Photopolymerization of Acrylated Epoxidized Soybean Oil: A Photocalorimetry-Based Kinetic Study.

Photocure kinetics of acrylated epoxidized soybean oil (AESO) was studied via photocalorimetry without adding any diluent/comonomer, in the presence of two different photoinitiators, namely, 2,2-dimethoxy phenylacetophenone and 1-hydroxycyclohexyl phenyl ketone. The effect of varying photoinitiator concentration, light intensity, and temperature on the extent of crosslinking was calculated from the ratio of experimentally measured reaction enthalpy to the theoretical enthalpy of reaction (ΔH theoretical). Photocuring of AESO was observed to be a second-order reaction exhibiting autocatalytic behavior. Nevertheless, due to the occurrence of vitrification, incomplete crosslinking (α ≠ 1) was observed in most curing conditions. Rate constants and activation energies were determined using both nonlinear model-fitting and model-free isoconversional methods. Activation energy, as determined from the model-free isoconversional method, was observed to increase as the reaction proceeded, indicating the shift in cure mechanism from kinetic-controlled to diffusion-controlled. Finally, the reaction termination mechanism was observed to be a combination of second-order and primary radical termination mechanisms.

Introduction

Photocuring of thermosets, especially multifunctional acrylates, is commonly used in coatings, lithography, liquid optically clear adhesives, and screen printing[1] due to its faster curing, superior performance, reduced emission of volatile organic compounds, and lower costs.[1−3] Conventionally, these multifunctional acrylates have been derived from crude oil, a fossil resource, thereby generating concerns regarding their ecological viability.[1] To address this, initial efforts[4−9] have been made to explore triglyceride-based bioacrylates as sustainable alternatives. Yet, among these efforts, only three studies[4,5,8] have attempted to monitor the progression of cure with time. Further, all of these efforts[4,5,8] fail to dwell in greater detail on understanding the photocure kinetics of these bioacrylates, be it determining their reaction propagation and termination mechanisms or understanding the influence of various process parameters (such as the type and concentration of the photoinitiator used, light intensity, or temperature) on activation energies and rate constants of the cure reaction.[10] This makes it challenging to exploit the full potential of these triglyceride-based acrylate systems, as a detailed knowledge base on their photocure kinetics is necessary for selecting appropriate process conditions. Unfortunately, the knowledge on cure kinetics of crude oil-based acrylates cannot be extended to triglyceride-based bioacrylates due to significant differences in their molecular architecture.[11−14] Typically, conventional acrylates possess a cyclic aliphatic or aromatic backbone with acrylic groups present as the end group. However, triglyceride-based bioacrylates possess an aliphatic backbone consisting of three fatty acid chains attached to the glycerol center, where the acrylic group is in the middle of the fatty acid chain.[14] This variance highlights the need for a detailed study that provides a holistic analysis of photocure kinetics of triglyceride-based acrylates.

Generally, photocure kinetics of any material is studied using two techniques: photocalorimetry (i.e., photo-differential scanning calorimetery (DSC) and real-time Fourier transform infrared (RT-FTIR) spectroscopy.[15,16] While photo-DSC enables us to monitor cure kinetics by measuring the change in enthalpy during curing, RT-FTIR monitors the change in intensity of IR absorption spectra of any specific functional group.[17] Although RT-FTIR provides useful insight about the rate of chemical reaction, particularly when more than one reacting species is involved, making it a complementary technique to photo-DSC, it also suffers from several limitations. First, RT-FTIR offers poor temperature control, which when combined with the exothermic nature of the cure reaction, makes it difficult to maintain isothermal conditions during the process, resulting in inaccurate kinetics calculations.[18] Second, this technique produces a large amount of raw spectra for just a single experiment, making it extremely difficult to analyze the entire gamut of results for any given sample/reaction, even as these results are susceptible to any phase or baseline changes.[19] Finally, RT-FTIR is a surface-based technique when carried out via ATR mode, i.e., it determines the extent of curing only at the surface for thick samples due to the inability of the infrared beam to penetrate through the sample.[20,21] In contrast, photo-DSC remains the oldest technique available to analyze the photocure kinetics of any reaction in a highly reliable and robust manner.[22,23] Additionally, photo-DSC helps us obtain the average degree of cure for the material as a whole. Together, all of these aspects render the photo-DSC technique advantageous over RT-FTIR in developing a holistic understanding of the kinetics of any material/substance.

Even as the photo-DSC technique can provide us adequate and optimal experimental data, any true study of reaction kinetics is only possible by analyzing such results using kinetic models. Typically, reaction kinetics is studied either via model-fitting method and/or model-free isoconversional method.[24] Of these two types of methods, model-free isoconversional methods are well-known for being more realistic and accurate in predicting reaction kinetics, as they are free from any assumptions and determine the variation in activation energy with progression of the reaction.[25−27] Nevertheless, since their inception, such methods have been mainly used to understand nonisothermal reaction kinetics, with model-fitting methods used predominantly in the case of their isothermal counterparts.[25,28] This can be ascribed to the premise that isoconversional methods are inaccurate for isothermal kinetics but are accurate for nonisothermal reactions, a premise that arose from the initial use of isoconversional methods for thermal degradation reactions.[25,28] However, photocuring of any thermoset under isothermal conditions leads to gelation and vitrification of the polymer, an isoconversional phenomenon.[29] This suggests that determining photocure kinetics via isoconversional methods may give useful insights about the reaction that may otherwise not be obtainable via use of model-fitting methods.

Hence, this work aims at comprehensively understanding the photocure kinetics of a triglyceride-based bioacrylate by monitoring it using photocalorimetry. In a prior work[30] from our group, photo-DSC was used to characterize the cure kinetics of an unsaturated polyester resin, containing 45 wt % styrene as a diluent, where the derivations were targeted at model-based process control and optimization. In this study, we chose acrylated epoxidized soybean oil (AESO) as the bioacrylate and used two different photoinitiators (PIs) belonging to the Type-I category (radical formation via unimolecular bond cleavage), namely, 2,2-dimethoxy phenylacetophenone (DMPA) and 1-hydroxycyclohexyl phenyl ketone (HCPK), without the presence of any diluent/comonomer. While DMPA possesses a short-lived excited triplet state and exhibits a yellowing character,[31] HCPK is a nonyellowing PI that is widely used for curing acrylate monomers.[32] Experimental observations, as noted for varying process conditions, were analyzed using both the model-fitting method and isothermal model-free isoconversional method[10,33] to assess their relative suitability in understanding the reaction kinetics of such systems. Finally, the light intensity exponent (β) was estimated to determine the termination mechanism of the curing reaction.

Background on Cure Kinetics

Model-Fitting Method

In general, heat flow, measured using photo-DSC, is assumed to occur solely due to one reaction, namely, the crosslinking of acrylate groups that are present in the AESO molecule. This assumption is valid in this work due to the absence of any solvent and/or comonomer. Hence, the rate of conversion (or crosslinking) (Rp or dα/dt) can be calculated using eq 1,(34) where ΔHtotal is the total enthalpy of the reaction at 100% crosslinking and dH/dt is the heat flow measured under any isothermal DSC condition.Upon integrating eq , the degree of conversion (α) can be obtained from eq , where ΔHtotal is the total enthalpy of a reaction at 100% crosslinking (i.e., α = 1).[20,34]To understand the variation in α with time, i.e., the progression of cure (or its kinetics), several cure kinetics models, both phenomenological and mechanistic ones, have been proposed and well discussed in the literature.[24,35,36] These models can be classified into two groups: nth order models that can be further categorized into accelerating and decelerating reaction models and autocatalytic reaction models. However, according to the International Confederation for Thermal Analysis and Calorimetry (ICTAC) Review Committee recommendations,[24] for a cure kinetics model to be reliable and robust, it is necessary that the model is capable of taking into account the variation in the extent of conversion (α) via both the nth order and autocatalytic reaction models.[24] One such model that has been widely used to understand cure kinetics of various systems is the Kamal–Sourour[36] model, which is expressed using eq . Here, dα/dt is the rate of reaction, k1 and k2 are the rate constants that correspond to the nth order and autocatalytic reaction models, while m and n refer to the orders of crosslinking and monomer consumption, respectively.However, it is widely known that during the isothermal cure of a thermoset, the cure reaction can cease due to the formation of a glassy phase that traps free radicals, thereby preventing cure completion (i.e., α ≠ 1). Hence, the Kamal–Sourour[36] model was modified, as shown in eq , to capture this incomplete cure, where the term “1” in eq is replaced by αmax, which refers to the maximum degree of conversion that can occur (such that αmax ≤ 1) during the reaction.Since the objective of this study was to determine the optimal values for all cure kinetic parameters, i.e., reaction rate constants (k1, k2) and reaction orders (m, n), these were obtained by matching model-predicted dα/dt (obtained using eq ) as closely possible with experimentally measured dα/dt values through curve fitting. Mathematically, such close matching between model-predicted and experimentally measured values of dα/dt is undertaken by using the cost function (defined in eq )Here, RSS is the residual sum of least squares, N is the total number of data samples, i is the time index, Rp exp is the experimentally measured rate of cure, and Rp calc is the model-predicted rate of cure. Essentially, the model functions on the premise of minimizing RSS, with the lowest value of RSS obtained for the most accurate values of all four desired parameters. Upon determining the values of (k1, k2), activation energy (Ea) of the curing reaction can be calculated using the Arrhenius equation (defined in eq ), where A is the pre-exponential factor, R is the universal gas constant (8.314 J/(mol K)), and T is the temperature (K).

Model-Free Isoconversional Method

Since model-fitting methods are well-known for giving Arrhenius parameter values (activation energy and rate constants) that are notoriously uncertain,[37] the recent ICTAC Review Committee has recommended the use of model-free isoconversional methods to predict the kinetic behavior of a chemical reaction in a realistic manner.[24] In this regard, activation energy (Ea,α) must be determined at different values of α = 0.05–0.95 with a step size of not larger than 0.05 for a better understanding of cure kinetics over time. Hence, a simplified form of the integral isoconversional method (eq ) was selected to understand and predict the photocure kinetics of AESO, where tα refers to the time taken to reach a particular extent of degree of conversion (α) at different temperatures (T) and Ea,α is the activation energy of reaction for the specific value of α.Using a linearfit for the plot between ln(tα,) and the reciprocal of isothermal test temperature (T), the slope was used to determine Ea,α as a function of α.

Termination Mechanism: Light Intensity Exponent

To determine the termination mechanism of the cure reaction, experimentally obtained dα/dt values were fitted as a function of α using eq to estimate the light intensity exponent (β).[1,38] Here, dα/dt is the rate of reaction, k(p) is a parameter that depends on the extent of conversion (α), x and β are exponents, and I0 refers to the intensity of UV radiation incident on the sample (in mW/cm2).

Results

Change in Reaction Enthalpy and Time under Different Process Conditions

Tables and 2, respectively, detail total reaction enthalpies and peak reaction times obtained during the photocuring of AESO under different processing conditions (varying PI concentration, UV light intensity, and temperatures), for both the low-intensity (50, 100, and 150 mW/cm2) (Supporting Information, Figure S1) and medium-intensity (1500, 2500, and 3500 mW/cm2) (Supporting Information, Figure S2) batches. As can be seen, an increase in intensity from the low-intensity to the medium-intensity regime led to an increase in reaction enthalpy along with a significant decrease (>50%) in reaction time. This observation was consistent, irrespective of the type of PI, UV intensity, and/or temperature used. Further, the reaction time for the HCPK-initiated reaction was much higher than for DMPA-initiated reaction at lower PI concentration and temperature in both low- and medium-intensity regimes.

Table 1

Enthalpy of Reaction and Peak Time for Photocuring of AESO at Different Photoinitiator Concentration, Intensity, and Temperature Obtained from Photo-DSC for the Low-Intensity Regime

			enthalpy (J/g)		peak time (s)
intensity (mW/cm2)	temp (°C)	concentration of PI (wt %)	DMPA	HCPK	DMPA	HCPK
50	25	0.5	–89.85	–59.8	25	67
		1	–113.2	–87.24	19	31
		2	–121.2	–116.5	13	13
		4	–127	–124.7	7	13
	50	2	–143	–124.3	7	13
	75		–155.7	–128.3	7	7
100	25		–134.4	–133.8	7	13
150			–139.2	–141	7	7

Table 2

Enthalpy of Reaction and Peak Time for Photocuring of AESO at Different Photoinitiator Concentration, Medium-Light Intensity, and Temperature Obtained from Photo-DSC

			enthalpy (J/g)		peak time (s)
intensity (mW/cm2)	temp (°C)	concentration of PI (wt %)	DMPA	HCPK	DMPA	HCPK
1500	25	0.5	–132.73	–123.8	3.3	5.4
		1	–143.36	–131.98	2.9	3.8
		2	–149.79	–130.83	2.3	3.3
		4	–144.80	–163.89	2.6	2.6
	50	2	–173.52	–178.46	2.4	2.6
	75		–187.28	–179.3	2.2	2.3
2500	25		–155.75	–180.32	2.6	2.7
3500			–153.35	–162.19	2.3	2.7

The extent of conversion (α), under all tested conditions, was calculated using eq . With regard to this equation, in the case of photocure kinetics, the general practice is to treat the enthalpy of a reaction obtained at the highest reaction temperature (tested for) as its total enthalpy (ΔHtotal).[20,30,34] This practice assumes that the cure reaction is complete at the highest reaction temperature, resulting in α = 1. Nonetheless, from Table , it is clear that the highest enthalpy of crosslinking reaction (−141 J/g) was obtained at the highest UV radiation intensity (150 mW/cm2) and lowest reaction temperature (25 °C), and not at the highest reaction temperature (75 °C), when HCPK was used as PI. Also, upon comparing Tables and 2, it is evident that for the same PI concentration and temperature, reaction enthalpy increased with an increase in light intensity (to 1500/2500/3500 mW/cm2). This indicates that any further increase in reaction temperature and/or UV intensity, beyond the tested conditions in this study, may lead to a further increase in enthalpy beyond values obtained in this work. Hence, the aforementioned practice in photocure kinetics studies[20,30,34] of choosing the highest enthalpy obtained as ΔHtotal of the reaction can lead to unrealistic conclusions on cure kinetics. Therefore, to obtain a realistic understanding of the extent of conversion (α), it is critical to calculate ΔHtotal using a theoretical method.

Theoretical Heat of Reaction

While the theoretical heat of reaction has been previously calculated in the literature for simple molecules, such as methyl acrylate and diepoxies,[39] there still exists a lack of clarity on determining the theoretical heat of reaction for complex molecules such as triglycerides. Typically, eq can be used to calculate the theoretical heat of reaction (ΔHtheoretical) for a complex molecule (such as triglyceride), where f is the number of reactive sites per mole of the monomer, C is the fraction of the monomer used in final chemical composition, H is the energy (in Joules) per mole of the reactive site, and MW is the molecular weight of the monomer (in grams/mole).While, in the literature,[40] it has been mentioned that there can be a maximum of 4.2 acrylate groups in the acrylated epoxidized soybean oil (AESO), it is difficult to attain complete acrylation of epoxidized soybean oil resulting in reduction in the number of acrylate groups. Hence, to determine the extent of acrylation in the AESO, 1H NMR spectroscopy was carried out. From Figure S3 (Supporting Information), the functionality of AESO used in this study was determined as 2.5. Based on this, the average molecular weight (MW) was calculated as ∼1120 g/mol. Enthalpy of the reaction (ΔH) was considered to be −86.2 kJ per acrylate double bond,[39,41] while the value of C (or fraction of monomer) was assumed to be 1, as no solvents or comonomers were used in this work. Based on these details and eq , the theoretical heat of reaction (ΔHtheoretical) for 100% conversion of the double bond in the AESO molecule was calculated to be −192.41 J/g. Using this value as ΔHtotal, the degree/extent of conversion (α) was calculated using eq for all test conditions and has been plotted for the low- and medium-intensity batches in Figures a–f and 2a–f, respectively.

Figure 1

Extent of conversion (α) as a function of time during photocuring of AESO at varying photoinitiator concentration, intensity, and temperature for two different photoinitiators in the low-intensity regime.

Figure 2

Extent of conversion (α) as a function of time during photocuring of AESO at varying photoinitiator concentration, intensity, and temperature for two different photoinitiators in the medium-intensity regime.

Effect of Photoinitiator Type and Concentration on the Extent of Cure

From Figures a–f and 2a–f, it is evident that the extent of crosslinking did not reach unity (α ≠ 1) for any tested conditions, while the maximum curing of 98% was obtained when DMPA was used as PI and cured at 75 °C. Also, upon comparing both these figures, it is clear that the curing reaction reached completion within 40 s for the medium-intensity regime, while it continued for the entire 120 s duration in the low-intensity regime. As can be seen from Figure (for the low-intensity regime), the extent of cure was observed to increase with the increase in PI concentration (be it DMPA or HCPK), UV intensity, and/or temperature. However, in the case of the medium-intensity regime (Figures a,b), while the increase in DMPA concentration did not alter the extent of curing (α), a significant increase in α was observed with the increase in HCPK concentration. On the other hand, Figure c,f shows that α increased with an increase in isothermal temperature, irrespective of the PI used. Further, the rate of reaction was also observed to differ in both intensity regimes, being much slower in the low-intensity regime compared to the medium-intensity regime (Figures and 2). Additionally, for both intensity regimes, DMPA-containing samples showed a higher reaction rate (i.e., faster reaction) than HCPK-containing samples. In contrast, for both PIs, an increase in intensity beyond 1500 mW/cm2 did not significantly influence either the curing rate (dα/dt) or the extent of curing (α).

Kinetic Analysis

Initially, experimentally obtained dα/dt vs α curves were fitted using the Kamal–Sourour model (eq ) to understand cure kinetics. The error between the model-predicted and experimentally obtained dα/dt values (as a function of α) was minimized using the objective function (value of RSS) shown in eq . Figure shows experimentally obtained and model-predicted curves for dα/dt as a function of α. A poor fit was observed between the two sets of values, indicating that the Kamal–Sourour model failed to predict the experimental observations in a realistic manner. This is mainly due to the assumption made by this model that α reaches unity (i.e., complete crosslinking occurs),[33,42] while Figures and 2 clearly show that crosslinking of AESO was not complete. Hence, to account for incomplete cure that occurs under isothermal conditions, the modified Kamal’s model (eq ) was used to fit experimentally obtained dα/dt values (as a function of α). The objective function (eq ) was used to minimize the error between model-predicted and experimentally observed values and accurately determine both reaction rate constants (k1, k2) and reaction orders (m, n). Figure a,b shows model-predicted and experimentally obtained dα/dt values (as a function of α) for both PIs (DMPA and HCPK) at varying isothermal temperature conditions. As can be seen, the modified Kamal–Sourour model exhibited good fit with experimental values, indicating its suitability in explaining the experimental observations of photocuring of AESO. Based upon this fitting, the values of rate constants (k1, k2) and reaction orders (m, n) were obtained and have been reported in Table .

Figure 3

Experimental and model-fitted (Kamal–Sourour model) values of dα/dt as a function of α for AESO containing 2 wt % DMPA photocured at 25 °C and UV intensity of 1500 mW/cm2.

Figure 4

Experimental data for dα/dt as a function of α at 25, 50, and 75 °C (1500 mW/cm2), fitted with the modified Kamal’s model, for two photoinitiators: (a) DMPA and (b) HCPK.

Table 3

Enthalpy of Reaction and Peak time for Photocuring of AESO at Different Photoinitiator Concentration, Light Intensity, and Temperature Obtained from Photo-DSC

sample	temp (°C)	k1 (s–1)	k2 (s–1)	m	n	m + n	αmax	sum of squares	activation energy (kJ/mol)	R2
DMPA	25	0	0.659	0.52	1.32	1.859	0.78	0.00082	0.76	0.84
	50	0	0.654	0.59	1.51	2.104	0.91	0.00187
	75	0	0.630	0.58	1.55	2.144	0.98	0.00205
HCPK	25	0	0.47	0.49	1.35	1.847	0.68	0.00053	5.54	1
	50	0	0.56	0.55	1.47	2.032	0.93	0.00110
	75	0	0.65	0.58	1.54	2.128	0.93	0.00137

As can be seen from Table , at all temperature conditions (25, 50, and 75 °C), the value of k1 is obtained as zero, indicating that the reaction has no nth-order component and is solely an autocatalytic reaction.[36] Also, m + n ≅ 2, indicating that the cure reaction is a second-order reaction. Thus, from the determined rate constant k2, the activation energy (Ea) of the autocatalyzed crosslinking reaction was calculated via the Arrhenius equation (eq ), and the respective values have been provided in Table .

To understand the variation in activation energy (Ea,α) with the progression of cure (i.e., the extent of conversion/crosslinking or α), along with cure kinetics in a holistic manner, a model-free isoconversional method was also employed in this study. Initially, plots between ln tα, and 1000/T were obtained, and a linear fit was attained for all experimental values at varying α values, as shown in Figure for DMPA as PI. The slope of this linear fit provided the activation energy values (Ea,α) at different α, which were then subsequently plotted (Figure ) to demonstrate the variation in activation energy (Ea,α) as a function of α. As shown, the activation energy of the curing reaction increased with the progression of the reaction, indicating that the reaction is very complex.

Figure 5

Isoconversional plots of the extent of conversion in the range of 0.05–0.75 for AESO containing 2 wt % DMPA as the photoinitiator.

Figure 6

Variation in activation energy with the extent of conversion for AESO samples photocured using two different photoinitiators (DMPA and HCPK).

Light Intensity Exponent: Termination Mechanism

The termination mechanism of the curing reaction was also analyzed by using the light intensity exponent method (eq ). In general, for this method, the value of x is assumed to be 1 during modeling to determine the value of light intensity exponent (β),[1,43] while keeping in mind the aforementioned RSS principle (eq ), i.e., obtaining the least sum of squared errors. Based on this assumption (x = 1), β was obtained in the range of 0.1–0.3 for all reaction conditions, i.e., β < 0.5. This indicates that termination of curing occurred via the combination of two mechanisms: (a) primary radical termination or the reaction between free radicals derived from the AESO molecule and PI radicals, which prevent crosslinking between two AESO-based radicals, and (b) second-order termination or the reaction between two AESO-based radicals, which actually stops further crosslinking from taking place.

Discussion

Effect of Vitrification on the Extent of Cure

Multiple studies have reported incomplete curing (i.e., α ≠ 1) during isothermal cure conditions, irrespective of the material used and/or the cure mechanism that occurs.[20,31,35,44] Such behavior is commonly known as “vitrification”, described as the transformation of a polymer from its liquid/rubbery state to its glassy state due to the crosslinking of polymeric chains. This is typically accompanied by an increase in its viscosity as well as reduced mobility of both PI and AESO species, which leads to a drastic decrease in the subsequent rate of reaction.[20,31,35,44] As a result, the reaction ceases prior to cure completion (i.e., α does not reach 1), while the remaining functional groups are left behind as unreacted groups.[45]

It is widely known and understood that for any photocuring reaction, an increase in any one of the three processing parameters, light (UV) intensity, photoinitiator (PI) concentration, or temperature, will increase the rate and extent of curing (i.e., dα/dt and α, respectively) due to the increase in the number of free radicals available for reaction propagation.[9,30] However, for the medium-intensity batch in this study, DMPA concentration was observed to not cause any change in either the cure rate or the extent of curing (Figure a). Similar observations have also been reported earlier by Mucci and Vallo[46] in their work on analyzing the photopolymerization of methacrylate monomers using DMPA as PI. They have attributed this behavior to the screening effect because of the increase in UV absorbance of samples that contained ≥ 0.25 wt % of DMPA. In other words, at higher UV intensities (comparable to those employed in the medium-intensity batch in this study), the optimum PI concentration for obtaining the maximum extent of curing is 0.25 wt % for DMPA.[46] Beyond this value, any further increase in PI concentration will accelerate the termination process and will not contribute toward reaction propagation (i.e., curing). Since this study employs DMPA at higher concentrations (0.5, 1, 2, and 4 wt %) than this limit, variation in DMPA amount was observed to have a negligible effect on crosslinking-related parameters for the medium-intensity batch.

In contrast with this noninfluence of DMPA concentration, the temperature was seen to play a determining role for both the curing rate and the extent of cure for the medium-intensity batch (Figure c). This is explained by the fact that an increase in temperature enhanced the rate of reaction by increasing the mobility of both PI and AESO species that hitherto remained unreacted at lower temperatures. This, in turn, improved the ability of PI molecules to cause photocuring of AESO. Interestingly, for the low-intensity batch, the curing reaction was observed to occur beyond 120 s (Figure ), thus indicating that AESO molecules had undergone dark polymerization despite their slower rate of curing.[47] This occurrence of dark polymerization, albeit at retarded rates, is at odds with the conventional thought that considers free-radical polymerization to stop upon switching off the UV light. However, such dark cure in free-radical photopolymerization has also been reported lately in difunctional methacrylate.[47] This dark cure was hypothesized to occur due to the retained activity of free radicals that initially got trapped inside the crosslinked polymeric network and subsequently cured any molecule that was available and freely accessible for crosslinking in their surroundings, a phenomenon commonly known as the “cage effect”.[48]

A stark difference was also observed in the reactivity of the two PIs (DMPA and HCPK) used in this study, as corroborated by lower peak times (Tables and 2) and higher/equivalent extent of curing (α, Figures and 2) for DMPA-containing AESO samples over their HCPK-containing counterparts. This can be ascribed to the faster cleavage of DMPA that occurs within 100–200 ps,[49] as measured via electron paramagnetic resonance spectroscopy that has a picosecond resolution.[50] Further, the decomposition rate constant (kd) of DMPA has been previously estimated as 1011 s–1 by Kurdikar and Peppas.[51] Such a high rate constant indicates that decomposition of DMPA is a very fast process. Conversely, HCPK needs more time for cleavage and subsequent reaction with triglyceride molecules,[32] which explains its slower reactivity vis-à-vis DMPA in this study.

Finally, despite observing the vitrification phenomenon under all photo-DSC conditions, the acrylated triglyceride system employed in this study (i.e., AESO) exhibited the highest rate constant till date among all acrylates that are commonly used in photocure coatings.[9,15] This means that AESO underwent faster curing (i.e., less time) than other existing acrylates. Yet, at the same time, the AESO sample did not show complete curing under any condition employed in this work, which can be explained by the sole major limiting factor with such systems, their higher functionality (f = 2.5). This is in line with the existing literature,[52] which shows that the functionality of an acrylate (f) is inversely proportional to its extent of conversion (or curing, α). Such a relationship is the logical outcome of two key aspects: an increase in total enthalpy (ΔHtotal) of the acrylate due to its higher functionality (eq ) and the occurrence of the vitrification phenomenon in the acrylate system upon its curing.

Activation Energy Dependence on Conversion

Based on the model-free isoconversional method employed in this work (Figure ), the activation energy (Ea,α) of the cure reaction was initially observed to increase linearly with α but subsequently showed a drastic increase during the later stages of the reaction. This increase was observed irrespective of the PI used (DMPA or HCPK) and can be entirely attributed to the occurrence of vitrification in AESO during its curing. An additional complementing factor is the trapping of primary radical into the molecular network of AESO, which inhibits its availability for further curing, thereby stopping the cure reaction from taking place further. Interestingly, activation energy was observed to be higher for HCPK-initiated samples than for DMPA-initiated samples, when determined from both model-fitting and model-free techniques. This can be ascribed to the fact that since the photolysis product of HCPK is bulkier than that of DMPA, HCPK molecules may experience severe steric hindrance upon migrating to acrylate groups for undergoing crosslinking reaction. Hence, HCPK-containing samples find it difficult to undergo crosslinking at higher α values vis-à-vis their DMPA-containing counterparts, which explains the difference in their respective activation energies at a higher extent of conversion (α).

This variation in activation energy with an increase in α also highlights the high complexity of the crosslinking reaction, as reported elsewhere,[24] from the point of view of cure kinetics, for it means that a single rate equation cannot be used to explain/describe the cure kinetics of AESO. This is because the vitrification of a polymer is accompanied by a shift in the reaction mechanism (from chemical- to diffusion-controlled) on account of change in its activation energy.[24] Hence, the combination of vitrification, primary radical trapping, and the subsequent transformation in the nature of the curing reaction leads to significant differences in activation energy (Ea,α) values obtained at different degrees of conversion (α) vis-à-vis the activation energy (Ea) value obtained using the modified Kamal’s model (reported in Table ). Further, it also establishes that the model-free isoconversional method is more accurate and realistic in predicting the cure reaction kinetics of AESO when compared to the model-fitting method, as it can better capture the complexity of the entire curing process. Thus, this study establishes the relatively higher suitability of isoconversional methods over model-fitting methods for analyzing the cure phenomenon of triglyceride-based bioacrylate.

Conclusions

Photocure kinetics of AESO, a biobased acrylated triglyceride, was studied via photo-DSC by using two different photoinitiators (DMPA and HCPK). A method to determine the theoretical heat of reaction for complex molecules is proposed. Irrespective of the photoinitiator used, AESO exhibited autocatalytic behavior with the reaction order obtained as ∼2. While modified Kamal’s model (or Kamal’s model that accounts for vitrification) was observed to well-fit experimentally obtained values for the extent of curing (α), the model-free isoconversional method was found to accurately predict the photocure kinetics of AESO. Variation in activation energy at varying degrees of conversion was observed via the use of the model-free isoconversional method. Finally, the light intensity exponent method indicated that termination of the curing reaction occurred via the combination of primary radical termination and second-order termination mechanisms. The outcome of this work will enable the selection of optimal cure conditions for triglyceride-based bioacrylates for various applications. In particular, the knowledge provided in this work on UV curability of AESO in the presence of a nonyellowing HCPK initiator will enable its use in diverse applications, such as liquid optically clear adhesives and screen printing.

Experimental Section

Materials

Acrylated epoxidized soybean oil (AESO), containing 4000 ppm of monomethyl ether hydroquinone as an inhibitor (Viscosity: 18000–32000 cps and an acid value of < 10 mg KOH/g), was used as a photocurable bioacrylate resin. Two photoinitiators, namely, 2,2-dimethoxy phenylacetophenone (DMPA) and 1-hydroxycyclohexyl phenyl ketone (HCPK), were used. All of the chemicals were purchased from Sigma-Aldrich, MI, USA and used in the as-received state. An appropriate quantity of the photoinitiators was added to the AESO and hand-mixed. Both the photoinitiators were observed to dissolve in the resin completely.

Photocalorimetry

NETZSCH Photo-DSC 204 F1 Differential Scanning Calorimeter (DSC), equipped with a UV lamp (OmniCure S2000) through a single light guide, was used to monitor the photocure kinetics of AESO under a nitrogen atmosphere at a flow rate of 40 ml/min. Approximately, 2–5 mg of samples was taken in an open aluminum pan for each DSC test/experiment. Both the reference and sample pans were exposed to UV radiation in the wavelength region of 320–500 nm for 120 s with a delay of 5 s. The influence of PI concentration (0.5, 1, 2, and 4 wt%), temperature (25, 50 and 75 °C), and light intensity on photocure kinetics of AESO was studied. Light intensity was monitored at the end of the light guide and was varied in two batches: a low-intensity batch/regime of 50, 100, and 150 mW/cm2 and a medium-intensity batch/regime of 1500, 2500, and 3500 mW/cm2. In both the cases, the intensity experienced by the sample was ∼ 50 times[53] lower. To monitor the occurrence of any chemical reaction solely due to thermal energy, samples were held in isothermal conditions (at their respective temperature profile) for 5 min prior to exposing them to UV radiation for 120 s. No heat flow was observed during the initial 5 min of isothermal condition, indicating that the thermal energy provided did not induce any curing reaction. Further, for each experiment, samples were subjected to a second UV irradiation cycle to monitor the occurrence of any residual reaction. The absence of peak(s) during this second irradiation cycle (under all conditions) indicates the completion of the reaction during the first irradiation cycle.

Nuclear Magnetic Resonance (NMR) Spectroscopy

1H NMR spectroscopy was carried out on AESO using 300 MHz Bruker Avance (Billerica, MA), at 5341 MHz spectral width and 3.0 s acquisition time. A concentration of 50 mg/ml of AESO to CDCl3 was utilized, and 16 scans were collected.