An Original Method to Characterize the Expansion of Thermally Self-Expandable Adhesive Thermoset Formulations Using Optical Microscopy

Abstract

The present work focuses on a characterization method using optical microscopy to follow the swelling of expanded adhesive formulations. The original aspect of the method consists of incorporating a heating plate into the optical microscope in order to apply a curing process to the adhesive and to directly observe its expansion between two glass plates with a defined gap. Different heating rates and curing temperatures were applied to observe the influence on the expansion process, characterized by the time needed to fill the gap between the glass plates, the homogeneity and quality of the contact on the upper plate and the stability of the adhesion. Bisphenol A diglycidyl ether/dicyandiamide adhesive formulations were particularly considered in this study. Adhesives were also analyzed by DSC to clearly define the curing process in function of the heating rate and curing temperature. The expansion process presented in this work implies the use of a physical expansion agent to validate the analytical method. Their expansion was also studied under an optical microscope to define a blowing profile matching the curing process of the epoxy-amine adhesive. Results showed that the synchronization between the expansion and the curing of the adhesive could be followed and analyzed qualitatively by the proposed method and could be applied to various expanded formulations.

1. Introduction

In order to bring long-term resistance and durability to metal assemblies used in the automotive industry, welding, threading and bonding methods are widely used to ensure the cohesion of numerous metal pieces. In the case of bonding, an adhesive formulation is applied on metallic surfaces and has to undergo curing between 160 and 200 °C to acquire its final mechanical properties. The curing reaction of adhesive joints is integrated in the assembly line during the cataphoresis step. Both the cataphoresis and the curing of adhesive joints processes apply a similar temperature range and can be realized in one step through a dedicated heating enclosure.

Structural adhesives used in the automotive industry are generally based on the chemistry of epoxides [1,2]. This kind of adhesive joint brings an efficient resistance to corrosion and fatigue, a weight reduction in the structure and a reduction in the stress compared to the perforation of metal sheets [3]. Epoxide adhesive formulations can be proposed as viscous liquids or as solid films. In this second case, the curing step softens the formulation and leads to cross-linking to ensure adhesion through the reaction between the adhesive and hydroxides at the surface of the substrate. Another point of interest is that the adhesive film can be shaped into well-defined or complex forms, adapted to the shape of metal pieces to bond. This aspect limits the spreading of the adhesive during assembly, when the pressure is applied by the two substrates. The quantity of adhesive joints and adhesive strength can be optimized. In the case of certain automotive parts, the metal/adhesive/metal assembly is first prepared as a multilayer system, and then the piece is stamped to gain its final form to be assembled on the vehicle structure. The stamping step can easily create gaps and disturb the parallelism of the metal/adhesive/metal assembly. Furthermore, it may also create defects and humidity diffusion paths leading to high loss of durability [4,5,6]. In order to compensate for the volume of eventual defects related to the assembly process, L&L Products (partner of the initial project) developed an alternative structural adhesive formulation able to expand up to 100% of its initial thickness. The expansion of the adhesive joint during the curing allows to homogenize the contact between the metallic substrate and the bonding formulation and to recover the adhesive strength eventually lost, due to the absence of contact between certain zones [7,8]. It is, however, tricky to estimate the quality of the contact between opaque substrates. Furthermore, the efficiency of the expansion could be related to the initial thickness of the joint, the heating rate, the curing temperature, the curing time, the quantity of expansion agent or the cross-linking kinetics of the formulation. The characterization of polymers able to expand and foam is quite common. A classical characterization is the observation of the morphology, the porosity and the internal structure of the polymer by techniques such as X-ray microtomography [9], SEM observation [10] and nitrogen adsorption using the Brunauer–Emmet–Teller and Barrett–Joyner–Halenda methods (BET-BJH) [11]. Measurements such as foam density and expansion ratio could also be done on final cross-linked polymer foams [12]. Even if the macroscopic/microscopic observation enables the efficient characterization of a polymer foam, it is however limited to materials obtained after curing and expansion and does not mimic the actual process in more complex systems such as adhesive assemblies. The literature contains a large range of expansion studies based on custom-built devices which are more adapted to specific applications [13]. Foaming, cell distribution, cell coalescence and cellular structure could be directly observed with high-speed cameras and optical microscopy through image analysis [14,15]. The use of heating plates in the case of thermoset polymers is quite common to follow the evolution of foaming parameters in different curing conditions and to mimic industrial processes such as injection molding and polymer extrusion [16,17].

In the case of a thermoset formulation able to expand, the most crucial aspect to focus on to optimize the properties is the understanding of both the cross-linking and the expansion mechanisms with temperature. Indeed, as both mechanisms highly depend on the temperature and the heating rate, it is essential to reach synchronization to enable the formation of a stable structure. However, the study of the synchronization of the cross-linking and the expansion is quite absent from the literature as the curing parameters have to correspond to the ones applied in automotive assembly lines, with few considerations for the efficiency of thermoset adhesive parts. This work presents an original method to estimate the expansion and the contact quality between two substrates with a fixed gap in the case of an epoxide adhesive formulation able to expand. The aim of this method is to mimic the expansion and adhesion processes occurring in substrate/polymer/substrate assemblies in the case of industrial curing cycles. In the particular case of this manuscript, the study focuses on solid adhesive films used in the automotive industry based on epoxy/dicyandiamide chemistry and containing a physical expansion agent presented as microspheres. This method is based on in situ microscopic observation of the expansion of a formulation with the use of a Peltier effect heating plate. Behaviors such as expansion, bubbling and spreading were characterized depending on the curing temperature and the heating rate to select the right conditions to obtain an optimal contact between two substrates to limit the lack of adhesion and prevent water diffusion pathways.

2. Materials and Methods

2.1. Adhesive Formulation

The formulation is mainly based on the reaction between bisphenol A diglycidyl ether (BADGE) and dicyandiamide (Dicy) and optimized to create a structural adhesive with a high resistance for metal assemblies. The expansion was possible due to the presence of a physical expansion agent (PEA) in the formulation. The commercial adhesive formulation was provided by L&L Products that also contains a urea-based accelerator (0.35 wt%) and mineral fillers (calcium carbonate and calcium hydroxide at 10–15 wt%). The Dicy was introduced in excess with a molar ratio of 1.35 compared to the stoichiometry, with an Amine Hydrogen Equivalent Weight of 14 g/eq. The model formulation was provided in two versions. One with all the components (TEAF) and one without the PEA (TEAFW). In TEAF formulation, PEA was introduced between 1 wt% and 3 wt%. All formulations were developed as solid films at room temperature with a Tg of about 85 °C and a thickness of 0.50 ± 0.03 mm. Due to trade secret, the exact quantities of all components were not disclosed by L&L Products, which may limit the reproducibility of experiments. Each formulation was prepared in batch and all samples of one formulation come from the same batch to enable comparison during the study. In order to use the same PEA for each experiment, it was decided to focus on the Unicell MS4600 system (Dongjin Italia, Casoli, Italy), a white powdery product at room temperature.

The PEA is a spherical bead composed of a solid polymeric shell and a liquid core. The core consists of a liquid alkane mixture (at room temperature) which evaporates and expands the shell at a defined temperature. When the expansion agent is heated above the Tg of the shell, the system starts to expand (Figure 1).

The expansion goes on when heated until reaching a maximum size of particles, about 40–50 times volumetrically according to the supplier. After this point, and if the system is still heated to a higher temperature or for a long time, the shell may shrink or collapse, due to a chemical degradation or a loss of the inner volatiles. The PEA used in TEAF is supposed to start blowing around 113–123 °C and to reach maximum size around 163–175 °C (supplier data). As the shell needs to expand and contain the alkane, it is also important that the polymer used has a glass transition (Tg) lower than the curing temperature of the adhesive. The shell polymer also needs to not degrade at the curing temperature.

2.2. Curing Process

The standard curing parameter industrially used for the TEAF formulation is 25 min at 175 °C in an oven. To estimate the influence of curing parameters on the expansion it was chosen to vary the curing temperature between 130 and 175 °C and the heating rate between 1 °C·min⁻¹ and 30 °C·min⁻¹. The curing time, when reaching the isothermal temperature, is 25 min for each experiment. Considering the different temperatures and heating rates, the time of the curing process will vary from one experiment to another. When studying the isothermal curing temperature, the heating rate was fixed at 20 °C·min⁻¹. When studying the heating rate, the curing temperature was fixed at 175 °C. Curing parameters variations were done both on adhesive formulations and on the PEA only.

2.3. Methods and Apparatus

Thermal properties were analyzed by DSC using DSC Q200 equipment (TA Instruments, Guyancourt, France) with heating parameters depending on the experiment. For the variation in the heating rate, the thermogram is recorded between 0 °C and 250 °C and two cycles were performed in order to ensure that the cross-linking of the adhesive is complete. Analyses were done at 1 °C·min⁻¹, 10 °C·min⁻¹, 20 °C·min⁻¹ and 30 °C·min⁻¹. Influence of the isothermal curing temperature was analyzed by DSC with a first ramp (cycle 1) between 0 °C and the final temperature at 10 °C·min⁻¹ followed by an isothermal step of 25 min before cooling at 0 °C. A second ramp (cycle 2) was then applied to 250 °C at 10 °C·min⁻¹ to ensure that the cross-linking was complete. These experiments were done on adhesive formulations.

Curing of the adhesive was studied using a Vertex 70 Fourier Transform InfraRed spectrometer (FTIR) (Bruker, Wissembourg, France) with an Attenuated Total Reflectance (ATR) device, between 4000 and 400 cm⁻¹, with a resolution of 4cm⁻¹.

The study of the expansion of the PEA during the heating process was done using an Olympus BX51 optical microscope with an Olympus DP20 CCD camera (Vision Engineering, Surrey, UK), with a magnification from ×50 to ×500 and used in transmission mode. The variation in the temperature were controlled under microscope using an adapted Peltier heating plate LTS350 (Linkam, Surrey, UK) controlled by a temperature module TP 94 (Linkam, Surrey, UK). This device enables the direct observation of samples under optical microscope during a specified heating process. For each experiment, around 0.5 g of PEA was deposited on the heating plate as the most homogeneous layer as possible.

Image analyses were performed using ImageJ 1.54g software (developed by Wayne Rasband, Open-Source, Public Domain) mainly for the observation of PEA particle sizes and their variation during heating processes. When heated at isothermal temperature, it was chosen to measure particle sizes each 5 min during 25 min after reaching the isothermal temperature. Experiments were done between 130 °C and 175 °C with a heating rate of 20 °C·min⁻¹. Diameter measurements were done manually on 250 particles for each heating condition in order to have a relevant standard deviation and representative values.

The original characterization method proposed in this work to estimate the expansion of the adhesive and the spreading quality with a model substrate also uses the optical microscope, the Peltier heating plate and the temperature module. A piece of raw adhesive of 10 × 10 × 0.50 mm³ (about 0.05 g) was intercalated between two microscope glass slides, separated by adhesive Teflon stripes in order to control the thickness. Teflon stripes were placed to form a 10 × 10 mm² space to contain the adhesive formulation and to ensure the sealing and avoid edge effects. The specimen was then placed on the heating plate and a pierced metal weight of 24.4 g is placed on the upper glass slide to maintain the thickness during the test. The hole pierced on the metal weight has a diameter of 5 mm. During the curing process, the adhesive sample will expand and eventually reach the upper glass slide. The microscopic observation of the upper slide enables to estimate the quality of the contact, the rate and the homogeneity of the spreading. The quantification of the spreaded surface was made using ImageJ software. Images of surfaces were first contrasted to obtain the surface in contact with the glass plate as black spaces and the surface not in contact as white spaces (brightness between 140 and 180 and contrast at maximum). The threshold was set using the Huang method to analyze the surface in contact as %Area (threshold between 40 and 60% depending on the image quality). The surface analysis was made on five images for each sample.

In the conditions of a free expansion, the adhesive joint’s thickness may increase by more than 100% (Figure 2). In the case of a metal/adhesive/metal assembly, the thickness of the joint is determined and fixed before the curing process. The present study considered two cases. The first case is a fixed thickness of 0.50 mm and the second corresponds to a fixed thickness of 0.70 mm. As the initial thickness of the adhesive joint is around 0.50 mm, only the second case needed theoretically a volume compensation associated with the expansion.

The spreading capacity of the adhesive joint was estimated using ImageJ for the characterization of the “spreaded surface”. The expansion of the adhesive joint between the glass slides was recorded by the CellSens software (Version 1.16) (in conjunction with the Olympus camera) and videos were edited using DaVinci Resolve software (Version 18.6.2) (BlackMagic Design, Melbourne, Australia) in order to observe changes and modifications during the experiment. An example of raw video is presented as supplementary information.

3. Results and Discussion

3.1. Expansion of the Physical Agent

As seen in Figure 1 under optical microscope in transmission, the PEA looks like relatively homogeneous particles, with an average diameter of about 15–20 µm. It remains stable until 140 °C, then the blowing quickly starts until 150 °C, creating hollow spheres. After a curing isotherm of 25 min at 175 °C, the expansion agent reaches an average diameter of about 60–70 µm. The PEA was observed under an optical microscope and a heating process was applied. The heating rate was fixed to 20 °C·min⁻¹ until the set isothermal temperature. Then, a curing isotherm of 25 min was carried out and the average diameter of particles was measured each 5 min. The average diameter value was calculated from 250 spheres (Figure 3a). For all measurements, the standard deviation was also calculated from 250 spheres and presented in

Table 1. Average particle diameter measurement values of PEA during 25 min at different curing temperatures.

Curing Temperature	130 °C	140 °C	150 °C	160 °C	165 °C	170 °C	175 °C
0 min	17.16 ± 0.86 µm	21.44 ± 0.72 µm	76.56 ± 1.12 µm	74.70 ± 3.43 µm	75.45 ± 2.04 µm	77.69 ± 2.37 µm	75.09 ± 2.88 µm
1 min	/	51.33 ± 2.33 µm	/	/	/	/	/
2 min	/	71.20 ± 1.57 µm	/	/	/	/	/
5 min	16.24 ± 0.70 µm	76.75 ± 2.83 µm	82.33 ± 3.11 µm	81.39 ± 2.86 µm	90.78 ± 4.32 µm	91.97 ± 4.44 µm	88.63 ± 3.97 µm
10 min	16.61 ± 0.97 µm	77.50 ± 2.68 µm	81.94 ± 4.10 µm	83.24 ± 3.16 µm	88.63 ± 3.43 µm	89.92 ± 4.47 µm	82.14 ± 3.11 µm
15 min	16.61 ± 0.95 µm	76.20 ± 3.81 µm	79.90 ± 2.99 µm	82.33 ± 3.48 µm	87.52 ± 2.37 µm	87.53 ± 3.38 µm	74.90 ± 3.42 µm
20 min	17.00 ± 1.01 µm	78.60 ± 2.93 µm	87.14 ± 3.35 µm	79.35 ± 2.97 µm	84.54 ± 3.22 µm	86.58 ± 4.33 µm	68.41 ± 3.48 µm
25 min	17.00 ± 0.76 µm	81.94 ± 3.10 µm	92.72 ± 4.23 µm	82.69 ± 4.13 µm	81.94 ± 4.82 µm	76.39 ± 6.10 µm	66.55 ± 7.33 µm

. Average sphere volumes as a function of time are also presented (Figure 3b).

Even after 25 min at 130 °C, above the Tg of the shell provided by the supplier (around 110 °C), the PEA does not blow and remains stable with an aspect close to its initial state. At 140 °C, blowing starts within a minute and reaches a stabilized average size after 5 min (76.75 ± 2.83 µm). The diameter slightly increases after 15 min at 140 °C, reaching a maximum of 81.97 ± 3.10 µm after 25 min. The observation is similar at 150 °C, with a higher value. The diameter stabilizes around 80 µm between 5 and 15 min and then increases to reach a maximum of about 92.72 ± 4.23 µm after 25 min. At 160 °C, the average diameter of spheres is relatively stable during the curing isotherm and is about 80 µm. At higher temperatures the behavior of the PEA differs, as the maximum average diameter is reached after 5 min and is followed by a constant decrease until 25 min of heating. At 165, 170 and 175 °C, the maximum average diameter reached after 5 min of heating is about 90 µm, and the following decrease is more significant while the temperature increases. So, the average diameter reaches 81.94 ± 4.82 µm at 165 °C, 76.39 ± 6.10 µm at 170 °C and 66.55 ± 7.33 µm at 175 °C. The increase in standard deviation indicates a larger size distribution of particles due to the shrinkage of the polymer shell.

Even if the average diameter of spheres decreases while heated, they keep a relatively perfect spherical shape. The decrease in volume can be related to the release of the inner volatiles by cracks in the shell or by diffusion through the wall. The aspect of the PEA after 25 min at 150 °C and 175 °C is presented in Figure 4.

At room temperature the remaining inner volatiles condense and create empty spaces inside the sphere. So, the shell may collapse in the available free space. Apart from the deformation of the shell, the PEA keeps a spherical shape after cooling (Figure 5). This confirms that the expansion mechanism is irreversible as the particles aspect after heating does not correspond to their initial shape, size and distribution (Figure 1, left picture).

If the PEA is heated at 175 °C for longer periods the spheres begin to coalesce and bond to each other. Shells progressively shrink due to the release of volatiles and the degradation of the polymeric shell. Figure 6 presented the example of the PEA heated at 175 °C during 1 h.

Another modification observed during the heating of the PEA was a change in color at macroscopic scale. When heated in oven during 25 min at 160 °C, the initial white powder strongly expanded and turned into a white translucent light powder. Under an optical microscope, this sample is very similar to the one observed at the maximum of expansion and is composed of various hollow spheres (Figure 7).

When heated at 175 °C during 25 min, the expanded hollow spheres coalesce and create interconnections between the shells. At the macroscopic scale, this interconnected network induces the formation of an orange foam-like structure, which is relatively cohesive. Under optical microscope, the average diameter of marbles decreases while shells coalesce. This effect may be due to the release of the inner volatile core through cracks of by diffusion through the shell (Figure 8).

3.2. Study of the Curing of the Adhesive Formulation

The curing reaction between BADGE and Dicy is quite complex, and seems to highly depends on the curing conditions such as the temperature, the heating cycle, the environment, the catalyst, the additives or also the stoichiometry. The major reaction consists of an addition leading to the opening of the BADGE oxirane ring by Dicy. Three main reactions may occur, due to the high functionality of the hardener, leading to the formation of polyether structures. In each case, the reaction leads to the formation of hydroxyl groups (Figure 9) [18,19,20].

Then, nitrile groups react with hydroxyl groups in order to create cross-linking bonds. However, various reaction mechanisms were proposed. According to Saunders et al., the addition of the hydroxyl with the nitrile leads to an iminoether, and then rearranged to form a substituted guanylurea (Figure 10). This reaction could be intermolecular or intramolecular, leading to formation of cyclic structures. Measured by IR, this reaction can be characterized by the appearance of a carbonyl peak around 1740 cm⁻¹ and the disappearance of the nitrile peak at 2180 cm⁻¹ [21].

Zahir favored the hypothesis of an intramolecular mechanism, leading to the elimination of cyanamide and the formation of 2-iminooxazolidine and 2-aminooxazoline. The C=N bonds of these particular cycles can be identified by IR with a peak at 1660 cm⁻¹. The previous structures may rearrange, leading to urea cyclic groups (1700 cm⁻¹) or react with new hydroxyl functions and open, forming urethane bonds (1740 cm⁻¹) [22] (Figure 11).

According to Gilbert, the addition of the hydroxyl group on the C=N bond of the Dicy is associated with the elimination of ammonia and the formation of a pentacycle of iminooxazolidine. The hydrolysis of the imine group then produces a cyclic urethane, revealed by an IR peak at around 1750 cm⁻¹ [19] (Figure 12).

According to the reaction mechanisms presented in the previous part, it can be possible to follow the cross-linking of the adhesive thanks to FTIR-ATR measurements. Figure 13 presents the FTIR-ATR analysis of the TEAF formulation before and after curing at 175 °C for 25 min.

The decrease in peaks at 910 and 874 cm⁻¹ is related to the opening of oxirane cycles, and the appearance of a peak between 1682 cm⁻¹ can be related to formation of cross-linked bonds (urea, urethane…). As the hardener was introduced in excess, it can be difficult to follow the cross-linking with the evolution of its characteristic peaks. Due to the existence of various mesomer forms of the Dicy, the modification of the double peak at 2200 cm⁻¹ to a single band can be related to a rearrangement of bonds in the hardener after the curing cycle [23,24]. The appearance of a peak at 1686 cm⁻¹ and the shift in a peak to 1647 cm⁻¹ are related to cross-linking bonds formed during the reaction. These peaks possibly correspond to C=O and C=N bonds or urea groups and 2-iminooxazolidine/2-aminooxazoline, according to the mechanism proposed by Zahir [22].

The absence of a characteristic peak at 1740 cm⁻¹ after curing supposes that the reaction with a hydroxyl group after the rearrangement step is not favored. It seems that the cross-linking reaction in the TEAF formulation follows the mechanism leading to cyclic urea groups. A possible hypothesis is that the observed reaction may be induced by the urea-based accelerator in the formulation. The study of a formulation without this accelerator could be useful to confirm this observation.

This diversity of hypothetical reactive paths explains the wide range of reaction temperatures that can be observed by DSC analysis. As the temperature ramp may strongly influence the curing reaction of the BADGE/Dicy system, DSC analyses were carried out on the TEAF adhesive formulation at 1 °C, 10 °C, 20 °C and 30 °C·min⁻¹ between 0 and 250 °C (Figure 14) for two identical ramps. The measured values are presented in

Table 2. Values of DSC analysis between 0 °C and 250 °C of the TEAF adhesive formulation at different heating rates.

	Measured Temperature (°C)	Identification	Enthalpy of Reaction (J·g−1)
1 °C·min−1	-	Tg of cured adhesive
	120	Tg of PEA
	-	Expansion of PEA
	102	Start of curing
	131	Max of curing	140
	169	End of curing
10 °C·min−1	49	Tg of raw adhesive
	87	Tg of cured adhesive
	119	Tg of PEA
	129	Start of curing
	138	Expansion of PEA
	160	Max of curing	120
	193	End of curing
20 °C·min−1	47	Tg of raw adhesive
	85	Tg of cured adhesive
	117	Tg of PEA
	140	Expansion of PEA
	148	Start of curing
	185	Max of curing	130
	247	End of curing
30 °C·min−1	51	Tg of raw adhesive
	88	Tg of cured adhesive
	117	Tg of PEA
	141	Expansion of PEA
	154	Start of curing
	194	Max of curing	120
	246	End of curing

.

Considering that the conversion of the initial material is total after the exothermic peak, it is possible to estimate the degree of conversion at each heating rate, when the set temperature of 175 °C is reached. At 1 °C·min⁻¹, the conversion is 100%, 55% at 10 °C·min⁻¹, 18% at 20 °C·min⁻¹ and 6% at 30 °C·min⁻¹.

Glass transitions of the raw adhesive and the shell of the PEA are relatively stable, whatever the heating rate. The Tg of the raw adhesive is around 45–50 °C in each case and the around 85–90 °C for the cured material, but it was not possible to determine the value with precision at 1 °C·min⁻¹, as the endotherm seems to be too little relative to the DSC sensitivity.

The Tg of the shell of the PEA is around 115–120 °C for each experiment. The variation in heat flow associated with expansion is also present in each test, except in the one at 1 °C·min⁻¹. Indeed, considering that the temperature related to the expansion is around 140 °C in each case, the possible variation may be hidden within the exothermic peak, between 100 and 170 °C, with a maximum at 130 °C. It is also possible that, as the curing started at 100 °C and reaches a maximum at 130 °C, the viscosity of the adhesive formulation increases too much preventing an efficient expansion.

For the test at 10 °C·min⁻¹, the expansion of the system occurs 10 °C after the beginning of curing. So, it can be considered that the viscosity of the formulation is increasing and that the expansion is not optimal. For both tests at 20 and 30 °C, the starting of the curing occurs after 140 °C, and the expansion of the PEA. Considering that, after the Tg, the viscosity of the system decreases progressively before the curing, the separation of the expansion and the curing may favor the blowing of the polymer layer as a minimum of viscosity is reached. As the beginning of the curing is reached, the cross-linking mechanism hardens the system.

Regarding the results, it seems that the modification of the heating rate delays the initiation step of the polymerization. When started, the curing presents a similar enthalpy of reaction, relatively independent of the heating rate and about 120 J.g⁻¹. The synchronization of the curing and the expansion mechanisms is an essential point to ensure an optimal spreading within the adhesive assembly.

DSC measurements at 10 °C·min⁻¹ show that the beginning of the curing process occurs around 130 °C, with a maximum peak at 160 °C. It is supposed that once the cross-linking is initiated, the reaction occurs until the stabilization of the structure. Isothermal DSC measurements at 130, 140, 160 and 175 °C during 25 min, followed by a second cycle between 0 and 250 °C, were carried out to study the curing reaction. The first ramp from 0 °C to the isothermal temperature, the isotherm during 25 min and the cooling to 0 °C are defined as “cycle 1”. The second ramp from 0 °C to 250 °C and the cooling to room temperature are defined as “cycle 2”. As the curing peak may not appear completely in the curve as a function of time, the measurement of the enthalpy of reaction has to be done on the curve expressed as a function of time. Results are presented on Figure 15 and

Table 3. Values DSC analysis the TEAF adhesive formulation at different heating temperatures.

	Measured Temperature (°C)	Identification	Enthalpy of Reaction (J·g−1)
130 °C	46	Tg of cured adhesive
	66	Tg of cured adhesive
	117	Tg of PEA
	141 (2nd cycle)	Expansion of PEA
	130	Start of curing
	After 9 min at 130 °C	Max of curing	85
	After 24 min at 130 °C	End of curing
140 °C	43	Tg of raw adhesive
	69	Tg of cured adhesive
	117	Tg of PEA
	138	Expansion of PEA
	130	Start of curing
	After 4 min at 140 °C	Max of curing	80
	After 15 min at 140 °C	End of curing
160 °C	44	Tg of raw adhesive
	84	Tg of cured adhesive
	114	Tg of PEA
	139	Expansion of PEA
	130	Start of curing
	After 40 s at 160 °C	Max of curing	115
	After 12 min at 160 °C	End of curing
175 °C	45	Tg of raw adhesive
	84	Tg of cured adhesive
	114	Tg of PEA
	139	Expansion of PEA
	130	Start of curing
	165	Max of curing	130
	After 10 min at 175 °C	End of curing

.

By expressing DSC curves as a function of time, all isothermal measurements present an entire curing peak, starting at 130 °C and ending after a certain time at the set temperature. Considering the ending time of the curing for each experiment, the curing lasts 24 min at 130 °C, 16 min at 140 °C, 15 min at 160 °C and 14.5 min at 175 °C (Figure 16).

Regarding the values of the enthalpy of reaction and the glass transition of the cured adhesive for experiments at 130 and 140 °C, it seems that the curing was not as advanced as for the experiments at 160 and 175 °C. The lower glass transition of the cured adhesive traduces a higher mobility of molecular chains and a lower concentration of cross-linked points. According to the results, it seems to be important to reach the maximum temperature of curing, at a certain heating rate, to favor the cross-linking of the BADGE /Dicy system.

3.3. Expansion and Spreading Characterization Under Optical Microscope

In the conditions of a free expansion, the adhesive joint’s thickness may increase by more than 100%. In the case of a metal/adhesive/metal assembly, the thickness of the joint is determined and fixed before the curing process. The present study considered two cases. The first case is a fixed thickness of 0.50 mm and the second corresponds to a fixed thickness of 0.70 mm. As the initial thickness of the adhesive joint is 0.50 mm, only the second case, theoretically, needed an expansion to fill the empty volume.

The TEAF formulation can be studied under an optical microscope using an adapted heating plate. In order to observe the surface of the sample during heating, and its capacity to compensate a gap with the expansion, a visualization setup was prepared. A piece of raw adhesive was intercalated between two glass plates, separated by Teflon stripes in order to fix the thickness. The specimen was then placed on the heating plate and a metal piece was placed on the upper plate to maintain the desired thickness during the test. The goal is to observe the modification at the surface of the adhesive while heated, depending on the heating ramp and the curing isothermal temperature. At a certain time, and due to the expansion, the surface of the adhesive will reach the upper glass plate and create contact. This step, named spreading, was visualized to study the quality of the spreading (and eventually the adhesion) and the capacity to fill the empty volume related to the expansion.

During heating tests, different steps of surface changes can be observed (Figure 17).

After the initial state, at a certain temperate higher than the Tg, the adhesive softens and cavities appeared. This step also corresponds to a slight decrease in the thickness due to a spreading on the lower glass plate. When the expansion of the physical agent starts, the surface of the adhesive presents cavities that appear progressively until a “pop” temperature. The step corresponds to a quick appearance of PEA at the surface of the adhesive, at a higher temperature than the formation of the porous structure. And then, the adhesive may reach the glass plate and spreads until stabilization and hardening related to the curing reaction. The dark areas represent the contact between the adhesive and the glass plate and the bright areas represent an absence of contact. When the spreading stops, it is possible to estimate the quality of the contact surface (and indirectly the adhesion) related to the curing conditions.

First, the spreading capacity was studied with a temperature ramp of 20 °C·min⁻¹ from 20 to 175 °C, and with an isothermal curing of 25 min at 175 °C. The samples were prepared according to the method previously presented. Two configurations were studied: 0.50 mm and 0.70 mm of thickness between the glass plates. The steps of surface change were described by their temperature ranges of occurrence (Figure 18,

Table 4. Temperature values of the expansion steps under optical microscope after 25 min at 175 °C with a heating rate of 20 °C.min⁻¹ and a 0.50 mm and 0.70 mm gap between glass plates.

Conditions	0.5 mm	0.7 mm
Cavities	120 °C	127 °C
Expansion	-	143 °C
“Pop”	-	165–167 °C
Spreading	138–147 °C	175 °C (t0–1 min)
Spreaded surface	99.4 ± 0.5%	56.4 ± 3.2%

).

It was not possible to observe the expansion and the “pop” steps for the specimens at 0.50 mm because the spreading occurs before, preventing other visualizations. At the end of the tests, the spreading quality seems to only depend on the set thickness. At 0.50 mm, the contact between the adhesive and the glass plate is total, whereas only the half of the surface seems to be in contact with the adhesive at 0.7mm. The formation of cavities occurs at a similar temperature for all samples (115–125 °C) such as the expansion and the “pop”. The contact between the adhesive and the plate begins at a similar temperature, around 135 °C. At 0.70 mm, the contact begins at 175 °C and lasts around 1 min. In order to optimize the adhesive/plate contact, it is necessary to synchronize the expansion and the curing mechanisms. The expansion provides a better effect when the viscosity of the adhesive remains relatively low.

In a second attempt, various temperature ramps were applied on the adhesive formulation. Like the first attempt, the changes at the surface were observed and associated with corresponding temperatures. The observation was made with temperature ramps of 1 °C.min⁻¹, 10 °C.min⁻¹, 20 °C.min⁻¹ and 30 °C.min⁻¹, with the specimens at 0.50 mm (Figure 19,

Table 5. Temperature values of the expansion steps under optical microscope after 25 min at 175 °C at different heating rates with 0.50 mm gap between glass plates.

Conditions	1 °C·min−1	10 °C·min−1	20 °C·min−1	30 °C·min−1
Cavities	120 °C	122 °C	120 °C	113 °C
Expansion	135 °C	149 °C	-	-
“Pop”	155 °C	-	-	-
Spreading	140	140–164 °C	138–147 °C	130–135 °C
Spreaded surface	5.7 ± 2.1%	99.1 ± 0.2%	99.4 ± 0.5%	98.8 ± 0.5%

) and 0.70 mm (Figure 20,

Table 6. Temperature values of the expansion steps under optical microscope after 25 min at 175 °C at different heating rates with 0.70 mm gap between glass plates.

Conditions	1 °C·min−1	10 °C·min−1	20 °C·min−1	30 °C·min−1
Cavities	120 °C	118 °C	127 °C	118 °C
Expansion	135 °C	146 °C	143 °C	150 °C
“Pop”	155 °C	155–159 °C	165–167 °C	-
Spreading	-	175 °C (40 s–4 min)	175 °C (t0–1 min)	148–170 °C
Spreaded surface	0.9 ± 0.3%	17.9 ± 8.6%	56.4 ± 3.2%	98.6 ± 0.3%

).

As previously mentioned, when the adhesive is heated above the Tg, the material softens and the thickness slightly decreases. So, the expansion is needed to compensate the height loss to ensure the contact. For the specimens at 10, 20 and 30 °C.min⁻¹, the spreading at the surface of the glass plate is total at the end of the cycle. However, this step begins at a lower temperature when the heating rate increases, and during a shorter interval. It seems that the increase in the heating rate favors the spreading capacity of the adhesive. For the specimen at 1 °C.min⁻¹, the spreading started at 140 °C but only consists of isolated spots and cannot be considered as a total spreading. The increase in viscosity due to the curing, associated with the decrease in thickness related to the softening, limits the expansion of the adhesive.

With a thickness of 0.70 mm between the glass plates, the influence of the heating rate is more obvious (Figure 20, Table 6).

The spreading quality increases with the temperature heating rate as the contact between the adhesive and the glass plate occurs at a lower temperature. A temperature ramp of 1 °C.min⁻¹ is insufficient to provide a contact, possibly related to too important increase in viscosity before the expansion of the PEA. Even if the contact occurs at 175 °C for the specimens at 10 and 20 °C.min⁻¹, the spreading mechanism begins when 175 °C is reached at 20 °C.min⁻¹ and lasts about 1min, whereas the specimen at 10 °C.min⁻¹ needed about 1min at 175 °C to start the spreading. In this case, the spreading stabilizes after about 4 min. The heating rate of 20 °C.min⁻¹ leads to a better spreading quality. In the case a higher heating rate (30 °C.min⁻¹), the spreading begins around 150 °C and stabilizes at 170 °C, leading to a complete contact between the adhesive and the glass plate.

4. Conclusions

This work presented an original in situ method to estimate the quality of the expansion of thermoset adhesive formulations between two substrates. A physical expansion agent (PEA) was chosen as the reference for this experiment. First, the expansion capacity of the PEA was estimated by optical microscope observation with a Peltier effect heating plate to determine the expansion profile as a function of curing temperature and heating rate. Results show a stable expansion at 20 °C.min⁻¹ for curing temperatures between 140 °C and 160 °C. Under 140 °C the PEA does not expand at an optimal rate and above 160 °C the PEA starts to degrade and collapse during the curing process. The most efficient temperature for the expansion was measured for a 150 °C isothermal process, which is lower than the ideal curing temperature of the reference adhesive formulation (175 °C). These observations are specific to the chosen reference and will differ when using a different expansion agent. The relevance of the proposed method for other expansion agents will be investigated in further experiments. The curing process was then characterized by DSC and showed that the curing temperature and the heating rate has a major influence on the adhesive properties and on the curing itself. A slow heating rate lowers the start of the curing and may limit the expansion and a fast heating rate leads to an expansion occurring before the curing. These situations are not optimal for the expansion process of the adhesive. A synchronization between the expansion and the curing has to be found and can be determined by both DSC and microscopic observations. Finally, these coupled methods that allow to characterize the quality of the expansion were used on thermoset adhesive formulations containing PEA by applying various heating rates and with 0.50 mm and 0.70 mm gaps between the two substrates to bond. The optical microscope and the heating plate allowed for the observation of the expansion, the contact and the spreading of the formulation depending on heating conditions. A difference was clearly observed, showing that heating rate may impact the quality of the adhesive contact for self-expanded formulations. The aim is to mimic curing conditions and gaps between substrates applied in the industry to ensure that the contact, and therefore the adhesive, is the most efficient for a specific application. In the case of metal/polymer/metal assembly used in the automotive industry, the aim of this method is to control that the curing conditions applied are able to fill gaps in stamped car pieces to optimize adhesion and to prevent a loss in durability due to preferential water diffusion pathways. At the current state, the proposed method is limited to adhesive formulations presented as a solid film before curing and may not apply to liquid epoxide or polyurethane formulations. More experiments will be carried out on this method with a wider range of formulations and expansion agents to validate its relevance and efficiency.