PVT Behavior of Thermoplastic Poly(styrene-co-acrylonitrile)-Modified

Oct 31, 2008 - School of Chemical Sciences, Mahatma Gandhi UniVersity, Priyadarshini ... The volume shrinkage during polymerization of a thermoplastic...
0 downloads 0 Views 2MB Size
J. Phys. Chem. B 2008, 112, 14793–14803

14793

PVT Behavior of Thermoplastic Poly(styrene-co-acrylonitrile)-Modified Epoxy Systems: Relating Polymerization-Induced Viscoelastic Phase Separation with the Cure Shrinkage Performance Jesmy Jose,† Kuruvilla Joseph,†,‡ Ju¨rgen Pionteck,§ and Sabu Thomas*,† School of Chemical Sciences, Mahatma Gandhi UniVersity, Priyadarshini Hills P.O., Kottayam, Kerala, India, Leibniz-Institute fu¨r Polymerforschung Dresden e.V., Hohe Straβe 6, 01069 Dresden, Germany ReceiVed: March 7, 2008; ReVised Manuscript ReceiVed: August 4, 2008

The volume shrinkage during polymerization of a thermoplastic modified epoxy resin undergoing a simultaneous viscoelastic phase separation was investigated for the first time by means of pressure-volume-temperature (PVT) analysis. Varying amounts (0-20%) of poly(styrene-co-acrylonitrile) (SAN) have been incorporated into a high-temperature epoxy-diamine system, diglycidyl ether of bisphenol A (DGEBA)-4,4′-diaminodiphenyl sulfone (DDS) mixture, and subsequently polymerized isothermally at a constant pressure of 10 MPa. Volume shrinkage is highest for the double-phased network-like bicontinuous morphology in the SAN15% system. Investigation of the epoxy reaction kinetics based on the conversions derived from PVT data established a phase-separation effect on the volume shrinkage behavior in these blends. From subsequent thermal transition studies of various epoxy-DDS/SAN systems, it has been suggested that the behavior of the highly intermixed thermoplastic SAN-rich phase is the key for in situ shrinkage control. Various microscopic characterizations including scanning electron microscopy, atomic force microscopy, and optical microscopy are combined to confirm that the shrinkage behavior is manipulated by a volume shrinkage of the thermoplastic SAN-rich phase undergoing a viscoelastic phase separation during cure. Consequently, a new mechanism for volume shrinkage has been visualized for the in situ polymerization of a thermoplastic-modified epoxy resin. Introduction Addition of low amounts (up to 20 wt %) of engineering thermoplastics (TP) is a promising way to overcome the intrinsic brittleness of the classical epoxy resins, with the least compromise on the processability. Various TP-modified epoxy resins possess enhanced fracture toughness,1-4 adhesion,5 thermomechanical stability,6 etc. on account of their interesting microstructural characteristics. With the epoxy glass transition as well as the modulus being least affected upon TP addition, these materials are now increasingly replacing the conventional unmodified high-temperature purpose epoxy resins. A key application area includes the aerospace industry, where, these high-impact resins make up superior matrices for the carbon, Kevlar, and Aramid fiber-reinforced composites that constitute the various structural components of aircraft and rockets.7,8 During epoxy polymerization, the continuous replacement of secondary van der Waals bonds by primary covalent bonds results in a volume shrinkage (ca. 4-7%). This is a serious problem in the area of material sciences, especially for composite materials. Being cured under geometrically constrained environments, cure shrinkage generates large residual stresses that further lead to severe manufacturing problems such as surface-quality flaws, shape distortions, warpage, etc., in the epoxy-based materials. When translated to thermoplastic incorporated epoxies, the polymerization however proceeds via a competitive series of transitions: the liquid-liquid phase separation giving separate epoxy and TP-rich phases, gelation of the * Corresponding author. E-mail: [email protected]. † Mahatma Gandhi University. ‡ Current address: Indian Institute of Space Science and Technology, Department of Space, IIST Post, Thiruvananthapuram, Kerala, India. § Leibniz-Institute fu ¨ r Polymerforschung Dresden e.V.

epoxy phase, or a subsequent vitrification (Tg ) Tcure) of either or both of the epoxy- and thermoplastic-rich phases. Consequently, in addition to chemical cross-linking, the phaseseparation characteristics, such as the mechanism and coarsening kinetics and nature and type of morphologies, would also presumably influence the in situ shrinkage behavior of these modified resins. Studies on internal stress in polyester-modified epoxies by Cao et al.9 revealed a composition-dependent behavior that was attributed to the influence of thermoplastic polyester on the epoxy cross-link density. In the case of TPmodified polyester resins, the phase separation favorably offered an effective shrinkage control via a “low profile” mechanism, which is determined by the morphological changes during cure.10-12 However, the volume-shrinkage behavior of epoxy/ thermoplastic systems remains unexplored, even though it is very significant from both scientific and industrial viewpoints. It is now generally accepted that a predominant Viscoelastic phase separation13 follows the spinodal demixing during the polymerization-induced phase separation (PIPS) in dynamically asymmetric epoxy/TP systems to give mechanically favorable final network-like and sponge-like morphologies.14-16 The large difference in Tg, molecular weight, or both between the epoxy oligomer and thermoplastic component induces a dynamic viscoelastic asymmetry, which on coupling with concentration fluctuations, results in a viscoelastic phase separation (VPS) during PIPS. Here, the relaxation time of the interaction network of slow fluid (thermoplastic) being longer because of the low mobility (or large size), therefore, fails to catch up with the characteristic time of phase separation and leads to a viscoelastic phase separation. VPS is characterized by unique coarsening behaviors like phase inversion, TP volume shrinkage, etc., which are never observed for conventional phase separation. The

10.1021/jp802015n CCC: $40.75  2008 American Chemical Society Published on Web 10/31/2008

14794 J. Phys. Chem. B, Vol. 112, No. 47, 2008 interplay between phase-separation (or critical) dynamics and the slow dynamics of the thermoplastic material initially generates transient phase-inverted systems having domains of the fast dynamic epoxy-rich phase in slow dynamic thermoplasticrich phase matrix. Then these domains grow further to give network-like or sponge-like structures. Finally, the slow dynamic epoxy-rich phase develops into a dispersed phase by volume shrinkage. Among the few researchers, Zheng et al.14 studied the coarsening dynamics for PEI-modified epoxy resin and described successfully the temperature-dependent relaxation time for the viscoelastic flow of epoxy oligomers using the WLF equation. Additionally, a simultaneous secondary phase separation (SPS), which yields complex phase-in-phase microstructures, has also been visualized in several epoxy/TP systems.17-20 Therefore, any investigation on the epoxy resin properties by considering the possible effects of these complex events would certainly serve toward better end-use applications. In this context, we describe the in situ volume shrinkage behavior of a thermoplastic SAN-modified diglycidyl ether of bisphenol A (DGEBA) epoxy by means of PVT analysis. The initially homogeneous epoxy-4,4′-diaminodiphenyl sulfone (DDS)/SAN mixtures undergo a reaction-induced phase separation during cure. Because the kinetic interplay between phase separation and cross-linking is the key to morphology selection, this paper exploits the phase separation-microstructure-cure shrinkage performance correlation in SAN-incorporated epoxy resins. The reasons for selecting SAN for the present study are as follows: (i) Since the SAN glass transition (Tg ) 105 °C) is lower than the isothermal cure temperatures (Tcure )140 and 160 °C) selected, the vitrification of the SAN-rich phase never happens during cure. (ii) Because of the very high SAN molecular weight (Mw) 54 800 g/mol), compared with the epoxy oligomer, the dynamic asymmetry would favorably induce a viscoelastic phase separation during cure. Moreover, from our recent studies, the SAN-modified epoxy matrices were found capable of imparting augmented mechanical strength as well as thermomechanical stabilities in glass fiber reinforced composites.21 Experimental Section Materials and Samples Preparation. The epoxy precursor used, a diglycidyl ether of bisphenol A (DGEBA), is a liquid at room temperature with a low degree of polymerization (n ) 0.15) (LY556 manufactured by Ciba Geigy and kindly supplied by Vantico Polymers, India). The curing agent was an aromatic diamine with intermediate reactivity, 4,4′-diaminodiphenyl sulfone (DDS), supplied by Atul Products India, Ltd. The amorphous thermoplastic styrene-co-acrylonitrile copolymer (Mw) 54 800 g/mol; Tg ) 105 °C, acrylonitrile content ) 33 wt %) was procured from Bayer Chemicals, Belgium. The blends were prepared by initially dissolving SAN in epoxy resin by continuous stirring at 180 °C in a N2 atmosphere. The DDS hardener, taken in stoichiometric ratio of epoxy to amino hydrogen’s (equal to one) was then added at a lower temperature of 140 °C. The clear solution was then quenched in liquid nitrogen and stored in a freezer to prevent any curing before the actual measurements started. Blends with SAN content varying between 0 and 20 g in the epoxy-hardener mixture (100 g DGEBA + 32 g DDS) were prepared, and the samples were named as neat epoxy, SAN-5%, SAN-10%, SAN15%, and SAN-20% blends. Measurements. Specific volumes were measured using a fully automated GNOMIX high-pressure dilatometer capable of detecting volume changes as small as 0.0002 cm3/g within

Jose et al. TABLE 1: Experimental and Calculated Initial Specific Volumes for Epoxy/SAN Blends at 160 °C % SAN

specific volume (exptl), cm3/g

specific volume (calcd), cm3/g

0 5 10 15 20

0.8867 0.8891 0.8887 0.8978 0.8983

0.8943 0.8972 0.8999 0.9025 0.9049

an accuracy limit of 0.002 cm3/g (below 200 °C). The apparatus is described elsewhere.22 Ultrahigh-vacuum samples embedded in a nickel foil cup were loaded in the liquid Hg filled PVT sample cell and were isothermally scanned for specific volumes at a constant pressure of 10 MPa. In order to check whether any cross-linking has taken place during the sample preparation stages, the experimental initial specific volumes were compared with those calculated from the densities of the individual components by assuming an additive behavior (Table 1). The lower values for experimental specific volumes indicate a certain degree of cross-linking; however, the deviation was within the acceptable range (less than 1%). The cloud-point measurements were performed on a built-in small-angle light scattering (SALLS) set up equipped with a 5 mW He-Ne laser (λ ) 638.2 nm). After necessary calibrations, a few milligrams of the sample was placed between two microscopic glass slides and cured on the hot stage at 160 °C. The onset time of liquid-liquid phase separation was determined from the surge in the scattering intensities being recorded as a function of time using built-in software (custom-made TestPoint application). Thermal transitions of the cured samples were determined by measuring the loss modulus (E′′) as a function of temperature using a dynamic mechanical thermal analyzer at a heating rate of 3 °C/min (TA instruments DMA 2980 operating in a dual cantilever mode at 1 Hz). Samples obtained after curing at 160 °C for 6 h with approximate dimensions of 20 mm × 3 mm × 5 mm were used. Morphology development during polymerization was studied using an optical microscope (Olympus BH2). A few milligrams of the sample was sealed between two glass slides and placed in a Mettler FP82-HT hot stage at 160 °C. Digital micrographs were taken at several curing times with a JVC TK-C1381 CCD camera and analyzed by the program Qwin from Leica Company. The final blend morphologies were investigated using scanning electron microscopy (SEM) and atomic force microscopy (AFM). The cured samples from PVT run were cleaned to remove the mercury. To get a better phase contrast, the thermoplastic SAN was preferentially etched out using THF at room temperature, and the samples were vaccuum-dried, sputtered with gold, and scanned using a Leo 435 VP scanning electron microscope. In blends where an etching with THF ruined the original blend morphology, atomic force microscopy (AFM) was performed on the unetched cryo-smoothened samples. AFM pictures were captured in tapping mode by means of a NanoScope Dimension 3100 (Veeco). Results and Discussion Determination of the Volume Shrinkage in Epoxy-DDS/ SAN Blends from PVT Results. Pressure-volume-temperature analysis is a well-known probe for studying the volume changes accompanying phase transitions in thermoplastic polymers. However, for thermosets, the specific volume variations (PVT behavior) could be used to investigate the volume shrinkage during polymerization. Under isothermal polymeri-

PVT Behavior of Thermoplastic SAN-Epoxy Systems

J. Phys. Chem. B, Vol. 112, No. 47, 2008 14795

Figure 1. PVT data for SAN-modified epoxy-DDS blends during isothermal cure at 140 (closed symbols) and 160 °C (open symbols) at a constant pressure of 10 MPa: (9,0) neat epoxy; ([,]) SAN-5%; (2,4) SAN-10%; (b,O) SAN-15%; (1,3) SAN-20%.

zation conditions, the specific volume changes (at fixed T and P) directly reflect the internal structure and volume-averaged interactions in the sample without any hindrance from the glass transition and thermal expansion coefficient. Figure 1 gives the specific volume data obtained for the various epoxy-DDS/SAN mixtures studied at 140 and 160 °C. Irrespective of blend composition, the volume decays exponentially due to polymerization and remains constant in the final stages of cure. The beginning of the final plateau region corresponds to gelation or vitrification processes; hereafter, the specific volume change is less sensitive. The raw specific volume data seems to give some information of a SAN concentration effect. For both isothermal cure temperatures (Figure 1), at certain intermediate stages of cure, the specific volume curves for the SAN-10% and SAN-15% systems show a cross-over, whereby the SAN-15% system gave slightly lower specific volumes than SAN-10% blend in the final cure stages. From the specific volume values, the percentage volume shrinkage at any time t could be calculated as

[

% volume shrinkage ) 1 -

blend V (t) SAN epoxy V (t)0) + V (t)0)

]

× 100 (1)

epoxy where V(tSAN ) 0) and V(t ) 0) are the initial volumes of pure SAN blend is the and neat epoxy-DDS mixtures at time t ) 0. V (t) volume of the blend at any time t calculated from the experimental blend specific volumes using eq 2:

blend blend V tblend ) V sp φ t

(2)

V blend spt is the experimental specific volume for blends at any time t and φblend represents the total weight of the blend in grams. Equation 3 gives the final shrinkage values normalized with respect to the epoxy volume content in the blend.

(

% shrinkagenormalized ) 1 -

epoxy V (t) epoxy V (t)0)

)

× 100

(3)

Figure 2 gives the shrinkage evolution profile for epoxy-DDS/ SAN samples cured isothermally at 160 °C. Among the blends, with increasing SAN content, the percentage shrinkage decreased progressively up to a certain cure time (around 50 min). Interestingly, thereafter, the SAN-15% system exhibited rela-

Figure 2. Volume shrinkage vs time plots for epoxy-DDS/SAN blends having different SAN concentrations (%) during the isothermal cure at 160 °C at a constant pressure of 10 MPa: (0) neat epoxy; (]) SAN-5%; (4) SAN-10%; (O) SAN-15%; (3) SAN-20%. The inset graph gives the net volume shrinkage in relation to the blend volume (b) and normalized to the volume of the epoxy phase (9) (the dotted lines are to guide the eyes).

tively higher volume shrinkage than the SAN-5% and SAN10% blends. In any case, in the absence of any phase separation, at any given cure time the thermoplastic SAN would merely dilute the epoxy-DDS reaction volume causing a progressive decrease in the relative volume shrinkage with the increasing SAN concentration. Therefore, in epoxy-DDS/SAN samples, the simultaneous thermoplastic phase separation during epoxy polymerization would possibly account for the experimentally observed (Figure 2) anomalous trend in the shrinkage evolution as determined by various factors, such as stoichiometric variations, phase compositions, and interfaces. The inset in Figure 2 gives the final cure shrinkage (from eq 2) together with the normalized values (from eq 3) as a function of the SAN content. In either case, the volume shrinkage decreased linearly up to the SAN-10% system, then shows an upturn giving a maximum shrinkage for the SAN-15% system, thereby further demonstrating the profound thermoplastic phase separation effect on volume shrinkage. Phase-Separation Effect on the Volume Shrinkage Kinetics of Epoxy-DDS/SAN Blends. In view of the PVT data a general equation of state for the specific volume (V) dependence on temperature (T), pressure (P), and conversion (X) for epoxy polymerization could be expressed as23

dV 1 ∂V ) V V ∂T

( )

P, X

dT +

1 ∂V V ∂X

( )

P,T

dX +

1 ∂V V ∂P

( )

T, X

dP

(4)

For curing at constant T and P the volume changes between the times 0 and t could be then given by eq 5:

∫0t dV ) ∫0X VtR(X,T,P) dX

(5)

where R(X,T,P) ) 1/V(∂V/∂X)P,T is a volumetric coefficient associated with polymerization. For neat epoxy systems, a likely hypothesis is to assume a linear relationship between epoxy conversion and cure shrinkage, and then the conversion X could be represented as24,25

X)

V t - V 0 V∞ - V0

(6)

where V ′t and V ′0 are the specific volumes at time t and at t ) 0, respectively, for each cure temperature. V ′0 and V∞ are the

14796 J. Phys. Chem. B, Vol. 112, No. 47, 2008

Jose et al.

SCHEME 1: Curing Reactions in DGEBA-DDS Systems

initial and final specific volumes at a reference temperature considered as the highest temperature used, which in our case was 160 °C. The step polymerization of an epoxy-amine system is well established in the literature (Scheme 1): an oxirane ring opening by the primary amine hydrogen (reaction 1) gives a secondary amine that further reacts with another epoxy group to create a tertiary amine (reaction 2). Recently Ramos et al.26 have derived the epoxy reaction kinetics based the conversion values obtained from the PVT analysis of an epoxy-MCDEA hardener mixture by measuring the specific volumes at varying isothermal temperatures and pressures. They noticed a predominant temperature effect on the kinetic rate constants. For epoxy-amine systems, the phenomenological autocatalytic model proposed by Kamal27,28 is found to be the most conceivable.29-31 The generalized rate expression for a non-zero initial reaction rate is given by eq 7:

dX ) (k1 + k2X m)(1 - X)n dt

(7)

where k1 and k2 are the apparent rate constants for the reactions autocatalyzed by the hydroxyl groups initially present and those generated during cure; m and n are the order of the reaction. The above model gave very good fitting for neat epoxy at all temperatures studied (results are not discussed here). In thermoplastic SAN-modified epoxy-DDS blends, an investigation of the cure kinetics based on conversions from PVT data would enable a quantitative determination of any phase separation effect on shrinkage. Figure 3a-c illustrates the results of the kinetics analysis for epoxy-DDS/SAN samples using the autocatalytic model. In the case of blends, the diluent effect of SAN has been taken into account during modeling because it does not affect the values of the apparent rate constants. Note that no chemical reactions between SAN and either epoxy or DDS were noticed from our earlier IR studies. The symbols (Figure 3a-c) represent experimental conversion (from PVT measurements at 160 °C) for the 10%, 15%, and 20% SANmodified blends measured at a constant pressure of 10 MPa. The solid lines in each graph represent the theoretical conversion values obtained by a differential of eq 7 resolved using a fourthorder semi-implicit Runge-Kutta method32,33 with k1 and k2 as fitting parameters. The cloud-point time determined by light scattering (indicated by arrows) represents the beginning of the phase-separation process. In the homogeneous region (before phase separation in the range of cloud-point arrows), a good agreement between the experimental and theoretical curves has been observed. In terms of volume shrinkage, the above finding evidently establishes the dilution effect by SAN on epoxy volume shrinkage in the miscible state. However, at the onset of phase separation (marked by arrows), the experimental conversion curves deviated from the theoretical values, thereby establishing for the first time a phase-separation effect on volume

shrinkage that is being manifested by a possible coupling between phase-coarsening kinetics and the reaction kinetics. This means that in the phase-separated regime, the experimental conversion values account for the combined volume changes associated with both phase separation and polymerization. As seen from Figure 3, beyond the onset of phase separation, the relative extent of conversion at a fixed time is found to vary as 10% < 15% > 20%, which agrees with our earlier findings on the relative shrinkage evolution among the blends (Figure 2). Since the extent of conversion due to volume changes may vary in the separate emerging epoxy- (R1) and SAN-rich (β1) phases, the term “average conversion” is better used to represent the experimental conversion values. Considered in terms of chemical reaction rates, the above observed deviations in experimental conversions during the phase separation of nonreactive SANmodified epoxy systems could be due to numerous factors, stoichiometric variations, phase compositions, etc., which are being further investigated. Remarkably, a failure of the presumed linear relationship between shrinkage and conversion would also contribute toward the altered experimental conversion obtained for epoxy-DDS/

Figure 3. Experimentally determined (symbols) average epoxy conversion (Xavg) vs time for various epoxy-DDS/SAN systems cured at 160 °C and 10 MPa having (a) SAN-10%, (b) SAN-15%, and (c) SAN20%. The solid lines are the theoretical values (see text). The arrows indicate the onset of phase separation obtained from light scattering studies.

PVT Behavior of Thermoplastic SAN-Epoxy Systems SAN blends. In reactive systems undergoing a polymerization induced phase separation, a possible difference in reaction rate and modulus of the emerging phases would generate internal stresses at the interface. In low profile additive thermoplastic (LPA)-modified polyester resins, during PIPS, the release of internal stresses via microvoids caused a substantial volume expansion, which further counteracts the simultaneous polymerization shrinkage and thereby causes a rapid decline in the net volume shrinkage values.10-12 However, in epoxy-DDS/ SAN systems, the rather monotonous evolution of the shrinkage curves (Figure 2) would indicate the absence of any such competitive volume expansion, which could be explained as follows. During cure, the development of internal stresses happens only when the polymerization shrinkage is greater than the “resistance” force possessed by the collective interfacial attributes constituting the phases. Generally in thermoplasticbased epoxy systems, the gelation happens at a much later stage of cure, usually around 57%, which is well beyond phase separation. Therefore in the initial and intermediate stages of curing, the stress generated at the interface is not sufficient to initiate any volume expansion. Besides, in epoxy-DDS/SAN blends, because the SAN glass transition is always much lower than the experimental isothermal temperature (140 and 160 °C), a higher stress relaxation due to the viscoelastic deformation of SAN chains tends to prevent the growth of microcracks. The topographical analysis of the SEM micrographs of the unetched epoxy-DDS/SAN blend sample surfaces (not given here) obtained after PVT run failed to display any microvoids in both epoxy- and SAN-rich phases and therefore further corroborates the proposed negligible interfacial contribution toward the volume shrinkage. Investigation of the Morphologies and Phase Behavior of SAN-Modified Epoxy-DDS Blends. The morphologies of the samples obtained after the isothermal PVT run at 160 °C were studied using SEM and are illustrated in Figure 4A-G. In the SEM micrographs B, C, D, and F, the dark areas correspond to the regions from which the SAN phase (β1) has been etched out using THF, and the relatively bright area indicates the epoxyrich (R1) phase. The unmodified neat epoxy gives an invariant morphology quite evident of a homogeneous system (Figure 4A). The microstructure of SAN-5% and SAN-10% blends given in Figure 4C,D consists of primary spherical SAN-rich domains dispersed in the continuous epoxy-rich matrix. At a higher magnification in the case of the SAN-10% sample (Figure 4D), the primary SAN domains disclosed secondary interconnected epoxy globules (R2) that have emerged via a secondary phase separation. For the sake of simplicity, the morphological implications of SPS will be discussed separately. The SAN-15% system exhibits “macrophase”-separated morphology (Figure 4E). In the figure, the relatively dark area represents the smoothened epoxy-rich (R1) region and the bright coarse regions constitute the SAN-rich (β1) phase. Such a primary network-like morphology consisting of large thermoplastic-rich domains exhibiting irregular shapes and dispersed in an epoxy-rich continuous phase that is very close to the cocontinuous phase structure is characteristic of a predominant viscoelastic phase separation during PIPS.34,35 At an even higher SAN content, that is, for the SAN-20% system, the final morphology consists of a phase-inverted sponge-like morphology, again distinctive for viscoelastic phase separation. Figure 5A shows the AFM picture of the unetched SAN-20% blend showing large dispersed epoxy globules of average size of 15-25 µm embedded in a continuous SAN-rich (β1) phase. Here it was noticed that the dispersed epoxy globules are not

J. Phys. Chem. B, Vol. 112, No. 47, 2008 14797 interconnected as evidenced by a complete destruction of the specimen while etching with THF. A careful analysis of the SEM, as well as AFM, micrographs reveals double-phase structures due to a secondary phase separation in either or both epoxy- and thermoplastic-rich primary phases. Figure 4C,D,G gives a closer view of secondary epoxy particles (R2) in SAN-rich (β1) matrix. Except for the SAN-20% system, the secondary epoxy particles existed as interconnected spherical globules in the primary thermoplastic SAN-rich phase as verified from the intact nature of the epoxy globular structure after THF treatment. Again, the secondary SAN particles were seen to be randomly dispersed domains in the epoxy-rich primary phases of SAN-15% (Figure 5F) and SAN-20% (Figure 5D) systems. The phase behavior of the above complex double-phase structures was investigated using dynamic mechanical thermal analysis (DMTA). Two separate transitions in the loss modulus vs temperature plot (Figure 6) indicate phase separation; the modulus peak at lower temperature corresponds to the SANrich phase (β1) and that at higher temperature is due to the relaxation of the cross-linked epoxy-rich phase (R1). The increase in the height of the R relaxation peak of SAN with increase in SAN content demonstrates the shift in primary morphology from a particulate to phase inversion and therefore corroborates the morphology results. The glass transition is a parameter sensitive to the degree of cure and, furthermore is a very good probe for stoichiometric imbalances. In any case, a deviation of 5% in the epoxy-amine mixture composition, for example, would result in a change of 5 °C in the full cure glass transition of neat epoxy-DDS system.36 The above notion was considered to examine the possible deviations in the epoxy-amine ratio of the primary epoxy-rich phase in stoichiometric epoxy-DDS/SAN blends. It has been found that the composition of the primary epoxy-DDS-rich phase of these systems is close to pure epoxy-amine for cross-linking beyond the onset of phase separation.37 For isothermal curing at 160 °C, upon the completion of cure, the absence of differential segregation would result in an epoxy-phase glass transition close to that of the neat epoxy-DDS. The glass transition of the epoxy-rich phase (at 160 ( 2 °C) in epoxy-DDS/SAN blends shows no significant deviation from that of the neat epoxy-DDS; hence, denying any possible epoxy-DDS stoichiometric imbalances in the epoxy-rich phase during reaction-induced phase separation. Similar unshifted final epoxy glass transitions have been often reported for low Tg (Tg of thermoplastic < Tcure) thermoplastic-modified epoxy blends.38,39 In blends, the SAN-rich phase exhibited broad relaxation peaks (their widths vary from 30 to 40 °C for 5% and 10% SAN containing blends, and 60 °C for the 15% SAN-containing system) having an onset around 100 °C that is very close to the Tg of neat SAN (Figure 6). Besides, the peak maxima were shifted to higher temperatures and the extent of Tg increase followed a descending trend with increasing SAN content. Earlier, the morphology studies on the various blend systems revealed secondary epoxy particles present in the highly intermixed SAN-rich phase. So, the broader relaxation peak as well as the increased thermoplastic glass transition observed for the highly intermixed SAN-rich primary phase in various epoxy-DDS/SAN blends could be attributed to the closer relaxation temperatures corresponding to SAN phase and that due to the partially/fully phase-separated cross-linked secondary epoxy substructures, the latter having a lower degree of crosslinking than the primary epoxy-rich phase. This finding is

14798 J. Phys. Chem. B, Vol. 112, No. 47, 2008

Jose et al.

Figure 4. SEM micrographs of various epoxy-DDS/SAN blends after 7 h of isothermal curing at 160 °C and 10 MPa from PVT analysis: (A) unmodified epoxy; (B) SAN-5% blend; (C) SAN-10% blend; (D) SAN-10% (magnified phase-in-phase structure); (E) SAN-15% blend; (F) SAN15% (magnified view of the epoxy-rich phase with secondary SAN domains); (G) SAN-15% (magnified view of the SAN-rich phase with secondary epoxy globules).

particularly relevant for SAN-15% blend system, where the primary epoxy phase relaxation, reduced to a shoulder of the main peak at 111 °C, contrary to an expected relatively dominant separate peak for the primary epoxy-rich major phase (Figure 6). As verified from the morphological results, in SAN-15% blend, the relatively large volume fraction and the possible high cross-link densities of the secondary epoxy substructure in the intermixed SAN-rich primary phase accounts for the above interesting relaxation behavior. The morphological results together with glass transition and the final volume shrinkage of the completely cured epoxy-DDS/

SAN blends are summarized in Table 2. The glass transition is directly related to specific volume. Thus, depending on the glass transition behavior, the corresponding cross-linking densities of the epoxy chains present in the separate primary epoxy-rich phase, as well as for those in the highly intermixed SAN-rich phase, would influence the relative volume shrinkage behavior of epoxy-DDS/SAN blends. Having established an intact glass transition for the primary epoxy-rich phase with an undisturbed epoxy-amine stoichiometry during phase separation and also considering the discussions pertaining to the broad relaxation behavior of the highly intermixed SAN phase, we suggest that

PVT Behavior of Thermoplastic SAN-Epoxy Systems

J. Phys. Chem. B, Vol. 112, No. 47, 2008 14799

Figure 5. AFM image of (A) SAN-20% blend, (B) SAN-20% (magnified view), (C) secondary structures in SAN-rich matrix, and (D) secondary structures in epoxy-rich dispersed phase.

Figure 6. Loss modulus vs temperature plots of the epoxy-DDS/ SAN samples having varying SAN contents being measured after 6 h cure at 160 °C: (0) neat epoxy; (9) neat SAN; (]) SAN-5%; (4) SAN10%; (O) SAN-15%; (3) SAN-20%.

the behavior of the highly intermixed SAN-rich phase illustrated by both the composition and morphological disposition during polymerization-induced phase separation would preferentially determine the relative volume shrinkage behavior of the various epoxy-DDS/SAN blends. Phase Structural Evolution during PIPS in Epoxy-DDS/ SAN Systems. It is common knowledge that the modifier concentration has a strong influence on both the phase-separation mechanisms and the microstructural attributes of reactive polymer systems. In epoxy-TP blends (usually for TP < 30%), a spinodal decomposition (SD) has been generally accepted as the most conceivable mechanism of phase separation. Figure 7 shows the optical micrographs during phase separation in SAN10% blend at 160 °C. At the onset of phase separation, as visualized by Figure 7a, the well-interconnected initial bicontinuous morphology establishes a spinodal demixing mechanism of phase separation. A consequent phase coarsening by interdiffusion looses the interconnectivity of the minor SAN-rich phase (Figure 7b) in the SAN-10% system to finally yield a SAN-dispersed microstructure (Figure 7c). Careful analysis of the primary SAN-rich phase disclosed fine secondary epoxy substructures at the early stages of primary phase separation. A consensus is yet to be established regarding the possible reasons behind this scientifically debated complex

phenomenon observed for phase-separating reactive systems. In a recent study, Tang et al.18 have suggested that a predominant hydrodynamic effect induced by the faster phase separation rate induced secondary phase separation in the phase-inverted morphologies of epoxy-PES blends. In a broader sense, generally for polymer mixtures, the phenomenon of secondary phase separation emerges through at least four processes: (a) noninstantaneous temperature changes;40,41 (b) altered thermodynamic conditions, for example, by cross-linking42 or conformational ordering;43 (c) mass transfer limitations;44 (d) rapid hydrodynamic coarsening.45 Process (a) is not feasible under the present experimental conditions of isothermal curing. Because the gelation is well-separated from phase separation, no semipermeable membrane is formed during phase separation in epoxy-SAN blends, and hence process (c) is not interesting. Since the curing of epoxy involves a continuous quench, the equilibrium compositions are not constant; they move apart. As a result, the coexisting phases have to change their composition continuously by material exchange. Because of a rapid hydrodynamic coarsening and the increasing distances, material exchange by diffusion becomes more and more difficult, and at a certain time, the diffusion becomes too slow to follow the structure coarsening resulting in secondary demixing inside the primary domains. Thus in epoxy-SAN blends, a combined effect due to cross-linking and the rapid hydrodynamic coarsening of the initial bicontinuous structures would probably be the main reason for secondary phase separation. Concomitantly, recent studies on the PIPS effect on reaction kinetics of epoxy-TP systems with a high TP content (TP > 20 wt %) have indicated that the relative interdiffusion rates between the coexisting heterogeneous primary epoxy- and thermoplasticrich phases competes with chemical reaction rate during PIPS.36,38,39 Due to its complexity, comprehensive analyses of SPS in epoxy-TP blends are limited, which would be interesting toward the development of novel materials with unique properties. As far as the present study is concerned, the preferential interdiffusion between the primary and secondary phases as a result of secondary phase separation might be the reason for the earlier findings of unshifted epoxy-rich phase glass transition as well as for the existence of a highly intermixed SAN-rich phase in the SAN-modified systems. Among the various blends, the SAN-15%, having a networklike macrophase structure, exhibited the maximum volume shrinkage during cure. The phase structural evolution for SAN15% blend at 160 °C was studied using optical microscopy and

14800 J. Phys. Chem. B, Vol. 112, No. 47, 2008

Jose et al.

TABLE 2: Summary of the Volume Shrinkage, Glass Transition Behavior, and Morphological Observations Obtained for Epoxy-DDS/ SAN Blends Cured at 160 °C glass transition temperature (°C) sample

temperature (°C)

morphology

neat epoxy SAN-5% SAN-10% SAN-15% SAN-20% SAN-100%

160 160 160 160 160 160

homogeneous SAN dispersed SAN dispersed network-like bi continuous sponge-like phase inversion homogeneous

a

epoxy-rich phase (R1) 160 160 161 158a

SAN-rich phase (β1) 116 113 111 108 104

volume shrinkage (%)

volume shrinkage normalized (%)

7.3 6.6 5.7 6.2 5.4

7.3 7 6.4 7.2 6.7

The relaxation at 158 °C is reduced to a shoulder of the broad main relaxation peak at 111 °C.

Figure 7. Optical microscopy pictures of SAN-10% blend cured during different times at 160 °C: (a) 18, (b) 20, and (c) 35 min.

Figure 8. Optical microscopy pictures of SAN-15% blend cured during different times at 160 °C: (a) 21, (b) 22, (c) 23, (d) 24, (e) 25, and (f) 36 min.

can be described as follows. As shown in Figure 8a, the onset of phase separation for the SAN-15% system is visualized by the formation of initial bicontinuous structures due to a spinodal demixing mechanism. Thereafter, in contrast to the expected spinodal coarsening dynamics, the minority SAN phase forms the matrix containing nucleated epoxy dispersed spheres (Figure 8b). This is due to viscoelastic phase separation.12 According to various theories of viscoelastic phase separation13 and as also recently reported by Li and co-workers15,34 for TP-modified epoxies, the pattern evolution during viscoelastic phase separation is governed by a competitive dominance between deformation (τd, characteristic time of deformation that represents the composition fluctuations between the SAN- and epoxy-rich phases) and the rheological time scales (τts, characteristic rheological time of the slower thermoplastic SAN phase representing the elastic field). Upon phase separation, τd decreases rapidly at first with increasing composition difference

and then increases with the domain size, while τts increases steeply with the increased composition difference and becomes almost constant in the late stage. Accordingly, VPS proceeds through three main stages as explained below for the SAN15% blend. Figure 8b represents the initial diffusion stage (τd > τt) where the SAN viscoelastic effects prevent the growth of the normal composition fluctuations resulting in epoxy nucleated spheres. Here it is noteworthy that the primary SAN-rich matrix exists as a coarse-grained highly intermixed phase comprising epoxy substructures due to an early secondary phase separation. Subsequently (Figure 8c-e), the primary epoxy spheres grow by the transport of the epoxy oligomers from the highly intermixed SAN-rich primary matrix toward the epoxy-rich holes making the SAN phase increasingly elastic. This is accompanied by a volume shrinkage of the highly intermixed SAN-rich phase in the elastic regime (τd < τts). Interestingly, a

PVT Behavior of Thermoplastic SAN-Epoxy Systems

J. Phys. Chem. B, Vol. 112, No. 47, 2008 14801

Figure 9. Schematic representation of the mechanism of volume-shrinkage evolution during reaction-induced viscoelastic phase separation in thermoplastic SAN-modified epoxy systems.

simultaneous diffusion of the epoxy oligomers toward the secondary epoxy phase also happens in conjunction with the growth of the primary epoxy spheres as evident from the rapid coarsening of the secondary epoxy globules visualized in Figure 8c-e. Furthermore, the viscoelastic phase separation theory necessitates the existence of a “transient gel-like” structure for the more viscoelastic phase in order to undergo the so-called volume-shrinkage phenomenon during VPS.13 Therefore, the rather “permanent gel-like” network structure constituting a highly intermixed SAN-rich phase containing interconnected secondary epoxy globules validates the observed volume shrinkage during viscoelastic phase separation in SAN structural evolution for SAN-15% blend. In the final stages, the relaxation of dynamic asymmetry occurs, and the interfacial tension dominated hydrodynamic regime (τd > τts) yields a final network-like morphology having a predominant epoxy major phase (Figure 8f). Thus during viscoelastic phase separation, in addition to the cure shrinkage owing to polymerization, a volume shrinkage of the highly viscoelastic SAN-rich phase likely contributes to the net volume shrinkage, which further toward the final cure stage is being manifested by an apparent cross-linking of the secondary epoxy substructures present in the intermixed SAN-rich primary phase. Volume Shrinkage Mechanism in Epoxy-DDS/SAN Blends. In view of the results and discussions so far, we can divide the dynamics of volume shrinkage control in epoxy/SAN blends into five stages as given by the schematic diagram in Figure 9. Stage 1 corresponds to the early induction stage, where the system starts as a homogeneous fluid mixture consisting of epoxy resin, thermoplastic, and the hardener (assuming no mixing effects). Hereafter, due to the epoxy cross-linking, the volume shrinkage increases progressively with time. The thermoplastic SAN dilutes the oxirane ring concentration, and, therefore, in the homogeneous regime, the relative rate of

shrinkage among the blends decreases proportionally with increasing SAN content. Here, no volume contributions from SAN chain deformation are expected by the pressure effect because SAN exists in the melt state (SAN Tg being lower than the cure temperature). At different varying extents of conversion, determined by various thermodynamic parameters, a deep quench into the unstable region through a spinodal mechanism (SD) yields initial bicontinuous morphologies as given by stage 2 in Figure 9. From this stage onward, the total volume shrinkage is the average of the volume changes due to the separate primary epoxy- and thermoplastic-rich phases. The initial bicontinuous structures undergo rapid coarsening, which when coupled with the continuous chemical quench induces a secondary phase separation to yield highly intermixed SAN and epoxy phases. Further, in the late stages of SD phase separation (stage 3), the volume shrinkage is determined by the contributions from both reaction and phase-coarsening kinetics. For a low TP content system (φSAN e 10%), a much faster percolation to cluster transition occurs according to the conventional SD late stage dynamics, whereby the continuous thermoplastic minor phase breaks up to reduce the interfacial tension. A different situation exists when the TP content is beyond that of the SAN10% system (φSAN g 15%). Here, a simultaneous growth of the dynamic asymmetry suppresses the normal SD coarsening dynamics causing a viscoelastic phase separation. As a result the slow dynamic SAN intermixed primary phase becomes more viscoelastic and forms the matrix, while the less viscoelastic epoxy-rich primary phase nucleates as spheres in the elastic regime (τd < τts) as shown in the pictures. In the case of low TP content system (φSAN e 10%), equilibrium is rapidly attained in stage 4 followed by gelation, which freezes the morphology and subsequent vitrification of the epoxy phase. Here the nonvitrified SAN dispersed domains

14802 J. Phys. Chem. B, Vol. 112, No. 47, 2008 might effectively resist the shrinkage. Meanwhile for higher SAN content systems in the elastic regime, the primary epoxy nucleates as well as the secondary epoxy globules further grow by normal diffusion. According to theories of viscoelasticity, this diffusion process must accompany a volume shrinkage of the more viscoelastic phase in order to avoid the inhomogeneous deformation of the elastic network.13 This additional shrinkage due to the thermoplastic-rich phase further contributes toward the total specific volume changes during polymerization and therefore accounts for the observed high volume shrinkage beyond the SAN-10% system, that is, for the SAN-15% system (stages 4 to 5 for φSAN ) 15%) during phase separation. Theoretical studies have further suggested that such volume shrinkage processes essentially involve a relaxation of the thermoplastic dynamic asymmetry.13,46 Now, the higher the TP concentration, the slower will be the relaxation of this dynamic asymmetry during phase separation. Thus, in the case of SAN20% blend (φSAN g 20%), the thermoplastic rheological time, τts, becomes longer than that for phase separation whereby the SAN dynamic asymmetry ceases to relax. The morphology then preferentially gets frozen in the elastic regime (stage 4 for φSAN g 20%), and no volume shrinkage of the SAN-rich phase occurs. In the case of intermediate TP content systems (stage 5 for φSAN ) 15%), the volume shrinkage of the more viscoelastic SAN thermoplastic phase further continues with a coarsening of both the anisotropic primary epoxy domains and the secondary epoxy substructures (stage 5), thereby inducing a phase inversion until the gelation/vitrification commences and finally yielding a network-like partially continuous SAN-rich primary phase with local phase inversion in the continuous macrosized epoxy-rich matrix. Conclusions The in situ volume shrinkage behavior, determined by the pressure-volume-temperature analysis of poly(styrene-coacrylonitrile)-modified epoxy thermosets has been shown to be governed by the viscoelastic phase separation kinetics during polymerization, a first investigation of this kind among thermoplastic-toughened epoxy resins. In contrast to an expected linear dependency, the shrinkage is highest for the SAN-15% system with a double-phase network-like bicontinuous microstructure. Considering a linear relationship between conversion and specific volume changes, an investigation of the shrinkage kinetics of blends have shown that in the miscible region, the diluent effect due to SAN causes a progressive decrease in relative shrinkage with its increasing concentration among the blends. However, beyond phase separation, a coupling between the phase coarsening and reaction kinetics decided the shrinkage behavior. Owing to a continuous quench depth, the phase separation in these blends are accompanied by secondary phase separation, and at higher SAN contents, the various double-phase morphologies also possessed clear imprints of viscoelastic phase separation. Concomitantly, similar cross-link densities of the stoichiometric epoxy-rich phase, as revealed from DMTA studies, as well as the broad relaxation behavior of the highly intermixed SAN-ric phase estabilshed a prominent role of the thermoplastic phase toward determining the shrinkage characteristics of epoxy-DDS/SAN systems. In the case of SAN-15% system, the morphology development studies visualizes a viscoelastic phase separation being characterized by an additional volume shrinkage of the coarsegrained thermoplastic SAN-rich phase during polymerization, which further accounts for its enhanced volume shrinkage.

Jose et al. Accordingly a new volume shrinkage mechanism has been formulated to understand the volume shrinkage in SAN-modified epoxy resins. In view of practical applications, it is noteworthy that the intrinsic viscoelastic properties of the thermoplastic modifier, even though present in a very low amount, do determine the processing characteristics of these multiphase resins. Acknowledgment. The authors wish to thank ISRO, India, for the financial support to J.J. and G. Groeninckx for the DMTA and optical microscopy measurements. References and Notes (1) Giannotti, M. I.; Bernal, C. R.; Oyanguren, P. A.; Galante, M. J. Polym. Eng. Sci. 2005, 45, 1312. (2) Nguyen, L. D.; Lowry, M. A.; Stone, D. L.; Underwood, J. A. US Patent, 5,434,226,1995. (3) Ragosta, G.; Musto, P.; Scarinzi, G.; Mascia, L. Polymer 2003, 44, 2081. (4) Francis, B.; Jose, J.; Thomas, S.; Ramaswamy, R.; Rao, V. L. Polymer 2005, 46, 12372. (5) Kim, S. H.; Lee, D. W.; Chung, K.; Park, J. K.; Jaung, J.; Jeong, S. H. J. Appl. Polym. Sci. 2002, 86, 812. (6) Abad, M. J.; Barral, L.; Cano, J.; Lopez, J.; Nogueria, P.; Ramı´rez, C.; Torres, A. Eur. Polym. J. 2001, 37, 1613. (7) Yang, L.; Xiao-Su, Y; Bang-Ming, T. Chin. J. Aeronautics 2000, 13, 242. (8) Woo, E. M.; Mao, K. L. Composites, Part A 1996, 27, 625. (9) Cao, Y.; Yu, D.; Chen, L.; Sun, J. Polym. Test. 2001, 20, 685. (10) Li, C.; Potter, K.; Wisnom, M. R.; Stringer, G. Compos. Sci. Technol. 2004, 64, 55. (11) Li, W.; Lee, L. J.; Hsu, K. H. Polymer 2000, 41, 711. (12) Li, W.; Lee, L. J. Polymer 2000, 41, 697. (13) Tanaka, H. Phys. ReV. Lett. 1996, 76, 787. Tanaka, H. Phys. ReV. E 1997, 56, 4451. Tanaka, H.; Araki, T. Phys. ReV. Lett. 1997, 78, 4966. Tanaka, H. J. Chem. Phys. 1994, 100 (7), 5323. Tanaka, H. J. Phys.: Condens. Matter 2000, 12, R207. (14) Zheng, Q.; Peng, M.; Song, Y. H.; Zhao, T. J. Macromolecules 2001, 34, 8483. (15) Gan, W.; Yu, Y.; Wang, M.; Tao, Q.; Li, S. Macromolecules 2003, 36, 7746. (16) Goossens, S.; Goderis, B.; Groeninckx, G. Macromolecules 2006, 39, 2953. (17) Park, J. W.; Kim, S. C. Polym. AdV. Technol. 1995, 7, 209. (18) Tang, X.; Zhang, L.; Wang, T.; Yu, Y.; Gan, W.; Li, S. Macromol. Rapid Commun. 2004, 25, 1419. (19) Chen, J.-L.; Chang, F.-C. Macromolecules 1999, 32, 5348. (20) Alig, I.; Rullmann, M.; Holst, M.; Xu, J. Macromol. Symp. 2003, 198, 245. (21) Hameed, N.; Sreekumar, P. A.; Francis, B.; Yang, W.; Thomas, S. Composites, Part A 2007, 38, 2422. (22) Zoller, P.; Walsh, D. Standard Pressure-Volume-Temperature Data for Polymers; Technomic Publishing, Lancaster: Chicago, IL, 1995. (23) Boyard, N.; Vayer, M.; Sinturel, C.; Eree, R.; Delaunay, D. J. Appl. Polym. Sci. 2003, 88, 1258. (24) Hill, R. R.; Muzumdar, S.; Lee, L. J. Polym. Eng. Sci. 1995, 35, 852. (25) Zarrelli, M.; Skordos, A. A.; Patridge, I. K. Plast., Rubber Compos. 2002, 31, 377. (26) Ramos, J. A.; Pagani, N.; Riccardi, C. C.; Borrajo, J.; Goyanes, S. N.; Mondragon, I. Polymer 2005, 46, 3323. (27) Kamal, M. R. Polym. Eng. Sci. 1974, 14, 231. (28) Sourour, S.; Kamal, M. R. Thermochim. Acta 1976, 14, 41. (29) Naffakh, M.; Dumon, M.; Dupuy, J.; Gerard, J. F. J. Appl. Polym. Sci. 2005, 96, 660. (30) Kim, M.; Kim, W.; Choe, Y.; Park, J. M.; Park, I. S. Polym. Int. 2002, 51, 1353. (31) Francis, B.; Vanden Poel, G.; Posada, F.; Groeninckx, G.; Rao, V. L.; Ramaswamy, R.; Thomas, S. Polymer 2003, 44, 3687. (32) Tura´nyi, T. Comput. Chem. 1990, 14, 253. (33) Gottwald, B. A.; Wanner, G. Simulation 1982, 37, 169. (34) Yu, Y.; Wang, M.; Gan, W.; Tao, Q.; Li, S. J. Phys. Chem. B 2004, 108, 6208. (35) Taniguchi, T.; Onuki, A. Phys. ReV. Lett. 1996, 77, 4910. (36) Swier, S.; Van Mele, B. Macromolecules 2003, 36, 4424. (37) Jose, J.; Goossens, S.; Groeninckx, G.; Thomas, S, manuscript in preparation. (38) Swier, S.; Van Mele, B. Polymer 2003, 44, 2689. (39) Swier, S.; Van Mele, B. Polymer 2003, 44, 6789.

PVT Behavior of Thermoplastic SAN-Epoxy Systems (40) Henderson, I. C.; Clarke, N. Macromolecules 2004, 37, 1952. (41) Inoue, T. Prog. Polym. Sci. 1995, 20, 119. (42) Clarke, N.; Mcleish, T. C. B.; Jenkins, S. D. Macromolecules 1995, 28, 4650. (43) Lore´n, N.; Hermansson, A.-M.; Williams, M. A. K.; Lundin, L.; Foster, T. J.; Hubbard, C. D.; Clark, A. H.; Norton, I. T.; Bergstro¨m, E. T.; Goodall, D. M. Macromolecules 2001, 34, 289.

J. Phys. Chem. B, Vol. 112, No. 47, 2008 14803 (44) Williams, R. J. J.; Rozenberg, B. A.; Pascault, J.-P. AdV. Polym. Sci. 1997, 128, 95. (45) Tanaka, H.; Araki, T. Phys. ReV. Lett. 1998, 81 (2), 389. (46) Jianing, Z.; Zhenli, Z.; Hongdong, Z.; Yuliang, Y. Phys. ReV. E 2001, 64, 051510.

JP802015N