High Temperature Epoxy Foam: Optimization of Process Parameters

For many years, reduction of fuel consumption has been a major aim in terms of both costs and environmental concerns. One option is to reduce the weight of fuel consumers. For this purpose, the use of a lightweight material based on rigid foams is a relevant choice. This paper deals with a new high temperature epoxy expanded material as substitution of phenolic resin, classified as potentially mutagenic by European directive Reach. The optimization of thermoset foam depends on two major parameters, the reticulation process and the expansion of the foaming agent. Controlling these two phenomena can lead to a fully expanded and cured material. The rheological behavior of epoxy resin is studied and gel time is determined at various temperatures. The expansion of foaming agent is investigated by thermomechanical analysis. Results are correlated and compared with samples foamed in the same temperature conditions. The ideal foaming/gelation temperature is then determined. The second part of this research concerns the optimization of curing cycle of a high temperature trifunctional epoxy resin. A two-step curing cycle was defined by considering the influence of different curing schedules on the glass transition temperature of the material. The final foamed material has a glass transition temperature of 270 °C.


Introduction
Polymer-fiber composite materials are widely used in many applications as they combine a low weight with satisfying mechanical properties. A decrease in weight is a recurrent requirement in industry, particularly in the aerospace field where fuel cost has a significant impact. Producing foamed materials is a way to achieve mass reduction.
There are different methods to produce foam: decomposition of chemical substance with gas formation, under heating [1][2][3][4]; bubble creation by vaporization of a liquid [5][6][7][8][9][10]; gas introduction in the resin under high-pressure follow-up with expansion when pressure drops [11][12][13]; and mechanical beating [14,15]. The low boiling temperature liquids used for the expansion by vaporization are usually hydrofluorocarbon, which is problematic because they contribute to global warming. Other methods allow lightening a structure, for example by filling a matrix with microballoons to produce syntactic foam [16][17][18], or with water-soluble crystals in a net immersed in water so that holes supplant the crystals [19,20]. However, these processes do not involve expansion, so they are not relevant for our application.
chosen. Indeed, the high energy chemical bonds of the aromatic cycle are responsible for the properties' stability at high temperature. Because it is an aromatic amine, diaminodiphenylsulfone (DDS) has been selected as curing agent. Moreover, TETM is known to have superior thermal oxidative stability over other types of epoxy resins [40]. Aronhime and Guillham estimated theoretically the glass transition temperature of TETM/DDS formulation at about 352˝C [41].
For the reasons previously exposed, the substitution of phenolic resin in the Roxalte was the first purpose of this study. Subsequently, different foaming processes were considered. Starting with the blowing agent used in the Roxalte, i.e. expandable microspheres, seemed logical. Indeed, as far as we know, the association of the preselected epoxy formulation with these microspheres has not been studied. Further developments on other foaming processes with different foaming agents will be the aim of another paper.
Wang et al. [42] produced epoxy/expandable microsphere foams in two steps. The epoxy was diglycidyl-ether of bisphenol-A. A precure step was first done on the epoxy/curing agent/blowing agent formulation in order to increase the viscosity by starting the polymerization process. The foaming process occurred during the second step where the formulation was put under a higher temperature. This process intended to improve the cell size distribution in the foam. In our study, we propose manufacturing the foam in one step only, by tuning the microsphere's foaming kinetic to the resin gelation kinetic. The crosslinking reaction and the expansion of microspheres take place simultaneously.
Obtaining an expanded material depends on two distinct phenomena, the foaming agent expansion on one hand and the resin gelation on the other hand. These two phenomena are independent of each other but both depend on time and temperature.
First, we propose a complete physicochemical study of expandable microspheres, in order to understand their behavior. In particular, the expansion is studied by thermomechanical analysis and correlated to the gel time of the epoxy formulation with the aim of identifying the ideal process temperature and producing the optimized foamed material. The optimization of the epoxy formulation curing cycle has been realized to achieve the manufacture of enhanced expanded foam.

Materials
The triglycidyl ether of tris(4-hydroxyphenyl)methane (TETM) and the diaminodiphenylsulfone (DDS) were kindly offered by Huntsman Advanced Materials (Basel, Switzerland). Their commercial names are Tactix 742 and Aradur 976, respectively. The foaming agents from the Expancel®range were supplied by Rossow Industries (Gennevilliers, France). The Expancel®950 DU 80 was selected (Particule size 18-24 µm; Density ď12 kg/m 3 ). The microsphere copolymer is made of acrylonitrile, methylmetacrylate, and methacrylonitrile and the volatile liquid inside is an isooctane-isopentane mix. Its technical data sheet also reveals the presence of magnesium hydroxide. The chemical structures of TETM and DDS are presented in Figure 1. triglycidyl ether of tris(4-hydroxyphenyl)methane (TETM), a trifunctional epoxy with high aromatic ratio, has been chosen. Indeed, the high energy chemical bonds of the aromatic cycle are responsible for the properties' stability at high temperature. Because it is an aromatic amine, diaminodiphenylsulfone (DDS) has been selected as curing agent. Moreover, TETM is known to have superior thermal oxidative stability over other types of epoxy resins [40]. Aronhime and Guillham estimated theoretically the glass transition temperature of TETM/DDS formulation at about 352 °C [41]. For the reasons previously exposed, the substitution of phenolic resin in the Roxalte was the first purpose of this study. Subsequently, different foaming processes were considered. Starting with the blowing agent used in the Roxalte, i.e. expandable microspheres, seemed logical. Indeed, as far as we know, the association of the preselected epoxy formulation with these microspheres has not been studied. Further developments on other foaming processes with different foaming agents will be the aim of another paper.
Wang et al. [42] produced epoxy/expandable microsphere foams in two steps. The epoxy was diglycidyl-ether of bisphenol-A. A precure step was first done on the epoxy/curing agent/blowing agent formulation in order to increase the viscosity by starting the polymerization process. The foaming process occurred during the second step where the formulation was put under a higher temperature. This process intended to improve the cell size distribution in the foam. In our study, we propose manufacturing the foam in one step only, by tuning the microsphere's foaming kinetic to the resin gelation kinetic. The crosslinking reaction and the expansion of microspheres take place simultaneously. Obtaining an expanded material depends on two distinct phenomena, the foaming agent expansion on one hand and the resin gelation on the other hand. These two phenomena are independent of each other but both depend on time and temperature.
First, we propose a complete physicochemical study of expandable microspheres, in order to understand their behavior. In particular, the expansion is studied by thermomechanical analysis and correlated to the gel time of the epoxy formulation with the aim of identifying the ideal process temperature and producing the optimized foamed material. The optimization of the epoxy formulation curing cycle has been realized to achieve the manufacture of enhanced expanded foam.

Materials
The triglycidyl ether of tris(4-hydroxyphenyl)methane (TETM) and the diaminodiphenylsulfone (DDS) were kindly offered by Huntsman Advanced Materials (Basel, Switzerland). Their commercial names are Tactix 742 and Aradur 976, respectively. The foaming agents from the Expancel® range were supplied by Rossow Industries (Gennevilliers, France). The Expancel® 950 DU 80 was selected (Particule size 18-24 μm; Density ≤12 kg/m 3 ). The microsphere copolymer is made of acrylonitrile, methylmetacrylate, and methacrylonitrile and the volatile liquid inside is an isooctane-isopentane mix. Its technical data sheet also reveals the presence of magnesium hydroxide. The chemical structures of TETM and DDS are presented in Figure 1.

Preparation of Epoxy Resin
The epoxy resin formulation was prepared in stoichiometric proportion: the stoichiometric ratio, denoted as r, is defined by the ratio between the number of amine and epoxy group and has to be equal to 1. Then, the mass proportions were 71% of TETM for 29% of DDS. The melting temperature of DDS is 178˝C. As evidence, the TETM was first heated at 180˝C and the DDS was added and mixed under mechanical stirring until apparent homogeneity. Indeed, we can see in Figure 2 the difference between the TETM/DDS formulations mixed at 100 and 180˝C. At 100˝C, the viscosity is small enough to allow mixing but the DDS is not melted, leading to a low homogeneity, compared to the mix realized at 180˝C.
The mix is cooled at ambient temperature before starting any analyses.

Preparation of Epoxy Resin
The epoxy resin formulation was prepared in stoichiometric proportion: the stoichiometric ratio, denoted as r, is defined by the ratio between the number of amine and epoxy group and has to be equal to 1. Then, the mass proportions were 71% of TETM for 29% of DDS. The melting temperature of DDS is 178 °C. As evidence, the TETM was first heated at 180 °C and the DDS was added and mixed under mechanical stirring until apparent homogeneity. Indeed, we can see in Figure 2 the difference between the TETM/DDS formulations mixed at 100 and 180 °C. At 100 °C, the viscosity is small enough to allow mixing but the DDS is not melted, leading to a low homogeneity, compared to the mix realized at 180 °C.
The mix is cooled at ambient temperature before starting any analyses.

Preparation of Epoxy Foam
To prepare the epoxy foam, TETM was first mixed into DDS at 180 °C, as above. Then the temperature was lowered to 100 °C and the microspheres were added and mixed under mechanical stirring again until apparent uniformity. First experiments were useful to evaluate the necessary quantity of blowing agent. Finally, the mixture was poured into an aluminum pot and cured in an oven at constant temperature.

Scientific Characterization
All differential scanning calorimetry (DSC Q 200 from TA Instruments, New Castle, DE, USA) analyses were performed under inert atmosphere (N2) using an aluminum crucible with a calibrated hole and an empty crucible as reference.
First, the temperature domain of curing reaction of epoxy system was determined by DSC. The reactive formulation was prepared as mentioned before and a small sample of about 5 mg was removed for testing.
To understand the expandable microspheres behavior, DSC experiments were performed in dynamic mode, on usual microspheres and on compressed microspheres. The compressed microspheres were obtained using a manual hydraulic press (Specac, 15 tons). A few grams of microspheres were introduced into the press and a 9-ton load was applied. This allows for throwing the liquid hydrocarbons out. The DSC analyses were performed between −20 and 400 °C, at a temperature ramp of 3 °C/min. To be free from liquid hydrocarbon evaporation and to focus on determination of the glass transition temperature, DSC analyses were carried out on another sample as follows: a sample of compressed microspheres has been heated from −50 to 200 °C, at a rate of 10 °C/min, then cooled before a second temperature scan at 10 °C/min to Tg determination. The first heating allows for evacuating the residual volatile liquid and the second one allows the Tg measurement. Thermogravimetric analyses (TGA-50 Shimadzu, Kyoto, Japan) were achieved in order to get the degradation temperature. Experiments were performed between 25 and 500 °C, at temperature ramps of 1, 5, 10, and 20 °C/min, under air atmosphere.

Preparation of Epoxy Foam
To prepare the epoxy foam, TETM was first mixed into DDS at 180˝C, as above. Then the temperature was lowered to 100˝C and the microspheres were added and mixed under mechanical stirring again until apparent uniformity. First experiments were useful to evaluate the necessary quantity of blowing agent. Finally, the mixture was poured into an aluminum pot and cured in an oven at constant temperature.

Scientific Characterization
All differential scanning calorimetry (DSC Q 200 from TA Instruments, New Castle, DE, USA) analyses were performed under inert atmosphere (N 2 ) using an aluminum crucible with a calibrated hole and an empty crucible as reference.
First, the temperature domain of curing reaction of epoxy system was determined by DSC. The reactive formulation was prepared as mentioned before and a small sample of about 5 mg was removed for testing.
To understand the expandable microspheres behavior, DSC experiments were performed in dynamic mode, on usual microspheres and on compressed microspheres. The compressed microspheres were obtained using a manual hydraulic press (Specac, 15 tons). A few grams of microspheres were introduced into the press and a 9-ton load was applied. This allows for throwing the liquid hydrocarbons out. The DSC analyses were performed between´20 and 400˝C, at a temperature ramp of 3˝C/min. To be free from liquid hydrocarbon evaporation and to focus on determination of the glass transition temperature, DSC analyses were carried out on another sample as follows: a sample of compressed microspheres has been heated from´50 to 200˝C, at a rate of 10˝C/min, then cooled before a second temperature scan at 10˝C/min to T g determination. The first heating allows for evacuating the residual volatile liquid and the second one allows the T g measurement.
Thermogravimetric analyses (TGA-50 Shimadzu, Kyoto, Japan) were achieved in order to get the degradation temperature. Experiments were performed between 25 and 500˝C, at temperature ramps of 1, 5, 10, and 20˝C/min, under air atmosphere.
The study of expansion of expandable microspheres was performed by thermomechanical analysis (TMA 7 from Perkin Elmer, Waltham, MA, USA). This instrument is used to measure the dimensional variation of a sample under constant strain, as a function of time or temperature. A low strain is applied by a probe on the sample to maintain the contact between the probe and the sample, and the position of the probe is registered. The tests were done in isothermal mode, chosen to be in the appropriate temperature range of the curing reaction. A temperature ramp of 30˝C/min was applied until the isothermal temperature was reached. The strain applied was 10 mN.
The determination of the gel time of the epoxy system was realized by viscosimetric experiment using a dynamic rheometer (MCR 302 from Anton Paar, Graz, Austria) equipped with a cup-plate geometry, in which the diameter of the upper plate is much smaller than the diameter of the cup (25 and 41 mm), in order to prevent side effects and slippage. The gel time is measured at the divergence of viscosity and strongly depends on the temperature used in the kinetic. So the experiment was performed in isothermal mode, chosen to be in the appropriate temperature range of the curing reaction, in steady state with a shear rate of 20 s´1. For these experiments, the reactive formulations were prepared at a temperature of 180˝C. The mixture was then poured into the dedicated cup in the instrument and the test started when the temperature was reached.
Scanning electronic microscopy (SEM) was used to study the morphology of the foam. SEM was done using model ZEISS EVO HD15 (Jena, Germany). The variable pressure (VP) imaging was carried out at 30 Pa. There was no need to coat non-conductive materials prior to analysis. The incident electron beam was kept at 10 kV.
The study of optimization of the curing cycle of the epoxy system was performed by DSC and thermomechanical analysis by using a dynamic rheometer equipped with rectangular torsion geometry. First, epoxy formulations were subjected to different curing times (30,40,50, 60, 120, and 180 min) at the same temperature. Then, around 6 mg of each material were used for DSC analysis. Tests were performed at a heating rate of 5˝C/min, from 25 to 400˝C. Samples for thermomechanical analysis were prepared as follows. The epoxy formulation was mixed, poured into an aluminum mold, and cured in an oven under a determined cycle. The samples stayed in the oven for slow cooling. The dimensions of the final samples were 2 mmˆ8 mmˆ45 mm. Tests were performed with a temperature ramp of 3˝C/min, from 25 to 350˝C, with a deformation of 0.1% and an angular frequency of 1 rad/s. Thermomechanical analyses using a dynamic rheometer equipped with rectangular torsion geometry were also performed on epoxy foam. The foamed material was cut to form a parallelepipedal sample (3 mmˆ12 mmˆ45 mm). Tests were performed with a temperature ramp of 3˝C/min, with an angular frequency of 1 rad/s and a deformation of 0.05%.

Preliminary Study of the Epoxy Formulation and Expandable Microspheres Characterization
First, the epoxy formulation was studied by DSC in order to determine the temperature range of the curing reaction. The DSC thermogram exposed in Figure 3 shows a first exothermic peak corresponding to the curing reaction and a second one, noisy and repeatable, probably due to the degradation. It is presumed that the exothermic peak of degradation is due to the condensation of polyaromatic species by radical carbonization, as was postulated by Rose et al. [43]. According to the DSC curve, this process starts around 330˝C. It can be observed that the first exothermic peak is included between 130 and 240˝C, for a heating rate of 3˝C/min. Since the reaction starts below 180 °C, the question was asked about the mixing temperature of the formulation. In order to know whether the gelation started during the mixing, the gel times were determined at different temperatures for two mixes: one mixed at 100 °C, and the second at 180 °C. Results are gathered in Table 1. The gel times prove the difference between the two mixes. Anyway, if the mix is realized at 100 °C, we may have some concerns about DDS precipitation as it is not well mixed into the epoxy prepolymer ( Figure 2). Thus, we have chosen to perform the mixing at 180 °C, aware that the gelation is only slight. However the gelation did not continue when the microspheres were added and mixed at 100 °C, as the curing reaction only starts at 130 °C (see Figure 3). Thermochemical characterizations (DSC and TGA) were performed on compressed and uncompressed microspheres to understand their behavior.
The DSC analyses performed on compressed and uncompressed microspheres are presented in Figure 4a. There are three major reactions. The first one, measured at 160 °C, appears for uncompressed microspheres only. This is certainly due to the hydrocarbon liquid vaporization. According to the technical data sheet, there are two hydrocarbons, isopentane and isooctane. Their vaporization temperatures are 28 and 99 °C, respectively, at atmospheric pressure. However, when the microspheres are unexpanded, the internal pressure in this confined space is higher than in atmospheric conditions, leading to a higher temperature of vaporization [44]. A pressure drop is possible when the microspheres expands, leading to the evaporation of the condensed blowing agent. The vaporization occurred here at around 160 °C, which is therefore the temperature of starting expansion. This vaporization should correspond to isooctane. Therefore, the vaporization of isopentane is not noticeable. It was supposed that the proportion of this hydrocarbon was too low to be clearly detected by DSC. Figure 4b exposes the analyses performed on compressed microspheres to reveal the glass transition temperature. A major Tg appears around 107 °C and a secondary one around 146 °C. There are various Tg values reported in the literature for the components of the copolymer capsule (PMMA, PAN, PMAN) [45,46], because the Tg is depending on molecular weight. We do not know the molecular weight of the copolymers so it is difficult to conclude which copolymer is related to these Tg. However, we assumed that the first glass transition (Tg1) offers initial softening to reduce local pressure in the inner part of cell wall; the second Tg (Tg2) leads to the softening of the outer part of the cell wall, allowing the hydrocarbon vaporization due to drop in pressure. One of the three copolymers is not associated with any Tg; we assume that its role is to  Since the reaction starts below 180˝C, the question was asked about the mixing temperature of the formulation. In order to know whether the gelation started during the mixing, the gel times were determined at different temperatures for two mixes: one mixed at 100˝C, and the second at 180˝C. Results are gathered in Table 1. The gel times prove the difference between the two mixes. Anyway, if the mix is realized at 100˝C, we may have some concerns about DDS precipitation as it is not well mixed into the epoxy prepolymer ( Figure 2). Thus, we have chosen to perform the mixing at 180˝C, aware that the gelation is only slight. However the gelation did not continue when the microspheres were added and mixed at 100˝C, as the curing reaction only starts at 130˝C (see Figure 3). Thermochemical characterizations (DSC and TGA) were performed on compressed and uncompressed microspheres to understand their behavior.
The DSC analyses performed on compressed and uncompressed microspheres are presented in Figure 4a. There are three major reactions. The first one, measured at 160˝C, appears for uncompressed microspheres only. This is certainly due to the hydrocarbon liquid vaporization. According to the technical data sheet, there are two hydrocarbons, isopentane and isooctane. Their vaporization temperatures are 28 and 99˝C, respectively, at atmospheric pressure. However, when the microspheres are unexpanded, the internal pressure in this confined space is higher than in atmospheric conditions, leading to a higher temperature of vaporization [44]. A pressure drop is possible when the microspheres expands, leading to the evaporation of the condensed blowing agent. The vaporization occurred here at around 160˝C, which is therefore the temperature of starting expansion. This vaporization should correspond to isooctane. Therefore, the vaporization of isopentane is not noticeable. It was supposed that the proportion of this hydrocarbon was too low to be clearly detected by DSC. Figure 4b exposes the analyses performed on compressed microspheres to reveal the glass transition temperature. A major T g appears around 107˝C and a secondary one around 146˝C. There are various T g values reported in the literature for the components of the copolymer capsule (PMMA, PAN, PMAN) [45,46], because the T g is depending on molecular weight. We do not know the molecular weight of the copolymers so it is difficult to conclude which copolymer is related to these T g . However, we assumed that the first glass transition (T g 1) offers initial softening to reduce local pressure in the inner part of cell wall; the second T g (T g 2) leads to the softening of the outer part of the cell wall, allowing the hydrocarbon vaporization due to drop in pressure. One of the three copolymers is not associated with any T g ; we assume that its role is to ensure relative stiffness. The proximity of these two T g values made us think of a T g gradient, in which the third copolymer T g should be contained. ensure relative stiffness. The proximity of these two Tg values made us think of a Tg gradient, in which the third copolymer Tg should be contained.
The second and third major reactions appear for both compressed and uncompressed microspheres. It is thus assumed that they are related to the capsule of the microspheres. The exothermic peak, between 220 and 290 °C, is due to the cyclization reaction in polyacrylonitrile and polymetacrylonitrile [47,48], exposed in Figure 5. The third reaction is an endothermic peak between 320 and 380 °C. It is attributed to the decomposition of magnesium hydroxide to form magnesium oxide and water, as illustrated in Figure 6, where the thermogram of magnesium hydroxide is represented. We can see that the endothermic peak is between 260 and 360 °C.    ensure relative stiffness. The proximity of these two Tg values made us think of a Tg gradient, in which the third copolymer Tg should be contained. The second and third major reactions appear for both compressed and uncompressed microspheres. It is thus assumed that they are related to the capsule of the microspheres. The exothermic peak, between 220 and 290 °C, is due to the cyclization reaction in polyacrylonitrile and polymetacrylonitrile [47,48], exposed in Figure 5. The third reaction is an endothermic peak between 320 and 380 °C. It is attributed to the decomposition of magnesium hydroxide to form magnesium oxide and water, as illustrated in Figure 6, where the thermogram of magnesium hydroxide is represented. We can see that the endothermic peak is between 260 and 360 °C.   The second and third major reactions appear for both compressed and uncompressed microspheres. It is thus assumed that they are related to the capsule of the microspheres. The exothermic peak, between 220 and 290˝C, is due to the cyclization reaction in polyacrylonitrile and polymetacrylonitrile [47,48], exposed in Figure 5. The third reaction is an endothermic peak between 320 and 380˝C. It is attributed to the decomposition of magnesium hydroxide to form magnesium oxide and water, as illustrated in Figure 6, where the thermogram of magnesium hydroxide is represented. We can see that the endothermic peak is between 260 and 360˝C. In order to confirm the phenomena responsible for the weight loss, the TGA curves, performed at 10 °C/min, of compressed microspheres, PAN, and magnesium hydroxide were compared ( Figure  7). PAN analysis is reported in a study by Korobeinyk et al. [47]. It is observed that the mass loss due to the PAN cyclization reaction matches the major microspheres' weight loss. A similar mechanism is observed for PMAN [48]. We can also see that magnesium hydroxide degradation occurs in the same temperature range. Thermogravimetric analyses were performed at various heating rates on compressed microspheres in order to determine the degradation temperature. Results are reported in Figure 8. The degradation temperature and polymer degradation depend on the temperature ramp. Moreover, when the heating is performed slowly, the degradation phenomenon happens gradually. On the contrary, when heating is performed quickly, the main degradation kinetics are graphically clear, which explains the shape differences between the curves.
For every curve, a small decrease of about 2%-3% appears around 200 °C; this is attributed to the vaporization of residual hydrocarbon liquid. Then, for every curve the degradation temperature is measured at a mass loss of 5% to take into account the polymer's degradations. The curve of the degradation temperature in function of the temperature ramp is drawn and the degradation temperature at a very low heating rate (close to 0 °C/min) is determined by extrapolation. As is shown in Figure 9, this value is around 235 °C. The 8.5% residual mass corresponds to the magnesium oxide formation, which occurs between 260 and 360 °C according to the heating rate and resists temperatures up to 700 °C, all the other carbon species being transformed to CO2. In order to confirm the phenomena responsible for the weight loss, the TGA curves, performed at 10˝C/min, of compressed microspheres, PAN, and magnesium hydroxide were compared ( Figure 7). PAN analysis is reported in a study by Korobeinyk et al. [47]. It is observed that the mass loss due to the PAN cyclization reaction matches the major microspheres' weight loss. A similar mechanism is observed for PMAN [48]. We can also see that magnesium hydroxide degradation occurs in the same temperature range. In order to confirm the phenomena responsible for the weight loss, the TGA curves, performed at 10 °C/min, of compressed microspheres, PAN, and magnesium hydroxide were compared ( Figure  7). PAN analysis is reported in a study by Korobeinyk et al. [47]. It is observed that the mass loss due to the PAN cyclization reaction matches the major microspheres' weight loss. A similar mechanism is observed for PMAN [48]. We can also see that magnesium hydroxide degradation occurs in the same temperature range. Thermogravimetric analyses were performed at various heating rates on compressed microspheres in order to determine the degradation temperature. Results are reported in Figure 8. The degradation temperature and polymer degradation depend on the temperature ramp. Moreover, when the heating is performed slowly, the degradation phenomenon happens gradually. On the contrary, when heating is performed quickly, the main degradation kinetics are graphically clear, which explains the shape differences between the curves.
For every curve, a small decrease of about 2%-3% appears around 200 °C; this is attributed to the vaporization of residual hydrocarbon liquid. Then, for every curve the degradation temperature is measured at a mass loss of 5% to take into account the polymer's degradations. The curve of the degradation temperature in function of the temperature ramp is drawn and the degradation temperature at a very low heating rate (close to 0 °C/min) is determined by extrapolation. As is shown in Figure 9, this value is around 235 °C. The 8.5% residual mass corresponds to the magnesium oxide formation, which occurs between 260 and 360 °C according to the heating rate and resists temperatures up to 700 °C, all the other carbon species being transformed to CO2. Thermogravimetric analyses were performed at various heating rates on compressed microspheres in order to determine the degradation temperature. Results are reported in Figure 8. The degradation temperature and polymer degradation depend on the temperature ramp. Moreover, when the heating is performed slowly, the degradation phenomenon happens gradually. On the contrary, when heating is performed quickly, the main degradation kinetics are graphically clear, which explains the shape differences between the curves.
For every curve, a small decrease of about 2%-3% appears around 200˝C; this is attributed to the vaporization of residual hydrocarbon liquid. Then, for every curve the degradation temperature is measured at a mass loss of 5% to take into account the polymer's degradations. The curve of the degradation temperature in function of the temperature ramp is drawn and the degradation temperature at a very low heating rate (close to 0˝C/min) is determined by extrapolation. As is shown in Figure 9, this value is around 235˝C. The 8.5% residual mass corresponds to the magnesium oxide formation, which occurs between 260 and 360˝C according to the heating rate and resists temperatures up to 700˝C, all the other carbon species being transformed to CO 2 .

Determination of the Optimal Foaming Temperature
In our process, we propose to manufacture the foam in only one step. Then we need to determine the temperature that leads to equilibrium between the microsphere's foaming kinetic and the resin gelation kinetic. First, the gel time of the epoxy system is determined at different temperatures, after which the microsphere's foaming is studied.
Gelation time is measured at the divergence of viscosity and obviously depended on the temperature, as shown in Figure 10.
Microsphere expansion study was performed by TMA at isothermal temperatures, from 160 °C to 200 °C. Results of analysis are given in Figure 11. The relative expansion is the ratio of the sample thickness to the initial sample thickness. For each temperature, this deformation of expansion reaches a maximum value then drops. There are two phases: expansion and shrinkage. The deformation of expansion is higher and is happening faster when the temperature increases. This is because of the higher foaming temperature generated by a higher pressure, which stretches the cell wall even more, as was reported by other authors [27,42]. At 180, 190, and 200 °C the expansion is quite important compared to at temperatures of 160 and 170 °C. The shrinkage phase is even faster when the temperature is high.

Determination of the Optimal Foaming Temperature
In our process, we propose to manufacture the foam in only one step. Then we need to determine the temperature that leads to equilibrium between the microsphere's foaming kinetic and the resin gelation kinetic. First, the gel time of the epoxy system is determined at different temperatures, after which the microsphere's foaming is studied.
Gelation time is measured at the divergence of viscosity and obviously depended on the temperature, as shown in Figure 10.
Microsphere expansion study was performed by TMA at isothermal temperatures, from 160 °C to 200 °C. Results of analysis are given in Figure 11. The relative expansion is the ratio of the sample thickness to the initial sample thickness. For each temperature, this deformation of expansion reaches a maximum value then drops. There are two phases: expansion and shrinkage. The deformation of expansion is higher and is happening faster when the temperature increases. This is because of the higher foaming temperature generated by a higher pressure, which stretches the cell wall even more, as was reported by other authors [27,42]. At 180, 190, and 200 °C the expansion is quite important compared to at temperatures of 160 and 170 °C. The shrinkage phase is even faster when the temperature is high.

Determination of the Optimal Foaming Temperature
In our process, we propose to manufacture the foam in only one step. Then we need to determine the temperature that leads to equilibrium between the microsphere's foaming kinetic and the resin gelation kinetic. First, the gel time of the epoxy system is determined at different temperatures, after which the microsphere's foaming is studied.
Gelation time is measured at the divergence of viscosity and obviously depended on the temperature, as shown in Figure 10.
Microsphere expansion study was performed by TMA at isothermal temperatures, from 160˝C to 200˝C. Results of analysis are given in Figure 11. The relative expansion is the ratio of the sample thickness to the initial sample thickness. For each temperature, this deformation of expansion reaches a maximum value then drops. There are two phases: expansion and shrinkage. The deformation of expansion is higher and is happening faster when the temperature increases. This is because of the higher foaming temperature generated by a higher pressure, which stretches the cell wall even more, as was reported by other authors [27,42]. At 180, 190, and 200˝C the expansion is quite important compared to at temperatures of 160 and 170˝C. The shrinkage phase is even faster when the temperature is high.  It has to be noted that the expansion starts during the temperature ramp after 4 min, that is to say at 150 °C. This temperature corresponds with the Tg2 (see Section 3.1) responsible for the expansion. The time of expansion corresponds to the expansion phase: from t = 4 min to the maximum value of the curve, as shown in Figure 11 for the temperature of 200 °C.
Results from the TMA and epoxy system gel time analysis are gathered in Figure 12. TETM/DDS/Microspheres foam samples were manufactured as described in Section 2 and cured at a constant temperature from 160 to 200 °C, with the aim of comparing with the previous analysis. The pictures are given in Figure 12, associated with each temperature.
Between 160 and 180 °C, the gel time is quite high compared to the time of expansion. At 160 and 170 °C the pictures show that expansion is very limited and a phase separation can be observed: the bubbles are lighter than the resin and go up to the surface. This phenomenon can occur because the resin gelation occurs later on. At 180 °C, the picture shows a great expansion, without phase separation. Indeed, as was shown in Figure 11, at this temperature the expansion is relatively important. Moreover, even if the gelation occurs late, the shrinkage kinetic is quite slow so the bubble size decreases slowly.  It has to be noted that the expansion starts during the temperature ramp after 4 min, that is to say at 150 °C. This temperature corresponds with the Tg2 (see Section 3.1) responsible for the expansion. The time of expansion corresponds to the expansion phase: from t = 4 min to the maximum value of the curve, as shown in Figure 11 for the temperature of 200 °C.
Results from the TMA and epoxy system gel time analysis are gathered in Figure 12. TETM/DDS/Microspheres foam samples were manufactured as described in Section 2 and cured at a constant temperature from 160 to 200 °C, with the aim of comparing with the previous analysis. The pictures are given in Figure 12, associated with each temperature.
Between 160 and 180 °C, the gel time is quite high compared to the time of expansion. At 160 and 170 °C the pictures show that expansion is very limited and a phase separation can be observed: the bubbles are lighter than the resin and go up to the surface. This phenomenon can occur because the resin gelation occurs later on. At 180 °C, the picture shows a great expansion, without phase separation. Indeed, as was shown in Figure 11, at this temperature the expansion is relatively important. Moreover, even if the gelation occurs late, the shrinkage kinetic is quite slow so the bubble size decreases slowly. It has to be noted that the expansion starts during the temperature ramp after 4 min, that is to say at 150˝C. This temperature corresponds with the T g 2 (see Section 3.1) responsible for the expansion. The time of expansion corresponds to the expansion phase: from t = 4 min to the maximum value of the curve, as shown in Figure 11 for the temperature of 200˝C.
Results from the TMA and epoxy system gel time analysis are gathered in Figure 12. TETM/DDS/Microspheres foam samples were manufactured as described in Section 2 and cured at a constant temperature from 160 to 200˝C, with the aim of comparing with the previous analysis. The pictures are given in Figure 12, associated with each temperature.
Between 160 and 180˝C, the gel time is quite high compared to the time of expansion. At 160 and 170˝C the pictures show that expansion is very limited and a phase separation can be observed: the bubbles are lighter than the resin and go up to the surface. This phenomenon can occur because the resin gelation occurs later on. At 180˝C, the picture shows a great expansion, without phase separation. Indeed, as was shown in Figure 11, at this temperature the expansion is relatively important. Moreover, even if the gelation occurs late, the shrinkage kinetic is quite slow so the bubble size decreases slowly.
Between 190 and 200˝C, the gel time is close to the time of expansion. The expansion is important and faster; there is no phase separation. At these temperatures the shrinkage kinetic is fast, as is shown in Figure 11, which confirms that the gelation has to be close to the time of expansion to avoid the shrinkage of the foam. Furthermore, a progressive darkening is observed inside the sample at 190 and 200˝C. This coloration is due to the exothermicity of the curing reaction so the temperature in the sample is greater than the isothermal temperature in the oven. Therefore it is reasonable to think that an oxidative reaction occurred in these samples. The cause of this oxidation is explained later in the text.
The density data for these samples were measured by using a pycnometer and are gathered in Table 2. The density decreases with the temperature, which confirms that the expansion is more important at 190 and 200˝C than at 180˝C. However, a better compromise would be a temperature of 180˝C, because it allies a great expansion with a slow shrinkage and avoids thermal degradation. Between 190 and 200 °C, the gel time is close to the time of expansion. The expansion is important and faster; there is no phase separation. At these temperatures the shrinkage kinetic is fast, as is shown in Figure 11, which confirms that the gelation has to be close to the time of expansion to avoid the shrinkage of the foam. Furthermore, a progressive darkening is observed inside the sample at 190 and 200 °C. This coloration is due to the exothermicity of the curing reaction so the temperature in the sample is greater than the isothermal temperature in the oven. Therefore it is reasonable to think that an oxidative reaction occurred in these samples. The cause of this oxidation is explained later in the text.
The density data for these samples were measured by using a pycnometer and are gathered in Table 2. The density decreases with the temperature, which confirms that the expansion is more important at 190 and 200 °C than at 180 °C. However, a better compromise would be a temperature of 180 °C, because it allies a great expansion with a slow shrinkage and avoids thermal degradation.  A SEM (scanning electronic microscopy) micrograph of the epoxy formulation/expandable microsphere foamed at 180 °C is given in Figure 13. The cell size distribution is homogeneous, which is a sign of good synchronization between the occurrence of the gelation and the time of expansion [42].
In order to observe the behavior of different grades of expandable microspheres, three additional expandable microspheres grades were selected and tested at 180 °C (920 WUF 40, 920 DU 40, 980 DUX 120). They were chosen according to their expansion temperature, which was around 180 °C. Referring to the supplier data sheet, these four grades are made of the same copolymer: acrylonitrile-methylmetacrylate-methacrylonitrile. However, the hydrocarbons inside the shell are not the same for all. For instance, 920 WUF 40 and 920 DU 40 contain isopentane, 980 DUX 120 contains isooctane, and 950 DU 80 contains an isopentane-isooctane mix. As was explained in Section 3.1, expansion is possible only if the shell is softened enough. Then the hydrocarbons are allowed to expand. Knowing the nature of the copolymer is not sufficient because the Tg is  A SEM (scanning electronic microscopy) micrograph of the epoxy formulation/expandable microsphere foamed at 180˝C is given in Figure 13. The cell size distribution is homogeneous, which is a sign of good synchronization between the occurrence of the gelation and the time of expansion [42].
In order to observe the behavior of different grades of expandable microspheres, three additional expandable microspheres grades were selected and tested at 180˝C (920 WUF 40, 920 DU 40, 980 DUX 120). They were chosen according to their expansion temperature, which was around 180˝C. Referring to the supplier data sheet, these four grades are made of the same copolymer: acrylonitrile-methylmetacrylate-methacrylonitrile. However, the hydrocarbons inside the shell are not the same for all. For instance, 920 WUF 40 and 920 DU 40 contain isopentane, 980 DUX 120 contains isooctane, and 950 DU 80 contains an isopentane-isooctane mix. As was explained in Section 3.1, expansion is possible only if the shell is softened enough. Then the hydrocarbons are allowed to expand. Knowing the nature of the copolymer is not sufficient because the T g is dependent on the molecular weight of each copolymer. So the nature of the gas inside the shell as well as the mass of each copolymer is needed to predict the temperature of expansion. dependent on the molecular weight of each copolymer. So the nature of the gas inside the shell as well as the mass of each copolymer is needed to predict the temperature of expansion. Four epoxy formulation/expandable microspheres mixes were prepared. The epoxy formulations were prepared as before, in an aluminum crucible, at 180 °C. The temperature was dropped to 100 °C, at which time the microspheres were added and mixed until apparent homogeneity. The four mixtures had exactly the same weight and a proportion of 5% of microspheres. Then, the crucibles were placed in an oven at 180 °C, for 30 min.
The pictures of the cured mixes with the four different expandable microspheres are presented in Figure 14. Tstart corresponds to the temperature when the foaming begins and Tmax corresponds to the temperature when the microsphere volume is maximal. These values are issued from the technical data sheet. The samples were cut down the middle into two parts; we can see the 5 cm length profiles. The samples containing the 920 WUF 40 and 920 DU 40 microspheres show average expansion and bubbles of heterogeneous and large dimensions. This is certainly due to coalescence. As the bubbles grow, adjacent ones tend to coalesce in order to reduce the total free energy by lowering the surface area of cells [49][50][51]. This is happening here because the thermoplastic capsule collapsed due to the high temperature while the crosslinking reaction was not really consequent. Indeed, it has been demonstrated by various authors that too low a viscosity increases the probability of coalescence [50,52]. The sample containing the 950 DU 80 microspheres presents greater expansion and bubbles of homogeneous and microscopic dimensions.
The sample containing the 980 DUX 120 microspheres shows a low expansion, on the surface only. Referring to producer data, this grade needs higher temperature to expand than previous ones. The temperature in the core is not high enough to produce a whole expansion compared to the surface because of the weak thermal conductivity of epoxy resins. Four epoxy formulation/expandable microspheres mixes were prepared. The epoxy formulations were prepared as before, in an aluminum crucible, at 180˝C. The temperature was dropped to 100˝C, at which time the microspheres were added and mixed until apparent homogeneity. The four mixtures had exactly the same weight and a proportion of 5% of microspheres. Then, the crucibles were placed in an oven at 180˝C, for 30 min.
The pictures of the cured mixes with the four different expandable microspheres are presented in Figure 14. T start corresponds to the temperature when the foaming begins and T max corresponds to the temperature when the microsphere volume is maximal. These values are issued from the technical data sheet. The samples were cut down the middle into two parts; we can see the 5 cm length profiles. The samples containing the 920 WUF 40 and 920 DU 40 microspheres show average expansion and bubbles of heterogeneous and large dimensions. This is certainly due to coalescence. As the bubbles grow, adjacent ones tend to coalesce in order to reduce the total free energy by lowering the surface area of cells [49][50][51]. This is happening here because the thermoplastic capsule collapsed due to the high temperature while the crosslinking reaction was not really consequent. Indeed, it has been demonstrated by various authors that too low a viscosity increases the probability of coalescence [50,52]. dependent on the molecular weight of each copolymer. So the nature of the gas inside the shell as well as the mass of each copolymer is needed to predict the temperature of expansion. Four epoxy formulation/expandable microspheres mixes were prepared. The epoxy formulations were prepared as before, in an aluminum crucible, at 180 °C. The temperature was dropped to 100 °C, at which time the microspheres were added and mixed until apparent homogeneity. The four mixtures had exactly the same weight and a proportion of 5% of microspheres. Then, the crucibles were placed in an oven at 180 °C, for 30 min.
The pictures of the cured mixes with the four different expandable microspheres are presented in Figure 14. Tstart corresponds to the temperature when the foaming begins and Tmax corresponds to the temperature when the microsphere volume is maximal. These values are issued from the technical data sheet. The samples were cut down the middle into two parts; we can see the 5 cm length profiles. The samples containing the 920 WUF 40 and 920 DU 40 microspheres show average expansion and bubbles of heterogeneous and large dimensions. This is certainly due to coalescence. As the bubbles grow, adjacent ones tend to coalesce in order to reduce the total free energy by lowering the surface area of cells [49][50][51]. This is happening here because the thermoplastic capsule collapsed due to the high temperature while the crosslinking reaction was not really consequent. Indeed, it has been demonstrated by various authors that too low a viscosity increases the probability of coalescence [50,52]. The sample containing the 950 DU 80 microspheres presents greater expansion and bubbles of homogeneous and microscopic dimensions.
The sample containing the 980 DUX 120 microspheres shows a low expansion, on the surface only. Referring to producer data, this grade needs higher temperature to expand than previous ones. The temperature in the core is not high enough to produce a whole expansion compared to the surface because of the weak thermal conductivity of epoxy resins. The sample containing the 980 DUX 120 microspheres shows a low expansion, on the surface only. Referring to producer data, this grade needs higher temperature to expand than previous ones. The temperature in the core is not high enough to produce a whole expansion compared to the surface because of the weak thermal conductivity of epoxy resins.

TETM/DDS Cure Cycle Optimization
Based on the previous study, a first temperature T 1 = 180 C was set for the curing process of the TETM/DDS system. DSC analyses were conducted as described before. Results are presented in Figure 15.

TETM/DDS Cure Cycle Optimization
Based on the previous study, a first temperature T1 = 180 C was set for the curing process of the TETM/DDS system. DSC analyses were conducted as described before. Results are presented in Figure 15. An exothermic peak is remained around 240 °C for each time of cure. The longer the curing lasts, the lower the remaining peak height. However, even after 3 h at 180 °C, the cure is not completed. The Tg depends on the degree of cure reaction [53]. Then the system needs to be fully cured to increase the Tg.
A post-curing temperature T2 = 240 °C was then added to the cycle. To determine the optimal cure thermal cycle, six curing cycles were prepared and tested by thermomechanical analysis using a dynamic rheometer equipped with rectangular torsion geometry: The temperature ramp between the two stages was 3 °C/min. DSC analyses of each sample after the two steps curing cycles were performed and no remaining peak was observed.
Results from rectangular torsion tests, performed at an angular frequency of 1 rad/s, are gathered in Figure 16. The samples that went through a second stage of 1 h at 240 °C (C1, C3, and C5) are similar, likewise the samples that had a second stage of 2 h at 240 °C (C2, C4, and C6). Two alpha relaxations, a rheological signature of glass transition temperature, are observed at the peak on the G'' curves of C1, C3, and C5 samples, α1 at 230 °C, and α2 at 330 °C. Indeed, G'' is sensitive to the long-range molecular motion occurring during the α relaxation [54]. α1 indicates molecular movements, which confirms that the curing process is not over and that more time for more energy is needed around this temperature. Evolution in the polymerization reaction is occurring during this analysis. The macromolecules' conformation allows for the reaction of chemical entities (α1). As the heating is slow during the test, the molecules have the time to react, leading to α2, which actually is at the same temperature as α. Then, this test is allowed to complete the curing reaction for samples that went through a second stage of only 1 h at 240 °C (C1, C3, and C5). The second stage of 2 h at 240 °C is then needed to complete the curing process. However, a first stage of 1 h at 180 °C is enough An exothermic peak is remained around 240˝C for each time of cure. The longer the curing lasts, the lower the remaining peak height. However, even after 3 h at 180˝C, the cure is not completed. The T g depends on the degree of cure reaction [53]. Then the system needs to be fully cured to increase the T g .
A post-curing temperature T 2 = 240˝C was then added to the cycle. To determine the optimal cure thermal cycle, six curing cycles were prepared and tested by thermomechanical analysis using a dynamic rheometer equipped with rectangular torsion geometry: The temperature ramp between the two stages was 3˝C/min. DSC analyses of each sample after the two steps curing cycles were performed and no remaining peak was observed.
Results from rectangular torsion tests, performed at an angular frequency of 1 rad/s, are gathered in Figure 16. The samples that went through a second stage of 1 h at 240˝C (C 1 , C 3 , and C 5 ) are similar, likewise the samples that had a second stage of 2 h at 240˝C (C 2 , C 4 , and C 6 ). Two alpha relaxations, a rheological signature of glass transition temperature, are observed at the peak on the G" curves of C 1 , C 3 , and C 5 samples, α 1 at 230˝C, and α 2 at 330˝C. Indeed, G" is sensitive to the long-range molecular motion occurring during the α relaxation [54]. α 1 indicates molecular movements, which confirms that the curing process is not over and that more time for more energy is needed around this temperature. Evolution in the polymerization reaction is occurring during this analysis. The macromolecules' conformation allows for the reaction of chemical entities (α 1 ). As the heating is slow during the test, the molecules have the time to react, leading to α 2 , which actually is at the same temperature as α. Then, this test is allowed to complete the curing reaction for samples that went through a second stage of only 1 h at 240˝C (C 1 , C 3 , and C 5 ). The second stage of 2 h at 240˝C is then needed to complete the curing process. However, a first stage of 1 h at 180˝C is enough since there is no difference when increasing the time of the first stage from 1 h to 3 h (C 2 , C 4 , and C 6 ). Finally, the optimized cure cycle is 1 h at 180˝C and 2 h at 240˝C (C 2 ). The alpha (α) temperature, evaluated at the peak on the G" curve, is then equal to 330˘2˝C.
Epoxy foam was realized as described in Section 2. The mixture was cured in an oven, where the cycle C 2 was applied. Parallelepipedal samples were cut from the foamed material and rectangular torsion analyses were performed. Results are given in Figure 17, with the curves of the non-foamed sample previously obtained. The moduli values of the foam are logically lower because of air presence. The T α of the foam is also less important. Indeed, the endothermic reaction due to the vaporization of the liquid during the expansion of the microspheres consumes a part of the heat released during the exothermic cure reaction of the TETM/DDS system. since there is no difference when increasing the time of the first stage from 1 h to 3 h (C2, C4, and C6). Finally, the optimized cure cycle is 1 h at 180 °C and 2 h at 240 °C (C2). The alpha (α) temperature, evaluated at the peak on the G'' curve, is then equal to 330 ± 2 °C. Epoxy foam was realized as described in Section 2. The mixture was cured in an oven, where the cycle C2 was applied. Parallelepipedal samples were cut from the foamed material and rectangular torsion analyses were performed. Results are given in Figure 17, with the curves of the non-foamed sample previously obtained. The moduli values of the foam are logically lower because of air presence. The Tα of the foam is also less important. Indeed, the endothermic reaction due to the vaporization of the liquid during the expansion of the microspheres consumes a part of the heat released during the exothermic cure reaction of the TETM/DDS system.  The DSC analyses in Figure 18 allow us to compare the thermograms of TETM/DDS formulation, one containing 5% of expandable microspheres and one without microspheres, with the thermogram of expandable microspheres only. The value of heat flow for microspheres was reduced to the weight of hydrocarbons contained in the epoxy formulation with expandable microspheres, as they are responsible for the endothermic reaction, or 1% of the epoxy formulation since there is no difference when increasing the time of the first stage from 1 h to 3 h (C2, C4, and C6). Finally, the optimized cure cycle is 1 h at 180 °C and 2 h at 240 °C (C2). The alpha (α) temperature, evaluated at the peak on the G'' curve, is then equal to 330 ± 2 °C. Epoxy foam was realized as described in Section 2. The mixture was cured in an oven, where the cycle C2 was applied. Parallelepipedal samples were cut from the foamed material and rectangular torsion analyses were performed. Results are given in Figure 17, with the curves of the non-foamed sample previously obtained. The moduli values of the foam are logically lower because of air presence. The Tα of the foam is also less important. Indeed, the endothermic reaction due to the vaporization of the liquid during the expansion of the microspheres consumes a part of the heat released during the exothermic cure reaction of the TETM/DDS system.  The DSC analyses in Figure 18 allow us to compare the thermograms of TETM/DDS formulation, one containing 5% of expandable microspheres and one without microspheres, with the thermogram of expandable microspheres only. The value of heat flow for microspheres was reduced to the weight of hydrocarbons contained in the epoxy formulation with expandable microspheres, as they are responsible for the endothermic reaction, or 1% of the epoxy formulation The DSC analyses in Figure 18 allow us to compare the thermograms of TETM/DDS formulation, one containing 5% of expandable microspheres and one without microspheres, with the thermogram of expandable microspheres only. The value of heat flow for microspheres was reduced to the weight of hydrocarbons contained in the epoxy formulation with expandable microspheres, as they are responsible for the endothermic reaction, or 1% of the epoxy formulation (20% of hydrocarbons in 5% of microspheres). There is a small interval of 5˝C between the peaks on the thermograms of the formulation without and with microspheres, corresponding to 15 min. This temperature shift is explained by the endothermic peak of vaporization, which delays the curing reaction. As the expansion happens during the first stage of the curing process (at 180˝C), it was postulated that more time at this temperature was needed to complete the curing reaction of the foam. At temperatures over 230˝C, the increased heat flow encircled in the thermogram of epoxy formulation with microspheres is related to a degradation, which is discussed later. (20% of hydrocarbons in 5% of microspheres). There is a small interval of 5 °C between the peaks on the thermograms of the formulation without and with microspheres, corresponding to 15 min. This temperature shift is explained by the endothermic peak of vaporization, which delays the curing reaction. As the expansion happens during the first stage of the curing process (at 180 °C), it was postulated that more time at this temperature was needed to complete the curing reaction of the foam. At temperatures over 230 °C, the increased heat flow encircled in the thermogram of epoxy formulation with microspheres is related to a degradation, which is discussed later. Then two other cycles were tested for 2 h at 180 °C + 2 h at 240 °C, and 3h at 180 °C + 2 h at 240 °C (C4 and C6). Results are given in Figure 19. Applying the cure cycle C2 allows for reaching 250 ± 2 °C as the glass transition temperature. Increasing the time of the first stage of 1 h allows for increasing the glass transition temperature by about 20 °C, leading to a Tα of 270 ± 2 °C. However, a first stage of 3 h at 180 °C is not needed. Then two other cycles were tested for 2 h at 180˝C + 2 h at 240˝C, and 3h at 180˝C + 2 h at 240˝C (C 4 and C 6 ). Results are given in Figure 19. Applying the cure cycle C 2 allows for reaching 250˘2˝C as the glass transition temperature. Increasing the time of the first stage of 1 h allows for increasing the glass transition temperature by about 20˝C, leading to a T α of 270˘2˝C. However, a first stage of 3 h at 180˝C is not needed. (20% of hydrocarbons in 5% of microspheres). There is a small interval of 5 °C between the peaks on the thermograms of the formulation without and with microspheres, corresponding to 15 min. This temperature shift is explained by the endothermic peak of vaporization, which delays the curing reaction. As the expansion happens during the first stage of the curing process (at 180 °C), it was postulated that more time at this temperature was needed to complete the curing reaction of the foam. At temperatures over 230 °C, the increased heat flow encircled in the thermogram of epoxy formulation with microspheres is related to a degradation, which is discussed later. Then two other cycles were tested for 2 h at 180 °C + 2 h at 240 °C, and 3h at 180 °C + 2 h at 240 °C (C4 and C6). Results are given in Figure 19. Applying the cure cycle C2 allows for reaching 250 ± 2 °C as the glass transition temperature. Increasing the time of the first stage of 1 h allows for increasing the glass transition temperature by about 20 °C, leading to a Tα of 270 ± 2 °C. However, a first stage of 3 h at 180 °C is not needed. The foams presented a dark coloration, as shown in Figure 20. We assume that this coloration is due to the degradation of expandable microspheres and particularly to the polymerization of sequences of adjacent nitrile groups [55], in the cyclization reaction occurring in polyacrylonitrile and polymethacryclonitrile [47,48]. Indeed, this darkening occurred during the post-curing at 240˝C, as is highlighted in Figure 21. The foams presented a dark coloration, as shown in Figure 20. We assume that this coloration is due to the degradation of expandable microspheres and particularly to the polymerization of sequences of adjacent nitrile groups [55], in the cyclization reaction occurring in polyacrylonitrile and polymethacryclonitrile [47,48]. Indeed, this darkening occurred during the post-curing at 240 °C, as is highlighted in Figure 21.  Moreover, in order to confirm that the TETM/DDS formulations do not start to degrade during the post-cure at 240 °C, TGA analyses were performed on two samples ( Figure 22). One sample has been subjected to the first stage at 180 °C only, and the second one has been subjected to the post-cure at 240 °C. We can observe that these curves are similar, which confirms that the post-cure does not initiate the degradation of the epoxy resin. The experimental result corroborates the fact that the increase in the thermogram ( Figure 18) of an epoxy formulation with microspheres at 230 °C is only related to the degradation of the microspheres.  The foams presented a dark coloration, as shown in Figure 20. We assume that this coloration is due to the degradation of expandable microspheres and particularly to the polymerization of sequences of adjacent nitrile groups [55], in the cyclization reaction occurring in polyacrylonitrile and polymethacryclonitrile [47,48]. Indeed, this darkening occurred during the post-curing at 240 °C, as is highlighted in Figure 21.  Moreover, in order to confirm that the TETM/DDS formulations do not start to degrade during the post-cure at 240 °C, TGA analyses were performed on two samples ( Figure 22). One sample has been subjected to the first stage at 180 °C only, and the second one has been subjected to the post-cure at 240 °C. We can observe that these curves are similar, which confirms that the post-cure does not initiate the degradation of the epoxy resin. The experimental result corroborates the fact that the increase in the thermogram ( Figure 18) of an epoxy formulation with microspheres at 230 °C is only related to the degradation of the microspheres.  Moreover, in order to confirm that the TETM/DDS formulations do not start to degrade during the post-cure at 240˝C, TGA analyses were performed on two samples ( Figure 22). One sample has been subjected to the first stage at 180˝C only, and the second one has been subjected to the post-cure at 240˝C. We can observe that these curves are similar, which confirms that the post-cure does not initiate the degradation of the epoxy resin. The experimental result corroborates the fact that the increase in the thermogram ( Figure 18) of an epoxy formulation with microspheres at 230˝C is only related to the degradation of the microspheres. The foams presented a dark coloration, as shown in Figure 20. We assume that this coloration is due to the degradation of expandable microspheres and particularly to the polymerization of sequences of adjacent nitrile groups [55], in the cyclization reaction occurring in polyacrylonitrile and polymethacryclonitrile [47,48]. Indeed, this darkening occurred during the post-curing at 240 °C, as is highlighted in Figure 21.  Moreover, in order to confirm that the TETM/DDS formulations do not start to degrade during the post-cure at 240 °C, TGA analyses were performed on two samples ( Figure 22). One sample has been subjected to the first stage at 180 °C only, and the second one has been subjected to the post-cure at 240 °C. We can observe that these curves are similar, which confirms that the post-cure does not initiate the degradation of the epoxy resin. The experimental result corroborates the fact that the increase in the thermogram ( Figure 18) of an epoxy formulation with microspheres at 230 °C is only related to the degradation of the microspheres.

Conclusions
This paper reports on a complete study leading to the manufacture of a high temperature epoxy foam. The analyses performed on the expandable microspheres highlighted the different steps leading to the expansion. The copolymer shell passes through two main glass transition temperatures: the first one offers initial softening to reduce local pressure in the inner part of cell wall, whereas the second one leads to the hydrocarbons' vaporization and the cell expansion. It was demonstrated that the expansion kinetics performed by TMA is even more important at higher temperatures. These results were correlated to the gel time of the epoxy formulation. Indeed, in order to maintain the bubble size and a homogeneous size distribution, gelation and foaming times have to be close. A good compromise temperature is 180˝C, because it allies great expansion with slow shrinkage and avoids thermal degradation.
In the second part of this paper, the TETM/DDS formulation curing process was optimized and a T α of 330˘2˝C was obtained after a two-step curing cycle. Finally, the epoxy foam was manufactured and its curing process was optimized, leading to a T α of 270˘2˝C.