High Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation

The merging of two research fields of thermoplastics and thermosets (TP/TS) is creating high performance composites with high rigidity, toughness and thermal stability etc. for versatile applications (Martuscelli,1996). However, most TP/TS mixtures do not always show the synergistic effects of rigidity and toughness as one expected (Bulknall, 1985; Douglas et al, 1991; Chen, 1995; Hayes et al, 200). On the other hand, the nature sets very elegant examples of strong solid structure, e.g. animal bones and wood plants, where sophisticated inhomogeneous structures are introduced to disperse the internal stress, yet keeping some interfacial strength to eliminating the micro-cracking or fracture propagation to achieve both high rigidity and toughness (Maximilien et al, 2010). Figure 1a is an example of SEM picture of animal bone, shown in Figure 1b is an illustrative example of a calculated bi-continuous phase separated structure of TP modified TS composites which resembles the nature occurring fine material structures.


Introduction
The merging of two research fields of thermoplastics and thermosets (TP/TS) is creating high performance composites with high rigidity, toughness and thermal stability etc. for versatile applications (Martuscelli,1996).However, most TP/TS mixtures do not always show the synergistic effects of rigidity and toughness as one expected (Bulknall, 1985;Douglas et al, 1991;Chen, 1995;Hayes et al, 200).On the other hand, the nature sets very elegant examples of strong solid structure, e.g.animal bones and wood plants, where sophisticated inhomogeneous structures are introduced to disperse the internal stress, yet keeping some interfacial strength to eliminating the micro-cracking or fracture propagation to achieve both high rigidity and toughness (Maximilien et al, 2010).Figure 1a is an example of SEM picture of animal bone, shown in Figure 1b is an illustrative example of a calculated bi-continuous phase separated structure of TP modified TS composites which resembles the nature occurring fine material structures.This inspires us to improve the structures of our TP/TS systems, which will phase separate in a controllable way thermodynamically.Even very compatible mixtures of TP in TS matrices of practical importance will become incompatible during curing at certain stage, which is a prerequisite of toughening the brittle TS matrix.This process is so called cure reaction induced phase separation (CIPS) (Inoue 1995).CIPS provides an ingenious approach to realize controllable multi-scale phase morphology from nano-, micro to macroscales (Hedrick et al, 1985;Ritzenthaler, et al, 2002a;2003b), especially, the bi-continuous and/or phase inverted morphologies generated via spinodal decomposition may create some favorable structures which are critical in toughing the TS matrices (Cho,1993;Girard-Reydet et al, 1997;Oyanguren et al, 1999).
The separated phase morphologies will freeze through the matrix gelation or vitrification near Tg which become the final structures in TP modified TS composite materials.In this sense, our major task is to find some general relations between the miscibility, phase separated morphology evolvement, on one side, and the chemical structures of components and cure Thermoplastic -Composite Materials 50 processing parameters in time-temperature windows on the other side.This is still a challenging goal up to now because of our limited knowledge of phase separation and the gaps between the CIPS research status and processing practices.To achieve the goal of morphology/process design, the combined efforts of multi-scientific approaches are employed both experimentally and theoretically on many practical systems of TP/TS/hardener.The results are promising as reflected in some of our recent publications (Zhang,2008;Zhang et al, 2008a;2007b;2008c;2006d;2006e) as well as in the new composite materials achieved based on this research for the structural parts of aircraft (Yi,2006;Yi et al, 2008;Yi＆An,2008;Yi,2009).The present chapter will provide a concise summary of our up-to-date original contributions with relevant literatures in this field emphasizing the breakthrough in approaches to understand and control the CIPS during processing.

Multi-scale design of the ideal composites
In macroscopic scale various reinforced TP or TS composites are designed with various graphite, glass and/other fibrous texture and alignments, where the matrices are treated as homogeneous.It is understood more and more that the rigidity and toughness of matrices are also of practical importance.The high rigidity & toughness require multi-scale controllable inhomogeneity of composites including bi-continuous and/or phase inverted micro-network with adjustable interface (Girard-Reydet et al, 1997;Oyanguren et al, 1999).
After the concept of CIPS was proposed in 80s of last century (Inoue 1995), many experimental and theoretical investigations have been carried out to study the CIPS process in mixtures of tightly crosslinked TS matrices with ductile TP which is now assumed a promising alternative to rubber toughening approach, particularly when high value of elastic modulus, strength and glass transition temperature Tg are required.
For the continuous and/or inverse minor TP phase generated during CIPS and fixed in the thermosetting matrix, the most efficient toughening candidates are those that are able to www.intechopen.com High Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 51 plastically deform and can withstand the crack bridging/pinning or crack deflection and yielding (Pearson ＆Yee, 1993).Proper adhesion between the interfaces of TP and TS phases is very crucial for effective toughening of the TS matrix.Poor adhesion will render the TP phase premature de-bonding before expected deformations occur, while over strong adhesion between TS and TP interface will restrain the extent of ductile deformation undergone by TP phase because of the excessive constraint brought by TS phase.This constraint would limit the amount of material which could be involved in the bridging and deformation process (Williams, 1997;Girard-Reydet et al 1997;Oyanguren et al, 1999).Therefore the modifier TP must be chose in such a way that optimal affinity between TP and TS components is ascertained properly whereby CIPS can occur freely while cracking and/or delaminating can be restraint.Good combination of TP/TS with proper miscibility and cure cycle can display synergistic toughening effect.Since the mechanical properties of the toughened materials are closely related to their final morphologies, many studies have been focused on the qualitative or pseudo-quantitative thermodynamic and kinetic analysis of the phase separation mechanism (Girard et al, 1998;Gan et al 2003;Zucchi et al,2004) and morphological microstructure control as we will describe in next sections.At the same time, lots of the researches on the CIPS are limited to academic concern, the control of the morphology during cure cycle of commercial application targeting predefined mechanical properties is to be disclosed.
The well studied process is to mix the high performance TP into the matrix to form a homogeneous solution of the two components, and the subsequent cure generates the phase separated morphological structures.This process is so called in situ approach, wherein the phase separated structures are distributed evenly in the matrix.Drawbacks of the in situ approach are the substantial increase of the matrix viscosity, whereby the handling and processing window of the composites are significantly deteriorated (Cheng et al, 2009) and, also, the adhesion and modulus match of TP/TS to strength reinforcer fiber may be not as good as TS alone.Contrary to the in situ toughing, ex situ toughing is an innovative concept of a spatially localized toughening concept (Yi et al), whereby the composites are fabricated by interleaving TP containing layers and TS impreged carbon fiber layers initially.During the CIPS and the interlaminar diffusion, formed a spatial gradient of toughness and rigidity between highly toughened layers to the non-toughened, high rigidity graphite plies.Thus, the basic idea of ex situ concept sets a good example of the multi-scale design by controlled processing.As a part of the ex situ manufacturing of composite, our work is focused on the CIPS morphology evolvement, chemorheology, the time-temperature dependence of CIPS process in broad time-temperature processin g w i n d o w w i t h a c o m b i n e d a p p r o a c h o f experiment and theory.

Strategies and approaches
The curing process involves several transformation, e.g. from the viscous clear liquid to a phase separated opaque dope solution and then to three dimensional chemical gels, finally to a vitrified solid.Phase separation occurs due to the change of the mixing enthalpy and entropy which arise from the size increase of the thermosetting oligomers.Upon gelation, TGDDM: N, N, N', N'tetraglycidyl-4, 4'diaminodiphenylmethane, used as the matrix resin Diglycidyl ether of phenol A, used as the matrix resin

Detection of the cure induced phase separation process
The CIPS process can be observed in situ by different techniques, e.g.rheology (Xu et al,  2007 ; Zhang et al, 2006; Bonnet et al, 1999), small angle light scattering (SALS) and turbidity (Girard et al, 1998).Rheology can measure certain abrupt mechanical change upon phase separation, while SALS and turbidity trace the change of the optical mismatch between TP and TS rich domains during phase separation.SALS is most widely used to characterize the evolvement of domain size quantitatively in TP and TS blends (Bucknall＆Gilbert, 1989;Girard et al, 1998;Gan et al 2003).In this work, we observed the early stage of CIPS in some TP/TS systems with a modified transmission optical microscope system(TOM) (Xu＆Zhang,2007) and compared the microscopy method with the rheology and SALS approaches in the whole CIPS process.It was approved that the phase separation times can be determined by TOM.This enabled us to focus on the phase separation time/temperature dependence by considering the effects of TP molecular size and content, TP and TS structure, cure rate and cure mechanism and the stoichiometric ratio.The description of the whole cure time-temperature window of the thermoplastic modified thermosetting systems with the phase separation is of great importance in the composites processing industry.

Transmission optical approach
To observe the initial stage of phase separation, a lab-made computerized transmission optical microscope (TOM) system equipped with inversed optical design, long focusing objective lens and well controllable heating chamber was created which allows long term observation and data collection with a high resolution of 0.2 m in a wide working temperature range of RT~250°C (Xu＆Zhang, 2007).The system can assign the onset of phase separation for systems with low refractive index difference and small domain size as low as 0.2 m, this won't be succeeded by usual TOM and SALS.
The samples for TOM observation were prepared by pressing the melt between two pieces of cover glass with a thickness of about 0.2mm.The moment when the morphological structure appeared was defined as the phase separation time t ps .The values of t ps at any particular temperature are the average of five measurements with observation errors of ±3% as measured in various TP modified TS systems (Zhang,2008;Zhang et al, 2008aZhang et al, , 2007b;;2008c;2006d;2006e).
We find that the phase separation times detected with different resolution optical lens are the same within a relative error of <±1.5% as shown in Table 2 for mixtures of DGEBA/DDM/PES and DGEBA/MTPHA/BDMA/PES with different PES content.The accuracy of the TOM approach to assign the onset of phase separation was also verified in various other systems like epoxy/DDS/PEK-C system, DGEBA/DDM/PEI, DGEBA/MTHPA/BDMA/PEK-C and (Cyanate ester)/TP systems.The fact that the t ps values are independent on the magnification of the TOM proves that the onset of phase separation has been observed.If the initial phase domains were smaller than the resolution of the TOM system, one should see them earlier with higher magnification.The physical reason of limited initial domain size was explained as the nature of the cure induced spinodal decomposition by Inoue (Inoue, 1995;Ohnaga et al, 1994).Theoretically, the growing of TS size will drive the system into the unstable region in the phase diagram from

Dynamic rheological measurements
The rheological experiments were performed with a rotational rheometer with disposable parallel plates (gap 1 mm and diameter 40 mm) (ARES of TA Instrument Co.).The multiple frequency dynamic time sweeps were conducted under the isothermal conditions using the time resolved rheometric technique to collect automatically multi-frequency data as function of time (Mours＆Winter,1994).The dynamic rheological data in Figure 2 were collected at frequencies 1, 2, 5, 10 and 20rad/s, respectively with an initial strain of 0.5 %.The strain level was automatically adjusted to maintain the torque response within the limit of the transducer.
Figure 2 shows the characteristic rheological profiles together with the corresponding morphological TOM micrographs of DGEBA/MTPHA/BDMA/PEK-C mixture along curing at 100 0 C.There are two critical transitions in the plots of tan δ at different frequencies in the left layout in Figure 2. The curves cross at the times of phase separation t ps and chemical gelation t gel respectively, wherein loss tangent becomes independent of frequency and the dynamic modulus shows power law function with frequency (Chambon et al,1986;Hess et al, 1988).This critical status corresponds to the gel point and the network shows fractal structure, as we have discussed more systematically in previous work (Zhang,2008;Zhang et al, 2008;Zhang＆Xu, 2006).It was found that the critical gel at phase separation exhibits lower fractal dimension, indicating the looser structure of TP network as will be explained in Section 4.

Small angel light scattering
The CIPS of systems with enough optical mismatches were observed in situ by SALS (small angel light scattering) using HeNe laser light ( = 632.8nm).Samples were mounted in a temperature controllable hot-stage.The scattering pattern generated by the sample was visualized on screen and recorded by a CCD.The time upon which the scattering ring appears is defined as the phase separation time.SALS is a good approach to monitor the structures in micron scale in size, wherein the phase separated structures have big enough optical mismatch (Girard-Reydet et al, 1998;Yu et al, 2004).SALS approach is unable in detecting the beginning time of phase separation process in some systems we studies like TGDDM/DGEBA/DDSPEK-C systems, whereas high resolution TOP and rheology take more effective role in characterizing the morphology evolvement (Zhang, 2008;Zhang et al, 2008;Zhang ＆ Xu, 2006).Actually, the SALS method observes heterogeneous structures with sizes above micron (Girard et al, 1998), while the present TOM approach has an optical resolution of 0. 2 m and can concurrently give the direct morphology evolvement information.
It is interesting to have the CIPS systems with enough optical mismatches where the combined detection approaches can be employed giving the comprehensive in situ information.As shown in Figure 3a of the DGEBA/MTHPA/BDMA/PES system, the phase

Differential scanning calorimetry
As a usual measurement method for TS curing study, DSC (differential scanning calorimetry) is a very convenient and effective means to monitor the thermosetting monomer cure kinetic, i.e. isothermal/dynamic cure rate, cure activation energy and cure enthalpy etc (Borrajo 1995;Bonnau, 2000).DSC is also widely used in the measurement of the polymeric material glass transition temperature Tg through the detection of heat capacity transition Cp of the material during temperature scanning.In our present work, DSC technique is used to test the Tg of TS matrix and TP modifier, measure the cure kinetic at different temperatures and the cure activation energy Ea.
Figure 4 shows the dynamic exothermic and isotheral cure kinetic curves of DGEBA/MTHPA/BDMA/PEK-C10 with 0.45% initiator BDMA.As shown in the left layout, the temperature scanning curves of systems with different scanning rates.It can be observed that the peak temperatures vary with the change of scanning rate.Displayed in Figure 4b is the isothermal cure kinetics of DGEBA/MTHAP/BDMA/PEK-C10phr at 100°CBy fitting such cure kinetic curve, information about cure reaction order and cure activation energy etc can be got.
By making the temperature scanning of the raw material with different scanning rate, like 5, 10, 15, 20, 25°C/min, then analyze the exothermic curve and recognize the peak temperatures T p , the cure activation energy is calculated as below: Where β is the heating rate, Tp is the peak temperature, Ea is the cure activation energy and R is the gas constant.increases at a relatively slow pace.Thus, different-sized species present in the reaction system and the large macromolecules appears only at the later stage of conversion with broadening molecular weight distribution.Thermodynamically, phase separation takes place because of the increase of molecular size of the TS component and possible change of the interaction energy between the TP and TS species.With the growing of TS in size, the entropy of mixing decreases which is a very critical contribution to the free energy of mixing.Besides the entropy change, the mixing enthalpy generated from the interaction energy variation, which may decreases and/or increases, relying on the specific interaction between TS and TP components.When the mixing free energy rise above zero, the system is driven into unstable region and spinodal phase separation takes place, bi-continuous or phase inverted phases precipitate.Usually, the phase separation, gelation and vitrification appear sequentially, i.e., the phase separation takes place prior to the chemical gelation and, finally, vitrification as Tg increases with crosslinking going on.
With this physical scheme in mind, the thermodynamics of the cure induced phase separation in TP modified TS systems could be approximated in the framework of Flory-Huggins lattice theory (Riccardi et al, 1996;Girard et al, 1998;Ileana et al, 2004;Riccardi et al, 2004a;Riccardi et al, 2004b): Where Δg is the mixing free energy in unit volume, R is gas constant, T is absolute temperature  TS,I ,  TP,j are the volume fractions of TS i and TP j respectively, B the interaction energy density.There are two parts in eq. 2 dominating the direction of CIPS process.In the first part, the entropy decreasing with growing of TS in size is always favorable to phase demixing.The second factor comes from mixing enthalpy by the exchange interaction between TP and TS, it may increase with cure of TS (favoring demixing), or keeps constant throughout cure, or decrease with polymerization (favoring mixing).The second part in Δg is very crucial for the pattern of phase diagram, whereby various polymer combinations manifest different diagram, like upper critical solution temperature (UCST), lower critical solution temperature (LCST) and even combined UCST/LCST phase diagram.It should be mentioned that the interaction energy coefficient in Flory-Huggins theory framework is not an effective characteristic measure in describing CIPS process, since is only correlated with temperature, not enough for the description of TP modified TS with continuous changing of TS density.Because of the limitation of the theoretical framework, up to know, most of the thermodynamic analysis of the polymer phase separation phenomena are qualitatively discussed with the help of .
In present work, we study the phase separation time-temperature dependence on the chemical environment, which may arise from the stoichiometric imbalance and component structures variation.We will use interaction energy density parameter B as shown in eq. 2 instead of to explain the interesting phenomena ever observed widely in the TP modified TS systems.Based on the Hildebrand-Scatchard-van Laar HSL theory, the interaction energy density was defined as below: www.intechopen.com

Thermoplastic -Composite Materials 60
Where V ref is a random reference volume.The interaction energy density B contains two parts, the first part is the energy coming from energy exchange during mixing of different components, the second part arises from other energy change, like the extra energy that generated from the incompressibility, hydrogen bonding etc.There are several approaches to calculate or estimate the interaction energy density B, such as experimental testing the binary phase diagram, then fitting the mixing diagram with certain theoretical equation; test the mixing energy with analogous chemicals and the estimation based on solubility parameters.Experimental determinations of the interaction energy density parameter B are very burdensome, sometimes difficult to do.In our present discussion, we will estimate B value by the solubility parameter approach.The solubility approximation simplifies the complex interaction energy expression between different components to the internal pure component interaction and inter-component interaction, by assuming that inter-component interaction is a geometric mean of each pure components interaction: Where  is Hildebrand solubility parameter and is defined as: , ΔE is the molar energy of vaporization, V m is the molar volume (Hildbrand, 1950).So it is possible to calculate the B value through the solubility parameters δ TP , δ TS by equation: For small molecules it is practical to test  value directly by measure the enthalpy change during vaporization which is extensively tabulated.Considering the fact that the heat of vaporization of a macromolecule can not be measured experimentally, it is a common practice to calculate δ using the corresponding value of the solubility parameter of its monomers as substitute.Fedor, Van Krevelen et al.'s group additivity, computer simulation (Lewin et al, 2010) and thermodynamic analysis (Fornasiero et al, 2002;Utracki＆Simha, 2004) are also used widely in literature.Bicerano (Bicerano, 2002) developed a topological approach to calculate δ values based on the interaction index method, whereby group contribution is not necessary, δ can be deduced based on the polymeric chemical structure.We will illustrate the basic process to calculate δ as follow.
First, draw the skeletal structure of MTHPA in Figure 5a: J 1/2 /cm 3/2 .Then put δ into eq.4 to obtain B value.In our following discussion, all the interaction energy density parameters B are calculated based on Bicerano's approach.

Rheological characterization of the CIPS process in TP/TS mixture
TS/TP/harder is a typical asymmetric mixture.Most TP systems which are of practical importance usually have much higher molecular mass, higher modulus and Tg than that of TS monomers.The large difference in modulus/Tg and molar mass of TS and TP are responsible for the dynamical asymmetry in the relaxation and diffusion of chain which play a critical role in the early stage of phase separation process (Gan et al, 2003;Yu et al, 2004).At the beginning, TP was dissolved in the TS monomers homogenously.With cure going on, TS grows in size and the storage modulus built up higher as shown in Figure 2.
When the miscibility between TP and TS jumped into the unstable region, TP precipitated out as inverse phase, or as nodular or dispersed phase depending on the TP volume fraction.The phase separation process is manifested by the abrupt increases of viscosity and modulus profiles.Then the viscosity and modulus evolve further upwards, which grow in a step way in the chemical gelation vicinity again.As can be found that phase separation occurs before chemical gelation -the infinite TS molecular structures originated from the percolation of the TS oligomers.As displayed in Figure 2b, two critical gel points were found, the first one occurs at the phase separation time t ps which was verified by the morphological observation by TOM, the second critical gelation phenomenon appears at the chemical structure percolation point t gel which was confirmed based on DSC isothermal exothermic process, solubility test and rheological study of the neat TS cure process (Macosko＆Millerlb,1976).
It was observed in the beginning of cure, the TP/TS/harder blends have a storage modulus G' much lower than the loss modulus G", which implies the viscous essentiality of the homogenous mixtures.Upon curing, G' grows in a more steep manner than G".At the vicinity of phase separation point t ps , log-log plots of G' and G" versus ω became parallel and the loss angel factor tan became independent on frequency, implying a critical gel transition.TOM proved the appearance of the bi-continuous phase structure at this first transition.As the reaction proceeds, the shape of G' plots is further altered by the second rapid increase of G' which corresponds to the chemical gelation of TS oligomers confirmed by solubility tests and rheological monitoring on neat TS cure.After gelation, the margin of the domains became clearer gradually due to further phase separation and the coarsening of phase pattern can be observed depending on the interfacial tension and the viscosities of the phases.

The critical gel behaviors at t ps and t gel
As shown in Figure 6, at the vicinity of phase separation point t ps and chemical gelation point t gel , G' and G" involve in a parallel manner versus frequency ω in log-log plot, while loss angel factor tan become independent on frequency, implying the characteristic gel transitions, which Winter et al named as critical gel state (Scanlan & Winter, 1991).We will discuss such critical gel behaviors and its applications more systematically in following sections.
At both t ps and t gel , the dynamic scaling behaviors of storage modulus G' and the loss modulus G'' appears as shown in Figure 6, where G' and G'' versus frequency in the vicinity of t ps and t gel for system of DGEBA/MTHPA/BDMA/PEK-C 15phr isothermally curing at 100°C are displayed, the curves have been shifted horizontally by a factor A (see insert) for easier comparison.As can be seen G' and G'' show a power low relation through eq.6； '' ' Where ω is frequency; n is the critical relaxation index.
The difference between the critical gel at t ps and t gel was noticed that G' is less than G'' at t ps while larger than G'' at t gel as shown in Figure 6 and also reported in other systems (Scanlan＆Winter, 1991).The second difference is in the slop of log G', G'' vs. log ω curves at critical states.The critical gel index at both of t ps and t gel are calculated and summarized in Table 4 for DGEBA/MTHPA/BDMA/PEK-C system with PEK-C of 15, 10, 5phr respectively.The scaling behaviors occur at both t ps and t gel and over the entire experimental temperature range of 90~130°C.
By analyzing the data in Table 4, it can be found that both of n ps and n gel are independent to cure temperatures within the experimental range, more significantly, n ps is reasonably larger than n gel .The existence of the power law index n value for a critical gel is a signature of its structure of self-similarity, for which the fractal geometry can be applied.Based on percolation theory, Muthukamar suggested that, the fractional dimension d f can be represented in terms of power law index (Muthukumar, 1989): Where d f is the fractal dimension and d is the space dimension.
Taking the values of n ps , n gel and space dimension of 3 to eq.  (Eloundou,1996).The PS network is clearly looser than the chemical gelation ones.The TS crosslinking expels the less mobile high mass TP to form a sponge-like loose structure, where physical entanglement of TP macromolecules appears, causing storage modulus G' jump or tangent loss drop.This may explain why the phase separation dimension fractional d f,ps values increase with the increase of TP content.It is also noticed that the chemical gelation factional dimension d f,gel also increases with the increase of TP content.The explanation may not be easy since multi-factors involved in the cure process with PS: component diffusion mobility, interface stability, hydrodynamic effect and the competition between phase separation dynamics and crosslinking mechanism (stepwise or chainwise polymerization) etc.It is understood that the phase separation is incomplete after t ps and some second component still left in the TP or TS rich phase for thermodynamic or kinetic reason.When the TP chain length is long enough like PEK-C we used in our study with molecular size far beyond the entanglement limit size, TP will behavior in such a way that the more the TP exists in TS rich phase, the denser the gel network will be.Similar scenarios can be imaged in the TP rich phase.Based on the kinetic and hydrodynamic approaches, we can explain the second phase separation observed usually in the CIPS process as was illustrative displayed in Figure 7, where TP and TS rich phase contain second phase respectively (Oyanguren et al, 1999).Here DGEBA reacts with hardener MTHPA at the presence of initiator BDMA through the chain polymerization, wherein large quantities of monomers exist upon phase separation, which will contribute greatly to the interface instability, second phase separations were observed in different temperature at different PEK-C contents as displayed in Figure 7a and Figure 7b.Low molecular weight TP show strong mobility after t ps , also enhance the domain coarsening and second phase separation.Wherein t ps is the phase separation time, Ea(ps) is the phase separation activation energy, T is the absolute temperature, and R is the universal gas constant.
The Arrhenius type dependence of CIPS time＆temperature was observed in a various TS/TP/hardener mixtures, including the UCST (up critical solution temperature) and LCST(low critical solution temperature) types mixtures.The phase separation processes were monitored with TOM and rheology, some of them were also observed with SALS techniques.As were shown in Figure 3 and Table 3, the phase separation timetemperature dependency represented by phase separation activation energy Ea(ps) is independent on the detection means.The generality and factors which impair influence on Ea(ps) values have been discussed in detail in our published/unpublished works (Zhang,2008;Zhang et al, 2008a;2007b;2008c;2006d;2006e).It was observed that Ea(ps) values relay on the TP/TS internal chemical environment, like TS cure stoichiometric ratio, chemical structures of TP and TS, cure activation energy Ea etc, while the TP content, TP molecular size and cure rate show no obvious effect on the value of Ea(ps) in the TP/TS systems as far as TPs were employed as the minor part to toughen the TS matrix and cure reaction kinetics follows the Arrhenius type temperature dependence.
Illustrative explanation on the time-temperature dependence of the CIPS in TP modified TS systems are made as following subsections.

Effect of TP contents and TP molecular size on Ea (ps)
Based on comprehensive study of the CIPS process of various TP/TS combinations in our research group, it was found that the phase separation energy Ea(ps) is independent on the TP content, wherein the TP is the minor modifier to toughen the TS matrix, systems with TS as minor component used as plasticizer are out of the scope of our research.
Illustratively Shown in Figure 8 are the drawings of lnt ps vs. 1/T for the LCST type system DGEBA/MTHPA/BDMA /PEK-C and UCST type system Cyanate ester AroCyL10/accelerator/PEI with different TP contents, respectively.The corresponding phase separation activation energy values of Ea(ps) derived from the slope of the lnt ps vs. 1/T plots are summarized in Table 5.As depicted in Figure 8, although both of the two systems contain different TP contents, phase separation times t ps can be well correlated with cure temperature in Arrhenius type in wide cure reaction temperature.The Ea(ps ) values shown in Table 5 are not sensitive to the TP content within the experimental range.The independency of Ea(ps) on the TP level can be understood from the physical origin of the CIPS process.Although the starting TP levels are different for the studied systems, but the driving force, and the chemical kinematics are same, which is manifested by the similar activation energy Ea.
Similar trends were also observed in other systems, e.g.Epoxy/anhydride/Initiator/PES (PEI) systems, Epoxy/amine /PES(PEI) systems, Epoxy/amine/PEK-C systems, Cyanate ester (AroCyL-10)/catalyst/PES system and Bismaleimide /TP systems.In all these systems, the phase separation times may vary because of the phase separation detection means, but the Arrhenius type phase separation time/temperature dependency Keeps unchanged.The exceptions could be mentioned that at high enough TP contents, the chemical reaction kinetics will be restraint because the high viscous environment will hinder the molecules free diffusion, under such event, Arrhenius type cure kinetic is not applicable, www.intechopen.comHigh Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 67 or when the cure temperature near glass temperature, diffusion incorporated model is necessary (Kim,2002).To understand the effect of the TP content on the component chain mobility and cure activation energy, the chemorhelogical processes of some TP modified TS systems were analyzed.The following chemorheological relationship (Halley＆Mackay,1996) is applied to the TP/TS system with cure.
Wherein η 0 the initial viscosity of the system without curing，k reaction kinetic constant which is dependent to reaction temperature and cure reaction kinematic constant,η 0 and k follow Arrhenius form equation： The cure reaction process accompanied by the phase separation process of the TP modified TS systems with different TP content was monitored, the viscosity evolvement before phase separation was fitting by the Arrhenius chemorheological relation.Illustratively shown in Figure 9a are the complex viscosity * profiles of DGEBA/MTHPA /PEK-C10phr with 0.9% initiator BDMA just before phase separation in the linear coordinate, displayed in Figure 9b is the same data but in the semi-log coordinator.The complete * profiles under different cure temperature were displayed in the up right corner of Figure 9a.In Figure 9b of the semi-log coordinate, the * curves were shifted on purpose by a factor of A, which is denoted by the number beside each curve.The straight forward trends of each curve in the semi-log coordinate imply that the cure process can well be fitted by the Arrhenius relationship.The cure activation energy E k and viscous flow energy E η were calculated and summarized in    (ps) are not altered obviously by the variation of TP molecular size.Theoretically, the miscibility of TP/TS changes with the size of TP as disclosed by Buckanall (Bucknall et al, 1994) in DGEBA/PES systems, which may change the startup of phase separation times t ps , but the chemical reaction kinetics has not been altered significantly as shown in Table 7 denoted by cure activation energy Ea, wherein the effect of the variation of TP content resembles the effect of TP molecular size effect.In fact we found that some of the phase separation time/temperature data in published in literatures also followed the Arrhenius equation, e.g. from the data based on turbidity of castor oil modified epoxy system (Ruseckaite et al, 1993) and those based on light scattering of epoxy/dicyandiamide /PES systems (Kim et al, 1993).

Effect of cure rate
In our research, initiator or catalyst was employed to accelerate the TS matrix cure rate, the more the initiator/catalyst content, the fast the cure reaction will be.It was observed that phase morphologies will change in size or patterns during the variation of cure rate, whereas the phase separation time-temperature dependences keep similar.Shown in Figure 10 are the linear correlations of lnt ps vs. 1/T for DGEBA/MTHPA/BDMA/PEK-C system containing different level of initiator BDMA and AroCyL10/PEK-C system with different catalyst content of Cu(AcAc) 2 .Since the initiator/catalyst contents are very limited, so the thermodynamics of the mixtures is supposed to approximately the same.According to the cure reaction mechanism, the higher the initiator or catalyst levels, the faster the consumption of the TS monomers, and the earlier the systems will be driven into the unstable region.Although the phase separation times are different for composites with different level of initiator/catalyst while the parallel layouts of curves in both of the two diagrams indicate the similar E a (ps) values as summarized in Table 8, even though the morphologies change with variation of initiator levels (Cui et al, 1997;Montserrat et al, 1995).
The cure reaction activation energy E a of systems with different level of initiator/catalyst was determined by Kisinger approach as shown in Table 8

Cure activation energy barrier and chemical environment effects on Ea(ps)
As depicted in Section 5.2 and 5.3 that Ea(ps) values are independent on the cure rate and also irrespective to the TP content and TP molecular size in the usual systems as far as TPs are used in minor part as toughening agents.But Ea(ps) varies with the cure activation energy barrier Ea and chemical environments, both of which will be discussed in the following subsections.

Cure path and cure activation energy effects on Ea(ps)
Shown in Figure 11 and Table 9 is the phase separation time/temperature relations in the range of 120-200℃ for systems of BMI/DBA/PEK-C10phr and BMI/DBA/PES 10phr respectively, together with the gel time/temperature data.It can be seen that the slope of Arrhenius plot of the phase separation and gelation in the 120-170℃ range is lower than that in the 180-200℃ range in both of the two systems.Correspondingly, the phase separation activation energy E a (ps) values in the range of 120-170 ℃ are lower than those in the range of 180-200℃.The change of temperature dependency on the phase separation times can be attributed to the complexity of the cure reaction between BMI and DBA.Although lots attempts have been made to elucidate the curing mechanism of BMI/DBA systems, these have been impeded by the complexities of the reaction multiple paths of the reaction.The following reaction types have been proposed to be involved in the curing process: ENE, Diels-Alder, homopolymerization, and alternating compolymerization.Allyl phenol compounds are expected to co-react with BMI to give linear chain extension by an ENE-type reaction in lower temperature range of 120-170℃ and show a lower cure activation energy barrier, and this is followed by a Diels-Alder reaction at a high temperature range of 180-200℃ with bigger cure activation energy (Mijovic＆Andjelic,1996;Shibahara et al, 1998;Rozenberg et a,2001;Xiong et al, 2003).(Fisch et al,1956;Fisch,1960).As shown in Table 10, both the DGEBA/MTHPA/PES and DGEBA/MTHPA/PEI systems without initiator BDMA show higher cure activation energy Ea values than the systems containing 0.9%BDMA, accordingly higher Ea(ps) values were observed.It should be pointed that in DGEBA/MTHPA/PEI system which shows UCST phase behavior, the Ea(ps) of composition without catalyst BDMA is much higher than the composition with BDMA.
The cure activation energy changes in the range of ⊿Ea=14.5kJ .mol -1 , while phase separation activation energy show the discrepancy of⊿Ea(ps)=22.3kJ.mol -1 , which is much higher than the former.It can be assumed that the significant increase of Ea(ps) of DGEBA/MTHPA/PEI composition is not only generated from the change of chemical reaction energy barrier Ea, while part of the increase of ⊿Ea(ps) may come from the increase of the cure reaction temperature.In DGEBA/MTHPA/PEI composition, because the absence of initiator BDMA, the cure temperature has been elevated quite lot for cure efficiency, as was reported, DGEBA/PEI is a system of UCST phase diagram (Bonnaud et al, 2002), the elevation of cure temperature will increase the quench depth which possibly causes the increase of Ea(ps).The physical scheme of the correlation and even similar value range implies the CIPS induced by the curing.

Ea(ps) dependence on stoichiometry in UCST TP/TS/hardener systems
Besides the chemical reaction energy barrier dependence of Ea(ps), it was observed that the Ea(ps) values also vary with the chemical environments, like stoichiometric balance, TP and TS monomer structures.We will discuss how the chemical environment factors will alter the phase separation activation energy in details.In this subsection, we only talk about the phase separation time-temperature dependence in TP modified TS systems with UCST type phase diagram, which can be explained quite well with the interaction energy density parameter in our research scope.
Shown in Figure 12 and Table 11 are the time-temperature dependence variation in DGEBA/DDM(DDS)/PEI10phr and DGEBA/MTHPA/BDMA/PEI10phr systems with different chemical stoichiometric ratio r.As can be seen in DGEBA/DDM/PEI system which shows UCST type phase behavior (Bonnaud et al, 2002), the phase separation activation energy values Ea(ps) of it increase with r, when r=1.5 the system even shows no phase separation.While in DGEBA/DDS/PEI system, Ea(ps) changes with r in an opposite way.In view of chemical reaction energy barrier, E a does not change with the stoichiometric ratio r for both of the DGEBA/DDM/PEI and DGEBA/DDS/PEI systems as displayed in Table 11 by value of Ea.The changes of Ea(ps) possibly originate from the variation of chemical environments with the change of stoichiometric ratio, e.g. the miscibility between components is altered with the increase of r values, which in turn influence the Ea(ps) substantially.We calculated the interaction energy density B values for different combination as shown in Table 12.It can be seen that the miscibility of PEI with DDS is worse than that it with the cured DGEBA/DDS.So it can be hypothesized that the more of www.intechopen.com High Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 73 DDS, the poorer the miscibility between TP and cured TS and the easier for the onset of phase separation with lower Ea(ps).The inverse scenarios were observed in DGEBA/DDM/PEI systems, wherein DDM is unfavorable for demixing of the components of TP and TS, DDM shows better affinity to PEI than the cured DGEBA/DDM matrix, so the more the DDM content, the better the compatibility between the components and the bigger Ea(ps) value it exhibits.Similar phenomena were observed in DGEBA/MTHPA/PEI systems which can also be explained in the frame of interaction energy density parameter.

TP and TS structure effects on Ea(ps) in UCST type TP/TS/hardener systems
Besides the chemical stoichiometric balance, TP or TS structure variation all take substantial effects on the phase separation time-temperature dependence.As shown in Table 13, the phase separation activation energy Ea(ps) varies seriously with the poly(ether imide) structures changing, either in epoxy/amine and epoxy/anhydride matrices, or in cyanate ester systems.PEI and PEI 1 have different chemical structures as shown in Table 1, both of them show different miscibility with matrices of DGEBA/DDM and DGEBA/DDS respectively.In DGEBA/DDM matrix, PEI 1 show higher affinity to the matrix than PEI does, so it can be understood that even no phase separation was observed in DGEBA/DDM/ PEI 1 for the excellent compatibility; while in DGEBA/DDS matrix, miscibility trend reverses, PEI is more favorable to phase mixing than PEI 1 as denoted by the higher Ea(ps) value of 78.6 k J .mol -1 .Similar phenomena were observed in the matrix of DGEBA/anhydride/PEI and AroCyL10/PEI respectively.In AroCyL10/Cu(AcAc)2/PEI1 system PEI1 even can't be dissolved in the matrix homogenously within the experimental condition for the poor miscibility of the TP and AroCyL10 monomers, the bigger interaction energy density parameter B in Table 15 gives the reasonable interpretation.In all these mixtures, the variations of Ea(ps) because of PEI structure dissimilarity can be well interpreted by the interaction energy density theory.The phase separation time-temperature dependencies also changed with the variation of TS monomer structures, as shown in Table 14, with the increase of epoxy molecular weight, Ea(ps) changes substantially.Here all the epoxy monomers have the structures referring to Table 1, only differing in the number of repeating unit n, the numbers beside "E" in Table 14 donating the epoxide value of the epoxy monomer, like E54 means monomer contains 0.54mol epoxide functional group per 100g DGEBA., The smaller epoxide value, the bigger the monomer size.As can be seen that the phase separation activation energy Ea(ps) decreases with the increase of molecular size of epoxy monomer DGEBA.Interaction energy density parameter B values between the cured DGEBA/DDM matrix and PEI were www.intechopen.com High Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 75 calculated and summarized in Table 15.As was shown that the growth of DGEBA monomer size give higher B value implying the widening of miscibility mismatch between TP and TS components.This is apparently contradictory to the finding in above section wherein TP content and size show limited effect on Ea(ps) values.It is true that the increase of DGEBA molecular size will change the phase diagram location as were observed, but except the DGEBA size effect on phase behavior, there is another favoring factor for phase separation is the volume fraction change of cure agent DDM because of the increase of DGEBA size.Bigger DGEBA gives higher molar volume fraction which means small volume fraction remained for cure agent DDM.DDM is a component which is very favorable of phase mixing, the less of DDM volume is, the worse the miscibility of TP and TS will be.

Stoichiometry and TP and TS structure effects on Ea(ps) in LCST type TP/TS/hardener systems
The interaction energy density theory works well in rationalization of the phase separation time-temperature dependence on the chemical environment change: ether coming from stoichiometric imbalance, or TP and TS structures in the TP/TS systems with UCST phase behavior.But such theory is not universal in TP/TS systems with LCST phase diagrams.
Shown in Table 16 are the phase separation activation energy Ea(ps) values of DGEBA/DDM/PES (Bonnaud et al, 2002) and DGEBA/MTHPA/PES with different stoichiometry r both of which show LCST type phase diagram.As was shown that both of Ea(ps) values increase with the descending of r value.The interaction energy density parameter B of PES/harder and PES/(cured TS) are calculated and displayed in Table 18.It was found that both of the two hardeners of DDM and MTHPA show higher affinity to PES than the corresponding cured resin, while the experimental observations did not support the theoretical prediction, the higher the hardener content was, the smaller the phase separation energy barrier appeared.The apparent paradoxical phenomenon comes from the limitation of the present thermodynamics in depicting polymer mixture phase separation process, which will be explained in following content within this section.Besides the stoichiometric effects on the variation of phase separation activation energy Ea(ps) in various TP modified TS systems which show LCST phase diagrams, the chemical environments effects arising from TP and TS structure alterations were also studied.Here we mainly talk about some representative systems, like DGEBA/DDM/TP, DGEBA/MTHPA/BDAM /TP and Cyanate ester system of AroCyL10/TP (Hwang et al, 1999) as in Table 17, wherein TP structure varies in repeating units.
Based on interaction energy density parameter calculation between different combinations as shown in All these experimental and theoretical discrepancies arise from the limitation of the present cohesive energy theory of interaction energy interaction density.Indeed, volume fraction change during cure, strong hydrogen bonding and strong polar interaction are not included in our calculation.These effects could be critical for the phase behavior, especially in the LCST systems, so further theoretical and experimental efforts are to be made.Up to now, all the theoretical analysis works only well in UCST systems where hydrogen bonding and other special energy effect are absence.But for the LCST systems, as have been found in this subsection, most of the experimental results contradict the theoretical hypothesis.

Time-temperature transformation diagram with phase separation
As was known that all the processing of thermosetting composites are carried out in wide time/temperature space as presented by TTT (time-temperature-transformation) diagram (Enns＆Gillham,1983;Gillham,1986).Phase separation usually occurs before the chemical For TP modified TS processing, the cure time-temperature routine has a great impact on the morphology generated during cure further to the final material properties (Inoue, 1995;Williarms, 1997).To get TP/TS material with desired structure and properties, it is necessary to design the cure time-temperature processing routine.The time-temperature dependence of gel and vitrification is well illustrated by the kinetic approach for the systems with unambiguous cure kinetics (Wisanrakkit ＆ Gillham,1990).It is desired to have the phase separation zone defined in the TTT diagram for TP/TS composite processing.So far only schematic diagram appeared in literatures (Williams, et al. 1997).This situation is obviously related to the issue of determination of phase separation time in broad temperature range.diagram).Figure 13 displays the example of experimental determined TTT-PS diagram, which defines the main events taking place during the curing of TP modified TS with the presence of CIPS in whole time-temperature window.Obviously the TTT-PS diagram may be of great importance for composites processing where morphology and structure control is of interests.It is shown in Figure 13 that in the measured Epoxy/DDS/PKE-C system with a LCST phase behavior, the phase separations take place before chemical gelation throughout the cure temperature range employed, and the phase separated structure was even observed at temperature where the epoxy matrix started degrading with volatiles coming out.In some TP/TS mixture, e.g.UCST systems, phase separations could only take place in certain time-temperature space within the up-merging curves of phase separation and gelation.Even in this area PS is not guarantied, because of the PS dynamics at early critical stages and the pattern developing dynamics.In Figure 13, the phase separation and gelation time-temperature curves in the semi-log coordinator at low temperature range are extrapolated in a convergent direction intentionally to depict to what will happen at low temperature.It can be assumed theoretically that the phase separation time-temperature curve and the gelation time-temperature will converge below certain temperature.This is because our present mixture shows LUST phase behavior.The decrease of temperature is favorable of phase mixing which possibly will contribute to a higher phase separation activation energy; on the other hand, the cure reaction will become difficult because the restrain on component mobility generated by the high viscosity accompanying the temperature decrease.

Dynamic theory for morphology evolvement
The thermodynamic theory reveals the possibility of phase separation (PS), the phase separation possible mechanism like spinodal & bimodal decompositions, and the conditions of PS emerging, while it does not answer how the PS pattern develops, e.g. the phase separation dynamics.To understand the phase pattern evolvement ( for example in Figure 7) and control the morphology in processed product, the phase separation dynamics has been intensively studied in the past three decades from both experimental and theoretical viewpoints (Onuki,1986;Onuki ＆ Taniguchi 1991;Tanaka, 1997;Onuki ＆ Taniguchi 1997).
Experimental observations of CIPS disclosed spectacular variable schemes of morphology changes depending on the composition and process parameters.Good news is that owing to the advances of condensed matter physics a unified scheme became possible to explain the complex phenomena.From the concept of dynamic universality of critical phenomena, phase separation phenomena have been classified into various theoretical models.For example, phase separation in solids is known as the "Solid Model or Model B" (Hohenberg ＆ Halperin, 1977), where the local concentration can be changed only by material diffusion, while phase separation in fluids is known as the "fluid model or model H" by both diffusion and flow (Doi＆Onuki, 1992).It has been established that within each group the behavior is universal and does not depend on the details of the material.The classic theories consider the same dynamics for the two components of a binary mixture, which we call "dynamic symmetry".However, such an assumption of dynamic symmetry is hardly valid in ''complex fluids''.For the example of TP/TS mixtures, their molecular mass, diffusivity and mobility are very different.For these ''dynamic asymmetric'' systems one needs to consider the interplay between critical dynamics and the slow dynamics of polymer itself which was in so called ''Viscoelastic Model'' (Tanaka, 1997).
Some researches have been focused on the CIPS process starting from homogenous solution using Onuki's Elastic solid model (Zhang, 1999) or "two fluid" model or Tanaka's viscoelastic model (Tanaka, 1997).Which model should one choose depends what dominant physical parameters are involved in the system.For example, at the early stage of spinodal Simulation is often a necessary approach to "image" the events take place in the black box where we can't tough directly by any means for its complexity, while numerical solutions work successfully.
In our present work, the modified Elastic Model of Onuki (Onuki＆ Taniguchi 1991) was employed, the phase separation dynamical equation was numerically solved, using the socalled cell dynamical system presented by Oono and Puri (Oono ＆ Puri 1988).
The free energy functional of the system F is given by (Onuki, 1997;Zhang, 1999): Wherein f () +1/2|▽| 2 is the free energy of unit volume in the Flory-Huggins average lattice theory or Ginzberg-Landau theory, a is a parameter represents the coupling constant of concentration gradient and elastic field, u is the deformation vector.In linear viscoelastic region, the bulk elastic energy originates from the bulk modulus K and symmetric conformation tensor Q, which are irrespective to volume fraction , and shear energy comes from the shear modulus μ and symmetric conformation tensor Q: Wherein d is the spatial dimension  and concentration  is supposed to be linear as: For simplicity, all the constant part in the free energy, the free energy functional can be expressed as And with equation 18, we can construct the TDGL time dependent Ginzberg Landau equation： 22 2 ,, [ is the modulus of TS phase, it changes with the cure time for the increase of TS size, for most of the TS monomers reaction through the second order reaction.According to the chemorheology of curing, one can get the time-temperature dependence of dynamic viscosity change.Therefore, in our simulation, we use the expression g E ～(TS conversion) The Ginzberg-Landau free energy is simplified as following: By numerical calculation of governing equation of 16, 20 and 21 we can get the qualitative information about what happened in the intermediate and later stager of CIPS using the modified "Solid Elastic Model" wherein modulus mismatch between TP and TS varied along time.Shown in Figure 14 is a simulation result for the systems with "Viscoelastic symmetric" modulus.The volume fractions of TP component is =0.5.As can be seen in the "dynamic viscelastic symmetric" system, bi-continuous phase was observed at early stage and some coarsening of the networks appear with time until at last TP network morphologies freeze , because of full gelation of TS phase In some TP/TS/Hardener systems when the TP molecular is not high enough or TS monomers show high hydrodynamic effect upon phase separation, second phase separations were always observed.By the present "Elastic Solid Model", such scenario can't be incorporated, wherein in the ''Two Fluid Model'' coupled with chemical reaction works www.intechopen.com High Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 81 better.But for the composites material fabrication, the final morphology though too much interface sharpening and phase coarsening via flow may not a favorite process.
In the real processing of composites material during fabrication of large parts, like composite aircraft wind and fuselage, incomplete diffusion with elastic misfit between the components does exist, wherein the TP and TS are interlaminated during RTM (resin transfer molding).Based on ex situ toughening concept, the thin TP layers are periodically interleaved the TS/graphite prepreg plies, the respective TP and TS components were separated.After the wet prepregs are laminated, the interaction between TP and TS phase and the diffusion couples with the cure reaction and phase separation start (Yi,2006;Yi et al, 2008;Yi＆An,2008;Yi, 2009).In such delicate, hard to monitor process, the computer simulation can play role.We will discuss such simulation process with modified Onuki "Elastic Solid Model" in subsequent section.In Section 6.1 we simulate what happens in the in situ phase separation process, wherein the phase separated structures are distributed spatially evenly in the matrix, phase inversion mechanism was clearly displayed.Here, we will use the same phase separation thermodynamic and similar kinetic frame to monitor the events happened in ex situ toughening process, where interleaved TP and TS layered are layout periodically in spatial direction.By a similar two dimension CDS process , a linear concentration gradient is assumed along y-coordinate, our computational space is graded in square of size 0<x<L and 0<y<L, wherein L is the dimensionless square grid length, the concentration gradient is shown as Other thermodynamics and phase separation dynamics descriptions keep the same as Section 6.1 wherein in situ phase separation process is calculated by the CDS approach.In the present system, Shown in Figure 15 are simulated morphology evolving process with varying TP and TS modulus mismatch value of Ge.It can be observed that the higher the asymmetry is, the more inverted structures formed in the interface between TP and TS matrices.In a Ge=100 system, bi-continuous and inversion structures appear throughout the direction perpendicular to the layer plane, and also the morphology distribution varies along the same direction.

www.intechopen.com
Thermoplastic -Composite Materials 82 Shown in Figure 16 is the experimental SEM (Scanning Electronic Microscope) observation on the fractured interface between the epoxy BEP/DDS matrix and thermoplastic PEK-C layer.The morphology size and pattern vary periodically in direction perpendicular to ply direction.The simulation scenarios fit the experimental results qualitatively.In the future, improvements of present model with incorporation of spatial temperature ingredient and cure cycle possibly will give more clues to how the morphologies evolve during the real composite manufacture process.

Concluding remarks
In the past three decades, numerous experimental and theoretical achievements dealing with reaction-induced phase separation (RIPS) of polymer systems were driven by versatile applications in spite of the complex nature of the subject from the theoretical point of view.Cure induced phase separation (CIPS) is an important part of RIPS due to its innovative applications to composite processing, or more generally, to the innovations of multi-phase polymers.Instead of fight with interfacial tension in polyblends, CIPS processes involve the creation of phases with granular or network morphologies, the interfacial sharpening & coarsening according to the balance of diffusion and convection of viscous mobile systems, the termination of PS or the secondary PS affected by TS gelation and vitrification.It seems to be a disadvantage to allow the morphology influenced by so many factors, but it's also an advantage to have many means to create the structure we want, if we know the control rules.The intention of this review was using our practice to provide a qualitative (or partially quantitative) basis for readers to rationalize the numerous factors in CIPS process to achieve the desired morphologies and composite properties.
In the first section it was shown that: a) the purpose of modifying a high performance TS polymer composite is calling on innovative multi-scale matrix design; b) to achieve high toughness and, at the same time, high rigidity, the controllable micro-inhomogeneity including bi-continuous network with adjustable interface could be a good option for composite matrices; c) cure induced phase separation of the mixture of thermosetting resin (TS) with engineering plastics (TP) provide promising approach to do so.d) a gradient design of PS fitted in layered composite can even more optimize its performance.

Thermoplastic -Composite Materials 84
To detect and study CIPS process comprehensively, one need multi measuring means, including necessary experimental innovations, which are summarized as: a) to detect the phase separation time (PS) in early stage, a patented transmission optical morphology analyzer (TOM) allows determining PS time and tracing the image evolvement of microphase separation at high T and long time; b) comparative studies using OM, SALS & dynamic rheology found two critical transitions: TP gelation by PS and a TS gelation, revealed the loose fractal network structure of TP at PS and TS fractal network at "critical gel states" and determined the network fractal dimensions.
Based on these "on-line" experimental approaches we were able to study the cure reaction of TP/TS with PS in broad time-temperature range.The temperature dependence of PS time was found experimentally to obey Arrhenius type rule for various TP/TS/hardener systems and PS activation energy may be defined and its relations to many compositional parameters were clarified experimentally.We were able to build the Time-Temperature-Transformation diagram including phase separation (TTT-PS) experimentally, which is of essential importance for TP/TS composite processing.There is still long way to go to answer the following questions thoroughly: why the CIPS happens?What governs the phase separated morphology development?How to predict the appearing of CIPS from the chemical formulation and processing parameters?However, some progresses in this direction can build a promising picture: a) like all mixing thermodynamics, Flory-Huggins mean field theory can be performed for CIPS.Instead using interaction parameter , the interaction energy density B was employed with more reasonable physical meaning and good relation with cohesive energy and, thus, chemical structures of the components; b) for the dynamically very asymmetric system, the viscoelastic phase separation theory by incorporating the reaction kinetics into time dependent Ginsburg-Landau (TDGL) equation and cellular-automata simulation can simulate the morphology development.If interlaminate diffusion is cooperated, the gradient morphology in layered composite can be simulated; c) our practice and the theoretical background are well situated in a multiphysics frame.Either thermodynamic or dynamic theory is all in mesoscopic level, with unified rules and leaving lower level chemical reactions and structures in parameters.This paradigm should work and be proved further in our efforts to build better composites.
Most of our researches were carried out under the support of National Key Basic Research Program (973 Program, 2003CB615600)

Fig. 1 .
Fig. 1.Skeletal structure of animal bone and simulated bi-continuous phase structure of some composite material.(a) Bone structure.(b) Bi-continuous phase of a composites in phase separation

Fig. 4 .
Fig. 4. DSC dynamic and isothermal cure kinetic monitor of DGEBA/MTHPA/BDMA/PEK-C10phr systems.(a) Exothermic curves.(b) Isothermal cure kinetics curve, r=1 at 100°C Fig. 5. (a) Hydrogen suppressed graph obtained by omitting the hydrogen atoms and connecting all of the remaining atoms (vertices of the graph) with bond(edges of the graph)

Fig. 12 .
Fig. 12. Plots of lnt ps vs. 1/T for TP/ TS/hardener systems with different stoichiometry r values www.intechopen.comHighPerformance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 77 gelation, which is always accompanied by the change of viscosity or modulus as shown by Figure2in Section 3. Following phase separation is the chemical gelation of TS rich phase, which retards the resin flow and limits the coarsening of the dispersed phase domain.Upon vitrification, curing ceased totally because the freezing of the component chain.

Fig. 13 .
Fig. 13.TTT diagram depicting occurring times for phase separation, chemical gelation and vitrification for Epoxy/DDS/ PEK-C 15phr systemBased on our efforts on the phase-separation time-temperature dependence relationship and the TS matrix cure kinetics, it is now convenient to build the Time-Temperature Transformation Diagram (TTT) accompanied with phase separation process (TTT-PS diagram).Figure13displays the example of experimental determined TTT-PS diagram, which defines the main events taking place during the curing of TP modified TS with the presence of CIPS in whole time-temperature window.Obviously the TTT-PS diagram may be of great importance for composites processing where morphology and structure control is of interests.It is shown in Figure13that in the measured Epoxy/DDS/PKE-C system with a LCST phase behavior, the phase separations take place before chemical gelation throughout the cure temperature range employed, and the phase separated structure was

Fig. 14 .
Fig. 14.Morphology evolvement of two components mixture with symmetric moduli.t donating the dimensionless time in the temporal space.Volume fraction is 50% where in  is the volume fraction of minor component TP.

Fig. 15 .
Fig. 15.Effect of modulus mismatch value Ge on the morphology distribution in direction perpendicular to ply

Table 1 .
Chemical structures of components employed in the present research www.intechopen.comThermoplastic-Composite Materials 54 High Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 55 the stable state, which will result in the decreasing of concentration fluctuation wave length and the rising quenching depth theoretically.But the successive increase of the quenching depth has not changed the regular concentration fluctuation sine wave length at intermediate cure rate, as observed in most of the CIPS processes with spinodal decomposition, where very regular and fine bi-continuous or inverse phase separation were observed, computer simulation also verified such phenomenon.Possibly the increasing quenching depth depresses the domain coarsening arising from interfacial tension and hydrodynamics in the early stage of CIPS.Experimentally, in the initial stage of phase separation, the domain size keeps constant because of the compromising effect of domain coarsening and quenching increasing.So it seems no size variation appears in the early stage of CIPS, the concentration fluctuation wave length kept constant, and the limited value of initial fluctuation wavelength becomes the periodic size of phase domains in the early stage of phase separation before domain growth driven by interfacial tension and hydrodynamic force so the early stage of the spinodal decomposition can be observed at similar t ps using different TOM magnification. www.intechopen.com

Table 3 .
Ea (ps)values of DGEBA/MTHPA/BDMA/TP systems detected by different means High Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 61 Then based on the simple atomic index  of skeletal structure, non-hydrogen vertexes and the atom valence index v www.intechopen.comatomic +5*N group )(N atom and N group correction)： 7, we obtained the value of fractional dimension at phase separation point d f,ps and chemical gelation point d f,gel www.intechopen.comHigh Performance Thermoplastic/Thermosetting Composites Microstructure and Processing Design Based on Phase Separation 63 respectively, both of which are summarized in Table 4.As was observed that bigger n gives a smaller fractional dimension implying looser crosslinking network.The values of d f,gel are within the range of most single TS gelation, as reported in the other networking systems

Table 4 .
Fig. 6.Frequency response of G' (the solid symbols) and G" (the open symbols) at the vicinity of t ps and t gel for DGEBA/MTHPA/BDMA/PEK-C 15 phr cure under 100°C Power law index at t ps and t gel under different temperatures for DGEBA/MTHPA/BDMA/PEK-C system 8. Effect of TP content on the relation of lnt ps vs. 1/T of DGEBA/MTHPA/BDMA/PEK- a: The reaction activity energy of both neat thermosetting monomer were determined by the gel point method Table5.Ea(ps) of DGEBA/MTHPA/BDMA/PEK-C and (AroCyL10)/Cu(AcAc) 2 /PEI with different TP content Table 6 respectively.It can be found in the present TP loading levels, E k and E η keep constant with the varying TP content.The results were confirmed in other characterization approaches, e.g.gelation and exothermic methods and in other systems.Flow viscosity itself is sensitive to the dragging effect exerting by TP macromolecules, while the viscous activation energy E η value is related to smaller scale of TP chain segments.Considering the related mechanisms of E η and Ea (ps), it is understandable that Ea(ps) keeps constant regardless the variation of TP content in low TP contents.The constant E η values of the systems with different PEK-C loading level indicate that TP has no effect on the diffusion mobility of the TS oligomers during phase separation.

Table 8 .
as Ea.It was observed that when initiator content is ≥0.9%, cure reaction activation energy Ea values are similar in DGEBA/MTHPA/BDMA/PEK-C systems of different BDMA content.This is different from Montserat's finding(Montserat et al, 1995), which said that Ea generally decreased with increase of initiator level.It is presumably because the initiator content employed in our present systems is higher enough to catalyze the chain-wise copolymerization of DGEBA and MTHPA during the temperature scan, and the higher Ea values in Montserat's systems probably arise from the uncatalyzed copolymerization for the deficiency of initiator BDMA.Similar results were observed in AroCyL10/Cu(AcAc) 2 /PES systems.Ea (ps) for Epoxy/hardener/TP and Cyanate ester/TP systems with different initiator /catalyst content

Table 10 .
Ea(ps)values for DGEBA/MTHPA/PES and DGEBA/MTHPA/PEI systems without and with BDMA The effects of cure reaction energy Ea(ps) were also observed in DGEBA/MTHPA/PES and DGEBA/MHTPA/PEI systems as displayed in Table10, wherein the initiator BDMA was employed or not.The reactions of DGEBA and MTHA with or without initiator BDMA are in different path.As reported that DGEBA and MTPHA reacts through the chain mode in the presence the catalyst such as tertiary amine BDMA with low activation energy compared to the stepwise polymerization of DGEBA and MTHPA

Table 12 .
Interaction energy density B values between different combinations

Table 14 .
Ea (ps)for epoxy/DDM/ PEI systems with different epoxy monomer structures Table18, it can be seen that in DGEBA/DDM/TP systems, PEK-S shows highest affinity to the matrix, then PSF and PES show dropping affinity to matrix in sequential.While the measured Ea(ps) values have not shown the corresponding theoretical sequence.It can be seen that DGEBA/DDM/PES-C with highest miscibility did show biggest phase separation energy barrier, e.g.no phase separation was observed upon completion of the cure.But DGEBA/DDM/PES shows a higher Ea(ps) value than that of DGEBA/DDM/PSF system which is contrary to theoretical prediction.Inconsistence was also observed in other systems, e.g.DGEBA/MTHPA/PEK-C systems, Ea(ps) doesn't show a monotonic relationship with the calculated B value.

Table 17 .
Ea(ps)variation for the change of TP structure in matrix of TS/hardener/TP mixtures

Table 18 .
Calculated B for PES, PSF and PES-C with TS components