C-ON bond homolysis of alkoxyamines: When too high polarity is detrimental

Article abstract of DOI:10.1039/c7ob01312d

Throughout the last decade, the effect of electron withdrawing groups (EWGs) has been known to play a role-minor or moderate depending on the nitroxyl fragment R₁R₂NO-in the change in the homolysis rate constant (k_d) for C-ON bond homolysis in alkoxyamines (R₁R₂NOR). It has been shown that the effect of EWGs on k_d is described by a linear relationship with the electrical Hammett constant σ_I. Since then, linear multi-parameter relationships f(σ_RS,ν,σ_I) have been developed to account for the effects involved in the changes in k_d, which are the stabilization of the released radical (σ_RS) and the bulkiness (ν) and polarity (σ_I) of the alkyl fragment. Since a decade ago, new results have been published highlighting the limits of such correlations. In this article, previous multi-parameter linear relationships are amended using a parabolic model, i.e. (σ_I,nitroxide - σ_I,alkyl)², to describe the effect of EWGs in the alkyl fragment on k_d. In contrast to previous studies, these improved linear multi-parameter relationships f(σ_RS,ν,Δσ_I²) are able to account for the presence of several EWGs on the alkyl fragment, R. An unexpectedly strong solvent effect-a ca. 1500-fold increase in k_d-from tert-butylbenzene to the water/methanol mixture is also observed for 3-((2,2,6,6-tetramethylpiperidin-1-yl)oxyl)pentane-2,4-dione 1b in comparison to a ca. 5-fold increase in k_d that is generally observed.

Full text of DOI:10.1039/c7ob01312d

Organic &
Biomolecular Chemistry
View Article Online
View Journal
PAPER
C–ON bond homolysis of alkoxyamines: when too
high polarity is detrimental†
Cite this: DOI: 10.1039/c7ob01312d
Paulin Nkolo,a Gérard Audran,*a Raphael Bikanga,b Paul Brémond,a
Sylvain R. A. Marque
*
a,c and Valérie Roubauda
Throughout the last decade, the effect of electron withdrawing groups (EWGs) has been known to play a
role – minor or moderate depending on the nitroxyl fragment R1R2NO – in the change in the homolysis
rate constant (kd) for C–ON bond homolysis in alkoxyamines (R1R2NOR). It has been shown that the
effect of EWGs on kd is described by a linear relationship with the electrical Hammett constant σI. Since
then, linear multi-parameter relationships f(σRS,ν,σI) have been developed to account for the effects
involved in the changes in kd, which are the stabilization of the released radical (σRS) and the bulkiness (ν)
and polarity (σI) of the alkyl fragment. Since a decade ago, new results have been published highlighting
the limits of such correlations. In this article, previous multi-parameter linear relationships are amended
using a parabolic model, i.e. (σI,nitroxide − σI,alkyl)2, to describe the effect of EWGs in the alkyl fragment on
kd. In contrast to previous studies, these improved linear multi-parameter relationships f(σRS,ν,ΔσI2) are
able to account for the presence of several EWGs on the alkyl fragment, R. An unexpectedly strong
solvent effect – a ca. 1500-fold increase in kd – from tert-butylbenzene to the water/methanol mixture is
also observed for 3-((2,2,6,6-tetramethylpiperidin-1-yl)oxyl)pentane-2,4-dione 1b in comparison to a
ca. 5-fold increase in kd that is generally observed.
Received 29th May 2017,
Accepted 16th June 2017
DOI: 10.1039/c7ob01312d
rsc.li/obc
molecules. In the last decade, empirical and theoretical
Introduction
models11,12 have been developed to account for the different
In the last few years,1 the cleavage/reformation property of effects that influence the rate constant (kd) of the C–ON bond
alkoxyamines, affording an unexpectedly long lifetime for this homolysis in alkoxyamines (Scheme 1) as well as to drive the
species, has been implemented in various applications such as design of new alkoxyamines that exhibit the best efficiency for
self-healing polymers,2,3 materials for photonics,4 and coding various applications.1–8 Hence, several empirical multiparameter
systems.5,6 Recently, the external triggering of the C–ON bond equations based on different types of Hammett or Taft constants
in alkoxyamines was used to highlight7 the potential use of have been developed to account for the effects observed in the
these molecules as agents for Theranostics.8 For three nitroxyl and the alkyl fragments.13,14 Thus, in eqn (1),‡ electrical
decades, alkoxyamines had been used as initiators/controllers Hammett constant σI describes the effect of electron withdrawing
in Nitroxide Mediated Polymerization.9 Today, restrictive regu- groups (EWGs) on both the stabilization of the released nitroxide
lations encourage the development of more efficient and more and the change in polarity in the alkoxyamine, and Es accounts
secure initiators.10 Hence, all these new developments require for the bulkiness of the nitroxyl fragment.§ For the alkyl frag-
the accurate and reliable knowledge of the factors controlling ment,15 two equations are developed depending on the nitroxyl
the C–ON bond homolysis, both to design new efficient mole- fragment: eqn (2) ¶ for 1• (so-called TEMPO)∥ and eqn (3) ** for
cules and to get deeper insight into the reactivity of these 2• (so-called SG1)∥ as released nitroxides – σRS to account for the
aAix Marseille Univ, CNRS, ICR, UMR 7273, case 551, Avenue Escadrille Normandie-
Niemen, 13397 Marseille Cedex 20, France. E-mail: sylvain.marque@univ-amu.fr,
g.audran@univ-amu.fr
bLaboratoire de Substances Naturelles et de Synthèse Organométalliques Université
des Sciences et Techniques de Masuku, B.P. 943, Franceville, Gabon
cN. N. Vorozhtsov Novosibirsk Institute of Organic Chemistry SB RAS, Pr. Lavrentjeva
9, 630090 Novosibirsk, Russia
‡N = 239, R2 = 0.95, t = 99.99%, F0.01 = 210. See ref. 13.
§This relationship was only developed for the released radical h•.
¶N = 14, R2 = 0.96, for ρRS and δ, t = 99.99%, for ρI, t = 99.45% and F0.01 = 80. See
ref. 15.
∥N = 19, R2 = 0.85, t = 99.99%, F0.01 = 29. See ref. 15.
**In ref. 15, alkoxyamines are distributed into two families: one for nitroxyl frag-
ments carrying weak EWGs or EDGs with 1-based alkoxyamines as representa-
tives, and one for nitroxyl fragments carrying EWGs with 2-based alkoxyamines
as representatives.
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
c7ob01312d
This journal is © The Royal Society of Chemistry 2017
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Scheme 1 Different effects that play roles in the changes in kd.
Reproduced by permission of the Royal Society of Chemistry (see
ref. 15).
Fig. 1 The alkyl radicals and nitroxides discussed.
stabilization of the released radical, σl for the effect of EWG and
ν for the bulkiness of the alkyl fragment.
logðkd=sꢀ1Þ ¼ ꢀ5:88ð+0:28Þ ꢀ 3:07ð+0:28Þ ꢁ σI ꢀ 0:88ð+0:04Þ
pared as already reported.22 Alkoxyamines 1b,c and 2a–c
(Scheme 2) were prepared by scavenging the alkyl radicals a•–c•
in the presence of nitroxides 1• and 2•, respectively. Radicals
a•–c• were generated by oxidation of the anions of malonic acid
derivatives in the presence of CuCl2, which were prepared
using lithium diisopropylamine (LDA) as a base. Yields ranged
from poor (ca. 20% for 2-based alkoxyamines) to moderate
(ca. 70% for 1-based alkoxyamines).22 Moreover, alkoxyamine
2b was only detected at low temperatures and it decomposed
upon warming to room temperature.
ꢁ Es
ð1Þ
logðkd=sꢀ1Þ ¼ ꢀ 14:8ð+0:7Þ þ 13:9ð+0:9Þ ꢁ σRS
ð2Þ
þ 13:6ð+3:2Þ ꢁ σI þ 6:6ð+0:7Þ ꢁ ν
logðkd=sꢀ1Þ ¼ ꢀ 14:3ð+1:3Þ þ 15:3ð+2:2Þ ꢁ σRS
ð3Þ
þ 19:5ð+3:0Þ ꢁ σI þ 7:0ð+1:1Þ ꢁ ν
These correlations were developed with more than 15
kd values each. It was shown that kd increases with increasing
stabilization of the released radical and with an increase in
both the polarity and bulkiness of the substituents R1,2 and X
(Scheme 1).16 The robustness, accuracy, and reliability of
eqn (2) and (3) have been tested over the years.17,18 However,
some years ago, Studer and coll.19 reported the reactivity of 1a,
1e and 1d (Fig. 1) and observed higher homolysis activation
energies Ea (Table 1) than expected from eqn (2). Taking into
account the aforementioned requirements, this led us to
thoroughly investigate the polar effect of the alkyl fragment
with 7 new alkoxyamines 1a–c,f,g, and 2a–c (Fig. 1), which all
carry polar alkyl fragments and release stabilized radicals.
Their kd values were measured in tert-butylbenzene and in a
water/MeOH mixture. The results showed an overestimated
polar effect for the TEMPO-based fragment carrying only elec-
tron donor groups (EDGs) and a rather good estimate of the
polar effect for the SG1-based fragment carrying EWGs. DFT
calculations were performed to get deeper insight into how the
polarity of the nitroxide influences the polar effect of the alkyl
fragment.
Hydrobromination of styrene oxide afforded the expected
bromohydrin.23 The latter, using the Atom Transfer Radical
Addition (ATRA) procedure in the presence of 1•,24 was trans-
formed into the corresponding alkoxyamine.25 After purifi-
cation, the alkoxyamine was oxidized26 into 1g or was trans-
formed25 into 1f in the presence of acetic anhydride (Scheme 3).
Kinetic measurements
The values of kd were measured by EPR and used O2 as the
alkyl radical scavenger, as previously reported (Fig. 1SI†).27
Except for 1a and 1c – for which the plateau was not reached
due to an excessively long experiment time†† – and for 2b –
which was not stable enough to be handled at room tempera-
ture – the plateau of the concentration of nitroxide was
reached at the expected level for all of the other alkoxyamines
(Table 1). The values of kd for alkoxyamines 1a, 1d, and 1e
have been measured by Studer and coll.19 A difference of ca.
6 kJ mol−1 for 1a cannot be accounted for so far. Alkoxyamines
1k and 1l have been measured by Megiel et al.28 in acetonitrile.
A difference of ca. 7 kJ mol−1 between 1g and 1k cannot be
thoroughly rationalized, because the same stabilization is
expected for radicals g• and k•‡‡ as their polarity is the same
Results
Preparation of 1a–c,f,g,m and 2a–c
††In such cases, the initial slope method was used. See ref. 27.
‡‡aHα = 19.0 G for HCOCH2 and aHα = 19.5 G for MeCOCH2, meaning that these
two radicals are similarly stabilized. Thus, it is assumed that σRS,g• ≈ σRS,k• or is
Alkoxyamine 1m was prepared as reported for 2m using nitrox-
ide 1• as a alkyl radical scavenger.20,21 Alkoxyamine 1a was pre- slightly lower.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 1 Experimental C–ON bond homolysis rate constant k’d at temperature T, activation energy Ea (kJ mol−1) and corresponding kd at 120 °C
Alkoxyamines
Solvent
T (°C)
k′d (10−4 s−1
)
Ea
kd (120 °C in s−1
)
Ref.a
1a
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
160
6.2
145.8
140.0
144.9
139.3
137.7
132.1
131.1
121.5
132.9
141.8
161.5
114.0b
132.4b
130.3
126.1
123.8
10−5
t.w.
19
t.w.
t.w.
19
6 × 10−5
1.3 × 10−5
7.3 × 10−5
1.2 × 10−4
6.6 × 10−4
9.0 × 10−4
0.017
1b
1c
1d
1e
1f
1g
1h
1i
1j
1k
1l
1m
1mH+
2a
2b
2c
2h
2i
2j
2m
2mH+
1a
1b
1c
150
150
3.1
15.1
19
130
70
0.98
0.76
t.w.
t.w.
27
27
27
28
28
t.w
t.w.
t.w.
n.d.
t.w.
27
27
27
20 and 21
20 and 21
t.w.
t.w.
t.w.
t.w.
t.w
t.w.
t.w.
t.w.
30
t-BuPh
t-BuPh
t-BuPh
5.2 × 10−4
3.4 × 10−5
8.2 × 10−8
0.17
Acetonitrile
Acetonitrile
t-BuPh
6.1 × 10−4
1.1 × 10−3
4.2 × 10−3
0.008
131
100
110
34.9
5.3
31.3
t-BuPhc
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
t-BuPh
Degradation at −30 °C
6.0/14.0
80
118.9/116.4
125.0/126.4
128.4/130.8
149.1
123.0
115.6
125.8
121.2
121.6
122.4
127.1
121.0
119.1
114.4/113.1
124.0/123.0
119.2/119.4
109.7/108.3
0.037/0.081
5.8 × 10−3/3.8 × 10−3
2.1 × 10−3/10−3
3.6 × 10−6
0.01
0.1
H2O/MeOHd
H2O/MeOHd
H2O/MeOHd
H2O/MeOHd
H2O/MeOHd,e
H2O/MeOHd,g
H2O/MeOHd
H2O/MeOHd
H2O/MeOHd
H2O/MeOHd
H2O/MeOHd,g
80
80
80
80
81
81
80
70
0.58
4.8
2.4
1.80
0.44
3.4
4.6 × 10−3
0.019
1.6 × 10−2
0.013
1g
1m
1mH+f
2a
2c
2h
2m
2mH+
3.0 × 10−3
2.0 × 10−2
0.035
0.6
9.0/14.4
0.149/0.222
7.9 × 10−3/0.011
3.4 × 10−2/3.2 × 10−2
0.63/0.97
32
31
a t.w. = this work and n.d. = not determined. b Values reported in acetonitrile. Original rate constants were divided by a factor of 2 to account for
the solvent effect. See ref. 27 and 29. c Protonation was performed in situ in the presence of 2 equivalents of TFA. See the ESI† for the 1H NMR
signals. d MeOH/H2O 1 : 1 v/v. e pH = 7.0. f pKa = 4.22, see the ESI.† g pH = 2.0.
Scheme 3 Preparation of 1f and 1g. (a) 47% HBr (aq), CHCl3, 30 min, rt;
(b) CuBr/Cu(0)/PMDETA, benzene, 12 h, rt; (c) TPAP, NMO, CH2Cl2, 2 h,
0 °C; (d) Ac2O, Et3N, CH2Cl2, 8 h, rt.
Scheme 2 Preparation of 1a–c and 2a–c.
and their bulkiness is not expected to differ too much
potential for application as agents for theranostics, the
(vide infra).§§ The solvent effect was taken into account as
kd values were also measured in a water/MeOH (v/v 1 : 1)
kd,acetonitrile/kd,tBuPh = 2.27,29 As alkoxyamines exhibit interesting
mixture.30 Whatever the solvent, diastereoisomers of 2c¶¶ only
exhibit a difference of 1–2 kJ mol−1 – a difference already
§§The kd value for 1k has been measured in acetonitrile (see ref. 28). The few
data available for type-1 alkoxyamines have been obtained using weakly polar
alkoxyamines. Thus, an unexpected moderate/strong solvent effect cannot be
straightforwardly disregarded.
¶¶As 2c is an oil, X-ray analysis was not performed and the configurations were
not determined.
This journal is © The Royal Society of Chemistry 2017
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
reported many times for diastereoisomers – and do not However, it was assumed that changes in the electronegativity
deserve further comment.
of the alkyl fragment, i.e., changes in χC, are described by the
electrical Hammett constant σI (eqn (4)). Therefore, a plot of σI
vs. log kd is equivalent to the plot of χC vs. (χO − χC)2which has
a parabolic shape (y = x2) as displayed in Fig. 3. Assuming a
nitroxyl fragment carrying a strong EWG for χC at position A
(Fig. 3), a large (χO − χC)2 value and hence a low kd value are
observed. Then, moving to position B (increasing χC by increas-
ing the polarity), (χO − χC)2 decreases and hence kd increases
linearly (red slope in Fig. 3) due to the shape of the curve.
Assuming a nitroxyl fragment with EDGs, the parabolic shape
is expected to flatten and is described by the violet line in the
case of eqn (2) and by dotted green lines and positions C–E in
Multi-parameter relationships
34,35
Parameters ν,33 σRSE
,
and σI36,37 were estimated as pre-
viously reported15 and are shown in Table 1SI.† ∥∥ The coeffi-
cients in eqn (2) were re-estimated using ν = 0.58 for
CH2COOMe,33 ν1 = νiPr = 0.76 for CHCH3COOMe, ν1 = νtBu
=
1.24 for CMe2COOMe, and νCOOMe
νCHMeCOOMe = 0.83 and νCMe COOMe = 1.25. Therefore, through
=
0.5,33 affording
2
using these new values and implementing 1f and 1l in eqn (2),
which are expected to be included in the correlation, very good
statistical outputs (eqn (4))*** were obtained, supporting our
new assumptions (Fig. 2). The weight of each effect was esti-
mated as previously reported15 and afforded 39%, 20%, and
41% for the stabilization, polar, and steric effects, respectively.
These values are very close to those previously reported (44%,
16%, and 40%, respectively)15 and do not deserve further
comment.
Fig. 3. Thus, for χC at position C, a small difference in (χO
−
χC)2 is observed, and, moving to position D, (χO − χC)2
decreases only a little, affording a small increase in kd (purple
slope in Fig. 3), and moving further to position E, the pre-
dicted polar effect is more important than observed, i.e., the
purple line is below the parabolic curve.
1
2
2
logðkd=sꢀ1Þ ¼ ꢀ 14:8ð+0:4Þ þ 12:4ð+0:9Þ ꢁ σRSE
BDEðA ꢀ BÞ ¼ ðBDEðA ꢀ AÞ þ BDEðB ꢀ BÞÞ þ aðχA ꢀ χBÞ
ð4Þ
þ 17:7ð+2:2Þ ꢁ σI þ 7:1ð+0:5Þ ꢁ ν
ð5Þ
In the case of homolysis, the activation energy Ea is very
similar to the Bond Dissociation Energy (BDE) which depends
on the enthalpic and polar terms, as proposed by Pauling38–40
(eqn (5)). The polar term is given by the square of the differ-
ence in electronegativity χ.
Hence, as log kd is proportional to Ea – and consequently to
the BDE – it has to be proportional to (χO − χC)2 which
describes the effect of the changes in polarity in eqn (5).
DFT calculations
To get deeper insight into the polar effect involved in the
changes in kd, DFT calculations were performed for 1a and 2a
in water and toluene as solvents at the M062X/6-31+G(d,p)
level of theory. The PCM standard calculations method has
been used for calculations in solvents.41 Typical bond lengths,
distances and angles are reported in Table 2SI† and do not
differ from those reported for the molecules in the same
family. The dihedral angles θ1–θ3 (as well as the θ′1–θ′3 needed
for their estimations, Fig. 4) for the interactions required at
the TS (vide infra) are reported in Table 2. The atoms that are
involved in dihedral angles θ are displayed in Fig. 4. Angle θ1
was chosen as the smallest value to reach 90° (vide infra), θ2 for
the second ester moiety, and θ3 to describe the position of the
alkyl fragment.
Calculated ΔHr and ΔΔHr values are different to the experi-
mental Ea and ΔEa values (see ESI†) but the trends are the
same. Consequently, the calculations and the subsequent
parameters reliably describe the reactivity observed for 1a and
Fig. 2 A plot of log(kd/s−1) vs. f(σRSE,σI,ν). ( ) represents the 1-based
■
alkoxyamines used in eqn (4) with the extrema displayed; the most
deviating data included in the correlation are highlighted by the vertical
●
red lines; ( ) represents k and g for which a range of parameters was
□
defined (horizontal black lines), and ( ) represents outlying data.
∥∥The parameters σRS, ν, and σI were used to account for the effect of the stabi-
lization of the released alkyl radical, the steric hindrance in alkoxyamines due to
the bulkiness of the alkyl fragment, and the electron donating or electron with-
drawing properties of the EDGs and EWGs, respectively, on kd. More details are
provided in the ESI.†
***R2 = 0.98, s = 0.51, N = 16, F99.99% = 202, and t = 99.99% for all coefficients.
Fig. 3 Parabolic display of the polar effect.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Organic & Biomolecular Chemistry
Paper
Discussion
Studer and coll.19 reported kd values for 1a and 1d that are very
close to those for 1i, although the methyl group was replaced
by the more polar and stabilizing EWGs COOEt and CON
(OMe)Me (Table 1SI† and dotted line in Fig. 2). This puzzling
result led us to re-investigate the stabilization and polar and
steric effects of 1a–c,f,g,m and 2a,c which carry groups with
very different polarities. Interestingly, kd values for 1a–e are
estimated by eqn (4) and are 180-, 1500-, 64-, 21- and 7-times
larger than when experimentally observed!††† All of these kd
values are clearly lower than expected, meaning that the polar,
steric and stabilization effects are not properly described by
their respective parameters. The stabilization parameter σRS is
estimated from hyperfine coupling constants and the polar
parameter is given by well tabulated Hammett constants σI. On
the other hand, the steric parameter ν is estimated using
several assumptions (vide supra). However, no realistic assump-
tions can account for the large difference that is observed.†††
In sharp contrast to what happens for type-1 alkoxyamines,
the predicted and the experimental values of Ea for 2c differ by
less than 1 kJ mol−1; 2b decomposes spontaneously despite
having an Ea value estimated at ca. 108 kJ mol−1, meaning that
the bulkiness of b is likely to be underestimated, and the Ea
for 2a is 5 kJ mol−1 – not 10 kJ mol−1 as expected – lower than
that for 2i.‡‡‡ Thus, eqn (3) provides a better description of
the effects involved in the C–ON bond homolysis in type-2
alkoxyamines than eqn (2) does for type-1 alkoxyamines.
Fig. 4 NBO charge distributions on the O atom of the nitroxyl moiety
and on the methine carbon in 1a (left) and 2a (right), in water and
toluene as solvents. The red, blue, and green colours highlight the di-
hedral angles θ’1, θ’2 and θ’3, respectively.
Table 2 Geometrical parametersa for 1a and 2a for the angles required
in the TS: θ1 (dihedral angle σO–Cπ*C1vO1), θ2 (dihedral angle
σO–Cπ*C2vO2), θ3 (dihedral angle LPσ*O–C) and the calculated comp-
lementary dihedral angles θ’1, θ’2, and θ’3
1a
2a
Solvent
Toluene
Water
Toluene
Water
θ′1 (<OCC1O1>)
θ1
114
24
173
83
128
8
114
24
174
84
129
9
52
38
14
76
113
8
31
59
26
64
127
−8
b
θ′2 (<OCC2O2>)
c
θ2
θ′3 (<CNOC>)
d
θ3
As shown by the DFT calculations, the difference in the
charge distribution is larger in 1a than in 2a, meaning that the
difference in electronegativity χ between the O and C atoms of
the C–ON bond is larger in 1a than in 2a. In our case, this
difference in charges highlights the difference in (χO − χC)2 of
the C–ON bond and is accounted for by the Hammett constant
σI. As the nitroxyl fragment 2 carries strong EWGs (σI = 0.28),13
a In degrees. Calculated by DFT at the M062X/6-31+G(d,p) level of
theory. The solvent effect was estimated using the PCM method. The
numbering is displayed in Fig. 4. b θ1 = |θ′1 − 90°|. c θ2 = |θ′2 − 90°|.
d θ3 = θ′3 − 120°.
2a qualitatively, in water as well as in toluene. Thus, some changes in kd are described by positions A and B in Fig. 3 and
interactions that result in stabilization at the TS and a differ- eqn (3) holds true, i.e. it is a linear description of the polar
ence in the conformation of the TS with respect to the starting effect. In contrast, for the nitroxyl fragment 1, which does not
materials are poorly accounted for by the calculated reaction carry EWGs (σI = −0.06),13 the change in kd, i.e., the large
enthalpy.
The NBO charges calculated for 1a and 2a in toluene E in Fig. 3 and it is eqn (6) that is valid instead of eqn (2).
(Fig. 4) show a larger difference (Δ) for 1a than for 2a, denoting
It is clear that these two examples describe the polar effect
difference in electronegativity, is described by positions C and
a larger difference in the electronegativity (χ) between the observed in type-1 and type-2 alkoxyamines: a strong polar
C and O atoms in 1a than that in 2a. A smaller difference is effect (red slope and dashed lines in Fig. 3) occurs for type-2
observed in water (Fig. 4), i.e., δΔ = 0.035 in toluene and δΔ = alkoxyamines, due to the presence of EWGs on the nitroxyl
0.019 in water, meaning that the stabilization of the polar car- fragment, described by a linear change in kd and is well pre-
bonyl moieties in water decreases the electronegativity of the C dicted for a broad range of values of σI, and a weak polar effect
atom, leading to an increase in Δ. No significant difference in (purple slope and dotted lines in Fig. 3) occurs for type-1
the Radical Stabilization Energy (RSE) was observed when alkoxyamines, due to the presence of EDGs on the nitroxyl
changing the solvent (see ESI†).
Dipole moments, charges at the proton of the methine
carbon in the alkyl fragment and energies E for bonding →
anti-bonding orbital interactions in the alkyl fragments are
reported for 1a and 2a in Table 3SI.† Geometrical parameters
†††Assuming CO(OMe), Ac, CO(N(OMe)Me) and COt-Bu groups are as sterically
demanding as an H atom, νCH Ac = 0.76 for a•, b•, c•, and e•. Except for 1e, which
2
lies on the correlation line, the others are clearly downward outliers.
(distances and valence angles) for intramolecular H-bonding
(IHB) are reported in Table 3SI† for 1a and 2a.
‡‡‡This difference might be due to a conformational effect as the conformations
of 1i and 2i are different.
This journal is © The Royal Society of Chemistry 2017
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
fragment, described by a linear change in kd on a very narrow
range of σI values.
This claim is nicely supported by the Ea for 2mH+ being
smaller by ca. 8 kJ mol−1 (ref. 20 and 21) (i.e. a 10-fold increase
in kd) than that for 2m (going from A to B in Fig. 3), whereas
the Ea for 1mH+ is only ca. 4 kJ mol−1 (i.e. a 4-fold increase in
kd, Table 1) smaller than the Ea for 1m (going from C to D in
Fig. 3). Keeping in mind that the protonation of m does not
change the stabilization of mH+• nor the steric demand in
mH+, as this protonation is performed at the para position of
the pyridyl ring;20 changes in kd are only caused by changes in
polarity. Thus, due to the parabolic change in electronegativity
as described above, the polar effect is clearly weaker in type-1
alkoxyamines than in type-2 alkoxyamines.
Fig. 6 (a) The charge distribution for a polar TS. (b) The orbital inter-
actions in the expected TS for the homolysis of the C–ON bond.
Taking these results into account and based on reasonable
assumptions,§§§ the data gathered in Table 1 combined with observed for 1g. The role of the solvent is to stabilize the reactant,
the data reported in the literature are fitted with eqn (6) (Fig. 5). products, and transition state (TS). As the homolysis displays the
Except for 1k¶¶¶ and 1b,∥∥∥ which are outliers, all of the other late TS, i.e., it is product-like, any stabilization of the products
data reported in Table 1 are nicely accounted for by eqn (6) **** affords the stabilization of the TS. The polar effect destabilizes
(Fig. 5), in sharp contrast to the poor efficiency of eqn (4) the reactant (alkoxyamine), thereby decreasing Ea. The insignifi-
(Fig. 2). As expected, the coefficient for the polar effect is nega- cant changes in Ea for 1g and 2h (Table 1) when changing the
tive, meaning the smaller (σI,TEMPO − σI)2 is, the higher kd is.
solvent from t-BuPh to water/MeOH show that the stabilization of
the nitroxide and alkyl radicals due to solvent effects is not
strong enough to overbalance the solvation of the alkoxyamine.
Moreover, the DFT calculations showed that the solvent effect on
the RSE (see the ESI†) can be disregarded. On the other hand,
calculations show significant charge separations in the hetero-
nuclear O–C bond (Fig. 4) due to the presence of strong EWGs
on the alkyl fragment. Consequently, a polar TS is expected
(Fig. 6a) to be more stabilized in water than in a non-polar
solvent such as t-BuPh, partly accounting for the striking ca.
20 kJ mol−1 decrease observed in the Ea for 1a. The same effect
should be expected for 2a, however, a strikingly lower effect is
observed, i.e., only a decrease of 4.7 kJ mol−1 in Ea.
logðkd=sꢀ1Þ ¼ ꢀ 12:2ð+0:4Þ þ 11:4ð+0:9Þ ꢁ σRS
2
ꢀ 95:0ð+0:9Þ ꢁ ðσI;TEMPO ꢀ σIÞ þ 7:0ð+0:5Þ ꢁ ν
ð6Þ
The values of Ea for 1a–c decrease dramatically by around
20 kJ mol−1 (Table 1) when the solvent is changed from tert-butyl-
benzene to water/MeOH, whereas no change (ca. 1 kJ mol−1) is
It has been proposed that the TS implies that there are
several orbital interactions between the nitrogen lone pair (n)
and the bonding and anti-bonding orbitals n → σ*O–C, σO–C
→
π*CvO, as well as re-hybridization at the N, O, and C atoms,
i.e., from sp3 to sp2 (Fig. 6). For optimal interactions in the TS
(Fig. 6b), the dihedral angles θ3 and θ1 need to be as close as
possible to 0° to favour the interactions n → σ*O–C and σO–C
→
π*CvO. Consequently, the entropic cost to reach a confor-
mation fulfilling these requirements in the TS is expected to
be low when these requirements are already observed in the
starting materials. Interestingly, in 1a and 2a, θ3 is very close
to 0°, meaning that the orbital and bond are already in the
required position in the starting materials for efficient n →
σ*O–C interaction in the TS, implying that the TS can be
reached without entropic cost.
Fig. 5 A plot of log(kd/s−1) vs. f(σRSE,f(σI),ν). ( ) represents 1-based
■
alkoxyamines used in eqn (6), ( , blue empty square) represents new
alkoxyamines and ( , red empty square) represents outlying data.
§§§The polar effect in the model is given by (χO − χC)2 and assumed to be equi-
valent to (σI,TEMPO − σI)2. As electron donating methyl groups are attached to the
By taking these statements into account, the ca. 20 kJ mol−1
nitroxyl moiety, it is assumed that σI,TEMPO = σI,H NO = 0.16. See ref. 37.
2
¶¶¶Taking into account footnote §§ and as σRS was estimated neither reliably nor decrease in Ea due to the solvent effect observed in 1a is nicely
accurately this data was removed from the correlation set. Nevertheless, kd is still
described by the stabilization of the polar TS, which is
enhanced by θ1 = 24° both in t-BuPh and in water (Fig. 7 and
Table 3SI†) and is close enough to 0° to generate a low entro-
14-fold underestimated.
∥∥∥The kd value of 1b is clearly an outlier, as it is estimated to be 153-fold stron-
ger than observed. This means that the steric hindrance of CH(CHO)2 is the
pic cost. The same stabilization of the TS is expected for 2a, in
sharp contrast to the 4 kJ mol−1 decrease in Ea (Table 1)
same as that of the ethyl group. At this stage, we have no rationale for this result.
****R2 = 0.94, t = 99.99% for all coefficients, F0.01 = 233, and N = 22.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 8 Calculated IHB interactions for 1a in toluene and water (a) and
for 2a in toluene (b) and in water (c). The blue lines show IHB.
Fig. 7 Newman projections along the N–O bond (a) and along the
C–C1 bond in the alkyl fragments for 1a both in t-BuPh and water
(b) and for 2a in t-BuPh (c) and in water (d). The dotted line shows the
expected position of the CvO moiety in the TS. The θ values (in red) are
the values that the angles have to close by in order to reach the
expected conformation in the TS.
observed by changing the solvent from t-BuPh to water/MeOH.
This weaker solvent effect is caused by both a smaller differ-
ence in the solvation between TS and alkoxyamine which
affords a weaker solvent effect and the entropic cost due to the
change in the conformation from 2a to the TS. First, in t-BuPh,
the steric strain in 2a affords a conformation exhibiting a
larger θ1 value than that of 1a, that is, θ1 = 38° and θ1 = 24°,
respectively (Table 2). Secondly, in water, θ1 = 59° is even larger
than that in t-BuPh which increases the entropic cost to reach
the required conformation in the TS. Consequently, gains in
energy due to the increase in steric strain and polarity and
solvent effects going from 1a to 2a and from t-BuPh to water/
MeOH are almost balanced by the energetic cost due to the
solvation and the detrimental changes in conformation in 2a.
Conformation changes are mainly ruled by 4 effects: (i) sta-
bilizing hyperconjugation, (ii) destabilizing steric strain which
has to be minimized, (iii) a dipole moment which has to be
minimized in non-polar solvents, and (iv) H-bonding.
A thorough analysis of the geometrical parameters and
interactions aforementioned (Table 3SI†) in alkoxyamines 1a
and 2a shows that, for 1a, the geometrical parameters (Tables
2SI and 3SI†) and the interactions are the same in water and
toluene, whereas for 2a the geometrical parameters (Tables 2SI
and 3SI†) are clearly different in water and toluene, as well as
the interactions which differ in number and strength and from
those reported for 1a. 3 hyperconjugation interactions (E =
49 kJ mol−1, Table 3SI†)†††† and 2 IHB interactions‡‡‡‡
Fig. 9 DFT calculated dipole moments μ for 1a in toluene (μ = 4.1 D)
and water (μ = 4.9 D): (a) and (b) respectively, and for 2a in toluene (μ =
4.0 D) and water (μ = 6.0 D): (c) and (d), respectively. The blue arrows
show the dipole moment vectors.
(Fig. 8 and Table 3SI†) stabilize 1a whatever the solvent which
means that there are no other conformations providing less
steric strain or better hyperconjugation interactions which
would overbalance the increase in the dipole moment μ
(Fig. 9).
In sharp contrast to 1a, the conformation of 2a in water
(Fig. 9) exhibits a larger dipole moment (μ = 6.0 D) and greater
steric strain (the shortest distances between the carbonyl moi-
eties and methyl groups are 3.11 Å and 3.17 Å, Fig. 8) than the
conformation in toluene (μ = 4.0 D, and 3.30 Å and 3.54 Å,
Fig. 8). These changes are governed by both the occurrence of
hyperconjugation interactions and IHB in toluene – 3 hyper-
conjugation interactions ((E = 50 kJ mol−1, Table 3SI†)††††
and 2 IHB interactions‡‡‡‡ (Table 3SI† and Fig. 8)) – and
those in water – 4 hyperconjugation interactions ((E = 53
kJ mol−1
(Table 3SI† and Fig. 8)).
, Table 3SI†)†††† and 3 IHB interactions‡‡‡‡
The solvent effect in 1m/1mH+ affords an increase in
kd that is similar to the one observed for 2m/2mH+. As
observed for 2m/2mH+, the solvent effect is stronger for 1mH+
than for 1m, which is likely to be due to the solvation of the
counter-anion (Table 1).31,32
††††The sum of the energies with significant hyperconjugation contributions in
the ester moieties.
‡‡‡‡Intramolecular H-bonding is observed between H-acceptors and H-donors
for distances smaller than the sum of the van der Waals radii of the atoms
involved in IHB: rH = 1.09 Å, rN = 1.50 Å and rO = 1.52 Å. IHB also depends on the
valence angle α, i.e., strong IHB is expected when α values that are larger than
150° and weak IHB is expected when α values are smaller than 120°. IHB when α
values that are lower than 90° requires strained conformations. See ref. 42 and 43.
Experimental section
All of the solvents and reactants for the preparation of the
alkoxyamines were used as received. Routine reaction monitor-
ing was performed using silica gel 60 F254 TLC plates; the
This journal is © The Royal Society of Chemistry 2017
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
spots were visualized upon exposure to UV light and a phos- (CH2, d, J = 7.3 Hz), 52.7 (CH3), 52.6 (CH3), 35.7 (C, d, J =
phomolybdic acid solution in EtOH, followed by heating. 5.4 Hz), 29.4 (3CH3, d, J = 5.8 Hz), 27.9 (3CH3), 16.8 (CH3, d,
Purifications were performed on chromatography columns J = 5.7 Hz), 16.4 (CH3, d, J = 6.8 Hz). HRMS (ESI) calc. for
1
with silica gel grade 60 (230–400 mesh). H, 13C, and 31P NMR C18H37NO8P+: 426.2251 [M + H]+; found: 426.2252.
spectra were recorded in CDCl3 on a 300 or 400 MHz spectro-
Methyl 2-((tert-butyl(1-(diethoxyphosphoryl)-2,2-dimethyl-
meter. Chemical shifts (δ) were reported in ppm using residual propyl)amino)oxy)-3-oxobutanoate 2c. Yield of 23% as an oil.
non-deuterated solvents as the internal reference for 1H and For the minor isomer: 1H NMR (400 MHz, CDCl3): δ 4.98 (s,
13C-NMR spectra, and 85% H3PO4 for 31P-NMR spectra. High- 1H), 4.35 (dd, J = 14.9, 7.4 Hz, 2H), 4.11–3.92 (m, 2H), 3.73 (s,
resolution mass spectra (HRMS) were recorded on a SYNAPT 3H), 3.27 (d, J = 25.9 Hz, 1H), 2.46 (s, 3H), 1.31 (dt, J = 15.8,
G2 HDMS (Waters) spectrometer equipped with a pneumati- 6.3 Hz, 7H), 1.14 (s, 9H), 1.07 (s, 9H). 31P NMR (162 MHz,
cally assisted atmospheric pressure ionization source (API). CDCl3): δ 24.81. 13C NMR (101 MHz, CDCl3): δ 202.9 (CO),
Positive mode electrospray ionization was used on the 168.4 (COO), 93.6 (CH), 68.9 (CH, d, J = 137.7 Hz), 62.7 (C),
samples: electrospray voltage (ISV): 2800 V; opening voltage 62.1 (CH2, d, J = 6.6 Hz), 59.5 (CH2, d, J = 7.5 Hz), 52.6 (CH3),
(OR): 20 V; nebulizer gas pressure (nitrogen): 800 L h−1. Low 35.8 (C, d, J = 5.9 Hz), 29.4 (3CH3), 28.2 (3CH3), 26.9 (CH3),
resolution mass spectra were recorded on an ion trap AB 16.6 (CH3, d, J = 6.3 Hz), 16.3 (CH3, d, J = 6.6 Hz). HRMS (ESI)
SCIEX 3200 QTRAP instrument equipped with an electrospray calc. for C18H37NO7P+: 410.2302 [M + H]+; found: 410.2299. For
1
source. The parent ion [M + H]+ is quoted.
the major isomer: H NMR (400 MHz, CDCl3): δ 5.10 (s, 1H),
4.47–4.36 (m, 1H), 4.34–4.24 (m, 1H), 4.13–4.02 (m, 1H),
4.00–3.91 (m, 1H), 3.74 (s, 3H), 3.24 (d, J = 25.1 Hz, 1H), 2.26
(s, 3H), 1.32 (t, J = 7.2 Hz, 3H), 1.28 (t, J = 7.1 Hz, 3H), 1.10 (s,
General procedure for the preparation of alkoxyamines 1b,c
and 2a,c
Lithium diisopropylamine (2 M in THF/hexane), (4.2 ml, 9H), 1.06 (s, 9H). 31P NMR (162 MHz, CDCl3): δ 24.70. 13C
8.33 mmol, 1.1 eq.) was dissolved in THF (30 ml) and cooled NMR (101 MHz, CDCl3): δ 201.4 (CO), 167.4 (COO), 94.3 (CH),
to −78 °C. Alkanedione (1.0 g, 7.57 mmol, 1 eq.) was added 69.3 (CH, d, J = 138.4 Hz), 62.8 (C), 62.4 (CH2, d, J = 6.1 Hz),
and the resulting solution was stirred for 45 min prior to the 59.1 (CH2, d, J = 7.3 Hz), 52.6 (CH3), 35.8 (C, d, J = 5.9 Hz), 29.0
addition of a solution of nitroxide (12.1 mmol, 1.6 eq.) in THF (3CH3), 28.1 (3CH3), 27.2 (CH3), 16.9 (CH3, d, J = 5.4 Hz), 16.4
(14 ml). Copper(II) chloride (2.5 g, 18.9 mmol, 2.5 eq.) was (CH3, d, J = 6.7 Hz). HRMS (ESI) calc. for C18H37NO7P+:
added and the reaction mixture was warmed to room tempera- 410.2302 [M + H]+; found: 410.2298.
ture. After stirring for another 18 h, the reaction was quenched
2-Phenyl-2-((2,2,6,6-tetramethylpiperidin-1-yl)oxyl)acetaldehyde
by the addition of NH4Cl (aq. sat, 10 ml) and Na2CO3 (aq. sat, 1g. 2-Phenyl-2-((2,2,6,6-tetramethylpiperidin-1-yl)oxyl)ethanol
10 ml). The aqueous layer was extracted with diethyl ether (100 mg, 0.36 mmol, 1 eq.) was dissolved in dichloromethane
(3 × 10 ml) and dried over MgSO4. After concentrating under (3 ml) and cooled to 0 °C. Tetrapropylammonium perruthe-
reduced pressure, the residue was purified by column nate (TPAP, 13 mg, 36 µmol, 0.1 eq.) was added and the result-
chromatography.
ing solution was stirred until it turned black prior to the
3-((2,2,6,6-Tetramethylpiperidin-1-yl)oxyl)pentane-2,4-dione addition of a solution of N-methyl morpholine oxide (NMO,
1b. Yield of 75%. 1H NMR (400 MHz, CDCl3): δ 4.92 (s, 1H), 130 mg, 1.08 mmol, 3 eq.). After stirring for another 2 h, the
2.21 (s, 6H), 1.63–1.26 (m, 6H), 1.19 (s, 6H), 0.96 (s, 6H). reaction was quenched by the addition of Na2CO3 (aq. sat,
13C NMR (101 MHz, CDCl3): δ 203.9 (2CO), 101.7 (CH), 3 ml). The aqueous layer was extracted with dichloromethane
60.1 (2C), 40.3 (2CH2), 33.1 (2CH3), 27.2 (2CH3), 20.3 (2CH3), (3 × 5 ml) and dried over MgSO4. After concentrating under
17.0 (CH2). HRMS (ESI) calc. for C14H26NO3: 256.1907 reduced pressure, the residue was purified by column chrom-
1
[M + H]+; found: 256.1908.
atography (56 mg, 56%). H NMR (400 MHz, CDCl3) δ 9.66 (d,
Methyl 3-oxo-2((2,2,6,6-tetramethyl piperidine-1-yl))butano- J = 3.8 Hz, 1H), 7.41–7.29 (m, 5H), 5.09 (d, J = 3.8 Hz, 1H),
ate 1c. Yield of 52%. 1H NMR (400 MHz, CDCl3): δ 4.82 (s, 1H), 1.63–1.33 (m, 6H), 1.27 (s, 3H), 1.20 (s, 3H), 1.14 (s, 3H), 0.87
3.75 (s, 3H), 2.31 (s, 3H), 1.65–1.37 (m, 4H), 1.36–1.08 (m, 8H), (s, 3H). 13C NMR (101 MHz, CDCl3) δ 199.1 (CO), 135.4 (CH),
1.01 (d, J = 16.6 Hz, 6H). 13C NMR (101 MHz, CDCl3): δ 202.9 128.7 (2CH2), 128.2 (CH2), 127.0 (2CH2), 93.1 (CH), 60.2 (C),
(CO), 168.4 (COO), 93.6 (CH), 60.3 (C), 60.1 (C), 52.5 (CH3), 59.9 (C), 40.0 (2CH2), 33.9 (CH3), 33.4 (CH3), 20.5 (CH3), 20.2
+
40.2 (2CH2), 33.1 (CH3), 32.6 (CH3), 26.6 (CH3), 20.3 (2CH3), (CH3), 17.0 (CH2). HRMS (ESI) calc. for C17H26NO2 : 276.1958
17.0 (CH2). HRMS (ESI) calc. for C14H26NO4 : 272.1856 [M + H]+; found: 276.1959.
+
[M + H]+; found: 272.1857.
4-[1-(2,2,6,6-Tetramethylpiperidin-1-yloxy)-ethyl]-pirydine 1m.
Dimethyl 2-((tert-butyl(1-(diethoxyphosphoryl)-2,2-dimethyl- In an open flask, MnCl2 (84 mg, 0.67 mmol, 0.1 eq.) was
propyl)amino)oxy)malonate 2a. Yield of 25%. 1H NMR added to
stirred solution of salen ligand (180 mg,
a
(400 MHz, CDCl3): δ 5.10 (s, 1H), 4.45–4.36 (m, 1H), 4.34–4.24 0.67 mmol, 0.1 eq.) in isopropanol (10 mL). After 30 min of
(m, 1H), 4.11–4.01 (m, 1H), 3.99–3.91 (m, 1H), 3.76 (s, 6H), stirring at room temperature, a solution of TEMPO (1.56 mg,
3.27 (d, J = 25.5 Hz, 1H), 1.36–1.24 (m, 6H), 1.13 (s, 9H), 1.09 10 mmol, 1.5 eq.) and 4-vinylpyridine (700 mg, 6.66 mmol,
(s, 9H). 31P NMR (162 MHz, CDCl3) δ 24.45. 13C NMR 1 eq.) in isopropanol (10 mL) was added, then solid NaBH4
(101 MHz, CDCl3): δ 167.7 (COO), 166.7 (COO), 86.0 (CH), 69.4 (490 mg, 5.55 mmol, 1 eq.) was added in small portions. The
(CH, d, J = 138.7 Hz), 62.7 (C), 62.5 (CH2, d, J = 6.2 Hz), 59.0 mixture was stirred for 24 h at room temperature. The solvent
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Organic & Biomolecular Chemistry
Paper
was evaporated to give the crude product. After concentrating
This improved description of the polar and solvent effects
under reduced pressure, the residue was purified by column in alkoxyamine C–ON bond homolysis is important and
chromatography to afford 544 mg of 1m (32%). 1H NMR needed in order to design new alkoxyamines suitable for more
(400 MHz, CDCl3) δ 8.53 (d, J = 6.0 Hz, 2H), 7.23 (d, J = 6.0 Hz, efficient NMP in various conditions,9,16 to develop smart
2H), 4.77 (q, J = 6.7 Hz, 1H), 1.62–1.35 (m, 9H), 1.27 (s, 3H), materials,3–6 or to favour new applications in biology.8
1.16 (s, 3H), 1.04 (s, 3H), 0.69 (s, 3H). 13C NMR (75 MHz,
CDCl3) δ 153.99 (C), 149.19 (2CH), 121.12 (2CH), 81.65 (CH),
59.39 (2C), 39.98 (4CH3), 22.93 (3CH2), 16.81 (CH3). HRMS
(ESI) calc. for C16H27N2O+: 263.2118 [M + H]+; found: 263.2119.
Acknowledgements
The authors thank Aix-Marseille University and the CNRS for
financial support. ANR granted funding for this project
(ANR-14-CE16-0023-01). PN thanks the Ministere de la
Recherche du Gabon and Campus France for the Ph.D. grant.
The authors thank the RENARD network for the EPR platform.
Kinetic measurements
The homolysis rate constants kd were measured by EPR27 as
previously reported and were given by eqn (7). Air was used as
the alkyl radical scavenger for the EPR experiments. Activation
energies Ea were estimated using eqn (8) and the average
frequency factor A = 2.4 × 1014 s−1. Values of kd and Ea are
listed in Table 1.
½nitroxideꢂ1 ꢀ ½nitroxideꢂt
Notes and references
ln
¼ ꢀkd ꢁ t
ð7Þ
ð8Þ
½nitroxideꢂ1
1 G. Audran, P. Brémond and S. R. A. Marque, Chem.
Commun., 2014, 50(59), 7921–7928.
2 C. Yuan, M. Z. Rong, M. Q. Zhang, Z. P. Zhang and
Y. C. Yuan, Chem. Mater., 2011, 23, 5076–5081.
2:4 ꢃ 1014
Ea ¼ 8:314 ꢁ T ꢁ ln
kd
3 Z. P. Zhang, M. Z. Rong, M. Q. Zhang and C. Yuan, Polym.
Chem., 2013, 4, 4648–4654.
Conclusion
4 B. Schulte, M. Tsotsalas, M. Becker, A. Studer and
L. De Cola, Angew. Chem., Int. Ed., 2010, 49, 6881–6884.
5 R. K. Roy, C. X. Laure, D. Fischer-Krauser, L. Charles and
J.-F. C. O. Lutz, Chem. Commun., 2015, 51, 15677–15680.
6 R. K. Roy, A. Meszynska, C. E. Laure, L. Charles, C. Verchin
and J.-F. C. O. Lutz, Nat. Commun., 2015, 6, 1–8.
7 D. Moncelet, P. Voisin, N. Koonjoo, V. Bouchaud,
P. Massot, E. Parzy, G. Audran, J.-M. Franconi,
E. Thiaudière, S. R. A. Marque, P. Brémond and P. Mellet,
Mol. Pharmaceutics, 2014, 11, 2412–2419.
8 G. Audran, P. Brémond, J.-M. Franconi, S. R. A. Marque,
P. Massot, P. Mellet, E. Parzy and E. Thiaudière, Org.
Biomol. Chem., 2014, 12, 719–723.
9 J. Nicolas, Y. Guillaneuf, C. Lefay, D. Bertin, D. Gigmes and
B. Charleux, Prog. Polym. Sci., 2013, 38, 63–235.
The Bond Dissociation Energy (BDE) is described by the sum
of enthalpic and polar terms given in eqn (5) as proposed by
Pauling.38–40 Often, and especially concerning the C–ON bond
homolysis in alkoxyamines, it is assumed that the Ea for homo-
lysis and BDE are very similar. Consequently, Ea or log kd
should be given by a multi-parameter relationship based on
parameters describing stabilization and steric effects by a
linear model and on a parabolic model describing the polar
effect. The results described above for 1-based alkoxyamines
show that the linear relationship using a parabolic model for
the polar effect (eqn (6) and Fig. 5) is able to account for all of
the data, in sharp contrast to the simple linear multi-para-
meter model (Fig. 2 and eqn (2)). Therefore, it should also be
applied to 2-based alkoxyamines. However, it is truly needed
when very strong EWGs are on the alkyl fragment, affording 10 Registration, Evaluation, Authorisation, and Restriction of
χC values that are larger than χO values. The main drawback for
the use of a parabolic model is that the electrical Hammett
constant σI,nitroxide of the nitroxyl fragment has to be deter-
Chemical Substances (REACH), EC, 1907/2006, http://eur.
lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32006
R1907:en:NOT, accessed Sept 19, 2013.
mined. As a rule of thumb, one may assume that it should be 11 J. L. Hodgson, C.-Y. Lin, M. L. Coote, S. R. A. Marque and
used when the alkyl and nitroxyl fragments carry strong EWGs
and EDGs, respectively.
K. Matyjaszewski, Macromolecules, 2010, 43(8), 3728–3743.
12 C. Y. Lin, S. R. A. Marque, K. Matyjaszewski and
M. L. Coote, Macromolecules, 2011, 44(19), 7568–7583.
In this work, a striking 1500-fold increase in kd due to the
solvent effect is also reported. As far as we know, it is the 13 S. Marque, J. Org. Chem., 2003, 68, 7582–7590.
largest solvent effect reported for alkoxyamines at the present 14 H. Fischer, A. Kramer, S. R. A. Marque and P. Nesvadba,
time. It is likely to be due to the stabilization in the TS and the
impact of this effect depends a lot on the favoured confor- 15 D. Bertin, D. Gigmes, S. R. A. Marque and P. Tordo,
mation of the alkyl fragment. This conformation is ruled by a Macromolecules, 2005, 38, 2638–2650.
subtle interplay of the occurrence of hyperconjugation, steric 16 D. Bertin, D. Gigmes, S. R. A. Marque and P. Tordo, Chem.
Macromolecules, 2005, 38, 9974–9984.
strain, dipole moments and H-bonding.
Soc. Rev., 2011, 40, 2189.
This journal is © The Royal Society of Chemistry 2017
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
17 D. Bertin, P.-E. Dufils, I. Durand, D. Gigmes, B. Giovanetti, 30 G. Audran, P. Brémond, S. R. A. Marque and T. Yamasaki,
Y. Guillaneuf, S. R. A. Marque, T. Phan and P. Tordo,
Macromol. Chem. Phys., 2008, 209, 220–224.
18 E. Guegain, V. Delplace, T. Trimaille, D. Gigmes, D. Siri,
J. Org. Chem., 2016, 81, 1981–1988.
31 G. Audran, P. Brémond, S. R. A. Marque and G. Obame,
J. Org. Chem., 2013, 78, 7754–7757.
S. R. A. Marque, Y. Guillaneuf and J. Nicolas, Polym. Chem., 32 G. Audran, P. Brémond, S. R. A. Marque and G. Obame,
2015, 6, 5693–5704. Polym. Chem., 2012, 3, 2901–2908.
19 K. Molawi, T. Schulte, K. O. Siegenthaler, C. Wetter and 33 M. Charton, Top. Curr. Chem., 1983, 58–91.
A. Studer, Chem. – Eur. J., 2005, 11, 2335–2350.
20 P. Brémond, A. Koïta, S. R. A. Marque, V. Pesce, V. Roubaud
and D. Siri, Org. Lett., 2012, 14(1), 358–361.
21 P. Brémond and S. R. A. Marque, Chem. Commun., 2011,
47(14), 4291–4293.
34 J. J. Brocks, H. D. Beckhaus and A. Beckwith, J. Org. Chem.,
1998, 63, 1935–1943.
35 F. A. Neugebauer, Carbon Radicals with Conjugated
π-System, in Magnetic Properties of Free Radicals, ed.
H. Fischer, Landolt-Börnstein, Group II, Springer-Verlag,
Berlin, 2002, vol. 26.
22 This procedure is derived from C. Wetter, K. Jantos,
K. Woithe and A. Studer, Org. Lett., 2003, 5(16), 36 C. Hansch, A. Leo and R. W. Taft, Chem. Rev., 1991, 91,
2899. 165–195.
23 B.-L. Chen, Y.-J. Wang, H. Guo and G.-Y. Zeng, Eur. J. Med. 37 M. Charton, Prog. Phys. Org. Chem., 1981, 13, 119–251.
Chem., 2016, 109, 199–205.
24 K. Matyjaszewski, B. E. Woodworth, X. Zhang, S. G. Gaynor
and Z. Metzner, Macromolecules, 1998, 31, 5955.
38 L. Pauling, The Nature of the Chemical Bond, Cornell
University Press, Ithaca, NY, 1960.
39 B. P. Roberts, J. Chem. Soc., Perkin Trans. 2, 1996, 2719–2725.
25 A. Prechter and M. R. Heinrich, Eur. J. Org. Chem., 2013, 40 A. A. Zavitsas, Thermochemistry and Hydrogen Transfer
5585–5589.
Kinetics, in Encyclopedia of Radicals in Chemistry, Biology,
and Materials, ed. C. Chatgilialoglu and A. Studer, Wiley,
Chichester, U.K., 2012, pp. 81–106.
26 D. Bianchi, P. Cesti and E. Battistel, J. Org. Chem., 1988, 53,
5531.
27 S. Marque, C. Le Mercier, P. Tordo and H. Fischer, 41 J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005,
Macromolecules, 2000, 33(12), 4403–4410. 105, 2999–3093.
28 E. Megiel, A. Kaim and M. K. Cyrański, J. Phys. Org. Chem., 42 R. S. Rowland and R. Taylor, J. Phys. Chem., 1996, 100,
2010, 23, 1146–1154. 7384–7391.
29 A. Goto, Y. Kwak, C. Yoshikawa, Y. Tsujii, Y. Sugiura and 43 G. A. Jeffrey and W. Saenger, Hydrogen Bonding in Biological
T. Fukuda, Macromolecules, 2002, 35, 3520.
Structures, Springer, Berlin, Heidelberg, 1994.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2017