polymers

Review
A Review on Modeling Cure Kinetics and Mechanisms
of Photopolymerization
Margit Lang 1, *, Stefan Hirner 2 , Frank Wiesbrock 2 and Peter Fuchs 1

1 Polymer Competence Center Leoben, 8700 Leoben, Austria; [email protected]

2 Institute for Chemistry and Technology of Materials, University of Technology Graz, NAWI Graz,
8010 Graz, Austria; [email protected] (S.H.); [email protected] (F.W.)
* Correspondence: [email protected]; Tel.: +43-384-242-962-753

Abstract: Photopolymerizations, in which the initiation of a chemical-physical reaction occurs by the

exposure of photosensitive monomers to a high-intensity light source, have become a well-accepted
technology for manufacturing polymers. Providing significant advantages over thermal-initiated
polymerizations, including fast and controllable reaction rates, as well as spatial and temporal control
over the formation of material, this technology has found a large variety of industrial applications.
The reaction mechanisms and kinetics are quite complex as the system moves quickly from a liquid
monomer mixture to a solid polymer. Therefore, the study of curing kinetics is of utmost importance
for industrial applications, providing both the understanding of the process development and
the improvement of the quality of parts manufactured via photopolymerization. Consequently,
this review aims at presenting the materials and curing chemistry of such ultrafast crosslinking
polymerization reactions as well as the research efforts on theoretical models to reproduce cure
kinetics and mechanisms for free-radical and cationic photopolymerizations including diffusion-
controlled phenomena and oxygen inhibition reactions in free-radical systems.

Keywords: free-radical photopolymerization; cationic photopolymerization; mechanistic model;

Citation: Lang, M.; Hirner, S.; phenomenological model; reaction kinetics
Wiesbrock, F.; Fuchs, P. A Review on
Modeling Cure Kinetics and
Mechanisms of Photopolymerization.
Polymers 2022, 14, 2074. https:// 1. Introduction
doi.org/10.3390/polym14102074
Photopolymerization is a method for manufacturing highly-crosslinked polymer
Academic Editor: Katarzyna networks, in which the initiation of a chemical-physical reaction occurs by exposing pho-
Szcześniak tosensitive, monofunctional or multifunctional monomers to a high-intensity, generally
Received: 27 March 2022
ultraviolet (UV), light source. According to Decker [1], photopolymerization is one of the
Accepted: 17 May 2022
most efficient methods for achieving quasi-instantaneous polymerization. Its huge potential
Published: 19 May 2022
in the simple and fast production of materials with special properties leads to a wide range
of potential applications [2]. Practical applications include, for instance, coatings [1,3],
Publisher’s Note: MDPI stays neutral
tissue engineering [4,5], photolithography [6–9], microfluidic device fabrication [10,11], 3D
with regard to jurisdictional claims in
prototyping [12–15], and 4D bioprinting [16,17]. Significant characteristics of photopoly-
published maps and institutional affil-
merization include the solvent-free formulation, ability to cure at ambient temperature
iations.
conditions (which is especially important for heat-sensitive materials), and low energy con-
sumption, as well as spatial and temporal control over the polymerization [1]. Furthermore,
the temperature rise resulting from the exothermic nature of the reaction can be controlled
Copyright: © 2022 by the authors.
by changing the irradiation intensity and wavelength [18]. However, there are also a num-
Licensee MDPI, Basel, Switzerland. ber of problems and complexities associated with photopolymerization including volume
This article is an open access article shrinkage and stress, the presence of unreacted, potentially extractable polymer, and the
distributed under the terms and susceptibility to oxygen inhibition in the case of free-radical photopolymerizations [19,20].
conditions of the Creative Commons One decisive factor in photopolymerizations is the curing degree achieved, as it
Attribution (CC BY) license (https:// significantly affects the mechanical properties of the polymers. The curing degree α is zero
creativecommons.org/licenses/by/ at the beginning of the chain-growth or crosslinking reaction. It increases with time in the
4.0/). course of the photopolymerization process due to the growth of the polymer chains.

Polymers 2022, 14, 2074. https://doi.org/10.3390/polym14102074 https://www.mdpi.com/journal/polymers

Polymers 2022, 14, 2074 2 of 58

Photocurable resins can be divided into two major classes, differing basically by
their polymerization mechanism: photoinitiated free-radical polymerizations (such as
the polymerizations of acrylates) and photoinitiated cationic polymerizations (such as
the polymerizations of epoxides, lactones and vinyl ethers, which are inactive towards
radicals) [21]. The latter have the distinct advantage that they lack sensitivity towards
atmospheric oxygen whereas the loss of radicals to oxygen, known as oxygen inhibition, is
a problem that is pervasive in free-radical photopolymerization [22–24]. Figure 1 shows the
conversion versus the exposure time for a cycloaliphatic diepoxy compound recorded in
the presence of air, in a N2 -saturated atmosphere and in the presence of air after covering
the monomer with a transparent polyethylene film (laminate). The kinetic curves clearly
show the lack of sensitivity towards oxygen for cationic photopolymerizations, as the
epoxy monomer polymerized at essentially the same rate independent of the experimental
conditions [25]. Operating in a N2 -saturated atmosphere and as a laminate, the polymer
even contains a larger amount of residual monomer, i.e., the extent of conversion is lower
compared to operating in the presence of O2 .

Figure 1. Conversion versus exposure time for the cationic photopolymerization of a cycloaliphatic
diepoxy compound in the presence of air, in a N2 -saturated atmosphere, and in the presence of air
after covering with a transparent polyethylene film (laminate). Reprinted with permission of John
Wiley and Sons from reference [25].

The lack of sensitivity towards oxygen further implies that, in contrast to free-radical
polymerization, the chain reactions continue to develop in the dark even in the presence of
air [25] and even in shadow regions (regions that had no illumination) [26]. This “dark reac-
tion” or “dark curing” offers the possibility of curing residual monomers in order to achieve
complete conversion in cationic polymerization [27]. According to Corcione et al. [28], the
“dark reaction” in free-radical polymerization resins is negligible. The “dark reaction”,
which takes place after the termination of UV exposure, is due to the fact that, unlike radicals
in free-radical photopolymerization, two cations cannot interact to undergo combination or
disproportionation and, hence, such types of termination are generally suppressed. The
living polymer chain will continue to grow in the dark until termination occurs by transfer
reaction or bimolecular interaction with another species present in the polymerization
mixture (e.g., water or bases added/present in the reaction mixture) [29].
According to Lin et al. [30], oxygen inhibition plays a critical role, especially for
optically thin polymers. However, various strategies to reduce oxygen inhibition in pho-
topolymerizations are proposed in the literature [31,32]: (1) using a higher light dose or
intensity, (2) using a higher photoinitiator concentration, (3) using co-initiators, (4) addition
of radical scavengers, (5) working in an inert environment, and (5) chemical mechanisms
such as the thiol-ene and thiol-acrylate-Michel systems which are insensitive to oxygen.
According to Kim et al. [29], cationic photopolymerization is becoming increasingly
important due to its applicability to rapid prototyping, i.e., the technique of stereolithog-
raphy. Although there is increased interest in cationic photopolymerizations, free-radical
photopolymerization is still the most popular and most widely used type [33]. According
to Decker et al. [25], one of the main limitations of photoinitiated cationic polymerization is
Polymers 2022, 14, 2074 3 of 58

the relatively low cure speed, compared to the very reactive acrylic systems polymerized
by free radical photoinitiated polymerization. Increasing the intensity of the UV radiation,
by using lasers, is a possible way to overcome this issue [34,35].
Besides free-radical and cationic photopolymerizations, so-called hybrid systems
have been reported as well, which implies mixing monomers that polymerize by different
mechanisms, i.e., in the presence of both, radical and cationic photoinitiators [36]. Using
such hybrid systems offers the possibility of producing interpenetrating polymer networks
within a few seconds of UV irradiation. Additionally, by a proper selection of the two
components, the final properties of the cured polymer can be controlled precisely [1].
Compared to thermal curing, photopolymerizations reach higher conversions as vol-
ume shrinkage occurs over a much longer timescale than the chemical reaction, i.e., a
temporary excess of free volume is generated. The heat instantly evolved by the exother-
mic reaction results in an increase in the sample temperature and therefore contributes
to a higher final degree of conversion [1]. Furthermore, one of the distinct advantages
of photoinduced polymerizations is the precise control over the initiation step with re-
spect to onset and end of the period of initiation as well as its magnitude (light inten-
sity). To evaluate the contribution of increasing light intensity, Decker et al. [1] recorded
temperature profiles of samples undergoing photopolymerization by real-time Fourier-
transformed infrared (RT-FTIR) spectroscopy. The temperature profiles along with the
conversion profiles for polyurethane-acrylates at two different UV radiation light intensities
(10 mW cm−2 , 80 mW cm−2 ) are shown in Figure 2. An increase in light intensity obviously
leads to a faster polymerization and a more extensive cure; the final product contains a
lower amount of unreacted functional groups. Decker and co-authors reported that a higher
initial light intensity I0 led to an increase in sample temperature which in turn provided
more molecular mobility and, consequently, led to higher ultimate conversion. As can be
seen in Figure 2, the temperature starts to rise as soon as the polymerization reaction begins
and reaches its maximum value once the polymerization reaction starts to slow down due
to gelation. The temperature decreases slowly as air cooling becomes predominant over
the exothermic polymerization reaction at that stage.

Figure 2. Temperature profile vs. exposure time (solid line) and conversion profile vs. exposure time
(dashed line) for different UV light intensities of photopolymerization of polyurethane-acrylates.
Reprinted with permission of John Wiley and Sons from reference [1].

The properties and structure of polymeric materials are governed to large extent by the
kinetics of their synthesis, in general highly non-equilibrium polymerization processes [37].
To determine the kinetics of photoinitiated polymerizations, two techniques are widely
used, namely RT-FTIR spectroscopy and photodifferential scanning calorimetry (photo-
DSC). Photo-DSC, which monitors the photopolymerization reaction’s heat flow rate over
time, is by far the most widely used technique in photocuring kinetic studies [20,33,38].
However, its main limitation lies in the relatively long response time, requiring operation
with low-intensity UV radiation. Due to its time resolution in the range of milliseconds,
Polymers 2022, 14, 2074 4 of 58

real-time FTIR is well suited to evaluate important kinetic parameters of ultrafast (intense
UV or laser irradiation) crosslinking polymerization reactions [1]. FTIR directly records
conversion versus time curves.
In order to study material property changes during the photopolymerization process,
it is necessary to model the kinetics of the photopolymerization’s chemical reactions and
describe the evolution of the curing solution’s species composition. Existing models to
analyze photopolymerizations can be broadly classified as energetic approaches, mech-
anistic approaches and phenomenological approaches [39]. Mechanistic kinetic models
offer a number of distinct advantages over phenomenological kinetic models. For instance,
mechanistic models offer the possibility to treat the effect of the type, concentration or
number of initiators on the overall curing rate separately, once the values of the various rate
constants (e.g., initiation, propagation, termination) have been determined. Subsequently,
there is no need to conduct curing experiments each time the type, concentration or number
of initiators is changed, as is the case for phenomenological models. Approaches to obtain
reasonably accurate and realistic kinetic expressions are discussed in detail in Section 3. Fur-
thermore, approaches regarding the implementation of kinetic models through numerical
simulations are shown in Section 4.
Due to the increasing interest in using nanocomposites to improve and tailor-fabricate
material properties, there is a demand for synthesizing nanocomposites with a uniform
distribution of nanoparticles in the polymer matrix. However, due to effects such as
high surface energy, unusual chemical activity, and immiscibility of the nanoparticles and
the monomer formulation/polymer matrix, nanoparticles tend to form large aggregates
(already) in the monomer solution, resulting in non-uniform materials with deteriorated
physicochemical characteristics [40]. Thus, one key point in the preparation of polymer
nanocomposites is the selection of a polymerization technique that ensures the fixation of
the initial uniform distribution of nanoparticles in the final nanocomposite, preventing the
agglomeration of nanoparticles during the polymerization process. Frontal polymerization
proved to be a positive technique, contributing to the uniform distribution of nanoparticles
in the resulting polymer composite [41,42].
The main shortcoming of photopolymerization lies in the limited penetration depth,
caused by decreasing light intensity along the height of the formulation to be cured, which
greatly limits the application potential of photopolymerization to thin films and adhesive
applications [27,43]. Layers with a typical thickness between 5 and 200 µm or, at the very
most, a few millimeters thickness can be polymerized [1,44]. Light absorption causes
the intensity of radiation to decay within the material according to the Beer–Lambert
law [45]. The incident light is mainly absorbed by the photoinitiator in the top layer
exposed to light, leading to a top-to-bottom gradient for the photogenerated initiating
active species and, consequently, leading to a sharp depth of the cure profile within the
sample undergoing polymerization. The presence of fibers as a reinforcing phase may
even contribute to this shortcoming by further reducing the transmission of light [45,46].
Therefore, photopolymerization can be effectively used for curing thin samples, but it is not
suitable for curing thick samples, especially those containing carbon fibers or other opaque
materials [43]. According to Decker [1], photoinitiation has proven to be well suited to
induce frontal polymerization, which is highly beneficial to curing thick specimens. The
basic idea behind frontal polymerization will be explained in Section 6.

2. Materials and Curing Chemistry

2.1. Radical Polymerizations
Free-radical polymerizations are initiated by free radicals. The polymerization begins
with the generation of primary radicals, which arise from the decomposition of the initiator,
and subsequent addition, for example to a carbon–carbon double bond of a monomer,
which yields the so-called initiation radicals [47]. The radicals are only released from
the monomers themselves in very rare cases. Usually, initiator molecules are added that
form the reacting species through thermochemical, electrochemical, and/or photochemical
Polymers 2022, 14, 2074 5 of 58

treatment. Basically, four different types of reactions can be associated with free radical poly-
merizations: generation of primary radicals from non-reactive species (initiation), radical
addition to a suitable monomer (propagation), atom-transfer and atom abstraction reactions
(chain-transfer reactions and disproportionation) and radical–radical recombination [48,49]
(Figure 3). Disproportionation and recombination of macro-radicals or initiator radicals rep-
resent two types of termination reactions. These termination reactions are responsible for a
low concentration of the active centers, on the order of about 10−8 mol L−1 [50]. Further
restrictions on the reaction kinetics are caused by side reactions of the active species with
the solvent, impurities, monomers, initiators and polymers, yielding radicals as well. These
different radicals can influence the reaction and form polymers with different constitutions,
configurations and molar mass distribution.

Figure 3. Mechanism of free-radical polymerizations, comprising the decomposition of the initiator,

the formation of the initiation radicals, and the propagation, chain-transfer and terminating reactions.

Polar, steric, stabilization, and thermodynamic effects have further influence on the
reaction and reactivity of the radicals or monomers [47]. In polar effects, for example, the
nucleophilicity of the radical and the electrophilicity of the monomer play an important
role. Due to the fact that head-to-head additions and head-to-tail additions are the most
common propagation reactions, steric effects of the monomers and radicals have great
influence. Stabilization effects occur if delocalization of the unpaired radical electron is
possible. The higher the delocalization of the electron, the lower the reactivity of the radical.
In principle, two types of monomers are able to perform free-radical polymerizations:
monomers bearing high-tension saturated rings or unstressed unsaturated rings, as well
as monomers bearing unsaturated bonds, i.e., double bonds [50]. Typical monomers for
free-radical polymerizations are alkenes, which are shown with other examples in the list
below (Figure 4).

Figure 4. Representation of the chemical structures of monomers for free-radical polymerizations

and of the corresponding repetition units.
Polymers 2022, 14, 2074 6 of 58

2.1.1. Thermally-Initiated Radical Polymerizations

In so-called real thermal polymerizations, the polymerization reactions of monomers
take place with the complete exclusion of external initiators such as atmospheric oxygen,
light sources and other impurities. Therefore, it is a spontaneous or self-initiated polymer-
ization without the targeted addition of initiators. A real thermal polymerization is likely
to occur in styrene, some of its derivatives, 2-vinylpyridine, 2-vinylfuran, acenaphthylene,
methyl acrylate and others [50].
In general, the thermal self-initiated polymerization is very rare and, therefore, thermal
homolytic dissociation of an initiator is the most common way to create reactive radical
species. The energy required to homolytically split a covalent bond can be brought in by
means of thermal, chemical and photochemical energy. In the case of thermal energy, a
so-called thermally initiated or thermally catalyzed polymerization reaction takes place.
Compounds used as thermal initiators commonly have bond dissociation energies in the
range of 100–170 kJ mol−1 [51]. Compounds that are not included in this dissociation
energy range either have too slow or too high dissociation speeds and are commonly
not practicable in polymerization reactions. Only a few compounds have bonds that are
homolytically cleavable in this energy range; these mainly have the O–O, S–S, or N–O
bonds [51]. The most commonly used initiators in thermal initiated radical polymeriza-
tion reactions are classified as organic peroxides including dicumyl peroxide, dibenzoyl
peroxide, cumyl hydroperoxide and methylethylketone peroxide (Figure 5). In addition
to the abundantly used organic peroxides, also so-called azo compounds are also of-
ten used as initiators. Examples of such azo compounds are 2,20 -azobisisobutyronitrile
(AIBN), 2,20 -azobis (2,4-dimethylpentanenitrile), 4,40 -azobis (4-cyanovaleric acid), and 1,10 -
azobis(cyclohexanecarbonitrile) (Figure 6). It should be noted that the driving force for
the homolytic dissociation of azo compounds is the formation of the very stable nitrogen
molecule due to the fact that the C–N bond dissociation energy is relatively high [51].

Figure 5. Chemical structures of most commonly used organic peroxide initiators.

Figure 6. Chemical structures of most commonly used azo initiators.

Polymers 2022, 14, 2074 7 of 58

2.1.2. Photoinitiated Radical Polymerizations

In recent years, photoinitiated polymerizations have become an important and efficient
polymerization type that has several advantages including the possibility of solvent-free
formulations, low-temperature conditions, and the spatial control of the light penetration
inside the photosensitive resin [52,53]. In principle, radical photoinitiators are compounds
that decompose into radicals during exposure to shortwave or visible light. By absorbing
the energy of the radiation, the π-electrons of the photoinitiator are raised to a higher
level. The π* excited state has a short lifespan only; it nonetheless offers the time range
for the molecule to decompose into radicals. These radical PIs can be distinguished into
two different systems, type I (decomposition) and type II (decomposition and subsequent
abstraction of protons). In type I, also related to as Norrish I cleavage, a light-induced
photolytic α-cleavage yields two radicals [54].
Common type I photoinitiators are benzoin, its derivatives, dialkoxy acetophenones,
aminoalkyl phenones, and bisacyl phosphinoxides (Figure 7). The main drawback of type I
photoinitiators is their irreversible consumption during photopolymerization [53]. In type
II photoinitiators, also referred to as H-abstraction type initiators, the photoinitiator is able
to abstract a hydrogen atom from adjacent molecules, forming two radicals.

Figure 7. Chemical structures of common type I photoinitiators.

Common type II photoinitiators are benzophenone, benzil, camphorquinone and

thioxantone in combination with proton donors (Figure 8). Photolytic cleavage of such
compounds in the presence of a hydrogen donor yields a ketyl radical and an additional
radical deduced from the hydrogen donor [55]. Due to the stabilization effect, ketyl radicals
are very stable radicals and normally do not react in the polymerization reaction [56].

Figure 8. Chemical structures of common type II photoinitiators (the proton donors are not repre-
sented here).
Polymers 2022, 14, 2074 8 of 58

2.2. Cationic Polymerizations

Cationic polymerizations are chain-growth polymerizations composed of addition
reactions of electrophilic and nucleophilic species. Initiators are the electrophilic part
E+ , whereas the monomer is the electron-donating part. These reactions are started by
the addition of the positively charged electrophile E+ to the electron-donating monomer,
resulting in a monomer cation EM+ , often a carbocation, which is an activated monomer that
performs the propagation reaction [50]. Polymerizable monomers can be distinguished into
two different classes: ethylenic monomers, in which the reactive species is the carbocation,
and heterocyclic monomers, in which the reactive species is the onium ion. The viable part
of the monomer must be the most nucleophilic part. As an example, acryl nitrile consists of
a vinyl group and a nitrile group; the nitrile group has the more pronounced nucleophilic
character. However, due to the resonance stabilization, no cationic polymerization is
possible [50].
In general, three different classes of initiators, BrØnsted acids, Lewis acids, and car-
bonium salts, are often used as initiators. The counter anions must not be nucleophilic,
as they would otherwise induce chain termination. However, since the monomer cations
are highly reactive, and therefore also unstable, an increased number of termination re-
actions and chain-transfer reactions occur. In addition, a solvent is required for almost
all cationic polymerizations. Therefore, cationic polymerization is used in only a few
industrial applications.

2.2.1. Cationic Polymerizations Initiated by Carbonium Ions, BrØnsted Acids, or

Lewis Acids
Three different initiation classes, BrØnsted acids, Lewis acids, and carbonium salts are
usually used in cationic polymerization reactions. BrØnsted acids and protonic acids are
able to start the polymerization reaction by protonation (of, e.g., the olefin). Acids with very
low pka values, which are associated with a low basicity of the counterions, are required
to protonate enough monomers to subsequently start the chain-growth reaction. Another
important requirement is the weak nucleophilicity of the counterions, which otherwise
would lead to chain termination. Commonly used BrØnsted acids for the initiation of a
cationic polymerization are perchloric acid, trifluoromethane sulfonic acid (triflic acid),
methane sulfonic acid and fluorosulfonic acid (Figure 9).

Figure 9. Chemical structures of common types of cationic initiators.

Lewis acids are another type of cationic initiator. Some of the Lewis acids, mostly
metal halides, are able to carry out self-ionization (2 AlCl3 ↔ [AlCl2 ]+ [AlCl4 ]− ), which
initiates the reaction. Examples of such metal halides are AlCl3 , TiCl4 , PF5 , SnCl4 , and
I2 [50]. However, the efficiency of self-ionization is extremely low, and the polymerization,
therefore, suffers from incomplete and slow initiation. To overcome this low efficiency, a so-
called co-catalysis initiation mechanism is preferred. A Lewis acid, representing an activator,
is mixed with a protonogen (H2 O) yielding a complex that initiates the polymerization
reaction. The polymerization rate is depending on the amount of the protonogen, due to
the possibility of chain termination by the protonogen [57].
Another way of initiation is the use of stable carbonium salts. An example of this
type of initiator is trityl chloride, which dissociates into a tripenhyl carbenium cation and a
chloride anion. The reactivity of the carbonium ion can be increased by complexing the
Polymers 2022, 14, 2074 9 of 58

counterion, e.g., the chloride anion can be trapped in a [SbCl6 ]− complex by adding SbCl5 ;
thus, the termination of the chain growth can be reduced [50].

2.2.2. Photoinitiated Cationic Polymerizations

The photoinitiated cationic polymerization is a widely used technique today and was
advanced in the 1970s by the discovery of photoacid generators (PAGs) by Crivello [58]. In
principle, photolysis of the photoinitiator is achieved by irradiation with UV light, due to
which superacids are formed that can initiate the polymerization [59]. Common cationic
photoinitiators are diaryliodonium salts, triarylsulfonium salts, diazonium salts and their
derivatives (Figure 10) [60,61].

Figure 10. Chemical structures of common types of photoinitiators.

As mentioned before, the counterion plays an important role in the activity and
stability of the formed carbonium cation and the polymerization efficiency. Large anions
such as SbF6 − , AsF6 − and PF6 − are very weak nucleophiles due to the charge distribution.
Therefore, superacids that are formed from PAGs are very stable initiators and have little
tendency to chain termination reactions. The main drawbacks are the low solubility of the
photoinitiators in non-polar monomers and the absorption band in the deep UV region,
which does not overlap with the emission band of visible light [62].

2.3. Radical and Cationic Polymerizations Highlighted in This Review

This review is focused on the research efforts on analytical models to accurately
describe the cure kinetics and mechanisms for curing behavior simulations of frontal
polymerization reactions. Hence, the types of polymerizations that have been subjected to
numerical analyses will be presented in this subchapter in brevity.

2.3.1. Radical Polymerization of Acrylates

Acrylates and methacrylates are the esters of acrylic acid and methacrylic acid, respec-
tively. Acrylates or methacrylates are a class of compounds that is most commonly poly-
merized using radical polymerization technology, constituting one of the most abundantly
produced classes of polymers. (Meth)acrylates are one of the most reactive monomers
in free-radical polymerization reactions. This high reactivity and diversity of chemical
structures provide a variety of mechanical, physical, chemical and optical properties; the
corresponding polymers and copolymers are used in a wide range of applications such
as coatings and adhesives [63]. The main difference between acrylates and methacry-
lates is the lower toxicity and the lower reactivity of methacrylates. Commonly used
monomers are methyl (meth)acrylate, ethyl (meth)acrylate, 2-ethylhexyl acrylate, hydrox-
yethyl methacrylate, and butyl (meth)acrylate, as well as acrylate-functionalized oligomers
such as polyurethanes, polyesters, polyethers, and polysiloxanes [64].
Polymers 2022, 14, 2074 10 of 58

In principle, the radical polymerization of acrylates starts with the decomposition of

the initiator yielding radicals. These radicals are able to attach to the vinyl group of the
(meth)acrylate leading to the initiation radical and starting the propagation (Figure 11).

Figure 11. Schematic representation of the radical polymerization reaction of acrylates. In represent-
ing the initiator and R* the radical.

2.3.2. Cationic Polymerization of Epoxides and Vinyl Ethers

Epoxy resins are one of the most commonly used thermosetting polymers and an
important matrix for polymer composites [65]. The epoxy ring is highly strained and
therefore exhibits high reactivity towards many substances [66]. The most important and
commercially available epoxy resins are the diglycidyl ethers of bisphenol A (DGEBA)
and bisphenol F (DGEBF). In Figure 12, the cationic polymerization of epoxide is shown.
The reaction starts with the decomposition of the photoinitiator, yielding the superacid
HX. The superacid protonates the epoxy ring, which subsequently ring-opens and reacts
with a non-protonated epoxy monomer. This reaction is also known as ring-opening
polymerization [67,68].

Figure 12. Reaction mechanism of the cationic ring-opening polymerization of epoxides.

Another important monomer class that polymerizes after cationic initiation is vinyl
ethers. The decisive factor for this behavior is the strong electron-donating alkoxy sub-
stituent, which also renders anionic or radical polymerizations impossible [69]. Common
vinyl ether monomers are ethyl vinyl ether (EVE), iso-butyl vinyl ether (IBVE), cyclohexyl
vinyl ether, and hydroxy butyl vinyl ether (HBVE). An overview of the reaction mecha-
nism is depicted in Figure 13. In principle, an acid is dissociated and protonates the vinyl
ether yielding a carbocation. This carbocation starts to propagate. Notably, cyclic addition
reactions, using a “cyclic initiator” are possible with vinyl ethers [70].
Polymers 2022, 14, 2074 11 of 58

Figure 13. Schematic representation of the cationic polymerization of vinyl ethers.

3. Models of Cure Kinetics

Curing behavior simulations require analytical models to accurately describe the
complex reaction mechanism and kinetics as the system moves quickly from a monomer
formulation to a polymer mixture. Models to describe the curing behavior can be broadly
classified as energetic or kinetic models [39].
Energetic cure models, as the name already implies, describe photopolymerizations
from an energetic point of view and are based on a direct relationship between the radiation
intensity, radiation profile and energy. These models assume that the curing process
starts when a critical value of energy, i.e., a material-dependent parameter, is reached.
Consequently, using energetic models, the photopolymer is assumed to solidify, i.e., the
resin is assumed to be cured, when, for instance, the irradiated energy into the resin reaches
or surpasses a certain threshold value [71]. In 2010, Tomeckova et al. [72] presented a model
for the photopolymerization of suspensions of ceramic powders in monomer solutions,
based on a critical energy dose in terms of the relative number of photo-generated radicals
and the concentration of inhibitors. The principal disadvantage of energetic models is that
they provide information on whether curing is achieved or not, but they do not provide
information about the curing degree, which, however, is crucial to correctly predict the
mechanical performance of the polymer [39].
In contrast, kinetic models are capable of distinguishing between different degrees of
cure, which in turn affect the mechanical properties of the polymer. Kinetic models, which
are usually mechanistic (non-empirical) or phenomenological (empirical or semi-empirical)
aim at predicting the evolution of the curing degree as a function of time.
Describing the photopolymerization from a mechanistic point of view implies that
the complete scheme of consecutive and competitive reactions which take place during
curing reaction (e.g., initiation, propagation, termination) is considered [73]. The curing
degree is evaluated by solving several differential equations. The set of ordinary and/or
partial differential equations is based on traditional mass action kinetics and can be used to
define the dynamic concentration, i.e., the evolution of one or more reactant variables in
the curing process as a function of time and space (depending on the model). To simplify
the complexity of the curing reaction, usually, several assumptions are introduced for
mechanistic models.
So-called phenomenological models are an alternative to mechanistic models. Phe-
nomenological models are formulated in terms of the curing degree and are much easier to
apply compared to mechanistic models [29]. The aim of phenomenological approaches is
to describe the curing process with one reaction. Although several simultaneous reactions
occur during the curing process a single simple empirical rate equation is used to describe
the curing kinetics. The rate equation has the following general form [29] (Equation (1)),

dα
= Kc ( T ) f ( α ) . (1)
dt
dα/
dt denotes the reaction rate, α denotes the curing degree, Kc denotes a chemical-
controlled rate constant as a function of temperature T, and f (α) is a function of the degree
of cure. Therefore, phenomenological models ignore the chemical details and fit the data to
Polymers 2022, 14, 2074 12 of 58

a mathematical functional form where the constants of the model are determined based on
experimental procedures.
Comparing the advantages and disadvantages of mechanical and phenomenological
models, Matias et al. [74] suggest that the mechanistic approach is not practicable for engi-
neering purposes, as the model equations typically require a large number of parameters
that must be determined from experimental data fitting or through numerical optimization
schemes in order to be solved. In contrast, phenomenological approaches usually require a
limited number of parameters and are therefore considered as being simple and suitable
for engineering applications. However, one major drawback associated with phenomeno-
logical models is the fact that these models capture the main features of reaction kinetics
but ignore how individual species react with each other. For instance, phenomenological
models are not capable of including the effect of the initiator (i.e., type, concentration or
number of initiators) on the rate of cure. Consequently, the kinetic parameter, the rate
constant Kc ( T ) shown in Equation (1), needs to be recalculated by conducting curing ex-
periments for each change of resin formulation [75]. In contrast, once the values of the
various constants (e.g., for initiation, propagation and termination) are determined for a
mechanistic model, the mechanistic model is capable of treating separately the effect of
type, concentration or number of initiators on the overall curing rate [76,77].
Another limitation of phenomenological models is their inability to predict post-curing
operations due to diffusion-controlled effects after vitrification [75]. Another alternative
is stochastic models which are based on kinetic Monte Carlo simulations, for instance,
proposed by Gillespie [78] or Ciftcioglu et al. [79]. These models determine the reaction
sequence based on the probability of each possible event and can be used to predict the
double bond conversion, molecular weight distribution and network connectivity [80].
The next sections briefly describe free-radical as well as cationic photopolymerization
and present a comprehensive review of the principal theoretical models developed for
predicting the kinetics of free-radical and cationic photopolymerization.

3.1. Free-Radical Photopolymerization

The traditional radical-initiated photopolymerization reaction can be divided into
the following steps: photodecomposition, photoinitiation, propagation and chain transfer
reactions and termination [81]. In a free-radical photopolymerization reaction, the following
typical reactants can be found: initiator molecules In, free radicals generated by the
photoinitiator R∗ , monomers M, polymer chains P, oxygen O2 and solvent. As light
interacts with the photoinitiator molecules, the initiator molecule is decomposed yielding
radicals. The asterisk * represents the active site of the radical species. Usually, one
photoinitiator molecule decomposes into two radicals (Equation (2)):

In → 2R∗ . (2)
kd

When a free radical reacts with a monomer, it transfers its active center to the monomer
and initiates a “macroradical”, to which monomers are added successively. An active site
does not vanish, until it is terminated; therefore, the polymer chain is also in an active state
(labeled by the asterisk *; Equation (3)):

R∗ + M → RM∗ . (3)
ki

The polymer chains propagate via reactions with other monomers or crosslinking with
other polymer chains (in the case of multifunctional acrylates; Equation (4)):

R − M∗ + n M → RMn∗+1 or P∗ + M → P∗ . (4)
kp kp

The general photopolymerization scheme is shown in Figure 14. At the initial state
(t = t0 ) the light source is off; the monomer is in a liquid phase (α(t = t0 ) = 0), and the
Polymers 2022, 14, 2074 13 of 58

photoinitiators are in an inactive stage. When the monomer resin is irradiated (t1 > t0 ),
the photoinitiators are decomposed yielding free radicals that are capable of initiating the
growth of a polymer chain. The polymer chain growth is quantified by the curing degree
(α1 (t1 > t0 ) > 0). Subsequently, the polymer chains propagate or crosslink with other
polymer chains, leading to an increase in the curing degree (α2 (t2 > t1 ) > α1 )).

Figure 14. General photopolymerization scheme.

Termination refers to the processes in which the reactive radical centers on polymer
molecules, as well as radicals, are terminated either by reacting with a free radical or
with a radical that is on a chain. Each termination reaction results in a “dead polymer
chain” Pdead or a “dead radical” Rdead , respectively. Termination occurs according to three
mechanisms: radical combination, radical disproportionation or radical trapping [82].
Radical combination refers to the combination of the ends of two growing polymer chains
(Equation (5)),
P∗ + P∗ → Pdead , (5)
k tc

or the combination of a growing end of a polymer chain with a free radical (Equation (6)):

P∗ + R∗ → Pdead (6)
k tc

Radical disproportionation refers to the removal of a hydrogen atom, forming an

unsaturated group and leading to the formation of two dead polymer chains (Equation (7)):

P∗ + P∗ → Pdead . (7)
k td

Radical combination and radical disproportionation are also referred to as bimolecular

termination. Termination by radical trapping, referred to as unimolecular termination,
occurs when active radicals become trapped between immobile polymer chains [83].
The termination rate coefficient k t in the Equations (5)–(7) does not only depend on
temperature and pressure, but on parameters such as polymer weight fraction, solvent vis-
cosity, polymer-monomer-solvent interactions, chain length of the macro-radicals involved
in the termination reaction, chain flexibility, chain entanglements and molecular weight
distribution as well [84].
In the presence of oxygen, propagation and termination reactions are inhibited as
oxygen in the reaction volume acts as a radical scavenger, i.e., it can react with radicals,
and reduces the quantum yields of the initiating radicals [80]. Oxygen reduces the effi-
ciency of initiation and generally leads to significant retardation or even inhibition of the
polymerization [33]. For instance, Decker et al. [22] stated that very little consumption of
the monomer occurred until most of the oxygen in the reaction volume was consumed
by reaction with radicals. Furthermore, it has been shown experimentally that the con-
centration of dissolved oxygen in the reaction volume must be lowered by at least two
orders of magnitude before polymerization begins, due to the high reactivity of oxygen
with radicals [82]. The inhibition mechanism can be described as follows (Equation (8)):

R∗ + O2 → Rdead or P∗ + O2 → Pdead . (8)

k t,oxy k t,oxy
Polymers 2022, 14, 2074 14 of 58

In order to avoid oxygen inhibition, Lovestead et al. [23] proposed to use a light source
with different wavelengths. A lower wavelength, which can only penetrate a few microns
deep into the sample, can be used to cure the top layer and limit/prevent any additional
oxygen diffusion into the sample. Subsequently, a higher intensity wavelength can be
used to cure the rest of the sample once the pre-dissolved oxygen is consumed. According
to Andrzejewska et al. [33], acrylates are generally more susceptible to oxygen inhibition
than methacrylates.
In the equations listed hereinabove, k d , k i , k p , k t , and k t,oxy are denoted as the respective
rate constants for decomposition, initiation, propagation, termination and termination by
oxygen inhibition. The propagation rate k p and the termination rate k t are not constant.
Both rates contribute to auto-acceleration and auto-deceleration, the two main regimes of
kinetic behavior during propagation [23,74]. In the course of the polymerization reaction,
the physical state of the medium changes from a viscous liquid to a viscoelastic rubber
(in some cases finally to glassy materials) causing drastic variations of the reactive species
mobility [1]. Both behaviors, auto-acceleration and auto-deceleration, are governed by
the mobility changes of radicals and unreacted double bonds as a result of the continuing
polymerization and crosslinking [19].
Auto-acceleration (gel effect or Trommsdorff–Norrish effect) denotes a reduction in the
termination kinetic constant k t and a significant increase in the polymerization rates due
to increasing viscosity. A few seconds after irradiation, this effect, in which the segmental
movement of radicals is restricted due to localized increases in the viscosity of the polymer-
izing system, can be observed [37]. Prior to that, chain termination by a combination of
two free-radical chains occurs at a high frequency. However, when the concentration of
“dead polymers” increases, i.e., the growing polymer molecules with active free-radical
ends are surrounded by an increasingly viscous medium, the reduction in mobility and
therefore hindered termination can be observed. The changes in viscosity affect the macro-
molecules but do not prevent smaller molecules, such as radicals and monomers, to move
freely. Consequently, as termination collisions are restricted, the concentration of active
polymerizing chains and the consumption of monomer rises rapidly, leading to a signifi-
cant polymerization rate. According to Batch et al. [85], bimolecular termination is even
more hindered in crosslinking polymerizations compared to linear polymerization. Con-
sequently, for crosslinking systems, diffusion-limited termination occurs at even lower
conversions compared to linear polymerization systems, and termination may be insignifi-
cant at this stage. Batch et al. [85] studied the influence of crosslinker concentrations on
the polymerization rates using a vinyl ester resin mixed with styrene cured isothermally in
DSC experiments at 60 ◦ C. The experiments indicate that increasing the concentrations of
crosslinkers increases both the initial slope of the polymerization rate Rr and its maximum
value Rr,max (Figure 15).

Figure 15. Polymerization rate Rr versus time for various mixtures of vinyl ester resin and styrene
with labels corresponding to the crosslinker concentration. Reprinted from reference [85] with
permission of John Wiley and Sons.
Polymers 2022, 14, 2074 15 of 58

When the reaction continues, the system becomes even more viscous and is governed
by a strong reduction in molecular mobility. The transition to a glassy state strongly affects
the polymerization kinetics, reducing the mobility of the monomers and radicals. At this
stage, the propagation reaction becomes diffusion-controlled leading to a decrease in the
rate constant for propagation k p , consequently leading to a decrease in the polymerization
rate R p . This decline in the polymerization rate is referred to as auto-deceleration or the
glass effect [19].
The photopolymerization profile, also referred to as the conversion-time curve, has a
characteristic S-shape and can be divided into four different regimes as for instance shown
for the polymerization of methyl methacrylate (MMA) by Achilias [84] (Figure 16). At the
very early stage after UV irradiation, the reactive species react with the inhibitors (e.g.,
oxygen in the case of free-radical photopolymerizations), leading to an induction period.
When all the inhibitors are consumed, the reactive species react with the monomers yielding
macroradicals that propagate. At this stage (Stage-I), the polymerization rate R p remains
almost constant. The crossover of Stage-I and Stage-II denotes the onset of the gel effect.
Therefore, Stage-II shows a sharp increase in the polymerization rate R p , followed by an
increase in the conversion X. The maximum polymerization rate occurs at the crossover
of Stage-II and Stage-III. Stage-III is characterized by a significant decrease in the reaction
rate R p (auto-deceleration). At very high conversions, R p tends asymptotically to zero. In
this regime of polymerization, the polymerization reaction slows down and finally stops
despite the continuing presence of both, radicals and monomer reactants.

Figure 16. Indicative time evolution of the free-radical photopolymerization reaction (polymeriza-
tion rate R p , conversion X and −ln(1 − X ) versus time) presenting the classification of the reaction
into four regimes from polymerization of methyl methacrylate (MMA) at 80 ◦ C with AIBN (azo-
bisisbutyronitrile) 0.03 mol L−1 . Reprinted from reference [84] with the permission of John Wiley
and Sons.

3.1.1. Mechanistic Models for Free-Radical Photopolymerizations

All mechanistic models have in common that they use first-order reaction equations to
describe the concentration variations of the individual species, including photoinitiators,
free radicals and monomers. However, they differ in how diffusion control is treated, i.e.,
by using propagation rate coefficients that decrease empirically with increasing conversion
(e.g., Wu et al. [82]), or by applying the free volume concept (e.g., Bowman et al. [19],
Anastasio et al. [86]).
In 2018, Wu et al. [82] proposed a spatial mechanistic kinetic model that employs
first-order chemical reaction differential equations to calculate the variation of the species
concentrations. Additionally, it includes the description of the spatial distribution of
reactants inside the continuum body during the process. The monomer used in their work is
poly(ethylene glycol) diacrylate (PEGDA) which is widely used in biomedical applications
and 3D printing. In the model, the generation of free radicals due to photoinitiator reactions
is considered according to Equation (9),
Polymers 2022, 14, 2074 16 of 58

∂C I (x, t)
= − βI (x, t)C I (x, t) , (9)
∂t
in which β denotes the decomposition rate.
The light intensity is the driving factor for the formation of free radicals. The Beer–
Lambert law describes the light propagation through a homogeneous medium without
internal sources or scattering and, therefore, the light intensity variation with changes in
the spatial position of the sample. Wu et al. [82] used the three-dimensional version of the
Beer–Lambert law, also referred to as the radiative-transfer equation [87] (Equation (10)):

Ω(x, t)·∇ I (x, t) = − A(x, t) I (x, t), (10)

in which Ω(x, t) represents the direction of light propagation, I (x, t) the light intensity
at position x and time t, and A(x, t) the local depletion rate of light intensity due to the
absorbance of the species.
According to Wu et al. [82], the variation of the light intensity depends on the con-
centration of the light-absorbing species and the respective molar absorptivity. Therefore,
the absorptivity is not simply that of the photoinitiators (as proposed for instance in a
publication from Anastasio et al. [86]), but instead is a combination of photoinitiator, free
radicals and polymer matrix which all can absorb photons and, hence, attenuate light as it
propagates through the material. Consequently, Wu et al. [82] calculated the local depletion
rate according to Equation (11),
h i
A(x, t) = εC I (x, t) + A absorber (x, t) + A polymer p(x, t) + Amonomer (x, t)(1 − α(x, t)) , (11)

in which ε denotes the molar absorptivity of the initiator, C I (x, t) the concentration of the
light-absorbing species, A absorber the absorption by photoabsorbers, Amonomer the absorption
by the unconverted monomer, A polymer the absorption by the converted polymer, and α(x, t)
the degree of cure.
Light refraction is not considered in the model, as for photocuring systems light is
usually irradiated in (or close to) the perpendicular direction [82]. Therefore, the light
intensity in the thickness direction can be calculated according to Equation (12):

∂I (z, t)
= − A(z, t)C I (z, t) I (z, t). (12)
∂z
The evolution of radical concentration CR is modeled according to Equation (13),

∂CR (x, t)
= mβI (x, t)C I (x, t) − 2k t (CR (x, t))2 − kO CR (x, t)CO (x, t) , (13)
∂t
with the termination rate k t , the concentration of radicals CR (x, t) (regardless of the chain
length), the reaction rate kO between oxygen and radicals, and the concentration of oxygen
CO (x, t). The parameter m denotes the number of radicals generated during photode-
composition, depending on the type of photoinitiator (e.g., two in the publication from
Anastasio et al. [86]). The second term of Equation (12) is a termination term that accounts
for the inactivation of radical species: if two active radicals react, they can recombine
yielding a dead polymer, reducing the concentration of active radicals. Wu et al. [82]
only considered a termination by radical combination but neglected chain length depen-
dence, effects of polymer heterogeneity, and radical trapping as, for instance, proposed by
Bowman et al. [19]. However, the authors stated that further termination mechanisms, e.g.,
monomolecular termination, could also be included in this model. The factor 2 that appears
in the second term of the equation accounts for two radical chains being “inactivated”
by the termination event. A reaction that further influences free-radical polymerizations
is the inhibition by oxygen [88]. Due to the high reactivity of oxygen towards radicals,
oxygen reacts very rapidly with the propagating radical, and the resulting peroxy radical is
very unreactive towards propagation. This by-reaction, hence, represents one of the most
Polymers 2022, 14, 2074 17 of 58

relevant limitations of free-radical polymerization [19]. Therefore, the third term in the
∂C (x,t)
above equation is added to describe the evolution of oxygen in the solution O∂t .
The monomers in the solution are gradually consumed by combining with radicals,
reducing the reactive functional groups. Therefore, the concentration of the unconverted
functional groups can be modeled as shown in Equation (14),

∂C M (x, t)
= − k p C M CR , (14)
∂t
in which k p denotes the propagation rate.
The curing degree does not explicitly appear in the system of differential equations,
since it can only be evaluated once the degree of monomer conversion has been solved.
The degree of cure can be calculated according to Equation (15),

C M (x, t)
α(x, t) = 1 − , (15)
C M (x, t = 0)

in which C M (x, t) is the concentration of the monomer molecules (unconverted double

bonds) at time t.
Although the volume shrinkage during the chemical reaction can increase the concen-
trations of different species, Wu et al. [82] observed that the effect of the volume concentra-
tions was largely canceled in the equations for C I (x, t), CR (x, t), CO (x, t) and C M (x, t) and
does not affect the curing degree noticeably. Consequently, the equations hereinabove use
the concentration in the initial configuration. The presented approach by Wu et al. [82] was
also taken up by Brighenti et al. [89] who developed a multiphysics modeling approach to
predict the mechanical properties of polymers obtained via photo-induced polymerization.
One challenge associated with the model presented by Wu and his co-authors is the
implementation of non-constant propagation and termination rates k p and k t , as both
parameters are a function of conversion (the molecular mobility in the reaction medium
decreases as the polymerization proceeds) [19,37,86]. Effects such as auto-acceleration
and auto-deceleration can only be accounted for when these coefficients are modeled as
functions of conversion. In particular, k t is highly dependent on conversion [19]. According
to Buback et al. [90], changes in the termination rate k t are caused by the following two
dominant mechanisms for termination that occur in parallel: termination due to species
translational diffusion k t,D and termination due to reaction-diffusion k t,RD . Diffusion-
controlled termination generally occurs when the mobility of the growing polymer chains
is hindered and can be summarized in Equation (16),

k t = k t,D + k t,RD , (16)

in which the species translational diffusion k t,D is subdivided into the center of mass
translational diffusion k t,TD and segmental diffusion k t,SD . These two mechanisms of
species translational diffusion occur consecutively, i.e., in order to react with each other,
radicals must be close enough to meet (referred to as center-of-mass translational diffusion
k t,TD ), and the reactive groups must be reoriented into proper position in order to react with
each other (referred to as segmental diffusion k t,SD ) (Figure 17). According to Wu et al. [82],
segmental diffusion k t,SD is often treated as a constant. The diffusion of the radical’s center
of mass, described by the center of mass translational diffusion k t,TD , depends on the
viscosity of the solution [91]. Increasing viscosity leads to a decrease in the center of mass
translational diffusion defined by Equation (17),

1
k t,TD = , (17)
exp(cα)

with the relative viscosity coefficient c and the curing degree α [82].
Polymers 2022, 14, 2074 18 of 58

Figure 17. Scheme of bimolecular termination reactions between two macroradicals (colored in red).
Two polymer coils must come into contact by center-of-mass translational diffusion (1), and segmental
reorientation (segmental diffusion) (2) has to occur in order to bring both reactive chains ends in
proximity and to form a radical-radical encounter pair (3).

The idea behind reaction-diffusion, described with the parameter k t,RD , is that the
radical site at the end of a growing chain does not only move as a result of diffusive motion
but also because chain growth occurs at this side, i.e., the radical end also moves when
the polymer chain grows due to the addition of monomer molecules at the chain end
(propagation) (Figure 18).

Figure 18. Scheme of reaction-diffusion where the movement of the growing radical site is attributed
to the addition of monomer molecules at the chain end (propagation).

According to Bowman et al. [19], reaction-diffusion termination is a unique mobil-

ity mechanism for radicals in crosslinked systems. As a result of the classical diffusion
mechanisms, the radical mobility drops to such an extent that their centers of mass are
essentially immobile, and the only possible diffusive motion of radical chain ends becomes
the propagation reaction [33]. The reaction-diffusion rate can be modeled to be proportional
to the rate coefficient for propagation k p and the monomer concentration. If any effect of
volume shrinkage is ignored, the monomer concentration is proportional to (1 − α) [90],
and the reaction-diffusion rate is defined according to Equation (18),

k t,RD = CRD k p (1 − α), (18)

in which CRD represents the reaction-diffusion proportion parameter and k p the propaga-
tion rate modeled according to the publications of Buback et al. [90,92], as well as Dickey
and Willson [93].
Figure 19 exemplarily shows the variation of the termination rate coefficient as
a function of conversion for the polymerization of methyl methacrylate published by
Achillias et al. [84]. As long as the increase in segmental diffusion is counterbalanced by
the decrease in translational diffusion, the termination rate coefficient k t remains con-
stant or decreases moderately only with increasing conversion. The initial conversion
range, in which the termination rate coefficient remains approximately constant, is con-
siderably dependent on the monomer type [84]. At the point at which the center-of-mass
(translational) diffusion becomes rate-determining, the termination rate constant decreases,
leading to an increase in the total macroradical concentration and the polymerization rate
(auto-acceleration). At higher conversions, the effect of auto-acceleration stops, and the
Polymers 2022, 14, 2074 19 of 58

termination-controlling mechanism changes to reaction-diffusion. The observed decrease

in k t is gradual only. At even higher conversions, diffusion-controlled propagation is
observed. Combining the equations hereinabove, the total termination rate used for the
kinetic model proposed by Wu et al. [82] is defined according to Equation (19),

1 CRD (1 − α)k p0
kt = exp(cα)
+ k p0
, (19)
1
k t,SD + k t,TD0 1+ k p,D0 exp(cα)

where k p0 denotes the polymerization rate at the beginning of the reaction (α = 0), k t,TD0
is the rate of mass translational diffusion at zero conversion, and k p,D0 corresponds to a
parameter used to characterize the diffusion-controlled propagation reaction.

Figure 19. Termination rate coefficient k t vs. conversion for methyl methacrylate at 0 ◦ C () and
50 ◦ C (). Reprinted from reference [84] with the permission of John Wiley and Sons.

The kinetic model by Wu et al. [82] was validated for PEGDA (Mn = 250 g mol−1 )
with 0.3 wt.% of 2,2-dimethoxy-2-phenylacetophenone as photoinitiator. The samples
were cured with a UV curing lamp with a 365 nm wavelength bandpass filter and a light
intensity of 5 mW cm−2 on the top surface of the solution. FTIR measurements were used
to obtain the parameters used for the reaction kinetics model. Figure 20 shows the degree of
conversion versus the reaction time for the experimental results and the kinetic model [82].
It reveals that the model results match the experimental conversion rate well and that the
model accurately captures the auto-acceleration effect where the degree of cure increases
rapidly after a conversion of about 12.0%.

Figure 20. Degree of conversion versus reaction time for free-radical photopolymerization of PEGDA.
Reprinted from reference [82] with permission of Elsevier.

Apart from modeling the photopolymerization kinetics in terms of the photopolymer-

ization reaction mechanism, Wu et al. [82] also modeled the material property evolution, i.e.,
Polymers 2022, 14, 2074 20 of 58

the evolution of the glass-transition temperature Tg as well as the volume shrinkage during
curing. The glass-transition temperature is an important indication of the curing extent, as
it increases with fractional conversion. In order to describe the relationship between Tg and
the curing degree, Wu et al. [82] used the model proposed by Gan et al. [94], which consid-
ers the crosslinking effects on the mobility of the curing system and, consequently, also the
significant changes above the glass-transition temperature at high curing degrees. Further-
more, it captures the evolution of Tg of a wide variety of curing systems [82]. Consequently,
the Tg change during the curing process is modeled according to Equation (20),

Er
Tg = h i, (20)
Rln g1 (1 − α)ξ + g2

in which Er represents the activation energy for the transition from the glassy to the rubbery
state, R is the gas constant, g1 and g2 are material constants, α is the curing degree, and ξ is
a parameter accounting for the effects of chain entanglement.
Another spatially dependent polymerization model, which is also used for a broad
range of industrial applications [88,95–97], was developed by Bowman et al. [19]. The model
of Bowman and co-authors further includes chain-length dependent termination (CLDT),
assuming that radicals diffuse and terminate according to their chain length [98]. According
to Lovestead et al. [98], the kinetic chain length is affected by the initiation rate, i.e., with an
increasing initiation rate, the kinetic chains become shorter. Shorter chains more readily
diffuse and terminate easier according to their length. Consequently, the termination kinetic
constant must incorporate all the different possible mechanisms that control termination:
translational diffusion, segmental diffusion, reaction-diffusion and chain-length dependent
termination. The model for incorporating chain-length dependent termination into the
termination kinetic equations, proposed by Bowman et al. [19], builds on models that
incorporate free volume theory and diffusion-controlled kinetics [38,95,96,99]. When the
fractional free volume v f of the system is greater than the critical free volume vc f , the poly-
merization is reaction limited. In the case that the fractional free volume v f of the system is
less than the critical free volume vc f , the polymerization is diffusion-controlled [95].
The termination kinetic constant incorporating chain-length dependent termination,
described as a function of conversion, can be summarized in Equation (21),
 " # −1  −1
k t,RD k p [ M ] − At ( v1 −v 1
 
 1 1 1 ) 
k t i,j = k1,1
t0 1 + + + γ e f f ,c f , (21)
k1,1
t0
2 iγ j 

in which k1,1
t0 is the termination kinetic constant between two radicals of length 1, [ M ] the
monomer concentration, γ an exponent that describes the relationship between mobility
and termination, At a constant that controls the onset and rate of auto-acceleration, v f the
fractional free volume, and v f ,c f the critical free volume for the regime in which termination
becomes controlled by the active species’ segmental motion, i.e., the termination transitions
to diffusion control.
The fractional free volume, considering only the case without excess free volume, is
assumed to be a function of conversion as proposed by Bowman et al. [99]. According to
Bowman et al. [19], the chain-length dependent termination kinetic constant k t i,j accounts
for radicals of length i terminating with radicals of length j and is able to predict a region
in which termination is dependent on the radical chain lengths. However, the model also
does not consider a limiting radical chain length to determine if the radical is incorporated
(“trapped”) in the gel and no longer capable of diffusion-limited termination [100].
A pointwise mechanistic model was proposed by Anastasio et al. [86] to describe
the monomer conversion by using the kinetics of the photopolymerization reaction for
a methacrylate resin. Again, the reaction scheme of free-radical photopolymerization,
Polymers 2022, 14, 2074 21 of 58

described earlier in this section, is presented as a set of differential equations, describing

the evolution of species concentrations over time.
However, as opposed to the model suggested by Wu et al. [82], this model does not in-
clude the description of the spatial distribution of reactants inside the continuum body dur-
ing the reaction process. In the set of differential Equations (22)–(26), [ In], [ R∗ ], [ M], [ P∗ ]
and [ Pdead ] correspond to the concentrations of initiator, free radicals, monomer, growing
polymer chains and dead polymer chains, respectively.

d[ In]
= −k d [ In] (22)
dt
d[ R∗ ]
= 2 f k d [ In] − k p [ M][ R∗ ] − k t [ P∗ ][ R∗ ] (23)
dt
d[ M]
= k p [ M][ R∗ ] − k p [ M][ P∗ ] (24)
dt
d[ P∗ ]
= k p [ M][ R∗ ] − k t [ P∗ ][ R∗ ] − 2k t [ P∗ ]2 (25)
dt
d[ Pdead ]
= k t [ P∗ ]2 + k t [ P∗ ][ R∗ ] (26)
dt
Again, like with the model proposed by Wu et al. [82], the curing degree does not
explicitly appear in the system of differential equations as it can only be evaluated after
the problem related to the monomer conversion has been solved using the relationship
summarized in Equation (27),

[ M(t)]
α(t) = 1 − , (27)
[ M(t = 0)]

in which [ M (t)] denotes the concentration of the monomer molecules at the time t. In
the equations hereinabove, f corresponds to the initiator efficiency, namely the fraction of
radicals that initiate the growth of a polymer chain, while k d , k p and k t correspond to the
reaction rate constants for decomposition, propagation and termination. In order to be able
to solve the set of equations, these parameters must be determined.
The initiator efficiency decreases as a function of conversion due to the “caging
effect” leading to the recombination of free radicals [84]. The caging effect depends on
the amount of the initiator radicals that are entrapped in the system during the curing
reaction. Entrapped initiator radicals are not likely to be available for participation in the
curing reaction. Han et al. [76] pointed out that the caging effect might be significant in the
curing reaction of unsaturated polyester resins due to the formation of a three-dimensional
network structure as opposed to the polymerization of methyl methacrylate or styrene that
yields linear (uncrosslinked) macromolecules.
In the model proposed by Anastasio et al. [86], the recombination (“trapping”) pro-
cess (R∗ + R∗ → 2Rdead ) is considered by reducing the initiator efficiency accordingly
kt
(Equation (28)):
1
f = h  i . (28)
1 + exp C v1 − 1
v f ,c f
f

At the beginning of the reaction, the initiator efficiency is assumed to be 1. C is an

adjustable parameter and represents the rate at which the initiator efficiency decreases with
increasing conversion. In their implementation of the model, Anastasio et al. [86] used a
value of C = 1 for the adjustable parameter to study the predictive capabilities with respect
to changing process conditions. However, Anseth and Bowman [38], for instance, use C
as a fitting parameter. v f is the fractional free volume of the system and is modeled as a
function of conversion as proposed by Bowman et al. [99]. v f ,c f is the critical free volume
at which propagation becomes diffusion controlled.
Polymers 2022, 14, 2074 22 of 58

According to Anastasio et al. [86], a critical aspect of the proposed model for poly-
merization kinetics concerns the determination of the initiator decomposition rate k d . The
authors determine the initiator decomposition rate k d using a modified Beer–Lambert
law for penetration of light into a medium, as suggested by Boddapati [80]. The initiator
decomposition rate depends on the concentration of the initiator [ In], the incident intensity
of the light source I0 and depth into the absorbing medium z according to Equation (29),

λ
k d = 2.3φεI0 · exp(−2.3ε[ In]z)· , (29)
NA hc

in which φ represents the quantum yield of the initiator, ε the molar absorptivity of the
initiator, λ the wavelength of light, h Planck’s constant, and c the speed of light.
Anastasio et al. [86] used termination reaction rates including the diffusion effects by
using a limited number of adjustable parameters as proposed by Anseth and Bowman [38].
The model of Anseth and Bowman includes reaction-diffusion, the transition from reaction-
controlled to diffusion-controlled reaction and volume relaxation and is in good agreement
with experimental results [101].
The model described hereinabove neglects the translational diffusion because, accord-
ing to Achillias [84], the translational diffusion of the polymer chains in a crosslinking
system is negligible (already) from the start of the reaction. Using this approach, the
termination rate constant is described by Equation (30),
  
1
k t = k t0 1 +   k  h  i  , (30)
R k p
p0
+ exp − A v1 − v 1
f f ,ct

with k t0 as the initial termination rate constant, R a constant, k p the propagation rate
constant, k p0 the initial propagation rate constant, v f the fractional free volume of the
system, and v f ,c f the critical free volume at which propagation becomes diffusion controlled.
Similar to C in Equation (8), A is an adjustable parameter and is used as a fitting parameter
in the model of Anseth and Bowman [38]. In the implementation of the model proposed by
Anastasio et al. [86], the parameter A is set to 1.
A similar model, where the main kinetic rate constants are defined as functions of the
fractional volume v f in order to consider their progressive diffusional control throughout
the photopolymerization reaction was proposed by Christmann et al. [102]. Christmann
states that most of the kinetic models to study the complex free radical photopolymeriza-
tion mechanism and its related effects do not consider simultaneously all the termination
pathways and/or neglect the evolution of terminations along the polymerization reaction.
However, according to Ibrahim et al. [103], the proportion of the termination mechanisms
that occur during free radical photopolymerization is expected to evolve during the poly-
merization because of the progressive increase in the medium viscosity which limits the
species motion as the tridimensional network evolves. The termination mechanisms, con-
sidered in the model of Christmann et al. [102], namely biomolecular termination, primary
radical termination, and monomolecular termination are shown in Figure 21. Bimolecular
termination can either occur by a combination (formation of a chemical bond between two
macroradicals) or disproportion (hydrogen abstraction from a macroradical to a second one
with the formation of a double bond on the former). As combination and disproportion
both involve a reaction between two macroradicals, they are lumped into a single termina-
tion mechanism [102]. Primary radical termination refers to the reaction between a primary
radical and a macroradical. At final conversion, the polymer network is vitrified, and it
can be assumed that all remaining macroradicals are trapped by occlusion in the polymer
network (monomolecular termination).
Polymers 2022, 14, 2074 23 of 58

Figure 21. Termination mechanisms considered in the kinetic model of Christmann et al. [102].

The above-mentioned termination reactions are modeled by a set of differential equa-

tions (Equations (31)–(33)):

d[bimol ] 2
= 2k t,b R(C = C )∗n .

(31)
dt
d[ R(C = C )n R A ] R∗A
R(C = C )∗n [ R∗A ].
 
= 2k t,PRT (32)
dt
d[ R(C = C )n R B ] R∗B
R(C = C )∗n [ R∗B ].
 
= 2k t,PRT (33)
dt
In Equation (31), [bimol ] represents the concentration of macroradicals terminated by bi-
molecular termination, either diffusional or through reaction-diffusion. In Equations (32) and (33),
[ R(C = C )n R A ] and [ R(C = C )n R B ] are the concentrations of macroradicals terminated by
primary radical termination by R∗A and R∗B , respectively. R∗A and R∗B denote the phosphonyl
and benzoyl radicals which are yielded by dissociation of TPO under light exposure. In
the publication of Christmann et al. [102], the main kinetic rate constants are defined as
functions of the fractional free volume v f in order to consider their progressive diffusional
control throughout the photopolymerization process. Christmann and his co-authors model
the decreasing fraction of unoccupied volume in the reaction medium, the free volume v f ,
according to Equation (34), as



v f = 0.025 + α M T − Tg,M ΦM + α P T − Tg,P (1 − ΦM ). (34)

α is the thermal expansion coefficient, and Tg is the glass transition temperature with
the respective subscripts M (monomer) and P (polymer). ΦM denotes the volume fraction
of the monomer and is defined in Equation (35),

1 − Conversion
ΦM = ρ , (35)
1 − Conversion + Conversion · ρMP

where ρ corresponds to the volumetric mass density of the monomer (M) and the polymer
(P). According to the authors, the propagation rate constant k p can be modeled using the
following relationship (Equation (36))

k p0
kp =     . (36)
1 1
1 + exp A P vf − v f ,cP

The above expression incorporates the propagation intrinsic rate constant k p0 (i.e.,
without any diffusional control), a parameter A p which governs the rate at which k p de-
creases with viscosity, the free volume v f , and the critical fractional free volume v f ,cp at
which propagation becomes diffusion-limited. The initiation rate constant k i , as well as
the rate constant for primary radical termination k t,PRT are modeled similarly to the propa-
gation rate constant, see Equation (36). Christmann et al. [102] assume that the initiation
rate constant k i and the rate constant for primary radical termination start decreasing at
the same time and at the same rate as the propagation rate constant k p . This implies that
Polymers 2022, 14, 2074 24 of 58

the respective exponential factors Ai and At,PRT and the critical fractional free volume
v f ,ci and v f ,ct,PRT are assumed to be equal to A p and v f ,cp . The values of the intrinsic rate
constants k i,0 and k t,PRT0 depend on the nature of the primary initiating radical. Diffusional
bimolecular termination and subsequent reaction-diffusion processes are modeled using
Equation (37),
  −1
1
k t,b = k t,b0 1 + Rrd k p
    , (37)
k t,b0 + exp − At,b v1f − 1
v f ,ct,b

where k t,b0 corresponds to the intrinsic bimolecular termination rate constant (i.e., without
any diffusional control), Rrd is a constant, At,b is an exponential factor that governs the rate
at which k t,b decreases with viscosity, and v f ,ct,b represents the critical free volume at which
bimolecular termination becomes diffusion-limited. Christmann et al. [102] successfully
applied the aforementioned kinetic model considering simultaneously all possible termina-
tion pathways (bimolecular termination, primary radical termination, and radical trapping
by occlusion) to the photopolymerization initiated by a type-I photoinitiator (cleavage type,
i.e., photoinitiators that dissociate into two radicals following photon absorption), showing
a good agreement with experimental results. Furthermore, the authors were capable of
identifying the relative contribution of the different termination pathways throughout the
photopolymerization process. Christmann et al. [102] showed that bimolecular termination
is the major termination reaction during the whole photopolymerization process. However,
due to the progressive diffusion control of the polymerization reactions as the polymer
network grows and due to the cessation of initiation when the photoinitiator is totally
consumed, the ratio of bimolecular termination as well as of primary radical termination
and macroradicals evolves. Figure 22 provides a deeper insight into the evolution of the
termination modes during the photopolymerization reaction. The figure reveals that bi-
molecular termination is the main termination process. According to the authors, the strong
growth of the bimolecular termination at the early stages of photopolymerization can be
explained by the continuous production of macroradicals, as initiation occurs. Only in
the last stages of the photopolymerization process does primary radical termination (PRT)
become efficient. Christmann et al. [102] relate this to the competition between primary
radical termination and initiation reactions for the primary radicals until the last stage of
the photopolymerization reaction.

Figure 22. Evolution of the fractions of terminated species, propagating or trapped macroradicals
(left scale) and TPO concentration (right scale) as a function of the acrylate conversion.

The model based on the free volume principle, presented by Christmann et al. [102],
was also used by Gao et al. [104] to combine polymerization kinetics with reaction condi-
tions to realize a 3D printing preview for stereolithography.
In 2017, Wang et al. [105] proposed a point-wise mechanistic model for modeling
the photopolymerization reaction kinetics of Exposure Controlled Projection Lithography
(ECPL). Oxygen diffusion effects, which were found to have a significant influence on the
Polymers 2022, 14, 2074 25 of 58

size, shape and properties of parts fabricated with stereolithography, are also incorporated
in the model. The authors consider oxygen diffusivity in two dimensions as described in
Equation (38),

∂[O2 ] ∂2 [O2 ] ∂2 [O2 ]

= −k t,O2 [ R∗ ][O2 ] + DO2 2
+ DO2 . (38)
∂t ∂x ∂z2
k t,O2 denotes the rate constant for termination of a radical with oxygen, [ R∗ ] denotes
the concentration of free radicals, [O2 ] denotes the concentration of oxygen, and DO2
denotes the oxygen diffusion coefficient.
The models presented so far, and also models published, for instance, by Goodner et al. [96],
Lee et al. [106], Buback et al. [90], Dickey et al. [93,107] and Long et al. [87] assume that the
termination is due to radical recombination, which is the most common assumption for
termination [108]. However, Wen et al. [109] state that a kinetic model that ignores radical
trapping fails to predict two important aspects of experimental observations: Firstly, the
concentration of trapped radicals increases monotonically with conversion, whereas the
concentration of active radicals increases initially and then drops at high conversions [110].
Secondly, a higher light intensity leads to a lower fraction of trapped radicals at a given
conversion but to a higher trapped radical concentration at the end of the reaction [110]. In
order to include radical trapping, Wen et al. [109] extended the functional-group reaction
scheme shown at the beginning of this section with the formation of trapped radicals like
in Equation (39),
R∗ → R∗b , (39)
kb

in which R∗b are trapped (buried) radicals and k b is the rate constant for radical trapping
(burying). Radical trapping is assumed to take place according to a unimolecular first-order
reaction. Consequently, the material  balance equations proposed by Wen et al. [109] include
the trapped radical concentration Rb and the active radical concentration [ R∗ ], according
∗


to Equation (40),
d R∗b
 
= k b [ R ∗ ]. (40)
dt
In the proposed model, the propagation rate constant k p , the termination rate constant
k t , as well as the rate constant for radical trapping k b , are simple functions of free volume
following the model developed by Anseth and Bowman [38,101]. The model proposed
by Wen et al. [109] presumes that the rate constant for radical trapping k b increases with
conversion as radical trapping occurs more and more often as the chain growth in the
course of the polymerization proceeds. Therefore, the rate constant k b is modeled to increase
exponentially with the inverse of the fractional free volume v f according to Equation (41),
!
Ab
k b = k b0 exp , (41)
vf

in which k b0 is the pre-exponential factor and Ab the dimensionless activation volume that
governs the rate at which radical trapping increases as a function of fractional free volume.
Wen et al. [109] compared their proposed model for predicting the reaction rate R p (with
and without radical trapping) to photo-DSC experimental results during the polymerization
of DEGDMA with 0.42 mW cm−2 light intensity and 0.1 wt.% DMPA (Figure 23). The
markers on the dashed curve for the model considering radical trapping show (a) the onset
of auto-acceleration, (b) reaction-diffusion becoming dominant for termination, (c) radical
trapping becoming dominant for termination, and (d) the propagation reaction becoming
reaction-diffusion controlled. The figure shows that the model, including radical trapping,
is consistent with experimental measurements of the polymerization rate R p . The reaction
rate before ∼ 25% conversion is not severely affected by radical trapping. However, for
conversions higher than 25%, the reaction rate appears to be higher without trapping,
finally resulting in the prediction of a higher final conversion.
Polymers 2022, 14, 2074 26 of 58

Figure 23. Predicted reaction rate (with and without radical trapping) and actual reaction rate mea-
sured by photo-DSC versus conversion during the polymerization of DEGDMA with 0.42 mW cm−2
light intensity and 0.1 wt.% DMPA. Reprinted from reference [109] with permission from the Ameri-
can Chemical Society.

Another model including the termination by radical trapping was proposed by Batch
and Macosko [85]. However, this model is rather limited as it requires the a priori knowl-
edge of the final monomer concentration and the monomer concentration when trapping
begins if it does not begin immediately. Perry et al. [108] suggested a model, building on the
model introduced by Batch and Macosko [85] that circumvents the limitations mentioned
above and that also includes dark reactions that occur in the course of the photopolymer-
ization, i.e., the polymerization does not stop once the light source is extinguished, but
reactions continue in the dark period.
The models discussed hereinabove give accurate and useful predictions for isothermal
systems. However, O’Brien and Bowman [97] stated that some polymerization systems
were more complex and the inclusion of additional factors such as heat generation, heat
transfer and mass transfer was necessary. The authors pointed out that in particular heat
effects were important, as the kinetic constants, as well as the diffusion coefficients, are
a function of temperature. Therefore, the one-dimensional kinetic photopolymerization
model proposed by O’Brien and Bowman, which is based on the work of Goodner and
Bowman [81,95,96], incorporates not only the temporal and spatial variation of species
concentration, temperature and light intensity through the sample depth but additionally
heat and mass transfer effects. As pointed out in the publication, additionally to the
model proposed by Goodner and Bowman [95], this model is capable of simulating both
photobleaching and non-photobleaching initiators.
In the kinetic model proposed by O’Brien and Bowman [97], mass transfer effects are
considered by adding a diffusive flux term to the species balance as shown in Equation (42),
 
dCi d d x̂i
= Ri + ĈDi , (42)
dt dz dz

in which the parameter Ci denotes the concentration of species (while the index i corre-
sponds to the respective component). The species balance is set up for initiator, primary
radical, monomer, polymer radical, and dead polymer where the mobile species are the
initiator, monomer and primary radicals. Ri is the reaction term, i.e., the term for species
generation or consumption by the reaction. The second term corresponds to the diffusive
flux, including the effective total concentration of mobile species Ĉ, the diffusion coefficient
Di , and the effective mole fraction of mobile species x̂i . The diffusion coefficient for the mo-
bile species is calculated using the equation provided by Bueche [111]. The mathematical
form of Ĉ and x̂i can be found in the publication of Goodner and Bowman [95].
Polymers 2022, 14, 2074 27 of 58

In addition to the species balance, the energy transport is incorporated in the model
using an energy balance that includes heat transfer, as well as heat generation by radiation
absorption and heat generation by reaction according to Equation (43):

dT d2 T d[C = C ]
ρc p = k 2 + ε I I0 c∗I − ∆H , (43)
dt dz dt

in which ρc p dT
dt represents the heat accumulation with the density ρ and the heat capacity c p .
The first term of Equation (36) represents the heat transfer, in which k is the thermal conduc-
tivity and z is the spatial coordinate for the sample depth. The second term of Equation (36)
represents the heat generation by radiation and assumes that all energy absorbed by the
system is converted into heat; ε I is the molar absorption coefficient of the initiator, I0 the
incident light intensity, and c∗I the concentration of all light-absorbing species. By adjusting
the initiator molar absorption coefficient, varying optical densities can be simulated. The
last term of Equation (36) corresponds to the heat generation by a reaction due to the
exothermic nature of the polymerization reaction; ∆H denotes the heat of polymerization
and d[C = C ]/dt the consumption of double bonds (monomer consumption).
Phenomena like diffusion-controlled kinetics and termination by reaction-diffusion are
described in terms of the fractional free volume theory of the polymerizing mixture. Chain-
length independent propagation, termination and inhibition are assumed. Furthermore,
bimolecular termination is realized by considering a lumped termination rate constant k t
that accounts for both combination and disproportionation. Physical and thermal properties
such as density, specific heat and conductivity are assumed to remain constant in the course
of the reaction. The attenuation of the curing light caused by the absorption of light by the
initiator is modeled according to Beer–Lambert law (Equation (44)),

I = I0 exp(−ε I c I z), (44)

where I denotes the light intensity at a depth z, and c I the unreacted initiator concentration.
O’Brien and Bowman [97] adjusted the overall light absorbance, which determines the
degree of attenuation, by varying one of the following parameters: initiator concentra-
tion, molar absorption coefficient and sample thickness. The authors pointed out that
different combinations of these three parameters led to the same absorbance and therefore
to the same initial light attenuation. However, in the course of the reaction, differences
would be apparent as each of the three variables has distinct influences on other aspects
of polymerization.
In order to solve the differential equations of the species and energy balances, O’Brien
and Bowman [97] considered different thermal boundary conditions (insulating, conducting
and constant temperature boundary conditions) affecting the heat transfer in a polymer-
izing sample. The inhibitory effect of oxygen on free-radical photopolymerization was
incorporated into the model presented above by O’Brien and Bowman in 2006 [88].
Intensive research with regard to modeling the curing kinetics and deriving analytical
relationships between curing depth and crosslink time, as well as considering the effects
of oxygen inhibition and viscosity was conducted by Lin and his co-authors [30,112–117].
For instance, in 2016 Lin et al. [112] proposed a comprehensive mechanistic model for
the cure kinetics of photopolymerization in optically thick polymers, providing useful
guidance for the parameters selection and optimization for predicting the curing time for
various polymer thicknesses in photoinitiated polymerization systems. In their publication,
the authors state that most kinetic models presented in the literature are based on the
oversimplified assumption that the photolysis product becomes completely transparent
after polymerization. However, the distribution of the photoinitiator is non-uniform
(depletion of the photoinitiator concentration) and the UV-light may still be absorbed by
the photolysis product besides the absorption of the monomer [112]. Consequently, the
authors derived kinetic equations for the concentration of the unreacted photoinitiator
C (z, t), see Equation (45), and the UV light intensity, see Equation (46).
Polymers 2022, 14, 2074 28 of 58

∂C (z, t)
= − a· I (z, t)·C (z, t), (45)
∂t
with a = 8.36λφε 1 , where λ is the light wavelength, φ is the quantum yield, and ε 1 is the
molar extinction coefficient of the initiator.
∂I (z, t)
= −2.303 [(ε 1 − ε 2 )·C (z, t) + ε 2 C0 F (z) + Q]· I (z, t), (46)
∂t
where ε 2 is the molar extinction coefficient of the photolysis product, C0 is the initial
concentration of the photoinitiator on the surface C0 = C (z = 0, t = 0), F (z) is a distribution
function for the initial photoinitiator concentration in the polymer system, and Q is the
absorption coefficient of the monomer and the polymer repeat unit. For the simplified
case reported, for instance, by Ivanov et al. [118], F (z) = 1, Q = 0, and ε 2 = 0 applies.
However, Lin et al. [112] report analytical equations for the general case of a non-uniform
photoinitiator concentration F (z) without the assumption of Q = 0, and ε 2 = 0. The
initial non-uniform photoinitiator concentration is given by the distribution function, see
Equation (47),
0.5z
F (z) = 1 − , (47)
D
where D is the half-width at half-maximum [112]. When D is much larger than the polymer
thickness F (z) = 1 applies, corresponding to the flat distribution or uniform case. The equa-
tions of Lin and his co-authors also include a time-dependent generalized Beer–Lambert
law, denoted as “Lin law”. According to the authors, the oversimplified assumption
that the light intensity in the polymer follows a conventional Beer–Lambert law (with
neglected depletion of photoinitiator concentration C (z, t)) is only valid for optically thin
materials. The authors state that the depletion of the photoinitiator C (z, t) will also affect
the time-dependent profiles of the intensity I (z, t) which, in general, will not follow the
Beer–Lambert law. Therefore, the so-called “Lin law” has two modifications compared to
the Beer–Lambert law: (1) it has a time-dependent term and (2) it has a z-dependent term
accounting for the fact that the photoinitiator concentration C (z, t) is an increasing function
of z (for t > 0), leading to a more accurate description for both C (z, t) and I (z, t).
Lin et al. [112] furthermore define the crosslinking time T ∗ based on the time needed
to deplete the photoinitiator concentration, i.e., the time needed for the completion of the
gelation procedure. T ∗ was found to be an exponentially increasing function of z and to
be inversely proportional to the UV light intensity. Another important key parameter of
photopolymerization, namely the local photoinitiation rate of production of free radicals
was modeled according to Equation (48):

R(z, t) = 167.2·λ·ε 1 ·φ· I (z, t)·C (z, t), (48)

According to the above equation, the photoinitiation rate is proportional to the product
ε 1 ·C (z, t) and the light intensity I (z, t), two competing factors. The authors derived an
analytic equation for the optimal product ε 1 C0∗ , expressed in Equation (49),

1
ε 1 C0∗ =   exp[ a( I0 t) A(z)]. (49)
ε2
1− ε1 z

Consequently, the optimal photoinitiation rate is an increasing function of the initial

light intensity I0 , the quantum yield φ, and the initiator absorption ε 1 . Furthermore, the
optimal photoinitiation rate is a decreasing function of the depth z, the initial photoinitiator
concentration C0 , and the photolysis product absorption ε 2 .
In 2019, Lin et al. [114] modeled the optimal conditions for maximum efficacy and
crosslink depth, presenting analytic formulas for curing depth and crosslink time of pho-
toinitiated polymerization, without the assumption of thin-film or spatial average. Fur-
thermore, the authors derived kinetic equations for the efficacy and curing depth for
Polymers 2022, 14, 2074 29 of 58

systems with type-I radical-mediated and type-II oxygen-mediated pathways under the
quasi-steady-state assumption and bimolecular termination [115].
More recently, Lin et al. [116] presented the theoretical modeling and kinetics of the
red-light controlled oxygen inhibition for improved UV-light initiated monomer conversion
based on a novel strategy presented by Childress et al. [119]. As mentioned earlier, free-
radical photopolymerizations are particularly sensitive to oxygen inhibition, i.e., oxygen
can react with primary radicals and thereby reduce the efficiency of photopolymerization
by quenching the primary initiating and propagating radicals. Conventional strategies to
reduce oxygen inhibition in photoinduced polymerizations include physical methods (e.g.,
working in an inert environment, use of multiple photoinitiators with different rates of ini-
tiation, . . . ) and chemical mechanisms (e.g., additives that are insensitive to oxygen, . . . ).
In 2019, Childress et al. [119] reported a novel strategy for red-light-controlled oxygen
inhibition. The authors used red-light to preirradiate the monomer, followed by the UV-
light excitation of the photoinitiator in order to independently achieve photosensitization
and photoinitiation via irradiation of the two distinct absorption bands. The technique of
Childress et al. [119] allows us to partially or completely eliminate the induction time using
red light.
The research of Lin et al. [30] also considers three-component photoinitiating systems
(A/B/C) where the co-initiators/additives serve the regeneration of the photoinitiator A
and the generation of extra radicals.

3.1.2. Phenomenological Models for Free-Radical Photopolymerizations

Phenomenological models are set up to treat the conversion of monomers as the only
variable to characterize the polymerization process and ignore the variations of other
species [82]. Therefore, Equation (1) represents the reaction that describes the whole
curing process.
The simplest and most common analytical form of the function of the degree of cure is
given by Equation (50),
f ( α ) = (1 − α ) n , (50)
in which n represents the reaction order and α the extent of the reaction [29]. Furthermore,
the rate constant Kc ( T ) in Equation (1) is assumed to follow the Arrhenius equation that
can be expressed according to Equation (51),

− Ea
 
Kc ( T ) = K0 exp , (51)
RTabs

with the so-called pre-exponential factor K0 , the activation energy Ea , the universal gas
constant R, and the absolute temperature Tabs .
Replacing Kc ( T ) and f (α) in Equation (1) with the correlations summarized in
Equations (38) and (39), the so-called n-th order kinetic model [74] according to Equation (52)
is obtained:
− Ea
 
dα
= K0 exp (1 − α ) n . (52)
dt RTabs
For an isothermal reaction, the n-th order kinetic reaction model predicts a maximum
of the reaction rate at the start of the reaction (t = 0). Obviously, this model cannot be
applied for photopolymerization reactions showing a maximum value of the reaction rate
at any point, rather than the reaction starting point [120]. An alternative to the n-th order
kinetic model is the so-called autocatalytic model proposed by Kamal et al. [120,121]. The
general model equation of the autocatalytic model is given by Equation (53),

dα
= kαm (αmax − α)n , (53)
dt
in which α is the relative conversion, k an Arrhenius-type rate constant, m the autocatalytic
exponent, n the reaction order exponent, and αmax the maximum conversion. αmax is consid-
Polymers 2022, 14, 2074 30 of 58

ered to be 1 assuming the completed reaction. The applicability of the autocatalytic model
in modeling free-radical photopolymerizations (as well as cationic photopolymerizations)
is due to the fact that an auto-catalyzed reaction assumes a propagation reaction that is
characterized by an accelerating conversion rate, with its maximum occurring well after
conversion initiation [122]. This implies that for systems according to this auto-catalytic
model, the reaction rate is initially equal to zero, and a maximum value of the reaction
rate occurs at intermediate conversion. According to Achilias [84], the autocatalytic model
of Kamal [120] is often used to describe diffusion-controlled polymerization reactions.
For instance, the autocatalytic model has been used to describe the photoinitiated poly-
merization of dental resin-monomers and composite systems as well as multifunctional
acrylates [33,46,123–126]. However, as mentioned earlier, this model is essentially phe-
nomenological and does not provide any mechanistic insight.
A phenomenological model for the isothermal kinetic behavior of an acrylic resin
accounting for the effect of auto-acceleration, vitrification and light intensity on the reaction
kinetics has been proposed by Maffezzoli et al. [127]. The authors used the autocatalytic rela-
tion introduced by Kamal [120] to describe the conversion state with Arrhenius and power-
law relationships for temperature and light intensity dependence. The kinetic behavior is
modeled using a simple pseudo-autocatalytic expression summarized in Equation (54),

dα
= K ( Ia , T )αm (αmax − α)n (1 − α), (54)
dt
in which K is a rate constant characterized by an Arrhenius-type dependence on tempera-
ture T and the absorbed light intensity Ia , m and n are positive fitting parameters indepen-
dent of temperature, and αmax is the maximum degree of reaction obtained in isothermal
DSC cure experiments. αmax is calculated as the ratio of the heat developed during the
experiments (Qisothermal ) and the maximum heat of reaction measured in a non-isothermal
experiment (Qtotal ). In their publication, Maffezzoli et al. [127] assumed that laser exposure
(wavelength λ∗ ) led to an absorbed light intensity according to Equation (55),
∗
 
Ia = I0 (λ∗ ) 1 − 10−ε(λ )[ In]z , (55)

in which ε denotes the molar absorbance of the photoinitiator depending on the wavelength
of the light source, [ In] the initiator concentration, and z the thickness of the sample. The
rate constant is consequently modeled according to Equation (56)

K = K0 ( T ) I0b , (56)

containing the Arrhenius type temperature-dependent kinetic constant K0 ( T ) and the

fitting parameter b.
The considerations of Kamal et al. [120], as well as Maffezzoli and colleagues [127],
were advanced by Rehbein et al. [128] for modeling the crosslinking reaction of pho-
topolymers used in additive manufacturing processes such as digital light processing and
stereolithography. The model of Rehbein et al. [128] describes the maximum attainable cur-
ing degree as a function of absolute temperature and light intensity. The model considers
both, isothermal and non-isothermal processes, as well as time-dependent light intensity
boundary conditions. Similar to the model proposed by Maffezzoli et al. [127], the model
of Rehbein considers not only time-dependent light intensity but also the fact that the
light intensity depends on the vertical location z(t) since it evolves during the printing
process. Therefore, the curing degree is modeled as an autocatalytic model of (m + n)-th
order according to Equation (57)

dα
= [k1 ( I (z(t), t), T ) + k2 ( I (z(t), t), T )αm ]·(αmax − α)n , (57)
dt
involving the location- and time-dependent light intensity I (z(t), t).
Polymers 2022, 14, 2074 31 of 58

In 2005, Bartolo [75] proposed a coupled photothermal phenomenological kinetic

model to describe photoinitiated curing reactions and to correctly model the physical and
chemical changes occurring in the bulk and in the surroundings of the material exposed
to UV light, as well as the rates at which these changes occur. One main feature of the
model is the modeling of the heat flow within the curing process, assuming that the
irradiated material volume absorbs energy and initiates the phase change in the material
(transformation from liquid to solid).
The temperature field in the region exposed to UV radiation can be described by
the equation of heat conduction in a general three-dimensional framework according to
Equation (58),
∂T
ρc p = ∇·(λ∇ T ) + q g , (58)
∂t
containing the density ρ, the specific heat c p , the thermal conductivity λ, the temperature
gradient ∇ T, and the volumetric heat generation q g . It is assumed that the internal heat
generation q g is influenced only by the heat of polymerization according to Equation (59),

dα
q g = ρH , (59)
dt
in which H represents the total heat release and dα/dt corresponds to the kinetic model.
Furthermore, appropriate boundary conditions regarding temperature, heat flux emitted
from the laser and convectional heat loss are assumed (for detailed information, see [129]).
For this model, the light intensity values at the resin surface were defined by assuming
a Gaussian intensity distribution. The absorption of UV radiation is defined by the Beer–
Lambert law and, consequently, the variation of the light intensity along the thickness
of the resin layer (decrease in light intensity with depth) can be modeled according to
Equation (60), "   #
s(t) 2
I (s, z, t) = I0 exp −2 exp(−ε[ In]z), (60)
w0

where s(t) represents the position in time of a point under irradiation, z the penetration
depth (z = 0 on the resin surface), I0 the peak light intensity, w0 the laser beam radius, ε
the absorptivity of the layer, and [ In] the initiator concentration.
The coupled photothermal phenomenological kinetic model by Bartolo [75] conse-
quently describes the evolution of the degree of cure according to Equation (61),

−E
 
dα 1
=  i p
ϕI exp [ In]q αm (1 − α)n , (61)
dt 1 + exp[ξ α − αd) RTabs

in which ξ is the diffusion constant, αd the critical value of the curing degree corresponding
to the onset of diffusion-controlled effects over the curing reaction, ϕ the pre-exponential
factor of the rate constant, I the light intensity, E the activation energy, R the gas constant,
and Tabs the absolute temperature. The parameters p and q are constants. The exponents
m and n denote the reaction orders; consequently, m + n is the overall reaction order.
The kinetic parameters ξ, αd , E as well as the exponents m, n are assumed to be non-
constant but to vary in a non-linear way with temperature, light intensity and initiator
concentration [129].
Bartolo et al. [75] also modeled the glass-transition temperature and suggested a
non-linear relationship between the glass-transition temperature Tg and the curing degree
according to Equation (62),
Tg = Tg0 − Tg0 α + Tg∞ α3 , (62)
with the glass-transition temperature Tg0 of the uncured polymer and the glass-transition
temperature Tg∞ of the fully-cured polymer. As crosslinking increases, the glass transition
Polymers 2022, 14, 2074 32 of 58

temperature increases due to the restriction of chain movements, associated with a decrease
in the free volume during the curing process.
In 2019, Yang et al. [130] proposed a phenomenological kinetic model where the
degree of cure is put in relation to the mechanical properties of laser-based additively
manufactured components. The authors presented a mathematical model to quantify the
tensile strength and hardness of stereolithography fabricated materials by estimating the
solidification level. The authors assume that a specified layer i can be cured by the UV light
more than once since printed layers can still be slightly targeted by the light that penetrates
through the new fresh layer. Therefore, the printed layer is assumed to be re-cured when it
is inside the photosensitive liquid resin implying that all printed layers are continuously
cured. The phenomenological expression for the degree of cure for a specific layer i when it
is solidified for the jth time is shown in Equation (63):
( ! )
j j jq j p −E j m
h
j
in
αi (d, θ ) = tci Si Ii exp j
αi (d, θ ) 1 − αi (d, θ ) . (63)
RTi (d, θ )

j
tci corresponds to the curing time for the i-th layer when it is being cured for the
j-th time, d denotes the thickness of the layer, θ is the stratification angle between surface
j
normal vector and build direction, Si is the photoinitiator concentration (which is assumed
j
to decrease in reverse proportion with the curing degree), Ti is the temperature of for the
i-th layer when it is being cured for the j-th time, and p, q, m, n are model parameters (e.g.,
related to environmental conditions, type of resin, a.s.o.) which are determined by best
fitting of the experimental results. The above expression is similar to the phenomenological
model derived by Bartolo et al. [75], see Equation (62), however, diffusion-controlled effects
are neglected in the model proposed by Yang et al. [130].
In the literature, several modifications of kinetic models are proposed to express the
diffusion limitations of reacting polymer chains with phenomenological models. A very
early approach by Kenny et al. [131] incorporated the diffusion-rate control into the reaction-
kinetic expression by using the maximum degree of conversion αm achieved by isothermal
curing. However, the authors assumed a linear relationship between the maximum degree
of conversion and the cure temperature Tcure , which led to an infinite value of αm with
increasing temperature. Park et al. [132] derived a phenomenological n-th order kinetic
model that incorporates the diffusion-rate control into the reaction-kinetic expression by
using the maximum degree of conversion achieved in isothermal curing, expressed in
Equation (64),  n
dα α
= k(T ) 1 − , (64)
dt αm ( T )
where k ( T ) is the rate constant, α is the degree of cure, αm is the maximum degree of
cure achieved by isothermal curing and n is the reaction order. The maximum degree of
cure as a function of isothermal cure temperature is expressed by an empirical equation
(Equation (65))
a
αm ( T ) = , (65)
1 + b · exp(−k m T )
where the values of the fitting parameters a, b, and k m are obtained by a curve fitting method.
A similar approach, using a simple autocatalytic expression, was presented in 2017 by
Kim et al. [133], see Equation (66),
 m  n
dα α α
=k 1− , (66)
dt αm ( T ) αm ( T )

expressing the maximum degree of cure by using an empirical equation in the form of Hill
functions (Equation (67))
ac
αm ( T ) = a c . (67)
b + ac
Polymers 2022, 14, 2074 33 of 58

Again, the values of the fitting parameters a, b, and c are obtained by a curve fitting method.

3.2. Cationic Photopolymerizations

UV-curable resins that are cured according to cationic photopolymerization mecha-
nisms behave differently than those cured according to free-radical mechanisms. In cationic
photopolymerization formulations, different photoinitiators and monomeric materials are
used. Cationic polymerization reactions start with the initiation step, during which active
centers are produced. The photoinitiation step is the only step in cationic photopolymer-
ization that depends on light. Once the active centers are produced, they propagate by
polymerization reactions without any further interaction with light.
Cationic photoinitiators are referred to as photoacid generators PAGs. Once a cationic
photoinitiator absorbs UV irradiation, the initiator molecule is converted into a superacid
that initiates the polymerization. In the so-called propagation step, the active center reacts
successively with a number of monomers such that they are covalently attached to the
growing polymer chain. The polymer chain length is determined by the number of propa-
gation steps that occur before the active center undergoes chain-transfer or termination.
Compared to free-radical systems, cationic systems exhibit a relatively long lifetime and
relatively small termination rates [134].

3.2.1. Mechanistic Models for Cationic Photopolymerizations

A mechanistic mathematical model to characterize cationic photopolymerization
kinetics of epoxy compounds as a function of temperature and exposure time has been
proposed by Nelson et al. [134,135]. The authors modeled the rate of change of the active
center concentration [ M+ ] as a combination of the rate of active center generation by
photosensitization (first term) and the rate of consumption by termination (second term)
according to Equation (68),

d[ M+ ]
= k i [ A][ I ] − k t M+ ,
 
(68)
dt
in which k i represents the initiation rate constant accounting for a number of photo-
physical steps including excitation, intersystem crossing, exciplex formation and electron
transfer [136], [ A] the concentration of photosensitizer, [ I ] the initiator concentration, and
k t the termination rate constant.
The model suggested by Nelson et al. [134] assumes that one active center is produced
per photosensitizer and initiator molecule. Furthermore, all reactive centers are capable of
propagating. Integrating the rate of change of the active center concentration, including
the initial condition [ M+ ]0 = 0 and assuming an exponentially decreasing photosensitizer
concentration, Equation (69) is obtained:

k i [ A][ I ]
M + = [ A ]0
 
(exp(−k i [ A][ I ]t) − exp(−k t t)). (69)
k t − k i [ A][ I ]

Based on this mathematical model, Corcione et al. [28,137] proposed a temperature-

and intensity-dependent mechanistic model for the photopolymerization kinetics of an
epoxy-based resin for stereolithography, in which the rate of monomer consumption is
expressed according to Equation (70),

d[ M]
= −k p [ M] M+ ,
 
(70)
dt
in which k p denotes the propagation rate constant, [ M+ ] the active center concentration,
and [ M] the unreacted monomer concentration. The unreacted monomer concentration can
Polymers 2022, 14, 2074 34 of 58

be expressed as a function of the curing degree α and the initial monomer concentration
[ M]0 according to Equation (71):

[ M ] = [ M ]0 (1 − α ). (71)

Consequently, the rate of conversion depends on the reaction time and the light
intensity according to Equation (72):

dα k p k i [ A][ I ]
= [ A ]0 (exp(−k i [ A][ I ]t) − exp(−k t t))(1 − α). (72)
dt k t − k i [ A][ I ]

Another mechanistic model was proposed by Pantiru et al. [138] for cyclic acetals
using several cationic photoinitiators. In this model, the rate of monomer consumption is
modeled according to Equation (73),

d[ M]
= k p I + ·([ M] − [ M]e ),
 
− (73)
dt
with the rate constant for propagation k p , the concentration of active species [ I + ], the
monomer concentration at time t [ M ] and the equilibrium concentration [ M ]e .

3.2.2. Phenomenological Models for Cationic Photopolymerization

Similar to free-radical photopolymerizations, the simplest phenomenological model
for cationic photopolymerizations is the n-th order reaction model (Equation (51)). Auto-
catalyzed reaction models are generally described by the relation shown in Equation (51).
In 2002, van Assche et al. [139] proposed to quantify the effects of diffusion and
consequently mobility restrictions on the cure kinetics with direct estimation of a so-
called diffusion factor DF (α, T ). The diffusion factor is defined as the ratio between the
experimentally measured conversion rate and the predicted conversion rate at the same
reaction conversion, but without mobility restrictions according to Equation (74):
 
dα
dt ( α, T ) measured
DF (α, T ) =   . (74)
dα
dt ( α, T ) kin_model

The measured conversion rate is proportional to the heat flow measured in DSC, α
denotes the conversion and T denotes the absolute temperature. Applying the autocatalytic
model, proposed by Kamal [120,140], the predicted conversion rate is modeled according
to Equation (75),
 
dα
(α, T ) = (k1 + k2 αm )(1 − α)n = k kinetic (α, T )(1 − α)n , (75)
dt kin_model

in which k kinetic (α, T ) corresponds to the phenomenological rate constant. Combining the
equations listed hereinabove yields an apparent rate constant k apparent (α, T ) that includes
the effect of diffusion on the phenomenological rate constant k kinetic (α, T ) according to
Equation (76),  
dα
(α, T ) = k apparent (α, T )(1 − α)n , (76)
dt measured
with k apparent (α, T ) = DF (α, T )k kinetic (α, T ).
The apparent rate is quantified using the equation proposed by Rabinowitch [141]
(Equation (77)):
1 1 1
= + . (77)
k apparent k Di f f k kinetic
Polymers 2022, 14, 2074 35 of 58

In order to calculate the temperature dependence of the diffusion-controlled rate constant,

a Williams–Lendel–Ferry (WLF) equation [142], as proposed by Wisanrakkit et al. [143] and
summarized in Equation (78) can be used,

!
C1 T − Tg (α)
k Di f f = k Di f f 0 ( T ) exp
, (78)
C2 + T − Tg (α)

in which k Di f f 0 ( T ) is a constant related to local conditions for the creation of chemical

bonds, T the absolute temperature, Tg the glass-transition temperature, and C1 , C2 are
constants depending on the epoxy system. k Di f f 0 ( T ) is a constant employing an Arrhenius
temperature dependency. Consequently, van Assche et al. [139] modeled the diffusion rate
constant according to Equation (79),


  ED C1 T − Tg (α)
ln k Di f f (α, T ) = ln( A D ) − +
, (79)
RT C2 + T − Tg (α)

in which A D denotes the pre-exponential factor and ED the activation energy for the
Arrhenius-dependent diffusion rate constant k Di f f 0 ( T ).
Kim et al. [29] used the same approach for considering diffusion-controlled reactions
in the phenomenological kinetic model. The experimental data and the prediction of the
autocatalytic model for the photopolymerization of ECH (epichlorohydrin) resin with 1%
of diaryliodonium hexafluoroantimonate (photoinitiator) showed a good agreement over
the entire conversion range (Figure 24). The kinetic parameters of the autocatalytic model
were obtained by non-linear regression analysis. The plot shows the polymerization rate
versus conversion at different temperatures: the experimental data at 30 ◦ C are shown as
circles, the experimental data at 70 ◦ C are shown as squares, and the model predictions are
shown as solid lines, respectively.

Figure 24. Comparison of experimental data and autocatalytic model of the cationic photopolymer-
ization of ECH for different temperatures: Experimental data at 30 ◦ C (◦), experimental data at 70 ◦ C
(), and prediction of the autocatalytic model (solid line). Reprinted from reference [29] with the
permission of John Wiley and Sons.

The autocatalytic model shown in Equation (58) was also successfully used for the
modeling of the cationic photopolymerization of cycloaliphatic diepoxide (CADE) systems
with different photosensitizer concentrations [144]. Harikrishna et al. [122] also used the
autocatalytic model for the cationic photopolymerization of 1,4-cyclohexane dimethanol
diglycidyl ether. However, in this case, the kinetic parameters were studied by Levenberg-
Marquardt [145,146] by a non-linear regression method instead of a conventional linear
method in order to obtain more accurate values of the apparent rate constant. Further
examples of modeling cationic photopolymerizations based on the autocatalytic model
include publications from Boey et al. [147], Abadie et al. [148], and Macan et al. [149].
Polymers 2022, 14, 2074 36 of 58

However, the dependency of the rate constant on the light intensity is neglected in all of
these models.
In 2012, Golaz et al. [150] investigated the polymerization kinetics for the cationic
photopolymerization of common difunctional cycloaliphatic epoxy monomers initiated by
the decomposition of diaryliodonium salt photoinitiators and an isopropyl thioxanthone
photosensitizer. The authors used an autocatalytic expression proposed by Sesták et al. [151]
according to Equation (80),
   m  n
dα α α
= k( T, I ) 1− , (80)
dt t,T,I αmax αmax

in which k ( T, I ) is the rate constant, m the autocatalytic exponent, n the reaction order, α
the degree of conversion, and αmax the maximum conversion.
The rate constant is dependent on the temperature T and light intensity I, represented
by Equation (81),
k ( T, I ) = k0 ( T ) I β , (81)
with the temperature-dependent kinetics constant k0 ( T ) following the Arrhenius equation
and the exponent β as a fitting parameter.
The transmitted light intensity I is assumed to vary with the path length z following
the Beer–Lambert law according to Equation (82),

I (λ, z) = I0 (λ) exp(−µz), (82)

in which I0 denotes the incident light intensity depending on the wavelength λ and µ the
attenuation coefficient.
Figure 25 shows the experimental photo-DSC and modeled conversion rates versus the
normalized conversion of a cycloaliphatic epoxy compound at a temperature of 30 ◦ C and
a light intensity of 50 mW cm−2 . The conversion is normalized with the final (maximum)
conversion, depending on temperature and intensity. The figure reveals that it was possible
to predict the conversion rate and the conversion with reasonable accuracy up to vitrifi-
cation applying Equation (81). However, after vitrification, the polymerization was faster
than predicted by the autocatalytic model, confirming observations by Corcione et al. [137]
about the excess of “free volume”.

Figure 25. Comparison of experimental photo-DSC (black line) and modeled (red line) conversion
rates versus normalized conversion of a cycloaliphatic epoxy compound at 30 ◦ C and a light intensity
of 50 mW cm−2 . Reprinted from reference [150] with permission of Elsevier.

The autocatalytic expression proposed by Sesták et al. [151] was also successfully
applied to the kinetic modeling of bifunctional epoxy monomers by Voytekunas et al. [152].
However, as opposed to the model suggested by Golaz et al. [150], in this case, the rate
coefficient k( T ) is assumed to depend on temperature and photoinitiator concentration.
Polymers 2022, 14, 2074 37 of 58

Another phenomenological approach, presented by Jiang et al. [153], studied the reac-
tion kinetics of the photopolymer using the Avrami theory of phase change for isothermal
phase transfer as already applied for describing the cure kinetics of epoxies for instance by
Xu et al. [154] or Chen et al. [155]. The Avrami theory in its general form is represented by
Equation (83),
α(t) = 1 − exp(−K ·tn ), (83)
in which α(t) represents the reaction time-dependent curing degree, K the reaction speed
constant, and n the reaction order. The Avrami theory was originally used to describe the
kinetic process of polymer crystallization [156]. However, Pollard et al. [157] argued that
it was possible to predict the curing process of thermosets using the Avrami equation as
crystallization can be considered as a physical form of crosslinking. In order to accurately
present the photocuring kinetics, Jiang et al. [153] replaced the reaction speed constant K,
as well as the reaction order n, with undetermined coefficients a and b that depend on the
light intensity I according to Equation (84),
 
α(t, I ) = 1 − exp − a·tb , (84)

for which a and b can be calculated according to Equation (85):

a( I ) = c· I + d ,
(85)
b( I ) = e· I + f .

The undetermined constants c, d, e, and f, therefore, the undetermined coefficient

kinetic parameters a and b were determined by linearly fitting isothermal photo-DSC
experiments at different light intensities.

4. Implementation of Photopolymerization in Numerical Simulations

According to Marschik et al. [158] numerical methods can be used to effectively derive
approximate solutions to models for which analytical solutions are not available. Numeri-
cal procedures are capable of handling large equation systems with different degrees of
nonlinearities. Therefore, numerical simulations provide a useful tool to study and exper-
iment with the complex interactions within photopolymerization reactions. Numerical
implementation of photopolymerization reactions can yield valuable insight into which
variables are more important than others and how these variables interact, e.g., exposure
time and intensity of radiation. Therefore, some approaches regarding the implementation
of photopolymerization models by numerical simulations are shown in this section.
As already presented in Section 3.1.2., Bartolo [75] proposed a coupled photothermal
phenomenological kinetic model for free-radical photopolymerizations. In this model, the
law of conservation of energy, describing the heat transfer phenomena, is coupled with an
advanced kinetic expression that describes the cure kinetics in detail. The authors assume
that the temperature field in the exposed region (laser light source) can be described by
the two-dimensional heat conduction equation in cylindrical coordinates according to
Equation (86),    
∂T 1 ∂ ∂T ∂ ∂T dα
ρc p = rkr + kz + ρH , (86)
∂t r ∂r ∂r ∂z ∂z dt
The above equation assumes a simplified isotropic/anisotropic material with density
ρ, specific heat c p , thermal conductivities kr and k z , and a total heat release H. Furthermore,
T represents the temperature, t the time, and dα dt the kinetic model. The internal heat


generation, expressed by ρH dα dt is only due to the heat of polymerization. The solution

of the heat conduction equation requires the knowledge of initial conditions (i.e., initial
temperature Ti and initial value of the fractional conversion αi ) and must satisfy specific
Polymers 2022, 14, 2074 38 of 58

boundary conditions (e.g., light intensity and convection). Bartolo [75] introduced the
initial temperature Ti , according to Equation (87),

T (v, 0) = Ti , (87)

as well as the initial value of the fractional conversion αi , according to Equation (88),

α(v, 0) = αi (88)

in the domain being studied, where v represents a generic point in space. The thermal
kinetic boundary conditions, namely a specified temperature, a specified light intensity,
and a convection boundary condition are illustrated schematically in Figure 26 and given
in Equations (89)–(91).

Figure 26. Schematic representation of the thermal-kinetic boundary conditions considered for the
different regions Γ1 , Γ2 , Γ3 of the domain Ω. Redrawn from reference [75].

The boundary conditions proposed in the publication include a specified temperature

Ts , according to Equation (89),
Ts = T (v, t) at Γ1 , (89)
a specified light intensity, according to Equation (91),

∂T
kn − I (v, t) = 0 at Γ2 , (90)
∂n
as well as a convection boundary condition, according to Equation (92),

∂T
kn + h( T (v, t) − T∞ ) = 0 at Γ3 . (91)
∂n
T (v, t) is the temperature at the generic point v in space at time t, ∂T/∂n is the
derivative of the temperature in the direction normal to the surface, I is the light intensity,
h is the coefficient of heat transfer, and T∞ is the temperature of the surrounding space.
The decrease in light intensity with depth is assumed to obey the Beer–Lambert law.
Furthermore, any optical scattering effects and the flow of material due to convection or
diffusion are both considered negligible. For the computer implementation of the thermal-
kinetic model and the numerical solution of the heat conduction equation, subject to initial
conditions and boundary conditions, two stages of approximation are involved: spatial
approximation and temporal approximation. The generic domain Ω is discretized into
an appropriate number of linear rectangular finite elements Ωei . Bartolo [75] uses the
Galerkin method, which is the most common weighted residual method to transfer the heat
conduction equation to a form suitable for Finite Element Analysis (FEA) and to rewrite
the heat conduction equation at the element level. The two-dimensional heat conduction
equation can then be written in matrix form as shown in Equation (92):
.
C T + KT = F, (92)
Polymers 2022, 14, 2074 39 of 58

where C is the heat capacity matrix, K is the global “stiffness” matrix (also called the
conductivity matrix) and F is the equivalent nodal heat flow vector. In practice, the above
matrices are established for each element Ωei separately, and then assembled to give the
global matrices. The global matrices can then be solved for the nodal temperatures by
any numerical solution technique. To integrate Equation (93) with respect to time, the
Crank–Nicholson method is used. According to the Crank–Nicholson algorithm, which is
a finite difference method to numerically solve the heat conduction equation and similar
partial differential equations, the unknown values of the temperature at the time tn+1 can
be determined through the known temperatures at the time tn considering a temporal
approximation shown in Equation (93):

1 . . 
Tn+1 = Tn + ∆t T n+1 + T n . (93)
2
The fractional conversion, predicted by the kinetic model, is obtained with a fourth-
order Runge–Kutta procedure. The computer implementation was organized in two levels:
on the main level, developed in Visual Basic, all necessary input parameters were defined,
whereas the routine level, developed in Fortran 77, represents the computational level.
Flowcharts of the implemented code can be reviewed in the original publication [75].
From the simulations, Bartolo observed that the exothermic polymerization reaction and
the irradiation process result in a temperature increase in the exposed region. As the
reaction starts to slow down due to diffusion limitations, the temperature decreases due to
conduction and convection dissipation effects until an equilibrium between the dissipated
heat and the heat generated by irradiation is obtained after vitrification. Bartolo furthermore
showed that the conversion typically decreases by increasing the distance from the light
beam, due to the decrease in light intensity.
In 2020, Taki et al. [159] presented a simplified two-dimensional numerical simulation
for the Continuous Liquid Interface Production (CLIP). The CLIP system is a 3D printing
process, in which the liquid photopolymer resin is selectively exposed to UV light and is
solidified into parts. The innovation of CLIP, which makes it unique from stereolithography
applications (SLA) and digital light processing (DLP), is an oxygen-permeable membrane
that creates a dead zone underneath the part which allows for continuous curing as the
part is drawn out of the resin.
The aim of Takis’ approach was to simulate the shape of a printed object on a computer
before printing the object in order to determine its final shape under influences such as
volume shrinkage. The mechanistic model for free-radical photopolymerization, proposed
in earlier publications by the same author [160,161], includes the photopolymerization
kinetics of the initiation, propagation and termination (ordinal differential equations) as
well as oxygen inhibition reactions (partial differential equations). The rate coefficient of
propagation and termination was considered using the model of Anseth and Bowman [38].
The model neglects the effect of fluid flow induced by lifting the product, and a vertical
movement of the light source was assumed instead of a realistic motion of the lifting
of the photopolymerized parts. Furthermore, the temperature of the UV-curable resin
was assumed to be constant, which implies another simplification. However, the author
suggests that non-isothermal simulations of 3D printing in the CLIP system are subject to
ongoing studies. Figure 27a shows the geometry of the numerical 2D simulation, as well as
the areas subjected to UV exposure (colored in violet) as the UV light is moved downwards
and is emitted upwards.
Polymers 2022, 14, 2074 40 of 58

Figure 27. Simplified two-dimensional numerical simulation of photopolymerization reactions for

the Continuous Liquid Interface Production (CLIP): (a) Geometry used for the numerical simulations;
(b) Contour plot showing the results of the numerical simulations of C=C bond conversion as well as
UV light position at a lift-up speed of 0.1 mm s−1 [159,160]. Reprinted from reference [159] with kind
permission of the authors.

The boundary conditions for the numerical simulations include a periodic bound-
ary condition on the concentration of oxygen in the horizontal direction, according to
Equation (94), with the oxygen concentration [O2 ] depending on the spatial position z and
the time t:
[O2 ](0, z, t) = [O2 ](W, z, t). (94)
The oxygen concentration at the UV light source was assumed to be constant and
therefore equal to the initial oxygen concentration [O2 ]0 according to Equation (95):

[O2 ]( x, b, t) = [O2 ]0 . (95)

Equation (96) furthermore considers that oxygen was not capable of diffusing away
from the top, as expressed by
∂[O]2
|z=0 = 0. (96)
∂z
Taki et al. [159] also imposed a boundary condition on the light intensity below the
light source. Depending on the position x and z(z > b), the light intensity below the light
source was assumed to be zero, according to Equation (97),

E( x, z) = 0, (97)

with b as the vertical position of the light source. For z ≤ b the light was attenuated
according to the Beer–Lambert law in z-direction expressed by Equation (98)

E( x, z) = E0 ( x )·10−ε[ PI ](b−z) , (98)

where E0 ( x ) is the intensity of the UV light source depending on the position x, ε is the mo-
lar absorption coefficient, and [ PI ] is the molar concentration of the photoinitiator. Despite
the simplifications stated hereinabove, the numerical simulation showed the characteristic
properties of a CLIP system, including the dead zone, in which the polymerization was
inhibited due to radical quenching by oxygen. The numerical simulations were imple-
mented using MATLAB. Figure 27b shows the contour plot of the numerical simulations
of C=C double bond conversion. The horizontal white line represents the position of the
UV light, which is moved downward at a speed of 0.1 mm s−1 . At t = 0.10 s, the oxygen
concentration equaled 0.9 (shown as a dotted line) due to the fact that the initiator radical
Polymers 2022, 14, 2074 41 of 58

produced at t = 0.0 s reacted with oxygen dissolved in the formulation. With increasing
time, the C=C bond conversion starts to increase, indicated by the expansion of the interior
of the dotted line and its coloring. However, a dead zone is clearly visible between the
position of the light source and the C=C double bond conversion where polymerization
does not occur since oxygen molecules quench the radicals. The size of the dead zone does
not change with time due to the fact that the UV light intensity and oxygen permeation rate
were assumed to be constant. As the lift-up speed of the light source is of major concern
to the production speed, Taki et al. [159] also used a lift-up speed of 1 mm/s. The results
shown in Figure 28 clearly reveal that the dead zone is expanded by an increase in the
lift-up speed. As the maximum range of the color bar of Figure 28 is 50 times smaller than
that in Figure 27b, the simulations also reveal that a faster lift-up speed leads to a decrease
in the magnitude of the C=C bond conversion.

Figure 28. Contour plot showing the results of the numerical simulation of normalized oxygen
concentration of 0.9 and C=C bond conversion distribution for a lift-up speed of 1 mm/s. Reprinted
from reference [159] with kind permission of the authors.

In 2018, Gao et al. [162] presented a multi-physics modeling approach based on

a chemical kinetics model as well as a classical thermo-mechanical model to simulate
the printing process of a Digital Light Projection (DLP) printer. The numerical model,
proposed by the authors, aims at linking process conditions and material properties to
understand how, for instance, exposure time and layer thickness, influence the mechanical
properties of photopolymers and final products. The model, which is also suitable for
modeling multi-layer products, can be used to determine the relation between the process
conditions and the resulting properties of printed parts, for instance, warpage. The chemical
(photopolymerization) kinetics model is based on the model of Goodner and Bowman [95].
Furthermore, Gao et al. [162] consider the influence of a dye on the light penetration
depth D p . Adding a dye slightly influences the formulation and allows for a higher
controlled resolution in the layer thickness. The light absorption is modeled according to
Beer–Lambert’s law, assuming a light intensity I, according to Equation (99),

∂I I
= i f z ≤ z0 , (99)
∂z Dp

where the light direction is pointing to the z-direction and z0 is the position of the surface
where the light is projected.
Polymers 2022, 14, 2074 42 of 58

As mentioned before, the light penetration depth depends on whether the resin is
the only light absorber (see Equation (100)), or if a dye is added to the formulation (see
Equation (101)):
1
Dp = (100)
ε PI [ PI ]0 ln10
1
Dp = (101)
ε PI [ PI ]0 ln10 + ε dye [dye]0 ln10
where ε PI and ε dye denote the molar extinction coefficient of the photoinitiator and the
dye, whereas [ PI ]0 and [dye]0 denote the initial molar concentration of the photoinitiator
and the dye. The thermo-mechanical model considers the temperature increase due to the
absorption of the UV light and due to the exothermic nature of the photopolymerization
process (energy conversion). Two types of strains are considered: thermal expansion and
chemical shrinkage. Chemical shrinkage due to conversion is assumed to be isotropic and
linearly depending on the conversion. Consequently, chemical shrinkage can be expressed
with Equation (102),
εchem = − pεchem
max , (102)
where p is the double-bond conversion and εchem max is the maximum chemical shrinkage
when the monomer is fully converted to the polymer. Furthermore, a mixture relation,
shown in Equation (103), is introduced for the density ρ, the heat capacity c p , the thermal
conductivity κ, and the coefficient of thermal expansion α:

χ = p·χ poly + (1 − p)·χmono with χ = ρ, c p , κ, α. (103)

To capture the generated residual stress and to predict the warpage of the printed
samples Gao et al. [162] introduced a plastic model by adding plasticity based on a purely
elastic model. The plastic model assumes that the yielding stress of the material is linearly
proportional to the conversion before a specified transition value ptran and becomes constant
after that value. Empty material, representing a layer that is not printed yet, was realized
with a Young’s modulus of 10−12 Pa and a thermal conductivity equal to zero. The printing
process, involving the chemical reaction as well as the thermo-mechanical deformation,
was simulated using COMSOL Multiphysics (Version 5.3a).
The model developed by Gao et al. [162] was validated by comparing the conversion
measured by FTIR with the predictions of the model. A schematic representation of the
beam and the light direction is shown in Figure 29.

Figure 29. Schematic representation of the front surface and back surface of the beam, as well as the
light direction. The beam shows a positive bending shape. Redrawn from reference [162].

Gao et al. [162] found that the conversion measured for the back surface, independent
of the exposure time, agreed very well with the predictions of the model. According to
the authors, slight deviations might be attributed to the fact that the change in the light
absorption capabilities of the photoinitiator and the polymer are not captured in the model.
The conversion on the front surface was found to increase with exposure time for
both experimental measurements and model predictions. However, the conversion at
the front surface showed significantly higher values compared to the predictions of the
model. Gao et al. [162] provided some explanations for the deviations: for instance, free
radicals trapped in the polymer chains cannot be consumed immediately, leading to a
lower conversion measured in the experiments compared to the model predictions which
do not consider trapped radicals. Furthermore, oxygen exposure of the front surface was
Polymers 2022, 14, 2074 43 of 58

not considered in the proposed model. Evaluating the warpage of the printed samples
in terms of the deflection of the back surface revealed that a purely elastic model is not
sufficient to describe the residual stress in the beam. The purely elastic model predicts a
negative bending shape (i.e., a negative deflection) whereas a positive deflection is correctly
predicted by the plastic model.
As outlined in Section 3.1.1., Wang et al. [105] proposed a model for the photopoly-
merization reaction kinetics of Exposure Controlled Projection Lithography (ECPL). ECPL
is an additive manufacturing process in which UV curing radiation, controlled by a dy-
namic mask is projected through a transparent substrate onto the photopolymer resin
to fabricate three-dimensional structures. Compared to similar techniques to model the
ECPL process, for instance, published by Mizukami et al. [163], Erdmann et al. [164], and
Jariwala et al. [165], Wang et al. [105] present a more accurate, experimentally validated
model with revised photopolymerization rates. The authors used the capabilities of COM-
SOL software to model the photopolymerization reaction kinetics for a two-dimensional
finite element (FE) model, predicting the cured part geometry based on certain process
parameters. Additionally, changes in the refractive index and degree of conversion were
modeled throughout the reaction. A schematic sketch of the reaction chamber, modeled
in COMSOL, aiming to predict the height and profile of the final cured part, based on
the exposure time and intensity of the radiation, is shown in Figure 30. The rectangular
reaction chamber is assumed to be filled with liquid resin. The red arrows indicate the
regions into which irradiation is received by the monomer. All boundaries of the reaction
chamber are assumed to be insulated, which, according to the authors, closely resembles
the actual experiment conditions [105].

Figure 30. Reaction chamber modeled in COMSOL for the photopolymerization simulation of ECPL
(Exposure Controlled Projection Lithography). Redrawn from reference [105].

The polymerization kinetic model of free-radical photopolymerization of Wang et al. [105]

incorporates the effect of oxygen inhibition and accounts for oxygen diffusion in two
dimensions, as shown in Equation (104)

∂[O2 ] ∂2 [O2 ] ∂2 [O2 ]

= −k t,O2 [ R·][O2 ] + DO2 2
+ DO2 . (104)
∂t ∂x ∂z2
To estimate the concentration of the individual species at a given time and location
within the resin chamber (double bonds, live radicals, oxygen) the rate constants k p (for the
propagation of a radical), k t (for termination between two radicals), k t,O2 (for termination
of a radical with an oxygen molecule), and oxygen diffusion constant DO2 were modeled
along with a diffusional model (chdi) in COMSOL. Compared to Jariwala et al. [165], who
estimated the rate constants k t , k p , and k t,O2 by fitting the simulation results with the
experimental data from Fourier-Transform Infrared (FTIR) experiments, Wang et al. [105]
suggest, that the individual rate constants are not unique and may vary. Therefore, the
authors provide revised values for k t , k p , and k t,O2 .
In order to validate the ability of the reaction process model to accurately predict the
geometric profile of the cured part, the COMSOL simulated profiles were compared to the
Polymers 2022, 14, 2074 44 of 58

experimental results for several samples at different exposure times (10 s, 20 s, and 30 s)
with a UV light intensity of 8.86 W m−2 . The results, which can be reviewed in detail in the
publication of Wang et al. [105], confirm that the process model is effective in predicting
the geometry of the cured part. From the results, it is evident that the revised simulation
prediction of the cured height agrees very well with the experimental data points, whereas
the simulation before optimization of the rate constants predicts significantly higher cured
heights. Furthermore, the results show that the cured height increases with increasing
exposure time. Therefore, the simulation approach proposed by Wang et al. [105] is effective
in predicting the geometry of the cured part. Similar simulations were also conducted
by Jariwala et al. [166] who modeled the effects of oxygen inhibition and diffusion on the
polymerization reaction in mask-based stereolithography for acrylate-based monomers.

5. Photopolymerization Composites
The curing of composite formulations in polymer photochemistry is particularly chal-
lenging due to the fact that the fillers in the formulations potentially lower the penetration
depth of the UV light due to their opaqueness and/or their absorbance (see hereinabove).
If inorganic fillers such as fiber reinforcing materials [167,168] and/or micro- and nanopar-
ticles [169,170] have to be considered, the aspect of thermal conductivity also comes into
play. While (unfilled) polymers commonly have low thermal conductivity in the range of
0.1–0.2 W m−1 K−1 [171–173] (and seldomly in the range of up to 0.4 W m−1 K−1 [174,175],
inorganic fillers commonly have a significantly higher thermal conductivity [168–170].
Despite the fact that heat dissipation can potentially be considered to occur from com-
posite formulations to a higher extent than from (unfilled) polymerization mixtures, the
high relevance of composite materials shall be briefly highlighted in the example of three
material classes:

5.1. Fiber-Reinforced Polymers

In many cases, fiber-reinforced polymers show important properties such as wear re-
sistance, high thermal stability, impact resistance, chemical resistance and high mechanical
stiffness and strength [176,177]. A distinction can be made between two different types
of fibers, (1) the synthetic fibers, which include glass fibers and carbon fibers, and (2) the
natural fibers including cotton, hemp, jute, flax and others [178,179]. Due to the high
mechanical load-bearing, glass fiber and carbon fibers are primarily used with an increase
up to 113% in Young’s modulus when using 30 wt.% of carbon fibers [178].

5.2. Nanodielectrics
Nanodielectric composites combine the high dielectric permittivity of ceramic ma-
terials and the low loss factors, high dielectric strength (>300 kV/mm) and mechanical
flexibility of polymeric materials [180]. As polymeric materials, polyethylene, polypropy-
lene, (meth)acrylates and epoxides are used, into which mostly SiO2 , ZnO, MgO, BaTiO3 ,
Al2 O3 or TiO2 nanoparticles are filled [181,182]. Qiao et al. [183] established a methacrylate
composite with high permittivity (~20) and low tanδ (<0.02) over a wide range of frequen-
cies (1 kHz to 1 MHz) using functionalized BaTiO3 nanoparticles. The permittivity was
increased by 10 compared to the pure polymer, while the dissipation factor only increased
by the minimum amount of 0.01 [183].

5.3. Electromagnetic Shielding Materials

Electromagnetic [EM] pollution caused through radio, cell phones, cellular networks
and others, has become a worldwide problem due to its environmental and possible
hazardous effects [184,185]. In order to prevent this, composite materials are used that either
reflect or absorb the electromagnetic radiation. In principle, it can be distinguished between
two absorption material classes. Materials with high dielectric constant, such as BaTiO3 and
carbon particles, absorb the electric energy and convert it to thermal energy, whilst materials
with high permeability, such as Fe3 O4 , absorb the magnetic energy and convert it to thermal
Polymers 2022, 14, 2074 45 of 58

energy [186]. Other materials, used for electromagnetic shielding, are metallic and magnetic
materials such as steel, aluminum, copper, nickel, tin or carbon fibers [187]. Heat dissipation
from polymerization mixtures is detrimental to frontal polymerizations (Section 6), as the
heat generated in the course of the frontal polymerizations is the driving/initiating stimulus
for the start of the polymerization reactions in areas in the vicinity of the polymerization
front. If too much heat is dissipated, the polymerizations will stop, and the front will
stop migrating.

6. Frontal Polymerization
Frontal polymerizations, in which the reaction process is mainly governed by the
chemical and physical properties of the reacting system, is a promising curing strategy
that substantially helps to reduce manufacturing burdens. According to Pojman [43], the
general definition for frontal polymerizations is the following: A frontal polymerization
is a way to convert liquid resin into a solid material with a self-propagating reaction. In
general, the frontal polymerization (FP) technique is a process, in which a localized reaction
zone (the so-called polymerization front), once initiated by an external stimulus, acts as a
switch for the crosslinking of the reactants in adjacent areas. The process yields uniform
network formation throughout the reaction mixture. The allocation of FP techniques into
isothermal frontal polymerization (IFP), thermal frontal polymerization (TFP) and frontal
photopolymerization (FPP) is based on the external stimuli that are employed to initiate
and trigger the reacting front [188].
Isothermal-FP relies on the so-called “gel effect” or Norrish–Trommsdorff effect [189]
that occurs when the monomer and a thermal initiator diffuse into a solid polymer fraction
of the reaction mixture (polymer seed). By diffusing into the polymer seed, the solution of
monomer and thermal initiator dissolve its topmost layer to create a viscous region that
propagates through the reaction vessel (Figure 31). The polymerization occurs in both the
monomer solution and the viscous region but occurs faster in the viscous region, which
consequently propagates up the reaction vessel.

Figure 31. Schematic representation of isothermal frontal polymerizations (IFPs).

Reaction termination is inhibited by the high viscosity of the polymerized medium [190].
Isothermal-FP has been successfully used to produce gradient materials by incorporating
a dopant material, for instance, a second monomer or dye, into the monomer solution or
polymer seed [189]. However, it is limited to resin systems that exhibit the gel effect and
whose polymers are soluble in their monomers. Furthermore, isothermal fronts propagate
at the order of 1 cm/day and only for total distances of about 1 cm [188].
Thermal-FP, which can be applied to the widest range of materials, is based on the
coupling of thermal transport and the Arrhenius dependence of the reaction rate of an
exothermic polymerization [43]. The exothermic reaction causes a self-propagating front
that separates the cured polymer and the uncured liquid monomer (Figure 32).
In order to sustain a traveling front, the heat produced by the exothermic reaction
must exceed the heat that is lost through the reaction vessel. Furthermore, to prevent
the formation of bubbles, the monomer should have a higher boiling point than the front
temperature. Thermal frontal polymerization is the most widely studied mechanism,
Polymers 2022, 14, 2074 46 of 58

having the widest range of materials to be used [188]. A comprehensive and accessible
review of isothermal frontal polymerization and thermal frontal polymerization is given by
Pojman et al. [43,188,191,192].

Figure 32. Schematic representation of thermal frontal polymerizations (TFPs) requiring an external
stimulus to start an exothermic reaction wave.

Photofrontal polymerization is a distinct mode of polymerization from IFP and TFP,

which are autocatalytic frontal polymerization reaction processes [193]. It is based on the
photobleaching effect and requires the continuous flux of radiation, usually UV light, to
create a propagating wavefront of network formation [45] (Figure 33).

Figure 33. Schematic representation of photofrontal polymerizations showing the photobleaching

effect during light exposure.

According to Rytov et al. [194], the following conditions must be fulfilled for a photo-
chemical reaction to proceed as a typical frontal process: high light absorbance, photoin-
duced bleaching and restricted mass transfer. Photobleaching is based on the effect that
the light absorption of the photoinitiator decay products (photolysis products) is lower
than the light absorption of the original photoinitiator molecule, thereby allowing more
light to pass through the system and allowing the polymerization front to move steadily
towards deeper layers [18]. Examples of photobleaching initiators include benzoin ethyl
ether (BEE), solutions of acyl and biacyl phosphine oxides and substituted titanocenes [195].
The major disadvantage of photo-FP is the need for continuous exposure of the reaction
mixture to light radiation, where the rate and the degree of conversion depend on the given
light intensity. However, as opposed to isothermal and thermal FP, the polymerization can
be stopped at any time by simply “turning off the light” and can be reactivated when the
light source is started [196].
Recently, mathematical modeling and numerical simulation of frontal polymerization
have been gaining more and more attention as they are proven to be a major contribution
to the investigation of the various parameters that influence the dynamic phenomenon
Polymers 2022, 14, 2074 47 of 58

of frontal polymerization and can help with process optimization [197]. Theoretical ap-
proaches to modeling the kinetics of photoinitiated frontal polymerization can be for
instance reviewed in publications from Hayki et al. [18], Ivanov [118], Miller et al. [195],
Cabral et al. [193], and Decker [1].
The challenges of frontal polymerization include the curing of composites and the
fabrication of small components [188]. According to Robertson et al. [198], especially the
fabrication of small components is challenging, as much of the heat of polymerization
is lost to the environment through air or tooling surfaces. Frontal curing of composites
is challenging since a high-volume fraction of fibers is required to produce composite
materials with good mechanical properties. The corresponding reduction in resin con-
tent consequently reduces the exothermic energy density available for polymerization.
Furthermore, the preparation of composites leads to challenges regarding the increased
thermal conductivity, compared with the unfilled polymers, and the associated energy
losses [174]. In 2021, Hirner et al. [199] investigated UV-induced frontal polymerizations
for the preparation of gradient magnetic composites and experimentally quantified the
heat losses during the cationic polymerization reactions of an unfilled and filled bisphenol
A diglycidyl ether. As filler, magnetite (Fe3 O4 ) nanoparticles were used. The experimental
results were compared to the results gained by modeling and simulation according to
the finite element method. The results revealed that the epoxy resins filled with Fe3 O4
nanoparticles show an increase in the thermal conductivity by a factor of 2.5 compared to
the unfilled samples. The significant differences in thermal conductivity of the filled and
unfilled resin were also clearly visible in the simulations. During frontal polymerization,
the epoxy formulations with Fe3 O4 nanoparticles showed a faster and more homogeneous
heat propagation zone compared to the unfilled samples.

7. Discussion
Photopolymerization, in which the initiation of a chemical-physical reaction occurs
by the exposure of photosensitive monomers to a high-intensity light source, has become
a well-accepted technology for manufacturing polymers and has found a large variety of
industrial applications. Photoinitiated polymerizations provide significant advantages over
thermal-initiated polymerizations, including fast and controllable reaction rates as well as
spatial and temporal control over the formation of the material.
In principle, three different types of polymerization, namely cationic, anionic and
free-radical polymerization can be performed. These polymerization reactions are often
initiated by thermal input, often requiring high-temperature conditions and solvents. With
the development of better photoinitiators, photoinduced polymerizations can be used as
more efficient polymerization techniques in terms of temperature conditions and solvent-
free polymerizations. This improvement has been achieved mainly due to the development
of photoacid generators for cationic polymerizations and in free radical polymerizations
due to the development of type I and type II photoinitiator.
The modeling of the curing process can be approached by so-called energetic mod-
els or kinetic models (either phenomenologically or mechanistically). Energetic models
play a minor role in modeling curing kinetics because, unlike kinetic models, they do
not provide information about the degree of cure, which is essential for predicting the
mechanical properties of the polymer. Phenomenological models, which are based on
empirical or semi-empirical rate laws, are widely employed because they are simple and
require a limited number of parameters. However, results obtained from phenomenologi-
cal models cannot be extrapolated to new initial compositions or curing conditions [73].
Another limitation of phenomenological models is their inability to predict post-curing
operations due to diffusion-controlled effects after vitrification [75]. In contrast, mecha-
nistic models are obtained from the balance of chemical species involved in the reaction.
Therefore, mechanistic models offer more flexibility to changes in formulation or curing
conditions, as they account for the individual species concentrations. Mechanistic models
consider the complete scheme of consecutive and competitive reactions (initiation, prop-
Polymers 2022, 14, 2074 48 of 58

agation, termination) which take place during curing, hence providing better prediction
and interpretation. The principal disadvantage of mechanistic models is the need for
an accurate description of all species and reactions involved in the system which is not
trivial due to the complex nature of curing reactions, for both free-radical and cationic
photopolymerizations [75]. Corcione et al. [137] state that cationic photopolymerizations
are even more complex compared to free-radical photopolymerizations as they are strongly
affected by the resin formulation. Hence, in contrast to free-radical photopolymerizations,
there are few mechanistic models for photoinitiated cationic photopolymerization [29]. To
conclude, the advantages and disadvantages of using mechanistic or phenomenological
models to obtain a simple and reliable kinetic model describing the reactions that take place
during photopolymerization must be carefully weighed for the particular application.
Section 3 presented different approaches, both mechanistic and phenomenological, to
model photoinduced free-radical and cationic polymerization. Free-radical polymerization
refers to the process of forming a polymer material via the addition of free radicals whereas,
in ionic polymerization, the polymer material is formed using ionic chemical species as
initial reactants.
Many researchers have realized detailed studies on the mechanistic modeling of pho-
toinduced free-radical polymerization, see Section 3.1.1. From these approaches, one can
draw the conclusion that a complete mechanistic model should contain a very large number
of differential equations and requires many kinetic parameters that reflect the curing be-
havior with sufficient accuracy. Approaches reported in the literature distinguish between
pointwise mechanistic and spatially dependent mechanistic models. Unlike pointwise
mechanistic models, spatial mechanistic models include the description of the spatial dis-
tribution of reactants inside the continuum body during the process. Consequently, for
spatially dependent mechanistic models, the differential problem directly provides the
evolution of the degree of cure in space and time. According to Christmann et al. [102],
kinetic models to study the complex free-radical photopolymerization mechanism and its
related effects must consider simultaneously all the termination pathways (bimolecular
termination, primary radical termination, and radical trapping) and must not neglect the
evolution of terminations along the polymerization reaction. Furthermore, the termination
kinetic constant should incorporate all different possible mechanisms that control termi-
nation: translational diffusion, segmental diffusion, reaction-diffusion, and chain-length
dependent termination [19]. The authors state that all main kinetic constants (propagation,
termination) should be modeled considering their progressive diffusional control. Effects
such as auto-acceleration and auto-deceleration can only be considered when the rate
constants for propagation and termination are modeled as a function of conversion. How-
ever, Wu et al. [82] state that the most sensitive to viscosity increase is the constant of the
termination rate. In their proposed model, Wu et al. [82] consider non-constant propagation
and termination rates with increasing conversion to account for the decreasing molecular
mobility in the reaction medium as polymerization proceeds. The model considers termi-
nation by radical combination but neglects chain-length dependency, effects of polymer
heterogeneity, and radical trapping. A chain-length dependent termination kinetic constant,
which assumes that radicals diffuse and terminate according to their chain length, was
proposed by Bowman et al. [19]. However, the model does not consider a limiting radical
chain length to determine if the radical is “trapped” in the gel and is no longer capable of
diffusion-limited termination. Wen et al. [109] note that kinetic models that ignore radical
trapping fail to predict important aspects of experimental investigation. Therefore, the
authors model the rate constant for radical trapping as a simple function of free volume,
following the model of Anseth et al. [38]. The free volume theory of Anseth et al. [38] is also
the basis for models proposed for instance by Anastasio et al. [86] and Bowman et al. [19].
The kinetic model described by Anastasio and his co-authors considers the recombination
(“trapping”) process by reducing the initiator efficiency.
Another important factor is the modeling of the light intensity which is the driving
factor for the formation of free radicals. Some models proposed in the literature assume
Polymers 2022, 14, 2074 49 of 58

that the light intensity follows a conventional Beer–Lambert law which describes the light
propagation through a homogeneous medium without internal sources of scattering. There-
fore, these approaches only consider the light intensity variation with changes in the spatial
position of the sample. The abovementioned approach implies the oversimplified assump-
tion that the photolysis product becomes completely transparent after polymerization. In
contrast, Lin et al. [112] consider a non-uniform distribution of the photoinitiator (depletion
of the photoinitiator concentration) as the UV-light might still be absorbed by the photolysis
product, besides the absorption of the monomer.
A critical aspect of modeling the curing kinetics of photoinitiated free-radical polymer-
ization concerns oxygen diffusion effects. Cationic photopolymerizations have the distinct
advantage that they lack sensitivity towards atmospheric oxygen, whereas the loss of radicals,
known as oxygen inhibition, is pervasive in free-radical photopolymerizations [22]. Oxygen
in the reaction volume acts as a “radical scavenger”, i.e., it reduces the efficiency of initiation
and generally leads to significant retardation or even inhibition of the polymerization [33].
Therefore, considering the inhibitory effect of oxygen on free-radical photopolymerization
plays an important role in correctly modeling curing kinetics and is considered in most kinetic
models presented in the literature [22,24,34,82,88,105,115,119,160].
O’Brien et al. [97] pointed out that for more complex polymerization systems including
additional factors such as heat generation, heat transfer and mass transfer is necessary. Heat
effects play an important role, as the kinetic constants, as well as the diffusion coefficients,
are a function of temperature. Therefore, the model proposed by O’Brien et al. [97] includes
not only the temporal and spatial distribution of species concentration, temperature and
light intensity but also heat and heat-transfer effects.
Due to the complexity of cationic polymerization reactions, in contrast to free-radical poly-
merization, there are view mechanistic models for photoinitiated cationic polymerization [29].
A mechanistic mathematical model to characterize cationic photopolymerization reactions
as a function of temperature and exposure time, proposed by Nelson et al. [135], was used
as the basis for more elaborate models which also consider the dependency of the rate of
conversion on the light intensity, for instance, models proposed by Corcione et al. [137] and
Pantiru et al. [138].
According to Wu et al. [82], in contrast to mechanistic models, in phenomenological
models the conversion of the monomer (degree of cure) is the only variable to characterize
the polymerization process. Thus, the variation of other species is ignored. The simplest
and most common analytical form of the function of the degree of cure, namely the n-th
order kinetic reaction model, cannot be applied to modeling photopolymerization reac-
tions. The reason lies in the fact, that the n-th order kinetic model predicts a maximum
of the reaction rate at the start of the reaction (t = 0), whereas photopolymerization reac-
tions show a maximum value of the reaction rate at any point, rather than the reaction
start point [140]. Therefore, the majority of phenomenological models used to describe
photopolymerization reactions are based on the so-called autocatalytic model, proposed
by Kamal et al. [120], assuming a reaction rate that is initially equal to zero, showing a
maximum value at intermediate conversion. Simple models, proposed for instance by
Maffezzoli et al. [127], assuming isothermal kinetic behavior, were improved by describing
the maximum attainable curing degree as a function of absolute temperature as well as
location- and time-dependent light intensity. The influence of diffusion-controlled effects on
the curing reactions were considered in phenomenological models presented, for instance,
by Kim et al. [29] and Bartolo [75] who presented a coupled photothermal phenomeno-
logical approach to correctly model the physical and chemical changes occurring in the
bulk and in the surroundings of the material exposed to UV light, as well as the rates at
which these changes occur by describing the temperature field in the region exposed to UV
radiation by the heat conduction equation.
Polymer composites have many advantages over traditional polymeric materials in
terms of mechanical properties and are therefore widely used in automotive, aerospace
and biomedical applications. However, polymer composites are often obtained by heat-
Polymers 2022, 14, 2074 50 of 58

curing processes that require high energy input and often long curing times [200]. These
disadvantages can be avoided by using the frontal polymerization technique, in which
the reaction process is mainly governed by the chemical and physical properties of the
reacting system. Frontal polymerization (FP) is a process in which a localized reaction
zone (the so-called polymerization front), once initiated by an external stimulus, acts
as a switch for the crosslinking of the reactants in adjacent areas. Depending on the
external stimuli that are employed to initiate and trigger the reaction front, the following
types of frontal polymerization are distinguished: isothermal frontal polymerization (IFP),
thermal frontal polymerization (TFP), and frontal photopolymerization (FPP). According to
Frulloni et al. [197], recently, mathematical modeling and numerical simulation of frontal
polymerization have been gaining more and more attention as they are proven to be a major
contribution to the investigation of the various parameters that influence the dynamic
phenomenon of frontal polymerization and can help with process optimization.

8. Concluding Remarks
In this paper, we have presented an extensive review of processes and models related
to photopolymerization reactions. Besides briefly discussing the materials and curing
chemistry, the main goal of the review has been to provide a comprehensive overview of
analytical models and numerical approaches to accurately describe the cure kinetics and
mechanisms for curing behavior simulations of such ultrafast crosslinking polymerization
reactions. The correct modeling of diffusion-controlled phenomena as well as the modeling
of oxygen diffusion-reaction in free-radical systems, in order to study material property
changes during the photopolymerization process, faces both theoretical and numerical
challenges and different approaches are used to face these obstacles. Besides briefly dis-
cussing the main characteristics of different modeling approaches, the main goal of the
review has been to provide a critical and comprehensive overview of the similarities and
differences between approaches to describe the photothermal kinetic process of the chemi-
cal conversion from monomers in a liquid state to polymeric chains. The review attempts
to offer an overall view of the limitations of current modeling, ranging from mechanistic to
phenomenological models based on either pointwise or spatial approaches, and suggest
possible improvements. Furthermore, approaches regarding the implementation of kinetic
models through numerical simulations, proposed in the literature, are presented.

Author Contributions: Conceptualization M.L., S.H., F.W. and P.F.; Methodology, M.L. and S.H.;
Project Administration, F.W.; Supervision, F.W. and P.F.; Writing—original draft, M.L. and S.H.;
Writing—review & editing, M.L., S.H. and F.W. All authors have read and agreed to the published
version of the manuscript.
Funding: The research work was performed within the COMET-project ’Photostructurable Encapsu-
lation Molds and Magnetic Composites’ (project-no.: VII-1.S2) at the Polymer Competence Center
Leoben GmbH (PCCL, Austria) within the framework of the COMET-program of the Federal Ministry
for Climate Action, Environment, Energy, Mobility, Innovation and Technology and the Federal
Ministry for Digital and Economic Affairs with contributions by the Graz University of Technology.
The PCCL is funded by the Austrian Government and the State Governments of Styria, Lower Austria
and Upper Austria.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: May be found from the cited references.
Conflicts of Interest: The authors declare no conflict of interest.
Polymers 2022, 14, 2074 51 of 58

Abbreviations
AIBN Azobisisbutyronitrile
CADE Cycloaliphatic Diepoxide
CLDT Chain Length Dependent Termination
CLIP Continuous Liquid Interface Production
DCPD Dicyclopentadiene
DEGDMA Diethylene Glycol Dimethacrylate
DGEBA Diglycidyl Ether of Bisphenol A
DGEBF Diglycidyl Ether of Bisphenol F
DLP Digital Light Processing
DMPA Dimethoxy Phenylacetophenone
DSC Differential Scanning Calorimetry
ECH Epichlorohydrin
ECPL Exposure Controlled Projection Lithography
EVE Ethyl Vinyl Ether
FE Finite Elements
FEA Finite Element Analysis
FP Frontal Polymerization
FPP Frontal Photo Polymerization
FROMP Frontal Ring-Opening Metathesis Polymerization
FTIR Fourier-Transformed Infrared
IBVE Iso-Butyl Vinyl Ether
IFP Isothermal Frontal Polymerization
HBVE Hydroxylbutyl Vinyl Ether
MMA Methyl Methacrylate
PAG Photoacid Generator
PEGDA Poly(ethylene glycol) Diacrylate
PRT Primary Radical Termination
RICFP Radical-Induced Frontal Polymerization
RT-FTIR Real-Time Fourier-Transformed Infrared
TFP Thermal Frontal Polymerization
TPO Trimethylbenzoyl diphenylphosphine oxide
UV Ultraviolet
SLA Stereolithography Applications
WLF Williams-Lendel-Ferry

References
1. Decker, C. The use of UV irradiation in polymerization. Polym. Int. 1998, 45, 133–141. [CrossRef]
2. Peiffer, R.W. Applications of Photopolymer Technology; ACS Symposium Series; American Chemical Society: Washington, DC, USA,
1997; pp. 1–14; ISBN 0097-6156.
3. Bednarczyk, P.; Nowak, M.; Mozelewska, K.; Czech, Z. Photocurable Coatings Based on Bio-Renewable Oligomers and Monomers.
Materials 2021, 14, 7731. [CrossRef] [PubMed]
4. Tytgat, L.; Markovic, M.; Qazi, T.H.; Vagenende, M.; Bray, F.; Martins, J.C.; Rolando, C.; Thienpont, H.; Ottevaere, H.;
Ovsianikov, A.; et al. Photo-crosslinkable recombinant collagen mimics for tissue engineering applications. J. Mater. Chem.
B 2019, 7, 3100–3108. [CrossRef] [PubMed]
5. Pereira, R.F.; Bártolo, P.J. 3D Photo-Fabrication for Tissue Engineering and Drug Delivery. Engineering 2015, 1, 90–112. [CrossRef]
6. Hutchison, J.B.; Haraldsson, K.T.; Good, B.T.; Sebra, R.P.; Luo, N.; Anseth, K.S.; Bowman, C.N. Robust polymer microfluidic
device fabrication via contact liquid photolithographic polymerization (CLiPP). Lab Chip 2004, 4, 658–662. [CrossRef]
7. Grant Willson, C.; Trinque, B.C. The Evolution of Materials for the Photolithographic Process. J. Photopol. Sci. Technol. 2003, 16,
621–627. [CrossRef]
8. Rogers, J.A.; Nuzzo, R.G. Recent progress in soft lithography. Mater. Today 2005, 8, 50–56. [CrossRef]
9. Del Barrio, J.; Sánchez-Somolinos, C. Light to Shape the Future: From Photolithography to 4D Printing. Adv. Opt. Mater. 2019, 7,
1900598. [CrossRef]
10. Haraldsson, K.T.; Hutchison, J.B.; Sebra, R.P.; Good, B.T.; Anseth, K.S.; Bowman, C.N. 3D polymeric microfluidic device fabrication
via contact liquid photolithographic polymerization (CLiPP). Sens. Actuators B Chem. 2006, 113, 454–460. [CrossRef]
11. Xu, Y.; Qi, F.; Mao, H.; Li, S.; Zhu, Y.; Gong, J.; Wang, L.; Malmstadt, N.; Chen, Y. In-situ transfer vat photopolymerization for
transparent microfluidic device fabrication. Nat. Commun. 2022, 13, 918. [CrossRef]
Polymers 2022, 14, 2074 52 of 58

12. Liska, R.; Schuster, M.; Inführ, R.; Turecek, C.; Fritscher, C.; Seidl, B.; Schmidt, V.; Kuna, L.; Haase, A.; Varga, F.; et al.
Photopolymers for rapid prototyping. J. Coat. Technol. Res. 2007, 4, 505–510. [CrossRef]
13. Rocheva, V.V.; Koroleva, A.V.; Savelyev, A.G.; Khaydukov, K.V.; Generalova, A.N.; Nechaev, A.V.; Guller, A.E.; Semchishen, V.A.;
Chichkov, B.N.; Khaydukov, E.V. High-resolution 3D photopolymerization assisted by upconversion nanoparticles for rapid
prototyping applications. Sci. Rep. 2018, 8, 3663. [CrossRef] [PubMed]
14. Klikovits, N.; Sinawehl, L.; Knaack, P.; Koch, T.; Stampfl, J.; Gorsche, C.; Liska, R. UV-Induced Cationic Ring-Opening Polymer-
ization of 2-Oxazolines for Hot Lithography. ACS Macro Lett. 2020, 9, 546–551. [CrossRef]
15. Pagac, M.; Hajnys, J.; Ma, Q.-P.; Jancar, L.; Jansa, J.; Stefek, P.; Mesicek, J. A Review of Vat Photopolymerization Technology:
Materials, Applications, Challenges, and Future Trends of 3D Printing. Polymers 2021, 13, 598. [CrossRef] [PubMed]
16. Ligon, S.C.; Liska, R.; Stampfl, J.; Gurr, M.; Mülhaupt, R. Polymers for 3D Printing and Customized Additive Manufacturing.
Chem. Rev. 2017, 117, 10212–10290. [CrossRef] [PubMed]
17. Shusteff, M.; Browar, A.E.M.; Kelly, B.E.; Henriksson, J.; Weisgraber, T.H.; Panas, R.M.; Fang, N.X.; Spadaccini, C.M. One-step
volumetric additive manufacturing of complex polymer structures. Sci. Adv. 2017, 3, eaao5496. [CrossRef]
18. Hayki, N.; Lecamp, L.; Désilles, N.; Lebaudy, P. Kinetic Study of Photoinitiated Frontal Polymerization. Influence of UV Light
Intensity Variations on the Conversion Profiles. Macromolecules 2010, 43, 177–184. [CrossRef]
19. Bowman, C.N.; Kloxin, C.J. Toward an enhanced understanding and implementation of photopolymerization reactions. AIChE J.
2008, 54, 2775–2795. [CrossRef]
20. Rusu, M.C.; Block, C.; van Assche, G.; van Mele, B. Influence of temperature and UV intensity on photo-polymerization reaction
studied by photo-DSC. J. Therm. Anal. Calorim. 2012, 110, 287–294. [CrossRef]
21. Kaur, M.; Srivastava, A.K. PHOTOPOLYMERIZATION: A REVIEW. J. Macromol. Sci. Part C Polym. Rev. 2002, 42, 481–512.
[CrossRef]
22. Decker, C.; Jenkins, A.D. Kinetic approach of oxygen inhibition in ultraviolet- and laser-induced polymerizations. Macromolecules
1985, 18, 1241–1244. [CrossRef]
23. Lovestead, T.M.; O’Brien, A.K.; Bowman, C.N. Models of multivinyl free radical photopolymerization kinetics. J. Photochem.
Photobiol. A Chem. 2003, 159, 135–143. [CrossRef]
24. Studer, K.; Decker, C.; Beck, E.; Schwalm, R. Overcoming oxygen inhibition in UV-curing of acrylate coatings by carbon dioxide
inerting, Part I. Prog. Org. Coat. 2003, 48, 92–100. [CrossRef]
25. Decker, C.; Moussa, K. Kinetic study of the cationic photopolymerization of epoxy monomers. J. Polym. Sci. Part A Polym. Chem.
1990, 28, 3429–3443. [CrossRef]
26. Ficek, B.A. The Potential of Cationic Photopolymerization’s Long Lived Active Centers. Master’s Thesis, University of Iowa,
Iowa, IA, USA, 2008.
27. Malik, M.S.; Schlögl, S.; Wolfahrt, M.; Sangermano, M. Review on UV-Induced Cationic Frontal Polymerization of Epoxy
Monomers. Polymers 2020, 12, 2146. [CrossRef] [PubMed]
28. Esposito Corcione, C.; Greco, A.; Maffezzoli, A. Photopolymerization Kinetics of an Epoxy Based Resin for Stereolithography—
Calorimetric Analysis. J. Therm. Anal. Calorim. 2003, 72, 687–693. [CrossRef]
29. Kim, Y.-M.; Kostanski, L.K.; MacGregor, J.F. Kinetic studies of cationic photopolymerizations of cycloaliphatic epoxide, tri-
ethyleneglycol methyl vinyl ether, and cyclohexene oxide. Polym. Eng. Sci. 2005, 45, 1546–1555. [CrossRef]
30. Lin, J.T.; Lalevee, J.; Cheng, D.C. A Critical Review for Synergic Kinetics and Strategies for Enhanced Photopolymerizations for
3D-Printing and Additive Manufacturing. Polymers 2021, 13, 2325. [CrossRef]
31. De Beer, M.P.; Van Der Laan, H.L.; Cole, M.A.; Whelan, R.J.; Burns, M.A.; Scott, T.F. Rapid, continuous additive manufacturing by
volumetric polymerization inhibition patterning. Sci. Adv. 2019, 5, eaau8723. [CrossRef]
32. Van Der Laan, H.L.; Burns, M.A.; Scott, T.F. Volumetric Photopolymerization Confinement through Dual-Wavelength Photoinitia-
tion and Photoinhibition. ACS Macro Lett. 2019, 8, 899–904. [CrossRef]
33. Andrzejewska, E.; Lindén, L.-Å.; Rabek, J.F. Modelling the Kinetics of Photoinitiated Polymerization of Di(meth)acrylates. Polym.
Int. 1997, 42, 179–187. [CrossRef]
34. Decker, C.; Moussa, K. Kinetic investigation of photopolymerizations induced by laser beams. Die Makromol. Chem. 1990, 191,
963–979. [CrossRef]
35. Decker, C. Ultra-fast polymerization of epoxy-acrylate resins by pulsed laser irradiation. J. Polym. Sci. Polym. Chem. Ed. 1983, 21,
2451–2461. [CrossRef]
36. Lin, Y.; Stansbury, J.W. Kinetics studies of hybrid structure formation by controlled photopolymerization. Polymer 2003, 44,
4781–4789. [CrossRef]
37. O’Shaughnessy, B.; Yu, J. Autoacceleration in free radical polymerization. Phys. Rev. Lett. 1994, 73, 1723–1726. [CrossRef]
[PubMed]
38. Anseth, K.S.; Bowman, C.N. Reaction Diffusion Enhanced Termination in Polymerizations of Multifunctional Monomers. Polym.
React. Eng. 1993, 1, 499–520. [CrossRef]
39. Brighenti, R.; Cosma, M.P.; Marsavina, L.; Spagnoli, A.; Terzano, M. Laser-based additively manufactured polymers: A review on
processes and mechanical models. J. Mater. Sci. 2021, 56, 961–998. [CrossRef]
40. Davtyan, S.P.; Berlin, A.A.; Tonoyan, A.O. Advances and problems of frontal polymerization processes. Ref. J. Chem. 2011, 1, 56–92.
[CrossRef]
Polymers 2022, 14, 2074 53 of 58

41. Davtyan, S.P.; Avetisyan, A.S.; Berlin, A.A.; Tonoyan, A.O. Synthesis and properties of particle-filled and intercalated polymer
nanocomposites. Ref. J. Chem. 2013, 3, 1–51. [CrossRef]
42. Cui, Y.; Yang, J.; Zhan, Y.; Zeng, Z.; Chen, Y. In situ fabrication of polyacrylate/nanozirconia hybrid material via frontal
photopolymerization. Colloid Polym. Sci. 2008, 286, 97–106. [CrossRef]
43. Pojman, J.A. Mathematical modeling of frontal polymerization. Math. Model. Nat. Phenom. 2019, 14, 604. [CrossRef]
44. Garra, P.; Dietlin, C.; Morlet-Savary, F.; Dumur, F.; Gigmes, D.; Fouassier, J.-P.; Lalevée, J. Photopolymerization processes of thick
films and in shadow areas: A review for the access to composites. Polym. Chem. 2017, 8, 7088–7101. [CrossRef]
45. Ebner, C.; Mitterer, J.; Eigruber, P.; Stieger, S.; Riess, G.; Kern, W. Ultra-High Through-Cure of (Meth)Acrylate Copolymers via
Photofrontal Polymerization. Polymers 2020, 12, 1291. [CrossRef] [PubMed]
46. Carion, P.; Ibrahim, A.; Allonas, X.; Croutxé-Barghorn, C.; L’Hostis, G. Frontal free-radical photopolymerization of thick samples:
Applications to LED-induced fiber-reinforced polymers. J. Polym. Sci. Part A Polym. Chem. 2019, 57, 898–906. [CrossRef]
47. Moad, G.; Solomon, D.H. The Chemistry of Radical Polymerization, 2nd ed.; Elsevier: Boston, MA, USA, 2006; ISBN 9780080442884.
48. Matyjaszewski, K.; Davis, T.P. Handbook of Radical Polymerization; Wiley-Interscience: Hoboken, NJ, USA, 2003; ISBN 9780471461579.
49. Guerrero-Santos, R.; Saldívar-Guerra, E.; Bonilla-Cruz, J. Free Radical Polymerization. In Handbook of Polymer Synthesis, Characteri-
zation, and Processing; Saldívar-Guerra, E., Vivaldo-Lima, E., Eds.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2013; pp. 65–83.
ISBN 9781118480793.
50. Elias, H.-G. Makromoleküle: Chemische Struktur und Synthesen-Sechste; John Wiley & Sons: Hoboken, NJ, USA, 2009;
ISBN 3-527-29872-X.
51. Odian, G.G. Principles of Polymerization, 4th ed.; John Wiley & Sons: Hoboken, NJ, USA, 2010; ISBN 0-471-27400-3.
52. Fouassier, J.; Allonas, X.; Burget, D. Photopolymerization reactions under visible lights: Principle, mechanisms and examples of
applications. Prog. Org. Coat. 2003, 47, 16–36. [CrossRef]
53. Elian, C.; Brezová, V.; Sautrot-Ba, P.; Breza, M.; Versace, D.-L. Lawsone Derivatives as Efficient Photopolymerizable Initiators for
Free-Radical, Cationic Photopolymerizations, and Thiol-Ene Reactions. Polymers 2021, 13, 2015. [CrossRef]
54. Daglen, B.C.; Tyler, D.R. Photodegradable plastics: End-of-life design principles. Green Chem. Lett. Rev. 2010, 3, 69–82. [CrossRef]
55. Yagci, Y.; Jockusch, S.; Turro, N.J. Photoinitiated Polymerization: Advances, Challenges, and Opportunities. Macromolecules 2010,
43, 6245–6260. [CrossRef]
56. Ibrahim, A.; Stefano, L.; Tarzi, O.; Tar, H.; Ley, C.; Allonas, X. High-performance photoinitiating systems for free radical
photopolymerization. Application to holographic recording. Photochem. Photobiol. 2013, 89, 1283–1290. [CrossRef]
57. Schlüter, A.-D. Synthesis of Polymers; John Wiley & Sons: Hoboken, NJ, USA, 1998; ISBN 9783527294497.
58. Sangermano, M. Advances in cationic photopolymerization. Pure Appl. Chem. 2012, 84, 2089–2101. [CrossRef]
59. Crivello, J.V. Design of Photoacid Generating Systems. J. Photopol. Sci. Technol. 2009, 22, 575–582. [CrossRef]
60. Crivello, J.V.; Lee, J.L. Alkoxy-substituted diaryliodonium salt cationic photoinitiators. J. Polym. Sci. Part A Polym. Chem. 1989, 27,
3951–3968. [CrossRef]
61. Akhtar, S.R.; Crivello, J.V.; Lee, J.L. Synthesis of aryl-substituted sulfonium salts by the phosphorus pentoxide-methanesulfonic
acid promoted condensation of sulfoxides with aromatic compounds. J. Org. Chem. 1990, 55, 4222–4225. [CrossRef]
62. Vitale, A.; Sangermano, M.; Bongiovanni, R.; Burtscher, P.; Moszner, N. Visible Light Curable Restorative Composites for Dental
Applications Based on Epoxy Monomer. Materials 2014, 7, 554–562. [CrossRef]
63. Barner-Kowollik, C. Acrylate free radical polymerization: From mechanism to polymer design. Macromol. Rapid Commun. 2009,
30, 1961–1963. [CrossRef]
64. Tripathy, R.; Crivello, J.V.; Faust, R. Photoinitiated polymerization of acrylate, methacrylate, and vinyl ether end-functional
polyisobutylene macromonomers. J. Polym. Sci. Part A Polym. Chem. 2013, 51, 305–317. [CrossRef]
65. Chen, J.; Chu, N.; Zhao, M.; Jin, F.-L.; Park, S.-J. Synthesis and application of thermal latent initiators of epoxy resins: A review. J.
Appl. Polym. Sci. 2020, 137, 49592. [CrossRef]
66. Brydson, J.A. Epoxide Resins. In Plastics Materials; Elsevier: Amsterdam, The Netherlands, 1999; pp. 744–777; ISBN 9780750641326.
67. McGrath, J.E. Ring-Opening Polymerization: Introduction. In Ring-Opening Polymerization; McGrath, J.E., Ed.; American Chemical
Society: Washington, DC, USA, 1985; pp. 1–22; ISBN 9780841209268.
68. Sanda, F.; Endo, T. Radical ring-opening polymerization. J. Polym. Sci. Part A Polym. Chem. 2001, 39, 265–276. [CrossRef]
69. Higashimura, T.; Sawamoto, M. Carbocationic Polymerization: Vinyl Ethers. In Comprehensive Polymer Science and Supplements;
Elsevier: Amsterdam, The Netherlands, 1989; pp. 673–696; ISBN 9780080967011.
70. Kammiyada, H.; Ouchi, M.; Sawamoto, M. A Study on Physical Properties of Cyclic Poly(vinyl ether)s Synthesized via Ring-
Expansion Cationic Polymerization. Macromolecules 2017, 50, 841–848. [CrossRef]
71. Yamaguchi, K.; Nakamoto, T. Micro Fabrication by UV Laser Photopolymerization; Nagoya University: Nagoya, Japan, 1998.
72. Tomeckova, V.; Halloran, J.W. Predictive models for the photopolymerization of ceramic suspensions. J. Eur. Ceram. Soc. 2010, 30,
2833–2840. [CrossRef]
73. Blanco, M.; Corcuera, M.A.; Riccardi, C.C.; Mondragon, I. Mechanistic kinetic model of an epoxy resin cured with a mixture of
amines of different functionalities. Polymer 2005, 46, 7989–8000. [CrossRef]
74. Matias, J.M.; Bartolo, P.J.; Pontes, A.V. Modeling and simulation of photofabrication processes using unsaturated polyester resins.
J. Appl. Polym. Sci. 2009, 114, 3673–3685. [CrossRef]
75. Da Silva Bartolo, P.J. Photo-curing modelling: Direct irradiation. Int. J. Adv. Manuf. Technol. 2007, 32, 480–491. [CrossRef]
Polymers 2022, 14, 2074 54 of 58

76. Han, C.D.; Lee, D.-S. Analysis of the curing behavior of unsaturated polyester resins using the approach of free radical
polymerization. J. Appl. Polym. Sci. 1987, 33, 2859–2876. [CrossRef]
77. Nogueira, T.R.; Goncalves, M.C.; Ferrareso Lona, L.M.; Vivaldo-Lima, E.; McManus, N.; Penlidis, A. Effect of initiator type and
concentration on polymerization rate and molecular weight in the bimolecular nitroxide-mediated radical polymerization of
styrene. Adv. Polym. Technol. J. Polym. Processing Inst. 2010, 29, 11–19. [CrossRef]
78. Gillespie, D.T. Exact stochastic simulation of coupled chemical reactions. J. Phys. Chem. 1977, 81, 2340–2361. [CrossRef]
79. Altun-Çiftçioğlu, G.A.; Ersoy-Meriçboyu, A.; Henderson, C.L. Stochastic modeling and simulation of photopolymerization
process. Polym. Eng. Sci. 2011, 51, 1710–1719. [CrossRef]
80. Boddapati, A. Modeling Cure Depth during Photopolymerization of Multifunctional Acrylates. Ph.D Thesis, Georgia Institute of
Technology, Atlanta, GA, USA, 2010.
81. Goodner, M.D.; Bowman, C.N. Modeling Primary Radical Termination and Its Effects on Autoacceleration in Photopolymerization
Kinetics. Macromolecules 1999, 32, 6552–6559. [CrossRef]
82. Wu, J.; Zhao, Z.; Hamel, C.M.; Mu, X.; Kuang, X.; Guo, Z.; Qi, H.J. Evolution of material properties during free radical
photopolymerization. J. Mech. Phys. Solids 2018, 112, 25–49. [CrossRef]
83. Zhang, Y.; Kranbuehl, D.E.; Sautereau, H.; Seytre, G.; Dupuy, J. Modeling and Measuring UV Cure Kinetics of Thick Dimethacry-
late Samples. Macromolecules 2009, 42, 203–210. [CrossRef]
84. Achilias, D.S. A Review of Modeling of Diffusion Controlled Polymerization Reactions. Macromol. Theory Simul. 2007, 16, 319–347.
[CrossRef]
85. Batch, G.L.; Macosko, C.W. Kinetic model for crosslinking free radical polymerization including diffusion limitations. J. Appl.
Polym. Sci. 1992, 44, 1711–1729. [CrossRef]
86. Anastasio, R.; Peerbooms, W.; Cardinaels, R.; van Breemen, L.C.A. Characterization of Ultraviolet-Cured Methacrylate Networks:
From Photopolymerization to Ultimate Mechanical Properties. Macromolecules 2019, 52, 9220–9231. [CrossRef]
87. Long, K.N.; Scott, T.F.; Jerry Qi, H.; Bowman, C.N.; Dunn, M.L. Photomechanics of light-activated polymers. J. Mech. Phys. Solids
2009, 57, 1103–1121. [CrossRef]
88. O’Brien, A.K.; Bowman, C.N. Modeling the Effect of Oxygen on Photopolymerization Kinetics. Macromol. Theory Simul. 2006, 15,
176–182. [CrossRef]
89. Brighenti, R.; Cosma, M.P.; Marsavina, L.; Spagnoli, A.; Terzano, M. Multiphysics modelling of the mechanical properties in
polymers obtained via photo-induced polymerization. Int. J. Adv. Manuf. Technol. 2021, 117, 481–499. [CrossRef]
90. Buback, M.; Huckestein, B.; Russell, G.T. Modeling of termination in intermediate and high conversion free radical polymeriza-
tions. Macromol. Chem. Phys. 1994, 195, 539–554. [CrossRef]
91. Buback, M.; Hesse, P.; Hutchinson, R.A.; Kasák, P.; Lacík, I.; Stach, M.; Utz, I. Kinetics and Modeling of Free-Radical Batch
Polymerization of Nonionized Methacrylic Acid in Aqueous Solution. Ind. Eng. Chem. Res. 2008, 47, 8197–8204. [CrossRef]
92. Buback, M. Free-radical polymerization up to high conversion. A general kinetic treatment. Die Makromol. Chem. 1990, 191,
1575–1587. [CrossRef]
93. Dickey, M.D.; Willson, C.G. Kinetic parameters for step and flash imprint lithography photopolymerization. AIChE J. 2006, 52,
777–784. [CrossRef]
94. Gan, S.; Seferis, J.C.; Prime, R.B. A viscoelastic description of the glass transition-conversion relationship for reactive polymers. J.
Therm. Anal. Calorim. 1991, 37, 463–478. [CrossRef]
95. Goodner, M.D.; Bowman, C.N. Development of a comprehensive free radical photopolymerization model incorporating heat and
mass transfer effects in thick films. Chem. Eng. Sci. 2002, 57, 887–900. [CrossRef]
96. Goodner, M.D.; Lee, H.R.; Bowman, C.N. Method for Determining the Kinetic Parameters in Diffusion-Controlled Free-Radical
Homopolymerizations. Ind. Eng. Chem. Res. 1997, 36, 1247–1252. [CrossRef]
97. O’Brien, A.K.; Bowman, C.N. Modeling Thermal and Optical Effects on Photopolymerization Systems. Macromolecules 2003, 36,
7777–7782. [CrossRef]
98. Lovestead, T.M.; Berchtold, K.A.; Bowman, C.N. Modeling the Effects of Chain Length on the Termination Kinetics in Multivinyl
Photopolymerizations. Macromol. Theory Simul. 2002, 11, 729–738. [CrossRef]
99. Bowman, C.N.; Peppas, N.A. Coupling of kinetics and volume relaxation during polymerizations of multiacrylates and multi-
methacrylates. Macromolecules 1991, 24, 1914–1920. [CrossRef]
100. Zhu, S.; Tian, Y.; Hamielec, A.E.; Eaton, D.R. Radical trapping and termination in free-radical polymerization of methyl
methacrylate. Macromolecules 1990, 23, 1144–1150. [CrossRef]
101. Anseth, K.S.; Wang, C.M.; Bowman, C.N. Kinetic evidence of reaction diffusion during the polymerization of multi(meth)acrylate
monomers. Macromolecules 1994, 27, 650–655. [CrossRef]
102. Christmann, J.; Ley, C.; Allonas, X.; Ibrahim, A.; Croutxé-Barghorn, C. Experimental and theoretical investigations of free radical
photopolymerization: Inhibition and termination reactions. Polymer 2019, 160, 254–264. [CrossRef]
103. Ibrahim, A.; Maurin, V.; Ley, C.; Allonas, X.; Croutxe-Barghorn, C.; Jasinski, F. Investigation of termination reactions in free radical
photopolymerization of UV powder formulations. Eur. Polym. J. 2012, 48, 1475–1484. [CrossRef]
104. Gao, Y.; Xu, L.; Zhao, Y.; You, Z.; Guan, Q. 3D printing preview for stereo-lithography based on photopolymerization kinetic
models. Bioact. Mater. 2020, 5, 798–807. [CrossRef] [PubMed]
Polymers 2022, 14, 2074 55 of 58

105. Wang, J.; Zhao, C.; Zhang, Y.; Jariwala, A.; Rosen, D. (Eds.) Process Modeling and In-Situ Monitoring of Polymerization for
Exposure Controlled Projection Lithography (ECPL). In Proceedings of the 28th Annual International Solid Free Form Fabrication
Symposium—An Additive Manufacturing Conference, Austin, TX, USA, 7–9 August 2017.
106. Lee, J.H.; Prud’homme, R.K.; Aksay, I.A. Cure depth in photopolymerization: Experiments and theory. J. Mater. Res. 2001, 16,
3536–3544. [CrossRef]
107. Dickey, M.D.; Burns, R.L.; Kim, E.K.; Johnson, S.C.; Stacey, N.A.; Willson, C.G. Study of the kinetics of step and flash imprint
lithography photopolymerization. AIChE J. 2005, 51, 2547–2555. [CrossRef]
108. Perry, M.F.; Young, G.W. A Mathematical Model for Photopolymerization From a Stationary Laser Light Source. Macromol. Theory
Simul. 2005, 14, 26–39. [CrossRef]
109. Wenand, M.; McCormick, A.V. A Kinetic Model for Radical Trapping in Photopolymerization of Multifunctional Monomers.
Macromolecules 2000, 33, 9247–9254. [CrossRef]
110. Anseth, K.S.; Anderson, K.J.; Bowman, C.N. Radical concentrations, environments, and reactivities during crosslinking polymer-
izations. Macromol. Chem. Phys. 1996, 197, 833–848. [CrossRef]
111. Bueche, F. Physical Properties of Polymers; Interscience: New York, NY, USA, 1962.
112. Lin, J.-T.; Wang, K.-C. Analytic formulas and numerical simulations for the dynamics of thick and non-uniform polymerization
by a UV light. J. Polym. Res. 2016, 23, 53. [CrossRef]
113. Lin, J.-T.; Cheng, D.-C. Modeling the efficacy profiles of UV-light activated corneal collagen crosslinking. PLoS ONE 2017, 12,
e0175002. [CrossRef]
114. Lin, J.-T.; Liu, H.-W.; Chen, K.-T.; Cheng, D.-C. Modeling the Optimal Conditions for Improved Efficacy and Crosslink Depth of
Photo-Initiated Polymerization. Polymers 2019, 11, 217. [CrossRef]
115. Lin, J.-T.; Liu, H.-W.; Chen, K.-T.; Cheng, D.-C. Modeling the Kinetics, Curing Depth, and Efficacy of Radical-Mediated
Photopolymerization: The Role of Oxygen Inhibition, Viscosity, and Dynamic Light Intensity. Front. Chem. 2019, 7, 760. [CrossRef]
[PubMed]
116. Lin, J.-T.; Cheng, D.-C.; Chen, K.-T.; Chiu, Y.-C.; Liu, H.-W. Enhancing UV Photopolymerization by a Red-light Preirradiation:
Kinetics and Modeling Strategies for Reduced Oxygen Inhibition. J. Polym. Sci. 2020, 58, 683–691. [CrossRef]
117. Lin, J.-T.; Lee, Y.-Z.; Lalevee, J.; Kao, C.-H.; Lin, K.-H.; Cheng, D.-C. Modeling the Enhanced Efficacy and Curing Depth of
Photo-Thermal Dual Polymerization in Metal (Fe) Polymer Composites for 3D Printing. Polymers 2022, 14, 1158. [CrossRef]
[PubMed]
118. Ivanov, V.V.; Decker, C. Kinetic study of photoinitiated frontal polymerization. Polym. Int. 2001, 50, 113–118. [CrossRef]
119. Childress, K.K.; Kim, K.; Glugla, D.J.; Musgrave, C.B.; Bowman, C.N.; Stansbury, J.W. Independent Control of Singlet Oxygen and
Radical Generation via Irradiation of a Two-Color Photosensitive Molecule. Macromolecules 2019, 52, 4968–4978. [CrossRef]
120. Kamal, M.R.; Sourour, S. Kinetics and thermal characterization of thermoset cure. Polym. Eng. Sci. 1973, 13, 59–64. [CrossRef]
121. Sourour, S.; Kamal, M.R. Differential scanning calorimetry of epoxy cure: Isothermal cure kinetics. Thermochim. Acta 1976,
14, 41–59. [CrossRef]
122. Harikrishna, R.; Ponrathnam, S.; Tambe, S.S. Reaction kinetics and modeling of photoinitiated cationic polymerization of an
alicyclic based diglycidyl ether. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. At. 2014, 318, 263–268. [CrossRef]
123. Atai, M.; Watts, D.C. A new kinetic model for the photopolymerization shrinkage-strain of dental composites and resin-monomers.
Dent. Mater. 2006, 22, 785–791. [CrossRef]
124. Khudyakov, I.V.; Legg, J.C.; Purvis, M.B.; Overton, B.J. Kinetics of Photopolymerization of Acrylates with Functionality of 1−6.
Ind. Eng. Chem. Res. 1999, 38, 3353–3359. [CrossRef]
125. Cook, W.D. Photopolymerization kinetics of dimethacrylates using the camphorquinone/amine initiator system. Polymer 1992,
33, 600–609. [CrossRef]
126. Ganglani, M.; Carr, S.H.; Torkelson, J.M. Influence of cure via network structure on mechanical properties of a free-radical
polymerizing thermoset. Polymer 2002, 43, 2747–2760. [CrossRef]
127. Maffezzoli, A.; Terzi, R. Effect of irradiation intensity on the isothermal photopolymerization kinetics of acrylic resins for
stereolithography. Thermochim. Acta 1998, 321, 111–121. [CrossRef]
128. Rehbein, T.; Lion, A.; Johlitz, M.; Constantinescu, A. Experimental investigation and modelling of the curing behaviour of
photopolymers. Polym. Test. 2020, 83, 106356. [CrossRef]
129. Bartolo, P.J.d.S. Optical Approaches to Macroscopic and Microscopic Engineering. Ph.D. Thesis, University of Reading, Reading,
UK, 2001.
130. Yang, Y.; Li, L.; Zhao, J. Mechanical property modeling of photosensitive liquid resin in stereolithography additive manufacturing:
Bridging degree of cure with tensile strength and hardness. Mater. Des. 2019, 162, 418–428. [CrossRef]
131. Kenny, J.M.; Trivisano, A. Isothermal and dynamic reaction kinetics of high performance epoxy matrices. Polym. Eng. Sci. 1991,
31, 1426–1433. [CrossRef]
132. Park, I.-K.; Lee, D.-S.; Nam, J.-D. Equivalent processing time analysis of glass transition development in epoxy/carbon fiber
composite systems. J. Appl. Polym. Sci. 2002, 84, 144–154. [CrossRef]
133. Kim, Y.C.; Hong, S.; Sun, H.; Kim, M.G.; Choi, K.; Cho, J.; Choi, H.R.; Koo, J.C.; Moon, H.; Byun, D.; et al. UV-curing kinetics and
performance development of in situ curable 3D printing materials. Eur. Polym. J. 2017, 93, 140–147. [CrossRef]
Polymers 2022, 14, 2074 56 of 58

134. Nelson, E.W.; Jacobs, J.L.; Scranton, A.B.; Anseth, K.S.; Bowman, C.N. Photo-differential scanning calorimetry studies of cationic
polymerizations of divinyl ethers. Polymer 1995, 36, 4651–4656. [CrossRef]
135. Nelson, E.W.; Scranton, A.B. Kinetics of cationic photopolymerizations of divinyl ethers characterized usingin situ Raman
spectroscopy. J. Polym. Sci. Part A Polym. Chem. 1996, 34, 403–411. [CrossRef]
136. Nelson, E.W.; Carter, T.P.; Scranton, A.B. The role of the triplet state in the photosensitization of cationic polymerizations by
anthracene. J. Polym. Sci. Part A Polym. Chem. 1995, 33, 247–256. [CrossRef]
137. Corcione, C.E.; Greco, A.; Maffezzoli, A. Time–temperature and time-irradiation intensity superposition for photopolymerization
of an epoxy based resin. Polymer 2005, 46, 8018–8027. [CrossRef]
138. Pantiru, M.; Vuluga, D.M.; Vasilescu, D.S.; Abadie, M.J.M. Study of the cationic photopolymerization kinetics of cyclic acetals.
Polym. Bull. 2002, 47, 485–492. [CrossRef]
139. Van Assche, G.; Swier, S.; van Mele, B. Modeling and experimental verification of the kinetics of reacting polymer systems.
Thermochim. Acta 2002, 388, 327–341. [CrossRef]
140. Kamal, M.R. Thermoset characterization for moldability analysis. Polym. Eng. Sci. 1974, 14, 231–239. [CrossRef]
141. Rabinowitch, E. Collision, co-ordination, diffusion and reaction velocity in condensed systems. Trans. Faraday Soc. 1937, 33, 1225.
[CrossRef]
142. Williams, M.L.; Landel, R.F.; Ferry, J.D. The Temperature Dependence of Relaxation Mechanisms in Amorphous Polymers and
Other Glass-forming Liquids. J. Am. Chem. Soc. 1955, 77, 3701–3707. [CrossRef]
143. Wisanrakkit, G.; Gillham, J.K. The glass transition temperature (Tg) as an index of chemical conversion for a high-Tg amine/epoxy
system: Chemical and diffusion-controlled reaction kinetics. J. Appl. Polym. Sci. 1990, 41, 2885–2929. [CrossRef]
144. Cho, J.-D.; Hong, J.-W. Photo-curing kinetics for the UV-initiated cationic polymerization of a cycloaliphatic diepoxide system
photosensitized by thioxanthone. Eur. Polym. J. 2005, 41, 367–374. [CrossRef]
145. Levenberg, K. A method for the solution of certain non-linear problems in least squares. Quart. Appl. Math. 1944, 2, 164–168.
[CrossRef]
146. Marquardt, D.W. An Algorithm for Least-Squares Estimation of Nonlinear Parameters. J. Soc. Ind. Appl. Math. 1963, 11, 431–441.
[CrossRef]
147. Boey, F.; Qiang, W. Experimental modeling of the cure kinetics of an epoxy-hexaanhydro-4-methylphthalicanhydride (MHHPA)
system. Polymer 2000, 41, 2081–2094. [CrossRef]
148. Abadie, M.J.M.; Chia, N.K.; Boey, F. Cure kinetics for the ultraviolet cationic polymerization of cycloliphatic and diglycidyl
ether of bisphenol-A (DGEBA) epoxy systems with sulfonium salt using an auto catalytic model. J. Appl. Polym. Sci. 2002, 86,
1587–1591. [CrossRef]
149. Macan, J.; Ivanković, H.; Ivanković, M.; Mencer, H.J. Study of cure kinetics of epoxy-silica organic–inorganic hybrid materials.
Thermochim. Acta 2004, 414, 219–225. [CrossRef]
150. Golaz, B.; Michaud, V.; Leterrier, Y.; Månson, J.-A. UV intensity, temperature and dark-curing effects in cationic photo-
polymerization of a cycloaliphatic epoxy resin. Polymer 2012, 53, 2038–2048. [CrossRef]
151. Šesták, J.; Berggren, G. Study of the kinetics of the mechanism of solid-state reactions at increasing temperatures. Thermochim.
Acta 1971, 3, 1–12. [CrossRef]
152. Voytekunas, V.Y.; Ng, F.L.; Abadie, M.J. Kinetics study of the UV-initiated cationic polymerization of cycloaliphatic diepoxide
resins. Eur. Polym. J. 2008, 44, 3640–3649. [CrossRef]
153. Jiang, F.; Drummer, D. Curing Kinetic Analysis of Acrylate Photopolymer for Additive Manufacturing by Photo-DSC. Polymers
2020, 12, 1080. [CrossRef]
154. Xu, W.; Bao, S.; Shen, S.; Wang, W.; Hang, G.; He, P. Differential scanning calorimetric study on the curing behavior of epoxy
resin/diethylenetriamine/organic montmorillonite nanocomposite. J. Polym. Sci. B Polym. Phys. 2003, 41, 378–386. [CrossRef]
155. Chen, D.Z.; He, P.S.; Pan, L.J. Cure kinetics of epoxy-based nanocomposites analyzed by Avrami theory of phase change. Polym.
Test. 2003, 22, 689–697. [CrossRef]
156. Murias, P.; Byczyński, Ł.; Maciejewski, H.; Galina, H. A quantitative approach to dynamic and isothermal curing of an epoxy
resin modified with oligomeric siloxanes. J. Therm. Anal. Calorim. 2015, 122, 215–226. [CrossRef]
157. Pollard, M.; Kardos, J.L. Analysis of epoxy resin curing kinetics using the Avrami theory of phase change. Polym. Eng. Sci. 1987,
27, 829–836. [CrossRef]
158. Marschik, C.; Roland, W.; Löw-Baselli, B.; Steinbichler, G. Application of Hybrid Modeling in Polymer Processing. In Proceedings
of the Annual Technical Conference for Plastic Professionals (SPE ANTEC), Online, 30 March–5 May 2020.
159. Taki, K. A Simplified 2D Numerical Simulation of Photopolymerization Kinetics and Oxygen Diffusion-Reaction for the Continu-
ous Liquid Interface Production (CLIP) System. Polymers 2020, 12, 875. [CrossRef] [PubMed]
160. Taki, K.; Watanabe, Y.; Ito, H.; Ohshima, M. Effect of Oxygen Inhibition on the Kinetic Constants of the UV-Radical Photopolymer-
ization of Diurethane Dimethacrylate/Photoinitiator Systems. Macromolecules 2014, 47, 1906–1913. [CrossRef]
161. Taki, K.; Watanabe, Y.; Tanabe, T.; Ito, H.; Ohshima, M. Oxygen concentration and conversion distributions in a layer-by-layer
UV-cured film used as a simplified model of a 3D UV inkjet printing system. Chem. Eng. Sci. 2017, 158, 569–579. [CrossRef]
162. Gao, K.; Ingenhut, B.L.J.; van de Ven, A.P.A.; Valega Mackenzie, F.O.; ten Cate, A.T. (Eds.) Multiphysics Modeling of Photo-
Polymerization in Stereolithography Printing Process and Validation. In Proceedings of the 2018 COMSOL Conference in
Lausanne, Lausanne, Switzerland, 22–24 October 2018.
Polymers 2022, 14, 2074 57 of 58

163. Mizukami, Y.; Rajniak, D.; Rajniak, A.; Nishimura, M. A novel microchip for capillary electrophoresis with acrylic microchannel
fabricated on photosensor array. Sens. Actuators B Chem. 2002, 81, 202–209. [CrossRef]
164. Erdmann, L. MOEMS-based lithography for the fabrication of micro-optical components. J. Micro/Nanolithography MEMS MOEMS
2005, 4, 41601. [CrossRef]
165. Jariwala, A.S.; Ding, F.; Zhao, X.; Rosen, D.W. A Film Fabrication Process on Transparent Substrate Using Mask Projection Micro-
Stereolithography; The University of Texas at Austin: Austin, TX, USA, 2008.
166. Jariwala, A.S.; Ding, F.; Boddapati, A.; Breedveld, V.; Grover, M.A.; Henderson, C.L.; Rosen, D.W. Modeling effects of oxygen
inhibition in mask-based stereolithography. Rapid Prototyp. J. 2011, 17, 168–175. [CrossRef]
167. Huang, Y.; Kormakov, S.; He, X.; Gao, X.; Zheng, X.; Liu, Y.; Sun, J.; Wu, D. Conductive polymer composites from renewable
resources: An overview of preparation, properties, and applications. Polymers 2019, 11, 187. [CrossRef]
168. Zhang, X.; Fujiwara, S.; Fujii, M. Measurements of Thermal Conductivity and Electrical Conductivity of a Single Carbon Fiber.
Int. J. Thermophys. 2000, 21, 965–980. [CrossRef]
169. Morak, M.; Marx, P.; Gschwandl, M.; Fuchs, P.F.; Pfost, M.; Wiesbrock, F. Heat Dissipation in Epoxy/Amine-Based Gradient
Composites with Alumina Particles: A Critical Evaluation of Thermal Conductivity Measurements. Polymers 2018, 10, 1131.
[CrossRef] [PubMed]
170. Moradi, S.; Calventus, Y.; Román, F.; Hutchinson, J.M. Achieving High Thermal Conductivity in Epoxy Composites: Effect of
Boron Nitride Particle Size and Matrix-Filler Interface. Polymers 2019, 11, 1156. [CrossRef] [PubMed]
171. Plesa, I.; Notingher, P.V.; Schlögl, S.; Sumereder, C.; Muhr, M. Properties of Polymer Composites Used in High-Voltage Applica-
tions. Polymers 2016, 8, 173. [CrossRef] [PubMed]
172. Kochetov, R.; Andritsch, T.; Lafont, U.; Morshuis, P.; Smit, J.J. Thermal Conductivity of Nano-Filled Epoxy Systems. In 2009
Annual Report, Proceedings of the Conference on Electrical Insulation and Dielectric Phenomena, CEIDP 2009: Virginia Beach, VA, USA,
18–21 October 2009; IEEE: Piscataway, NJ, USA, 2009; pp. 658–661; ISBN 978-1-4244-4557-8.
173. Huang, X.; Jiang, P.; Tanaka, T. A review of dielectric polymer composites with high thermal conductivity. IEEE Elects. Insul. Mag.
2011, 27, 8–16. [CrossRef]
174. Windberger, M.S.; Dimitriou, E.; Rendl, S.; Wewerka, K.; Wiesbrock, F. Temperature-Triggered/Switchable Thermal Conductivity
of Epoxy Resins. Polymers 2020, 13, 65. [CrossRef]
175. Ma, H.; Gao, B.; Wang, M.; Yuan, Z.; Shen, J.; Zhao, J.; Feng, Y. Strategies for enhancing thermal conductivity of polymer-based
thermal interface materials: A review. J. Mater. Sci. 2021, 56, 1064–1086. [CrossRef]
176. Friedrich, K.; Breuer, U. Multifunctionality of Polymer Composites: Challenges and New Solutions; Elsevier: Kidlington, UK, 2015;
ISBN 9780323265034.
177. Unterweger, C.; Brüggemann, O.; Fürst, C. Synthetic fibers and thermoplastic short-fiber-reinforced polymers: Properties and
characterization. Polym. Compos. 2014, 35, 227–236. [CrossRef]
178. Rajak, D.K.; Pagar, D.D.; Menezes, P.L.; Linul, E. Fiber-Reinforced Polymer Composites: Manufacturing, Properties, and
Applications. Polymers 2019, 11, 1667. [CrossRef]
179. Holbery, J.; Houston, D. Natural-fiber-reinforced polymer composites in automotive applications. JOM 2006, 58, 80–86. [CrossRef]
180. Tan, D.; Irwin, P.; Sikalidis, C. Polymer Based Nanodielectric Composites. In Advances in Ceramics: Electric and Magnetic Ceramics,
Bioceramics, Ceramics and Environment; InTech: London, UK, 2011; p. 115; ISBN 978-953-307-350-7.
181. Song, W.; Han, B.; Zhang, D.; Sun, Z.; Wang, X.; Lei, Q. Preparation and properties of BiFeO3/LDPE nanocomposite. In
Proceedings of the 2015 IEEE 11th International Conference on the Properties and Applications of Dielectric Materials (ICPADM
2015), Sydney, Australia, 19–22 July 2015; IEEE: Piscataway, NJ, USA, 2015; pp. 800–803, ISBN 978-1-4799-8903-4.
182. Keith Nelson, J. Overview of Nanodielectrics: Insulating Materials of the Future. In Proceedings of the 2007 Electrical Insulation
Conference and Electrical Manufacturing Expo (EIC/EME), Nashville, TN, USA, 22–24 October 2007; IEEE Service Center:
Piscataway, NJ, USA, 2007; pp. 229–235, ISBN 978-1-4244-0446-9.
183. Qiao, Y.; Islam, M.S.; Wang, L.; Yan, Y.; Zhang, J.; Benicewicz, B.C.; Ploehn, H.J.; Tang, C. Thiophene Polymer-Grafted Barium
Titanate Nanoparticles toward Nanodielectric Composites. Chem. Mater. 2014, 26, 5319–5326. [CrossRef]
184. King, R.W.P. Electric currents and fields induced in cells in the human brain by radiation from hand-held cellular telephones. J.
Appl. Phys. 2000, 87, 893–900. [CrossRef]
185. Gandhi, O.P. Electromagnetic fields: Human safety issues. Annu. Rev. Biomed. Eng. 2002, 4, 211–234. [CrossRef] [PubMed]
186. Roh, J.-S.; Chi, Y.-S.; Kang, T.J.; Nam, S. Electromagnetic Shielding Effectiveness of Multifunctional Metal Composite Fabrics. Text.
Res. J. 2008, 78, 825–835. [CrossRef]
187. Sathish Kumar, K.; Rengaraj, R.; Venkatakrishnan, G.R.; Chandramohan, A. Polymeric materials for electromagnetic shielding—A
review. Mater. Today Proc. 2021, 47, 4925–4928. [CrossRef]
188. Pojman, J.A. Frontal Polymerization. In Polymer Science, 2nd ed.; Moeller, M., Matyjaszewski, K., Eds.; Elsevier Science: Burlington,
NJ, USA, 2012; pp. 957–980; ISBN 9780080878621.
189. Lewis, L.L.; DeBisschop, C.S.; Pojman, J.A.; Volpert, V.A. Isothermal frontal polymerization: Confirmation of the mechanism and
determination of factors affecting the front velocity, front shape, and propagation distance with comparison to mathematical
modeling. J. Polym. Sci. Part A Polym. Chem. 2005, 43, 5774–5786. [CrossRef]
190. Warren, J.A.; Cabral, J.T.; Douglas, J.F. Solution of a field theory model of frontal photopolymerization. Phys. Rev. E Stat. Nonlin.
Soft Matter Phys. 2005, 72, 21801. [CrossRef]
Polymers 2022, 14, 2074 58 of 58

191. Pojman, J.A.; Ilyashenko, V.M.; Khan, A.M. Free-radical frontal polymerization: Self-propagating thermal reaction waves. J. Chem.
Soc. Faraday Trans. 1996, 92, 2825. [CrossRef]
192. Khan, A.M.; Pojman, J.A. The Use of Frontal Polymerization in Polymer Synthesis. Trends Polym. Sci. 1996, 8, 253–257.
193. Cabral, J.T.; Douglas, J.F. Propagating waves of network formation induced by light. Polymer 2005, 46, 4230–4241. [CrossRef]
194. Rytov, B.L.; Ivanov, V.B.; Ivanov, V.V.; Anisimov, V.M. Mechanisms of front propagation of photochemical reactions in polymer
containing media: 1. Frontal regimes of photochemical reactions in polymer matrices with bleaching of specimen behind the
front. Polymer 1996, 37, 5695–5698. [CrossRef]
195. Miller, G.A.; Gou, L.; Narayanan, V.; Scranton, A.B. Modeling of photobleaching for the photoinitiation of thick polymerization
systems. J. Polym. Sci. Part A Polym. Chem. 2002, 40, 793–808. [CrossRef]
196. Petko, F.; Świeży, A.; Ortyl, J. Photoinitiating systems and kinetics of frontal photopolymerization processes—The prospects for
efficient preparation of composites and thick 3D structures. Polym. Chem. 2021, 12, 4593–4612. [CrossRef]
197. Frulloni, E.; Salinas, M.M.; Torre, L.; Mariani, A.; Kenny, J.M. Numerical modeling and experimental study of the frontal
polymerization of the diglycidyl ether of bisphenol A/diethylenetriamine epoxy system. J. Appl. Polym. Sci. 2005, 96, 1756–1766.
[CrossRef]
198. Robertson, I.D.; Yourdkhani, M.; Centellas, P.J.; Aw, J.E.; Ivanoff, D.G.; Goli, E.; Lloyd, E.M.; Dean, L.M.; Sottos, N.R.;
Geubelle, P.H.; et al. Rapid energy-efficient manufacturing of polymers and composites via frontal polymerization. Nature
2018, 557, 223–227. [CrossRef]
199. Slugovc, C.; Trimmel, G. Polymer Meeting 14—Book of Abstracts; Verlag der Technischen Universität Graz: Graz, Austria, 2021.
200. Ghosh, A.K.; Dwivedi, M. Advantages and Applications of Polymeric Composites. In Processability of Polymeric Composites;
Springer: New Delhi, India, 2020; pp. 29–57.