Coarsening of the Pore Network in Drying Latex Films upon

Article pubs.acs.org/Langmuir

Coarsening of the Pore Network in Drying Latex Films upon Interparticle Aggregation Katja Pohl, Rebekka König, Heike Römermann, Malin Schulz, and Diethelm Johannsmann* Institute of Physical Chemistry, Clausthal University of Technology, 38678 Clausthal-Zellerfeld, Germany ABSTRACT: The lateral drying front observed during film formation from latex dispersions with a Tg of the polymer around room temperature is composed of three three distinct lines. The lines are characterized by a decrease in turbidity, a renewed sharp increase in turbidity, and a more gradual decrease in turbidity at the end of what can be called a “halo”. Microcracks with herringbone morphology develop at the first line, where the turbidity decreases. If macrocracks are present, these nucleate close to the end of the halo. At the line, where the turbidity sharply increases, one also observes an increase in stress birefringence. The substructure of the drying front is characteristically different from the structures described previously for films drying from hard particles. In particular, the renewed increase in turbidity cannot be explained as pore-opening, but rather is the consequence of a coarsening of the pore network after the particles jump into contact. A capillary instability sets in, by which the small pores collapse under the polymer/water interfacial energy, while the larger pores expand correspondingly. The instability (related to the Rayleigh instability of liquid jets) makes the films appear turbid. Also, the induced mechanical heterogeneity prevents straight macrocracks from penetrating into the halo because crack deflection and crack branching would result, which is energetically unfavorable.

I. INTRODUCTION The ability to form clear films with smooth surfaces upon drying either from solution or from dispersion is among the distinctive properties of polymers.1 Unlike most inorganic materials, polymers can relax drying-induced stresses and go through what is called “film formation”.2,3 Film formation from polymer dispersions is often depicted as a three-stage process, comprising the evaporation of water until the particles touch (stage I), particle deformation (stage II), and polymer interdiffusion across the boundaries between particles (stage III). Within this picture, the film is treated as homogeneous, meaning that the transport of water is fast enough to let spatial heterogeneities be insignificant. Heterogeneities may occur both along the normal and the lateral direction. Along the surface normal, they can be avoided by making the Péclet number, Pe, sufficiently small. Pe is the ratio of the diffusion time (the time needed to equilibrate concentration gradients between the top and the bottom of the film) and the drying time.4,5 If Pe ≫ 1, a drying front propagates from the top of the film to the bottom. Importantly, a drying front may also occur in the plane of the film.6,7 Drying fronts propagating laterally across the film (often in the form of “edge-in drying”) are rather common because the rate of evaporation usually varies over the surface of the film. Edge-in drying gives rise to the well-known coffee-stain effect, meaning that there is a flow of material toward the edge.8,9 After drying, one finds a bulge at the rim. Occasionally, there also is a small depression at the center (a “crater”), where the lateral drying fronts eventually close in. Edge-in drying is attractive from an experimental point of view because it allows to visualize to progress of compaction. As previously noted in refs 10 and 11, the drying front often has a substructure. References 10 and 11 were concerned with particles, which do not film form at room temperature because © 2014 American Chemical Society

of their high Tg. For such samples, the different lines can be assigned to particle ordering, jump into contact, pore opening, and cracking. A substructure of the drying front is also found for softer particles, but there are characteristic differences. The drying front is composed of three separate lines, labeled “1”, “2”, and “3” in Figure 1. At line 1, the turbidity decreases. The

Figure 1. Image of a drying dispersion showing a decrease of turbidity at line 1, followed by a renewed increase at line 2. Panel B is an enlargement of the white square in panel A. Since the material’s Tg is around 12 °C (B in Table 1), there are no cracks. The lines in panel B denote the three transitions under discussion. The third transition is more gradual than transitions 1 and 2.

material ahead of line 1 clearly is in a fluid state. Microcracks with herringbone morphology develop at line 1. At line 2, the turbidity sharply increases. At the same time, there is an increase in stress birefringence. The height profile of the drying films displays kinks at both line 1 and line 2. Around line 3, the turbidity again decreases gradually, until the film eventually becomes clear. If the film displays macrocracks (some films do), these terminate around line 3. We call the bright region between lines 2 and 3 “halo” in the following. Received: April 8, 2014 Revised: July 4, 2014 Published: July 17, 2014 9384

dx.doi.org/10.1021/la501354k | Langmuir 2014, 30, 9384−9389

Langmuir

Article

Following the arguments raised in ref 10, line 1 might be associated with jump into contact. Within this picture, line 2 would be associated with pore opening, but this latter interpretation contradicts the experimental evidence in a few regards, elaborated on in the Discussion section. Clearly, the film is heterogeneous at the scale of the wavelength of light behind line 2, but this renewed increase in turbidity is consequence of a coarsening of the pore network rather than pore opening. As we argue below, the coarsening is the consequence of a capillary instability, driven by the polymer− water interfacial energy. It only occurs for sufficiently deformable particles and is important for the film-formation process because it arrests crack propagation.

Figure 2. Images taken on a drying front: (A) diffuse reflection; (B) transmission with crossed polarizers. The material has a glass temperature of 20 °C (C in Table 1). The solid, the dashed, and the dotted line indicate lines 1, 2, and 3, respectively. As panel B shows, there is stress birefingence behind line 2.

upper left). Importantly, there is a sharp increase in stress birefringence at line 2. Stress birefringence behind line 2 has been observed for many different samples. There was a reason to orient the sample such that the drying front makes an angle with the vertical. The principal axes of the stress birefringence are along and perpendicular to the drying front. If the drying front is lined up with one of the polarizers, the birefringent area becomes dark. Figure 3 addresses the height profile of the film around lines 1 and 2. It is difficult to obtain quantitative maps of film height

II. MATERIALS AND SAMPLE PREPARATION The experiments described below were carried out with acrylic latexes. Latex dispersions were prepared by miniemulsion polymerization. The monomers employed were methyl methacrylate (MMA, Carl Roth, 99%), n-butyl acrylate (BA, Aldrich, 99%), and acrylic acid (AA, Fluka, 99%). The ratio of n-butyl acrylate (BA) and methyl methacrylate (MMA) was adjusted such that the Tg of the polymer was around room temperature. Sodium dodecyl sulfate (SDS, Carl Roth, 99%) was used as surfactant, hexadecane (HD, Aldrich, 99%) as costabilizer and azobis(isobutyronitile) (AIBN, recrystallized from ethanol) as the initiator. The monomer emulsions were sonicated for 4 min and polymerized for 20 h at 70 °C. A polymer latex with a broad particle size distribution was synthesized with 1 wt % surfactant instead of 2 wt %. Also, the time of sonication was reduced from 4 min to 15 s. Table 1 shows the various values of Tg as calculated with the Fox

Table 1. Composition and Glass Temperatures of Various Samples As Calculated with the Fox Equation A B C

MMA [wt %]

BA [wt %]

AA [wt %]

Tg [°C]

39.5 54 59

59 44.5 39.5

1.5 1.5 1.5

−10 12 20

equation.1,12 The solids content of the latex dispersions was 25 wt %. The dispersions were not formulated in any way. In particular, there were no thickeners added to the recipe for the samples discussed below. Generally speaking, addition of thickeners reduces the effects discussed below. Films were cast at room temperature onto glass slides. The dispersion was spread onto an area of 1.5 × 1.5 cm2, limited by Scotch tape. The dry film thickness was 80 μm. A commercial dispersion (Brilliant Acryl Klarlack, F.0910, AkzoNobel Farbe and Heimtex GmbH) was studied for comparison. The experimental data mostly consist of images taken with a conventional optical microscope (Zeiss Axioplan) and a stereoscope (Will, Wetzlar). Height profiles were acquired with the Tencor P-1 long scan profilometer (Tencor Instruments). In some cases, the serum was stained with a water-soluble dye (Congo Red, Riedel-de Haën), which aids the analysis of the liquid transport during film formation. Dynamic light scattering (DLS) was undertaken with an instrument from ALV (Langen, Germany).

Figure 3. (A) Optical image of a drying film (Tg ∼ 20 °C, C in Table 1) taken in reflection. The light source and the camera were arranged such that the camera sees a reflected beam, when the film surface is inclined relative to the substrate. Clearly, there is a decrease of thickness ahead of both line 1 and line 2. The film is flat behind line 2. Both steps are linked to a tangential flow of water. (B) Sketch of the film height.

versus time. However, the central piece of information is readily extracted from images taken in reflection as sketched in Figure 4B. The lamp (a point source) and the camera were mounted such that a specular reflection reaches the camera if the sample surface is inclined with respect to the substrate. This is, for instance, the case on the lower left in panel A, which is part of the droplet-shaped wet portion of the film. Importantly, the region between lines 1 and 2 also produces a reflection: This part of the film surface is inclined, as well. The surface is flat (within the limited resolution of the technique) behind line 2. There are kinks in the height profile at lines 1 and 2. Both Figures 2 and 3 show microcracks with herringbone morphology. These nucleate at line 1. The kinks in the height profile as sketched in Figure 3B suggest that liquid is pumped away from the respective transition toward the liquid portion of the film. This type of flow has a direction opposite to the flow caused by the coffee-

III. RESULTS A result of central importance is displayed in Figure 2, which shows the drying front of an acrylic latex with a Tg of 20 °C (C in Table 1). The front propagates from right to left. Lines 1, 2, and 3 are indicated by solid, dashed, and dotted lines, respectively. The image in panel A was taken in diffuse reflection; the image in panel B was taken in transmission with crossed polarizers. The time lapse between panels A and B is about 30 s. One can actually see that the drying front has propagated between the two images (see the dark speck on the 9385

dx.doi.org/10.1021/la501354k | Langmuir 2014, 30, 9384−9389

Langmuir

Article

Figure 4. (A) Film where the serum had been stained with a dye. The Tg of the dispersion is 20 °C (C in Table 1). The dye distribution shows that liquid has been flowing toward the center during film formation. There is a depression at the center, resulting from the coffee stain effect. (B) Height profile taken with a profilometer along the dotted line in panel A.

stain effect. Figure 4 shows that the flow toward the center can be strong enough to result in a transport of polymeric material toward the center of the film (rather than the edge). The sample shown in Figure 4 was dried from a dispersion stained with a water-soluble dye (Congo Red). The gray value therefore encodes film thickness. More precisely, it encodes the amount of dye at a certain position, which is related to film thickness but not necessarily proportional to film thickness. Still, the film clearly is darker in the center than at the edges, which proves that the serum has experienced a flow toward the center. Figure 4B shows a height profile acquired with a profilometer. The profile also shows a slope in thickness, where the edge of the film is thinner than the center. This particular sample shows cracks and a crater at the very middle of the film. The flow of polymer (in addition to a flow of liquid serum) is most impressive for films with a Tg close to room temperature. These also tend to crack. Figure 5 addresses the question of whether line 1 can be interpreted as an ordering transition. Such an ordering transition has been previously observed on films containing particles with a Tg much above room temperature.10 Colloidal crystallization is absent here, as proven with films dried from dispersions with a broad particle size distribution (Figure 5B). This broad size distribution was obtained by reducing the amount of emulsifier relative to the standard recipe (see the Materials and Sample Preparation section) and, also, by reducing the time of sonication. Figure 5C shows the distribution of hydrodynamic radii as determined with dynamic light scattering (DLS). The distribution obtained with the standard recipe is shown for comparison. Colloidal crystallization requires narrow particle size distributions, but the appearance of line 1 is the same in panels A and B. We also tested for reversibility of the first transition as demonstrated in ref 10. For hard particles the ordering transition was found to be reversible in ref 10. The respective line moved backward when exposing the sample to humid air. For the soft films studied here, tests along the same lines did not give clear answers. In some cases line 1 could be pushed back with humid air, in others it could not.

Figure 5. Appearance of line 1 does not depend on whether the particle size distribution is narrow (panel A) or broad (panel B). Since a transition to an ordered colloidal crystal is not expected for a latex with large size polydispersity, this experiment rules out colloidal crystallization as the mechanism giving rise to the first drying front. Panel C shows the particle size distributions as determined with dynamic light scattering (DLS).

Most of this work was carried out with unformulated latexes with a Tg around room temperature. However, the occurrence of a substructure is by no means limited to this particular material, as Figures 6 and 7 demonstrate. Figure 6 was taken on

Figure 6. Images taken on a drying dispersion with a Tg of −10 °C (A in Table 1). The white lines sketch the height profiles of the film. (A) Diffuse reflection. (B) Reflection with the lamp and the camera arranged such that the camera sees a reflection when the film surface is inclined relative to the substrate. (Scattering portions of the film also appear bright.)

a soft latex (Tg ∼ −10 °C, A in Table 1). Panel A (acquired in diffuse reflection) shows the familiar halo. The halo is narrower than for the samples shown in the previous figures, but it is still present. Figure 6B was acquired in reflection with a point source illuminating the film surface at an angle. Inclined areas produce a specular reflection in the same way as in Figure 3. For the liquid portion, the specular reflection is superimposed onto strong scattering. It is hardly discerned. The two bright bands further to the right are the reflection from the edge ahead of line 2 and the halo. The height profile displays two kinks. 9386

dx.doi.org/10.1021/la501354k | Langmuir 2014, 30, 9384−9389

Langmuir

Article

renewed increase in turbidity should not be attributed to pore opening because the particles are too soft to withstand the large capillary pressure occurring while a meniscus recedes into a porous structure.13 Also, is it is not clear why the film thickness should rapidly decrease prior to pore opening (Figure 3) and why the stress birefringence should be increased after pore opening (Figure 2). Pore opening should relax the stress rather than increasing it. Clearly, either the first or the second transition must be the jump into contact, which in both cases leaves the question of what the other transition should be assigned to. If the second transition is viewed the jump into contact, the first transition can be assigned to a jump into the secondary DLVO minimum. The jump into the secondary minimum is not necessarily connected to an ordering transition (but it can be). Before further discussing this interpretation, we briefly go through the other possible assignment. If the first transition is the jump into contact, the second transition must be assigned to another process of compaction, whichat least in principlemight be a self-accelerating collapse of the void space in-between the particles. A void collapse under capillary pressure is known from hydrodynamics, where it carries the name “cavitation”. Differing from cavitation, however, the pores in latex films are filled with an incompressible medium, which has to escape from the collapsing region. As the size of the pores decreases, the hydrodynamic drag forces increase as well. For the latex films studied here, there actually is experimental evidence in favor of the first assignment (first transition being the jump into the secondary DLVO minimum, second transition being the jump into contact). It rests on the effect of salt, displayed in Figure 8A−D. When adding salt, the distance between the two transitions becomes smaller (although it never goes to zero). This finding is easily explained in the frame of DLVO theory. Figure 8E sketches the interaction potential between two particles according to DLVO theory, using a particle radius of a = 50 nm, a Hamaker constant of A = 10−20 J, an electrostatic energy at contact of V0 = 100 kBT, and three different Debye lengths, which are rD = 0.5, 1, and 2 nm. The following equation was used:

Figure 7. Image taken on a drying commercial lacquer. The lamp and the camera were arranged such that the camera sees a reflected beam when the film surface is inclined relative to the substrate.

Figure 7 was taken on a commercial lacquer purchased at a department store (see the Materials and Sample Preparation section). It is a formulated product. The material shows neither stress birefringence nor a halo. However, even this material shows the two kinks in the height profile. The image was taken in reflection with the lamp and the camera arranged as sketched in Figure 3B. One observes a bright band next to the edge of the liquid region. Figure 8 shows the influence of the salt content. Such an influence does exist. All images were taken in reflection under

V (D) = − Figure 8. Images taken in reflection on drying films (Tg ∼ 20 °C, C in Table 1), where salt was added in concentrations of 38, 60, and 150 mM/L prior to film formation in panels B, C, and D, respectively. Panel A shows the dispersion with no added salt. The width of the images corresponds to 1 mm. The gap between lines to 1 and 2 decreases in width with increasing ion strength. Panel E displays a model calculation based on DLVO theory, which can explain the narrowing of the region between lines 1 and 2 (see the main text).

⎛ D⎞ Aa + V0 exp⎜ − ⎟ 12D ⎝ rD ⎠

(1)

V is the interaction potential, and D is the distance between the particle surfaces. The first and the second term on the righthand side are the van der Waals attraction and the electrostatic repulsion, respectively. With decreasing Debye length (that is, with increasing salt content), the secondary minimum becomes deeper and moves toward shorter distances. Consequently, the particle volume fraction of an assembly of spheres with an interparticle spacing corresponding to the secondary minimum increases and the difference between the solids contents in the metastable state and the final, aggregated state decreases. The time spent between the two transitions shortens, and the gap between the two lines in the images of drying films narrows down. Accepting this interpretation, it must be explained why there is stress birefringence after the jump into contact and why the film’s turbidity increases rather than decreases. Since jump into contact occurs out of a state, where weak bonds have been established already, aggregation is a transition between two elastic states. The elastic nature of the material before jump into contact is evidenced by the herringbone cracks. Cracking

illumination with a point source as in Figure 3. The white vertical bands are the region between the first and the second transition. The distance between the two transitions decreases with increasing ion strength. Electrostatics evidently plays a role in the process.

IV. DISCUSSION As already mentioned in the Introduction, the assignments of the different lines made in ref 10 for films with Tg much above room temperature cannot be applied to the samples studied here. The first transition cannot be attributed to colloidal crystallization because of the size polydispersity. Also, the 9387

dx.doi.org/10.1021/la501354k | Langmuir 2014, 30, 9384−9389

Langmuir

Article

thickness. Both findings speak against skin formation driving the reverse flow, but again, skin formation may have an effect. There is a second peculiarity associated with wet sintering immediately after formation of a direct contact, which is an instability with regard to the pore size distribution. The capillary pressure is highest in the smallest pores, which induces a collapse of the small pores and a corresponding expansion of the larger pores. The phenomenon is the analogue of the Rayleigh instability observed with liquid jets. There is a size distribution of the pores even without the instability because of the random arrangement of the spheres. (Remember that colloidal crystallization was suppressed by the size polydispersity.) We call this process pore-coarsening. Pore-coarsening has the immediate consequence that a film, which was clear already (because the pores were too small to scatter light efficiently), turns into a turbid film after the collapse. Pore-coarsening is of much practical importance because it impedes cracking. As the stress birefringence data show, the tensile stress is largest immediately after the jump into contact. At the same time the internal cohesion inside the film is weak because the particles have established a partial contact only and because there was little time for polymer interdiffusion across the interparticle boundaries. Still, the straight macrocracks do not reach to the aggregation line (line 2); they terminate at the end of the halo (line 3). This is naturally explained by the material’s mechanical heterogeneity. Microscale mechanical heterogeneity is a well-known method to stop crack growth and brittle facture.20 It is, for example, employed in the preparation of high-impact polystyrene (HIPS).21 If the material contains hard and soft regions in parallel, these deflect cracks and induce branching, which is energetically unfavorable and therefore impedes crack growth.

requires a cohesive material. Because the jump into contact is a solid−solid transition, it induces a stress and stress birefringence. Assuming that soft and hard spheres behave similarly in this regard, an elastic state of the material before the jump into contact is also inferred from the experiments by Boulogne et al.,14 who studied drying films with small-angle neutron scattering and found an anisotropic structure factor. An anisotropic structure factor is most easily explained with an elastic medium and a nonzero deviatoric stress. A note on the side: The images in ref 10 show herringbone cracks to nucleate at the jump into contact, not at the first transition, which corresponds to ordering and to the jump into the secondary DLVO minimum. With regard to microcracks, the drying phenomenology differs substantially between hard and soft spheres. A second note on films formed from hard particles: When using PMMA latexes with a Tg of around 100 °C, we did not observe transition 1 (contrasting to the experiments reported in ref 10). The difference presumably is in the size polydispersity (which is broader here). This finding corroborates the interpretation given to the first transition in ref 10, which emphasizes particle ordering. One might argue that a jump into the secondary DLVO minimum contributes to gain in free energy at transition 1 even in the experiments reported in ref 10. That is plausible, but if this contribution were dominant, we would observe transition 1 with the PMMA latexes (Tg ∼ 100 °C), as well. The enthalpy gain associated to the jump into the secondary DLVO minimum is expected to increase with particle softness because of the flattening of the sphere at the point of closest approach. “Closest approach” here implies an interparticle distance corresponding to the DLVO minimum (not a direct contact). Put differently, the first transition is mostly of entropic origin (ordering) for hard spheres with a narrow size distribution, while it is mostly of enthalpic origin (jump into the secondary minimum) for soft spheres with broad size distribution. It would require soft spheres with a narrow size distribution to have both effects operating. The two effects might act synergistically at the same time (both contributing to transition 1), or they might give rise to an even more complicated substructure of the drying front, where ordering and jump into the secondary minimum happen one after the other. Increasing turbidity and the flow of material toward the center of the film can be explained with an immediate deformation of the spheres after the formation of the contact. The enthalpic gain associated with the jump into contact of soft spheres is larger than the enthalpy gain of corresponding hard spheres because a significant fraction of the polymer−water interface turns into a polymer−polymer interface shortly after the contact is formed. The contact radius scales as t1/2,15 which implies a fast progress of the initial phase of wet sintering. Because of the large energy again associated with wet sintering, it can revert the flow induced by the coffee-stain effect. Admittedly, a flow toward the center of a drying film might also be connected to skin formation.16−18 The coffee-stain effect is caused by a pinning of the three-phase line2 and increased evaporation rate at the edge of the drop.19 If a skin forms at the edge first (it usually does), it lowers the evaporation rate there, which may cause a flow toward the center. We varied the film thickness and the hardness of the films and found the flow toward the center to be more prominent for the high-Tg materials. There was no effect of film

V. CONCLUSIONS Films drying from latex dispersions with a Tg around room temperature show a turbid halo after the jump into contact, which is caused by a coarsening of the voids in between the latex spheres, driven by a capillary instability. This type of heterogeneity is beneficial for film formation because it hinders crack propagation.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (D.J.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Helpful discussions with Jörg Adams are much appreciated as well as financial support by the DFG (Contract DFG Jo278/181).

■

REFERENCES

(1) Mischke, P.; Brown, R. Film Formation in Modern Paint Systems; Vincentz Network: Berlin, 2010. (2) Routh, A. F. Drying of thin colloidal films. Rep. Prog. Phys. 2013, 76 (4), 046603. (3) Keddie, J. L.; Routh, A. F. Fundamentals of Latex Film Formation: Processes and Properties; Springer: Berlin, 2010. (4) Sheetz, D. P. Formation of films by drying of latex. J. Appl. Polym. Sci. 1965, 9 (11), 3759. 9388

dx.doi.org/10.1021/la501354k | Langmuir 2014, 30, 9384−9389

Langmuir

Article

(5) Routh, A. F.; Russel, W. B. A process model for latex film formation: Limiting regimes for individual driving forces. Langmuir 1999, 15 (22), 7762−7773. (6) Routh, A. F.; Russel, W. B. Horizontal drying fronts during solvent evaporation from latex films. AIChE J. 1998, 44 (9), 2088− 2098. (7) Salamanca, J. M.; Ciampi, E.; Faux, D. A.; Glover, P. M.; McDonald, P. J.; A, F.; Peters, A.; Satguru, R.; Keddie, J. L. Lateral drying in thick films of waterborne colloidal particles. Langmuir 2001, 17 (11), 3202−3207. (8) Berteloot, G.; Hoang, A.; Daerr, A.; Kavehpour, H. P.; Lequeux, F.; Limat, L. Evaporation of a sessile droplet: Inside the coffee stain. J. Colloid Interface Sci. 2011, 370, 155−161. (9) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.; Witten, T. A. Capillary flow as the cause of ring stains from dried liquid drops. Nature 1997, 389 (6653), 827−829. (10) Goehring, L.; Clegg, W. J.; Routh, A. F. Solidification and ordering during directional drying of a colloidal dispersion. Langmuir 2010, 26 (12), 9269−9275. (11) Holmes, D. M.; Kumar, R. V.; Clegg, W. J. Cracking during lateral drying of alumina suspensions. J. Am. Ceram. Soc. 2006, 89 (6), 1908−1913. (12) Brandrup, J. Polymer Handbook; Wiley: New York, 2003. (13) Routh, A. F.; Russel, W. B. Deformation mechanisms during latex film formation: Experimental evidence. Ind. Eng. Chem. Res. 2001, 40 (20), 4302−4308. (14) Boulogne, F.; Pauchard, L.; Giorgiutti-Dauphine, F.; Botet, R.; Schweins, R.; Sztucki, M.; Li, J.; Cabane, B.; Goehring, L., Structural anisotropy of directionally dried colloids. Europhys. Lett. 2014, 105, (3). (15) Frenkel, J. Viscous flow of crystalline bodies under action of surface tension. J. Phys. 1945, 9 (5), 385−391. (16) Erkselius, S.; Wadso, L.; Karlsson, O. J. Drying rate variations of latex dispersions due to salt induced skin formation. J. Colloid Interface Sci. 2008, 317 (1), 83−95. (17) Arda, E.; Pekcan, O. Time and temperature dependence of void closure, healing and interdiffusion during latex film formation. Polymer 2001, 42 (17), 7419−7428. (18) Tari, O.; Pekcan, O. Drying Technol. 2008, 26, 101−107. (19) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.; Witten, T. A. Capillary flow as the cause of ring stains from dried liquid drops. Nature 1997, 389 (6653), 827−829. (20) Argon, A. S.; Cohen, R. E. Toughenability of polymers. Polymer 2003, 44 (19), 6013−6032. (21) Hobbs, S. Y. The effect of rubber particle-size on the impact properties of high-impact polystyrene (hips) blends. Polym. Eng. Sci. 1986, 26 (1), 74−81.

9389

dx.doi.org/10.1021/la501354k | Langmuir 2014, 30, 9384−9389