Mechanism of autocatalysis in the thermal dehydrochlorination of poly(vinyl chloride)

Article abstract of DOI:10.1021/ma0352835

Autocatalysis during the thermal dehydrochlorination of poly(vinyl chloride) (PVC) is shown to be a free-radical process that converts the ordinary monomer units of the polymer into chloroallylic structures that have low thermal stabilities. In the first stage of dehydrochlorination, conjugated polyene sequences are created by a nonfree-radical route. They react with HCl to give cation monoradicals and/or excited cation diradicals. One or both of these species, or other radicals formed them, can then abstract methylene hydrogen in order to produce new radicals that are also carbon-centered. These are converted by chlorine-atom β scission into the chloroallylic segments, which start the growth of new polyenes in the usual (nonradical) way. At 180°C in solid PVC, autocatalysis was inhibited by free-radical scavengers (a hindered phenol, triphenylmethane, and metallic mercury) but greatly enhanced by an increased concentration of HCl when all-trans-β-carotene, a model for PVC polyene sequences, was introduced simultaneously. When the were subjected to autocatalytic conditions, other model compounds gave products that apparently resulted from the abstraction of hydrogen by free-radical intermediates.

Full text of DOI:10.1021/ma0352835

352
Macromolecules 2004, 37, 352-359
Mechanism of Autocatalysis in the Thermal Dehydrochlorination of
Poly(vinyl chloride)
Willia m H. Sta r n es, J r .* a n d Xia n lon g Ge†
Department of Chemistry and Department of Applied Science, College of William and Mary,
Williamsburg, Virginia 23187-8795
Received August 29, 2003; Revised Manuscript Received November 6, 2003
ABSTRACT: Autocatalysis during the thermal dehydrochlorination of poly(vinyl chloride) (PVC) is shown
to be a free-radical process that converts the ordinary monomer units of the polymer into chloroallylic
structures that have low thermal stabilities. In the first stage of dehydrochlorination, conjugated polyene
sequences are created by a nonfree-radical route. They react with HCl to give cation monoradicals and/
or excited cation diradicals. One or both of these species, or other radicals formed from them, can then
abstract methylene hydrogen in order to produce new radicals that are also carbon-centered. These are
converted by chlorine-atom â scission into the chloroallylic segments, which start the growth of new
polyenes in the usual (nonradical) way. At 180 °C in solid PVC, autocatalysis was inhibited by free-
radical scavengers (a hindered phenol, triphenylmethane, and metallic mercury) but greatly enhanced
by an increased concentration of HCl when all-trans-â-carotene, a model for PVC polyene sequences,
was introduced simultaneously. When they were subjected to autocatalytic conditions, other model
compounds gave products that apparently resulted from the abstraction of hydrogen by free-radical
intermediates.
In tr od u ction
Poly(vinyl chloride) (PVC) continues to be an ex-
tremely important commercial polymer, despite its
relatively low thermal stability, which fortunately can
be enhanced considerably by the use of protective
additives. Technological advances in the stabilization
of PVC obviously should be expedited by the availability
of information about the molecular mechanism by which
the polymer degrades. For that reason, the mechanism
of PVC thermolysis has been subjected to an enormous
amount of detailed study (and speculation) for more
than 50 years. Nevertheless, there are significant
aspects of the mechanism that still have not been
clarified. Here we propose solutions to some of the
remaining major problems.
steps whose transition states are highly polarized, as
shown in eq 3.2 Termination refers to the cessation of
polyene growth, which occurs when the polyene se-
quences still are rather short. In fact, all thermal
degradations of PVC engender a polyene sequence
length distribution wherein the average number of
double bonds per sequence typically ranges from only
about 3 to 20, depending on conditions.1 The termination
reactions have not been identified. However, several
possibilities are apparent, including various intra- or
intermolecular cyclizations of the polyenes themselves.2-4
Other facets of thermal degradation that are, if
anything, even more subtle will concern us here. One
of these has to do with initiation from the ordinary
(head-to-tail) monomer units. Such units are generally
believed to be much more stable than the structural
defects mentioned above.2,5,6 Hence, in the absence of
perturbing factors, loss of labile defects in the initial
stage of degradation should lead to autodeceleration in
the rate of evolution of HCl. This behavior is, indeed,
observed under some conditions.5,6 Yet, in other cases,
when most or all of the original defects have reacted,
dehydrochlorination does not slow appreciably and may
even autoaccelerate, sometimes drastically.5,6 The rea-
son for these kinetic effects has been the subject of much
debate, and they have been said by some researchers7,8
to result from the presence of R,â-unsaturated ketone
groups in the polymer. Supposedly formed by adventi-
Ba ck gr ou n d . In its initial phase, the thermal deg-
radation of PVC is primarily a process of sequential
dehydrochlorination that forms conjugated polyene
sequences (eq 1).1 This process is now well-recognized
-(CH2CHCl)n- 9
∆8 -(CHdCH)n- + nHCl (1)
to be a chain reaction in which initiation, propagation,
and termination steps occur.1 During initiation, various
structural segments of the polymer lose HCl and thereby
are converted into allylic chloride groups. During propa-
gation, these groups elongate very rapidly into polyenes
as dehydrochlorination progresses. At the outset of
degradation, the principal initiating species are highly
reactive “structural defects”.2 They are believed by most
researchers to be tertiary chloride moieties or allylic
chloride structures that are not at the ends of polymer
chains.2 Both initiation and propagation occur exclu-
sively either via ion pairs (eq 2) or by way of concerted
* Corresponding author. E-mail: whstar@wm.edu.
† E-mail: gex@nsula.edu.
10.1021/ma0352835 CCC: $27.50 © 2004 American Chemical Society
Published on Web 12/19/2003

Macromolecules, Vol. 37, No. 2, 2004
Mechanism of Autocatalysis 353
tious air oxidation, the cisoid forms of such structures
were suggested to serve as true catalysts for the
dehydrochlorination of ordinary monomer units, a pro-
cess that would continually create thermally labile
allylic sites and thereby prevent deceleration.7,8 On the
other hand, there is no conclusive evidence for the
presence of enone groups in commercial PVC,2,3 and
their proposed catalytic proficiency could not be con-
firmed by other workers in model-compound studies.2,9,10
and that mercury was thereby converted, to some
extent, into mercurous chloride.
Other relevant observations were reported by Hjert-
berg and So¨rvik,16 who found that thermal degradation
in an atmosphere containing 10 vol % of HCl caused
the average polyene length to increase significantly.
This result was reasonably attributed to the catalysis
of propagation by HCl, although it was pointed out later3
that HCl inhibition of polyene shortening through
cyclization might have been involved. At higher atmo-
spheric HCl contents (up to 40 vol %), the polyene
distribution underwent no further change, but the
autoacceleration of dehydrochlorination occurred to an
extent that increased with increasing concentrations of
polyenes and HCl.16 The autoacceleration was ascribed
to enhanced initiation that was caused by polyene-HCl
complexes whose formation somehow improved the
ability of HCl to promote the dehydrochlorination of
ordinary monomer units.16 Direct evidence for this
increased initiation (but not for its mechanistic ratio-
nale) was provided later by measurements which showed
convincingly that autoacceleration was accompanied by
an increase in the number of polyene sequences.17
An alternative explanation for the absence of decel-
eration is that the dehydrochlorination of ordinary units
is catalyzed by the evolving HCl. The presence of this
substance does indeed tend to increase the rate dramati-
cally.1-3,5,6,11 However, the reasons for such behavior are
complex, in that HCl is potentially able to alter the rates
of propagation and termination as well as that of
initiation. The effects of HCl on the degradation of PVC
have been discussed in several reviews,1,3,11 which show
that much still remains to be understood about the
mechanisms through which these effects occur.
Another perplexing aspect of PVC thermolysis is how
and to what extent free radicals are involved in the
process. As noted above, radical species do not intervene
in initiation or polyene elongation. Nevertheless, the
evidence for their presence in degrading PVC is exten-
sive, and it has been summarized and/or analyzed in
some recent publications.2,5,6,12,13
Recent studies in this area by Troitskii et al. are of
particular interest. One of these investigations involved
the use of literature kinetic data in order to develop
mathematical models of several schemes for thermal
dehydrochlorination. The results strongly supported a
mechanism for autocatalysis in which the interaction
of HCl with polyenes forms a chemical species whose
reaction with ordinary monomer units gives new sites
for initiation.18 That the autocatalysis involves free
radicals was indicated by other findings, one of which
was that fullerene C60, an effective radical trap19,20 that
does not react with HCl,21 retarded thermal dehydro-
chlorination when HCl was present but not when it was
removed by freezing.21 Also, an ESR study of thermally
degraded PVC showed that the number of stable spins
was much greater in samples that had experienced
autocatalysis caused by exposure to HCl.22 This result
was considered to be indicative of higher spin concen-
trations in these specimens during the degradation
process itself.
Au toca ta lysis Retr osp ective: P r eviou s Stu d ies
a n d Ra tion a les. A principal thesis of this paper is that
several features of PVC degradation, identified above,
that have not been explained heretofore can now be
understood easily in terms of a novel mechanism for
autocatalysis during thermal dehydrochlorination. This
mechanism requires the simultaneous presence of poly-
enes and HCl, and it involves cation radicals as essential
intermediates. The evidence for its operation will be
considered here in detail, beginning with an assessment
of work by previous investigators.
More than 30 years ago, Rasuvaev et al.14 reported
the results of experiments in which PVC was first
degraded thermally under vacuum for various lengths
of time while the HCl formed was removed by freezing.
The degraded specimens then were placed under an HCl
pressure of 150 mm and subjected to further thermolysis
while the HCl formed was allowed to accumulate in the
system. This sequential treatment led to huge accelera-
tions in the rate of dehydrochlorination whose magni-
tudes increased with increasing time of predegradation.
The autoaccelerations were attributed to an interaction
between HCl and polyene segments to form free radicals
that stimulated dehydrochlorination by a radical route,
and this proposal was supported by (a) the complete
inhibition of autocatalysis by maleic anhydride, which
was thought to destroy the polyenes by Diels-Alder
cycloaddition, and (b) the retardation of autocatalysis
by metallic mercury, toluene, and triphenylmethane,
which were believed to serve as radical scavengers.
Curiously, at the time when it was suggested, this
autoacceleration scenario does not seem to have at-
tracted much support from other researchers, perhaps
because the specific chemical mechanism proposed to
explain it14 was unprecedented and highly speculative.
The same fate seems to have befallen some additional
arguments for a radical pathway that were made later
by Troitskii and co-workers,15 who reported that silver
and glass, as well as mercury, retarded dehydrochlori-
nation in the presence of HCl, though not in its absence,
New detailed mechanisms for autocatalysis have now
been proposed by Troitskii and Troitskaya.5,6,13 These
mechanisms are heavily based on theoretical and ex-
perimental evidence in the literature which shows that
the energies needed for the conversion of conjugated
polyenyl cations into their first excited singlet and
triplet states are much less than those required for the
analogous excitations of the corresponding neutral
polyenes. One of the schemes considered5,6,13 begins with
the reversible transformation of labile structure 1 into
an ion pair (2) whose cationic component is created in
an excited diradical state. Discharge of this ion pair by
proton transfer produces the excited neutral diradical,
3 (eq 4), which in theory could abstract a methylene
hydrogen atom from an ordinary monomer unit to give
monoradical 4 and a â-chloroalkyl radical (5) (eq 5). A
conventional â-cleavage reaction (eq 6) then forms a new
allylic chloride unit that can initiate the growth of a new
polyene sequence, thereby causing autocatalysis. How-
ever, since autocatalysis requires the presence of HCl,
reaction 5 was considered to be much less important
than reaction 7, which produces a chlorine atom that
was said to initiate a homolytic mechanism for degrada-
tion.13 The exact nature of this mechanism was not

354 Starnes and Ge
Macromolecules, Vol. 37, No. 2, 2004
explained, but an obvious possibility would involve H
atom abstraction by Cl• from an ordinary monomer unit,
as in eq 8.
F igu r e 1. Dehydrochlorination of PVC (100 mg) at 180 °C
(mL/min of Ar, mg of BHT): a (10, 0); b (80, 0); c (10, 10).
Another autocatalysis mechanism proposed by Troitskii
and Troitskaya5,6,13 was actually the one that they
preferred. It involves new reactions 9-11 and, as
presented in one publication,13 includes reaction 7.
Formation of 3 via eqs 9 and 10 was thought to be much
more facile than its formation by eq 4, owing to the
ability of HCl to serve as an electrophilic catalyst for
C-Cl heterolysis.
F igu r e 2. Dehydrochlorination of PVC (100 mg) at 180 °C
(mL/min of Ar, mg of 8, mg of BHT): a (10, 4, 0); b (10, 4, 10);
c (10, 0, 0); d (80, 4, 0); e (10, 4, 100).
The suggested5,6,13 formation of cation-diradical-
containing ion pairs (2 and 7) is intriguing and is
considered further below. Difficulties arise, however,
with other aspects of the mechanisms just described.
First, it is not clear why the suggested pathways leading
to 3 would necessarily produce a higher concentration
of this intermediate than that resulting from the simple
thermal excitation of a conjugated polyene sequence.
The overall thermodynamics of all of these routes to 3
are the same, and the thermal formation of diradicals
(disolitons) from polyacetylene is documented convinc-
ingly.23 Thermal production of solitons from much
shorter polyene sequences, such as that in â-carotene,
has also been described.24,25 Moreover, reaction 7 ap-
pears even less likely than the unobserved reaction 5,
owing to the bond strength difference between HCl and
sec-C-H (103 vs ca. 98 kcal/mol)26 and to the much
lower concentration of HCl as compared to that of the
ordinary monomer units. Finally, reaction 11 is entirely
without precedent.
was not affected significantly by an 8-fold increase in
the rate of argon flow, which would have reduced the
HCl content of the atmosphere surrounding the sample.
Since no autoacceleration occurred in these two runs,
their observed rates were determined entirely by the
rate of initiation and the propagation/termination rate
ratio, with the amounts of HCl dissolved in the samples
being sufficient to maximize these kinetic factors in both
cases. A similar situation was described by Hjertberg
et al.17 Furthermore, at 200 min, the polyene concentra-
tions still were quite low, as the extent of dehydrochlo-
rination at that time was only ∼1.5 mol %. This
observation tends to support the proposition that au-
tocatalysis arises from an interaction of polyenes with
HCl. Last, and very importantly, plot c of Figure 1
shows that the well-known free-radical scavenger “BHT”
(2,6-di-tert-butyl-4-methylphenol) had no appreciable
effect on the rate when autocatalysis was not occurring.
In runs a, b, d, and e of Figure 2, the initial presence
of polyene sequences was ensured by incorporating a
small amount of â-carotene (8, nominally all-trans at
Resu lts a n d Discu ssion
Au toca ta lysis Revisited : New Exp er im en ta l Ob-
ser va tion s. In an attempt to obtain more insight into
the mechanism for autocatalysis, we have measured the
rates of HCl evolution from powdered PVC and PVC-
additive blends at 180 °C under flowing argon, using a
procedure that is described in detail elsewhere.27 The
HCl was swept by the gas stream into a vessel contain-
ing water and titrated there automatically with aqueous
sodium hydroxide.
the outset) into the starting samples. When the ambient
concentration of HCl was minimized (in run d), the
addition of 8 did not increase the rate (cf. run c of Figure
2). On the other hand, when 8 was present and the HCl
content was maximized by reducing the argon flow, a
Plots a and b of Figure 1 show that the rate of
dehydrochlorination (as determined from the slopes)

Macromolecules, Vol. 37, No. 2, 2004
Mechanism of Autocatalysis 355
F igu r e 3. Dehydrochlorination of PVC (100 mg) at 180 °C
under Ar (10 mL/min) with addition of BHT (100 mg) at 24 h.
See text for discussion.
F igu r e 5. Dehydrochlorination of PVC (100 mg) at 180 °C
under Ar (10 mL/min) (mg of 8, mg of triphenylmethane): a
(4, 0); b (4, 20); c (4, 60); d (0, 0); e (4, 100).
low strength of the C-Hg bond in RHg• radicals.28 On
the other hand, reaction 15 (where R• is carbon-
centered) seems quite reasonable, and its organometallic
product should readily scavenge HCl (eq 16).
Cl• + Hg f ClHg•
2 ClHg• f Hg2Cl2
(12)
(13)
Cl• + ClHg• f HgCl2
(14)
R• + ClHg• f RHgCl
(15)
(16)
F igu r e 4. Dehydrochlorination of PVC (100 mg) at 180 °C
under Ar (10 mL/min) (mg of 8, mg of Hg): a (4, 0); b (4, 30);
c (0, 0); d (4, 100).
RHgCl + HCl f RH + HgCl2
powerful autocatalytic effect was observed. Its temporal
profile (run a) closely resembles that reported24 for the
generation of free spins in a similar experiment with 8
and PVC. Moreover, the autocatalysis was greatly
reduced (in run b) and eventually obliterated (in run e)
by adding increasingly large amounts of the radical
scavenger. In the latter run, however, a part of the rate
reduction might have resulted from a decrease in the
polarity of the medium.
Autocatalysis was also found in the absence of 8 when
reaction times were of sufficient length to allow the
creation in situ of high polyene concentrations. After 24
h of reaction, the extent of dehydrochlorination was ca.
15 mol %, and the homogeneous blending of BHT with
the heavily cross-linked polymer was impossible. Nev-
ertheless, as Figure 3 reveals, the addition of BHT at
this point essentially stopped the autoacceleration and
thereby provided further evidence for its instigation by
free radicals.
As noted above, mercury has been reported to retard
the thermal dehydrochlorination of PVC14,15 and, in so
doing, to be converted, in part, into mercurous chlo-
ride.15 In our hands, the preparation of thoroughly
homogenized mixtures of PVC and mercury could not
be achieved. However, we found that, even so, rather
large amounts of mercury effectively thwarted the
acceleration of degradation that was caused by 8. In fact,
the rate of dehydrochlorination remained very close to
its original (uncatalyzed) level when identical weights
of mercury and PVC were used. These effects are
illustrated graphically by plots a-d in Figure 4.
The inhibition caused by mercury undoubtedly re-
sults, to some extent, from chlorine-atom scavenging
involving eqs 12 and 13.5,6 Reaction 14 is another likely
possibility, but the direct scavenging of carbon-centered
radicals by mercury is improbable, owing to the very
To confirm the expected inability of HCl to react
directly with mercury, 0.1 g of the metal was heated in
a sealed tube at 180 °C with 2.0 g of 5-chloro-5-
methylnonane. After 9 h, analysis by GC/MS showed
that 62% (14 mol equiv) of the alkyl chloride had
undergone dehydrochlorination. However, visual inspec-
tion strongly suggested that the mercury had survived
the reaction entirely unscathed.
Various workers have reported the use of triphenyl-
methane to retard the thermal dehydrochlorination of
PVC.5,6,14,29,30 The retardation mechanism has not been
firmly established but is reasonably presumed to involve
the abstraction of hydrogen by radicals from the ali-
phatic C-H bond. Curves a-c and e in Figure 5 show
that triphenylmethane was able to reduce the autocata-
lytic effect of 8. Though not as efficient in this respect
as BHT (cf. the curves in Figure 2), it completely
removed the autocatalysis when a sufficient amount of
it was present (cf. curves d and e of Figure 5). Signifi-
cantly, as was the case for BHT and mercury, it had no
effect on the rate of dehydrochlorination when autoca-
talysis was not occurring, apart from a slight decrease
in run e that a reduction in medium polarity could have
caused.
Does autoacceleration require the presence of HCl at
the outset, or will another Bro¨nsted acid suffice? This
question was addressed by studying the dehydrochlo-
rination of 4-chloroheptane at 180 °C in the presence
of 8 and/or phosphoric acid. The total yield of heptenes
was measured, and the data are summarized in Table
1. Runs 1 and 2 listed therein show that, in the absence
of H3PO4, the extent of dehydrochlorination was insig-
nificant and was not increased by the presence of 8. The
latter observation indicates that neutral diradicals
formed from the â-carotene by thermal excitation did
not abstract hydrogen from the substrate to form

356 Starnes and Ge
Macromolecules, Vol. 37, No. 2, 2004
Ta ble 1. Th er m a l Deh yd r och lor in a tion of
4-Ch lor oh ep ta n ea
The product mixtures formed in runs 3 and 4 of Table
2 were complex and could not be analyzed fully. Possible
chemical complications were, inter alia, (a) loss of the
OdC•R and R• radicals via their rapid addition to
polyenes31 and (b) aldol condensation of the aldehyde,
which would have been promoted by HCl. In similar
experiments, our use of an aldehyde stable to acid, 2,2-
dimethyloctanal, afforded intractable mixtures contain-
ing none of the 2-methyloctane that would have resulted
from reactions 19-21.
additive wt,b mg
H3PO4
run
8
heptenes, mol %
0.36
yield ratio
1
2
3
4
0
0
1.0
0.9
13
40
0
40
0
40
40
0.34
4.6 ( 0.9c
18.4 ( 1.5d
51
a
At 180 °C for
8
h
in a sealed tube under argon. See
b
Experimental Section for details. Weight of 4-chloroheptane was
Mech a n ism of Au toca ta lysis: A New P r op osa l.
When considered within the context of credible earlier
findings that were discussed above, our new experimen-
tal observations conclusively show that autocatalysis
during the thermal dehydrochlorination of PVC involves
the reaction of HCl with polyenes in order to form free
radicals that attack the ordinary monomer units of the
polymer. These units are thereby converted into ther-
mally labile structures that initiate the growth of
additional polyene sequences. The attack on the ordi-
nary units is logically formulated as a hydrogen ab-
straction process that creates radical 5, which then
produces the unstable moiety 1 (n ) 1) by chlorine-atom
â scission (eq 6). What still remains to be addressed is
the nature of the initial reactive radical species.
A defensible possibility was previously discussed by
one of us.2 This is the excited cation diradical component
of ion pair 7, which is shown in eq 10. Its potential
intermediacy follows by analogy from a speculated32
thermal equilibration of â-carotene with an orthogonal
neutral diradical formed by twisting about the central
double bond. Energy comparisons summarized else-
where13 and calculated energy barriers for rotation33
show that the formation of analogous diradicals from
conjugated polyenyl cations should be much more
facile.13 As a result, in PVC, the equilibrium concentra-
tions of these cation diradicals might be higher than
those of the corresponding neutral diradicals (3) formed
from neutral polyenes by thermal excitation. Moreover,
owing to electrophilicity conferred by the positive charge,
the cation diradicals could well be more reactive than
3 in the abstraction of methylene hydrogen from the
polymer.
400 mg in all runs. c Average deviation for five experiments.
d
Average deviation for four experiments.
Ta ble 2. Deca r bon yla tion of 2-Eth ylh exa n a la
additive wt,b mg
run
8
PVC
heptane, mol %
1
2
3
4
0
40
0
0
0
160
160
0
0
0.15 ( 0.05c
0.28 ( 0.02c
40
a
At 180 °C for
8 h in a sealed tube under argon. See
b
Experimental Section for details. Weight of 2-ethylhexanal was
400 mg in all runs. c Average for two experiments.
â-chloroalkyl radicals (cf. 5) that were converted by
thermolysis into alkenes (cf. eq 6). This finding argues
strongly against the occurrence of reaction 5 in degrad-
ing PVC. The increased yield of heptenes found in run
3 of Table 1 can be rationalized easily by invoking the
acid catalysis of C-Cl heterolysis (eq 17) followed by
proton loss (eq 18).2 However, the strong synergism
between 8 and phosphoric acid in run 4 seems best
explained by an interaction of these two additives to
form one or more free-radical intermediates that are not
analogous to 3 but can abstract hydrogen from the 3-
and 5-positions of the alkyl chloride, nonetheless.
Abstraction from the other positions of 4-chlorohep-
tane may have occurred as well. In that event, the
resulting radicals could not have formed heptenes but
probably were scavenged by 8 instead.
As a further test for the presence of radicals, the
radical-induced decarbonylation of an aldehyde was
explored. Such reactions are known to proceed by the
mechanism shown in eqs 19-21,31 where for our system,
Cation diradicals conceivably could arise from grow-
ing polyenes containing a chloroallylic terminus, as
shown in eqs 9 and 10. However, polyenes, once initi-
ated, grow and terminate very rapidly.1,2,4 Hence, even
at rather low levels of dehydrochlorination, the concen-
tration of “dead” polyenes (those that no longer have a
reactive chloroallylic group) should be much greater
than that of 1. For this reason, the protonation of dead
polyenes, P (eq 22), followed by thermal excitation (eq
23) would seem to be a more likely route to cation
P + H+ a PH+
PH+ a PH••+
(22)
(23)
R is 3-heptyl; Y• represents the radical(s) formed from
polyenes and HCl; and [H•] is an abstractable hydrogen
atom in the reaction matrix. Runs 1 and 2 of Table 2
show that, under our conditions, the aldehyde failed to
decarbonylate thermally or in the presence of 8. How-
ever, a small but reproducible yield of heptane was
produced when PVC was added (run 3), and the yield
was higher when 8 was present as well (run 4). We
regard these results as prima facie evidence for radical
intermediates.
diradicals than reactions 9 and 10. The diradicals could
exist primarily as triplets, but they might be formed
from excited singlet diradical intermediates.
Cation monoradicals comprise another type of reactive
radical species that could result from the interaction of
polyenes with HCl. Ample precedent for their formation
in similar systems exists. For example, in a seminal
study of the protonic acid doping of conjugated polymers,
Han and Elsenbaumer34 showed that, in such situations,
cation radical production is a general phenomenon that

Macromolecules, Vol. 37, No. 2, 2004
Mechanism of Autocatalysis 357
apparently is effected by reactions 24 and 25. Reactions
analogous to the latter one have also been reported by
Kispert and co-workers,35-38 who generated dications
from various carotenoids by electrochemical oxidation
and provided strong evidence for their subsequent
comproportionation with their neutral precursors, C, as
in eq 26. In the case of carotenoid 8, the equilibrium
constant for comproportionation (Kcom) was found to
chlorine-atom entrapment per se, because chlorine
atoms and labile defects are generated simultaneously
(eq 6). Chlorine-atom scavenging would, however, tend
to prevent the formation of additional labile sites in a
chain reaction consisting of eqs 6 and 8. Reaction 15
(where R• ) 5, PH••+, PH2•+, P•+) and reaction 16 could
also contribute to the inhibition caused by mercury, but
even so, the ability of this additive to prevent all of the
autocatalysis (see Figure 4) compels one to address the
possibility that the initial H abstracters are actually
chlorine atoms. These are unlikely to be formed directly
from HCl by hydrogen transfer, owing to the high
dissociation energy of the H-Cl bond,26 but an alterna-
tive possibility is the â scission (eq 6) of allylic C-Cl
bonds in cation radicals. On the other hand, the pres-
ence of such bonds in the cation radicals formed from
dead polyenes is uncertain, and even though the read-
dition of HCl to dead polyenes can occur to some
extent,49 it may form carbenium chloride ion pairs50
instead of covalently bonded chlorine when it takes
place at the temperatures that degradation requires.
P + 2H+ a PH2
(24)
(25)
++
PH2++ + P a PH2•+ + P•+
C
++ + C {Kcom} 2C•+
(26)
have a value of 1.22 at ambient temperature in dichlo-
romethane solution.38 Moreover, Kispert et al.39 dem-
onstrated that various carotenoids, including 8, were
converted into polyenyl cation monoradicals when they
were simply exposed to an excess of HCl (concentration,
∼1 M) in dichloromethane solution. This result also
strongly supports the occurrence of reactions 24 and 25
in thermally degrading PVC.
How do the cation di- and/or monoradicals transform
the ordinary monomer units into thermally labile sites?
As noted above, the simplest and by far the most
probable explanation involves the abstraction of meth-
Summarizing the discussion in this section, we con-
clude that cation radical intermediates are intimately
involved in autocatalysis and that this process involves
the abstraction of methylene hydrogen atoms from
ordinary monomer units. However, the identity of the
abstracting radicals remains to be established. The
literature contains conclusive evidence for the formation
of cation monoradicals from polyenes and HCl, and the
presence of excited cation diradicals in such systems is
still a reasonable possibility.
Rela ted Stu d ies. Zuoyun et al.51 have argued em-
phatically that free radicals are not involved in the
thermal dehydrochlorination of PVC. This conclusion
was based primarily on (a) the failure of a hindered
phenol and a polymeric nitroxyl radical to affect the
rate, (b) a lack of ESR evidence for consumption of the
nitroxyl during degradation, and (c) the failure of an
arylnitroso spin trap to scavenge any radicals while the
reaction was in progress. However, all of these results
were obtained in experiments where autocatalysis did
not occur, apparently because it was prevented by the
use of very dilute solutions of PVC, very low extents of
conversion, and low concentrations of HCl caused by
rapid bubbling with nitrogen. For this reason, the
authors’ findings51 do not relate in any way to the
mechanism for autocatalysis, though they do confirm
the absence of radicals from the earlier stages of
degradation.
ylene hydrogen to form 5, which then undergoes reac-
tion 6. Potential abstracting species are PH••+, PH2
,
•+
P•+, and neutral polyenyl monoradicals resulting from
the deprotonation of cation monoradicals. However, the
concentrations of these neutral radicals are likely to be
insignificant, in view of the very low Bro¨nsted acidity
of the cation monoradical formed from 8.38 Moreover,
the abstraction reactivity of the neutral monoradicals
may be much less than those of the related charged
species, a suggestion that is consistent with our failure
to detect any reaction of PVC, 4-chloroheptane, or
2-ethylhexanal with excited neutral diradicals (cf. 3)
formed from 8 by heating. In any case, it now is
apparent that the thermal degradation of PVC will
produce many different cation radicals whose specific
structures are determined by factors such as the lengths
of the precursory neutral polyenes, the position(s) of
polyene protonation,33 cis-trans isomerism,32,33,39-41
and various reactions of the polyenes and cation radicals
that change their sequence lengths.33,40,42-44 All of the
cationic radical species are potentially able to abstract
a hydrogen atom from the polymer.
In a study reported by van Hoang and Guyot,49 small
amounts of 8 decreased the rate of HCl evolution from
heated solutions of PVC in 1,2,4-trichlorobenzene. This
effect apparently resulted from the reversible addition
of HCl to the carotenoid polyene system. At 195 °C,
increasing amounts of 8 caused further rate reductions
until the initial concentration of the additive had
reached a certain level. Surprisingly, the use of ad-
ditional 8 at this point then caused a rate enhancement.
Apparently inexplicable when it was first obtained, this
result now seems consistent (qualitatively) with the
occurrence of autocatalysis by the mechanism we pro-
pose, the importance of which would increase, of course,
with increasing amounts of additive 8.
Conventional wisdom holds that phenols such as BHT
are ineffective traps for carbon-centered radicals.45
However, this belief is based primarily on the behavior
of phenols as antioxidants. In autoxidations that are not
oxygen-starved, carbon-centered radicals are not ef-
ficiently trapped by most phenols, because they are
converted very rapidly into peroxy radicals, highly
electrophilic species that are quickly scavenged in-
stead.45-47 Reactions of carbon-centered radicals with
phenols can actually be quite fast,46,48 and they may be
even faster when the radicals are made electrophilic by
the presence of a positive charge. Thus, the inhibition
of autocatalysis by large amounts of BHT, as revealed
in Figure 2, is by no means inconsistent with the
abstraction of hydrogen from the polymer by carbon-
centered cation radicals.
Tran et al.12,24,25,52 have suggested a polaron (cation
radical) mechanism for the thermal degradation of PVC
and have argued for its operation in several publications
that were recently reviewed in detail.12 As formulated
If such intermediates were indeed the only initial H
abstracters, autocatalysis would not be avoided by

358 Starnes and Ge
Macromolecules, Vol. 37, No. 2, 2004
Under an argon atmosphere, thionyl chloride (20.5 g, 172
mmol) was added dropwise with stirring to a mixture of
pyridine (3 mL, 37 mmol) and 4-heptanol (10.0 g, 86.1 mmol)
while the temperature was kept at 0-5 °C with an ice bath.
When the addition was complete, the mixture was stirred and
heated at 85-90 °C for 3 h, cooled to room temperature,
treated with ice water (100 mL), and extracted with ether (3
× 200 mL). The combined extracts were washed in succession
with 3% aqueous NaHCO3 (2 × 150 mL) and water (2 × 100
mL), dried over anhydrous MgSO4, and concentrated in a
rotary evaporator under aspirator vacuum. Short-path distil-
lation of the residue afforded 6.0 g (yield based on starting
ketone, 51%) of 4-chloroheptane (bp 78-80 °C at 60 Torr) in a
by its originators, the polaron scheme is strictly a
mechanism for normal chain propagation. It does not
address the question of autocatalysis, and it suffers from
several deficiencies that have been discussed else-
where.2,3 Nevertheless, its postulation has had heuristic
value, for it apparently was the first mechanism (and,
for several years, the only one) to consider the possibility
that cation radicals are involved in the thermolysis of
PVC.
Con clu sion s
purity of ∼100%, according to GC/MS analysis: m/e 98 (M+
-
The new mechanism for PVC degradation that has
been presented here accounts for several features of the
process that have been puzzling heretofore. These relate
not only to catalysis by HCl but also to the evidence for
the involvement of free radicals and to the initiation of
polyene growth from ordinary monomer units.
1
HCl). H NMR: δ 0.93 (t, 6H, 2CH3), 1.4-1.6 (m, 4H, 2CH2-
CH3), 1.6-1.7 (m, 4H, 2CH2CHCl), and 3.91 ppm (qnt, 1H,
CHCl). 13C NMR: δ 13.91 (C-1, -7), 20.04 (C-2, -6), 40.94
(C-3, -5), and 63.87 ppm (C-4).
P VC Deh yd r och lor in a tion Kin etics. Dehydrochlorina-
tions were carried out under flowing argon in a glass vessel
thermostated at 180 ( 1 °C, using a Metrohm 702 SM Titrino
apparatus for automatic titration by a procedure that was
previously described.27 The evolving HCl was swept by argon
into water and titrated periodically with 0.01 N NaOH in order
to maintain a solution pH of 7.10. Each sample, including those
with additives, contained 0.100 g of PVC. The PVC-additive
mixtures were prepared by grinding with a mortar and pestle,
and they afforded aliquots whose titration curves were highly
reproducible in the replicate experiments (at least 2-3) that
were performed for each composition. Compound 8 was stored
in the dark under argon at 0-5 °C. Blends containing this
additive were made in a drybox under argon and kept under
argon protection before being used in kinetic runs. Deteriora-
tion of 8 with time was ruled out by the excellent reproduc-
ibility of the results obtained in control runs that were
performed at frequent intervals with mixtures of 8 and PVC.
Deh yd r och lor in a tion of 4-Ch lor oh ep ta n e. The reac-
tants were thoroughly degassed on a vacuum manifold, sealed
in a heavy-walled glass tube in a drybox under argon, and
heated for 8 h at 180 ( 1 °C. The percentage conversion into
heptene isomers was then determined by GC/MS analysis.
Details of these experiments are summarized in Table 1.
Deca r bon yla tion of 2-Eth ylh exa n a l. The procedure used
was identical to that for the experiments with 4-chloroheptane.
After 8 h of heating under argon at 180 ( 1 °C, GC/MS
analysis gave the results in Table 2.
Previous work has shown that dehydrochlorination
begins primarily at thermally labile defect sites. They
start the growth of conjugated polyenes in a process that
is either ionic or quasiionic but certainly does not
involve free radicals. When the concentrations of HCl
and polyenes have reached a certain level, these prod-
ucts react to form polyenyl cation radicals that lead to
autocatalysis. The cation radicals are either monoradi-
cals or excited diradicals, and the autocatalysis involves
the abstraction of methylene hydrogen from ordinary
monomer units in order to form new radicals that are
also carbon-centered. However, the identity of the
abstracting species is not entirely clear. The new
radicals are converted by chlorine-atom â scission into
allylic chloride structures that can start the growth of
new polyene sequences in the conventional way.
Exp er im en ta l Section
Ma ter ia ls. Poly(vinyl chloride) with Mn ) 1.96 × 104 and
Mw ) 3.88 × 104, 2,6-di-tert-butyl-4-methylphenol (BHT,
nominal purity, 99+%), triphenylmethane (nominal purity,
99%), all-trans-â-carotene (8, nominal purity, 95%), mercury
(nominal purity, 99.9995%), and phosphoric acid (nominal
purity, 99.999%) were obtained from Aldrich. The other
chemicals were purchased from various suppliers and had the
highest available purities. Purities of organic materials were
confirmed by NMR and/or (gas chromatography)/(mass spec-
trometry) (GC/MS) measurements.
Ack n ow led gm en t. This work was supported by the
National Science Foundation under Grant No. DMR-
9610361. We thank Dr. L. D. Kispert for a stimulating
discussion.
In str u m en ta l An a lysis. A Varian Mercury VX-400 spec-
1
trometer was used to record H NMR spectra at 400.128 MHz
and proton-decoupled 13C NMR spectra at 100.623 MHz. The
spectra were obtained at ambient temperature from chloro-
form-d solutions containing Me4Si (δ ) 0.00 ppm) for internal
standardization.
Refer en ces a n d Notes
(1) Iva´n, B.; Kelen, T.; Tu¨do¨s, F. In Degradation and Stabiliza-
tion of Polymers; J ellinek, H. H. G., Kachi, H., Eds.; Elsevier
Science: New York, 1989; Vol. 2; Chapter
references therein).
(2) Starnes, W. H., J r. Prog. Polym. Sci. 2002, 27, 2133 (see also
references therein).
(3) Starnes, W. H., J r.; Girois, S. Polym. Yearb. 1995, 12, 105
(see also references therein).
(4) Shapiro, J . S.; Starnes, W. H., J r.; Plitz, I. M.; Hische, D. C.
Macromolecules 1986, 19, 230.
(5) Troitskii, B. B.; Troitskaya, L. S. Int. J . Polym. Mater. 1998,
41, 285 (see also references therein).
(6) Troitskii, B. B.; Troitskaya, L. S. Polym. Yearb. 1999, 16, 237
(see also references therein).
(7) Luka´sˇ, R.; Prˇa´dova´, O. Makromol. Chem. 1986, 187, 2111.
(8) Luka´sˇ, R. Makromol. Chem., Macromol. Symp. 1989, 29, 21.
(9) Starnes, W. H., J r. Polym. Mater. Sci. Eng. 1988, 58, 220.
(10) Starnes, W. H., J r.; Plitz, I. M. Unpublished work.
(11) Hjertberg, T.; So¨rvik, E. M. In Degradation and Stabilisation
of PVC; Owen, E. D., Ed.; Elsevier: New York, 1984; Chapter
2 (see also references therein).
8 (see also
The GC/MS measurements were made with a Hewlett-
Packard HP 5890A/5988A apparatus that was equipped with
a fused-silica HP-1 capillary column [cross-linked methylsili-
cone, 12 m × 0.2 mm (i.d.), 0.33 mm film thickness]. Helium
was used as the carrier gas, and the column temperature was
increased from 50 to 300 °C at the rate of 20 °C/min.
P r ep a r a tion of 4-Ch lor oh ep ta n e. A solution of sodium
borohydride (7.13 g, 189 mmol) and sodium hydroxide (1.5 g)
in water (300 mL) was added dropwise to a stirred solution of
4-heptanone (10.0 g, 87.6 mmol) in ethanol (400 mL) while
the temperature was kept at ca. 10 °C by external cooling. The
resulting mixture was stirred for 12 h at room temperature
and then acidified with dilute aqueous HCl and extracted with
ether (3 × 300 mL). The combined extracts were washed with
brine (3 × 300 mL), dried over anhydrous MgSO4, and freed
of solvent by rotary evaporation under vacuum, to obtain 10.0
g (yield, 98%) of residual 4-heptanol that was used without
purification in the following step.

Macromolecules, Vol. 37, No. 2, 2004
Mechanism of Autocatalysis 359
(12) Tran, V. H. J . Macromol. Sci., Rev. Macromol. Chem. Phys.
1998, 38, 1 (see also references therein).
(13) Troitskii, B. B.; Troitskaya, L. S. Eur. Polym. J . 1999, 35,
2215.
(14) Rasuvaev, G. A.; Troitskaya, L. S.; Troitskii, B. B. J . Polym.
Sci., Part A-1 1971, 9, 2673.
(15) Troitskii, B. B.; Troitskaya, L. S.; Myakov, V. N.; Lepaev, A.
F. J . Polym. Sci., Polym. Symp. 1973, 42, 1347.
(16) Hjertberg, T.; So¨rvik, E. M. J . Appl. Polym. Sci. 1978, 22,
2415.
(17) Martinsson, E.; Hjertberg, T.; So¨rvik, E. Macromolecules
1988, 21, 136.
(18) Troitskii, B. B.; Troitskaya, L. S. Eur. Polym. J . 1995, 31,
533.
(19) Krusic, P. J .; Wasserman, E.; Parkinson, B. A.; Malone, B.;
Holler, E. R., J r.; Keizer, P. N.; Morton, J . R.; Preston, K. F.
J . Am. Chem. Soc. 1991, 113, 6274.
(30) Gupta, V. P.; St. Pierre, L. E. J . Polym. Sci., Polym. Chem.
Ed. 1973, 11, 1841.
(31) Abell, P. I. Free Radicals 1973, 2, 63.
(32) Doering, W. von E.; Kitagawa, T. J . Am. Chem. Soc. 1991,
113, 4288.
(33) Konovalov, V. V.; Kispert, L. D. J . Chem. Soc., Perkin Trans.
2 1999, 901.
(34) Han, C. C.; Elsenbaumer, R. L. Synth. Met. 1989, 30, 123.
(35) Grant, J . L.; Kramer, V. J .; Ding, R.; Kispert, L. D. J . Am.
Chem. Soc. 1988, 110, 2151.
(36) Khaled, M.; Hadjipetrou, A.; Kispert, L. J . Phys. Chem. 1990,
94, 5164.
(37) Khaled, M.; Hadjipetrou, A.; Kispert, L. D.; Allendoerfer, R.
D. J . Phys. Chem. 1991, 95, 2438.
(38) J eevarajan, J . A.; J eevarajan, A. S.; Kispert, L. D. J . Chem.
Soc., Faraday Trans. 1996, 92, 1757.
(39) J eevarajan, A. S.; Wei, C.-C.; Kispert, L. D. J . Chem. Soc.,
Perkin Trans. 2 1994, 861.
(20) Morton, J . R.; Preston, K. F.; Krusic, P. J .; Hill, S. A.;
Wasserman, E. J . Phys. Chem. 1992, 96, 3576.
(21) Troitskii, B. B.; Troitskaya, L. S.; Yakhnov, A. S.; Lopatin,
M. A.; Novikova, M. A. Eur. Polym. J . 1997, 33, 1587.
(22) Troitskii, B. B.; Troitskaya, L. S.; Yakhnov, A. S.; Novikova,
M. A.; Denisova, V. N.; Cherkasov, V. K.; Bubnov, M. P.
Polym. Degrad. Stab. 1997, 58, 83.
(23) Chien, J . C. W. Polyacetylene Chemistry, Physics, and Mate-
rial Science; Academic Press: Orlando, FL, 1984; Chapter 5
(see also references therein).
(24) Tran, V. H.; Guyot, A.; Nguyen, T. P.; Molinie´, P. Polym.
Degrad. Stab. 1992, 37, 209.
(25) Tran, V. H.; Garrigues, C.; Nguyen, T. P.; Molinie´, P. Polym.
Degrad. Stab. 1993, 42, 189.
(26) Kerr, J . A.; Stocker, D. W. In CRC Handbook of Chemistry
and Physics, 83rd ed.; Lide, D. R., Ed.; CRC Press: Boca
Raton, FL, 2002; p 9-52.
(27) Starnes, W. H., J r.; Plitz, I. M. Macromolecules 1976, 9, 633.
Correction: Macromolecules 1976, 9, 878.
(28) Walling, C. Free Radicals in Solution; Wiley & Sons: New
York, 1957; p 345.
(29) Nolan, K. P.; Shapiro, J . S. J . Polym. Sci., Polym. Symp. 1976,
55, 201.
(40) Knoll, K.; Schrock, R. R. J . Am. Chem. Soc. 1989, 111, 7989.
(41) Doering, W. von E.; Sarma, K. J . Am. Chem. Soc. 1992, 114,
6037.
(42) Starnes, W. H., J r.; Edelson, D. Macromolecules 1979, 12,
797.
(43) Ishiwatari, M. J . Anal. Appl. Pyrolysis 1980, 2, 153.
(44) Byers, J . J . Org. Chem. 1983, 48, 1515.
(45) Ohkatsu, Y.; Matsuura, T.; Yamato, M. Polym. Degrad. Stab.
2003, 81, 151.
(46) Ingold, K. U. Free Radicals 1973, 1, 37.
(47) Walling, C. Free Radicals in Solution; Wiley & Sons: New
York, 1957; Chapter 9.
(48) Ueda, A.; Nagai, S. Polymer Handbook, 4th ed.; Brandrup,
J ., Immergut, E. H., Grulke, E. A., Abe, A., Block, D. R., Eds.;
Wiley-Interscience: New York, 1999; p II-97.
(49) van Hoang, T.; Guyot, A. Polym. Degrad. Stab. 1988, 21, 165
(see also references therein).
(50) Haddon, R. C.; Starnes, W. H., J r. Adv. Chem. Ser. 1978, 169,
333.
(51) Zuoyun, H.; Xingzhou, H.; Gang, S. Polym. Degrad. Stab.
1989, 24, 127.
(52) van Hoang, T.; Guyot, A. Polym. Degrad. Stab. 1991, 32, 93.
MA0352835