Mechanism of Photoinduced Metal-Free Atom Transfer Radical Polymerization: Experimental and Computational Studies

Article abstract of DOI:10.1021/jacs.5b13455

Photoinduced metal-free atom transfer radical polymerization (ATRP) of methyl methacrylate was investigated using several phenothiazine derivatives and other related compounds as photoredox catalysts. The experiments show that all selected catalysts can b

Full text of DOI:10.1021/jacs.5b13455

Subscriber access provided by University of South Dakota
Article
Mechanism of Photoinduced Metal-Free Atom Transfer Radical
Polymerization: Experimental and Computational Studies
Xiangcheng Pan, Cheng Fang, Marco Fantin, Nikhil Malhotra , Woong Young So, Linda
A. Peteanu, Abdirisak A. Isse, Armando Gennaro, Peng Liu, and Krzysztof Matyjaszewski
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.5b13455 • Publication Date (Web): 28 Jan 2016
Downloaded from http://pubs.acs.org on January 28, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of the American Chemical Society is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 16
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Mechanism of Photoinduced Metal‐Free Atom Transfer Radical
Polymerization: Experimental and Computational Studies
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Xiangcheng Pan1, Cheng Fang2, Marco Fantin1,3, Nikhil Malhotra1, Woong Young So1, Linda A. Pe-
teanu1, Abdirisak A. Isse3, Armando Gennaro3,*, Peng Liu2,*, and Krzysztof Matyjaszewski1,*
1Department of Chemistry, Carnegie Mellon University, Pittsburgh, Pennsylvania, 15213, United States
2Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania, 15260, United States
3Department of Chemical Sciences, University of Padova, via Marzolo 1, 35131 Padova, Italy
ABSTRACT: Photoinduced metal-free atom transfer radical polymerization (ATRP) of methyl methacrylate was investigated us-
ing several phenothiazine derivatives and other related compounds as photoredox catalysts. The experiments show that all selected
catalysts can be involved in the activation step, but not all of them participated efficiently in the deactivation step. The redox prop-
erties and the stability of radical cations derived from the catalysts were evaluated by cyclic voltammetry. Laser flash photolysis
(LFP) was used to determine the lifetime and activity of photoexcited catalysts. Kinetic analysis of the activation reaction according
to dissociative electron transfer (DET) theory suggests that the activation occurs only with an excited state of catalyst. Density
functional theory (DFT) calculations revealed the structures and stabilities of the radical cation intermediates as well as the reaction
energy profiles of deactivation pathways with different photoredox catalysts. Both experiments and calculations suggest that the
activation process undergoes a DET mechanism while an associative electron transfer involving a termolecular encounter (the exact
reverse of DET pathway) is favored in the deactivation process. This detailed study provides a deeper understanding of the chemi-
cal processes of metal-free ATRP that can aid the design of better catalytic systems. Additionally, this work elucidates several im-
portant common pathways involved in synthetically useful organic reactions catalyzed by photoredox catalysts.
INTRODUCTION
chemically mediated ATRP (eATRP)11 and photochemically
mediated ATRP (photoATRP)12 were developed to regulate
the polymerization by controlling the external stimuli. In the
last two cases, the regeneration of CuI activator is controlled
by external stimulation, but the rate constant of radical for-
mation (activation) is not affected.12k Therefore, the polymeri-
zation stops after activator is consumed by oxidation or termi-
nation but not immediately after removing the external stimuli.
Reversible deactivation radical polymerization (RDRP) proce-
dures, also termed controlled or living radical polymerization,
(CRP or LRP), provide well-defined polymers with complex
architectures.1 RDRP methods include nitroxide-mediated
polymerization (NMP),2 atom transfer radical polymerization
(ATRP),3 and reversible addition−fragmentation chain-transfer
(RAFT) polymerization.4 ATRP is the most extensively used
and widely investigated method due to the commercial availa-
bility of various initiators and catalysts.
Scheme 1. ATRP equilibria in a) normal ATRP with Cu-
based catalyst, b) low ppm Cu system. M: monomer.
Control over the polymer structure and suppression of radical
termination in an ATRP are due to a concurrent growth of all
chains and fast activation/deactivation equilibrium in which a
transition metal complex (usually [CuIL]+, L = ligand) acti-
vates reversibly an alkyl halide (Pn-X, X = Br or Cl), providing
an alkyl radical and the metal complex in a higher oxidation
state [X–CuIIL]+ (Scheme 1a).5 The alkyl radical could add up
to a few monomer units before it abstracts the halogen back
from a deactivator [X–CuIIL]+ to reform the dormant alkyl hal-
ide and the activator [CuIL]+. Originally, ATRP required rela-
tively large concentrations (1,000 to 10,000 ppm) of Cu-based
catalysts to compensate for radical termination reactions and
due to relatively low activities of the catalyst complexes. Sig-
nificant progress has been recently achieved allowing use of
low ppm levels of Cu catalysts in ATRP (Scheme 1b).6 These
systems employ reducing agents for activators regenerated by
electron transfer (ARGET) ATRP,7 conventional radical initia-
tors as in initiators for continuous activator regeneration
(ICAR) ATRP,8 or zerovalent metals as supplemental activa-
tors and reducing agents (SARA) ATRP,9 also termed SET-
LRP.10 Additionally, nonchemical methods such as electro-
Activation rate constants for some catalysts can be enhanced
by light,13 especially for photoredox catalysts.14 Photoredox
catalysts have been extensively investigated for water split-
ting,15 solar cells16 and photodynamic therapy17 in inorganic
and materials chemistry. Recently, photoredox catalysts were
also used in organic synthesis18 and polymerization,14,19 espe-
cially photoinduced electron transfer (PET)-RAFT.20 General-
ly, photoredox catalysts behave as strong oxidants and/or re-
ductants upon irradiation but they are poor oxidants and re-
ductants in the ground state. Therefore, reactions with photo-
redox catalysts could be precisely controlled by light.
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 16
A photoinduced ATRP was successfully catalyzed by fac-
The ISET vs. OSET dichotomy for Cu-based ATRP was pre-
[Ir(ppy)3] (1, ppy = 2-pyridylphenyl, in Figure 1).14 A simpli-
fied proposed mechanism for this process is shown in Scheme
2. Upon irradiation with visible light, excited fac-[Ir(ppy)3]*
viously analyzed using modified Marcus theory.27 It was con-
cluded that OSET should be ~ 109 times slower than experi-
mentally measured ISET.26 This is due to very high activation
energy of ET to alkyl halides that are typically used in Cu-
based ATRP systems. ET to these alkyl halides proceeds via
the dissociative process with a high contribution of the break-
ing bond to the activation free energy of the reaction.28 fac-
[Ir(ppy)3] 1 is a coordinatively saturated metal complex, there-
fore it cannot form any additional bond with the metal center,
and thus the ET most likely takes place via an OSET.29 In pho-
toinduced Ph-PTZ-catalyzed metal-free ATRP system, the
electron transfer (activation step) from excited Ph-PTZ* to
alkyl halide most likely proceeds via OSET due to the tremen-
dously negative redox potentials of the excited species (ꢀꢄꢃ
1
2
3
4
5
6
7
8
(1*), a very strong reductant, ꢀꢁꢃ
= - 1.73 V vs. SCE, can
ꢂ
∗
/ꢁ
reduce an alkyl bromide to generate an IrIV complex and an
organic radical which initiates polymerization. The IrIV com-
plex is a strong oxidant (ꢀꢁꢃ /ꢁ = 0.77 V vs. SCE), which could
ꢂ
react with the propagating radical to provide the ground state
catalyst 1 and polymer chain with a bromine at a chain end.
The radical could be either oxidized to RX by a concerted at-
om transfer step, or oxidized to the carbocation which then
recombines with bromide anion in a stepwise manner. Well-
defined poly(methyl methacrylate) (PMMA, Mn = 22,900,
Mw/Mn = 1.25) was obtained using low ppm amounts of 1 un-
der visible light irradiation. This system was then extended to
polymerize several acrylates21 as well as provide three-
dimensionally controlled nanostructures in a single step.22
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
•ꢂ
∗
/ꢄ
= – 2.10 V vs. SCE for 2, Ph-PTZ vs. Eo = – 0.18 V vs. SCE
for 3, [CuI(TPMA*3)]+). It should be noted that 59 mV corre-
sponds to one order of magnitude difference in equilibrium
constants for electron transfer reactions.
The mechanism of the deactivation process also plays an im-
portant role in controlling an ATRP reaction. In atom transfer
radical addition (ATRA) reactions by photoredox catalysts
such as 1, it was proposed that R is first oxidized to a carbo-
cation,30 which subsequently traps a nucleophilic halide anion
to yield the product. However, in the polymerization of methyl
methacrylate (MMA) and acrylonitrile (AN), the derived car-
bocations should be unstable and would be involved in side
reactions with residual water or elimination to form short oli-
gomers rather than polymers. No such products were observed
in photoinduced metal-free ATRP, indicating that the deacti-
vation step might not involve a carbocation as a key interme-
diate.
Figure 1. Structures of photoredox catalysts 1, 2 and a traditional
copper-based ATRP catalyst 3.
A metal-free ATRP process was subsequently developed by
using 10-phenylphenothiazine (2, Ph-PTZ, in Figure 1) as the
organic-based photoredox catalyst to synthesize well-defined
polymethacrylates23 and polyacrylonitrile.24 Analogous to 1,
Ph-PTZ 2 is also excited to form a very strong reductant Ph-
We performed a detailed mechanistic study on photoinduced
metal-free ATRP to identify structure-reactivity relationships,
using cyclic voltammetry (CV) and laser flash photolysis
(LFP) experiments. Kinetic analysis of both activation and de-
activation steps, according to Marcus theory and further de-
velopments27 and density functional theory (DFT) calculations
elucidated the following questions: 1) Does the activation step
follow ISET or OSET mechanism? 2) What is the mechanism
of the deactivation process during the controlled polymeriza-
tion? 3) What are the key intermediates in these reactions? 4)
What side reactions are involved? 5) How does this photoin-
duced metal-free ATRP system compare to classic Cu-
catalyzed ATRP?
PTZ* (ꢀꢄꢃ
= - 2.10 V vs. SCE in MeCN). The oxidized
•ꢂ
∗
/ꢄ
radical cation Ph-PTZ●+, formed upon reaction of Ph-PTZ*
with the alkyl halide, is a strong oxidant (ꢀꢄꢃ /ꢄ = 0.68 V vs.
•ꢂ
SCE in MeCN), able to deactivate the propagating alkyl radi-
cals and regenerate the ground state catalyst 2.
Scheme 2. Simplified activation/deactivation mechanism
for photoredox mediated ATRP reactions. Cat = 1 or 2.
RESULTS AND DISCUSSION
Polymerization reactions
Polymerization of MMA with Ph-PTZ 2. The results of pho-
toinduced metal-free ATRP of MMA with Ph-PTZ 2 under
different conditions, in various solvents, under different light
intensities, and in the presence of different ATRP initiators,
are summarized in Table 1. The standard polymerization under
conditions: [MMA]0:[EBPA]0:[2]0 = 100:1:0.1, MMA/DMA =
1/1 (v/v), (EBPA: ethyl α-bromophenylacetate, DMA: dime-
thylacetamide), at room temperature with irradiation at 365
nm (2.1 mW/cm2) reached 16% conversion of MMA after 4 h,
yielding PMMA with Mn = 2070, and Mw/Mn = 1.50 (entry 1,
Table 1) which is close to the theoretical value Mn,th (predicted
for a transferless process with a quantitative initiation). A
polymerization with stronger light intensity source (4.9
mW/cm2 at 365 nm) was faster, reaching 45 % conversion af-
ter 4 h (entry 2). The polymerization carried out in DMSO
The most active Cu-based ATRP catalyst reported so far is
[CuI(TPMA*3)]+ (3, TPMA*3
=
tris((4-methoxy-3,5-
dimethylpyridin-2-yl)methyl)-amine)) which has ꢀꢃ = -0.18 V
vs. SCE.25 Compared to these values, both 1 and 2 have much
more negative potential values (Figure 1), indicating much
greater reactivity in the activation of alkyl halides. They are so
active that they can participate in both outer- and inner-sphere
electron transfer (OSET and ISET) processes, while activation
in Cu-based ATRP occurs via ISET only.26
2
ACS Paragon Plus Environment

Page 3 of 16
Journal of the American Chemical Society
Table 1. Selected results of metal-free ATRP of MMA under different conditions.a
1
2
3
4
5
6
c
c
d
Entry
Conditions
[MMA]:[EBPA]:[2] = 100:1:0.1
[MMA]:[EBPA]:[2] = 100:1:0.1e
[MMA]:[EBPA]:[2] = 100:1:0.1f
Time
4 h
Conv.b
16%
Mn,th
Mn,GPC
2,070
5,440
2,500
Mw/Mn
1.50
1
2
3
1,800
4,700
1,400
4 h
45%
1.44
6 h
12%
1.98
[MMA]:[EBPA]:[2] = 100:1:0.1g
[MMA]:[EBiB]:[2] = 100:1:0.1
[MMA]:[EClPA]:[2] = 100:1:0.1e
[MMA]:[EBPA]:[2] = 100:1:0
[MMA]:[EClPA]:[2] = 100:1:0e
[MMA]:[EClPA]:[2]:[TBABr] = 100:1:0.1:1e
20 h
4 h
4 h
4 h
4 h
4 h
9%
1,100
2,200
5,700
2,900
1,700
8,100
1,120
3,840
1.78
1.79
3.44
2.25
2.02
2.47
7
8
9
4
5
6
7
8
9
20%
55%
27%
15%
79%
16,000
28,700
17,900
10,800
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a Reaction conditions: MMA/DMA = 1/1 (v/v), under room temperature irradiation by 2.1 mW/cm2, 365 nm. b Determined by 1H NMR.
c
1
d
Calculated based on conversion obtained by H NMR (i.e. Mn,th = MEBPA + 100 × conversion × MMMA). Determined by GPC in THF,
based on linear PMMA as calibration standards. e With 4.9 mW/cm2, 365 nm irradiation. f In DMSO. g In MeCN.
Figure 2. a) Left, semilogarithmic kinetic plots of polymerization of MMA with or without 2; b) right, number-average molecular weight
(Mn, filled symbols), and dispersity (Mw/Mn, open symbols) versus conversion.
resulted in a polymer with Mn higher than Mn,th and with a
broader molecular weight distribution (Mw/Mn = 1.98, entry 3)
than the reaction in DMA, indicating limited initiation effi-
ciency. The reaction in MeCN reached only 9% conversion
after 20 h (entry 4), showing a much slower polymerization
than in DMA and DMSO. With ethyl α-bromoisobutyrate
(EBiB) as ATRP initiator instead of EBPA, the Mn = 3840 of
PMMA was higher than Mn,th with broader distribution (Mw/Mn
= 1.79, entry 5). Indeed, activation of EBiB in ATRP is slower
than activation of PMMA-Br, due to the penultimate unit ef-
fect.31 The polymerization using ethyl α-chlorophenylacetate
(EClPA) was not controlled at all, resulting in 55% conversion
after 4 h with the formation of a polymer with bimodal distri-
bution Mn = 16000, Mw/Mn = 3.44, (entry 6), indicating that a
chloride-based initiator was not suitable in this photoinduced
metal-free system.
zation. These reactions also provide a clear indication that a
radical could be formed from EBPA under irradiation condi-
tions. The radical generated from EBPA could initiate the
polymerization before it terminates or abstracts the bromine
atom from a dormant species, Pn-Br. Therefore, EBPA could
act as both polymerization initiator and transfer agent in this
photomediated process.
The rates of polymerization with 2 under both irradiation con-
ditions were slightly slower than the one without 2 (Figure
2a), indicating that the concentration of radicals was decreased
and a radical deactivation process was involved in the pres-
ence of 2. The metal-free ATRP of MMA with 2 gave PMMA
with predictable Mn, growing with conversion and low disper-
sity, suggesting that the process is well controlled (Table 1,
entries 1 and 2).
Similarly, the background reaction for polymerization of
MMA with EClPA in the absence of 2 reached only 15% con-
version after 4 h of irradiation with 365 nm at 4.9 mW/cm2,
providing PMMA with Mn = 17900, and Mw/Mn = 2.02 (entry
8, Table 1). However, the same reaction with 2 was much fast-
er (entry 8 vs. 6, Table 1, 15% vs. 55% conversion at 4 h),
though with the same poor control, indicating that Ph-PTZ
catalyzed system is efficient to activate alkyl chloride but inef-
ficient to deactivate the propagating radicals.
Background reactions. The activation step should involve the
reaction between excited state of metal-free photoredox cata-
lyst and alkyl bromide, but under strong irradiation, the radical
could also be potentially formed by homolytic cleavage of the
C-Br bond in a conventional ATRP initiator or the polymer-Br
chain end.32 Polymerizations of MMA were conducted with
EBPA both in the absence and presence of Ph-PTZ 2 under the
following conditions: [MMA]0:[EBPA]0:[2]0 = 100:1:0 or
100:1:0.1, MMA/DMA = 1/1 (v/v), and irradiation with 365
nm at 2.1 or 4.9 mW/cm2. Polymerization of MMA without 2
provided PMMA with much higher Mn than the theoretical
value and Mw/Mn values as high as 2.2 (Table 1, entry 7 and
Figure 2b), suggesting an uncontrolled free radical polymeri-
Salt effects. As in metal-catalyzed ATRP, one possible deacti-
vation mechanism is the transfer of a halogen atom from the
radical cation-anion ion pair Cat+X– (X = Br or Cl) formed in
the activation step to the propagating radical (eq. 1). In a polar
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 16
solvent such as DMA, the ion pair would dissociate to the free
radical cation (Cat+) and a halide anion (Br- or Cl-) and could
reach an equilibrium state. Therefore, if deactivation occurs
according to eq. 1, the overall rate of polymerization and con-
trol over molecular weight distribution would be strongly in-
fluenced by the dissociation equilibrium Cat+X– = Cat+ + X–,
which can be shifted to the left if a large excess of halide ions
is added. Additionally, it must be noted that halide anions,
whether linked to Cat●+ or free in solution, are a fundamental
reagent of the deactivation step. Therefore, excess tetra-n-
butylammonium bromide (TBABr) was added to improve de-
activation in the polymerization of MMA when using Ph-PTZ
2 as a catalyst. However, the polymerization of MMA with
background reaction (entries 7 and 8 vs. entry 1, Table 2), in-
1
2
3
4
5
6
7
8
dicating that these catalysts efficiently activated the C-Br bond
but could not deactivate the propagating radical efficiently.
This is further supported by the evidence that the Mn of syn-
thesized PMMA using these catalysts was always much higher
than theoretical Mn (Figures 4b and 10b).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
added
TBABr
under
reaction
conditions
[MMA]0:[EBPA]0:[2]0:[TBABr]0= 100:1:0.1:x, x = 0.2, 1 or 2,
50% DMA, irradiation with 365 nm, did not give any observa-
ble difference in polymerization rates and dispersities (see
Figure S5).
On the other hand, when an excess TBABr was added to the
reaction using EClPA, the Mn (10,800) of obtained PMMA
was close to the theoretical value (Mn,th = 8,100), although the
dispersity was still high (Mw/Mn = 2.47, entry 9, Table 1). The
polymerization with EClPA and TBABr showed better deacti-
vation of the growing chains, indicating that deactivation was
more efficient in the presence of bromide ions.
Figure 3. The structures of catalysts studied in metal-free ATRP.
(a)
Polymerization with different catalysts. The set of compounds
shown in Figure 3 was chosen to study the effect of key struc-
tural features on their properties and reactivity. Compounds 2
and 4-7 are N-aryl phenothiazine derivatives; they were syn-
thesized from phenothiazine 9 and corresponding aryl halides
using Buchwald amination.33 10-Methylphenothiazine 8 and 9
are commercially available. The reaction of 2,3-
dihydroxynaphthalene and 2-aminothiophenol provided ben-
zo[b]phenothiazine,34 which subsequently reacted with chlo-
robenzene under Buchwald amination, yielding phenyl ben-
zo[b]phenothiazine 10. 9-Phenylcarbazole 11 and thianthrene
12 were selected due to the structural similarity to phenothia-
zine, and compounds 13 and 14 were selected due to photo-
sensitivities.35
(b)
All the catalysts shown in Figure 3 were investigated for pho-
toinduced metal-free ATRP of MMA with EBPA as initiator
under standard conditions. The results are summarized in Ta-
ble 2 and Figures 4 and S10. The reaction conditions were
[MMA]0:[EBPA]0:[Cat]0 = 100:1:0.1, in MMA/DMA = 1/1
(v/v), at room temperature with irradiation of 365 nm at 2.1
mW/cm2. The reactions with all phenothiazine-based com-
pounds, 2 and 4-10, were all slower than the background reac-
tion, indicating that some deactivation was involved. All N-
aryl phenothiazines 2, 4-7 and benzo[b]phenothiazine 10 un-
der photoinduced metal-free conditions provided well-defined
PMMA with predetermined Mn and dispersities Đ = 1.4-1.5
(entries 2-4 and 6, Table 2). The metal-free ATRP with cata-
lysts 8 (Me-PTZ) and 9 (H-PTZ) only provided limited con-
trol. The obtained Mn were close to the theoretical values at
low conversion of MMA (<30 %), but they became signifi-
cantly higher at higher conversion (entry 5, Table 2). This ob-
servation indicates that both 8 and 9 decomposed during the
later stage of the reaction and could not deactivate the radicals,
as previously suggested.23
Figure 4. a) Semilogarithmic kinetic plots of polymerization of
MMA
with
catalysts
2,
8
and 11,
conditions:
[MMA]0:[EBPA]0:[Cat]0 = 100:1:0.1, in MMA/DMA = 1/1 (v/v),
at room temperature with irradiation of 365 nm (2.1 mW/cm2); b)
number-average molecular weight (Mn, filled symbols), and dis-
persity (Mw/Mn, open symbols) versus conversion; black dot line:
linear fit for theoretical molecular weight.
The polymerizations with 11 (Ph-CBZ), 12 (TH), 13 (TIPS-
AN) and 14 ((CBZ-Ph)2) provided faster reactions than the
4
ACS Paragon Plus Environment

Page 5 of 16
Journal of the American Chemical Society
Table 2. Selected results of metal-free ATRP of MMA with different catalysts.a
1
2
3
4
5
6
7
8
9
c
c
d
Entry
Catalyst
Time
4 h
4 h
4 h
4 h
4 h
8 h
4 h
4 h
4 h
Conv.b
27%
16%
11%
10%
20%
42%
9%
Mn,th
Mn,GPC
Mw/Mn
2.25
1.50
1.48
1.40
2.24
1.79
1.47
1.89
2.56
1
2
3
4
5
-
2,900
1,800
1,300
1,200
2,200
4,400
1,100
4,100
3,100
28,700
2,070
1,580
1,600
4,520
58,400
1,670
18,500
51,550
2, Ph-PTZ
5, Cl-Ph-PTZ
7, Nap-PTZ
8, Me-PTZ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
6
7
8
10, Ph-benzoPTZ
11, Ph-CBZ
39%
29%
13, TIPS-AN
a Reaction conditions: [MMA]0:[EBPA]0:[Cat]0 = 100:1:0.1, in MMA/DMA = 1/1 (v/v), at room temperature with irradiation of 365 nm
(2.1 mW/cm2). b Determined by 1H NMR. c Calculated based on conversion obtained by 1H NMR (i.e. Mn,th = MEBPA + 100 × conversion ×
M
MMA). d Determined by GPC in THF, based on linear PMMA as calibration standards.
Since the background reaction using EBPA as initiator was a
fast process, it was not possible to determine whether all cata-
lysts were involved in the activation step. Therefore, EBiB
was used as the initiator rather than EBPA under metal-free
ATRP conditions. All the polymerizations with any of the cat-
alysts shown in Figure 3 were faster than the background reac-
tion using only EBiB, strongly indicating that all the catalysts
efficiently photoactivated alkyl halides (Figures 5 and S11).
try conducted at different scan rates, a lifetime of the order of
10 s was estimated for the radical cation 9●+ (see Figure S9).
Also 12 showed a similar behavior with a faster decay rate,
with a lifetime < 5 ms. Other tested carbazole derivatives (11
and 14) exhibited an irreversible oxidation peak, indicating
that their specific radical cations were not stable in DMA.
4
6
9
12
14
40
20
0
20
10
0
10
-20
a)
b)
0.2
0.4
0.6
0.8
1.0 0.2 0.4 0.6
1.2
1.4
1.6
E vs SCE (V)
Figure 6. Selected cyclic voltammograms for compounds charac-
terized by (a) reversible and (b) partially reversible or irreversible
oxidation, C = 2×10-3 M, in DMA + 0.1 M Et4NClO4 at 25 °C. (a)
v = 0.2 V s-1; (b) v = 0.05 V s-1.
Figure 5. Semilogarithmic kinetic plots of polymerization of
A stable radical cation is necessary for the efficient deactiva-
tion of the growing radicals, thus only the compounds with a
reversible redox behavior should efficiently control the
polymerization. The experimental results confirmed this con-
clusion as effective deactivation was observed only for com-
pounds with reversible redox properties. Moreover, CV analy-
sis confirmed the limited stability of 9●+ (H-PTZ●+, Figure 6b),
which indeed could efficiently deactivate the growing radicals
only at the beginning of the experiment, since a fraction of 9●+
decomposed by side reactions leading to the progressive con-
sumption of the catalyst. Therefore, cyclic voltammetry was a
reliable technique that allowed rapid screening of the analyzed
photoactive molecules as candidates for radical deactivation
and control of photoATRP experiments. CV experiments also
demonstrated that stable radical cations Cat●+ should be in-
volved in the deactivation reaction mechanism.
MMA
with
catalysts
2,
8
and 11,
conditions:
[MMA]0:[EBiB]0:[Cat]0 = 100:1:0.1, in MMA/DMA = 1/1 (v/v),
at room temperature with irradiation of 365 nm (2.1 mW/cm2);
black dot line: linear fit for background polymerization.
Characterizations and Properties of Catalysts
Redox potentials. Various catalyst properties were determined
to better understand their different reactivities. The results are
summarized in Table 3. Cyclic voltammetry (CV) was used to
measure the oxidation potentials of the catalysts and to assess
the stability of their radical cations (Cat●+). All CVs were rec-
orded in DMA in the presence of 0.1 M Et4NClO4 and some
examples are reported in Figure 6. Within the electrochemical
potential window of the solvent, all compounds, except 13,
could be oxidized to form a radical cation (Cat = Cat●+ + e–).
The reversibility of the voltammetric pattern is a direct indica-
tion of the stability of the radical cations. All substituted phe-
nothiazines (2, 4-8, 10) showed a reversible oxidation wave
(Figure 6a), indicating that the electrogenerated radical cation
was a stable species (lifetime ≥ 1 min). The CV of unsubstitut-
ed phenothiazine 9 had limited chemical reversibility: the rad-
ical cation quickly decomposed to form a product that was re-
duced at a lower potential (Figure 6b). From cyclic voltamme-
For the compounds exhibiting a reversible redox behavior, the
oxidation potential of the catalyst in the excited state
∗) could be estimated from the excitation energy of
the photocatalysts (Ehν), according to the following equation:16
(ꢀꢅꢃꢆꢇ
•ꢂ
/ꢅꢆꢇ
ꢊꢌ
ꢀꢅꢃꢆꢇ
ꢈ ꢀꢅꢃꢆꢇ
ꢉ ꢀꢊꢋ ꢈ ꢀꢅꢃꢆꢇ /ꢅꢆꢇ ꢉ
ꢑ2ꢒ
•ꢂ
∗
•ꢂ
•ꢂ
/ꢅꢆꢇ
/ꢅꢆꢇ
ꢍ
ꢎꢏꢐ
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 16
Table 3. Characterizations and reactivities of catalysts studied in metal-free ATRP in DMA.
1
2
3
4
5
6
7
8
ꢀꢅꢃꢆꢇ•ꢂ
ꢀꢅꢃꢆꢇ•ꢂ
/ꢅꢆꢇ
/ꢅꢆꢇ∗
λmax
Lifetime ꢓꢔ
CV
Catalyst
Activationc
Deactivationd
(ns)b
reversibility
(V vs. SCE)
0.815e
0.797
(V vs. SCE)a
-1.97
(nm)
2, Ph-PTZ
445
445
445
510
405
445
450
4.5
6.0
3.0
7.4
7.6
2.3
2.1
+
+
+
+
+
+
+
+
+
+
+
4, MeOPh-PTZ
5, ClPh-PTZ
6, Py-PTZ
-1.99
0.830
-1.96
+
+
0.903
-1.53
+
+
9
7, Nap-PTZ
8, Me-PTZ
9, H-PTZ
0.833
-2.23
+
+
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
0.826
-1.97
+
+/–
+/–
0.606
-2.15
+/–
10, Ph-benzoPTZ
11, Ph-CBZ
12, TH
0.902
1.423f
1.393g
>1.5h
440
375
445
445
410
-1.92
-1.91
-1.36
-
12.9
4.7
4.9
2.1
3.4
+
–
–
–
–
+
+
+
+
+
+
–
–
–
–
13, TIPS-AN
14, (CBZ-Ph-)2
1.392f
-
a From eq. 2. b Lifetime of the excited catalyst, Cat*. c Activation based on whether polymerization was faster than the background reaction
with EBiB as ATRP initiator (cf. Figures 5 and S11). d Deactivation evaluation based on whether Mn,GPC was close to Mn,th (cf. Figures 4b
and 10b). e For comparison, values in MeCN are ꢀꢄꢃ•ꢂ/ꢄ = 0.68 V vs. SCE and ꢀꢄꢃ•ꢂ = -2.10 V vs. SCE.23 f Potential of the anodic peak at
/ꢄ∗
v = 0.2 V s-1. g ꢀꢅꢃꢆꢇ•ꢂ
was estimated at high scan rates (v > 10 V s-1), where partial reversibility in CV could be achieved (Figure S9). h
/ꢅꢆꢇ
No oxidation wave was observed inside the potential range of DMA.
where h is the Planck constant, c is the speed of light, λmax is
of the excited state against the concentration of EBPA provid-
ed the rate constant kact = 5.7 × 109 M-1s-1. The quenching ex-
periment strongly supports the fact that the excited state of 2 is
a strong reductant and efficiently reduces the ATRP initiator.
the wavelength of maximum emission intensity of excited
state and ꢀꢅꢃꢆꢇ /ꢅꢆꢇ is the standard reduction potential of Cat+
•ꢂ
in the ground state. ꢀꢅꢃꢆꢇ /ꢅꢆꢇ was obtained from cyclic volt-
•ꢂ
Kinetic Evaluation of the Activation Mechanism by Modi-
fied Marcus Theory
The strongly negative values for ꢀꢅꢃꢆꢇ
bility of an OSET (eq. 4). This reaction involves a concerted
dissociative electron transfer (DET) to RX, as consolidated in
the literature for the reductive cleavage of alkyl halides.28,37
Therefore, assessment of ET kinetics cannot be made by a
straight forward application of the well-known Marcus theory
for electron-transfer processes.27a A modified model of Marcus
theory, developed by Savéant,27b-d is available and is currently
used to analyze the dynamics of DET processes.37b-e,38
ammetry as the half sum of anodic (Epa) and cathodic (Epc)
peak potentials, ꢀꢅꢃꢆꢇ /ꢅꢆꢇ ≈ E1/2 = (Epa + Epc)/2.
•ꢂ
suggested the via-
•ꢂ
∗
/ꢅꢆꢇ
LFP Measurements. Laser flash photolysis (LFP) experi-
ments were used to measure the lifetime of the excited states
of the catalysts 2, 4-14. All of them are near or below 10 ns
(Table 3);36 they are much short-lived, at least hundreds of ns,
than the excited states of transition-metal based photoredox
catalysts. For example, the lifetime of the excited state of 1 is
1,900 ns.29
Lifetime: 2 with EBPA Quencher
Stern-Volmer Fit: 2 with EBPA Quencher
1.6
100
1.5
10x EBPA
20x EBPA
30x EBPA
10-1
1.4
(4)
1.3
10-2
According to the DET theory,27c a quadratic activation-driving
force relationship similar to that of Marcus theory of OSET
exists also for DET processes (eq. 5).
1.2
Stern-Volmer Data
1.1
Fitted Equation: y = 25.81x + 1.02
(R2=0.993)
1.0
20
40
60
(ns)
80
100
0.000 0.005 0.010 0.015 0.020 0.025
[EBPA] (mM)
t

2
rG o
4G0‡
G‡  G‡ 1
0 

Figure 7. a) Left, decay of the excited state of 2 with increasing
concentrations of EBPA; b) right, results of Stern-Volmer treat-
ment.


(5)
where ∆ꢞꢔ‡ is the intrinsic barrier of the reaction, i.e. the acti-
vation free energy when = 0. The intrinsic barrier is
rG o
The rate constant for the reaction between an excited catalyst
and EBPA, a conventional ATRP initiator, was then measured.
Ph-PTZ 2 was excited upon irradiation and allowed to react
with increasing concentrations of EBPA. Figure 7 summarizes
the results of these experiments. The rate constant for the reac-
tion of the excited state of 2 with EBPA was determined ac-
cording to the Stern−Volmer equation (eq 3):18f
given by ∆ꢞꢔ‡ ꢈ ꢑꢃ ꢘ ꢟꢠꢡꢒ/4, where o is the solvent reor-
ganization energy and DRX is the R-X bond energy. The prin-
cipal difference between OSET and DET is that the intrinsic
barrier of the latter mainly comes from the energy of the
breaking bond. When the two fragments of DET, R and X,
are able to give rise to ion-dipole interactions in the solvent
cage, the dynamics of ET is significantly affected and eq. 5
does not correctly predict the activation free energy. The
“sticky” model of DET takes into account formation of
ꢕ
ꢕ ꢈ ꢖ ꢘ ꢙꢆꢚꢇꢛꢜꢝ
(3)
ꢖ
ꢗ
where kact is the rate constant, [Q] is the concentration of the
quencher and ꢓꢔ is the lifetime of the excited state of 2 in the
absence of a quencher. A plot of the reciprocal of the lifetime
6
ACS Paragon Plus Environment

Page 7 of 16
Journal of the American Chemical Society
Table 4. Activation rate constants and relevant thermodynamic parameters for reaction (4) in DMA.
1
2
3
ꢀꢅꢃꢆꢇ•ꢂ
ꢀꢠꢃꢡ/ꢠ•ꢢꢡꢣ
b
Z×1011
rGo
G‡
G‡
Donor
RX
kact
φact
/ꢅꢆꢇ∗
0
4
5
6
7
8
9
V vs. SCE
-1.73c
-1.97
V vs. SCEa
-0.52
kcal mol-1
-28.9
-34.6
-41.5
-29.1
29.6
kcal mol-1
16.4
kcal mol-1
5.1
M-1s-1
3.0
M-1s-1
5.8×107
5.8×108
2.0×1010
1.5×106
1.0×10-14
4.3×108
3.3×108
*
1*, Ir(ppy)3
MBiB
MBiB
EBPA
MCiB
MBiB
MBiB
MBiB
0.4
2*, Ph-PTZ*
2*, Ph-PTZ*
2*, Ph-PTZ*
2, Ph-PTZ
8*, Me-PTZ*
11*, Ph-CBZ*
-0.52
-0.22d
16.5
3.7
2.9
1.3×10-3
4.6×10-2
3.5×10-6
-
5.0×10-4
7.7×10-4
-1.97
15.3
1.6
2.8
-1.97
0.82e
-0.76
19.2
7.2
3.2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
-0.52
16.5
34.5
3.8
2.7
-1.96
-0.52
-34.4
-33.3
16.6
2.6
-1.91
-0.52
15.5
4.0
2.8
a In DMF.39 b Calculated from eq. 8 or 9. CRX = 5·10-2 M; τ0 from Table 3; ΦF was determined to be roughly constant and on average
0.01 for a large set of phenothiazine derivatives,40 therefore ΦF = 0.01 was used for 2*, 8* and 11*; ΦF = 0.40 for 1*.41 c In MeCN.20g d in
DMF, calculated as in Ref. 39, using thermodynamic data from Ref. 26a. e ꢀꢅꢃꢆꢇ•ꢂ
.
/ꢅꢆꢇ
the ion-dipole adduct by introducing the interaction energy,
in polar solvents like DMF and CH3CN (0.24–0.50 kcal mol-
1).26b Radical–anion interactions depend on the dielectric con-
stant, which are very similar for CH3CN, DMF and DMA.
Therefore, we considered that this contribution to the activa-
tion energy should be similar to the one reported for the me-
thyl propionate radical (MP●) and Br– in CH3CN (0.24 kcal
mol-1).
Dp, into eq. 5:

2

rG o  DP
4G0‡
G‡  G‡ 1
0 


(6)
Although Dp is of electrostatic nature and is often very
small,37d it decreases significantly the intrinsic barrier, now
given by ∆ꢞꢔ‡ ꢈ ꢛꢃ ꢘ ꢤꢟR1/X2 ꢉ ꢟp1/2ꢥꢦꢝ/4 , resulting in en-
All excited catalysts show high reactivity with RX with kact
values in the 106-1010 M-1s-1 range. For the DET reaction be-
tween EBPA and 2*, kact = 5.7×109 M-1 s-1 was obtained from
FLP measurements in DMA, whereas the calculated value is
2.0×1010 M-1 s-1. Considering that a series of approximations
had been forcefully introduced into the calculation, the agree-
ment between experiment and theory can be considered to be
satisfactory. Therefore, unlike transition-metal catalyzed
ATRP which involves activation via an atom transfer (or
ISET) mechanism, activation in photoinduced ATRP follows a
concerted dissociative electron transfer mechanism (an OSET
mechanism).
hanced rate of electron transfer.
Eq. 6 was used to calculate G‡ of reaction 4 for a series of
catalysts and two ATRP initiators. The activation free energy
was then used to calculate the activation rate constant, kact, ac-
cording to eq. 7:
G‡
RT




kET  kact  Z exp 


(7)
where Z is the pre-exponential factor. The results are presented
in Table 4, whereas details of the calculations as well as all
parameters used in eqs. 6 and 7 are reported in the Supporting
Information. In the examined cases, Cat can be an organic
molecule in the excited state (e. g. 2*), the same organic mole-
cule in the ground state (2), or the excited state metal complex
1*. RX is methyl 2-bromoisobutyrate (MBiB), which mimics
All the analyzed phenothiazine derivatives have redox proper-
ties that are relatively similar to each other. Table 4 shows that
2* (Ph-PTZ*) and 8* (Me-PTZ*) should react with MBiB
with similar high rate constants (5.8×108 and 4.3×108 M-1s-1,
respectively). Such values are higher than kact reported for ex-
tremely active Cu-based ATRP systems (activation of tertiary
RBr initiators by [CuIMe6TREN]+ in water43 or DMSO44), and
are typical of fast polymerizations that are often difficult to
control.45 Nevertheless, these values cannot be directly com-
pared to the kact of a traditional ATRP, because reactions that
occur from an excited state are usually less than 100% effi-
cient. ATRP activation by Cat* must compete with all decay
pathways (radiative and non-radiative) that can bring the mol-
ecules back to their ground state. The quantum yield for a
first-order reaction from a given excited state is:46
the PMMA growing chain end. The driving force for the pho-
o
toinduced electron transfer (
) is estimated from the
rG
standard potentials of the redox couples of the donor
(Cat+/Cat) and acceptor (RX/R+X–), ꢀꢅꢃꢆꢇ
and
•ꢂ
/ꢅꢆꢇ
ꢀꢠꢃꢡ/ꢠ ꢢꢡ
the Weller equation:42
•
ꢣ
, and the energy to excite the catalyst, Ehv, by using
ꢬ
ꢫ
ꢩ
ꢪ
o ꢈ ꢧꢑꢀꢅꢃꢆꢇ /ꢅꢆꢇ ꢉ ꢀꢠꢃꢡ/ꢠ ꢢꢡ
•
ꢣ
ꢉ ꢀꢊꢨꢒ ꢉ ꢭꢮꢯ ꢯꢰ
(8)
•ꢂ
rG
ꢗ
where NA is the Avogadro constant, e is the elementary charge,
ε0 is the permittivity of vacuum and ε the relative permittivity
of the solvent at 25 °C. The last term is the Columbic energy
experienced by the radical ion pair at distance r.
k '
k'
 
F  F  k'0F
k'k0
k0
(9)
Unfortunately, not all the data required for estimating the fre-
where k’ is the rate constant of the first-order reaction that oc-
curs from the excited state, ΦF is the quantum efficiency for
the formation of the excited state, k0 is the rate constant of ra-
diative decay, and τ0 = 1/k0 is the lifetime of the excited state.
ATRP activation can be considered a pseudo-first-order reac-
tion, with rate constant k’, if we take into account that the
polymerization is living and that therefore RX concentration is
quency factor Z,
‡ and DP in DMA are available; therefore,
G0
it was assumed that the thermodynamic data for RX reduction
and bond dissociation were similar in DMF and in DMA. Al-
so, the “sticky” interaction energy between methyl isobutyrate
radical (MiB●) and Br– is unknown in DMA but this interac-
tion, for activated alkyl bromides, like MBiB, is always small
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 16
roughly constant during the reaction (k’ = kact[RX]). Therefore
eq. (9) can be written as
act kact [RX]0F
MCiB led to dissociation to the free MiB radical and Br– or
Cl–. The instability of the RX radical anion confirms that ET
from Cat* to RX (eq. 4) is a concerted dissociative ET pro-
cess. The standard free energies obtained from DFT calcula-
tions for the dissociative ET to form R● and X- were used to
estimate the activation free energies and rate constants accord-
ing to the sticky model of DET (eqs. 6 and 7). Table 5 reports
the computed reaction free energies and the activation free en-
ergies calculated from eq. 6. The agreement between the com-
puted activation energies and those derived from experimental
data, is not very good in some cases. This could be due to the
uncertainty in the computed free energy of solvation of ions
(see Supporting Information, Table S2). All reactions are high-
ly exergonic and have low barrier for the dissociative electron
transfer. This confirms that DET activation is highly likely
with all three catalysts.
1
2
3
4
5
6
7
8
(10)
Quantum yields for metal-free ATRP activation are reported in
Table 4. For example, with φact = 1.3×10-3, only 1 out of ca.
103 molecules of 2* survives for a sufficiently long time in the
excited state to be able to activate MBiB. In other words, even
if 2* is able to react with MBiB with a rate constant of 5.8×108
M s-1, the actual rate of activation is significantly decreased by
the low lifetime τ0 of the excited state and the fluorescence
quantum efficiency ΦF. Since the rate of activation is also the
rate of formation of the deactivator, these parameters affect
also the deactivation steps. In particular, deactivation can oc-
cur only if the rate of activation by Cat* is higher than the
background reaction, which is the case for all analyzed com-
pounds, as shown in Figures 5 and S11.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 5. Computed reaction energies, and activation free
energies in the activation processes, with the excited cata-
lysts 2*, 8*, and 11*.
A further insight into the efficiency of activation (eq. 9 and
10) can indicate why the photoinduced ATRP of MMA re-
quired only 50-100 ppm of 1, but 1000 ppm of 2. When com-
paring 1* and 2*, the former has both longer lifetime (1900 vs.
4.5 ns) and higher quantum efficiency (0.40 vs. 0.01). As
shown in Table 4, 1* can activate the RX bond much more
efficiently than 2*. Therefore, a much higher portion of the Ir
complex will be part of the activation/deactivation process,
while under the same conditions most of 2* will quickly decay
back to the ground state, thereby being unable to participate in
any activation/deactivation process.
The standard reduction potential of PMMA● is expected to be
similar to (or only slightly more negative than) that of MiB●, -
0.70 V vs. SCE.47 Moreover, the activation energy of this reac-
tion is low, because the reduction of the radicals does not re-
quire the scission of any bond. As a result, radicals can be
quickly reduced to carbanions by Cat*, with a diffusion-
controlled rate constant. However, in a controlled ATRP pro-
cess, like that under investigation, the concentration of R● is
very small, and hardly ever exceeds 10-6 M. Therefore, the rate
of radical reduction, which is proportional to both Cat* and R●
concentrations, is essentially too slow to compete with other
radical reactions such as propagation and deactivation back to
the dormant state. The preserved chain end functionality is
high, as confirmed by several successful chain extension
tests.23-24
(4)
a
b
c
c
rGo
G‡
rGo
G‡
Cat
R-X
2*, Ph-PTZ
2*, Ph-PTZ
8*, Me-PTZ
MBiB
MCiB
MBiB
−30.4
−30.9
−32.0
−25.1
4.7
6.8
4.4
6.3
−34.6
−29.1
−34.4
−33.3
3.7
7.2
3.8
4.0
11*, Ph-CBZ MBiB
a Reaction free energies (in kcal mol-1) are computed at the
M06-2X/6-311++G(3df,2p)//B3LYP/6-31G(d) level of theory.
The SMD solvation model with DMF solvent was used in geome-
try optimizations and single point energy calculations.
b Activation barriers of DET pathway calculated from eq. 6 using
DFT-calculated reaction energies (see Supporting Information for
details). c Calculated from eqs. 6 (G‡) and 8 (rGo).
Structures and stabilities of intermediates Cat●+ and Cat●+X−.
A key factor that determines whether the deactivation occurs
through an inner-sphere or an outer-sphere ET mechanism is
the structural stability of the resulting radical cation and halide
anion complex Cat●+X− formed after the activation. The opti-
mized geometries and energies of the radical cation 2●+ and
two lowest energy isomers of the 2●+Br− complex are shown
in Figure 8. The DFT calculations indicate that both isomers
of 2●+Br− have similar Gibbs free energies in solution as the
dissociated radical cation 2●+ and Br−. Isomer 2●+Br−-C (C
stands for covalent) has stronger covalent interactions between
the S atom in the catalyst and the bromide anion, while the in-
teractions between the catalyst radical cation and the bromide
in isomer 2●+Br−-I (I stands for ionic) are mostly ionic. This
difference is supported by the shorter S−Br distance (3.03 and
4.04 Å, respectively) and a greater S−Br Wiberg bond index
(0.12 and 0.01, respectively) in 2●+Br−-C than in 2●+Br−-I. In
addition, the Br atom in the covalent complex 2●+Br−-C is less
negatively charged and has greater spin density than the Br in
the ionic complex (Figure 8). Due to the ionic character of
2●+Br−-I, the geometry of the phenothiazine rings in the ionic
complex 2●+Br−-I is almost completely planar, the same as the
dissociated radical cation.50 In contrast, the phenothiazine is
bent in 2●+Br−-C, which resembles the geometry of 2 in the
ground state. Nonetheless, the relative Gibbs free energies, the
S-Br distances and bond orders all indicate that the interac-
tions between the catalyst radical cation and the bromide anion
From Table 4, it is clear that the standard potential of the 2+/2
couple in the ground state is too positive to effectively reduce
RX and generate radicals. Therefore, the reaction cannot pro-
ceed in the absence of light. When the light source is switched
off, activation stops almost instantly because of the very fast
decay of Cat* to its ground state.
DFT Calculations of the Activation and Deactivation
Mechanisms
In order to obtain further insights into the mechanisms of the
activation and the deactivation processes, and to explore fac-
tors that affect the efficiency of the metal-free photoredox cat-
alysts, DFT calculations were carried out for the reactions with
selected catalysts 2 (Ph-PTZ), 8 (Me-PTZ) and 11 (Ph-
CBZ).48,49
Activation Mechanism. The activation processes involving the
reactions of excited 2*, 8*, and 11* with MBiB and 2* with
MCiB (eq. 4, Cat = 2*, 8*, or 11*, RX = MBiB or MCiB)
were examined through DFT calculations. Here, MBiB and
MCiB were used as a model of the PMMA growing chain end.
Geometry optimizations of the radical anions of MBiB and
8
ACS Paragon Plus Environment

Page 9 of 16
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Figure 8. Optimized geometries of the radical cation 2●+ and the zwitterionic radical complex 2●+Br−. The bond lengths are provided in Å.
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
NPA atomic charges of Br, S, and N atoms are shown in red in square brackets. Spin densities are shown in blue in parentheses.
Scheme 3. Possible deactivation mechanisms in photoinduced metal-free ATRP.
Br-
Ph
O
N
S
+
O
I
-
MiB+
2-Br-
T
E
S
O
Br-
O
O
DET
S•+
N
Ph
+
+
S
N
Ph
Br-
+
O
O
MiB+
MiB•
2•+ Br--C
2
I
I
-
T
E
S
O
O
I
S
E
Br-
S•+
N
Ph
T
+
O
+
(
c
o
n
c
e
MiB•
2•+
r
t
e
d
a
t
o
m
t
r
AET-ter
a
n
s
f
e
r
)
O
Br
S
N
Ph
+
O
2
MBiB
in 2●+Br− in solution are relatively weak, and 2●+, 2●+Br−-C,
and 2●+Br−-I may all exist in equilibrium.51
mechanisms of the MiB● with 2●+ or with 2●+Br−-C52 were
evaluated (Scheme 3): (a) inner-sphere electron transfer
(ISET) mechanism through a concerted Br atom transfer from
2●+Br−-C to MiB● via transition state TS1; (b) dissociative
electron transfer (DET) from MiB● to 2●+Br−-C to form the
carbocation, 2 and Br−, followed by recombination of MiB+
and Br- to generate MBiB; (c) outer-sphere electron transfer
(OSET-I) from MiB● to 2●+Br−-C to form an anionic 2 Br−
complex and MiB+, followed by dissociation to the catalyst 2
and Br−, and counteriaons recombination; (d) outer-sphere
electron transfer from MiB● to the dissociated radical cation
2●+ (OSET-II); and (e) associative electron transfer from 2●+ to
MiB● and Br− to form the ground-state catalyst 2 and MBiB,
involving a termolecular encounter (AET-ter). ISET and AET-
ter pathways produce RX without the formation of any inter-
mediate, while all other ET pathways (DET, OSET-I, and
OSET-II) generate the R+ cation, which then rapidly
Similarly, calculations on other Cat●+X− complexes indicate
that their dissociation to the separated radical cation and halide
anion are all facile. The most stable isomers of 2●+Cl−, 8●+Br−,
and 11●+Br− complexes are all within ±1 kcal/mol of the sepa-
rated ionic species in terms of Gibbs free energies. The opti-
mized geometries, computed spin densities, charges, and Wi-
berg bond indices of these ion pair complexes are shown in
Figure S12. One of the lowest energy isomers of 8●+Br− (Me-
PTZ●+Br−) shows strong covalent interactions between S and
Br atoms, similar to those in 2●+Br−-C. However, no isomers
with clear covalent interactions between Cat●+ and X− were
located for 11●+Br− or 2●+Cl−.
Deactivation mechanisms. Since the DFT calculations have
shown that both the catalyst radical cation 2●+ and the ion pair
complex 2●+Br− exist in solution, five possible deactivation
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 16
13.8
(4.7)
OSET-I
13.8
2 Br-
1
2
3
4
5
6
7
8
+ MiB+
G
(H)
kcal/mol
O
DET
13.6
O
ISET
MiB+
10.7 (-9.6)
8.6
(7.8)
TS1
OSET-II
9.2
2 + Br-
9
MiB+
+
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AET-ter
3.9
O
O
0.2
(-6.3)
Br-
0.0
(0.0)
S•+
N Ph
2•+Br--C
2•+ + Br-
MiB•
MiB•
+
+
TS1
Br-
S•+
2•+Br--C
O
S•+
N Ph
N
Ph
O
-33.6
(-44.1)
O
Br
S
N Ph
O
MiB•
2•+
2 + MBiB
MBiB
2
Figure 9. Computed reaction energy profiles for the reaction of 2●+Br−-C with MiB●. Magenta: associative electron transfer involving a
termolecular encounter (AET-ter). Black: inner-sphere electron transfer (ISET) (i.e. concerted atom transfer); Green: dissociative electron
transfer from MiB● to 2●+Br−-C (DET); Red: stepwise outer-sphere electron transfer from MiB● to 2●+Br−-C to form 2 Br− (OSET-I);
Blue: outer-sphere electron transfer from MiB● to the dissociated radical cation 2●+ (OSET-II). Activation free energies in the OSET path-
ways are calculated using the Marcus theory.
recombines with the halide anion to form RX. The transition
state for the concerted Br atom transfer (ISET) was optimized
with DFT calculations. The barriers for the outer-sphere elec-
tron transfer pathways (OSET-I, OSET-II) were calculated
using the Marcus theory, whereas the activation free energy of
the DET pathway was calculated by eq. 6. The AET-ter path-
way (eq 11) is the exact reverse process of DET of the activa-
tion step with the catalyst at the ground state (eq 4).53
ISET pathway (10.5 kcal/mol). The other two reaction path-
ways, OSET-I and DET, have higher barriers of 13.6 and 13.4
kcal/mol, respectively.
These calculations suggest AET-ter to be the most favored
pathway. Additionally, some experimental observations are in
contrast with OSET-I, OSET-II and DET. First, the reaction of
2 with EClPA (alkyl chloride) was not as controlled as the re-
action between 2 and EBPA (alkyl bromide). The effects of
the halide (better control with RBr than RCl) rule out the pos-
sibility of OSET-II, which should not be affected by the nature
of X−. Also, the lack of oligomer formation during the
polymerization provides further evidence against the for-
mation of MiB+, thus ruling out not only OSET-II pathway but
also DET and OSET-I pathways.
kAET-ter
Cat
+
R
+ X
Cat
+
RX
(11)
Since the intrinsic barrier is defined as the activation free en-
ergy at zero driving force, reactions 4 and 11 have the same
∆ꢞꢔ‡ value. ∆ꢞꢔ‡ and ΔG‡ were already calculated for Ph-PTZ
+ MBiB (Table 4). Nevertheless we recalculated ΔG‡ using
DFT data for a better comparison of this reaction route with
the other reaction pathways for which only DFT data are
available. The activation free energy can then be calculated by
eq. 6. The computed reaction energy profiles of the five path-
ways are summarized in Figure 9 (detailed calculations in the
Supporting Information).
Computed Barriers for Deactivation Processes with Different
Catalysts. The computed activation energies of deactivation
reactions with different catalysts are summarized in Table 6.
In reactions with MBiB, the computed G‡ values are only
minimally affected when switching the catalyst from Ph-PTZ
2 to Me-PTZ 8 (entries 1 and 3). However, catalyst Me-PTZ 8
did not perform as well as catalyst 2, probably due to the slow
decomposition of 8●+ (vide supra).54 In the reaction with cata-
lyst Ph-CBZ 11 (entry 4), all five possible pathways have
much lower G‡ than the corresponding pathways with 2 and
8, due to the greater oxidizing power of 11●+ compared to 2●+
and 8●+. However, 11 is a very inefficient deactivator in the
polymerization of MMA (cf. entry 7, Table 2). The poor per-
formance of 11 is due to the instability of the radical cation
11●+ shown by CV experiments. In fact, CV of 11 has shown
an irreversible response even at high scan rates indicating that
The ISET pathway requires 10.5 kcal/mol of activation free
energy with respect to the ion pair complex 2●+Br−-C. The op-
timized geometry of the ISET transition state (TS1) is shown
in Figure 9. Among the other four outer-sphere electron trans-
fer pathways, AET-ter pathway of radical cation 2●+, MiB●
and Br− forming 2 and MBiB has the lowest activation energy,
3.9 kcal/mol (see the Supporting Information for detailed cal-
culations). The electron transfer from MiB● to the dissociated
radical cation 2●+ (OSET-II) requires 9.2 kcal/mol of activa-
tion free energy, which is close to the activation energy of
10
ACS Paragon Plus Environment

Page 11 of 16
Journal of the American Chemical Society
Table 6. Computed activation energies for possible deactivation pathways in metal-free ATRP with photoredox catalysts 2,
8, and 11.
1
2
3
4
5
6
7
8
activation energy for deactivation pathways ΔG‡ (ΔH‡) kcal/mol
Entry
catalyst
initiator
ISETa
AET-terb
DETa
OSET-Ia
OSET-IIb
1
2
3
4
a
2, Ph-PTZ
2, Ph-PTZ
8, Me-PTZ
11, Ph-CBZ
MBiB
MCiB
MBiB
MBiB
10.5 (−3.3)
12.5 (−2.4)
9.0 (−2.6)
8.3 (−3.5)
3.9
6.1
3.7
1.0
b
13.4
16.5
13.2
2.5
13.6
11.2
11.5
2.0
9.2
9.2
8.3
0.8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Activation energies with respect to the ion pair complex Cat●+X−; Activation energies with respect to separated ions Cat●+ and X−,
which energies are within 1 kcal/mol of the ion pair complex.
11●+ has a very short lifetime (vide supra). Computational re-
sults indicate that the homolytic cleavage of 11●+Br− to form
ground state 11 and a Br radical is exergonic by -1.5 kcal/mol
(Scheme 4). In contrast, the homolytic dissociation of other
Cat●+X− complexes to form free halogen radical is much more
unfavorable, and requires 14.0 kcal/mol, 13.8 kcal/mol, and
30.8 kcal/mol for 2●+Br−, 8●+Br−, and 2●+Cl−, respectively.
was obtained from the polymerization of MMA with 2 under
4.9 mW/cm2 irradiation (Figure 2).
For a termolecular reaction pathway AET-ter, the frequency
factor Zter is calculated following Tolman’s approach:55
1/2




1/2


1/2 



2RT
mA m
mAmB
mB m
mBmC


2
2
Zter  NA2 82

dABdBC
B 
C 












(12)
where A, B and C are the three species involved in the reac-
tion, d is the distance between the centers of the spheres
equivalent to the subscript particles, and δ is the distance be-
tween the two first spheres when hit by the third. Usually δ is
taken to be between 0.3 Å53 and 1 Å.55c The smaller value of
0.3 Å was used to avoid overestimating ZTER. The hard sphere
diameters of the species involved in the reaction were estimat-
ed from their computed volumes or taken from the literature
Scheme 4. Homolytic cleavage of the 11●+Br− complex indi-
cates the instability of 11●+.
The effects of halides on the barriers of ISET and all other
four deactivation pathways were then explored. When MCiB
is used in place of MBiB as the initiator in the reaction with
catalyst 2 (entry 2, Table 6), the barriers of the ISET, AET-ter
and DET pathways increase, whereas that of OSET-I decreas-
es. As expected, the halide has no effect on the barrier of the
OSET-II pathway. While the most preferred pathway with
MCiB is still AET-ter, the activation free energy is 2.2
kcal/mol higher than the reaction with MBiB.
(see Supporting Information). Then, using G‡
= 3.9
AET-ter
kcal/mol (Table 6), the rate constant was calculated from eq.
13 as kAET-ter = 4.6×107 M-2 s-1. This deactivation rate constant
is as high as or even higher than kdeact in typical Cu-based
ATRP.56
GA‡ET-ter
RT
kAET-ter  Ztere
(13)
In summary, the computed activation energies indicate that the
AET-ter pathway is preferred in the deactivation process. A
combination of more effective catalysts and initiator, such as 2
and 8 with MBiB (entries 1 and 3) leads to low barriers for the
AET-ter pathway. The performance of catalyst 11 is impeded
by the instability of both radical cation 11+ and complex
11+Br− that can readily dissociate to form a free bromine radi-
cal. The poor control of polymerization of MMA with alkyl
chloride as ATRP initiator provides a further support for the
AET-ter deactivation mechanism. With 2 as catalyst, G‡ of
AET-ter increases by 2.2 kcal/mol when Cl− is used in place of
Br−. This will result in a considerable lowering of the deactiva-
tion rate, which might not be able to outrun radical-radical
termination reactions (vide infra).
The rate of termolecular deactivation is given by:
RAET-ter  kAET-ter [2 ][Br ][R ]
(14)
However, both 2●+ and Br− concentration need to be estimated.
Considering around 10% of termination, [Br−] should be ca. 5
mM. Moreover, the CV registered during a metal-free ATRP
confirmed that around 5 mM of Br− was generated after a few
hours (Figure S14). The radical cation 2●+ could not be direct-
ly detected during the electrochemical measurements, and it
slowly decomposed by the reaction with Br−.54 Therefore, a
low value of 5×10-4 M was chosen for the concentration of 2●+.
Using these concentrations together with [R●]  4.6 ×10-8 M
gives RAET-ter = 5.3×10-6 M s-1.
A similar approach was used to calculate the rate constants
and rates of all other deactivation pathways (see the Support-
ing Information for the detailed calculations). Although the
concentration of 2●+ has a high uncertainty, the ratio of deacti-
vation rates is independent of [2●+] and [R●]. To single out the
effective deactivation pathway, the rate of all deactivation re-
actions must be first compared to Rt and then to each other.
R/Rt values calculated for all deactivation pathways are report-
ed in Table 7 (last column). ISET, DET and OSET-I are to be
discarded as possible deactivation pathways as they are at least
2 orders of magnitude slower than termination. The rate of
OSET-II is comparable with Rt, but clearly this deactivation
pathway cannot provide good control. This leaves AET-ter,
Comparison of Rates of All Deactivation Pathways
In a controlled radical polymerization, the deactivation reac-
tion should be faster than radical-radical termination to main-
tain the living character. Therefore, the rate constants and re-
action rates for different deactivation pathways were calculat-
ed for catalyst 2, and the results are summarized in Table 7.
The rate of radical termination could be obtained from
2 , where kt is the rate constant of radical-radical
Rt  kt[R• ]
termination and [R●] is the concentration of the propagating
radical. [R●]  4.6 ×10-8 M could be estimated from kp
=
app
kp[R•], where kp , the apparent rate constant of propagation,
app
11
ACS Paragon Plus Environment

Journal of the American Chemical Society
Table 7. Rate constants and rates of proposed deactivation pathways.
Page 12 of 16
1
2
3
4
5
6
7
8
9
k
R
R/Rt
Reaction
rate law
Rt  kt[R ]2
RAET-ter  kAET-ter [2+ ][Br][R]
RAET-ter  kAET-ter [2+ ][Cl ][R ]
(M-1 s-1)
(M s-1)
107
2.1×10-8
3.9×10-6
1
Termination
AET,ter
2R• → R-R
2•+ + R• + Br– → RBr + 2
2•+ + R• + Cl– → RCl + 2
2•+Br–-C + R• → RBr + 2
2•+ + R• → R+ + 2
3.4×107 a
1.9×102
<4.3
7.8×105 a <9.0×10-8
 kISET[2+ Br -C][R ]
6.1×103
3.5×103
45
5.0×10-10
7.7×10-8
3.7×10-12
2.4×10-2
3.7
ISET
R
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ISET
ROSET-II  kOSET-II[2+ ][R ]
RDET  kDET[2+ Br -C][R ]
ROSET-I  kOSET-I[2+ Br -C][R ]
OSET-II
DET
2•+Br–-C + R• → 2 + Br– + R+
1.8×10-4
<2.7×10-12 <1.3×10-4
OSET-I
2•+Br–-C + R• → 2-Br–-C +
32.4
R+
a Unit: M-2 s-1.b Activation of MCiB is much slower than activation of MBiB, kact,MCiB/kact,MBiB = 2.6×10-3. This implies that [2+] is much
smaller with MCiB than with MBiB. For the same reason [Cl-] < [Br-]. Therefore, RAET-ter and RAET-ter/Rt are overestimated.
Scheme 5. Proposed overall mechanism for photoinduced metal-free ATRP and optimized geometries of related intermediates,
based on DFT calculations.
which is more than 2 orders of magnitude faster than radical-
radical termination when RBr is used as initiator or bromide
ions are added, as the only possible deactivation pathway.
Overall Mechanism
A proposed overall mechanism can be constructed by combin-
ing all the information from experimental data, LFP measure-
ments, and calculations based on electron transfer (Marcus and
further developments) and DFT theories, and is illustrated in
Scheme 5. The optimized geometries of related intermediates
based on DFT calculations are also shown. After Ph-PTZ 2 is
excited to the excited state 2*, a dissociative electron transfer
occurs from 2* to the conventional ATRP initiator (alkyl bro-
mide, MBiB), forming the alkyl radical required to initiate the
polymerization. In this process, 2 is oxidized to the radical
cation, 2●+, which exists in equilibrium with 2●+Br-. The asso-
ciative electron transfer (AET-ter) from 2●+ to the propagating
radical and bromide anion finishes the catalytic cycle to re-
The same conclusion is reached if RAET-ter is calculated on the
basis of experimental data. In this case, using Go = 29.6
kcal/mol (Table 4) gives ΔG‡
= 5.0 kcal/mol. It follows
AET-ter
that kAET-ter = 5.8×106 M-2 s-1, RAET-ter = 6.8×10-7 M s-1 and
RAET-ter/Rt = 32 (see Supporting Information). It is clear that
AET-ter is the fastest deactivation pathway and, in particular,
at least one order of magnitude faster than all other deactiva-
tion reactions. When RCl is used as initiator, AET-ter is only
four times faster than termination, which explains why control
is lost with a chloride polymerization initiator (EClPA).
12
ACS Paragon Plus Environment

Page 13 of 16
Journal of the American Chemical Society
generate ground-state catalyst 2 and polymer chain with bro- high quantum efficiency (ΦF ≥ 0.01) to ensure efficient activa-
1
2
3
4
5
6
7
8
mine as chain-end fidelity.
tion of the R-X bond. The generated radical cation should also
be stable (lifetime >> 10 s) and have a high reduction potential
It is important to note that although the catalytic cycle shown
in Scheme 5 is the main pathway for the photoinduced metal-
free ATRP process, several background and side reactions also
need to be taken into account. Photoinduced homolytic cleav-
age of the C-Br bond, in the ATRP initiator or polymer chain
end, provides alkyl and bromine radicals and both radicals
could initiate a polymerization. This background reaction
might result in the formation of some dead chains from radi-
cal-radical termination. However, this type of termination
should be limited. In most cases, the alkyl radicals formed di-
rectly from background reaction would abstract the bromine
atom either from 2●+Br- or C-Br in EBPA to regenerate the
dormant chain end. Since the excited catalyst 2* is a strong
reductant, it is also plausible that radicals could be reduced to
carbanions. However, due to the very low concentration of
both radical and catalyst in the excited state, the radical reduc-
tion pathway is very slow. Based on the redox property of 2●+,
it could potentially oxidize Br- to bromine (Br2). Upon irradia-
tion Br2 could generate Br● to initiate the polymerization or
Br2 could add to a double bond to form an active ATRP initia-
tor. In both cases, the formation of Br2 should not stop the
polymerization; however, bromination of the catalyst might
turn off its reactivity and 8 might be more easily brominated
than 2.
(ꢀꢅꢃꢆꢇ
≈ 0.8 V vs. SCE) to ensure the quick oxidative
•ꢂ
∗
/ꢅꢆꢇ
trapping of R● and Br–.
ASSOCIATED CONTENT
Supporting Information. Experimental procedures for prepara-
tion, characterization of all compounds, additional experimental
data, supplementary figures, detailed kinetics and DFT calcula-
tions. This material is available free of charge via the Internet at
http://pubs.acs.org.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AUTHOR INFORMATION
Corresponding Author
A.G.: armando.gennaro@unipd.it
P.L.: pengliu@pitt.edu
K.M.: km3b@andrew.cmu.edu
Notes
The authors declare no competing financial interests.
ACKNOWLEDGMENT
We thank the NSF (CHE-1400052) and members of the CRP
Consortium (K.M., X.P., N.M.), University of Padova (A.G.,
A.A.I., M.F.), Fondazione Ing. Aldo Gini (M.F.). DFT Calcula-
tions were performed at the Center for Simulation and Modeling
at the University of Pittsburgh and the Extreme Science and Engi-
neering Discovery Environment (XSEDE) supported by NSF
(P.L., C.F.). We also acknowledge Dr. Michelle Coote for helpful
discussion.
CONCLUSIONS
The mechanism of photoinduced metal-free ATRP was inves-
tigated via a combination of polymerization, kinetics, cyclic
voltammetry, laser flash photolysis, dissociative electron
transfer, Marcus and DFT calculations. A controlled radical
polymerization needs to meet two criteria: fast initia-
tion/activation and efficient deactivation. All selected catalysts
are involved in the activation process, and generate alkyl radi-
cals upon irradiation, but not all are efficient deactivators. All
phenothiazine derivatives participate in the deactivation pro-
cess; however, only N-aryl phenothiazine derivatives are sta-
ble enough to survive until the later stages of the polymeriza-
tion. Alkyl chlorides could not be successfully used as ATRP
initiators and provide an uncontrolled radical polymerization.
REFERENCES
1. a) Goto, A.; Fukuda, T. Prog. Polym. Sci. 2004, 29, 329-385; b)
Braunecker, W. A.; Matyjaszewski, K. Prog. Polym. Sci. 2007, 32,
93-146; c) Matyjaszewski, K.; Davis, T. P. Handbook of radical
polymerization; Wiley Online Library, 2002.
2. a) Nicolas, J.; Guillaneuf, Y.; Lefay, C.; Bertin, D.; Gigmes, D.;
Charleux, B. Prog. Polym. Sci. 2013, 38, 63-235; b) Hawker, C. J.;
Bosman, A. W.; Harth, E. Chem. Rev. 2001, 101, 3661-3688.
3. a) Wang, J.-S.; Matyjaszewski, K. J. Am. Chem. Soc. 1995, 117,
5614-5615; b) Kato, M.; Kamigaito, M.; Sawamoto, M.;
Higashimura, T. Macromolecules 1995, 28, 1721-1723; c) Patten, T.
E.; Xia, J.; Abernathy, T.; Matyjaszewski, K. Science 1996, 272, 866-
868; d) Matyjaszewski, K.; Xia, J. Chem. Rev. 2001, 101, 2921-2990;
e) Matyjaszewski, K.; Tsarevsky, N. V. J. Am. Chem. Soc. 2014, 136,
6513-6533; f) Matyjaszewski, K. Macromolecules 2012, 45, 4015-
4039.
4. a) 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. Macromolecules 1998, 31, 5559-5562;
b) Hill, M. R.; Carmean, R. N.; Sumerlin, B. S. Macromolecules
2015, 48, 5459-5469.
5. a) De Paoli, P.; Isse, A. A.; Bortolamei, N.; Gennaro, A. Chem.
Commun. 2011, 47, 3580-3582; b) Matyjaszewski, K.; Patten, T. E.;
Xia, J. J. Am. Chem. Soc. 1997, 119, 674-680.
6. Pintauer, T.; Matyjaszewski, K. Chem. Soc. Rev. 2008, 37,
1087-1097.
7. Jakubowski, W.; Matyjaszewski, K. Angew. Chem. Int. Ed.
2006, 45, 4482-4486.
8. Matyjaszewski, K.; Jakubowski, W.; Min, K.; Tang, W.; Huang,
J.; Braunecker, W. A.; Tsarevsky, N. V. Proc. Natl. Acad. Sci. U.S.A.
2006, 103, 15309-15314.
Photoinduced metal-free ATRP provides a fascinating avenue
to synthesize well-defined polymers in the absence of residual
transition metals. For the analyzed phenothiazine-based pho-
tocatalysts, activation involves a dissociative electron transfer
to RX. Activation rate constants are higher than for classic Cu-
based ATRP systems. However, due to a short lifetime of the
excited states, activation of RX is quite slow and relatively
large amounts of catalyst should be used. Comparison of all
reasonable deactivation pathways showed that the most fa-
vored reaction route is the termolecular reaction of 2●+, R● and
Br–. Similarly to activation reaction, deactivation rate con-
stants are at least as high as that reported for copper complex-
es, but in this case the rate is severely reduced by the low like-
lihood of three-center encounters. Therefore, precise control
over macromolecular architecture by metal-free ATRP appears
to be limited by the establishment of a fairly slow activa-
tion/deactivation process.
In order to obtain a well-controlled metal-free ATRP, the cata-
lyst should efficiently absorb photons, and therefore must be
excited at the proper wavelength. In addition, photoexcitation
should produce a strongly reducing excited state (ꢀꢅꢃꢆꢇ
≈
•ꢂ
∗
/ꢅꢆꢇ
9. a) Konkolewicz, D.; Wang, Y.; Krys, P.; Zhong, M.; Isse, A. A.;
Gennaro, A.; Matyjaszewski, K. Polym. Chem. 2014, 5, 4396-4417;
-2 V vs. SCE), with a sufficiently long lifetime (τ0 ≥ 5 ns) and
13
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 16
b) Konkolewicz, D.; Wang, Y.; Zhong, M.; Krys, P.; Isse, A. A.;
21. Treat, N. J.; Fors, B. P.; Kramer, J. W.; Christianson, M.;
Gennaro, A.; Matyjaszewski, K. Macromolecules 2013, 46, 8749-
8772.
Chiu, C.-Y.; Alaniz, J. R. d.; Hawker, C. J. ACS Macro Lett. 2014, 3,
580-584.
1
2
3
4
5
6
7
8
10. Rosen, B. M.; Percec, V. Chem. Rev. 2009, 109, 5069-5119.
11. a) Magenau, A. J. D.; Strandwitz, N. C.; Gennaro, A.;
Matyjaszewski, K. Science 2011, 332, 81-84; b) Park, S.; Chmielarz,
P.; Gennaro, A.; Matyjaszewski, K. Angew. Chem. Int. Ed. 2015, 54,
2388-2392; c) Bortolamei, N.; Isse, A. A.; Magenau, A. J. D.;
Gennaro, A.; Matyjaszewski, K. Angew. Chem. Int. Ed. 2011, 50,
11391-11394.
22. Poelma, J. E.; Fors, B. P.; Meyers, G. F.; Kramer, J. W.;
Hawker, C. J. Angew. Chem. Int. Ed. 2013, 52, 6844-6848.
23. Treat, N. J.; Sprafke, H.; Kramer, J. W.; Clark, P. G.; Barton,
B. E.; Read de Alaniz, J.; Fors, B. P.; Hawker, C. J. J. Am. Chem. Soc.
2014, 136, 16096-16101.
24. Pan, X.; Lamson, M.; Yan, J.; Matyjaszewski, K. ACS Macro
Lett. 2015, 4, 192-196.
25. Kaur, A.; Ribelli, T. G.; Schröder, K.; Matyjaszewski, K.;
Pintauer, T. Inorg. Chem. 2015, 54, 1474-1486.
12. a) Konkolewicz, D.; Schroder, K.; Buback, J.; Bernhard, S.;
Matyjaszewski, K. ACS Macro Lett. 2012, 1, 1219-1223; b) Ribelli,
T. G.; Konkolewicz, D.; Pan, X.; Matyjaszewski, K. Macromolecules
2014, 47, 6316-6321; c) Mosnáček, J.; Ilčíková, M. Macromolecules
2012, 45, 5859–5865; d) Anastasaki, A.; Nikolaou, V.; Zhang, Q.;
Burns, J.; Samanta, S. R.; Waldron, C.; Haddleton, A. J.; McHale, R.;
Fox, D.; Percec, V.; Wilson, P.; Haddleton, D. M. J. Am. Chem. Soc.
2014, 136, 1141-1149; e) Pan, X.; Malhotra, N.; Simakova, A.; Wang,
Z.; Konkolewicz, D.; Matyjaszewski, K. J. Am. Chem. Soc. 2015, 137,
15430 - 15433; f) Dadashi-Silab, S.; Atilla Tasdelen, M.; Yagci, Y. J.
Polym. Sci. A Polym. Chem. 2014, 52, 2878-2888; g) Ciftci, M.;
Tasdelen, M. A.; Li, W.; Matyjaszewski, K.; Yagci, Y.
Macromolecules 2013, 46, 9537-9543; h) Tasdelen, M. A.; Ciftci, M.;
Yagci, Y. Macromol. Chem. Phys. 2012, 213, 1391-1396; i) Tasdelen,
M. A.; Uygun, M.; Yagci, Y. Macromol. Rapid Commun. 2011, 32,
58-62; j) Tasdelen, M. A.; Uygun, M.; Yagci, Y. Macromol. Chem.
Phys. 2010, 211, 2271-2275; k) Kwak, Y.; Matyjaszewski, K.
Macromolecules 2010, 43, 5180-5183.
13. a) Pan, X.; Malhotra, N.; Zhang, J.; Matyjaszewski, K.
Macromolecules 2015, 48, 6948-6954; b) Yang, Q.; Dumur, F.;
Morlet-Savary, F.; Poly, J.; Lalevée, J. Macromolecules 2015, 48,
1972-1980.
14. Fors, B. P.; Hawker, C. J. Angew. Chem. Int. Ed. 2012, 51,
8850–8853.
15. a) Meyer, T. J. Accounts of Chemical Research 1989, 22, 163-
170; b) Lalrempuia, R.; McDaniel, N. D.; Müller-Bunz, H.; Bernhard,
S.; Albrecht, M. Angew. Chem. Int. Ed. 2010, 49, 9765-9768; c)
DiSalle, B. F.; Bernhard, S. J. Am. Chem. Soc. 2011, 133, 11819-
11821.
16. Kalyanasundaram, K. Coord. Chem. Rev. 1982, 46, 159-244.
17. Howerton, B. S.; Heidary, D. K.; Glazer, E. C. J. Am. Chem.
Soc. 2012, 134, 8324-8327.
18. a) Kärkäs, M. D.; Matsuura, B. S.; Stephenson, C. R. J. Science
2015, 349, 1285-1286; b) Le, C. C.; MacMillan, D. W. C. J. Am.
Chem. Soc. 2015, 137, 11938-11941; c) Nawrat, C. C.; Jamison, C.
R.; Slutskyy, Y.; MacMillan, D. W. C.; Overman, L. E. J. Am. Chem.
Soc. 2015, 137, 11270-11273; d) Jin, J.; MacMillan, D. W. C. Nature
2015, 525, 87-90; e) Terrett, J. A.; Cuthbertson, J. D.; Shurtleff, V.
W.; MacMillan, D. W. C. Nature 2015, 524, 330-334; f) Prier, C. K.;
Rankic, D. A.; MacMillan, D. W. C. Chem. Rev. 2013, 113, 5322-
5363; g) Romero, N. A.; Margrey, K. A.; Tay, N. E.; Nicewicz, D. A.
Science 2015, 349, 1326-1330; h) Griffin, J. D.; Zeller, M. A.;
Nicewicz, D. A. J. Am. Chem. Soc. 2015, 137, 11340-11348.
19. a) Lalevée, J.; Peter, M.; Dumur, F.; Gigmes, D.; Blanchard,
N.; Tehfe, M.-A.; Morlet-Savary, F.; Fouassier, J. P. Chem. Eur. J.
2011, 17, 15027-15031; b) Lalevée, J.; Blanchard, N.; Tehfe, M.-A.;
Morlet-Savary, F.; Fouassier, J. P. Macromolecules 2010, 43, 10191-
10195; c) Ogawa, K. A.; Goetz, A. E.; Boydston, A. J. J. Am. Chem.
Soc. 2015, 137, 1400-1403; d) Xiao, P.; Zhang, J.; Dumur, F.; Tehfe,
M. A.; Morlet-Savary, F.; Graff, B.; Gigmes, D.; Fouassier, J. P.;
Lalevée, J. Prog. Polym. Sci. 2015, 41, 32-66; e) Perkowski, A. J.;
You, W.; Nicewicz, D. A. J. Am. Chem. Soc. 2015, 137, 7580-7583.
20. a) Xu, J.; Jung, K.; Atme, A.; Shanmugam, S.; Boyer, C. J. Am.
Chem. Soc. 2014, 136, 5508-5519; b) Shanmugam, S.; Xu, J.; Boyer,
C. J. Am. Chem. Soc. 2015, 137, 9174-9185; c) Shanmugam, S.; Xu,
J.; Boyer, C. Chem. Sci. 2015, 6, 1341-1349; d) Xu, J.; Shanmugam,
S.; Duong, H. T.; Boyer, C. Polym. Chem. 2015, 6, 5615-5624; e)
Shanmugam, S.; Boyer, C. J. Am. Chem. Soc. 2015, 137, 9988-9999;
f) Yeow, J.; Xu, J.; Boyer, C. ACS Macro Lett. 2015, 4, 984-990; g)
Chen, M.; MacLeod, M. J.; Johnson, J. A. ACS Macro Lett. 2015, 4,
566-569.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
26. a) Lin, C. Y.; Coote, M. L.; Gennaro, A.; Matyjaszewski, K. J.
Am. Chem. Soc. 2008, 130, 12762-12774; b) Isse, A. A.; Bortolamei,
N.; De Paoli, P.; Gennaro, A. Electrochim. Acta 2013, 110, 655-662.
27. a) Marcus, R. A. J. Chem. Phys 1956, 24, 966-978; b) Pause,
L.; Robert, M.; Savéant, J.-M. J. Am. Chem. Soc. 2000, 122, 9829-
9835; c) Saveant, J. M. J. Am. Chem. Soc. 1987, 109, 6788-6795; d)
Saveant, J. M. J. Am. Chem. Soc. 1992, 114, 10595-10602.
28. Isse, A. A.; Gennaro, A.; Lin, C. Y.; Hodgson, J. L.; Coote, M.
L.; Guliashvili, T. J. Am. Chem. Soc. 2011, 133, 6254-6264.
29. Flamigni, L.; Barbieri, A.; Sabatini, C.; Ventura, B.;
Barigelletti, F. In Photochemistry and Photophysics of Coordination
Compounds II; Balzani, V., Campagna, S., Eds.; Springer Berlin
Heidelberg, 2007; Topics in Current Chemistry, Vol. 281.
30. a) Wallentin, C.-J.; Nguyen, J. D.; Finkbeiner, P.; Stephenson,
C. R. J. J. Am. Chem. Soc. 2012, 134, 8875-8884; b) Nguyen, J. D.;
Tucker, J. W.; Konieczynska, M. D.; Stephenson, C. R. J. J. Am.
Chem. Soc. 2011, 133, 4160-4163.
31. a) Lin, C. Y.; Coote, M. L.; Petit, A.; Richard, P.; Poli, R.;
Matyjaszewski, K. Macromolecules 2007, 40, 5985-5994; b) Nanda,
A. K.; Matyjaszewski, K. Macromolecules 2003, 36, 8222-8224.
32. Pérez-Prieto, J.; Miranda, M. A. In Encyclopedia of Radicals
in Chemistry, Biology and Materials; John Wiley & Sons, Ltd, 2012.
33. Surry, D. S.; Buchwald, S. L. Chem. Sci. 2011, 2, 27-50.
34. Crivello, J. V. J. Polym. Sci. A Polym. Chem. 2008, 46, 3820-
3829.
35. a) Tehfe, M.-A.; Lalevée, J.; Morlet-Savary, F.; Graff, B.;
Blanchard, N.; Fouassier, J.-P. ACS Macro Lett. 2012, 1, 198-203; b)
Kolosov, D.; Adamovich, V.; Djurovich, P.; Thompson, M. E.;
Adachi, C. J. Am. Chem. Soc. 2002, 124, 9945-9954.
36. Lifetime of Ph-PTZ* is 2.3 ns, measued in MeCN, see:
Discekici, E. H.; Treat, N. J.; Poelma, S. O.; Mattson, K. M.; Hudson,
Z. M.; Luo, Y.; Hawker, C. J.; de Alaniz, J. R. Chem. Commun. 2015,
51, 11705.
37. a) Isse, A. A.; Sandonà, G.; Durante, C.; Gennaro, A.
Electrochim. Acta 2009, 54, 3235-3243; b) Costentin, C.; Robert, M.;
Savéant, J.-M. J. Am. Chem. Soc. 2003, 125, 10729-10739; c) Isse, A.
A.; Gennaro, A. J. Phys. Chem. A 2004, 108, 4180-4186; d)
Cardinale, A.; Isse, A. A.; Gennaro, A.; Robert, M.; Savéant, J.-M. J.
Am. Chem. Soc. 2002, 124, 13533-13539; e) Andrieux, C. P.; Le
Gorande, A.; Saveant, J. M. J. Am. Chem. Soc. 1992, 114, 6892-6904;
f) Andrieux, C. P.; Gallardo, I.; Savaent, J. M.; Su, K. B. J. Am.
Chem. Soc. 1986, 108, 638-647.
38. Pause, L.; Robert, M.; Savéant, J.-M. J. Am. Chem. Soc. 2001,
123, 4886-4895.
39. Isse, A. A.; Lin, C. Y.; Coote, M. L.; Gennaro, A. J. Phys.
Chem. B 2011, 115, 678-684.
40. Saucin, M.; Van de Vorst, A. Radiat. Environ. Biophys. 1980,
17, 159-168.
41. Jung, S. O.; Kim, Y.-H.; Kim, H.-S.; Kwon, S.-K.; Oh, H.-Y.
Mol. Cryst. Liq. Cryst. 2006, 444, 95-101.
42. Rehm, D.; Weller, A. Isr. J. Chem. 1970, 8, 259-271.
43. Konkolewicz, D.; Krys, P.; Góis, J. R.; Mendonça, P. V.;
Zhong, M.; Wang, Y.; Gennaro, A.; Isse, A. A.; Fantin, M.;
Matyjaszewski, K. Macromolecules 2014, 47, 560-570.
44. Lorandi, F.; Fantin, M.; Isse, A. A.; Gennaro, A. Polymer
2015, 72, 238-245.
45. Fantin, M.; Isse, A. A.; Gennaro, A.; Matyjaszewski, K.
Macromolecules 2015, 48, 6862-6875.
46. Turro, N. J., Modern Molecular Photochemistry 1978,
University Science Books, Herndon, VA, USA.
14
ACS Paragon Plus Environment

Page 15 of 16
Journal of the American Chemical Society
47. Bortolamei, N.; Isse, A. A.; Gennaro, A. Electrochim. Acta
2010, 55, 8312-8318.
1
48. For computational studies on Cu-catalyzed ATRP reactions,
see refs 26a and 28.
49. For computational studies of photoredox reactions with
transition metal-based catalysts: (a) Gutierrez, O.; Tellis, J. C.;
Primer, D. N.; Molander, G. A.; Kozlowski, M. C. J. Am. Chem. Soc.
2015, 137, 4896–4899. (b) Demissie, T. B.; Ruud, K.; Hansen, J. H.
2
3
4
5
6
Organometallics, 2015, 34, 4218–4228.
7
50. a) Pan, D.; Phillips, D. L. J. Phys. Chem. A 1999, 103, 4737-
4743; b) Okamoto, T.; Kuratsu, M.; Kozaki, M.; Hirotsu, K.;
Ichimura, A.; Matsushita, T.; Okada, K. Org. Lett. 2004, 6, 3493-
3496.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
51. The relative stabilities of 2●+, 2●+Br−-C, and 2●+Br−-I are
computed at different level of theories to benchmark the accuracy of
the computational methods. The method used in this study gives good
agreement
with
the
high-level
DF-LUCCSD(T)/cc-pVTZ
calculations. The computed stability of the ion pair complexes is also
affected by the implicit solvation model. Calculations using the
CPCM solvation model in place of SMD predicted a more favorable
dissociation of the 2●+Br− ion pair due to greater stability of the
bromide anion in solution. See the Supporting Information for details.
52. Among the two low energy isomers of ion pair 2●+Br−, only
the covalent isomer 2●+Br−-C is considered in the calculations of the
deactivation mechanisms. Due to the weak interactions between 2●+
and Br− in the ionic isomer of the ion pair, the reactivity of 2●+Br−-I
is expected to be similar to the separated catalyst radical cation and
halide anion.
53. Savéant, J.-M. J. Electroanal. Chem. 2000, 485, 86-88.
54. Shine, H. J.; Silber, J. J.; Bussey, R. J.; Okuyama, T. J. Org.
Chem. 1972, 37, 2691-2697.
55. (a) R.C. Tolman, Statistical Mechanics, Chemical Catalog Co,
New York, 1927, p. 247. (b) J.R. Partington, An Advanced Treatise
on Physical Chemistry, vol. 1, Longmans, London, 1967, p. 292. (c)
J.W. Moore, R.N. Pearson, Kinetics and Mechanism, vol. 131, Wiley,
New York, 1981, p. 130.
56. a) Matyjaszewski, K.; Paik, H.-j.; Zhou, P.; Diamanti, S. J.
Macromolecules 2001, 34, 5125-5131; b) Tang, W.; Kwak, Y.;
Braunecker, W.; Tsarevsky, N. V.; Coote, M. L.; Matyjaszewski, K.
J. Am. Chem. Soc.2008, 130, 10702-10713; c) Zerk, T. J.; Bernhardt,
P. V. Inorg. Chem. 2014, 53, 11351-11353.
15
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 16
1
2
3
Insert Table of Contents artwork here
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
16
ACS Paragon Plus Environment