Preparation and glass transition temperatures of elastomeric PolyHIPE materials Neil R. Cameron and David C. Sherrington* Department of Pure and Applied Chemistry, University of Strathclyde, T homas Graham Building, 295 Cathedral Street, Glasgow, UK G1 1XL Highly porous elastomeric PolyHIPE materials have been prepared by polymerisation of the continuous phase of high internal phase emulsions (HIPEs) containing styrene, divinylbenzene and varying amounts of either 2-ethylhexyl acrylate or the corresponding methacrylate monomer.The glass transition temperatures of the resulting copolymers were investigated by diVerential scanning calorimetry (DSC). For each series, Tg was found to vary non-linearly with copolymer composition. By consideration of the comonomer unit sequencing along the polymer backbone, as deduced from reactivity ratios, the relationship between Tg and composition in each case has been qualitatively explained. Comparison between experimental results and values calculated using an entropic approach have been made.It is thought that the (meth)acrylate comonomers reduce the overall Tg by three mechanisms: by introducing large amounts of free volume, due to their bulky side groups; by providing a much higher degree of flexibility to the polymer chains; and by reducing the proportion of adjacent styrene units, which display particularly unfavourable steric interactions. PolyHIPE1 polymers belong to a class of materials known as Investigations into the factors which aVect the cellular emulsion-derived foams.Their preparation involves the polym- structure and morphology of PS/DVB PolyHIPE polymers erisation of the continuous (external) phase of a high internal revealed that surfactant concentration plays a pivotal role,2 phase emulsion (HIPE), consisting of a large volume of particularly in determining whether an open- or closed-cell porogenic liquid dispersed in a relatively small volume of structure is formed.The volume fraction of the water droplets monomer(s) or monomeric solution. The volume fraction is less critical in this context. Recent cryo-SEM studies3 have of the dispersed phase must be at least 0.74, this being the shown that the appearance of holes between adjacent water maximum volume that can be occupied by uniform undeformed droplets corresponds to the gel-point of the polymerisation.spheres, and can be as high as 0.99. The HIPE is stabilised by The cell sizes of PS/DVB PolyHIPE polymers were found to a suYciently high concentration of a carefully chosen surfac- be strongly aVected by the concentration of electrolytes in the tant, such that inversion to the corresponding dilute emulsion aqueous phase,4 increasing salt concentration tending to lower is prevented.Following polymerisation, the dispersed phase is the cell-size. The presence of salts is known to enhance waterremoved yielding, generally, an open-cellular, low-density foam. in-oil emulsion stability via inhibition of the Ostwald ripening A considerable volume of literature exists on PolyHIPE process, due to the decreased miscibility of the oil and aqueous polymers, most of which deals with styrene/divinylbenzene phases.5 Additionally, electrolytes lower the interfacial tension (S/DVB) materials.In this case, an aqueous droplet phase is thus giving greater surfactant adsorption at the interface and dispersed in a mixture of monomers plus nonionic surfactant an increased resistance to droplet coalescence.6 (e.g.sorbitan monooleate). A water-soluble free-radical Open-cell PolyHIPE polymers are characterised by a low initiator, such as potassium persulfate, is used to facilitate bulk density, typically less than 0.1 g cm-3, and cell sizes range polymerisation. The open-cellular morphology of the resulting between 5 and 100 mm, depending on the preparation conmaterial can clearly be seen by scanning electron microscopy ditions, as mentioned above.Consequently, the internal surface (Fig. 1). Compared to conventional gas-blown polystyrene areas tend to be much lower than those observed in typical foams, the cells are more spherical and of smaller size. PS/DVB resins, 5 m2 g-1 being an average figure. However,7 using porogens in the monomeric phase allows PolyHIPE materials with surface areas up to 350 m2 g-1 to be prepared, in which the cell walls have pores similar in size and character to those observed in conventional macroporous polymer resins.8 PolyHIPE materials have recently been successfully employed as supports in solid phase peptide synthesis, in which the porous structure acts as a scaVold for a soft polyamide gel.9 Additionally, PolyHIPE monoliths have been used to immobilise flavin,10 and granulated PS/DVB PolyHIPE has been used as a catalyst support.11 The preparation, properties and applications of PolyHIPE polymers have recently been comprehensively reviewed.12 PS/DVB PolyHIPE polymers have similar overall mechanical properties to gas-blown PS foams, although the smaller size and increased spherical symmetry of the cells results in higher compressive strengths.2 They are, however, rather hard and brittle, due to the relatively high glass transition temperature (Tg) of polystyrene (100 °C).The mechanical properties of PolyHIPE polymers might be improved by introducing an Fig. 1 Scanning electron micrograph (SEM) of typical poly(styrene/DVB) PolyHIPE (bar�10 mm) elastomeric comonomer to reduce the overall Tg and thus lend J.Mater. Chem., 1997, 7(11), 2209–2212 2209some flexibility to the material. This comonomer must also be cellular morphology was confirmed by SEM (Fig. 2). The copolymer Tg values were determined by DSC; plots of Tg suYciently hydrophobic to form stable HIPEs, should copolymerise readily with styrene and DVB and preferentially should against composition are given in Figs. 3 and 4. Each Tg value was obtained from an average of two or three runs. be available commercially at low cost. Both 2-ethylhexyl acrylate (EHA) and methacrylate (EHMA) possess these requirements. The Tg values of the homopolymers are 223 and Theoretical treatments 263 K, respectively.13 Acrylates giving low Tg polymers are The Tg values of alkyl acrylate and alkyl methacrylate homooften used as comonomers to plasticise harder materials.14 polymers are largely governed by the nature of the alcoholic In this publication the preparation of PS/DVB PolyHIPE moiety of the ester; Tg decreases as the number of carbon materials incorporating increasing quantities of either EHA or atoms in the n-alkyl series increases.However, as the chain EHMA is described.In addition, the thermal properties of the resulting foams are investigated by means of diVerential scanning calorimetry (DSC). Experimental Materials Styrene (Fisons), divinylbenzene (55% DVB, tech. grade, Aldrich), (±)-2-ethylhexyl acrylate (Aldrich), (±)-2-ethylhexyl methacrylate (Aldrich), sorbitan monooleate (Span 80, Koch- Light), potassium persulfate (Fisons) and calcium chloride hexahydrate (Fisons) were used as received.Synthesis of poly[styrene/2-ethylhexyl (meth)acrylate/- DVB] PolyHIPE copolymers. Two series of materials were prepared, with varying contents of EHA or EHMA, relative to the total monomer content. A procedure for the preparation of a material containing 40% v/v EHMA is given below. Styrene (5 ml, 0.05 mol), DVB (1 ml), EHMA (4 ml, 0.02 mol) and Span 80 (2.0 g) were placed in a 300 ml wide-necked Fig. 2 Scanning electron micrograph (SEM) of poly(styrene/ DVB/EHA) PolyHIPE (vol%: S, 70; DVB, 10; EHA, 20. Internal polyethylene bottle, and were stirred with a glass stirring rod Phase vol%=90) (bar�25 mm) fitted with a D-shaped PTFE paddle, connected to an overhead stirrer motor, at 300 rpm.The aqueous solution, comprising potassium persulfate (0.2 g, 0.7 mmol) and calcium chloride hexahydrate (2.0 g, 9.1 mmol) in de-ionised water (90 ml ) was added dropwise, with constant stirring, to form the HIPE. As the aqueous phase was added, the bottle was lowered to maintain stirring just below the surface of the developing HIPE, ensuring that no water pockets formed.Once all the aqueous phase had been added, stirring was continued for a further 5 min, to produce as uniform aulsion as possible. The HIPE was then transferred to a reusable polymerisation mould. This consisted of a PVC cylinder, itself composed of two halves, which were screwed together, with separate screwon base and lid. When assembled, the mould had an internal height of 14 cm and an internal diameter of 4.5 cm.Prior to use, the inner surfaces were sprayed with PTFE spray to prevent the polymer from adhering. The HIPE was poured into the assembled mould, which was then immersed in a Fig. 3 Tg vs. EHA content of poly(styrene/DVB/EHA) PolyHIPE copolymers: (#) experimental values and (—) values predicted by water bath at 65 °C for 48 h, to form the PolyHIPE.The Barton’s model polymer monolith was retrieved by disassembling the vessel and was then extracted in a Soxhlet apparatus with water, for 24 h, to remove inorganic materials, followed by a lower alcohol, for a further 48 h. The bulk of the liquid was removed from the porous material in vacuo at room temp., and drying was completed in vacuo at 50 °C for 24 h. Instrumentation.The Tg values of the copolymers were determined by DSC using a DuPont 910 DiVerential Scanning Calorimeter fitted with a DuPont 9900 Computer/Thermal Analyser unit, under a steady flow of nitrogen and at a heating rate of 5 °C min-1. Monolithic PolyHIPE samples were ground to a powder and dried in a vacuum oven at 40 °C for 2 h prior to DSC analysis. Each DSC sample contained between 5 and 10 mg of polymer material.Results and Discussion Fig. 4 Tg vs. EHMA content of poly(styrene/DVB/EHMA) PolyHIPE Upon cleaning and drying, the EH(M)A PolyHIPE materials copolymers: (#) experimental values and (—) values predicted by Barton’s model appeared identical to their PS/DVB analogues. The open- 2210 J. Mater. Chem., 1997, 7(11), 2209–2212length increases beyond a certain value, crystallisation of the the only copolymer system examined by Barton which would not conform to his theory, which predicted a negative deviation side-chain occurs and Tg starts to increase once more. from linearity.Recent work20 with styrene/n-butyl methacrylate Branching of the alcohol chain restricts the rotation of the copolymers indicated that the Tg-composition relationship is a side-chain bonds so Tg is raised.Thus, PEHA has a Tg of simple curve of negative curvature. In this case, the experimen- 223 K compared to that of poly(n-hexyl acrylate) which is tal results were found to coincide with theoretical predictions. 216 K.13 Polymethacrylates are harder and therefore have The negative curvature predicted by Barton17 for styrene/n- higher Tg values than the corresponding acrylates, due to butyl acrylate copolymers, and observed by Schellenberg and increased restriction to rotation of the polymer backbone from Vogel20 in styrene/n-butyl methacrylate systems, can be the methyl group.For this reason, PEHMA has a Tg of 263 K, explained qualitatively by the reduction in steric hindrance higher than that of PEHA.between adjacent copolymer chains, due to the increase in free The glass transition temperatures of random copolymers volume on incorporation of comonomers with long n-butyl were initially described by a simple additive formula, known side chains. In comparison, styrene/methyl acrylate systems as the Fox equation [eqn. (1)],15 show a positive curvature in Tg-composition plots.The much smaller methoxy groups do not disturb the chain packing to 1 Tg = W1 T g1 + W2 T g2 +…+ Wi T gi (1) such a great extent, so steric hindrance to backbone motion is still encountered. The positive deviation has been attributed where Wi is the weight fraction of monomer i in the copolymer by Barton to the ability of methyl acrylate diads to form a and T gi is the glass transition temperature of its homopolymer.ring-type conformation: However, DiMarzio and Gibbs16 suggested that the Tg of a copolymer would be determined largely by the stiVness of the polymer chains, and arrived at the following relation (eqn. 2), naaa(Tg-T ga )+nbab(Tg-T gb )=0 (2) H H O O OMe OMe where ni is the mole fraction of repeat unit i, ai is the number of rotatable bonds in repeat unit i and T gi is the glass transition Experimental results temperature of the homopolymer of i. In reality, many copolymers were found to deviate from The Tg-copolymer composition plots for both the acrylatethese simple linear relationships.Barton17 modified the theory and methacrylate-containing copolymers are non-linear. The of DiMarzio and Gibbs to take account of the sequencing of transition temperatures of the EHA materials decrease steadily monomer units in a random copolymer. An AB-type copolymer until about 12% EHA content, followed by a plateau region has four possible pairs of repeat units, or diads, which can be up to 30%. After this, the Tg falls steadily towards that of expressed as aa, ab, ba or bb.The properties of an ab or ba PEHA (223 K).With the EHMA copolymers, however, there diad may indeed diVer from those of either an aa or bb is an initial dramatic decrease in Tg upon addition of only 5% sequence; the linear expression can therefore be extended to comonomer; subsequently, hardly any decrease is shown over give eqn. (3), almost the entire composition range, before a further sharp drop at the highest EHMA content.Barton’s copolymer Tg Tg=naa¾Tgab+nbb¾+(nab¾+nba¾)T gab (3) model17 was used to predict the Tg value of the PolyHIPE where n¾j is given by eqn. (4), copolymers, employing eqns. (3)–(5). For the EHA/S/DVB system, aaa=22, aab=14 and abb=6, whereas for the nij¾=nijaij/. ij (nijaij) (4) EHMA/S/DVB system, aaa=24, aab=15 and abb=6. Since the materials in the present study are all crosslinked and ~100% and nij is the molar fraction of ij sequences, aij is the number converted, the following assumptions have been made: (1) for of rotatable bonds in sequence ij, T gaa and T gbb are the Tg determining a values, DVB (and Et-styrene isomers) units have values of the respective homopolymers and T gab is the Tg of been ignored; (2) the concentrations of styrene and DVB in an exactly alternating 151 copolymer of a and b.This can be the comonomer feed are combined to give a value for styrenic rearranged to eqn. (5). monomers, designated species b in this case; (3) the mole Tg-naa¾T gaa-nbb¾T gbb=(nab¾+nba¾)T gab (5) fraction of each monomer incorporated in the copolymer, na, is equal to the mole fraction of monomer in the feed, nA.T gab can then be found from the slope of the plot of the left- Tg values for the 151 alternating copolymers (T gab) were hand side of eqn. (5) against (n¾ab+n¾ba). The sequencing of diad calculated using the linearised Barton eqn. (5), as described in units is controlled by the reactivity ratios of each monomer in the previous section. This gave T gab=282 K for EHA copolythe copolymer feed, therefore the Tg values of the copolymers mers and 310 K for the EHMA materials.These parameters are also aVected. were then used in eqn. (3) to calculate the theoretical Tg at Similarly, the Fox equation was modified by Johnston and each feed ratio. The results of the model predictions are shown co-workers18 to include the influence of monomer unit sequenwith the experimental data in Figs. 3 and 4. cing on copolymer Tg [eqn. (6)]. Both the experimental and model results indicate that the introduction of only 5% (meth)acrylate comonomer causes a 1 Tg = WaPaa T gaa + (WaPab+WbPba) T gab + WbPbb T gbb (6) considerable decrease in copolymer Tg. This could be due to an increase in free volume caused by the bulky ethylhexyl side Wi is the weight fraction of i in the copolymer, Pij is the chains, reducing the temperature at which the onset of local probability of unit ij occurring in the copolymer chain and segmental motion occurs.In addition, the introduction of some T gij is the contribution to the overall Tg of unit ij. These (meth)acrylate units into the backbone reduces the number of expressions, or modifications thereof, have been shown to be styrene–styrene diads, which experience unfavourable steric applicable to a wide range of copolymer systems.interactions between adjacent phenyl groups, thus also lower- Rather surprisingly, Tg values of styrene/2-ethylhexyl ing the Tg. However, further increases in EH(M)A content (meth)acrylate linear copolymers could not be found in the causes only a slight decrease in the Tg since the packing of the literature.However, the relatively similar styrene/n-butyl acry- chains is already severely disrupted. late system has been studied. Illers19 observed a non-linear S- After approximately 30% EH(M)A monomer content, shaped curve in the graph of Tg against composition, with an diVerences start to appear between the curves of acrylate and methacrylate copolymers. From Fig. 3, it can be seen that the inflection point at around 60% styrene. Interestingly, this was J. Mater. Chem., 1997, 7(11), 2209–2212 2211rate of decrease of Tg with increasing EHA content increases may explain why the Tg of the most lightly crosslinked sample at high EHMA content coincides with the calculated value. after a mole fraction of 30%.This can be explained as follows. The distances between adjacent chains do not increase greatly; however, the chance of bb diads of EHA occurring increases. Conclusions The reactivity ratios of styrene (M1) and EHA (M2) are The preparation of elastomeric PolyHIPE polymers from high reported as being r1=0.91 and r2=0.29.13 Therefore, at low internal phase emulsions containing styrene, DVB and 2- EHA content, it is highly probable that no EHA diads exist, ethylhexyl (meth)acrylate has been described. The glass trans- only ab styrene-EHA sequences.As the EHA concentration ition temperatures of the resulting materials, as determined by increases, EHA diads become increasingly more probable. DSC, were found to vary in a non-linear fashion with composi- Since EHA monomer repeat units possess many more rotatable tion.It is proposed that the (meth)acrylate comonomer units bonds than styrene units, the introduction of EHA diads would influence Tg in three ways: first, they introduce significant result in a large enhancement of the freedom of rotation and amounts of free volume due to their bulky side-groups; second, motion of the polymer chain.Thus, the Tg starts to decrease the high flexibility of the side-groups results in increased more quickly. segmental motion; and third, introduction of EH(M)A units The Tg values of the methacrylate copolymers (Fig. 4) do dilutes the concentration of sterically unfavourable styrene– not drop rapidly after 30% EHMA content, but remain more styrene diads. All of these eVects conspire to lower the Tg or less constant.This reflects the hindrance to rotation of the values further than is predicted by the Fox equation. The polymer backbone in the vicinity of EHMA units, caused by shapes of the individual curves have been explained qualitat- the a-methyl groups. The reactivity ratios of styrene and ively by considering the comonomer unit diad sequencing, EHMA could not be located in the literature.However, for which can be predicted from the corresponding reactivity styrene (M1) and n-hexyl methacrylate (M2) the corresponding ratios. Comparisons with Tg values calculated using Barton’s values are: r1=0.45 and r2=0.65,13 i.e. each polymer radical model indicate that the overall Tg-composition trends are type has approximately equal reactivity.Assuming styrene and reproduced in each case, despite the lack of exact correlations. EHMA have similar r values to those above, at low EHMA The complexity of the systems studied here may preclude the content bb EHMA diads are quite possible. Both intermolecuaccurate use of any predictive model. lar (chain spacing) and intramolecular (rotation of groups in the backbone) eVects therefore occur even at low EHMA NRC gratefully acknowledges the (then) SERC for the procontent, so initially there is a large drop in Tg (Fig. 4).vision of a CASE studentship and Unilever Research, Port Subsequently, interchain spacing does not drastically increase, Sunlight, the Cooperating Body. The discussions with Dr D. while the content of EHMA diads grows steadily, resulting in Gregory are particularly appreciated.a gradual drop in Tg with increasing EHMA content. Comparison of the experimental and model results indicates References that, in this case, Barton’s theoretical treatment is only partially successful at predicting copolymer Tg. However, a number of 1 D. Barby and Z. Haq, Eur. Pat. 0,060,138 (1982) (to Unilever). observations can be made. First, the trends shown by the 2 J.M.Williams and D. A. Wrobleski, L angmuir, 1988, 4, 656. experimental results are generally reproduced, that is to say 3 N. R. Cameron and D. C. Sherrington, Colloid Polym. Sci., 1996, 274, 592. that the EHA copolymer series is expected to have an S- 4 J. M. Williams, A. J. Gray and M. H. Wilkerson, L angmuir, 1990, shaped Tg-copolymer composition relationship, whereas the 6, 437.EHMA series shows simple negative curvature. Second, the 5 J. Kizling and B. Kronberg, Colloids Surf., 1990, 50, 131. model predicts a sharper initial drop in Tg for the methacrylate 6 M. P. Aronson and M. F. Petko, J. Colloid Interface Sci., 1993, series, which is also observed experimentally. Fig. 3 indicates 159, 134. 7 P. Hainey, I. M. Huxham, B. Rowatt, D. C. Sherrington and that the experimental Tg values of the EHA copolymers above L. Tetley,Macromolecules, 1991, 24, 117. nA=0.4 fall below the model curve. This may be due to 8 A. Guyot, Synthesis and Structure of Polymer Supports, in Syntheses plasticisation by residual surfactant, which is notoriously and Separations using Functional Polymers, ed.D. C. Sherrington diYcult to remove from PolyHIPE materials.21 This could and P. Hodge, Wiley, New York, 1988, ch. 1. counteract any raising of Tg caused by crosslinking, which is 9 P.W. Small and D. C. Sherrington, J. Chem. Soc., Chem. Commun., probably the reason for the underestimation of the Tg results 1989, 1589. 10 H. F. M. Schoo, G. Challa, B. Rowatt and D. C. Sherrington, by the model in Fig. 4. React. Polym., 1992, 16, 125. Crosslinking aVects the glass transition temperatures of 11 E. Ruckenstein and L. Hong, Chem. Mater., 1992, 4, 122. copolymers by two independent mechanisms:22,23 (1) the 12 N. R. Cameron and D. C. Sherrington, Adv. Polym. Sci., 1996, crosslinking eVect, which invariably raises Tg, and (2) the 126, 163. copolymer eVect, due to the chemical nature of the crosslinking 13 Polymer Handbook, ed.J. Brandrup and E. H. Immergut, 2nd edn., Wiley Interscience, USA, 1975. species. The latter can either raise or lower Tg. Since DVB is 14 B. B. Kine and R. W. Novak, Acrylic and Methacrylic Ester almost identical chemically to styrene, the copolymer eVect Polymers, in Encyclopedia of Polymer Science and Engineering can largely be neglected,24 especially for the low crosslink Vol. 1, Wiley-Interscience, USA, 1985. density of the materials in the present study. Recently, it has 15 T. G. Fox, Bull. Am. Phys. Soc., 1956, 1, 123. been reported that approximately 5% of crosslinked PS/DVB 16 E. A. DiMarzio and J. H. Gibbs, J. Polym. Sci., 1959, 40, 121. resins have Tg values about 25 K higher than linear poly- 17 J. M. Barton, J. Polym. Sci., Polym. Symp., 1970, 30, 573. 18 N. W. Johnston, J. Macromol. Sci., Rev. Macromol. Chem., 1976, styrene.25 In the same publication, however, model calculations C14, 215 and references therein. indicate that, for a given crosslink density, the increase in Tg 19 K. H. Illers, Kolloid-Z., 1963, 190, 16. caused by crosslinking is reduced as the flexibility of the 20 J. Schellenberg and J. Vogel, J. Polym. Sci. B, Polym. Phys., 1995, polymer backbone increases. For the copolymers investigated 32, 1969. here, this implies that increasing the concentration of the more 21 D. P. Gregory, Unilever Research, Port Sunlight, personal flexible component lessens the crosslinking eVect. This is the communication. 22 S. Loshaek, J. Polym. Sci., 1955, 15, 391. opposite to the trend shown in Fig. 4, which may reflect the 23 L. E. Nielsen, J.Macromol. Sci., Rev.Macromol. Chem., 1969, C3, 69. shortcomings of the various models (composition-Tg and 24 T. G. Fox and S. Loshaek, J. Polym. Sci., 1955, 15, 371. crosslinking-Tg) in adequately describing such complex sys- 25 J. Bicerano, R. L. Sammler, C. J. Carriere and J. T. Seitz, J. Polym. tems. The copolymers containing the highest amounts of Sci. B, Polym. Phys., 1996, 34, 2247. EH(M)A, which consequently are only 1% crosslinked, would be expected to mimic the model behaviour more closely. This Paper 7/02030I; Received 24thMarch, 1997 2212 J. Mater. Chem., 1997, 7(11), 2209–2212