the lack of crystallinity. [32] On the other hand, the decrease of the electrostatic repulsion with decrease in mAAc-as less acrylic acid groups are present in the pre-gel solution-promotes particle growth and aggregation, as noticed for gels with 60% and 70% mAAc. Furthermore, the increase in electrostatic repulsion as more carboxylic groups dissociate with increase in pH destabilizes the colloidal network. We thus conclude that the balance between shortrange hydrogen bonding (2-3 Å) [17] and hydrophobic interactions, and long-range electrostatic repulsion (Debye length ≈175 nm for water at pH ≈5.6) -the SALR pair potential-dictates the formation and stabilization of the fine network of colloidal particles that constitutes the gels with 80% mAAc.
Both the relative strength and the number density of the inter-particle interactions determine the particle arrangement and connectivity, and thereby, the emergence of the gel elasticity (G′ > G″) and storage moduli in the MPa-range. [9] This range is of relevance for applications in soft robotics and biomedical implants and devices, for which the materials need to be soft enough to not damage biological tissue yet still provide structural support. Indeed, the moduli are comparable to that of biological cartilages and skins, as well as elastomers. [33] Furthermore, the frequency (time) dependent rheological response reflects the dynamic nature of interpolymer/interparticle forces. It is worth mentioning that, although hydrophobic interactions arising from the methyl backbone are not dynamic, we believe that their role is key in protecting the hydrogen bonds from water molecules and thereby stabilizing them in the aqueous solution and enabling their dynamic and reversible nature.
We also present a novel method based on AFM imaging to resolve the spatial distribution of friction and adhesion of gels in situ (in liquid) and to relate the interfacial response to the surface structure and dynamics. The investigated gels yield colloidal friction, i.e., friction reflects the viscous dissipation of the colloidal network upon shear, while the low adhesion to the Si tip is mainly dictated by the electrostatic interaction, which exhibit a lamellar pattern. Furthermore, the optimal design of soft interfaces for biomimetic robots, biomedical devices, and human-machine interfaces relies on the ability to dynamically control their interfacial properties including topography, stiffness, adhesion, and friction. This work provides a proof of concept for the electrotunability of adhesion via changes of the lamellar pattern and their strength, and of the gel's colloidal friction. Because the colloidal system is arrested in a non-equilibrium state arising from the temporary balance between multiple interactions, altering the force balance can result in irreversible changes of the colloidal structure. Hence, it is important to consider effects of hysteresis and ageing of colloidal gels when shifts in the force balance are applied. Conditions that eliminate the observed ageing would be more interesting for real applications.
The implications of this work from a theoretical perspective are also important. Even though the application of equilibrium statistical mechanics to colloidal systems has been successful  Figure S13 (Supporting Information). www.advancedsciencenews.com by treating non-ergodic behavior within an arrested or static framework, these arguments are underdeveloped. [34] Current models generally consider only the relative strength of attractive and repulsive inter-colloidal forces to predict the structural behavior and mechanical properties of colloidal gels. This work does not only provide an experimental study for comparisons to theory, but also one which encompasses various interactions with varying relative strengths. Due to experimental challenges to in situ probe colloidal systems (e.g., due to solvent interference and the inability to accurately track particle movements on the experimental timescale), current experimental investigations are limited mainly to time-resolved spectroscopy techniques, [35] and confocal microscopy. [36] Hence, our experimental studies represent a significant advance in understanding the structure-property relationship of this type of gel.

Conclusions
Structural and dynamic properties of copolymer colloidal gels have been investigated through a combination of characterization methods. Depending on the ratio of mAAc and mAAm monomers, the gel structure evolves from micrometer-large aggregated colloidal particles to a fine network of interconnected submicron copolymer particles. Focusing on the colloidal gels with 80% mAAc, which exhibit the second type of microstructure, we propose that the colloidal microstructure is stabilized by an SALR pair potential. The pair potential consists of short-ranged hydrogen bonding between mAAc and mAAc monomers, and hydrophobic attraction-due to the methylated polymer backbone-as well as a long-ranged electrostatic repulsion from the fraction of charged methacrylic acid monomers. The reversible and dynamic nature of this SALR pair potential justifies the dissipative rheological behavior of the gels, while they exhibit a large storage modulus of ≈1 MPa at room temperature and certain (semi)crystallinity.
In situ AFM imaging spatially resolves the lamellar adhesion of 80% mAAc gels as well as their frictional characteristics, which reflect the colloidal topography-labeled as colloidal friction. Friction is determined by both the viscous dissipation upon shear of the near-surface colloidal network and the contact area, which decreases with sliding velocity due to the gel's viscoelastic deformation. These two competing mechanisms lead to a colloidal friction minimum at a sliding velocity ≈20 µm s −1 . At this minimum, adhesion-originated lamellar friction becomes relevant. The dynamic and reversible nature of the interactions involved in the stabilization of the gel microstructure not only influences the viscoelastic behavior of the gels but also introduces a stimulus responsive interfacial behavior. The electrotunability of gel adhesion and friction is demonstrated in proof-of-concept experiments. The results of this work help to improve the understanding of structure-property relations of colloidal gels and ultimately guide the design principles of gels with responsive behaviors by manipulating the fundamental interactions that stabilize the polymer network.

Experimental Section
Sample Preparation: 98% methacrylamide, 99% methacrylic acid with 250 ppm monomethyl ether hydroquinone (MEHQ) as an inhibitor, and N,N,N″,N″-tetramethylethane-1,2-diamine (TEMED) were used as-received from Sigma-Aldrich. The synthesis protocol was adapted from previous work by Zhang et al. [14a] by using Ammonium Persulfate (APS) in DI water as initiator and a higher concentration of tetramethyl ethylenediamine for a total monomer concentration of 5 m, which led to faster polymerization kinetics. Instead of using glass surfaces as molds, [14a] the monomer solution was poured on a polycarbonate substrate and polymerization happened in a nitrogen atmosphere. For this work, molar ratios of 60:40, 70:30, 80:20, and 95:05 methacrylic acid:methacrylamide were chosen. Corresponding amounts of methacrylic acid, methacrylamide, and 10 wt./vol.% APS were added to DI water to reach a total monomer concentration of 5 m, and an APS final concentration of 0.5 mol% relative to the total monomer concentration. The monomer solution was degassed using in-house vacuum for 15 min, and 1 vol% TEMED was added to the prepolymer solution. 3 mL of the activated prepolymer solution was pipetted into 40 mm diameter polycarbonate petri dishes and left to react at 45 °C in a nitrogen environment for 2 h until a gel was formed, resulting in discs ≈2-3 mm thick. Samples were transferred to a DI water bath and left at 45 °C for 20 h.
Water Content: The water content of the hydrogels was determined using gravimetric measurements. Three samples were weighed fully hydrated and again after being dried completely at 60 °C. Elevated temperatures were used to remove any bound water. The difference in mass of the two measurements was attributed to water loss, and the water content was calculated using: where w h and w d were the weight of the hydrated and dried sample, respectively. Scanning Electron Microscopy (SEM) A Hitachi 4800 was used for SEM imaging using mixed upper and lower SE detection. Samples were snap-frozen using liquid nitrogen and lyophilized to reduce structural changes during the drying process. However, the surfaces were observed to be significantly changed, and hence, only the cross-sections are considered here. The cross-sections were then sputter-coated in 5-10 nm of Au-Pd using an Emscope SC 500 to reduce surface charging and damage and obtain higher resolution images.
FTIR: An ATR-FTIR (Frontier FT-IR, PerkinElmer, USA) was used to characterize the chemical composition of the gels. The depth of penetration for this gel was minimal due its opacity. Spectra were taken on both the top and bottom surfaces of the gels and multiple spots to account for structural gradients. 10 scans were taken per spectra. Polymethylacrylamide and polymethylacrylic acid were synthesized and mixed at known mol% in the range 60-95% mAAc. The carboxyl/ amide peak intensity ratio at 1655 cm −1 /1690 cm −1 was used to define the calibration curve as a function of mAAc% in solution. The calibration curve was applied to the peak intensity ratio at 1655 cm −1 /1690 cm −1 of the gels to determine the actual mAAc % incorporated into the gel network.
Surface Imaging by Atomic Force Microscopy: An atomic force microscope (Nano Wizard, JPK Instruments, Germany) was used for Quantitative Imaging (QI). To avoid dehydration and structural changes due to evaporation, all AFM measurements were conducted in the respective aqueous fluid. Sharp tips for QI imaging were Si tips with a nominal radius of 8 nm (HQ:CSC37/No Al, MikroMash, nominal spring constant 0.6-1.2 N m −1 ). Due to the limitation of the image size, all experiments were conducted at multiple spots on the sample surface, and reproducibility of the hydrogel structure was tested by imaging multiple samples. The imaged surfaces were those gelated against the polycarbonate petri dish. Samples were mounted on to glass slides by slicing off small pieces (≈6 × 6 mm) using a razor blade. Excess water was aspirated from the bottom of the sample using a Kimwipe and bound to the glass surface using an epoxy glue (J-B KwikWeld, USA). The setup for electrochemical measurements can be found in Figure S11 www.advancedsciencenews.com (Supporting Information). Cyclic Voltammogramswere collected prior to each measurement so that constant biases were applied and electrochemical stability windows could be evaluated.
X-Ray Diffraction: A Bruker D8 Advance XRD System which emits Cu K-α 1 radiation was used (2D Eiger2 R 500K detector, 0.2 mm Goebel slit, 40 kV, 40 mA). Hydrated samples were placed on a low-background single-crystal Si holder at room temperature. Scans were taken at 0.1° s −1 to limit dehydration and 0.01° step increments. Background scans taken with water were subtracted from all measurements, and z-cradles were performed for each sample so that intensities are comparable and account for differences in sample roughness, thickness, and opacity.
Rheology: A TA Discovery HR-3 rheometer was used to measure the viscoelastic properties of the gels. 25 mm discs were cut out using a steel punch and sandwiched between a textured 50 mm bottom plate and a smooth 25 mm top parallel plate geometries. The textured plate was used to mitigate slip, and the edges of the gel were immersed in excess DI water to prevent dehydration. Frequency sweeps were performed at 0.05% strain with frequencies from 0.1 to 100 rad s −1 . Amplitude sweeps were performed at 10 rad s −1 with amplitudes ranging from 0.01 until yielding behavior was significant. Each data point was taken as an average of a minimum of four periods of oscillation, and a minimum of three samples were analyzed per composition. Oscillation waveforms were monitored to check for slip.
Solid-State HNMR: A Varian Unity Inova 300 was used for solid-state proton 1D MAS NMR data acquisition. Gels equilibrated in DI water were cut into long, 3 mm diameter cylinders (≈70 mg) and capped to prevent dehydration. Samples were rotated at 10k Hz to obtain significant peak resolution. Spectra were analyzed using MNova software.
Contact Angle: Reverse-phase contact angle measurements were conducted using a Rame-Hart Model 250 Contact Angle Goniometer using Dimethyl Carbonate as received (99%, Sigma-Aldrich) placed on the surface of the gel submerged in aqueous phase.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.