Article pubs.acs.org/Langmuir
Synthesis of Macroporous Polymer Particles Using Reactive Gelation under Shear Alexandros Lamprou, Itır Köse, Giuseppe Storti, Massimo Morbidelli, and Miroslav Soos* Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, 8093 Zurich, Switzerland S Supporting Information *
ABSTRACT: By combining elements from colloidal and polymer reaction engineering a new approach toward macroporous, mechanically robust polymer particles is presented, which does not require any porogenic additives. Specifically, aggregation and breakage in turbulent conditions of aggregates originating from fully destabilized primary latex particles is applied to produce compact, micrometer-sized clusters. Post-polymerization of monomer introduced initially to swell the primary particles is imparting mechanical rigidity and permanence to the internal structure. The resulting microclusters exhibit an internal porosity on the order of 70% and relatively broad pore size distribution, with exceptionally large pores, ranging from about 50 nm to 10 μm in diameter. These particulate microclusters, produced via reactive gelation under shear, are fractal objects with fractal dimension around 2.7, as opposed to the more open fractal structure of a monolith produced via stagnant reactive gelation, with fractal dimension of 1.9. Such macroporous particles are thought to be useful in applications requiring pores on the micrometer scale, e.g., in the chromatography of biomolecules or for packing beds perfusive to convective flow.
■
INTRODUCTION The synthesis of macroporous copolymer resins has been extensively investigated over half a century, for both fundamental and applied purposes.1,2 Applications of porous resins include, among others, solid-phase synthesis,3 extraction, ion exchange, catalysis,3 pollutant adsorption,4 chromatography,5−7 and insulators.1 Irrespectively of how sophisticated the synthetic procedure is, a pore-generating system is typically employed in the production of macroporous polymer resins.1 This may be a traditional inert porogen for generic applications, or a specifically interacting template for specific applications, e.g., affinity matrixes.1 In the first case, a good or a poor solvent for the incipient polymer network, a nonreactive linear polymer, or a combination of the three can be applied.2 While initially the porogenic mixture forms a single phase with the monomer and the cross-linker, at some point during the course of polymerization phase separation of the growing polymer network occurs, which might be due either to extensive cross-linking or simply to porogen−polymer matrix incompatibility. The polymeric nuclei thus formed organize themselves into microspheres, which subsequently come together into larger agglomerates that constitute the main building blocks of the final polymer network. The resulting pores are ultimately filled with the porogen, which is eventually extracted with an appropriate solvent or series of solvents at the end of the polymerization.2,3,8−11 The resulting pore morphology, pore size distribution, and specific surface area depend on the © 2014 American Chemical Society
amount and composition of the porogenic mixture, cross-linker, temperature, and time of polymerization, as well as whether the polymerization is run in a dispersed system or in the bulk.2,11,12 Obviously dispersed systems yield particles (typically micrometer-sized), while bulk systems yield continuous monoliths. Despite the growing scientific attention to monoliths, their commercial application remains limited. For example, in the area of chromatography,13,14 there are limited examples of commercially available monolithic stationary phases, with particles being still the most widespread packing material.13 For the production of particulate resins several methods have been employed, ranging from simple surfactant-free dispersion polymerization7 to classical suspension polymerization15 or more elaborated procedures. Examples of the latter include the activated swelling of Ugelstad,16 the seeded emulsion polymerization of Vanderhoff, El-Aasser, et al.10,17 and the staged templated suspension polymerization by Frechet et al.6 Novel porogens including oligomers,18 ionic and steric surfactants,19 and others4,7 have also been proposed in the literature. Yet, in all cases the relation between the factors affecting the pore generation and properties of the final material is still not well understood and requires a good deal of empiricism.2,11,12,14 Reactive Gelation is an alternative, porogen-free method toward macroporous monoliths.20 It is based on the colloidal Received: February 24, 2014 Revised: May 15, 2014 Published: May 22, 2014 6946
dx.doi.org/10.1021/la5000793 | Langmuir 2014, 30, 6946−6953
Langmuir
Article
Figure 1. Scheme of the macroporous polymeric microclusters synthesis.
principle of reaction-limited cluster aggregation (RLCA),21 i.e., the aggregation of polymer particles in a stagnant suspension, induced by the addition of salt at a concentration below the suspension’s critical coagulation concentration (CCC). The product is a colloidal gel, that is a space-filling, percolating network of weakly interconnected particles with a fractal structure.22 To increase mechanical strength of the formed gel prior to gelation, the latex particles are swollen by a welldefined amount of a monomeric mixture containing a thermal radical initiator. After gel formation the temperature is raised, inducing polymerization of the swelling monomers and a further rearrangement of the network structure, along with a substantial increase of its mechanical strength, yielding a macroporous monolith. Such polystyrene20 as well as poly(methyl methacrylate) based23 materials have been used for HPLC applications. Thereby, no size exclusion during inverse size-exclusion chromatography (ISEC) experiments20 and attractive mass transport properties by Van Deemter analysis23 have been demonstrated. However, a complication associated with monoliths synthesized via Reactive Gelation is their significant shrinkage occurring during their preparation. The main reasons for this shrinkage are partial interpenetration of the swollen primary particles during their aggregation,24 combined with further compaction during post-polymerization, as an inherent effect of the vinyl groups’ polymerization.11,25 Consequently, accommodating a single monolithic column prepared by Reactive Gelation inside an appropriate HPLC housing requires a laborious and failure-prone procedure. Specifically, the monolith has to be first machined to the right dimensions in a specialized workshop and then sealed into a special fitting ring, which can be finally packed as a whole in an HPLC housing.23 In our attempts to overcome this problem we developed a porogen-free method toward rigid macroporous microparticles, which can be readily packed in conventional chromatographic columns using standard equipment. This new approach relies on the combination of aggregation and post-polymerization under proper shear conditions and, similarly to Reactive Gelation,20 is divided into four distinct, successive steps: latex preparation, latex swelling, latex aggregation, and breakage and post-polymerization. As in stagnant Reactive Gelation outlined above, primary particles are subjected to swelling treatment with a monomeric mixture prior to aggregation, while postpolymerization is employed to rigidify the initially fragile aggregates. The main differences to stagnant Reactive Gelation lie in the aggregation step, which is realized under shear, and in the primary particle destabilization, which is realized by completely screening electrostatic repulsions. To this end, salt is added at concentrations above the latex CCC and aggregation of the primary particles occurs under fully destabilized conditions, i.e., diffusion-limited cluster aggregation (DLCA).26 This aggregation leads to the formation of clusters, which grow in size until they start being affected, i.e., broken by the shear,27 hence the new method is named Reactive Gelation
under shear.28 Subsequently, the system evolves through multiple aggregation-breakage events, which eventually lead to steady-state conditions.29,30 A dispersion of particles in the micrometer size range is thus obtained, whose size is determined by this dynamic equilibrium between aggregation and breakage,30 referred to in the following as microclusters. It is well-accepted that colloidal aggregates, although exhibiting a disordered structure, are actually self-similar objects, obeying the fractal geometry. This means that independently of the length scale used during their characterization, their mass scales with their size following a certain power law, with a scaling exponent called the fractal dimension, df, which indicates how compact the internal aggregate structure is.22,31 Here, it should be noted that this is universally applicable, including both aggregation mechanisms discussed in the context of this work: the production of colloidal gels via RLCA under stagnant conditions,21 as well as that of clusters via DLCA under shear, whereby multiple aggregation and breakage events occur.27,29 In our case, the lower and upper limits of this length scale are obviously the finite sizes of the primary particles and of the generated microclusters, respectively. The size and fractal dimension of the synthesized microclusters are resolved by static light scattering (SLS). A combination of SEM, N2 sorption, and Hg intrusion porosimetry, as well as ISEC, is used to characterize their internal porous structure. As far as the final surface properties are concerned, functionalization of the surface of the pores is in principle possible following common practices with polymer materials.1,3,9,14 Specifically, having incorporated in the outer shell a comonomer which can initiate atom transfer radical polymerization (ATRP)32 from the clusters surface, functionalization with grafted polymer chains is possible at a later stage. All the above steps are schematically visualized in Figure 1.
■
EXPERIMENTAL SECTION
Materials. All chemicals were used as received. Styrene (St, Fluka, >99%), the cross-linker divinylbenzene (DVB, Fluka, ∼80%), and the functional ATRP initiator methyl α-chloroacrylate (MCA, Acros, 99%) as monomers, 2,2′-azobis(isobutyronitrile) (AIBN, Aldrich, >98%) as initiator, hexadecane (HD, ABCR, 99%) as hydrophobe, sodium dodecyl sulfate (SDS, Fluka, >96%) as surfactant, and magnesium chloride hexahydrate (MgCl2·6H2O, Fluka, >98%) as aggregation inducer. Deionized water was further treated by a Millipore Simpack2 purification device. Latex Preparation. In this study, miniemulsion polymerization was employed,33 because it provides better control over the particle size with respect to conventional emulsion polymerization, since droplet nucleation dominates and the latex formation is independent of polymerization kinetics. As miniemulsions are critically stabilized systems, droplet and consequently final polymer particle size depend directly on the surfactant amount with respect to monomer,33 while using an oil-soluble radical initiator averts the formation of ionic chain ends. Details for the primary particle synthesis are reported in Table 1. Using a semibatch protocol, a core is first prepared in a minimeulsion batch (“initial core batch” column in Table 1), followed by two starved 6947
dx.doi.org/10.1021/la5000793 | Langmuir 2014, 30, 6946−6953
Langmuir
Article
Table 1. Recipe for the Preparation of Primary Particlesa batch
init.
1st
2nd
St (g) DVB (g) MCA (g) HD (g) AIBN (g) H2O (g) SDStot (g) SDSpost(wt %) tsonic (min)
26.2 5.8 1.28 0.64 426 0.96 33 5
15.8 0.20 -
7.2 0.16 5.44 -
10%. The magnetic stirrer was then removed and the latex gelation was left to complete overnight. It is interesting to note that the selection of electrolyte type and concentration was based on an optimization, in order to obtain a single-piece, mechanically rigid monolith. Actually, MgCl2 was found to have a smoother effect on the latex gelation than CaCl2, presumably due to the larger hydration number of Mg2+.35 On the other hand, large amounts of monovalent cations (i.e., Li+, Na+, K+) were required to destabilize the latex, which could result in salt crystals inducing cavity formation.1 Moreover, the gel dry fraction would be smaller than the targeted 10% if the salt solution concentration was to be retained well below 1 M, so as to avoid improper local mixing effects. Post-Polymerization and Post-Treatment. Post-polymerization was conducted by keeping the stirred reactor (respectively the gel vial) at 65 °C for 16 h. Afterward the aqueous layer was decanted (in the case of microclusters following centrifugation at 4600 rpm for 15 min) and the resins were washed 6 times with water and dried slowly at ambient conditions. For subsequent analyses the microclusters were used as such, while the monolith was further ground using a universal mill (Ika M20) into micrometer-sized particles for SLS, SEM, and N2 sorption characterization, while a few millimeter-sized pieces were reserved for Hg porosimetry (v.i.). Analytical Tools. Gravimetric analysis of the latex was done at defined time instances along, as well as at the end of, the polymerization. To this end, 0.7 mL of the latex was withdrawn and cooled instantly in an ice bath to suppress polymerization and evaporation of the liquid components. The sample was then spread over quartz sand and heated rapidly at 120 °C in air, using a HG53 Mettler-Toledo Moisture Analyzer. This provided the dry mass fraction of the sample within 2−3 min, from which the corresponding monomer conversion was calculated. The average size of the latex particles (core, particles with the first shell, or second shell/primary particles) was determined by dynamic light scattering (DLS, Malvern Zetasizer Nano ZS). The same device was used to measure the electrophoretic mobility of primary particles and estimate the corresponding ζ potential through the Smoluchowski model.35 Furthermore, the MgCl2 CCC was experimentally determined for every latex by monitoring the aggregation kinetics under quiescent conditions. To capture the early onset of aggregation each experiment was carried out at solid volume fraction of 1.0 × 10−6 using DLS. These CCC measurements are further described in the SI. Since the clusters produced are in the micrometer range, their size and internal structure was characterized by SLS (Malvern Mastersizer 2000). Details on the evaluation of the mean square radius of gyration ⟨Rg⟩ and the fractal dimension df of the clusters can be found in the SI. Scanning electron microscopy (SEM, Zeiss Gemini 1530 with field emission gun operated at 1 kV) was conducted to assess the shape, surface morphology, and size of microclusters, as well as to estimate an order of magnitude of the pore sizes. The specific surface area was calculated from the Brunauer− Emmett−Teller equation (BET)36 by analyzing the initial slope of the nitrogen adsorption isotherm obtained through nitrogen sorption measurements (Micromeritics TriStar Series 320). Pore size distribution and corresponding pore volumes were assessed by combining the data obtained through nitrogen sorption and mercury intrusion porosimetry (MIP, Porosimeter 2000). The complete N2 adsorption−desorption cycle was evaluated, especially the desorption isotherm by applying the Barret-Joyner-Halenda (BJH) method36 for the pore size distribution of small pores (20 nm), using a Hg contact angle of 130°,17 under the assumption of cylindrical pores. All measurements were conducted according to standard operating protocols. Last, details on the treatment of data obtained by ISEC for calculating the total porosity of chromatographic columns packed with microclusters and the peak asymmetry factor can be found in the SI.
a
Init.: initial core batch, 1st: 1st starved feeding loop, 2nd: 2nd starved feeding loop, SDStot: total SDS amount applied, SDSpost: wt % of SDS aliquot applied for poststabilization dissolved in 16 mL H2O, tsonic: ultrasonication time. Procedure followed described in Latex Preparation section.
feeding loops (“1st and 2nd starved feeding loop” columns in Table 1) for growing the two respective outer shells. Thus, control of crosslinking degree along the particle radius and of surface functionality is also possible. The procedure was as follows: The oil phase was prepared by mixing styrene, DVB, HD, and AIBN; this was subsequently added to the water phase containing SDS; immediately after ultrasonication (90% amplitude, full cycle, using a Hielscher UP400S ultrasonicator) in an ice bath for a given time, a second aliquot of surfactant was added for poststabilization. 34 The miniemulsion thus produced was transferred inside a 3-neck 500 mL round flask, equipped with reflux condenser and magnetic stirrer, degassed and heated to 70 °C. When conversion reached ∼80%, the two monomeric mixtures corresponding to the two feeding loops were successively fed using a Lamda Vit-Fit programmable syringe pump operated at 0.056 mL/min, while heating was maintained overnight. Applied procedure results in latex with solid content of 12 wt % and primary particles with diameter equal to 161 nm with core diameter equal to 139 nm. Latex Swelling. Similarly to stagnant reactive gelation,20,23 a 15% wt. mixture of monomers (with respect to the latex dry fraction) containing 40% wt. St, 20% DVB, 40% MCA, and 1% AIBN was added dropwise over 20 min into a flat-bottom 30 mL vial containing typically 1.15 g of latex particles (dry basis), under vigorous stirring. The latex was kept under these conditions for another 60 min. Aggregation under Shear. For microcluster preparation the previously swollen latex was diluted down to 1% wt. with water inside a completely filled 150 mL jacketed reactor. To ensure good mixing the reactor was equipped with a four paddle impeller and two baffles (see Figure SI 1 in Supporting Information). Aggregation was induced under defined stirring speed, i.e., shear rate, by two successive gradient salt addition steps, using a Lamda Vit-Fit programmable syringe pump operated at 4.23 mL/min: first 10 mL of 0.5 M MgCl2, followed by 10 mL of 4.9 M MgCl2. A small opening on the reactor lid was used to allow liquid overflow and prevent air from entering the system. The amount of added salt was defined to reach a concentration ∼10 times above the latex CCC, in order to ensure complete destabilization, leading to rapid aggregation and incorporation of primary particles into clusters. To reach dynamic equilibrium between aggregation and breakage30 the system was stirred at constant rotation speed for 4 h. Stagnant Aggregation. For comparison purposes a monolith has also been prepared in this work by stagnant reactive gelation. The final salt concentration was kept well below CCC (13 mM) to ensure partial destabilization and therefore rather slow aggregation. In this case, in order to obtain a uniform macroscopic gel the mixing of salt solution with the highly concentrated latex has to be much faster than the subsequent gelation process.20 Therefore, a 7 mM MgCl2 solution was added dropwise under stirring during ∼30 s into a 15 mL flatbottom vial containing 100 mg of previously swollen latex particles (dry basis), in an amount such as to reach a final gel dry fraction of 6948
dx.doi.org/10.1021/la5000793 | Langmuir 2014, 30, 6946−6953
Langmuir
■
Article
RESULTS AND DISCUSSION Characterization of Microclusters. It is well-known that aggregates composed of fully destabilized rigid primary particles and held together only by van der Waals attractive forces are extremely fragile, and when exposed to different shear rates, i.e., stirring speeds, evolve in size until reaching distinctly different steady-state sizes.27,29 In this regard, aiming to assess the effect of post-polymerization on mechanical stability, the produced microclusters were exposed to two different stirring rates. In a first group of experiments, microclusters were prepared at 200 rpm from swollen primary particles without applying postpolymerization. A sample was withdrawn and diluted from 10−2 w/w to ∼10−5 w/w so as to ultimately suppress aggregation. Following SLS analysis, these aggregates were stirred inside an identical reactor at 800 rpm until a new steady state was reached and were then reanalyzed by SLS. In a second group of experiments the procedure was repeated according to the same protocol, with the only difference that post-polymerization was carried out after the aggregation at 200 rpm. A comparison of the obtained particle size distributions (PSD) by SLS for all four samples is presented in Figure 2. As
microclusters are shown in Figure 3a,b. At low magnification (see Figure 3a and Figure SI 3) the clusters appear to be rather
Figure 2. Effect of post-polymerization on the mechanical strength of formed microclusters. Particle size distribution of microclusters prepared at 200 rpm by aggregation/breakage and stirred at 200 rpm (solid line) and 800 rpm (dashed line) after post-polymerization and at 200 rpm (short dashed line) and 800 rpm (dash-dotted line) before post-polymerization.
Figure 3. Microclusters from nanoparticles of dp = 161 nm prepared at stirring rate of 200 rpm: (a,b) SEM with circles indicating rather large throughpores; (c) corresponding structure factor.
expected, there is substantial size reduction for the microclusters which did not undergo post-polymerization when the stirring speed is increased from 200 to 800 rpm (see short dashed line vs dash-dotted line in Figure 2). On the other hand, post-polymerization does in fact increase substantially the clusters’ strength, as clearly indicated by the comparable PSD obtained for both stirring speeds of 200 and 800 rpm (solid and dashed line in Figure 2). In other words, post-polymerization practically eliminates the dependence of the microclusters’ steady-state size on increasing stirring speed. A direct comparison of the absolute sizes and the broadness of the distributions before and after post-polymerization is, however, difficult (solid vs short-dashed line in Figure 2). This is because during the short time needed for sample withdrawal and dilution (
Synthesis of Macroporous Polymer Particles Using Reactive Gelation under Shear Alexandros Lamprou, Itır Köse, Giuseppe Storti, Massimo Morbidelli, and Miroslav Soos* Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, 8093 Zurich, Switzerland S Supporting Information *
ABSTRACT: By combining elements from colloidal and polymer reaction engineering a new approach toward macroporous, mechanically robust polymer particles is presented, which does not require any porogenic additives. Specifically, aggregation and breakage in turbulent conditions of aggregates originating from fully destabilized primary latex particles is applied to produce compact, micrometer-sized clusters. Post-polymerization of monomer introduced initially to swell the primary particles is imparting mechanical rigidity and permanence to the internal structure. The resulting microclusters exhibit an internal porosity on the order of 70% and relatively broad pore size distribution, with exceptionally large pores, ranging from about 50 nm to 10 μm in diameter. These particulate microclusters, produced via reactive gelation under shear, are fractal objects with fractal dimension around 2.7, as opposed to the more open fractal structure of a monolith produced via stagnant reactive gelation, with fractal dimension of 1.9. Such macroporous particles are thought to be useful in applications requiring pores on the micrometer scale, e.g., in the chromatography of biomolecules or for packing beds perfusive to convective flow.
■
INTRODUCTION The synthesis of macroporous copolymer resins has been extensively investigated over half a century, for both fundamental and applied purposes.1,2 Applications of porous resins include, among others, solid-phase synthesis,3 extraction, ion exchange, catalysis,3 pollutant adsorption,4 chromatography,5−7 and insulators.1 Irrespectively of how sophisticated the synthetic procedure is, a pore-generating system is typically employed in the production of macroporous polymer resins.1 This may be a traditional inert porogen for generic applications, or a specifically interacting template for specific applications, e.g., affinity matrixes.1 In the first case, a good or a poor solvent for the incipient polymer network, a nonreactive linear polymer, or a combination of the three can be applied.2 While initially the porogenic mixture forms a single phase with the monomer and the cross-linker, at some point during the course of polymerization phase separation of the growing polymer network occurs, which might be due either to extensive cross-linking or simply to porogen−polymer matrix incompatibility. The polymeric nuclei thus formed organize themselves into microspheres, which subsequently come together into larger agglomerates that constitute the main building blocks of the final polymer network. The resulting pores are ultimately filled with the porogen, which is eventually extracted with an appropriate solvent or series of solvents at the end of the polymerization.2,3,8−11 The resulting pore morphology, pore size distribution, and specific surface area depend on the © 2014 American Chemical Society
amount and composition of the porogenic mixture, cross-linker, temperature, and time of polymerization, as well as whether the polymerization is run in a dispersed system or in the bulk.2,11,12 Obviously dispersed systems yield particles (typically micrometer-sized), while bulk systems yield continuous monoliths. Despite the growing scientific attention to monoliths, their commercial application remains limited. For example, in the area of chromatography,13,14 there are limited examples of commercially available monolithic stationary phases, with particles being still the most widespread packing material.13 For the production of particulate resins several methods have been employed, ranging from simple surfactant-free dispersion polymerization7 to classical suspension polymerization15 or more elaborated procedures. Examples of the latter include the activated swelling of Ugelstad,16 the seeded emulsion polymerization of Vanderhoff, El-Aasser, et al.10,17 and the staged templated suspension polymerization by Frechet et al.6 Novel porogens including oligomers,18 ionic and steric surfactants,19 and others4,7 have also been proposed in the literature. Yet, in all cases the relation between the factors affecting the pore generation and properties of the final material is still not well understood and requires a good deal of empiricism.2,11,12,14 Reactive Gelation is an alternative, porogen-free method toward macroporous monoliths.20 It is based on the colloidal Received: February 24, 2014 Revised: May 15, 2014 Published: May 22, 2014 6946
dx.doi.org/10.1021/la5000793 | Langmuir 2014, 30, 6946−6953
Langmuir
Article
Figure 1. Scheme of the macroporous polymeric microclusters synthesis.
principle of reaction-limited cluster aggregation (RLCA),21 i.e., the aggregation of polymer particles in a stagnant suspension, induced by the addition of salt at a concentration below the suspension’s critical coagulation concentration (CCC). The product is a colloidal gel, that is a space-filling, percolating network of weakly interconnected particles with a fractal structure.22 To increase mechanical strength of the formed gel prior to gelation, the latex particles are swollen by a welldefined amount of a monomeric mixture containing a thermal radical initiator. After gel formation the temperature is raised, inducing polymerization of the swelling monomers and a further rearrangement of the network structure, along with a substantial increase of its mechanical strength, yielding a macroporous monolith. Such polystyrene20 as well as poly(methyl methacrylate) based23 materials have been used for HPLC applications. Thereby, no size exclusion during inverse size-exclusion chromatography (ISEC) experiments20 and attractive mass transport properties by Van Deemter analysis23 have been demonstrated. However, a complication associated with monoliths synthesized via Reactive Gelation is their significant shrinkage occurring during their preparation. The main reasons for this shrinkage are partial interpenetration of the swollen primary particles during their aggregation,24 combined with further compaction during post-polymerization, as an inherent effect of the vinyl groups’ polymerization.11,25 Consequently, accommodating a single monolithic column prepared by Reactive Gelation inside an appropriate HPLC housing requires a laborious and failure-prone procedure. Specifically, the monolith has to be first machined to the right dimensions in a specialized workshop and then sealed into a special fitting ring, which can be finally packed as a whole in an HPLC housing.23 In our attempts to overcome this problem we developed a porogen-free method toward rigid macroporous microparticles, which can be readily packed in conventional chromatographic columns using standard equipment. This new approach relies on the combination of aggregation and post-polymerization under proper shear conditions and, similarly to Reactive Gelation,20 is divided into four distinct, successive steps: latex preparation, latex swelling, latex aggregation, and breakage and post-polymerization. As in stagnant Reactive Gelation outlined above, primary particles are subjected to swelling treatment with a monomeric mixture prior to aggregation, while postpolymerization is employed to rigidify the initially fragile aggregates. The main differences to stagnant Reactive Gelation lie in the aggregation step, which is realized under shear, and in the primary particle destabilization, which is realized by completely screening electrostatic repulsions. To this end, salt is added at concentrations above the latex CCC and aggregation of the primary particles occurs under fully destabilized conditions, i.e., diffusion-limited cluster aggregation (DLCA).26 This aggregation leads to the formation of clusters, which grow in size until they start being affected, i.e., broken by the shear,27 hence the new method is named Reactive Gelation
under shear.28 Subsequently, the system evolves through multiple aggregation-breakage events, which eventually lead to steady-state conditions.29,30 A dispersion of particles in the micrometer size range is thus obtained, whose size is determined by this dynamic equilibrium between aggregation and breakage,30 referred to in the following as microclusters. It is well-accepted that colloidal aggregates, although exhibiting a disordered structure, are actually self-similar objects, obeying the fractal geometry. This means that independently of the length scale used during their characterization, their mass scales with their size following a certain power law, with a scaling exponent called the fractal dimension, df, which indicates how compact the internal aggregate structure is.22,31 Here, it should be noted that this is universally applicable, including both aggregation mechanisms discussed in the context of this work: the production of colloidal gels via RLCA under stagnant conditions,21 as well as that of clusters via DLCA under shear, whereby multiple aggregation and breakage events occur.27,29 In our case, the lower and upper limits of this length scale are obviously the finite sizes of the primary particles and of the generated microclusters, respectively. The size and fractal dimension of the synthesized microclusters are resolved by static light scattering (SLS). A combination of SEM, N2 sorption, and Hg intrusion porosimetry, as well as ISEC, is used to characterize their internal porous structure. As far as the final surface properties are concerned, functionalization of the surface of the pores is in principle possible following common practices with polymer materials.1,3,9,14 Specifically, having incorporated in the outer shell a comonomer which can initiate atom transfer radical polymerization (ATRP)32 from the clusters surface, functionalization with grafted polymer chains is possible at a later stage. All the above steps are schematically visualized in Figure 1.
■
EXPERIMENTAL SECTION
Materials. All chemicals were used as received. Styrene (St, Fluka, >99%), the cross-linker divinylbenzene (DVB, Fluka, ∼80%), and the functional ATRP initiator methyl α-chloroacrylate (MCA, Acros, 99%) as monomers, 2,2′-azobis(isobutyronitrile) (AIBN, Aldrich, >98%) as initiator, hexadecane (HD, ABCR, 99%) as hydrophobe, sodium dodecyl sulfate (SDS, Fluka, >96%) as surfactant, and magnesium chloride hexahydrate (MgCl2·6H2O, Fluka, >98%) as aggregation inducer. Deionized water was further treated by a Millipore Simpack2 purification device. Latex Preparation. In this study, miniemulsion polymerization was employed,33 because it provides better control over the particle size with respect to conventional emulsion polymerization, since droplet nucleation dominates and the latex formation is independent of polymerization kinetics. As miniemulsions are critically stabilized systems, droplet and consequently final polymer particle size depend directly on the surfactant amount with respect to monomer,33 while using an oil-soluble radical initiator averts the formation of ionic chain ends. Details for the primary particle synthesis are reported in Table 1. Using a semibatch protocol, a core is first prepared in a minimeulsion batch (“initial core batch” column in Table 1), followed by two starved 6947
dx.doi.org/10.1021/la5000793 | Langmuir 2014, 30, 6946−6953
Langmuir
Article
Table 1. Recipe for the Preparation of Primary Particlesa batch
init.
1st
2nd
St (g) DVB (g) MCA (g) HD (g) AIBN (g) H2O (g) SDStot (g) SDSpost(wt %) tsonic (min)
26.2 5.8 1.28 0.64 426 0.96 33 5
15.8 0.20 -
7.2 0.16 5.44 -
10%. The magnetic stirrer was then removed and the latex gelation was left to complete overnight. It is interesting to note that the selection of electrolyte type and concentration was based on an optimization, in order to obtain a single-piece, mechanically rigid monolith. Actually, MgCl2 was found to have a smoother effect on the latex gelation than CaCl2, presumably due to the larger hydration number of Mg2+.35 On the other hand, large amounts of monovalent cations (i.e., Li+, Na+, K+) were required to destabilize the latex, which could result in salt crystals inducing cavity formation.1 Moreover, the gel dry fraction would be smaller than the targeted 10% if the salt solution concentration was to be retained well below 1 M, so as to avoid improper local mixing effects. Post-Polymerization and Post-Treatment. Post-polymerization was conducted by keeping the stirred reactor (respectively the gel vial) at 65 °C for 16 h. Afterward the aqueous layer was decanted (in the case of microclusters following centrifugation at 4600 rpm for 15 min) and the resins were washed 6 times with water and dried slowly at ambient conditions. For subsequent analyses the microclusters were used as such, while the monolith was further ground using a universal mill (Ika M20) into micrometer-sized particles for SLS, SEM, and N2 sorption characterization, while a few millimeter-sized pieces were reserved for Hg porosimetry (v.i.). Analytical Tools. Gravimetric analysis of the latex was done at defined time instances along, as well as at the end of, the polymerization. To this end, 0.7 mL of the latex was withdrawn and cooled instantly in an ice bath to suppress polymerization and evaporation of the liquid components. The sample was then spread over quartz sand and heated rapidly at 120 °C in air, using a HG53 Mettler-Toledo Moisture Analyzer. This provided the dry mass fraction of the sample within 2−3 min, from which the corresponding monomer conversion was calculated. The average size of the latex particles (core, particles with the first shell, or second shell/primary particles) was determined by dynamic light scattering (DLS, Malvern Zetasizer Nano ZS). The same device was used to measure the electrophoretic mobility of primary particles and estimate the corresponding ζ potential through the Smoluchowski model.35 Furthermore, the MgCl2 CCC was experimentally determined for every latex by monitoring the aggregation kinetics under quiescent conditions. To capture the early onset of aggregation each experiment was carried out at solid volume fraction of 1.0 × 10−6 using DLS. These CCC measurements are further described in the SI. Since the clusters produced are in the micrometer range, their size and internal structure was characterized by SLS (Malvern Mastersizer 2000). Details on the evaluation of the mean square radius of gyration ⟨Rg⟩ and the fractal dimension df of the clusters can be found in the SI. Scanning electron microscopy (SEM, Zeiss Gemini 1530 with field emission gun operated at 1 kV) was conducted to assess the shape, surface morphology, and size of microclusters, as well as to estimate an order of magnitude of the pore sizes. The specific surface area was calculated from the Brunauer− Emmett−Teller equation (BET)36 by analyzing the initial slope of the nitrogen adsorption isotherm obtained through nitrogen sorption measurements (Micromeritics TriStar Series 320). Pore size distribution and corresponding pore volumes were assessed by combining the data obtained through nitrogen sorption and mercury intrusion porosimetry (MIP, Porosimeter 2000). The complete N2 adsorption−desorption cycle was evaluated, especially the desorption isotherm by applying the Barret-Joyner-Halenda (BJH) method36 for the pore size distribution of small pores (20 nm), using a Hg contact angle of 130°,17 under the assumption of cylindrical pores. All measurements were conducted according to standard operating protocols. Last, details on the treatment of data obtained by ISEC for calculating the total porosity of chromatographic columns packed with microclusters and the peak asymmetry factor can be found in the SI.
a
Init.: initial core batch, 1st: 1st starved feeding loop, 2nd: 2nd starved feeding loop, SDStot: total SDS amount applied, SDSpost: wt % of SDS aliquot applied for poststabilization dissolved in 16 mL H2O, tsonic: ultrasonication time. Procedure followed described in Latex Preparation section.
feeding loops (“1st and 2nd starved feeding loop” columns in Table 1) for growing the two respective outer shells. Thus, control of crosslinking degree along the particle radius and of surface functionality is also possible. The procedure was as follows: The oil phase was prepared by mixing styrene, DVB, HD, and AIBN; this was subsequently added to the water phase containing SDS; immediately after ultrasonication (90% amplitude, full cycle, using a Hielscher UP400S ultrasonicator) in an ice bath for a given time, a second aliquot of surfactant was added for poststabilization. 34 The miniemulsion thus produced was transferred inside a 3-neck 500 mL round flask, equipped with reflux condenser and magnetic stirrer, degassed and heated to 70 °C. When conversion reached ∼80%, the two monomeric mixtures corresponding to the two feeding loops were successively fed using a Lamda Vit-Fit programmable syringe pump operated at 0.056 mL/min, while heating was maintained overnight. Applied procedure results in latex with solid content of 12 wt % and primary particles with diameter equal to 161 nm with core diameter equal to 139 nm. Latex Swelling. Similarly to stagnant reactive gelation,20,23 a 15% wt. mixture of monomers (with respect to the latex dry fraction) containing 40% wt. St, 20% DVB, 40% MCA, and 1% AIBN was added dropwise over 20 min into a flat-bottom 30 mL vial containing typically 1.15 g of latex particles (dry basis), under vigorous stirring. The latex was kept under these conditions for another 60 min. Aggregation under Shear. For microcluster preparation the previously swollen latex was diluted down to 1% wt. with water inside a completely filled 150 mL jacketed reactor. To ensure good mixing the reactor was equipped with a four paddle impeller and two baffles (see Figure SI 1 in Supporting Information). Aggregation was induced under defined stirring speed, i.e., shear rate, by two successive gradient salt addition steps, using a Lamda Vit-Fit programmable syringe pump operated at 4.23 mL/min: first 10 mL of 0.5 M MgCl2, followed by 10 mL of 4.9 M MgCl2. A small opening on the reactor lid was used to allow liquid overflow and prevent air from entering the system. The amount of added salt was defined to reach a concentration ∼10 times above the latex CCC, in order to ensure complete destabilization, leading to rapid aggregation and incorporation of primary particles into clusters. To reach dynamic equilibrium between aggregation and breakage30 the system was stirred at constant rotation speed for 4 h. Stagnant Aggregation. For comparison purposes a monolith has also been prepared in this work by stagnant reactive gelation. The final salt concentration was kept well below CCC (13 mM) to ensure partial destabilization and therefore rather slow aggregation. In this case, in order to obtain a uniform macroscopic gel the mixing of salt solution with the highly concentrated latex has to be much faster than the subsequent gelation process.20 Therefore, a 7 mM MgCl2 solution was added dropwise under stirring during ∼30 s into a 15 mL flatbottom vial containing 100 mg of previously swollen latex particles (dry basis), in an amount such as to reach a final gel dry fraction of 6948
dx.doi.org/10.1021/la5000793 | Langmuir 2014, 30, 6946−6953
Langmuir
■
Article
RESULTS AND DISCUSSION Characterization of Microclusters. It is well-known that aggregates composed of fully destabilized rigid primary particles and held together only by van der Waals attractive forces are extremely fragile, and when exposed to different shear rates, i.e., stirring speeds, evolve in size until reaching distinctly different steady-state sizes.27,29 In this regard, aiming to assess the effect of post-polymerization on mechanical stability, the produced microclusters were exposed to two different stirring rates. In a first group of experiments, microclusters were prepared at 200 rpm from swollen primary particles without applying postpolymerization. A sample was withdrawn and diluted from 10−2 w/w to ∼10−5 w/w so as to ultimately suppress aggregation. Following SLS analysis, these aggregates were stirred inside an identical reactor at 800 rpm until a new steady state was reached and were then reanalyzed by SLS. In a second group of experiments the procedure was repeated according to the same protocol, with the only difference that post-polymerization was carried out after the aggregation at 200 rpm. A comparison of the obtained particle size distributions (PSD) by SLS for all four samples is presented in Figure 2. As
microclusters are shown in Figure 3a,b. At low magnification (see Figure 3a and Figure SI 3) the clusters appear to be rather
Figure 2. Effect of post-polymerization on the mechanical strength of formed microclusters. Particle size distribution of microclusters prepared at 200 rpm by aggregation/breakage and stirred at 200 rpm (solid line) and 800 rpm (dashed line) after post-polymerization and at 200 rpm (short dashed line) and 800 rpm (dash-dotted line) before post-polymerization.
Figure 3. Microclusters from nanoparticles of dp = 161 nm prepared at stirring rate of 200 rpm: (a,b) SEM with circles indicating rather large throughpores; (c) corresponding structure factor.
expected, there is substantial size reduction for the microclusters which did not undergo post-polymerization when the stirring speed is increased from 200 to 800 rpm (see short dashed line vs dash-dotted line in Figure 2). On the other hand, post-polymerization does in fact increase substantially the clusters’ strength, as clearly indicated by the comparable PSD obtained for both stirring speeds of 200 and 800 rpm (solid and dashed line in Figure 2). In other words, post-polymerization practically eliminates the dependence of the microclusters’ steady-state size on increasing stirring speed. A direct comparison of the absolute sizes and the broadness of the distributions before and after post-polymerization is, however, difficult (solid vs short-dashed line in Figure 2). This is because during the short time needed for sample withdrawal and dilution (
