Article pubs.acs.org/Langmuir
Shear-Induced Reactive Gelation Bastian Brand, 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: In this work, we describe a method for the production of porous polymer materials in the form of particles characterized by narrow pore size distribution using the principle of shear-induced reactive gelation. Poly(styreneco-divinylbenzene) primary particles with diameter ranging from 80 to 200 nm are used as building blocks, which are assembled into fractal-like clusters when exposed to high shear rates generated in a microchannel. It was found that independent of the primary particle size, it is possible to modulate the internal structure of formed fractal-like aggregates having fractal dimension ranging from 2.4 to 2.7 by varying the residence time in the microchannel. Thermally induced postpolymerization was used to increase the mechanical resilience of such formed clusters. Primary particle interpenetration was observed by SEM and confirmed by light scattering resulting in an increase of fractal dimension. Nitrogen sorption measurements and mercury porosimetry confirmed formation of a porous material with surface area ranging from 20 to 40 m2/g characterized by porosity of 70% and narrow pore size distribution with an average diameter around 700 nm without the presence of any micropores. The strong perfusive character of the synthesized material was confirmed by the existence of a plateau of the height equivalent to a theoretical plate measured at high reduced velocities using a chromatographic column packed with the synthesized microclusters.
■
INTRODUCTION Porous polymers are used for a variety of different applications ranging from catalysis,1,2 separation processes,3,4 thermal insulators,5 to scaffolding in medical tissue engineering.6 Each application demands different properties and different manufacturing procedures depending on the amount and morphology of the material to be produced. When considering porous materials based on polymers, these are commonly prepared using pore-generating systems, usually in the form of a porogen.1,5,7−16 This is commonly a good solvent for the monomer but not for the polymer resulting in its segregation from the reacting monomer phase along the course of the polymerization reaction. This means that a dispersed phase of polymer forms during polymerization through a process referred to as nucleation. Given enough monomer, these nuclei grow and connect to each other, eventually forming a continuous polymer phase. At the end of the reaction, the porogen is extracted from the pores using an appropriate solvent.1,7−11 By proper choice of porogen and reaction conditions, this method allows for the tuning of the pore size as well as final material morphology. Furthermore, the initial system of porogen and monomer can be dispersed in a noncompatible continuous phase to form droplets that later turn into particles.10,11,17 It appears that in this process, a number of phenomena occur at once and their complex interaction is not fully understood.10,11,17,18 More control over the process could be achieved by separating the process steps in time applying a method referred to as reactive gelation.19−24 Using this technique, macroporous polymers can be prepared in a very controlled, stepwise manner © 2015 American Chemical Society
that mimics the different steps of the porogen method to some extent and only produces brine as waste-product. Both monoliths and particulate porous materials have been prepared and functionalized in a number of ways to obtain chromatographic media; other applications have not been explored yet. Starting from a spent polymer latex, the polymer particles are swollen by a mixture containing monomer and initiator, aggregated, and finally hardened by postpolymerization through heating. The last two steps can be carried out in stagnant conditions (thus preparing monoliths), or with agitation where aggregation and breakage by shear yield particles on the micrometer scale. The obtained porous materials are characterized by rather different pore size distributions (PSD). In particular, under stagnant conditions using partially destabilized particles, that is, stagnant reaction limited cluster− cluster aggregation (RLCA),25 the PSD is rather narrow, while in the presence of shear and using fully destabilized primary particles (i.e., shear diffusion limited cluster−cluster aggregation (DLCA)),26 a rather broad PSD covering approximately 2 orders of magnitude is generated.22,23 Production condition, in particular the presence or absence of breakup, has also an impact on the cluster internal structure. While in the case of stagnant RLCA conditions fractal dimension (df) is equal to 1.9, indicating rather open internal structure, in the shear DLCA conditions df is around 2.7, thus leading to rather compact clusters.27,28 Received: July 29, 2015 Revised: October 17, 2015 Published: October 21, 2015 12727
DOI: 10.1021/acs.langmuir.5b02805 Langmuir 2015, 31, 12727−12735
Article
Langmuir
ZS 3600 (Malvern Instruments, Malvern, Worcestershire, UK)). The reaction was stopped when reaching the desired core size. Such particles are subsequently used as a seed in the second step, a seeded emulsion polymerization. There, a 1% cross-linked shell was prepared around the core in a semibatch mode with slow monomer feed to achieve a radially more homogeneous cross-linkage. This shell preparation is carried out following the same steps above, with the previously prepared seed latex in the reactor’s initial charge, as described in Tables SI1−3 (shell). Aggregate Preparation. Depending on the experiment, the latex was taken either plain or swollen beforehand. If not stated otherwise, the latex was not swollen. The swelling was carried out by preparing a mixture of hydrophobic monomers Sty and DVB and hydrophobic initiator AIBN and subsequently adding it dropwise into the latex. The suspension was then agitated for at least 4 h. Aggregation of the latex was carried out using a high-shear device HC-2000 from Microfluidics (Newton, MA) equipped with a L30Z microchannel with the rectangular cross section of 5.26 × 10−8 m2 and a length of 5.8 mm. The sketch of the whole apparatus is shown in Figure 1, from hereon called “microchannel”.
Because the static RLCA method allows one to produce porous material with narrow PSD in the form of monolith, while the shear DLCA results in porous particles in broad PSD, the goal of the presented work is to develop a method combining the advantages of both of the above-mentioned, resulting in a particulate porous material with narrow PSD and good mechanical stability. Building on the previous work of our group,29,30 to induce aggregation of primary particles even in the presence of a repulsive barrier will be carried out using high shear, which is thus referred to as shear induced RLCA. Such condition represents a compromise between previously used conditions, which could have an impact on the final properties of the formed porous material, that is, narrow PSD and high fractal dimension. An additional advantage of using partially stabilized polymer dispersions is that the final postpolymerization step, used to enhance the mechanical stability of the formed clusters, can be carried out at higher solid fraction than in the original stirred tank process based on fully destabilized particles22,23 without the risk of the formation of large blocks of polymer. The characterization of the formed clusters, herein termed “microclusters”, is carried out using a combination of various techniques, that is, static light scattering, nitrogen sorption, mercury porosimetry, and scanning electron microscopy. Because previous attempts to produce material using shear DLCA conditions indicate that porous particles are characterize by rather large pores with perfusive characteristics,31 the obtained microclusters are packed into a HPLC column to evaluate their chromatographic properties using tracer injection experiments.
■
EXPERIMENTAL SECTION
Materials. The following chemicals have been employed in this work: 2−2′-azo(2-methylpropionitrile) (AIBN, Fluka, purum), divinylbenzene (DVB, Aldrich, 80% technical), potassium persulfate (KPS, Fluka, puriss p.a.), magnesium chloride (VWR, 99.9%), sodium chloride (Merck, for analysis), sodium dodecyl sulfate (SDS, Fluka, ≥98%), styrene (Sty, Fisher Scientific, general purpose grade), sodium phosphate monobasic (Fluka, purum p.a. anhydrous, ≥99.0%), and sodium phosphate dibasic (Fluka, purum p.a. anhydrous, ≥98.0%). All chemicals have been used as supplied without further purification. Ultrapure grade water for chromatography has been prepared by Millipore Synergy (Millipore, Billerica, MA). Deionized water for synthesis has been stripped of oxygen by degassing under vacuum and subsequent saturation with nitrogen. Primary Particle Preparation. Three latexes with primary particles diameters equal to 80 (L1), 125 (L2), and 200 nm (L3) were produced in two steps to investigate the effect of the primary particles size on the aggregation and gelation process (see Tables SI1− 3). In the first step, 20% cross-linked core particles were produced by semibatch emulsion polymerization under nitrogen atmosphere. A 4 L Mettler-Toledo LabMax was initially charged with water and surfactant (SDS) according to the recipe reported in Tables SI1−3 (initial charge 1, IC1). The temperature was set to 70 °C using the oil heating jacket. In a second flask, an emulsion of styrene, divinylbenzene, water, and surfactant (SDS) was prepared according to Tables SI1−3 (continuous feed 1, CF1) and kept emulsified using a magnetic stirrer. When the reactor temperature reached 70 °C, aqueous initiator (KPS) solution was injected through a septum into the reactor using a syringe and hypodermic needle according to Tables SI1−3 (initiator solution 1, IS1). The monomer/water emulsion CF1 was fed over the time specified in Tables SI1−3 using the Labmax membrane pump. In cases with reaction times significantly longer than the half-life time of KPS, initiator solution was fed using a syringe pump (IF1 in Tables SI1−3). The reaction progress was monitored with thermogravimetric dry content analysis (HG53 Halogen Moisture Analyzer (Mettler Toledo, Greifensee, Switzerland)) and dynamic light scattering (Zetasizer nano
Figure 1. Scheme of the microchannel equipment used. The pump applies a maximum pressure of 160 bar resulting in an average shear rate equal to 5.76 × 106 1/s. According to the manufacturer specification, the generated shear rate values can be correlated to the applied pressure drop with γ = 2.27 × 105·ΔP0.64, covering a range from 3.70 × 106 to 5.76 × 106 1/s corresponding to ΔP values from 80 to 160 bar. The corresponding flow rates and residence times of the latex inside the microchannel are summarized in Table 1.
Table 1. Hydrodynamic Conditions Used during the Aggregation Runs ΔP (bar)
Q (mL/min)
residence time (ms)
shear rate (γ) (1/s)
80 120 160
748 918 1059
1.15 0.94 0.81
3.70 × 106 4.80 × 106 5.76 × 106
Prior to the aggregation experiment, the latex was filtered through a cellulose filter to remove aggregates that form when the latex dries on the storage bottle wall and neck. It was then added to a well-stirred salt solution or water, depending on the targeted salt concentration. The suspension is then quickly transferred into the microchannel’s reservoir and finally passed through the microchannel. Between experiments, the microchannel was filled with water, so the first half of the leaving product was discarded and the sample taken from the middle of the latex plug obtained from steady-state operation was used for further characterization. In the case when several passes through the microchannel have been performed, the second half of the stream leaving the microchannel was collected and used as a starting material 12728
DOI: 10.1021/acs.langmuir.5b02805 Langmuir 2015, 31, 12727−12735
Article
Langmuir
The hardened aggregates allow for mechanically more demanding tests, so the Brunauer−Emmett−Teller (BET)36 equation was used to estimate the total surface area from nitrogen sorption measurements. Furthermore, mercury intrusion porosimetry37 was used to measure porosity and pore size distribution of the produced clusters. For visual inspection of the formed clusters, SEM pictures were taken using a Gemini 1530 FEG (Zeiss, Oberkochen, Germany). The dependency of the height equivalent to a theoretical plate (HETP) on the flow rate was measured by injecting pulses of dextrans at different flow rates into a 25 mM, pH 7 phosphate buffer using an Agilent Series 1200 (Agilent Technologies, Santa Clara, CA) equipped with a quaternary pump and degasser, an autosampler with integrated cooling, a refractive index detector, and a diode array detector. HETP values were calculated from the refractive index peak’s first and second moments.3,4
for the consequent passage. The swollen latex was aggregated in an identical way as described above and subsequently postpolymerized overnight at 70 °C in a 4 L Mettler-Toledo LabMax with diameter of 20 cm equipped with anchor impeller (diameter equal to 7.5 cm and height of 11 cm) without baffles. Considering the stirring speed of the impeller, which was 120 rpm, the estimated average shear rate applied to the latex in this operation was around 50 1/s, which is approximately 5 orders of magnitude lower than that in the microchannel. Sample Characterization. Two samples from the aggregate plug’s center were collected from the microchannel exit and stored in a pill flask. One sample was transferred to the rheometer (equipped with cone (stator)/plate (rotor) geometry), and the viscosity was measured at shear rates of 15.85 1/s. The second sample was diluted as a whole with deionized water until the obscuration was within the limits of the Mastersizer 2000 (Malvern Instruments, Malvern, Worcestershire, UK) used for cluster size and internal structure analysis. This ensured representative sampling where neither large nor small particles are preferred. The diluted sample was then slowly passed through the Mastersizer 2000 to prevent sedimentation using a syringe with a large outlet opening (5.5 mm) at 1 mL/min employing the syringe pump Lambda VitFit (LAMBDA Laboratory Instruments, CH). In all steps of the process, utmost care was taken to avoid large shear rates that could break the aggregates. Measured intensity of the scattered light was used to evaluate the particle size of large aggregates as well as their (fractal) internal morphology. The average structure factor ⟨S(q)⟩ of the produced aggregates was evaluated from the scattered light intensity I(q) according to32,33 ⟨S(q)⟩ =
I(0)P(q) I(q)
■
RESULTS AND DISCUSSION Determination of a Gel Zone. All latexes used in this work are stabilized by the presence of electrostatic repulsive forces originating from the added surfactant (SDS), which prevents primary particles from aggregating. To induce their aggregation, the repulsive barrier between primary particles has to be removed, that is, by properly increasing ionic strength, commonly referred to as diffusion limited cluster−cluster aggregation (DLCA).26,38 Alternatively, one can use sufficiently high shear rate to bring the primary particles so close that van der Waals attraction forces would overcome the repulsive electrostatic energy barrier, thus leading to aggregation, that is, shear induced reaction limited cluster−cluster aggregation (RLCA).29 In the first scenario, primary particles become fully destabilized, and due to their small size they start to aggregate following Brownian aggregation mechanism.39,40 In the second case, the aggregation process is a function of the energy barrier between primary particles (i.e., amount and type of used surfactant, and ionic strength of the surrounding solution), and the kinetic energy involved in their collision, which can be expressed in the form of a Peclet number, Pe = 3πμγR3p/kBT,29 with Rp as the primary particle size, kB as the Boltzmann constant, and T as temperature. In addition, the extent of aggregation is affected by the number of collisions between primary particles, which depends on the residence time inside the microchannel and the primary particles’ concentration. Accordingly, we can construct a demarcation curve separating the region where aggregation occurs from the region of no aggregation using ionic strength and the amount of energy added into the system (in our case being directly proportional to the applied shear rate) as independent operating conditions. Even though both of these parameters can trigger the aggregation process, there is a fundamental difference in the formed cluster mass distribution. Under DLCA conditions, electrostatic repulsion is completely suppressed, and each collision leads to an aggregation event. We therefore have aggregation between primary particles forming doublets and smaller oligomers, which aggregate further and eventually result in the formation of larger clusters. Because the highest aggregation rate is obtained for unequal aggregates, this results in a rather narrow cluster mass distribution (CMD).39,41 In the case of shear induced RLCA, there exists electrostatic repulsion between primary particles, and due to the action of shear the highest aggregation rate is obtained for collision of the largest aggregates.29 This, at least at the beginning of the process, results in the formation of a bimodal distribution where very large aggregates exist together with a large amount
(1)
where P(q) is the form factor of primary particles and q is the scattering wave vector defined as
q = 4π
n ⎛⎜ θ ⎞⎟ sin λ ⎝2⎠
(2)
Here, n refers to the refractive index of the continuous phase (in all cases water), λ is the laser wavelength, and θ is the scattering angle. The Guinier approximation34 was used to relate the structure factor to the radius of gyration:
⎛ q2⟨R 2⟩ ⎞ g S(q) ⎟ , for q⟨R g⟩S(q) < 1 ⟨S(q)⟩ = exp⎜⎜ ⎟ 3 ⎝ ⎠ ⟨R2g⟩
⟨R2g⟩S(q)
⟨R2g,p⟩
(3)
2
= + and Rg,p = (3/5) Rp being the radius of with gyration of the primary particles. It is worth noting that most aggregates studied in this work were several tens of micrometers in diameter, making the contribution of primary particles rather insignificant. Because the used systems obey the Rayleigh−Debye−Gans theory,32 we can extract from the light scattering measurements additional information about the micro aggregate internal structure, that is, their fractal dimension. In particular, for the monodisperse population of aggregates, when plotting the average structure factor ⟨S(q)⟩ as a function of q according to34
⟨S(q)⟩ ∝ q−df , for
1 1
