Radical Cations of Phenoxazine and Dihydrophenazine Photoredox Catalysts and Their Role as Deactivators in Organocatalyzed Atom Transfer Radical Polymerization

Article abstract of DOI:10.1021/acs.macromol.1c00640

Radical cations of photoredox catalysts used in organocatalyzed atom transfer radical polymerization (O-ATRP) have been synthesized and investigated to gain insight into deactivation in O-ATRP. The stability and reactivity of these compounds were studied in two solvents, N,N-dimethylacetamide and ethyl acetate, to identify possible side reactions in O-ATRP and to investigate the ability of these radical cations to deactivate alkyl radicals. A number of other factors that could influence deactivation in O-ATRP were also probed, such as ion pairing with the radical cations, radical cation oxidation potential, and halide oxidation potential. Ultimately, these studies enabled radical cations to be employed as reagents during O-ATRP to demonstrate improvements in polymerization control with increasing radical cation concentrations. In the polymerization of acrylates, this approach enabled superior molecular weight control, a decrease in polymer dispersity from 1.90 to 1.44, and an increase in initiator efficiency from 78 to 102%. This work highlights the importance of understanding the mechanism and side reactions of O-ATRP, as well as the importance of catalyst radical cations for successful O-ATRP.

Full text of DOI:10.1021/acs.macromol.1c00640

pubs.acs.org/Macromolecules
Article
Radical Cations of Phenoxazine and Dihydrophenazine Photoredox
Catalysts and Their Role as Deactivators in Organocatalyzed Atom
Transfer Radical Polymerization
Daniel A. Corbin, Blaine G. McCarthy, Zach van de Lindt, and Garret M. Miyake*
Cite This: Macromolecules 2021, 54, 4726−4738
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: Radical cations of photoredox catalysts used in
organocatalyzed atom transfer radical polymerization (O-ATRP)
have been synthesized and investigated to gain insight into
deactivation in O-ATRP. The stability and reactivity of these
compounds were studied in two solvents, N,N-dimethylacetamide
and ethyl acetate, to identify possible side reactions in O-ATRP
and to investigate the ability of these radical cations to deactivate
alkyl radicals. A number of other factors that could influence
deactivation in O-ATRP were also probed, such as ion pairing with
the radical cations, radical cation oxidation potential, and halide
oxidation potential. Ultimately, these studies enabled radical
cations to be employed as reagents during O-ATRP to demonstrate
improvements in polymerization control with increasing radical cation concentrations. In the polymerization of acrylates, this
approach enabled superior molecular weight control, a decrease in polymer dispersity from 1.90 to 1.44, and an increase in initiator
efficiency from 78 to 102%. This work highlights the importance of understanding the mechanism and side reactions of O-ATRP, as
well as the importance of catalyst radical cations for successful O-ATRP.
concentration of radicals in solution. Macroscopically, this
process minimizes irreversible termination reactions while
allowing the polymer chain to be reactivated for subsequent
chain growth, enabling control over the polymer structure.
While O-ATRP retains many of the advantages of
traditional ATRP as well as some added benefits, such as
mild and metal-free reaction conditions, it remains relatively
limited in monomer scope and mechanistic understanding.
Previously, limitations in monomer scope have generally been
addressed through the development of new catalysts. For
example, Matyjaszewski extended the scope of O-ATRP from
methacrylates to acrylonitrile by developing new phenothia-
zine catalysts,15 although phenyl phenothiazine, first reported
by Hawker for the O-ATRP of methacrylates,11 exhibited the
best performance.15 The development of dihydrophenazine16
and phenoxazine17 catalysts ultimately led to the controlled
polymerization of vinylcyclopropanes with tunable polymer
backbone composition.18 Through further development of the
dihydrophenazine family, the controlled polymerizations of
INTRODUCTION
■
The development of polymerization methods that exhibit
precise control over polymer molecular weight, dispersity (Đ),
and structure has long been a focus of polymer chemistry.1−4
Early examples of controlled polymerizations, also referred to
as “living” polymerizations, were technically challenging to
execute and required demanding reaction conditions,5 limiting
their broad utility. However, with the advent of controlled
radical polymerization (CRP) methods,6−9 precision polymer
synthesis has become more powerful and accessible. One
recently developed CRP is organocatalyzed atom transfer
radical polymerization (O-ATRP), which employs organic
photoredox catalysts (PCs) to synthesize well-defined
polymers under mild, metal-free conditions.10,11
Similar to traditional atom transfer radical polymerization
(ATRP) methods,12−14 O-ATRP controls polymer growth
through a reversible deactivation mechanism (Figure 1a).
During this process, the PC activates a “dormant” polymer
possessing a terminal C−Br bond by reduction of the polymer
chain end, generating a carbon-centered radical capable of
propagation as well as Br− and the PC radical cation (PC•+).
As with all radical polymerizations, the propagating radical is
susceptible to irreversible termination by reaction with other
radicals in solution. As such, a key feature of O-ATRP is
reversible deactivation, wherein the PC•+ mediates reinstalla-
tion of Br on the polymer chain end to lower the
Received: March 23, 2021
Published: May 3, 2021
https://doi.org/10.1021/acs.macromol.1c00640
© 2021 American Chemical Society
Macromolecules 2021, 54, 4726−4738
4726

Macromolecules
pubs.acs.org/Macromolecules
Article
primarily on computational methods rather than experimental
evidence to probe the mechanism of deactivation26 and
impact of ion pairing30 in this process. While the results of
these studies were certainly informative and served as useful
guides for future development, experimental investigation of
the deactivation process and methods to manipulate this
process are still needed.
In the present work (Figure 1c), we attempt to address
these limits in our understanding of deactivation in O-ATRP
through the investigation of PC radical cationsthe key
catalytic species we propose mediate this process. To do so,
several radical cations of O-ATRP PCs are synthesized and
characterized for the first time. Through investigation of the
reactivity of these compounds, new side reactions are
identified that can inhibit deactivation in O-ATRP. Further,
through the development of a deactivation model reaction,
evidence is found supporting the role of PC•+Br− as the
deactivator, and factors influencing this process are identified.
The most notable factors include the radical cation oxidation
potential [E°(PC•+/PC)] and the oxidation potential of the
halide [E°(X•/X−)], both of which can directly impact the
rate of deactivation. By investigating ion pairing with PC•+, it
is found that the choice of reaction solvent is far more
influential than PC•+ structure for the formation of a PC•+ ion
pair. Finally, by employing an isolated radical cation in O-
ATRP, it is demonstrated that these compounds can be used
to improve polymerization control, further supporting their
role in deactivation. Altogether, this work highlights the utility
of PC radical cations in O-ATRP, as well as the importance of
mechanistic understanding for the continued development of
O-ATRP.
Figure 1. Proposed mechanism of O-ATRP (a) and previous work to
improve deactivation during O-ATRP (b). This work (c) aims to
develop a better understanding of deactivation, the species involved
in this step, and how they can be used to improve polymerization
control in O-ATRP.
RESULTS AND DISCUSSION
■
Synthesis and Characterization of Radical Cations.
Synthesis. Radical cations of 1−11 (Figure 2a) were
synthesized using nitrosonium hexafluorophosphate, as the
oxidation potential of NO+ is more than sufficient to oxidize
1−11 [E°(NO+/NO) = 1.25 V,31 E1/2 ∼ E°(PC•+/PC) =
0.14−0.73 V, both vs saturated calomel electrode (SCE) in
MeCN]. Since the byproduct of this reaction is NO(g), the
product can be easily isolated by precipitation and washing
with hexanes. For PCs 1−5, it should be noted that the cyclic
voltammograms in MeCN exhibit two reversible oxidations
that could be accessible using NOPF6 (see Section S3 of the
Supporting Information). As such, precise stoichiometry is
necessary to avoid overoxidation of these compounds to the
dicationic species. For crystallography, crystal growth was
styrene19 and various acrylates20 were achieved. Finally, the
first example of the controlled polymerization of acrylate
monomers via O-ATRP came through the introduction of
dihydroacridine catalysts in 2020, which feature strongly
oxidizing radical cations [E°(PC•+/PC)] capable of control-
ling the fast propagation of acrylates through deactivation.21
In an alternative approach, we recently reported the first
application of electrolysis in O-ATRP in an attempt to gain
external control of deactivation during a polymerization.22 We
reasoned that applying an oxidizing electrochemical potential
to a polymerization solution would increase the concentration
of the PC radical cation (PC•+), which would, in turn, lead to
improved deactivation and polymerization control during O-
ATRP (Figure 1b). While this hypothesis was ultimately
supported, this work highlighted limitations in our under-
standing of the mechanism of this method, and it inspired new
questions to guide future experimentation, namely, is PC•+ the
deactivator in O-ATRP, does PC•+ engage in side reactions
that inhibit deactivation, and what factors influence the ability
of a PC•+ to effectively mediate deactivation?22 In other
words, this work highlighted the necessity of furthering our
mechanistic understanding of O-ATRP.
−
attempted using various methods with a range of PC•+PF6
salts. However, crystals of these compounds suitable for
single-crystal X-ray diffractometry (SCXRD) could not be
obtained. Given the large size of the radical cations, we
−
envisioned that the PF6 anion might be too small to enable
effective crystal packing and that a larger counterion might be
beneficial. As such, radical cations for SCXRD analysis were
synthesized using tris(4-bromophenyl)ammoniumyl hexa-
chloroantimonate, yielding 4•+SbCl6 and 9•+SbCl6 . In
both cases, crystallization by vapor diffusion (see Section S2
of the Supporting Information for details) gave needles of
suitable quality for X-ray diffraction studies.
−
−
Although several investigations of the O-ATRP mechanism
have been previously reported, the majority of these reports
focus on activation23−26 and the impact of PC photophysics
on this step.27−29 With regard to deactivation, only a handful
of reports exist,26,30 despite this step being critical to
polymerization control. Further, these investigations relied
Spectroscopic Characterization. To verify the identity of
each PC•+, absorption spectra of the isolated compounds were
compared to PC•+ spectra obtained using spectro-electro-
chemistry (Figure 2b). The spectra of the isolated compounds
4727
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
Figure 2. (a) Structures of photoredox catalysts studied in this work. (b) Representative example of radical cation spectra obtained by spectro-
electrochemistry in acetonitrile. (c) Representative comparison of UV−vis spectra for a PC (solid, yellow), the PC•+ obtained by spectro-
electrochemistry (dashed, light blue), PC•+PF6− (solid, dark blue), and a redissolved PC•+SbCl6− crystal (dashed, teal). (d) Crystal structure of a
phenazine radical cation (solvent molecule is removed for clarity).
were found to agree well with the reference spectra (Figure
2c), supporting the successful synthesis of each PC•+. In
equation, an expression giving the ratio of PC•+ to PC based
on the Eocp and E1/2 can be written (eq 2). Using eq 2, the
addition, the spectra of 4•+PF6 and 4•+SbCl6 were nearly
identical, suggesting that the identity of the counteranion has
a negligible impact on the spectroscopic properties of the
radical cation.
purity of each PC•+PF6 salt was estimated to be ∼97% or
−
−
−
greater. To verify the accuracy of this method, elemental
analysis was also performed for 11•+PF6 , which agreed well
−
with the calculated elemental composition of this compound
(see the Supporting Information)
Since crystals of 4•+SbCl6 and 9•+SbCl6 were obtained
and analyzed by SCXRD to determine their crystal structures
(see the Crystallography section), we wondered if the solid-
state spectra of these compounds would match their solution
spectra. Unfortunately, the crystals obtained were insufficiently
transparent to obtain well-resolved absorption spectra in the
solid state, so this comparison could not be made. Instead, the
crystals were redissolved in MeCN and their spectra were
measured in solution (Figure 2c). The agreement of these
−
−
[PC•+]
[PC]
ocp−E1/2)/RT
= eF(E
(2)
Crystallography. Single-crystal X-ray diffractometry was
performed using crystals obtained for 4•+SbCl6 (Figure 2d)
−
−
−
and 9•+SbCl6 . In each case, the SbCl6 anion was found
centered above the aromatic core of the PC•+ and the PC•+
was found to cocrystallize with 1 equiv of the solvent molecule
spectra and the PC•+PF6 spectra further support the identity
−
(Figures S54 and S55). While 4•+SbCl6 exhibited minimal
−
of the PC•+ salts.
disorder and was easily refined, 9•+SbCl6− exhibited significant
disorder that had to be modeled during refinement, namely,
the 2-naphthyl ring at the N-aryl position was disordered over
two positions, presumably because the substituent can rotate
about the C−N bond. In both cases, the refined crystal
structures matched the structures anticipated for the radical
cations. Combined with the spectroscopic data presented
above, these data provide further support for the identities of
these PC•+ salts.
Electrochemical Characterization. For each of the radical
cations synthesized, estimates of their purities were obtained
by the measurement of their open-circuit potentials (Eocp) in
MeCN. According to the Nernst equation (eq 1), the Eocp of a
PC•+ solution is dependent on the E1/2 of the redox couple
and the relative quantities of PC and PC•+
•+
i
y
RT
F
[PC ]
[PC]
j
j
z
z
Eocp = E1/2
+
ln
j
z
z
j
(1)
k
{
Stability of Radical Cations in Solution. During initial
work with these compounds, it was noticed that the stability
of PC•+ in solution was strongly dependent on the solvent. To
where R is the ideal gas constant, T is the absolute
temperature, and F is Faraday’s constant. Rearranging this
4728
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
understand the factors influencing the stability of PC•+, a
series of experiments were performed to follow the
decomposition of PC•+ using UV−vis spectroscopy in solvents
relevant to O-ATRP. Since it was previously observed that
dihydrophenazine PCs can undergo side reactions to be
substituted at the PC core,20 5•+ was primarily used for these
investigations as its core positions are protected by 2-naphthyl
substituents.
radical cation (10•+; Figures S62 and S64). The same
behavior was found but with a greater rate for the
disappearance of 10•+ relative to 5•+ and 3•+. As 10•+ is
significantly more oxidizing in the ground state [E1/2
∼
E°(10•+/10) = 0.66 V vs SCE in MeCN], its greater reactivity
may be due to its stronger oxidation potential.
In every case, an isosbestic point was observed during the
disappearance of PC•+, indicating conversion to a single
product. Further, when the reaction was carried out to high
conversion (as indicated by complete loss of the PC•+ signal),
the product spectrum closely resembled that of the PC. These
data, combined with the observation of pseudo-first-order
kinetics, led us to hypothesize that decomposition of PC•+
occurs by single electron transfer from DMAc to generate the
neutral PC. Under irradiation, this reaction might proceed
through a more oxidizing PC•+ excited state, which would
explain why the rate of the reaction increases. However, this
possibility will be discussed further in a later section (see the
Investigation of Radical Cation Side Reactions section).
While the oxidation of DMAc by PC•+ is consistent with
the data presented above, it is surprising given that the
oxidation potential of DMAc [E°calc(DMAc+/DMAc) = 1.98
V vs SCE] is significantly more positive than that of any PC•+
in this work [E1/2 ∼ E°(PC•+/PC) = 0.14−0.73 V vs SCE].
To probe this reaction further, a kinetic isotope study was
performed using DMF and deuterated dimethylformamide
(d7-DMF), assuming similar reactivity would be observed as
with DMAc (see the Section S4 in the Supporting
Information). A normal kinetic isotope effect was observed
in the dark (kH/kD = 5.9), and an inverse isotope effect was
observed under irradiation (kH/kD = 0.18 0.04). As inverse
equilibrium isotope effects are more common than inverse
kinetic isotope effects,33 this result led us to believe that an
equilibrium might be involved in the excited-state oxidation of
DMAc and DMF. We hypothesize that PC•+ preassociates
with a solvent molecule prior to photoinduced electron
transfer. However, regardless of the mechanism of this
reaction, the observation of this isotope effect supports a
direct reaction between DMF and 5•+.
In N,N-dimethylacetamide (DMAc), decomposition of 5•+
was observed with and without irradiation (Figure 3).
Figure 3. Representative example of UV−vis spectra following the
disappearance of 5•+ in DMAc under irradiation with a white light-
emitting diode (LED). The inset shows linear pseudo-first-order
kinetics following the absorption at the λmax = 682 nm.
Following the kinetics of these reactions by UV−vis revealed
that the decomposition reaction is accelerated by irradiation
with white light (Figures S56 and S58) and exhibits pseudo-
first-order kinetics in the dark (kobs = 0.00064 M−1 s−1) and
under irradiation (kobs = 0.39 M−1 s−1). The observed increase
in the rate of 5•+ disappearance with light suggests the
possibility of excited-state reactivity, which has previously
been observed for similar radical cations with electron-rich
substrates tethered at the N-aryl position.32 By contrast, 5•+
exhibited excellent stability in ethyl acetate (EtAc) regardless
of irradiation (Figures S65 and S67).
Several alternative hypotheses explaining the decomposition
of 5•+ were also investigated, including the oxidation of the
To investigate whether this behavior is unique to 5•+, the
same study was performed with a noncore-substituted
phenazine (3•+; Figures S59 and S61) and a phenoxazine
−
PF6 anion and the possibility of solvent impurities. To rule
−
−
out a reaction with PF6 , a solution of 5•+PF6 was prepared
Figure 4. (a) Representative example of UV−vis spectra following the disappearance of 5•+ in the presence of LiBr. The inset shows a
1
comparison of pseudo-first-order kinetics demonstrating the impact of irradiation and the halide identity. (b) Identification by H NMR of
bromocyclohexane formed from Br• after oxidation of Br− by 5•+.
4729
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
in deuterated DMF and irradiated with white LEDs. After the
solution turned from dark blue to yellow, indicating
conversion of 5•+ to 5, the reaction products were analyzed
by 19F NMR (Figure S68). The resulting spectrum was
consistent with the preservation of the PF6− anion. To test for
solvent impurities, DMF and d7-DMF were analyzed by gas
chromatography (Figures S70 and S71). Neither analysis
revealed volatile impurities that could account for the
observed reactivity, supporting a direct reaction between 5•+
and DMF.
Investigation of Radical Cation Side Reactions.
Impact of Irradiation on Radical Cation Reactivity. To
further investigate possible reactivity from the excited state of
5•+, a series of reactions were performed in the presence of
substrates with increasing oxidation potentials. For substrates
with lower oxidation potentials [E°(S+/S) ≤ 1 V vs SCE], the
slow disappearance of 5•+ was observed with equal kinetics
under irradiation and in the dark (Table S6). Since irradiation
did not impact this reaction, a ground-state mechanism is
proposed to be most likely with these substrates. Instead, for
substrates with greater oxidation potentials, no reactivity was
observed either in the dark or under irradiation, with DMAc
being the only exception at high concentrations. This
observation may be linked to the excited-state lifetime of
5•+, which could be too short to engage in bimolecular
reactions in solution unless the substrate is present in high
enough concentration (i.e., solvent quantities) to overcome
the lifetime of this species.
0.08 M−1 s−1). Instead, the reaction between 10•+ and Br− was
too rapid to follow by UV−vis (Figure S110), even in the
absence of light. These results broadly correlate with the
oxidation potentials of these compounds [E1/2(3•+/3) = 0.18
V; E1/2(5•+/5) = 0.32 V; E1/2(10•+/10) = 0.66 V, all vs SCE
in MeCN], possibly yielding insight into their capabilities as
deactivators in O-ATRP. This possibility will be discussed in
greater detail later in the text (see the Factors Influencing the
Deactivation of Alkyl Radicals section).
In the Presence of Chloride. While metal-catalyzed ATRP
is often performed in the presence of bromide and chloride
(either by using alkyl bromide or chloride initiators, or
through the addition of halide salts),12 O-ATRP in the
presence of chloride has remained challenging. One difference
between these halides is that alkyl chloride bond strengths are
typically greater than alkyl bromides, which would make
activation more challenging with alkyl chlorides. However,
previous investigations by Matyjaszewski and co-workers have
suggested that the issue with chloride may be ineffective
deactivation,26 though the origin of this issue remains a
mystery. To investigate this limitation of O-ATRP further, the
reactivity of 5•+ was studied in the presence of LiCl. Unlike
the reaction with Br−, that with Cl− in the dark exhibited only
a minor increase in the rate of disappearance of 5•+ (kCl‑dark
=
0.0020 0.0007 M−1 s−1) relative to the background reaction
in DMAc (kDMAc‑dark = 0.00064 M−1 s−1). Irradiation with
white LEDs again increased the rate of 5•+ disappearance
(kCl‑light = 0.074 0.013 M−1 s−1), though this reaction was
still slower than with Br− both in the dark (kBr‑dark = 0.14
In the Presence of Bromide. The reactivity of 5•+ was also
investigated in the presence of halides (Figure 4a), given the
relevance of these ions to O-ATRP. Although some
experiments were performed in ethyl acetate (Figures S104
and S106), the reaction of 5•+ with halides proved challenging
to track due to the rate of the reaction. As such, DMAc was
used instead for these investigations.
0.02 M−1 s−1) and under irradiation (kBr‑light = 0.31
0.06
M−1 s−1).
Under both irradiation conditions, the oxidation of Cl−
appears to be significantly slower than the oxidation of Br−.
This observation is consistent with the oxidation potentials of
these ions [E°(Br3 /Br−) = 0.7 V, E°(Cl3 /Cl−) = 1.1 V, both
vs SCE in MeCN34]. As such, a possible explanation for poor
deactivation in O-ATRP using Cl− could be that it is more
challenging to oxidize this ion, which leads to an overall
slower rate of deactivation with Cl− relative to Br−. This
hypothesis is further supported by experiments measuring the
rate of deactivation in the presence of Br− vs Cl−, although
these data will be discussed later (see the Factors Influencing
the Deactivation of Alkyl Radicals section).
−
−
In the presence of 0.1 M LiBr in the dark, 5•+ exhibited
reactivity (kBr‑dark = 0.14
0.02 M−1 s−1) that was
distinguishable from the background reaction with DMAc
(kDMAc‑dark = 0.00064 M−1 s−1), suggesting a possible ground-
state reaction between 5•+ and Br−. An increase in the rate of
disappearance for 5•+ was observed under irradiation (kBr‑light
= 0.31 0.06 M−1 s−1), although it is difficult to distinguish
whether this change in rate was due to a reaction with Br− or
simply with DMAc (kDMAc‑light = 0.39 M−1 s−1). Regardless of
irradiation, the formation of 5 was observed by UV−vis in
each case (Figures S100 and S102), suggesting a single
electron transfer mechanism between 5•+ and Br−.
Despite the kinetic differences observed between Cl− and
Br−, irradiation of 5•+ in the presence of Cl− again led to the
recovery of the ground-state UV−vis spectrum of 5 (Figure
S115), indicating a similar redox mechanism leading to the
formation of 5 and Cl•. A trapping experiment was attempted
to provide evidence for the formation of Cl•, but this
experiment was unsuccessful. While this result does not rule
out the formation of Cl•, it further highlights the inefficiency
of Cl− oxidation by 5•+.
Proposed Mechanism of Substrate Oxidation. Consider-
ing the reactivity studies discussed thus far, we propose the
following mechanisms for substrate oxidation by PC•+. In the
ground state, substrate oxidation appears to proceed through a
bimolecular electron transfer reaction, which results in the
formation of neutral 5 and the oxidized substrate. Instead, in
the excited state, association of the substrate with 5•+ prior to
photoexcitation may facilitate electron transfer (Figure 5).
After preassociation, irradiation of 5•+ could lead to
photoinduced electron transfer, which is likely followed by
dissociation of the product complex to yield free 5 and
Since such a reaction would be expected to generate
bromine radical (Br•), an experiment was devised to probe for
the presence of Br• in this reaction. To do so, the radical
halogenation of alkanes was employed, wherein a halogen
radical performs hydrogen atom abstraction from an alkane to
generate an alkyl radical, followed by radical coupling of the
alkyl radical with another halogen radical to give the
halogenated alkane. The reaction of 5•+ and Br− was
performed in the presence of cyclohexane and monitored by
1H NMR for the formation of bromocyclohexane. Excitingly, a
small quantity of bromocyclohexane was observed (Figure
4b), supporting the hypothesized oxidation of Br− by 5•+.
Finally, the kinetics of this reaction were investigated with
two other radical cations. When the reaction of 3•+ and Br−
was followed, similar results were observed as with 5•+,
although at a reduced rate (k3‑dark = 0.02 M−1 s−1, k3‑light
=
4730
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
PC•+Br− ion pair might lead to the oxidation of Br−,
generating a free equivalent of Br• that could then undergo
radical coupling with the radical on the polymer chain end in
a subsequent step (Figure 6; “stepwise mechanism”). In either
case, the products of deactivation would be the same. It
should be noted that in this work, the primary catalyst family
investigated was dihydrophenazines, which differ structurally
from the previously investigated phenothiazines by a second
N-aryl substituent. In addition, the radical cations of
dihydrophenazines are typically much less oxidizing than
those of phenothiazines. Both these properties could
ultimately impact the mechanism of deactivation, leading to
differences between various catalyst families.
To investigate which of these mechanisms might predom-
inate, another model reaction employing AIBN was employed.
Cyclohexane was also added to the reaction in an attempt to
trap Br• during deactivation. It was reasoned that the
formation of bromocyclohexane should only be observed if
free Br• forms during deactivation through a stepwise
mechanism. Instead, if deactivation proceeds through a
concerted mechanism, a reaction between Br• and cyclo-
hexane should be sufficiently challenging to prevent the
formation of bromocyclohexane. Indeed, when this experi-
ment was carried out using 5•+, the primary product of the
reaction was found to be that from deactivation (2-bromo-2-
methylpropanenitrile), with little-to-no bromocyclohexane
detectable by 1H NMR (Figures S120 and S121). This
experiment suggests that a concerted mechanism may be most
likely.
Bromide vs Chloride. As was discussed briefly above, O-
ATRP in the presence of chloride has remained challenging,
presumably due to an issue during deactivation with Cl−.26
Based on our investigations of radical cation reactivity in the
presence of Br− and Cl−, one possible explanation for why
deactivation is successful with Br− but not Cl− is based on
their difference in oxidation potentials. Since Cl− is more
challenging to oxidize, the deactivation reaction with Cl− is
likely slower, leading to ineffective deactivation and poor
polymerization control in O-ATRP.
To test this hypothesis more directly, the deactivation
model reaction employed above was followed in situ using
UV−vis spectroscopy (Figure 7a). By doing so, the
disappearance of 5•+ during deactivation could be monitored
to measure the kinetics of deactivation, yielding direct insight
into factors that might influence the rate of deactivation.
Unsurprisingly, when the reaction was performed in the
presence of Cl−, the disappearance of 5•+ was much slower
than with Br− (Figure 7b), indicating less efficient
deactivation. In the absence of halide, the disappearance of
5•+ was only slightly slower than in the presence of Cl−.
Therefore, while deactivation still appears to occur in the
presence of Cl−, it is very slow. In O-ATRP, slow deactivation
would promote a higher concentration of radicals during the
reaction, ultimately increasing termination reactions and
inhibiting polymerization control.
Figure 5. One proposed mechanism for substrate oxidation by
photoexcited PC•+ facilitated by preassociation of the PC•+ and
substrate.
oxidized substrate. While such an association with 5•+ is not
surprising for Cl− or Br−, it is perhaps less anticipated for a
neutral substrate such as DMAc. However, DMAc contains a
lone pair of electrons at the nitrogen position as well as
significant electron density around the carbonyl oxygen, which
might be susceptible to a weak interaction with the positively
charged PC•+. While this weak interaction might be negligible
at low concentrations, higher concentrations may enable a
small degree of association between DMAc and 5•+, enabling
excited-state reactivity. Alternatively, a bimolecular reaction
between the excited state of 5•+ and the substrate may also be
feasible, depending on the lifetime of this excited state and the
concentration of the substrate in solution. However, deeper
investigation of the photophysical properties of these radical
cations is necessary to probe this possible reactivity further.
Factors Influencing the Deactivation of Alkyl
Radicals. Deactivation of Alkyl Radicals. To better under-
stand these radical cations in the context of deactivation, their
reactions with alkyl radicals were investigated. A reaction to
model deactivation in O-ATRP was devised using 5•+ in the
presence of Br− to deactivate thermally generated radicals
from azobisisobutyronitrile (AIBN) (Figure 7a). To first
determine whether 5•+ could operate as a radical deactivator,
the formation of the brominated deactivation product was
1
monitored by H NMR (Figures S118 and S119). The NMR
spectrum of the model reaction showed a peak matching the
expected chemical shift of the deactivation product (δ = 2.07
ppm in CD3CN), suggesting 5•+ can indeed deactivate alkyl
radicals.
Previous computational investigations have attempted to
determine the mechanism of deactivation using density
functional theory and Marcus theory to predict the rate of
deactivation via various mechanisms.26 This work concluded
that a termolecular mechanism was most favorable with
phenyl phenothiazine as the catalyst.35 Instead, we hypothe-
size that a bimolecular deactivation mechanism is operative,
wherein PC•+ and Br− form an ion pair (PC•+Br−) prior to
reaction with the propagating radical (Figure 6; “concerted
mechanism”). Based on the observed reactivity between 5•+
and Br−, it was also envisioned that deactivation could occur
through a stepwise mechanism. In this case, formation of the
In addition to the difference in oxidation potentials of the
halides, we hypothesized that the propensity of Br• and Cl• to
undergo side reactions might also be important. Cl• could be
more prone to H-atom abstraction than Br• based on the
greater bond strength of H−Cl than H−Br,36 which makes
H−Cl formation more thermodynamically favorable. In turn,
this greater driving force might lead to more side reactions in
O-ATRP. To probe this possibility, a collector−generator
Figure 6. Hypothesized mechanisms of deactivation investigated in
this work.
4731
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
Figure 8. In situ kinetics of deactivation with three dihydrophenazine
radical cations demonstrating the impact of PC oxidation potential
[E1/2 ∼ E°(PC•+/PC), all in V vs SCE] on the rate of deactivation.
Kinetics monitored at 682 nm (1•+), 680 nm (3•+), and 677 nm
(4•+). Data normalized to the maximum absorbance at the time of
PC•+ addition (t = 0).
measured using conductometry. While PC•+PF6 is not the
−
Figure 7. (a) Model reaction used in this work to investigate
deactivation in O-ATRP. (b) In situ kinetics of deactivation with 5•+
(monitored at 677 nm) in the presence of halides (Br−, dark blue;
Cl−, purple) and in their absence. Data normalized to maximum
absorbance at the time of PC•+ addition (t = 0).
true deactivator in O-ATRP, we hypothesized that these salts
would exhibit similar trends in ion pairing as PC•+Br−,
allowing broad conclusions to be drawn.
To understand how the radical cation structure impacts ion
pairing, Kassoc was measured for each PC•+PF6 synthesized
−
experiment was performed using a rotating ring-disc electrode
with LiBr or LiCl in a mixture of DMAc and methyl
methacrylate (MMA) to mimic O-ATRP conditions (Figure
S129). The results of these experiments revealed that 0.7% of
Br• was collected, whereas 5.9% of Cl• was collected. In other
words, Cl• underwent fewer side reactions than Br•. Further, if
the mechanism of deactivation is in fact concerted as previous
experiments suggested, the possibility of side reactions from
free Br• or Cl• is likely reduced. Therefore, such side reactions
may not be responsible for poor control in O-ATRP using
Cl−.
Radical Cation Structure. One long-standing hypothesis in
the design of O-ATRP PCs is that increasing the oxidation
potential of PC•+ increases the rate of deactivation.21,37 While
this hypothesis has motivated the development of new PCs
with strongly oxidizing radical cations,17,21 it has never been
tested. As such, another series of model reactions was
performed using 4•+ [E1/2(4•+/4) = 0.14 V vs SCE], 3•+
[E1/2(3•+/3) = 0.18 V vs SCE] and 1•+ [E1/2(1•+/1) = 0.33 V
vs SCE], which feature increasing oxidation potentials (Figure
8). A correlation was observed between the oxidation
potential of PC•+ and the rate of deactivation, supporting
the validity of this hypothesis.
(Table S8). To our surprise, all of the radical cations
investigated showed Kassoc values within roughly 1 order of
magnitude of each other (ΔΔGassoc ∼ 1 kcal mol−1), which is
only slightly outside of the error of the measurement (Table
S10). Coupled with the fact that no trends in the
conductometry data were observed, these results suggest
that the impact of PC•+ structure on ion pairing in PC•+PF6
−
is minimal at best. By contrast, the solvent appears to have a
−
much greater impact on ion pairing in PC•+PF6 . When
conductometry was performed with 1•+PF6 in four different
−
solvent systems, from DMAc to tetrahydrofuran (THF)
(Table S9), Kassoc varied over several orders of magnitude
(102−106 M−1, ΔΔGassoc ∼ 5 kcal mol−1). Further, Kassoc
varied predictably as a function of the solvent dielectric
constant, as suggested by theory (Figure S147).38 Thus, while
the structure of PC•+ appears to have only a minor influence
on ion pairing, the choice of the solvent can be very impactful.
Impact of Radical Cations in O-ATRP. Polymerization
of Methyl Methacrylate. To better understand the role of
radical cations in O-ATRP, polymerizations were conducted
in the presence of increasing quantities of PC•+. In each case,
the kinetics of the polymerizations and the resulting polymers
were characterized to understand how the addition of PC•+
impacted the reaction. Since PC•+ is the hypothesized
deactivator in O-ATRP, we anticipated that adding
supplemental PC•+ to O-ATRP would result in (1) a lower
observed rate of the polymerization; (2) more linear
molecular weight growth; (3) lower Đ (1 < Đ ≤ 1.5)
throughout the polymerization, especially at low monomer
conversions; and (4) improved initiator efficiency (I* ∼
100%).
Since core substitution of dihydrophenazine PCs by alkyl
radicals has been reported as a possible side reaction,20
control experiments were performed in the absence of LiBr to
rule out interference from these reactions (Figures S125 and
S127). Further, this experiment was also attempted with
radical cations of phenoxazines 6−8, although these reactions
proceeded too rapidly to be measured quantitatively (Figure
S128).
Ion Pairing in Radical Cations. Given that the PC•+Br−
ion pair is the hypothesized deactivator in O-ATRP, the
susceptibility of PC•+ to form this ion pair could be important
for effective deactivation during a polymerization. To probe
the variables impacting ion pairing with PC•+, the equilibrium
For initial investigations, the polymerization of MMA using
5 under published conditions39 was targeted. While DMAc
has typically been the solvent of choice for O-ATRP, ethyl
acetate was chosen given the greater stability of 5•+PF6 in
−
ethyl acetate relative to DMAc. In addition, LiBr was added to
−
association constants (Kassoc) of various PC•+PF6 salts were
these polymerizations ([LiBr] = [5] + [5•+]) to facilitate the
4732
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
a
Table 1. Initial Polymerization Results for the O-ATRP of MMA with Increasing Quantities of Supplemental Deactivator
b
c
d
d
e
entry
[5]:[5•+
]
time (h)
conv. (%)
kobs (M−1 h−1
)
Mn,theo (kDa)
Mn,exp (kDa)
Đ
I* (%)
f
1
1:0
1:0
3:1
1:1
1:3
0:1
24
24
24
24
24
24
67.5
71.3
64.8
68.1
55.4
75.8
0.046
0.053
0.046
0.044
0.039
0.037
7.01
7.40
6.74
7.07
5.80
7.84
6.19
6.71
6.38
6.87
5.60
7.36
1.19
1.14
1.16
1.12
1.16
1.10
113
110
106
103
103
107
2
3
4
5
6
a
Unless stated otherwise, [MMA]/[DBMM]/[5/5•+]/[LiBr] = [1000]:[10]:[0.1]:[0.1] (see Section S11 of the Supporting Information for full
b
c
d
1
experimental details). Determined by H NMR. Determined from the first three time points (1, 2, and 4 h). Determined by gel permeation
chromatography (GPC). Initiator efficiency (I*) = (Mn,theo/Mn,exp) × 100%. Reaction run without LiBr.
e
f
a
Table 2. Initial Polymerization Results for the O-ATRP of MA with Increasing Quantities of Supplemental Deactivator
b
c
d
d
e
entry
[5]:[5•+
]
time (h)
conv. (%)
kobs (M−1 h−1
)
Mn,theo (kDa)
Mn,exp (kDa)
Đ
I* (%)
f
7
1:0
1:0
3:1
1:1
1:3
0:1
6
6
6
6
6
6
80.7
91.6
84.2
70.6
73.3
61.4
0.40
0.89
0.58
0.29
0.28
0.17
7.20
8.14
7.50
6.33
6.57
5.54
8.59
11.6
8.09
8.40
6.98
6.45
2.21
1.77
1.88
1.67
1.65
1.55
84
70
93
75
94
86
8
9
10
11
12
a
Unless stated otherwise, [MA]/[DBMM]/[5/5•+]/[LiBr] = [1000]:[10]:[0.1]:[0.1] (see Section S12 of the Supporting Information for full
b
c
d
e
experimental details). Determined by 1H NMR. Determined from the first three time points (0.5, 1, and 1.5 h). Determined by GPC. Initiator
efficiency (I*) = (Mn,theo/Mn,exp) × 100%. Reaction run without LiBr.
f
formation of the PC•+Br− ion pair. Since the addition of Br−
salts alone has been shown to improve deactivation,26 we first
investigated the impact of this reagent on polymerization
control (Table 1, entries 1 and 2). Unsurprisingly, adding LiBr
resulted in a slight decrease in Đ (Đ = 1.14 with LiBr vs 1.19
without), although otherwise similar polymerization results.
Although high-molecular-weight polymers have been synthe-
sized by a number of other controlled radical polymerization
methods,40−45 they have remained elusive in O-ATRP. In part,
this issue may be because the concentration of initiator must
be reduced to target higher molecular weights, but the
percentage of terminated chains in ATRP is predicted to vary
inversely with the initiator concentration.46 As such, to target
higher-molecular-weight polymers, better deactivation may be
necessary to control termination reactions.47
To this end, a series of polymerizations were performed
increasing the target molecular weight (Mn,target) of the
polymer by varying the ratio of monomer to initiator. When
these polymerizations were performed starting with 5,
polymers with low Đ were consistently obtained, although
I* increased undesirably over 100% with Mn,target (Table S12).
Further, the number average molecular weight (Mn) measured
for the polymers produced was limited to about 45 kDa, after
which molecular weight growth began to plateau (Figures
S158 and S163). We anticipated that performing these
polymerizations using 5•+ instead of 5 would improve these
results, but no significant improvements were observed (Table
S12). Even after varying the [LiBr] (Table S13) and 5•+
loading (Table S14), the Mn of the product polymers
remained limited to about 45 kDa. Further work is ongoing
to determine the mechanistic cause of this limitation.
Polymerization of Acrylates. Another limitation of O-
ATRP is its monomer scope, which remains narrow in
comparison to traditional ATRP.12 While different monomers
present different challenges, acrylate monomers have been
difficult to access in O-ATRP because of their large
propagation rate constants. To compensate for an increase
in the rate of propagation with acrylates relative to
methacrylates, faster deactivation is necessary. Previously,
our group reported two strategies to access the O-ATRP of
acrylates. In the first, a new class of organic PCs,
dihydroacridines, was developed, which featured strongly
oxidizing radical cations to increase the thermodynamic
To then understand the impact of adding 5•+PF6 to this
−
polymerization, the ratio of [5]:[5•+] was varied while
maintaining the overall catalyst loading ([5] + [5•+] = 100
ppm) constant (Table 1, entries 2−6) in the presence of 100
ppm LiBr. Overall, no significant improvements in polymer-
ization control were observed upon increasing [5•+] (ex. Đ =
1.14 for [5]:[5•+] = 1:0 vs Đ = 1.10 for [5]:[5•+] = 0:1),
presumably because the polymerization with 5 already exhibits
excellent polymerization control. However, a decrease in the
rate of the polymerizations was observed, especially during the
first several hours (kobs = 0.053 M−1 s−1 for [5]:[5•+] = 1:0 vs
kobs = 0.037 M−1 s−1 for [5]:[5•+] = 0:1), consistent with
improved deactivation.
For polymerizations started with only 5•+PF6 , one might
−
expect activation to be inaccessible due to the lack of 5.
However, we hypothesized that the reaction between 5•+ and
Br− would generate a small quantity of 5, allowing the
polymerization to begin upon irradiation. Support for this
hypothesis was found when control reactions were performed
(Table S11), which showed significantly reduced conversion
in the absence of LiBr (8.3 vs 75.8%). Further, visual
inspection of this polymerization revealed that the reaction
remained dark blue even after 8 h of irradiation with white
LEDs, indicating persistence of 5•+ in solution. By contrast,
when the same polymerization was performed with LiBr
present, the dark blue solution gradually turned light green
(Figure S157), indicating a mixture of 5•+ (blue) and 5
(yellow).
In an effort to improve polymerization control in a more
challenging system, the synthesis of high-molecular-weight
poly(methyl methacrylate) (PMMA) was undertaken.
4733
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
driving force for deactivation.21 More recently, a second
strategy was reported, in which a dihydrophenazine PC was
first reacted with diethyl 2-bromo-2-methylmalonate
(DBMM) to generate a new substituted catalyst, followed
by O-ATRP using this new PC. It was discovered that the
reaction of the PC and DBMM not only led to the formation
of a more oxidizing catalyst but also generated an excess of
PC•+ prior to O-ATRP, which likely improved deactivation
during the polymerization of acrylates.20 In the present work,
we hypothesized that the addition of isolated radical cations in
the O-ATRP of acrylates would also improve polymerization
control.
= 86% for [5]:[5•+] = 0:1). However, the most significant
improvement was in the evolution of molecular weight during
the polymerization, which decreased with only 5, indicating
no molecular weight control, but increased with 5•+ (Figure
9b). Together, these results represent a significant improve-
ment in the polymerization of MA using O-ATRP, although
there are still several indicators of poor control in these
results. For example, while Đ was reduced through the use of
5•+, a controlled polymerization should exhibit Đ ≤ 1.5. In
addition, I* = 86% for the O-ATRP of MA using 5•+, but I* =
100% is most desirable.
In an effort to further improve these results, an experiment
was performed increasing [5•+] along with [LiBr] such that
[LiBr] = [5•+]. Again, increasing the concentration of the
radical cation resulted in a decrease in the rate of the
polymerization (Figure 10), indicating improved deactivation
To first probe the impact of adding PC•+ to the
polymerization of an acrylate (methyl acrylate, MA), a series
of polymerizations were performed in which [5•+] was
increased while keeping the overall catalyst loading ([5] +
[5•+] = 100 ppm) constant (Table 2). With regard to the
polymerization kinetics, increasing [5•+] resulted in increas-
ingly linear pseudo-first-order kinetics (Figure 9a) and a lower
Figure 10. Polymerization kinetics for the O-ATRP of MA catalyzed
by 5, demonstrating a decrease in polymerization rate with increasing
quantities of 5•+.
with more 5•+. In addition, improvements in polymerization
control were observed up to 200 ppm 5•+, resulting in Đ =
1.44 and I* = 102% (Table 3) vs Đ = 1.90 and I* = 78% with
200 ppm 5 and LiBr. Importantly, polymerizations performed
by O-ATRP with 200 ppm 5 remained completely
uncontrolled (Figure S172), indicating that these improve-
ments are directly attributable to the presence of PC•+.
Unfortunately, further increasing [5•+] above 200 ppm did
not provide better control in the polymerization of MA;
instead, it decreased control (Table 3, entry 16). We
hypothesized that this decrease in control might be due to a
background polymerization of MA, which is insignificant over
shorter reaction times (6−14 h) but becomes competitive at
longer reaction times (24 h). Control reactions support this
hypothesis. When both 5•+ and LiBr were removed (Table
S15, entry S22), significant conversion of the monomer to
polymer was still observed (57.1% at 14 h), and a high-
molecular-weight polymer was recovered (Mn = 467 kDa).
Significant gelling of the reaction mixture was also observed
(Figure S176), which is consistent with the free-radical
polymerization of MA. For comparison, when the same
control experiment was performed using MMA, only a 2%
conversion and no gelling of the reaction mixture were
observed.
Figure 9. (a) Polymerization kinetics for the O-ATRP of MA
catalyzed by 5, demonstrating increasingly linear pseudo-first-order
kinetics with increasing quantities of 5•+. (b) Evolution of polymer
molecular weight (Mn, filled shapes) and Đ (hollow shapes) for the
polymerization of MA by O-ATRP with 5 (blue) and 5•+ (red).
rate of the polymerization (kobs = 0.89 M−1 s−1 for [5]:[5•+] =
1:0 vs kobs = 0.17 M−1 s−1 for [5]:[5•+] = 0:1). In particular, it
is interesting that polymerizations with low [5•+] exhibited
downward sloping pseudo-first-order kinetics, as this feature is
consistent with a prevalence of termination reactions due to
poor deactivation.12 The disappearance of this feature and the
lowering of kobs with increasing [5•+] are consistent with
improved deactivation.
To balance improvements in polymerization control but
suppress this background reaction, all remaining acrylate
polymerizations were performed using 200 ppm 5•+. Under
these conditions, reaction variables were tuned in an effort to
further improve polymerization control. For example, the
Increasing [5•+] also improved control during these
polymerizations (Table 2). As [5•+] increased, Đ decreased
(Đ = 1.77 for [5]:[5•+] = 1:0 vs Đ = 1.55 for [5]:[5•+] = 0:1)
and I* approached 100% (I* = 70% for [5]:[5•+] = 1:0 vs I*
4734
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
a
Table 3. Polymerization Results for the O-ATRP of MA with Increasing Quantities of 5•+
b
c
d
d
e
entry
[5•+] (ppm)
time (h)
conv. (%)
kobs (M−1 h−1
)
Mn,theo (kDa)
Mn,exp (kDa)
Đ
I* (%)
13
14
15
16
100
150
200
250
6
10
14
24
61.4
69.3
71.3
53.5
0.16
0.11
0.08
0.02
5.54
6.22
6.39
4.86
6.45
6.97
6.25
4.26
1.55
1.44
1.44
1.83
86
89
102
114
a
For all polymerizations, [MMA]/[DBMM] = [1000]:[10] and [LiBr] = [5•+] (see Section S12 of the Supporting Information for full
b
c
d
e
experimental details). Determined by 1H NMR. Determined from the first 6 h. Determined by GPC. Initiator efficiency (I*) = (Mn,theo/Mn,exp
× 100%.
)
a
Table 4. Results from the Polymerization of Various Acrylate Monomers by O-ATRP Using 5•+
b
c
c
d
entry
monomer
time (h)
conv. (%)
Mn,theo (kDa)
Mn,exp (kDa)
Đ
I* (%)
17
18
19
20
21
EA
nBA
tBA
EHA
EGMEA
14
14
14
14
14
64.2
64.4
84.1
87.1
87.0
6.68
8.51
11.0
16.3
11.6
6.72
8.89
9.36
17.6
13.5
1.47
1.47
1.74
1.85
1.60
99
96
118
93
86
a
In all cases, [monomer]/[DBMM]/[5•+]/[LiBr] = [1000]:[10]:[0.2]:[0.2] (see Section S12 of the Supporting Information for full experimental
b
c
d
1
details). Determined by H NMR. Determined by GPC. Initiator efficiency (I*) = (Mn,theo/Mn,exp) × 100%.
choice of the solvent can have significant effects in O-
ATRP,20,30,48 presumably by impacting the photophysics of
the PC and ion pairing in PC•+Br−. However, no improve-
ments in O-ATRP using 5•+ were obtained by changing the
solvent. Using THF, similar results were obtained as with
ethyl acetate (Table S16). By contrast, using DMAc led to a
complete loss of control (Table S16, entry S31), likely
because 5•+ reacts with DMAc and decomposes to 5. As a
result, polymerizations in this solvent are more analogous to
traditional O-ATRP using 5.
In addition, the quantity of LiBr was varied while
maintaining a constant [5•+]. We hypothesized that increasing
[LiBr] would improve polymerization control by further
encouraging deactivation. However, it is also possible that
adding more LiBr to the polymerization might increase the
rate of the side reaction between 5•+ and Br−, leading to faster
decomposition of 5•+ to 5. In this case, decreasing [LiBr]
might be more advantageous, as it might increase the lifetime
of 5•+ and improve deactivation during later reaction times.
To test these hypotheses, polymerizations were performed
varying the ratio of [5•+]/[LiBr] from [1]:[0.1] to [1]:[10]
(Table S17), but no improvements in polymerization control
were observed.
Finally, one important feature of all ATRP methods is the
retention of the C−Br bonds at the ends of the polymer
chains. This feature, termed chain-end group fidelity, is key for
subsequent functionalization of the polymers produced by
ATRP, such as by chain extension or block copolymer
synthesis. As such, the chain-end fidelity of poly(methyl
acrylate) (pMA) synthesized using 5•+ was characterized and
compared to pMA synthesized with 5. We anticipated that the
use of 5•+ would yield superior chain-end fidelity since
improving deactivation suppresses the termination reactions
that cause loss of the Br functionality.
To investigate this property, we first synthesized pMA
under optimized conditions using both 5 and 5•+ (Table S18).
The resulting polymers were characterized by 1H NMR
(Figures S188 and S189), which was consistent with the
expected spectrum for pMA. To identify the chain ends
arising from each set of polymerization conditions, the
polymers were characterized by matrix-assisted laser desorp-
tion ionization time-of-flight mass spectrometry (MALDI-
TOF MS). For pMA synthesized using 5, two peak
distributions were observed, corresponding to two sets of
end groups. The first distribution corresponded to polymers
capped by the DBMM-derived malonate moiety and a Br end
group. Instead, the second set of peaks was consistent with
polymers containing the same malonate group and a H end
group. Together, these results indicate that while some of the
Br chain-end groups are retained, some loss of the Br
functionality is also present. For pMA synthesized with 5•+,
two distinct peak distributions were also observed. One set
corresponded to the DBMM-derived malonate group on one
end and Br on the other, the expected end groups. The other
peak distribution corresponded to H and Br end groups,
which can be explained by the proposed background
polymerization, a free-radical polymerization that is ultimately
suppressed by deactivation. Alternatively, the same end groups
We next sought to understand whether 5•+ could be applied
to the O-ATRP of other acrylate monomers. In total, five
other acrylates were polymerized in this manner (Table 4).
For monomers with shorter alkyl chains, ethyl acrylate (EA)
and n-butyl acrylate (nBA), similar polymerization results
were obtained as with MA. However, increasing the length of
the alkyl chain led to a decrease in polymerization control
(entry 20). In part, this observation can be attributed to the
increase in the rate of propagation of acrylate monomers with
longer alkyl chains.49 In addition, increasing the length of the
monomer alkyl substituent likely lowers the overall polarity of
the polymerization solution, which might impact PC photo-
physics and ion pairing in PC•+Br−.
4735
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
Figure 11. Synthesis of acrylate block copolymers by O-ATRP with 5 (a) and 5•+ (b). GPC traces correspond to isolated and dried polymers as
measured using a differential refractive index detector (see Section S12 of the Supporting Information for full experimental details).
could also arise from polymers initiated by Br• that ultimately
undergo irreversible termination.
chemical, and X-ray diffraction techniques. To understand
their role and possible side reactions in O-ATRP, the
reactivity of these compounds was investigated in solution,
in deactivation model reactions, and in O-ATRP. Under the
appropriate conditions, we discovered that these compounds
can exhibit reactivity from both the ground state and a
photoexcited state. However, the mechanism of this excited-
state reactivity remains unclear, and deeper investigation of
radical cation photophysics is necessary to understand this
interesting phenomenon.
To further investigate chain-end fidelity in the absence and
presence of 5•+, the synthesis of block copolymers was
attempted using previously isolated pMA as a macroinitiator
in place of DBMM. When pMA synthesized using 5 was
resubjected to polymerization conditions in the presence of
MA, the resulting polymer was nearly identical to the pMA
macroinitiator (Figure 11a, blue). Similarly, when MA was
replaced by tert-butyl acrylate (tBA), only a minor shift in the
chromatogram of the block copolymer was observed relative
to the macroinitiator (Figure 11a, red). These results indicate
a significant loss of the Br chain ends during the O-ATRP of
MA.
Using a deactivation model reaction, the ability of one PC•+
to deactivate alkyl radicals was demonstrated by identification
of the expected deactivation product. This model reaction was
further used to investigate the impact of various factors on
deactivation kinetics, such as the identity of the halide or the
structure of PC•+. Ultimately, four main conclusions were
drawn from these experiments: (1) PC•+Br− is likely the
deactivator in O-ATRP; (2) the mechanism of deactivation
appears to be concerted, where PC•+Br− undergoes a
bimolecular reaction with the propagating radical; (3)
deactivation with Br− is faster than with Cl−, likely because
Cl− is more challenging to oxidize; and (4) the oxidation
potential of PC•+ correlates with the rate of deactivation, such
that more oxidizing radical cations exhibit faster deactivation.
By contrast, when pMA synthesized with 5•+ was
resubjected to polymerization conditions in the presence of
MA and tBA, clear evidence was found supporting the chain
extension of this macroinitiator. In the case of pMA-block-
pMA, a small shift in the chromatogram was observed (Figure
11b, blue) and the polymer molecular weight (Mn = 7.80
kDa) increased relative to the macroinitiator (Mn = 4.21 kDa).
For pMA-b-ptBA, a more significant shift in the chromato-
gram (Figure 11b, red) and an increase in the copolymer
molecular weight (Mn = 13.1 vs 2.96 kDa for pMA) were
observed, providing evidence for improved chain-end fidelity
for the polymerization using 5•+.
−
When ion pairing in PC•+PF6 was investigated by
conductometry, the structure of PC•+ was found to have
only a minor impact on the strength of ion pairing. However,
the polarity of the solvent significantly influences ion pairing,
supporting the importance of solvent choice in O-ATRP.
CONCLUSIONS
Radical cations of O-ATRP catalysts were synthesized and
characterized by a combination of spectroscopic, electro-
■
4736
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
Finally, the impact of radical cations on polymerization
control in O-ATRP was investigated with two different
monomers. While only limited improvements in polymer-
ization control were observed with MMA, presumably because
this system is already well controlled in the absence of added
PC•+, significant improvements in the polymerization of MA
were achieved by performing O-ATRP with PC•+ instead of
PC. Ultimately, this work demonstrates the importance of
radical cations for deactivation in O-ATRP and shows how
limitations in this polymerization method can be overcome by
understanding their reactivity.
Award). B.G.M. acknowledges support from the NSF GRFP.
D.A.C. would like to thank Dr. Scott Folkman for helpful
discussions of electrochemistry; Dr. Brian Newell, Cassidy
Jackson, and Anthony Campanella for assistance with
crystallography; and Cassidy Jackson for support throughout
this work.
REFERENCES
■
(1) Flory, P. J. Molecular size distribution in ethylene oxide
polymers. J. Am. Chem. Soc. 1940, 62, 1561−1565.
(2) Szwarc, M. Living polymers. Nature 1956, 178, 1168−1169.
(3) Matyjaszewski, K.; Muller, A. H. E. 50 Years of living
polymerization. Prog. Polym. Sci. 2006, 31, 1039−1040.
(4) Grubbs, R. B.; Grubbs, R. H. 50th Anniversary perspective:
living polymerization-emphasizing the molecule in macromolecules.
Macromolecules 2017, 50, 6979−6997.
ASSOCIATED CONTENT
■
sı
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.macromol.1c00640.
(5) Hawker, C. J.; Wooley, K. L. The convergence of synthetic
organic and polymer chemistries. Science 2005, 309, 1200−1205.
(6) Kato, M.; Kamigaito, M.; Sawamoto, M.; Higashimura, T.
Polymerization of methyl-methacrylate with the carbon-tetrachloride
dichlorotris(triphenylphosphine)ruthenium(ii) methylaluminum bis-
(2,6-di-tert-butylphenoxide) initiating system - possibility of living
radical polymerization. Macromolecules 1995, 28, 1721−1723.
(7) Wang, J. S.; Matyjaszewski, K. Controlled living radical
polymerization - atom-transfer radical polymerization in the presence
of transition-metal complexes. J. Am. Chem. Soc. 1995, 117, 5614−
5615.
(8) Solomon, D. H.; Rizzardo, E.; Cacioli, P. Polymerization
Process and Polymers Produced Thereby. US4581429, July 11, 1984.
(9) Chiefari, J.; Chong, Y. K.; Ercole, F.; Krstina, J.; Jeffery, J.; Le,
T. P. T.; Mayadunne, R. T. A.; Meijs, G. F.; Moad, C. L.; Moad, G.;
Rizzardo, E.; Thang, S. H. Living free-radical polymerization by
reversible addition-fragmentation chain transfer: the RAFT process.
Macromolecules 1998, 31, 5559−5562.
(10) Miyake, G. M.; Theriot, J. C. Perylene as an organic
photocatalyst for the radical polymerization of functionalized vinyl
monomers through oxidative quenching with alkyl bromides and
visible light. Macromolecules 2014, 47, 8255−8261.
(11) Treat, N. J.; Sprafke, H.; Kramer, J. W.; Clark, P. G.; Barton,
B. E.; de Alaniz, J. R.; Fors, B. P.; Hawker, C. J. Metal-free atom
transfer radical polymerization. J. Am. Chem. Soc. 2014, 136, 16096−
16101.
Materials and methods; procedures; characterization of
photocatalysts and radical cations; radical cation
stability studies; investigation of radical cation excited-
state reactivity; radical cation reactivity toward halides;
deactivation of alkyl radicals by radical cations;
investigation of hydrogen atom abstraction side
reactions; ion pairing in radical cation hexafluorophos-
phate salts; general polymerization procedures; supple-
mental data for the polymerization of methyl meth-
acrylate and methyl acrylate; and molecular coordinates
for substrate oxidation calculations (PDF)
−
Crystallographic data file of 4•+SbCl6− (CIF)
Crystallographic data file of 9•+SbCl6 (CIF)
AUTHOR INFORMATION
Corresponding Author
Garret M. Miyake − Department of Chemistry, Colorado
State University, Fort Collins, Colorado 80523-1872, United
States; orcid.org/0000-0003-2451-7090;
Email: garret.miyake@colostate.edu
■
Authors
Daniel A. Corbin − Department of Chemistry, Colorado State
University, Fort Collins, Colorado 80523-1872, United
States
Blaine G. McCarthy − Department of Chemistry, Colorado
State University, Fort Collins, Colorado 80523-1872,
United States
Zach van de Lindt − Department of Chemistry, Colorado
State University, Fort Collins, Colorado 80523-1872,
United States
(12) Krys, P.; Matyjaszewski, K. Kinetics of atom transfer radical
polymerization. Eur. Polym. J. 2017, 89, 482−523.
(13) Matyjaszewski, K. Atom transfer radical polymerization
(ATRP): current status and future perspectives. Macromolecules
2012, 45, 4015−4039.
(14) Matyjaszewski, K. Advanced materials by atom transfer radical
polymerization. Adv. Mater. 2018, 30, No. 1706441.
(15) Pan, X. C.; Lamson, M.; Yan, J. J.; Matyjaszewski, K.
Photoinduced metal-free atom transfer radical polymerization of
acrylonitrile. ACS Macro Lett. 2015, 4, 192−196.
(16) Theriot, J. C.; Lim, C. H.; Yang, H.; Ryan, M. D.; Musgrave,
C. B.; Miyake, G. M. Organocatalyzed atom transfer radical
polymerization driven by visible light. Science 2016, 352, 1082−1086.
(17) Pearson, R. M.; Lim, C. H.; McCarthy, B. G.; Musgrave, C. B.;
Miyake, G. M. Organocatalyzed atom transfer radical polymerization
using N-aryl phenoxazines as photoredox catalysts. J. Am. Chem. Soc.
2016, 138, 11399−11407.
(18) Chen, D. F.; Boyle, B. M.; McCarthy, B. G.; Lim, C. H.;
Miyake, G. M. Controlling polymer composition in organocatalyzed
photoredox radical ring-opening polymerization of vinylcyclopro-
panes. J. Am. Chem. Soc. 2019, 141, 13268−13277.
(19) Su, X.; Jessop, P. G.; Cunningham, M. F. Versatility of
organocatalyzed atom transfer radical polymerization and CO2-
switching for preparing both hydrophobic and hydrophilic polymers
with the recycling of a photocatalyst. Macromolecules 2019, 52,
6725−6733.
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.macromol.1c00640
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Research reported in this publication was supported by the
National Institute of General Medical Sciences (Award
R35GM119702) of the National Institutes of Health. The
content is solely the responsibility of the authors and does not
necessarily represent the official views of the National
Institutes of Health. The authors are grateful for support
provided by Colorado State University and the Research
Corporation for Science Advancement (Cottrell Scholar
4737
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738

Macromolecules
pubs.acs.org/Macromolecules
Article
(20) McCarthy, B.; Sartor, S.; Cole, J.; Damrauer, N.; Miyake, G.
M. Solvent effects and side reactions in organocatalyzed atom
transfer radical polymerization for enabling the controlled polymer-
ization of acrylates catalyzed by diaryl dihydrophenazines. Macro-
molecules 2020, 53, 9208−9219.
(21) Buss, B. L.; Lim, C.-H.; Miyake, G. M. Dimethyl
dihydroacridines as photocatalysts in organocatalyzed atom transfer
radical polymerization of acrylate monomers. Angew. Chem., Int. Ed.
2020, 59, 3209−3217.
(22) Corbin, D. A.; McCarthy, B. G.; Miyake, G. M. Impacts of
performing electrolysis during organocatalyzed atom transfer radical
polymerization. Polym. Chem. 2020, 11, 4978−4985.
(23) Jockusch, S.; Yagci, Y. The active role of excited states of
phenothiazines in photoinduced metal free atom transfer radical
polymerization: singlet or triplet excited states? Polym. Chem. 2016,
7, 6039−6043.
(24) Koyama, D.; Dale, H. J. A.; Orr-Ewing, A. J. Ultrafast
observation of a photoredox reaction mechanism: photoinitiation in
organocatalyzed atom-transfer radical polymerization. J. Am. Chem.
Soc. 2018, 140, 1285−1293.
(25) Lewis-Borrell, L.; Sneha, M.; Bhattacherjee, A.; Clark, I. P.;
Orr-Ewing, A. J. Mapping the multi-step mechanism of a photoredox
catalyzed atom-transfer radical polymerization reaction by direct
observation of the reactive intermediates. Chem. Sci. 2020, 11, 4475−
4481.
(26) Pan, X. C.; Fang, C.; Fantin, M.; Malhotra, N.; So, W. Y.;
Peteanu, L. A.; Isse, A. A.; Gennaro, A.; Liu, P.; Matyjaszewskit, K.
Mechanism of photoinduced metal-free atom transfer radical
polymerization: experimental and computational studies. J. Am.
Chem. Soc. 2016, 138, 2411−2425.
(27) Sartor, S. M.; McCarthy, B. G.; Pearson, R. M.; Miyake, G. M.;
Damrauer, N. H. Exploiting charge-transfer states for maximizing
intersystem crossing yields in organic photoredox catalysts. J. Am.
Chem. Soc. 2018, 140, 4778−4781.
(28) Sartor, S. M.; Lattke, Y. M.; McCarthy, B. G.; Miyake, G. M.;
Damrauer, N. H. Effects of naphthyl connectivity on the photo-
physics of compact organic charge-transfer photoredox catalysts. J.
Phys. Chem. A 2019, 123, 4727−4736.
(29) Sartor, S. M.; Chrisman, C. H.; Pearson, R. M.; Miyake, G. M.;
Damrauer, N. H. Designing high-triplet-yield phenothiazine donor-
acceptor complexes for photoredox catalysis. J. Phys. Chem. A 2020,
124, 817−823.
(30) Lim, C. H.; Ryan, M. D.; McCarthy, B. G.; Theriot, J. C.;
Sartor, S. M.; Damrauer, N. H.; Musgrave, C. B.; Miyake, G. M.
Intramolecular charge transfer and ion pairing in N,N-diaryl
dihydrophenazine photoredox catalysts for efficient organocatalyzed
atom transfer radical polymerization. J. Am. Chem. Soc. 2017, 139,
348−355.
on catalyst design and performance. Macromol. Rapid Commun. 2017,
38, No. 1700040.
(38) Fuoss, R. M. Ionic association.3. The equilibrium between ion
pairs and free ions. J. Am. Chem. Soc. 1958, 80, 5059−5061.
(39) Cole, J. P.; Federico, C. R.; Lim, C. H.; Miyake, G. M.
Photoinduced organocatalyzed atom transfer radical polymerization
using low ppm catalyst loading. Macromolecules 2019, 52, 747−754.
(40) Carmean, R. N.; Becker, T. E.; Sims, M. B.; Sumerlin, B. S.
Ultra-high molecular weights via aqueous reversible-deactivation
radical polymerization. Chem 2017, 2, 93−101.
(41) Xu, J. T.; Jung, K.; Atme, A.; Shanmugam, S.; Boyer, C. A
robust and versatile photoinduced living polymerization of
conjugated and unconjugated monomers and its oxygen tolerance.
J. Am. Chem. Soc. 2014, 136, 5508−5519.
(42) Gong, H. H.; Gu, Y.; Zhao, Y. C.; Quan, Q. Z.; Han, S. T.;
Chen, M. Precise synthesis of ultra-high-molecular-weight fluoropol-
ymers enabled by chain-transfer-agent differentiation under visible-
light irradiation. Angew. Chem., Int. Ed. 2020, 59, 919−927.
(43) Mueller, L.; Jakubowski, W.; Matyjaszewski, K.; Pietrasik, J.;
Kwiatkowski, P.; Chaladaj, W.; Jurczak, J. Synthesis of high molecular
weight polystyrene using AGET ATRP under high pressure. Eur.
Polym. J. 2011, 47, 730−734.
(44) Percec, V.; Guliashvili, T.; Ladislaw, J. S.; Wistrand, A.;
Stjerndahl, A.; Sienkowska, M. J.; Monteiro, M. J.; Sahoo, S. Ultrafast
synthesis of ultrahigh molar mass polymers by metal-catalyzed living
radical polymerization of acrylates, methacrylates, and vinyl chloride
mediated by SET at 25 degrees C. J. Am. Chem. Soc. 2006, 128,
14156−14165.
(45) Rzayev, J.; Penelle, J. HP-RAFT: A free-radical polymerization
technique for obtaining living polymers of ultrahigh molecular
weights. Angew. Chem., Int. Ed. 2004, 43, 1691−1694.
(46) Matyjaszewski, K. Comparison and Classification of Con-
trolled/Living Radical Polymerizations. In Controlled/Living Radical
Polymerization: Progress in ATRP, NMP, and RAFT; Matyjaszewski,
K., Ed.; ACS Symposium Series; American Chemical Society, 2000;
Vol. 768, pp 2−26.
(47) Kwiatkowski, P.; Jurczak, J.; Pietrasik, J.; Jakubowski, W.;
Mueller, L.; Matyjaszewski, K. High molecular weight polymetha-
crylates by AGET ATRP under high pressure. Macromolecules 2008,
41, 1067−1069.
(48) Ryan, M. D.; Theriot, J. C.; Lim, C. H.; Yang, H. S.;
Lockwood, A. G.; Garrison, N. G.; Lincoln, S. R.; Musgrave, C. B.;
Miyake, G. M. Solvent effects on the intramolecular charge transfer
character of N,N-diaryl dihydrophenazine catalysts for organo-
catalyzed atom transfer radical polymerization. J. Polym. Sci., Part
A: Polym. Chem. 2017, 55, 3017−3027.
(49) Beuermann, S.; Buback, M. Rate coefficients of free-radical
polymerization deduced from pulsed laser experiments. Prog. Polym.
Sci. 2002, 27, 191−254.
(31) Connelly, N. G.; Geiger, W. E. Chemical redox agents for
organometallic chemistry. Chem. Rev. 1996, 96, 877−910.
(32) Christensen, J. A.; Phelan, B. T.; Chaudhuri, S.; Acharya, A.;
Batista, V. S.; Wasielewski, M. R. Phenothiazine radical cation excited
states as super-oxidants for energy-demanding reactions. J. Am. Chem.
Soc. 2018, 140, 5290−5299.
(33) Parkin, G. Applications of deuterium isotope effects for
probing aspects of reactions involving oxidative addition and
reductive elimination of H-H and C-H bonds. J. Labelled Compd.
Radiopharm. 2007, 50, 1088−1114.
(34) Kolthoff, I. M.; Coetzee, J. F. Polarography in acetonitrile.2.
Metal ions which have significantly different polarographic properties
in acetonitrile and in water - anodic waves - voltammetry at rotated
platinum electrode. J. Am. Chem. Soc. 1957, 79, 1852−1858.
(35) Espenson, J. H. Chemical Kinetics and Reaction Mechanisms,
2nd ed.; McGraw-Hill: New York, NY, 2002.
(36) Luo, Y.-R. Comprehensive Handbook of Chemical Bond Energies;
CRC Press: Boca Raton, 2007.
(37) Theriot, J. C.; McCarthy, B. G.; Lim, C. H.; Miyake, G. M.
Organocatalyzed atom transfer radical polymerization: perspectives
4738
https://doi.org/10.1021/acs.macromol.1c00640
Macromolecules 2021, 54, 4726−4738