Role of Thiol-disulfide Exchange in Episulfide

Polymerization
GIONA KILCHER, LEI WANG, NICOLA TIRELLI
Laboratory of Polymers and Biomaterials, School of Pharmacy and Pharmaceutical Sciences,
University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom

Received 16 May 2007; accepted 26 November 2007

DOI: 10.1002/pola.22559
Published online in Wiley InterScience (www.interscience.wiley.com).

ABSTRACT: Episulfide polymerization offers a number of features that are uncommon

in other ring-opening anionic mechanisms. Besides the negligible sensitivity to water,
the most distinctive and novel one is likely to be the role of disulfides, which may act
both at the levels of chain transfer and end-capping, producing polymers that feature
both terminal and internal disulfides. In this article, we have qualitatively studied
the kinetics of chain transfer and measured the thiolate–disulfide exchange equilib-
rium constants. V C 2008 Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem 46: 2233–

2249, 2008
Keywords: chain transfer; episulfides; living polymerization; ring-opening polymer-
ization; synthesis

INTRODUCTION 10, allowing the polymerization (differently

Our group has extensively used the ring-opening
anionic polymerization of episulfides1,2 for pro-
ducing oxidation-sensitive materials; this fea-
ture is based on the conversion of apolar sulfur
(II) thioethers to more polar sulfur (IV) or (VI)
sulfoxides or sulfones and, due to the oxidative
character of many inflammatory reactions, can
find application in the design of carriers in
inflammation-responsive drug delivery.3,4
In episulfide polymerization, both initiating
and propagating species are thiolates; since thi-
ols have pKa values ranging from 7 to 11 in
water, deactivation due to protonation is sub-
stantially excluded in environments with pH

This article contains Supplementary Material available
via the Internet at http://www.interscience.wiley.com/jpages/
0887-624X/suppmat.
Correspondence to: N. Tirelli (E-mail: nicola.tirelli@
manchester.ac.uk)
Journal of Polymer Science: Part A: Polymer Chemistry, Vol. 46, 2233–2249 (2008)
V
C 2008 Wiley Periodicals, Inc.
from the structurally similar epoxide polymer-
ization) to preserve a living character also at
not excessively basic pH and under nonanhy-
drous conditions.
This significant advantage allows to produce
polymers under less harsh conditions and with a
broader range of chemical functions (protic
groups, esters) than it is possible with other ani-
onic mechanisms.
On the other hand, the use of thiols is often
complicated by the presence of disulfide impur-
ities, which may have a detrimental effect on
episulfide polymerization: firstly, the higher the
presence of disulfides, the lower the amount of
initiating groups and thus also the final degree
of polymerization; more importantly, disulfides
could exchange with thiolates at different stages
of the polymerization, with the effects high-
lighted in Scheme 1. It is noteworthy that the
concentration of disulfides is often very difficult
to predict, since the kinetics of thiol oxidation
can be dramatically altered by environmental
2233
2234 KILCHER, WANG, AND TIRELLI

Scheme 1. In episulfide polymerization thiolates are both initiating and propagat-

ing species. In an ideal case (horizontal sequence of events), the initiator (R0 ) and
end-capping agent (R00 ) are present at the two termini of the final polymer chain
and, in principle, are different in nature. If the exchange with a disulfide present in
the reaction environment is faster than propagation, this may lead to a substantive
replacement of initiating species, with the introduction of R residues as terminal
groups. If the exchange is as quick as or slower than propagation, two effects will be
recorded: a decrease in degree of polymerization, due to thiolate transfer and reinitia-
tion, and a mix-up in terminal groups, with the contemporary presence of polymers
asymmetrically terminated with R and R0 , or with R0 and R00 . It is noteworthy that
the macromolecular disulfides may be part of these equilibria (R may be a polysulfide
chain).

conditions (a number of catalysts, e.g., traces of
metal ions, can speed it up).
In previous occasions, we have overcome
these problems by generating thiolates in situ
through methanolysis of thioacetates1,4; these
derivatives, however, need to be ad hoc synthe-
sized, and the process can be expensive and
lengthy. On the contrary, it would be often bene-
ficial, for example, for scale-up purposes, to
directly employ commercially available thiols,
with their non-negligible amounts of disulfides.
In this study we have therefore studied the
effects of disulfides on episulfide polymerization,
specifically focusing on thiol-disulfide exchange
(Scheme 1 and explanation in the caption).
The Thiol-disulfide Exchange
This is a classical exchange of REDOX status
between two disulfide þ thiol couples and is a
phenomenon of the outmost importance in bio-
chemistry, governing the oxidation state of thiols
and thus also protein folding and activity.5,6 It
Journal of Polymer Science: Part A: Polymer Chemistry
has therefore been studied mostly in water
environment.
The outcome of this reaction depends on the con-
centrations of the reaction partners and on their
REDOX potentials. On their turn, both factors
depend also on the pKas of the involved thiols: as it
has been shown since the ’50s,7 not only the thiol-
disulfide exchange is a pH- dependent reaction,
but deprotonated thiols (and not thiols) are the
active species in the equilibrium, which can there-
fore be described as a thiolate–disulfide exchange.
As elegantly demonstrated by Whitesides for the
exchange of (di)thiols with the Ellmans’ reagent,8
the nucleophilicity of thiols is linearly related
to their basicity (inversely proportional to the
amount of thiolates at a given pH); therefore the
DpKa of the involved thiols is a primary factor for
determining the equilibrium constant of the
exchange (and the kinetics too).
In the most general case, when a homodisulfide
is exposed to a thiolate one must consider two equi-
libria, the formation of a heterodisulfide (mixed or-
ganic residues) (1) and, from there, that of a homo-
disulfide (identical organic residues) (2).
DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION
A few important remarks:
½R0 S
1. In both equilibria, the term represents
½R S
in identical manner the dependence on the
pKa difference (see Supplementary Infor-
mation, section 1), indicating the preferen-
tial formation of the salt of the most acid
thiol; this term pushes both equilibria to
left or right, hence, it will always favor the
formation of homodisulfides (RSSR or
R0 SSR0 , when respectively, R0 SH or RSH is
the most acid thiol).
2. It is also easy to infer that an increasing
concentration of RS will always favor the
formation of its homodisulfide RSSR (and
the same applies to the other thiol).
3. Finally, it is well-known that entropic fac-
tors, such as the possibility of intramolecu-
lar cyclization of the disulfide, as in the
case of dithiothreitol (DTT) and many
other dithiols,9,10 do also favor the forma-
tion of RSSR structures.
An open question is therefore whether there
are conditions that would favor the formation of
heterodisulfides over that of homodisulfides
ðK1exc > 1 > K2exc Þ.
Relevance to Episulfide Polymerization
The possible exchanges of a thiolate-terminated
polymer with a low MW disulfide would produce
a new macromolecule with the same mass in the
case of heterodisulfide formation, or would dou-
ble it in the case of dimerization to homodisul-
fide (the same happens in free radical polymer-
ization with termination by disproportionation
or recombination, respectively). In addition, the
first kind of macromolecule would present the
(hetero)disulfide in terminal position, which be
easier to manipulate, for example, to reduce; the
second kind of polymer would feature a central,
more hindered (homo)disulfide.
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
2235
EXPERIMENTAL
Materials
All materials were used as received from the
supplier (Aldrich, Gillingham, United Kingdom,
for benzyl mercaptan, propylene sulfide, 1,8-di-
azabicyclo [5.4.0]undec-7-ene, iodoacetamide and
tributyl phosphine; Fluka, Gillingham, United
Kingdom, for diallyl disulfide). THF and metha-
nol were degassed by bubbling argon under
inert atmosphere for 1 h before use.
Abbreviations. Benzyl mercaptan: BM, pro-
pylene sulfide: PS, diallyl disulfide: DDS, 1,8-
diazabicyclo [5.4.0]undec-7-ene: DBU, iodoacet-
amide: IA, tributylphosphine: TBP.
Polymer Synthesis
Three main kinds of experiments were run in
this study; typically polymerization experiments
were run in parallel in a FirstMate parallel re-
actor (Argonaut Technologies) at room tempera-
ture under argon atmosphere to yield families of
polymers with identical molecular weight but
different ratios between thiolates and diallyl
disulfide.
Polymerization of Propylene Sulfide
BM and PS were dissolved under inert atmo-
sphere in degassed THF to prepare a stock solu-
tion with a fixed ratio between monomer and
initiator (e.g., containing 10 mg/mL of BM and
120 mg/mL of PS for a 1:20 thiol-to-monomer ra-
tio). Five milliliters of stock solution were then
introduced under inert argon atmosphere in the
reactor. The polymerization was then started by
adding via a syringe 200 lL of a 300 mg/mL
DBU solution in degassed THF (for a 1:1 DBU/
thiol ratio). The reagents were allowed to react
for 45 min, 0.5 mL of a THF stock solution (300
mg/mL) of IA was then added to provide a twice
2236 KILCHER, WANG, AND TIRELLI
as many IA as thiolates. The mixture was
allowed to react for further 60 min, and then fil-
tered over glass wool completely evaporated at
the rotary evaporator. The crude product was
then dissolved in 1 mL CH2Cl2 and added drop-
wise to 10 mL MeOH under vigorous stirring.
The white dispersion was centrifuged for 10 min
at 5 8C and 4500 rpm allowing a clear phase
separation. The transparent methanol phase
was removed and the step was repeated. The
polymer was dried for 24 h under reduced pres-
sure (20 mbar, 30 8C), and a viscous colorless oil
was finally collected. Yield: 80–85%.
1
H-NMR (CDCl3): d ¼ 1.35–1.37 (CH3 in PPS
chain), 2.51–2.61 (m, 1 diastereotopic H of CH2
in PPS chain), 2.75–2.93 (m, CH and 1 diaster-
eotopic H of CH2 in PPS chain), 3.17–3.25 (m,
2H, SCH2CONH2), 3.68 (s, 2H, Ph
CH2S), 5.50 (b, 1H, NH2), 6.71 (b, 1H,
NH2), 7.25–7.26 (d, 5 H, PhCH2 ) ppm.
ATR-IR (thin film): 3330–3180 (m NH), 2958
(mas CH3), 2919 (mas CH2), 2867 (ms CH3), 1683 (m
C¼¼O, amide I), 1580 (d NH, amide II), 1495 (m
CC ring), 1450 and 1372 (das and ds CH3 bend-
ing), 1173 (m CSC) cm1.
Polymerization of Propylene Sulfide in the
Presence of Other Compounds
These polymerization experiments were con-
ducted identically to the experiments described
in 1, with the only difference that DDS or TBP
were added to the monomer before initiating the
polymerization. BM and PS were dissolved
under inert atmosphere in degassed THF to pre-
pare a stock solution with a fixed ratio between
monomer and initiator (containing 10 mg/mL of
BM and 120 mg/mL of PS for a 1:20 thiol-to-
monomer ratio). 5 mL of stock solution were
then introduced under inert argon atmosphere
in a reactor.
To each reactor was added a variable quantity
of a 370 mg/mL stock solution of DDS in THF
(for obtaining DDS/thiol ratios of 0.04/1, 0.2/1, 1/1
and 5/1) or of a 200 mg/mL stock solution of TBP
in THF (for obtaining TBP/thiol ratios of 0.5/1
and 3/1) and the polymerization was then started
by adding via a syringe 200 lL of a DBU solution
in degassed THF (containing 300 mg/mL for a 1:1
DBU/thiol ratio). The polymerization time and
work-up procedure were identical to the previous
experiments. Recovery yields ranged between 2
and 55% (calculated as weight of recovered poly-
mer/weight of monomers þ initiators).
1
H NMR (CDCl3): d ¼ 1.35–1.37 (CH3 in PPS
chain), 2.51–2.61 (m, 1 diastereotopic H of CH2
in PPS chain), 2.75–2.93 (m, CH and 1 diaster-
eotopic H of CH2 in PPS chain), 3.27–3.29 (d, 2H,
SCH2CH¼ ¼CH2), 3.68 (s, 2H, PhCH2S),
5.08–5.12 (t, SCH2CH¼ ¼CH2), 5.74–5.82 (m,
1H, SCH2CH¼ ¼CH2), 7.25–7.26 (d, 5 H,
PhCH2) ppm.
ATR-IR (thin film): 3330–3180 (m NH), 2960
(mas CH3), 2920 (mas CH2), 2865 (ms CH3), 1685 (m
C¼¼O, amide I), 1580 (d NH, amide II), 1495 (m
CC ring), 1450 and 1373 (das and ds CH3 bend-
ing), 1172 (m CSC) cm1.
Polymerization of Propylene Sulfide Followed
by Thiolate–disulfide Exchange
These polymerization experiments were con-
ducted identically to the experiments described
in 1, with the only difference that DDS was
added at the end of the polymerization. The po-
lymerization was prepared as described in 1 and
started with the addition of 200 lL of a 300 mg/
mL DBU solution in degassed THF (1:1 DBU/
thiol ratio). The reagents were allowed to react
for 45 min, then a different quantity of a 370
mg/mL stock solution of DDS was added to each
reactor (for DDS/thiol ratios of 0.3/1, 1/1, and 1/
2). The mixture was stirred for 15 min; 0.5 mL
of a THF stock solution (300 mg/mL) of IA was
then added to provide twice as many IA as thio-
lates. The mixture was allowed to react for fur-
ther 60 min, and then filtered over glass wool
completely evaporated at the rotary evaporator.
The crude product was then dissolved in 1 mL
CH2Cl2 and added dropwise to 10 mL MeOH
under vigorous stirring. The white dispersion
was centrifuged for 10 min at 5 8C and 4500
rpm allowing a clear phase separation. The
transparent methanol phase was removed and
the step was repeated. The polymer was dried
for 24 h under reduced pressure (20 mbar,
30 8C), and a viscous colorless oil was finally col-
lected. Recovery yields ranged between 77 and
86% (calculated as weight of recovered polymer/
weight of monomers þ initiators).
1
H NMR (CDCl3): d ¼ 1.35–1.37 (CH3 in PPS
chain), 2.51–2.61 (m, 1 diastereotopic H of CH2
in PPS chain), 2.75–2.93 (m, CH, and 1 diaster-
eotopic H of CH2 in PPS chain), 3.17–3.25 (m,
2H, SCH2CONH2), 3.27–3.29 (d, 2H,
SCH2CH¼ ¼CH2), 3.68 (s, 2H, PhCH2S),
5.08–5.12 (m, 2H, SCH2CH¼ ¼CH2), 5.50 (b,
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION
1H, NH2), 5.74–5.82 (m, 1H, SCH2CH
¼¼CH2), 6.71 (b, 1H, NH2), 7.25–7.26 (d, 5 H,
Ph CH2) ppm.
ATR-IR (thin film): 3330–3180 (m NH), 2958
(mas CH3), 2920 (mas CH2), 2867 (ms CH3), 1682 (m
C¼¼O, amide I), 1580 (d NH, amide II), 1495 (m
CC ring), 1450 and 1372 (das and ds CH3 bend-
ing), 1171 (m CSC) cm1.
Reduction/End-Capping Experiments
Hundred milligrams of poly(propylene disulfide)
(corresponding to 0.06 mmol of PPS block for a
theoretical MW of 1650 g/mol) were introduced
into the reactor under dry argon atmosphere,
followed by 5 mL of degassed THF. Upon com-
plete dissolution, the required amount of a 100
mg/mL stock solution of tributyl phosphine
(TBP) was introduced into the reactor via a
microliter syringe for a molar ratio TBP/S-S of 4
or 16 (the number of disulfides was calculated
from the sum P1 þ P3). A double molar excess
of H2O to TBP was also added in some control
experiments. The reduction reaction was then
allowed for 4 h; two procedures were then fol-
lowed: (a) a stock solution of DBU (300 mg/mL)
and EBA (100 mg/mL) were subsequently intro-
duced into the reactor for a molar ratio DBU/S-
S and EBA/S-S: 4:1; (b) the solvent was com-
pletely evaporated at a rotary evaporator (still
using the same vessel) and under a flow of inert
gas the crude was extracted twice under stirring
with 1 mL of degassed methanol; the polymer
was finally dissolved in 5-mL degassed THF and
procedure (a) was then followed.
The end-capping agent was allowed under
stirring for 60 min, following then a work-up
procedure identical to the previous experiments.
1
H NMR (CDCl3): d ¼ 1.26–1.30 (t, CH3,
CH3CH2COO ), 1.35–1.37 (CH3 in PPS chain),
2.51–2.61 (m, 1 diastereotopic H of CH2 in PPS
chain), 2.75–2.93 (m, CH and 1 diastereotopic H
of CH2 in PPS chain), 3.17–3.25 (m, 2H,
S CH2CONH2), 3.20–3.37 (m, 2H,  S
CH2COO), 3.68 (s, 2H, PhCH2 S),
4.15–4.20 (q, 2H, CH3CH2COO ), 5.50 (b, 1H,
NH2), 6.71 (b, 1H, NH2), 7.25–7.26 (d, 5 H,
Ph CH2) ppm.
ATR-IR (thin film): 3330–3180 (m NH), 2958
(mas CH3), 2919 (mas CH2), 2867 (ms CH3), 1735 (m
C¼¼O, ester), 1683 (m C¼ ¼O, amide I), 1580 (d
NH, amide II), 1495 (m CC ring), 1450 and 1372
(das and ds CH3 bending), 1173 (m CSC) cm1.
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
2237
Physicochemical Characterization
1
H NMR spectra were recorded with a Bruker
500 MHz spectrometer on CDCl3 solutions. ATR-
IR spectra were recorded on polymer films
obtained by evaporation of dichloromethane
polymer solutions directly on the diamond of a
Bruker Tensor 27 equipped with a temperature-
controlled Golden Gate ATR accessory. GPC
analysis was performed in THF solution on a
Polymer Laboratories GPC50 equipped with re-
fractive index and viscosimetric detectors, using
monodisperse polystyrene standards.
RESULTS AND DISCUSSION
Do Disulfides Undergo Chain Transfer During
Episulfide Polymerization?
In this study we have always employed benzyl
mercaptan as an initiator, thus producing propa-
gating species that are terminated at one end
with an aromatic group and at the other with a
reactive thiolate.
The possible equilibria arising from the pres-
ence of disulfides, either as impurities of the
thiol or because produced or introduced in situ
during initiation or propagation, are summar-
ized in a simplified form in Scheme 2: the key
point is that disulfides may chain transfer while
also producing polymers with terminal (KexcA ) or
internal disulfides (Kexc
B ). It is noteworthy that
the last ones must be characterized by larger
molecular weights.
Hypotheses and Assumptions
For a qualitative description of the system, it is
possible to
 estimate kprop  5  10  103 min1 mol1 l,
by extrapolating literature values to the po-
lymerization temperature, 20 8C (monomer
¼ PS, solvent ¼ THF, counterion ¼ tetralkyl
ammonium, T ¼ 0 8C and 20 8C11, extra-
polation supposing Arrhenius dependence).
ƒƒ! ƒƒ!
 suppose, in a general case, that kexcA > kB ,
exc
due to the higher steric hindrance carried
by the two partners (both polymeric in na-
ture) of the second reaction.
 distinguish two limiting cases:
(a) If the propagation rate is considerably
the transfer rate (kprop 
larger than ƒƒ!
½monomer > kexc
A ½dibenzyldisulfide,
ƒƒ! ƒƒ! ei-
ther because kprop >> kexc A ð> k exc
B Þ or
because ½monomer >> [dibenzyldisul-
2238 KILCHER, WANG, AND TIRELLI

Scheme 2. During the propagation stage (rate constant kprop), the growing chain
may be involved in two possible thiolate–disulfide exchange reactions: the production
of a terminal heterodisulfide
ƒƒ! that
ƒƒ! of an internal homodisulfide, with, respectively, for-
mation rate constant kexc exc exc exc
A and kB and equilibrium constants kA and kB . In a more
complete form all these constants should depend on the size of the polymer chain
(the degree of polymerization), since steric hindrance should play a major role, above
all in the second reaction; for seek of simplicity, however, we will ignore this depend-
ency. Please note that we omit the initiation step in our description, again for seek of
simplicity. Finally, we will also ignore any disulfide–disulfide equilibria. Chains ter-
minated with thiolates were finally end-capped with iodoacetamide, to avoid any
postpolymerization formation of disulfides.

fide]), the thiolate–disulfide exchanges
will mostly take place at the end of the po-
lymerization. If an end-capping agent
(e.g., iodoacetamide) quenches the thio-
late-terminated polymer, we therefore
expect to see a clear bimodal molecular
weight distribution: the end-capped poly-
mers (A) and those with terminal disul-
fides (B) would be present in a first peak,
with an average degree of polymerization
equal to the monomer/initiator ratio (as
in a living polymerization without chain
transfer); the polymers with internal
disulfides (C) will, on the other hand, be
present in a second peak with double mo-
lecular weight.

Table 1. Polymerization Experiments in the Presence or Absence of Tributyl Phosphine

GPC-Peak 1d GPC-Peak 2d

Samplea RSHb R0 SSR0 b DPc Mn Mw Mw =Mn Mn Mw M w=Mn Apeak2=Apeak1

Pblank 0.4 0 19.5 6 0.7 1450 1550 1.07 3100 3450 1.11 >0.05
e
PRED2:1
blank 0.4 0 20 1500 1650 1.10 3100 3500 1.13 0.11
PRED1:2
blank
e
0.4 0 20.5 1500 1710 1.14 ¼ ¼ ¼ n.a.
a
Monomer/thiol ratio ¼ 20. Polymerization yields (weight ratio recovered material/monomer) are always in the range 80–85%.
b
mmoles of benzyl mercaptan and diallyl disulfide.
c
Calculated as the ratio between the integral value of a PPS methyl proton (1.30 ppm) and that of a benzyl mercaptan meth-
ylene proton (3.67 ppm).
d
The GPC curves were fitted with two Gaussian peaks, which were then separately analyzed.
e
Polymerizations performed in the presence of 0.2 mmol and 0.8 mmol of tributyl phosphine, respectively for PRED2:1
blank and for
PRED1:2
blank (in both cases in large excess compared to the amount of disulfides).

Journal of Polymer Science: Part A: Polymer Chemistry

DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION 2239

a clear bimodal distribution, with peaks corre-

Figure 1. GPC trace for polymer Pblank (see Table 1
for polymer composition).
ƒƒ!
(b) If kprop  ½monomer  kexc
A [dibenzyldi-
sulfide], chain transfer will take place
during the polymerization, correspond-
ing to a broader but possibly mono-
modal molecular weight distribution,
which would be peaked at a much lower
value than in case a.
First Evidences
Poly(propylene sulfide) (Pblank in Table 1) initi-
ated with nonpurified (therefore likely to con-
tain disulfides) benzyl mercaptan indeed shows
sponding to the theoretical DP and to its double
(Fig. 1). The area of the second peak is, however,
not more than 5% of that of the first one.
Note: In Table 1, we report the ratio between
peak areas (from the integration of the weight
fraction versus molecular weight graphs). The
peak at lower molecular weight (the ‘‘mono-
meric’’ peak) is, however, considerably asymmet-
ric, with a long tail at high molar masses and
this likely causes an overestimation of the area
of the other (‘‘dimeric’’) peak.
Furthermore, the peak at higher molar
masses can be quantitatively removed by con-
ducting the polymerization in the presence of
appropriate quantities of a non thiol-based
reducing agent (tributyl phosphine, PRED2:1 and
blank
PRED1:2
blank in Table 1): it ought therefore to be
related to polymers where disulfides play a
major role in determining their molar mass,
that is, polymers with internal disulfides. These
two evidences thus seem to suggest propagation
rate to be considerably higher than the
exchange rates (Case a) for the benzyl mercap-
tan-initiated polymerization.
There are, however, issues of quantification.
As previously mentioned, this effect can be
due either to a high monomer/disulfide ratio
throughout most of the polymerization (disul-
fides are impurities of the thiols, but likely not
the major constituents of the mixture), or to a
propagation rate constant much larger than the

Table 2. Polymerization Experiments Conducted in the Presence of Disulfides

NMR GPCf

Samplea RSHb R0 SSR0 b DPc ( Mn ) P1d P2e Mn Mw MwM n Yield (%)

Ptransf
blank 0.2 0 30 (2220) 0 0.97 2370 2630 1.11 77
Ptransf
1 0.2 0.008 29.5 (2180) 0.15 0.88 2170 2950 1.36 55
Ptransf
2 0.2 0.04 25.4 (1900) 0.49 0.87 1880 2920 1.55 50
Ptransf
3 0.2 0.2 16.4 (1220) 1.93 0.62 1750 2860 1.63 42
Ptransf
4 0.2 1 10.7 (800) 5.4 0.29 1200 1500 1.25 2
a
Monomer/thiol ratio ¼ 20. Polymerization yields (weight ratio recovered material/monomer) are always in the range 80–
85%.
b
mmoles of benzyl mercaptan and allyl disulfide in the feed.
c
Calculated as the ratio between the integral value of a PPS methyl proton (1.30 ppm) and (a) that of a benzyl mercaptan
methylene proton (3.67 ppm) for Ptransf
blank , or (b) the half of the sum over all terminal groups (double bonds, aromatic rings, amides)
of the integral of a proton for the remaining polymers.
d
Ratio between terminal groups in form of double bonds (the integral value of a ethenyl proton, 5.72–5.83 ppm) and those in
0
þ½B0 =2þ½C0 =2
form of aromatic groups (the integral value of a benzyl mercaptan methylene proton, 3.67 ppm). P1 ¼ ½Eþ½Fþ½A
½Aþ½B=2þ½C=2þ½Eþ½F .
e
Ratio between terminal groups in form of amides (the integral value of an amide a proton, 3.15–3.24 ppm) and those in
½Aþ½A0 
form of aromatic rings (the integral value of a benzyl mercaptan methylene proton, 3.67 ppm). P2 ¼ ½Aþ½B=2þ½C=2þ½Eþ½F .
f
The average molecular weights were calculated on the whole range of MWs, without separating the ‘‘monomer’’ and ‘‘dimer’’
curves.

Journal of Polymer Science: Part A: Polymer Chemistry

DOI 10.1002/pola
2240 KILCHER, WANG, AND TIRELLI

Scheme 3. In addition to the reactions presented in Scheme 2, a number of other

equilibria arise following the use of diallyl disulfide. A key point is that all of them
determine reinitiation of polymer chains that may not feature any longer aromatic
rings as terminal groups. Chains terminated with thiolates were finally end-capped
with iodoacetamide, to avoid any postpolymerization formation of disulfides. For
clarity of presentation we have omitted both in the graph and in the text any refer-
ence to the equilibrium between allyl thiolate and B, which would regenerate benzyl
thiolate, and to those between benzyl thiolate and the polymeric disulfides containing
allyl groups; at the qualitative level of description that we are using, they would not
change the overall picture. Furthermore, their values can be easily calculated once
those of Kexc exc exc
A and K1 to K5 are determined.

exchange rate constant (or both). It is possible
to discriminate between these two effects by
varying the monomer/disulfide ratio.
Addition of Disulfides
We have therefore run polymerization experi-
ments adding increasing quantities of disulfides
(Table 2). More specifically, we have employed
diallyl disulfide, a disulfide chemically different
from the initiator (double bonds versus aromatic
rings), which makes possible to distinguish
between terminal groups present as initiators
and those introduced during chain transfer.
This approach apparently increases the com-
plexity of the system: a number of new thiol-di-
sulfide equilibria are added (Scheme 3) without
removing the old ones, since the disulfide impur-
ities of benzyl mercaptan are difficult to elimi-
nate; indeed it is impossible to reduce them in
situ, because this would reduce diallyl disulfide
too, while the ex situ reduction would require
Journal of Polymer Science: Part A: Polymer Chemistry
isolation procedures that can easily lead to par-
tial reoxidation.
At a more careful observation, however, it is
apparent that the system has not qualitatively
changed.
The species involved in similar equilibria
have always similar steric hindrance and pKa:
for example, both benzyl and allyl mercaptan
have markedly lower pKa than the secondary
thiol of the growing chain (respectively, 9.43 and
9.9612 versus more than 11). Since, as men-
tioned in the introduction, the major factor
influencing these equilibrium constants is the
DpKa between thiols, there should be little dif-
ference whether benzyl or allyl mercaptan is
involved; in addition, the reactivity of the propa-
gating species should be only marginally influ-
enced by the terminus at the other end of the
polymer. It is therefore reasonable to assume
KAexc  K1exc  K5exc . and KBexc  K2exc  K3exc  K4exc
and this is likely to apply also to the correspond-
ing kinetic constants. In particular, kprop ¼ kprop 2 :
DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION 2241

column P1 with twice the molar amount of

disulfide/thiol), suggesting that double
bonds are quantitatively incorporated in
our samples. If allyl mercaptan was pre-
dominantly produced at the end of the po-
lymerization, the scarcity of monomer
would not allow it to produce anything
except oligomers, which would be lost dur-
ing the work-up, loosing therefore at least
50% of the double bonds. The 1:1 olefin/
phenyl ratio in the polymers seems, on the
other hand, to be a strong indication that
the exchange takes place throughout the
polymerization.
Figure 2. Incorporation of double bonds in the poly- 2. By increasing the amount of disulfides, the
mer chains (black squares) and number average mo- average degree of polymerization drasti-
lecular weight (from NMR terminal group analysis
cally decreases, while the number of chains
and from GPC) as a function of the double bonds (in
diallyl disulfide-chain transfer agent)/phenyl groups
(in benzyl mercaptan–initiator) ratio in the polymer-
ization mixture; lines are only guides for eyes. Assum-
ing that the samples after methanol extraction are
representative of the whole polymerization product,
the incorporation of diallyl disulfide in the polymers
is quantitative. This assumption is clearly unaccept-
able for the last sample (Ptransf
4 , recovered only in 2%
yield), which is therefore added to the graph only for
completeness. The discrepancy between NMR- and
GPC-derived M n data for Ptransf
3 and Ptransf
4 is due to
the fact that the bimodal GPC distributions were here
treated as a single peak, with negligible consequences
for the samples with a small fraction of internal
disulfides.

a difference may exist in the initiation step,

then the propagating species is identical.
Finally, we still expect to have the above two
limiting cases, depending on the relative ratio
between propagation and chain transfer; in case
of bimodal GPC traces, A, A0 , B, B0 , and E
(chains with end-capping or terminal disulfides)
would be in the lower molecular weight peak,
while C, C0 , and F (chains with internal disul-
fides) the one at higher mass.
Two main results can be gathered from these
experiments:

1. Since the polymers are purified by precipi-

tation, these samples are composed of poly-
mers with a MW > 1000 g/mol (at lower
molar mass the materials are completely
soluble in methanol). In the first three
samples the double bond/benzyl group ratio
in the polymers is roughly the same as in
the feed (Fig. 2 and Table 2, compare the
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
Figure 3. GPC traces of polysulfides obtained using
benzyl mercaptan as an initiator, 20 equivalents of
propylene sulfide as a monomer and variable quanti-
ties (reported in relation to q equivalent of benzyl
mercaptan) of diallyl disulfide as a chain transfer
agent.
2242 KILCHER, WANG, AND TIRELLI

Scheme 4. The addition of a disulfide at the end of the polymerization would deter-
mine only two thiolate–disulfide equilibria, with no chain transfer. We have omitted
any reference to the equilibrium between G and B (terminal disulfide polymer gener-
ated by dibenzyl disulfide), as well as any link of the concentration of C to the pres-
ence of dibenzyl disulfide. This is a legitimate assumption on the basis that in our
experiments the concentration of dibenzyl mercaptan is at least one order of magni-
tude lower than that of diallyl disulfide, which should therefore overwhelm any other
effect.

with internal disulfides (the higher molec-
ular weight peak) seems to increase (Fig.
3). In this case the ratios between the
areas of the two peaks are little informa-
tive: due to the difficult recovery of the low
polymers (fairly high solubility in metha-
nol), some of the samples may be artifi-
cially enriched in the longest chains, that
is, those with internal disulfides; this is
specifically true in the case of Ktransf 4
where we presume the sample (recovered
in very low yields) to be composed only by
chains with internal disulfides.
These findings indicate the transfer efficiency
to be sharply dependent on the amount of added
disulfides, with very clear results already with a
5:1 thiol/disulfide molar ratio (Ptransf ). It is
2
therefore logical to conclude that the transfer
rate constants are at least of the same order of
magnitude as the propagation rate constants or,
maybe more likely, higher than those.
Partial Conclusions
(a) disulfides have a strong chain transfer effect
in propylene sulfide polymerization, (b) it is
likely that thiolates and disulfides are always
at or close to the equilibrium throughout the
polymerization.
As a logical sequel, on the other hand, the
system will not be sufficiently described as long
as the equilibrium constants are not known.
Note: Since in Ptransf
blank (no added disulfide) the
presence of dibenzyl disulfide impurities has
considerably less influence on the molecular
weight distribution than the addition of a 4%
diallyl disulfide (Ptransf
1 ), we can estimate a pos-
teriori that much less than 4% of benzyl mercap-
, tan is present in the form of disulfide.
Thiolate–disulfide Equilibrium Constants
If the polymerization is conducted in the pres-
ence of only small amounts of a disulfide (as in
the case of the impurities of benzyl mercaptan,
see the note at the end of the previous section),
in a first approximation we can neglect its influ-
ence if a second disulfide is added in much more
conspicuous quantities at the end of propagation
(see caption to Scheme 4).
Doing so we have removed any analytical
complication due to chain transfer, thus allowing
to study what would have happened to a grow-
ing chain if propagation was ‘‘quenched.’’ We
have also drastically reduced the number of
equilibria to the basic two: one leading to a
monomeric polymer with a terminal disulfide
(eq. constant Kexc
A ), the other one leading to the
dimeric polymer with an internal disulfide (eq.
constant Kexc
B ).
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION 2243

Table 3. Chain-End Transfer Experiments: Effect of Order of Addition and Time of Reaction

GPC-Peak 1d GPC-Peak 2d

Samplea RSHb R0 SSR0 b DPc Mn Mw M w=Mn Mn Mw M w=Mn Apeak2=Apeak1

Pblank 0.4 0 19.5 6 0.7 1450 1550 1.07 3100 3450 1.11 0.05
P1:2 150 0.4 0.2 20.3 6 1.5 1550 1730 1.11 3150 3580 1.14 0.21
P1:2 600 0.4 0.2 18.0 6 1.2 1500 1680 1.12 3100 3560 1.15 0.20
P1:2 1200 0.4 0.2 19.16 1.3 1500 1660 1.11 3150 3620 1.15 0.21
P1:2 after 0.4 0.2 18.9 6 0.8 1500 1640 1.09 3100 3500 1.13 0.05
a
All data are averages over three samples; where significant, the standard deviation is provided too. Polymerization yields
(weight ratio recovered material/monomer) are always in the range 78–85%.
b
mmoles of benzyl mercaptan and diallyl disulfide.
c
Calculated as the ratio between the integral value of a PPS methyl proton (1.30 ppm) and that of a benzyl mercaptan
methylene proton (3.67 ppm).
d
The GPC curves were fitted with two Gaussian peaks, which were then separately analyzed.

Firstly, we have prepared a series of samples
to optimize the time and conditions of reaction
(Table 3). A constant amount of diallyl disulfide
was left equilibrating for different times prior to
end-capping: no difference in molecular weight
(both from NMR and GPC analysis) and compo-
sition was apparent between samples kept 15,
60, and 120 min (P1:2_150 , P1:2_60, P1:2_1200 , respec-
tively). On the other hand, no double bond was
incorporated if diallyl disulfide was added imme-
diately after the end-capping agent, confirming
the effectiveness of iodoacetamide as a thiolate
quencher. It is also apparent that the molecular
weight distribution of this last sample was iden-
tical to that of a blank (end-capping without any
addition of diallyl disulfide), while the other
three samples exhibit a marked change in the
ratio between dimeric and monomeric peak.
Using the shortest equilibration time (15
min), we have then synthesized a series of poly-
mers by varying the thiolate/disulfide ratio
(Table 4).
In all these samples the degree of polymer-
ization from NMR (number of monomeric unit
per initiator) remains constant, as well as M n
and M w (for both peaks), calculated from GPC
data; this was indeed expected, since the only
effects of diallyl disulfide addition are suppos-
edly the introduction of different end groups
and the dimerization, with no influence on the
kinetic chain length. On the other hand, the
amount of added disulfide shows a profound
effect on the ratio between the two peaks, with
an apparently counter-intuitive decrease in rela-
tive intensity with increasing diallyl disulfide
concentration (Fig. 4, samples P1:2_150 , P1:1_150
and P2:1_150 ). In reality, as explained in the intro-
duction (section thiolate–disulfide exchange), it
is by increasing the thiolate concentration, and
not that of diallyl disulfide, that internal di-

Table 4. Chain-End Transfer Experiments: Effect of the Thiolate/Disulfide Ratio

GPC-Peak 1d GPC-Peak 2d

Samplea RSHb R0 SSR0 b DPc Mn Mw M w=M n Mn Mw Mw=M n Apeak2=Apeak1 e

Pblank 0.4 0 19.5 6 0.7 1450 1550 1.07 3100 3450 1.11 >0.05
P1:2 150 0.4 0.2 20.3 6 1.5 1550 1730 1.12 3150 3580 1.14 0.23 (0.28)
P1:1 150 0.4 0.4 19.3 6 0.6 1480 1640 1.11 3080 3580 1.16 0.20 (0.25)
P2:1 150 0.4 0.6 19.7 6 0.6 1480 1640 1.11 3180 3650 1.15 0.18 (0.19)
a
All data are averages over three samples, where significant, the standard deviation is provided too. Polymerization yields
(weight ratio recovered material/monomer) are always in the range 77–86%.
b
mmoles of benzyl mercaptan and diallyl disulfide.
c
Calculated as the ratio between the integral value of a PPS methyl proton (1.30 ppm) and that of a benzyl mercaptan
methylene proton (3.67 ppm).
d
The GPC curves were fitted with two Gaussian peaks, which were then separately analyzed.
e
In brackets the values obtained from NMR data (calculated as 2*[C]/([A] þ [E]).

Journal of Polymer Science: Part A: Polymer Chemistry

DOI 10.1002/pola
2244 KILCHER, WANG, AND TIRELLI

sulfide is necessarily increased. Also the ratio

Figure 4. GPC traces of four polysulfides differing
only in the amount of disulfide added at the end of
the polymerization (see Table 4 for polymer composi-
tions).
of the two peaks for the Pblank sample must
not be surprising, since it can be shown that
dependence of this ratio is strongly nonlinear on
the disulfide concentration.
As clearly visible in Figure 5, the resonances
of the three different terminal groups featured
in these polymers are easily recognizable in 1H
NMR spectra; this allows a quantitative evalua-
tion of the terminal groups in the recovered
polymer mixture (A@, E, and C). The concentra-
tions of the five species participating the two
equilibria ([A], [D], [E], [G], and [C]) can then
be easily gathered combining the molar fractions
of differently end-capped polysulfide chains P1 ¼
Xdoublebonds and P2 ¼ Xamides (from end group
NMR analysis) and the equations for the conser-
vation of the moles of disulfides, thiolates and
double bonds in the equilibria (see Supplemen-
tary Information, section 2).

Figure 5. 1H-NMR spectra (sections showing the terminal groups) for polymers
prepared adding different quantities of diallyl disulfide after polymerization and
before end-capping with iodoacetamide (see Table 2 for polymer compositions). With
increasing disulfide concentrations (from bottom up), the final polymer displays less
amides and more double bonds. The amount of initiators (resonance of the benzylic
CH2 at 3.65–3.70), on the other hand remains constant.
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION 2245

0.4760.17

0.5060.09
0.5760.26

Percentage of benzyl mercaptan-initiated PPS blocks terminated with allyl groups, calculated as the ratio between the integral value of a ethenyl proton (5.72–5.83

Percentage of benzyl mercaptan-initiated PPS blocks terminated with amide groups, calculated as the ratio between the integral value of a amide a proton (3.15–3.24
KBexc e

0.41
0.34
Table 5. NMR-Derived Equilibrium Concentrations of the Various Thiolate and Disulfide Species and Corresponding Equilibrium Constants for All

6.5 62.9

3.0 60.4
3.5 60.9
KAexc e

6.7
5.3
0.047 60.009

0.041 60.004
0.033 60.010
0.044
0.040
C

0.22 60.009

0.282 60.008
0.322 60.012

Figure 6. Equilibrium constants for the chain ter-

0.212
Equilibrium concentrations (mmol)

0.22

minus thiolate–disulfide exchange. As expected for

G

equilibrium constants, their values are negligibly de-

pendent on the disulfide/thiolate ratio.
0.12660.002

0.200 60.001
0.256 60.019
0.132
0.132

The resulting equilibrium concentrations for

E

the five species can be finally introduced into

All data are averages over three samples; where significant, the standard deviation is provided too.

eqs 1 and 2, providing the two equilibrium con-

stants (Table 5 and Fig. 6).
It is worth mentioning that we recover close
0.02760.009

0.159 60.004
0.311 60.013

or above 80% of the theoretical material. An im-

0.024
0.028

portant issue is whether these samples are

D

really representative of what has happened

before isolation, or the polymer had been fractio-
nated, for example loosing low molecular
Please note that K1 and K2 are dimensionless constants (see eqs 1 and 2).
0.18 60.02

0.1260.01
0.0860.01

weights in the methanol extraction (as it hap-

0.18
0.19

pens when disulfides are added during polymer-

A

ization). First, we must consider that, as a con-

ppm) and that of a benzyl mercaptan methylene proton (3.67 ppm).

ppm) and that of a benzyl mercaptan methylene proton (3.67 ppm).

sequence of transfer and purification operations,

mmoles of benzyl mercaptan and allyl disulfide in the feed.

there are unavoidable losses (polymer sticking

0.07
0.03
0.04
0.02
0.03

to glass wool, to flask walls) that are relatively

P2d

important due to the fairly small scale of our

6
6
6
6
6

experiments (half a gram of material). A con-

0.41
0.45
0.47
0.29
0.19

servative estimate is these losses to account for

NMR

5–10% of the material and not to be related to a

0.01
0.01
0.01
0.01
0.05

fractionation of the material. Considering also a

P1c

somehow less than quantitative monomer con-

6
6
6
6
6
Chain-End Transfer Experiments

version (its smell is still present at the end of

0.31
0.32
0.33
0.50
0.64

the reaction), we may assume that up to 10% of

the material to be possibly lost due to fractiona-
R0 SSR0 b

tion. The error of our measurements (see Table

0.2
0.2
0.2
0.4
0.6

5), however, is always well above 10%, we can

assume our data to be representative of the
RSHb

whole polymer.
0.4
0.4
0.4
0.4
0.4

The values obtained for Kexc exc

A and KB are rel-
atively constant at different disulfide/thiolate
Samplea

A KB , this seems to indicate

ratios. Since Kexc exc
1200
150
600

150
150

a fairly strong tendency to the formation of the

d
a
b

e
P1:2
P1:2
P1:2
P1:1
P2:1

mixed, terminal disulfide.

Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
2246 KILCHER, WANG, AND TIRELLI

Scheme 5. In this approach a mixture of polymers (end-capped, monomeric and di-

meric disulfides) is treated with tributyl phosphine, which should affect only the di-
sulfide-containing macromolecules. The end product should therefore be composed
only by identical polymers differing only in the terminal groups (amides from the
first end-capping, esters from the second one).

A comparison with literature seems to con-
firm the scale of our results. The Kexc exc
A KB prod-
uct sums up the overall exchange between sym-
metric structures and values are available in lit-
erature for this product in the formation of a
symmetric disulfide from two thiolates; indeed
our value (in between 2 and 4) situates in the
same range as that recorded (in water or metha-
nol) for the mercaptoethanol-induced reduction
of a large number of disulfides.10
If we now return to the only apparent incon-
gruence shown in Figure 4, that is, the relative
increase in dimeric peak with decreasing disul-
fide amounts, it is possible to qualitatively con-
firm the GPC results by calculating the
Apeak2 =Apeak1 ratio (weight fraction of the di-
meric polymers divided by the weight fraction of
monomeric polymers) from NMR data as the ra-
tio 2*[C]/([A]þ[E] (Table 4, last column, values
in brackets).
Proof of Principle of the Reusability of Disulfides
Finally, we have also investigated the possibility
to modify the disulfides at a later stage by
reducing them to thiols and then performing
chemistry on these groups. Reductions of inter-
nal disulfides have been used as biomimetic
means to change molecular weight and associa-
tion of amphiphilic block copolymers.13–16
Journal of Polymer Science: Part A: Polymer Chemistry
In addition, this approach would allow to
reduce the polydispersity of a sample prepared
in the presence of disulfide impurities: after
reduction the dimeric and monomeric polymers
would provide thiol-terminated chains with
identical molecular weight.
We have employed some of the polymers
described in the last section (polymerization,
then exchange with disulfide, then end-capping
with iodoacetamide), which are therefore pres-
ent as mixtures of macromolecules containing
end-capping agents, terminal disulfides and in-
ternal disulfides (Scheme 5). These polymers
were firstly reduced with tributyl phosphine:
since tributyl phosphine mode of action is based
on the presence of water, the reduction step was
attempted with or without addition of water to
degassed, but not anhydrous THF solution.
However, no appreciable difference was recorded
in the polymers by GPC or NMR, indicating the
traces of water present in THF to be sufficient
for effectively performing the reduction.
We have here employed a different end-cap-
ping agent, ethyl 2-bromoacetate (as opposed to
iodoacetamide), to highlight the yield of the sec-
ond end-capping and also any degradation of
amide groups. The end-capping was performed
in situ after the reduction (samples PRED 1 to
PRED
4 ), or in alternative, the polymers were pre-
cipitated in degassed methanol under nitrogen
DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION 2247

Table 6. NMR Characterization Data for Various Polysulfide Samples Before and After Reduction/End-Capping
Experiments

Starting material After reduction & end-capping

Samplea TBP/S-S P1b P2c Pd P1b P2c P3d P4e

PRED
1 4:1 0.62 0.20 0.18 0.04 0.20 0.62 0.14
PRED
2 (from a P1:2_150 sample) 16:1 0.32 0.43 0.25 0.03 0.38 0.41 0.18
PRED
3 (from a P1:1_150 sample) 16:1 0.50 0.29 0.21 0.03 0.28 0.52 0.17
PRED
4 (from a P2:1_150 sample) 16:1 0.63 0.24 0.13 0.04 0.24 0.50 0.22
f
PRED
5 (from a P1:1_150 sample) 16:1 0.50 0.31 0.19 0.03 0.31 0.29 0.41
f
PRED
6 (from a P2:1_150 sample) 16:1 0.60 0.17 0.23 0.03 0.17 0.30 0.50
a
DBU/S-S and ethyl 2-bromoacetate/S-S: 4:1 (for a theoretical DBU/SH or EBA:S-S ratio of 2:1).
b
Ratio between terminal groups in form of double bonds and those in form of aromatic groups ¼ fraction of double bond-
terminated chains.
c
Ratio between terminal groups in form of amides and those in form of aromatic groups ¼ fraction of amide-terminated
chains.
d
Fraction of chains with thiols or dimerized through internal disulfides, calculated as 1  (P1 þ P2) before reduction and as
1  (P1 þ P2 þ P3) after reduction and end-capping.
e
Ratio between terminal groups in form of ester and those in form of aromatic groups ¼ fraction of ester terminated chains.
f
These samples were first reduced, then precipitated twice to remove tributyl phosphine and finally end-capped.

Figure 7. 1H NMR spectra (sections showing the terminal groups) for P2:1_150 before
and after treatment with tributyl phosphine, and after end-capping with ethyl bro-
moacetate (sample PRED
4 ). The signals associated to the double bonds disappear
almost completely after reduction. All samples were precipitated twice in methanol
before NMR analysis.
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
2248 KILCHER, WANG, AND TIRELLI
Figure 8. Overlaid GPC traces for a few polysulfide
samples before and after reduction/end-capping.
Please note that the episulfide degree of polymeriza-
tion does not change during this operation, but any
difference is attributable to changes in terminal
groups and/or dimerization. It is indeed noticeable
that the main peak slightly shifts to higher molecular
weight, as a reasonable consequence of the introduc-
tion of acetic esters in the place of allylic double
bonds. The relative increase of the second peak is dis-
cussed in the text.
atmosphere, redissolved in degassed THF and
end-capped (samples PRED and PRED ). The char-
5 6
acterization data are provided in Table 6.
Allyl signals disappear almost completely
from NMR spectra (Fig. 7), showing tributyl
phosphine almost quantitatively reduces at least
the terminal disulfides (the only ones containing
double bonds). After end-capping, esters are
introduced in the polymer chains.
However, this end-capping procedure gives
much less than quantitative yields (the method,
on the other hand, generally provides quantita-
tive yields both on protein thiols17 and also on
our polysulfides (manuscript in preparation)).
This unsatisfactory result is not related to insuf-
ficient amounts of reducing agent: samples with
1:4 or 1:16 phosphine/disulfide ratios gave anal-
ogous yields.
A possible reason is the degradation of bro-
moacetates by tributyl phosphine: indeed, the
end-capping yield is significantly increased up
to >60% (samples PRED5 and PRED
6 ) by removing
most of this reducing agent through double pre-
cipitation in methanol. Please note that this
degradation/reaction must involve the 2-bromo-
acetate groups, not the reacted esters: there is
no sign of carboxylates on the polymer chains,
neither in IR (stretching C¼ ¼O) nor in NMR (a
protons) spectra.
It is also noticeable that, even if improved,
the end-capping stage is not yet quantitative;
this can, however, been explained by partial oxi-
dation occurring during the complex procedure
(e.g., during solvent evaporation) or by the pres-
ence of residual phosphine (additional extrac-
tions with methanol, however, have not
improved the yield).
GPC analysis (Fig. 8) showed that after end-
capping, with both procedures, the dimeric peak is
still present and its area moderately increases af-
ter reduction/end-capping (but the increase is less
for polymers with higher end-capping yields).
We therefore know that (a) the treatment
with tributyl phosphine is effective at least
towards terminal disulfides, (b) end-capping had
been partially hampered by parasite reactions
mostly related to the reactivity of the phosphine,
(c) even in the best cases, the number of non-
end-capped chains is higher than that of chains
originally involved in internal disulfides. We can
therefore conclude that at least part of the thiols
originated from terminal disulfides have in part
recombined to internal disulfides.
On the other hand, it is not clear whether the
internal disulfides have been reduced, but, if
this has happened, the resulting thiols have,
analogously to the former case (the two disul-
fides do generate the same kind of thiols), in
part recombined.
CONCLUSIONS
Disulfides have a chain transfer action in episul-
fide polymerization.
Qualitatively, the thiolate–disulfide exchange
rate constant
ƒƒ! is proplikely higher than that of propa-
gation ðkexc
A > k  5  10 3 103 min1 mol1 lÞ.
However, when only small amounts of disulfides
are present, a high monomer/disulfide ratio is
sufficient to render the exchange rate signifi-
cantly smaller than that of propagation (if we
consider in Pfransf
blank the propagation rate at least
one order of magnitude higher than the transfer
rate and max. 4% thiols to be present as disul-
fides, we could set an upper limit of
ƒƒ!
kexc
A  10 5
min1 mol1 lÞ. In this case the
exchange would mainly occur at the end of the
polymerization, that is, between chains pro-
duced with a substantially controlled (¼ no
transfer, low polydispersity) mechanism. This
finding suggests that therefore any complication
arising from disulfide impurities (internal or
terminal macromolecular disulfides) could be
avoided by interrupting propagation (e.g.,
through end-capping) at moderate conversions.
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
 THIOL-DISULFIDE EXCHANGE IN POLYMERIZATION
On the other hand, when the polymerization
is performed in the presence of conspicuous
amounts of disulfides, it is apparent that chain
transfer happens rapidly and that the equilib-
rium conditions between the different forms of
disulfides and thiolates are reached. We have
therefore focused on measuring the equilibrium
constants for the exchange between a thiolate-
terminated polysulfide chain and a model disul-
fide, showing that the first exchange (leading to
a terminal disulfide) is more thermodynamically
favored, while the second exchange (between
two polymer chains and leading to an internal
disulfides) has an equilibrium constant <1. We
ascribe this result mostly to entropic reasons,
since the product of the second reaction, charac-
terized by a double molar mass, should have
lower translational and rotational freedom (dif-
fusion), leading to a negative DS. Although not
measured, we also expect a similar behavior for
the kinetic constants, due to the increased steric
hindrance in a reaction between two polymers.
Finally, we have demonstrated that at least
the terminal disulfides can be reduced and
derivatized, although the overall functionaliz-
ability of the resulting thiols was revealed to be
difficult and samples with a higher fraction of
internal disulfides were recovered.
The authors thank Micap Plc. (Runcorn, UK) and spe-
cifically Craig Duckham for the fundamental financial
support. BBSRC is gratefully acknowledged for finan-
cial support (grant No. BBS/B/01,537). NT is indebted
to EPSRC for an Advanced Research Fellowship.
Journal of Polymer Science: Part A: Polymer Chemistry
DOI 10.1002/pola
2249
REFERENCES AND NOTES
1. Napoli, A.; Tirelli, N.; Kilcher, G.; Hubbell, J. A.
Macromolecules 2001, 34, 8913–8917.
2. Rehor, A.; Tirelli, N.; Hubbell, J. A. Macromole-
cules 2002, 35, 8688–8693.
3. Napoli, A.; Valentini, M.; Tirelli, N.; Muller, M.;
Hubbell, J. A.; Nat Mater 2004, 3, 183–189.
4. Rehor, A.; Hubbell, J. A.; Tirelli, N. Langmuir
2005, 21, 411–417.
5. Gilbert, H. F. Methods Enzymol 1984, 107, 330–351.
6. Gilbert, H. F. Adv Enzymol Relat Areas Mol Biol
1990, 63, 69–172.
7. Fava, A.; Iliceto, A.; Camera, E. J Am Chem Soc
1957, 79, 833–838.
8. Whitesides, G. M.; Lilburn, J. E.; Szajewski, R. P.
J Org Chem 1977, 42, 332–338.
9. Houk, J.; Singh, R.; Whitesides, G. M. Methods
Enzymol 1987, 143, 129–140.
10. Lees, W. J.; Whitesides, G. M. J Org Chem 1993,
58, 642–647.
11. Bonnansplaisance, C.; Levesque, G.; Midrak, A.;
Eur Polym J 1994, 30, 239–244.
12. Kreevoy, M. M.; Harper, E. T.; Duvall, R. E.;
Wilgus, H. S.; Ditsch, L. T. J Am Chem Soc 1960,
82, 4899–4902.
13. Wang, L.; Li, C. M.; Ryan, A. J.; Armes, S. P. Adv
Mater 2006, 18, 1566–1570.
14. Li, C. M.; Madsen, J.; Armes, S. P.; Lewis, A. L.
Angew Chem Int Ed 2006, 45, 3510–3513.
15. Oh, J. K.; Tang, C. B.; Gao, H. F.; Tsarevsky, N.
V.; Matyjaszewski, K. J Am Chem Soc 2006, 128,
5578–5584.
16. Tsarevsky, N. V.; Matyjaszewski, K. Macromole-
cules 2005, 38, 3087–3092.
17. Caruso, F.; Rodda, E.; Furlong, D. N. J Colloid
Interface Sci 1996, 178, 104–115.