-
ELSEVIER 0141-3910(95)00064-X Printed in Northern Ireland. All
rights reserved
0141-3910/95/$09.50
A studv of the aroducts of PVC thermal
Polymer Degradation and Stability 49 (1995) 181-191 0 1995
Elsevier Science Limited
Ian C. McNeil& Livia Memetea & William J. Cole Polymer
Research, Chemistry Department, University of Glasgow, Glasgow G12
8QQ, UK
(Received 3 January 1995; accepted 3 February 1995)
PVC thermal degradation in vacuum up to 500C has been followed
by recording the relative rate of volatile product formation by
thermal volatilisa- tion analysis while monitoring by mass
spectrometry the formation of the main products: HCl, aromatic and
aliphatic hydrocarbons, and non-condensable gases (CH,, HZ). The
material balance after pyrolysis has been evaluated. The liquid
fraction collected during pyrolysis was analysed by GC-MS and its
composition determined by the integration of the ion current under
the peaks due to different compounds. After HCl (53% of the PVC
sample), the tar is the major fraction (24%). The liquid fraction
(of which 80% is benzene) accounts for 7% of the original polymer.
The other fractions are the char (9.5%) and gas fraction (6.6%).
10% of the Cl remained trapped in the polymer until higher
degradation temperatures giving rise then to the chlorinated
compounds which account for 1.75% of the liquid fraction and 0.14%
of the polymer.
PVC shows two stages of degradation: during the first stage,
between 200 and 360C mainly HCl and benzene and very little alkyl
aromatic or condensed ring aromatic hydrocarbons are formed. It was
evaluated that 15% of the polyene generates benzene, the main part
accumulating in the polymer and being active in intermolecular and
intramolecular condensation reactions by which cyclohexene and
cyclohexadiene rings embedded in an aliphatic matrix are formed.
Alkyl aromatic and condensed ring aromatic hydrocarbons are formed
in the second stage of degradation, between 360 and 500C when very
little HCl and benzene are formed. In this stage the polymeric
network formed by polyene condensation breaks down in the process
of aromatisation of the above C, rings. The mechanism of benzene
formation at different temperatures was considered.
1 INTRODUCTION
The present investigation is a study of PVC pyrolysis which has
relevance to waste disposal by incineration and the environmental
problem it poses. The work is part of a larger project which aims
at: (i) identification and quantitative determination of the
pyrolysis products of PVC as a reference material and of its
formulations with additives such as plasticisers, stabilisers and
flame-retardants; (ii) better understanding of the pyrolysis
mechanism; and (iii) intervention into the pyrolysis mechanism in
order to minimize toxic product formation, offering thereby a less
polluting solution to disposal of PVC scrap by incineration.
The present paper, the second in the series, is concerned with
the separation, identification and quantitative determination of
the pyrolysis products of PVC.
2 EXPERIMENTAL
The PVC used was a typical industrial, suspension grade sample
from European Vinyls Corporation with molecular weight (A4,) 40,000
as determined by GPC. The pyrolysis was conducted under continuous
evacuation (initial vacuum 10v5 mm Hg) at a heating rate of lOC/min
up to SWC, when a hold time of 20 min was allowed. PVC samples
(powder) of
-
182 I. C. McNeil1 et al.
20-200 mg, spread as a thin layer at the bottom of a silica
cell, were subjected to pyrolysis. The pyrolysis products were
collected, separated and analysed according to the procedures of
thermal volatilisation analysis (TVA).. An account of this
technique has been presented in the first paper of the series.
The products evolved through pyrolysis were collected in
fractions: the cold ring or tar fraction, the liquid fraction, the
residue or char, and the gas fraction. The tar contains condensed
ring aromatic hydrocarbons (MW 128-250) and scission fragments of
the network formed through the crosslinking of polyene. The tar has
been qualitatively analysed in the earlier study. The liquid
fraction is made up of compounds which condense at room
temperature. The gas fraction consists of HCl together with noncon-
densable gases (CH,, H,) and C,-C, hydrocar- bons. The fractions
(apart from the gases) were weighed for material balance after
pyrolysis. The gases were determined by difference. HCl was
determined in a separate experiment in which it was retained on a
short column containing CaO mixed with glass wool which allowed for
the free passage of the other gases.
The liquid fraction containing products up to a molecular weight
of 150 was analysed by CC-MS using a Hewlett-Packard 5971 mass
selective detector interfaced to a 5890 series 11 gas chromatograph
and computer (Vectra QS/16s). Separations were effected with an HP1
fused silica capillary column (12.5m X 0.2mm X 0.33 pm)
temperature-programmed from 50C (5 min hold) to 220C (1 min hold)
at SC/min. The Grob-type injector (225C) was operated in split mode
(5O:l) and the He carrier gas flow rate was 1 ml/min. Mass spectra
(70 eV) were recorded in continuous scanning mode. The
concentration of the components in the analysed fraction was
determined by the integration of the ion current under the
peaks.
In other experiments, the evolution of the volatile pyrolysis
products was monitored during heating in vacuum to 500C by means of
Pirani gauges using the conventional four parallel line TVA
experimental arrangement. The pyrolysis products were then
condensed on a liquid nitrogen trap. Another Pirani gauge placed
after a liquid nitrogen trap recorded the non- condensable gases
(H, and CH, in this case). The Pirani output was recorded as a
function of time or temperature. For temperature measurement,
two thermocouples were used, one placed in the combustion tube
just above the sample, the other in the oven, in order to assess
the temperature lag between the sample and the oven, which is a
common phenomenon in dynamic heating. Typically, at the beginning
of the experiment the temperature lag was lo-15C, it decreased to
5-6C at 200C and then was steadily reduced to I-2C towards 500C as
shown in the corres- ponding diagrams. The Pirani recording gives a
measure of the relative rate of evolution of volatile products and
is useful in identifying the degradation stages of the polymer,
which together with the identification of the nature of the
products can provide valuable information on the degradation
mechanism.
A bleed to a Leda-Mass quadrupole mass spectrometer placed on
line with the TVA system enabled the mass spectra of the products
to be recorded in continuous scan mode (3.3 scans/min). The stored
MS scans were subse- quently further analysed for the
identification of the evolved products. The lines in the mass
spectra produced by ions of m/e which are characteristic of certain
products and which are not subject to interference from other
products were identified and their intensity plotted against time
(temperature) using Microsoft Excel. The intensity of the above
lines is expressed as partial pressure produced in the mass
spectrometer by ions of specific m/e.
3 RESULTS
3.1 Main product frictions
Table 1 presents the material balance of the products obtained
through the pyrolysis of PVC in vacuum up to 500C. As can be seen,
apart from HCl (53%), the cold ring products form the major
fraction (24%). Of the polymer remaining
Table 1. Material balance after PVC pyrolysis
NO FrXiKCl %
I MCI (2 6 2 Cold (tar) fradm nng 24 3 3 Lquld fncttal 7.0 4 Gas
fmdim (apart fim HCI) 66 5 Char 9s
-
A study of the products of PVC thermal degradation 183
after HCl emission, 50% degrades to tar, 20% to char, 30% to gas
fraction and 30% to liquid fraction. The amount of Cl in the
evolved HCl compared to the theoretical Cl content of the PVC shows
that 10% of the original Cl remains in the polymer to higher
temperatures than those involved in dehydrochlorination.
Table 2 presents the composition of the liquid fraction obtained
through the pyrolysis of PVC, expressed as main classes of
compounds. The following features can be noted.
Aromatic hydrocarbons are the main class (88%), of which the
major part is benzene (80%). Aliphatic hydrocarbons account for
5.2% of theliquid fraction. These have been identified in the
previous paper. Chlorinated hydrocarbons make up 1.75% of the
liquid fraction and correspond to 0.14% of the original
polymer.
Table 3 presents the main aromatic con- stituents of the liquid
fraction obtained through the pyrolysis of PVC alone. Benzene
accounts for 80% of the mixture, as already shown (and for 5.6% of
the PVC). It is followed by toluene (1*8%), o&o-xylene (l-3%),
ethylbenzene (1%) and 1,2 methylethylbenzene (l%), the percen- tage
being relative to the liquid fraction. A surprising feature is that
among the alkyl benzenes, the ortho-derivatives are formed in a
higher concentration than metu- and paru- derivatives: ortho-xylene
accounts for l-3%, me&- and puru-xylene for O-6%,
1,2-methylethyl benzene for l%, while 1,3- plus 1,4- derivatives
make up 0.3%. This trend reflects a higher probability for the
ortho-derivatives to be formed.
Table 4 presents the constituents of the class of chlorinated
hydrocarbons produced through the pyrolysis of PVC. The main
chlorinated com- pounds are chlorinated alkenes and chlorinated
aromatics, all regarded as hazardous to health.
Table 2. The composition of the liquid fraction (classes of
compounds) collected during the pyrolysis
of the PVC sample
c- % 96
Bmrsns 19.14 Alkyl amna,c 5.22
Alkenyl .mmat,c 0 83
Cc.l&nd rmg alonlatic 2.40
T&al .-t,c 88 20 AlhO 2.47 C!yClWlkNWS 0 80 Alkmes I .52
cyckmkmm 0.40 Total aliphraic 5 19 Chlorinated hydmxbms I IS others
4.87 Total IOOGQ
Table 3. Quantitative determination of the main aromatic
products in the liquid fraction
No PWdW %
1 796
/ \ 5 -o- 06
-
3.2 Thermal volatilisation analysis
The evolution of the volatile compounds constituting fractions
l-4 of Table 1 was followed by TVA and the results are presented in
Fig. 1, which shows the response of five Pirani gauges and two
thermocouple outputs as a function of temperature. The evolution of
non-condensable
by the response placed after the
gases (HZ and U-L,) is indicated in the trace for the Pirani
gauge -196C trap (curve 5).
T&le 4. Chlorinated hydrocarbons pyrolysis of PVC
NO Crnpamd % NO
Cl
16 +
01 25
Cl CI
20 +,a
SO1 26
Cl
21 II
02 27
Cl
22 ?I
01 26
23 /t/ o-2
Cl 24 w 02
formed in the
%
o-15
01
02
016
02
036
-
184 1. C. McNeil1 et al.
500 5
01 : : 04 120 1.30 200 240 230 320 300 4w 440 480
Temperature (C)
Fig. 1. TVA diagram recorded as Pirani gauge output (relative
rate) vs temperature for PVC thermal degradation. Key to traces:
1-4, volatile products (condensable) recorded with four Pirani
gauges placed on parallel lines: 5, noncondensable gases (Hz, CH,);
6,7, thermocouples output.
Figure 1 shows that the volatile compounds are evolved in two
degradation stages: between 200 and 360C and between 360 and 500C.
These have already been identified by TVA. HCl and benzene are the
major degradation products in the first stage. During the second
stage, breakdown products of the crosslinked polyene are formed
which were qualitatively analysed as components of the tar, liquid
and gas fractions. The amounts of volatile products at the second
stage are smaller, as can be seen in Fig. 1. A small amount of
non-condensable gases is formed
(4 1 WE.03 P.w3m
O.OOE-03
7 WE.00
g o.ocEm
E 5.cQE-m
j 4.OOE-09
3OoE-08
2wE-05
1OOE-a
O.OOE+W
30
E
@I
0 50 loo 150 200 250 300 350 4Ou 450 5OQ
Tempemtun (C)
I OOE-08
3OOE-00
2wE-09
1 .JOEJm 82
OOOE+OO _s
0 50 KM 150 200 250 300 350 UXI 450 5al
Tsmperalum (C)
Fig. 2. Ion monitoring during PVC thermal degradation: (a)
benzene (78) HCI (36-38); (b) toluene (91,92).
towards the end of the second stage and very little if any, in
the first stage.
3.3 Evolution of individual products
3.3.1 Hydrogen chloride and benzene The evolution of the
individual products with temperature was followed by plotting the
intensity of the lines produced in the mass spectra by their
characteristic ions vs tempera- ture. Figure 2(a) shows the
evolution of benzene (m/e 78) and HCl (m/e 36, 38). It is evident
that these are evolved mainly during the first degradation stage
and only to a minor extent during the second stage. However, Fig.
2(a) shows that although HCl is produced at the same time as
benzene (at 200C) it appears that the HCI curve (m/e 36) initially
lags behind the benzene curve, which was a reproducible trend in
all the present experiments. Although the delay is expressed in
terms of temperature, it is believed that time and not temperature
governs the phenomenon, which may possibly anomaly due to initial
HCl adsorption glassware. The effect is being investigated.
be an on the further
3.3.2 Toluene and other aromatics Toluene formation as a
function of temperature was followed using its ions at m/e 91 and
92 (Fig. 2(b)). While in the first degradation stage there are no
other compounds contributing to the 91 line, in the second stage
this is no longer valid. Compounds such as xylene, ethylbenzene,
propylbenzene, butylbenzene and their isomers are the main sources
of the 91 line in the second stage. However, the 92 line can be
safely regarded as characteristic only of toluene, since other
substances giving the same ion (especially C,-alkylbenzenes) are
formed in very low concentrations. Figure 2(b) shows that a small
part of the toluene is formed in the same process as benzene. The
major part of the toluene is formed in a different process in the
second stage. This conclusion has been also verified by the GC
analysis of the liquid fractions collected separ- ately for each
degradation stage. Styrene formation was followed by its molecular
ion (104) and so was ethylbenzene and the isomers of xylene (106),
all presented in Fig. 3(a). They are all formed in a minor amount
during the first stage and almost entirely during the second
stage.
-
A study of the products of PVC thermal degradation 185
(4 3.wE-00 3.OcE-w
2.sOE-OQ
F ~2.OE-00
j 1.sOE-09
l.WE-OQ
0 s 10 1s m 25 30 35 40 45 50 TemmmlumrC)x0.1
.m .
: ?I
I ;
0 5 lo 15 m 25 30 35 40 45 50
Tanpentun CC)xO.l
Fig. 3. Ion monitoring during PVC thermal degradation: (a)
styrene (NM), C3-C5 alkyl benzene (105), ethylbenzene and
xylene isomers (106); (b) indene (115, 116).
Figure 3(a) shows also the intensity for evolution of the ion
with m/e 105 characteristic of C,-C, alkyl benzenes which are
present in this system. These products are also formed mainly in
the
(4 ZOOE-OQ- .d
lmlE-o~- .
:. .
l.eE-OS ..
1.4E-00 8 ;
0
- 12oE-O~-
S I l.OEoo ..
1 8.OOE-10 ..
' cs 117
. -a .
e.OE-10
4.Oms10 : z.OOE-10
_*pQ?,,,
__...yy : , '4Al8
O.OOEIoo L, .: . 118
0 5 lo 15 m 25 30 35 40 45 50 Temmrmum~~IxO.1
(b) T o.ooE-10
moE-10
7oOE.10 I
0 s 10 is m 25 30 35 40 45 50 T.mpemlum~C)xO.~
Fig. 4. Ion monitoring during PVC thermal degradation: (a)
indane, methylindane and C,-alkenylbenzenes (117, 118);
(b) naphthalene (128).
second degradation stage. The same observation is valid for
indene (ions 115, 116, Fig. 3(b)) and also for indane and methyl
indanes (structures 15 and 16, Table 3) and C,-alkenylbenzenes
(structures 7 and 8, Table 3) having contributions to the ions 117
and 118, the formation of which during degradation is presented in
Fig. 4(a). Naphthalene (128, Fig. 4(b)) is formed only during the
second degradation stage.
In the case of indene, indane (and other compounds) and
naphthalene, the curves are very noisy due to the low amount in
which these compounds are formed. In Figs 3(b), 4(a) and 4(b),
which show their formation as a function of temperature, the
experimental data are displayed in two curves representing the
upper and lower levels within which the experimental results are
scattered. Nevertheless, the trend in the intensity variation is
clear.
3.3.3 Aliphatic hydrocarbons Figures 5(a) and 5(b) show
aliphatic hydrocarbons through
the evolution of their lines at m/e
67 (cyclopentene), 55-57 (C,,-C,, alkenes) and 57 (C,,,C,,
alkanes). As can be seen, they all have small peaks situated
towards the higher
1 .la5oo I\ _,_ l o.OOE*w : _ ,_--,
0 SO 100 150 200 2x) 3aO 350 4Oa 450 SaJ
50 100150 200 250 MO 350 400 *50 500
T-W)
Fig. 5. Ion monitoring during PVC thermal degradation: (a)
cyclopentene (67); (b) C,0-C,3 alkenes (55-57) and C,,-C,,
alkanes (57).
-
186 I. C. McNeil1 et al.
5mE-10-~ 2
,j>
0ooE.M).
0 so 1w 150 xl0 25n c?m 350 400 450 !m Tempermum CC)
Fig. 6. Ion monitoring during PVC thermal degradation: methane
(16) and H, (2).
temperatures of the first degradation stage, between 300 and
37oC, which develop into important peaks in the second stage.
3.3.4 Non-condensable gases Figure 6 illustrates the evolution
of HZ (m/e 2) and CH, (m/e 16) with temperature. The amount of H,
might be underestimated due to the low sensitivity of the
quadrupole mass spectrometer for m/e under 10 amu. However, the
general tendency of H, evolution can be observed. H, and CH, are
formed only in minor amount, if any, in the first degradation
stage. They are evolved in detectable amounts in the second stage,
at temperatures above 400C.
4 DISCUSSION
4.1 Aspects of PVC degradation from previous work
It is known that the thermal degradation of PVC begins at a
relatively low temperature, soon after T. (8OoC).4 The initiation
of the dehydrochlorina- tion (DHC) was correlated with the labile
sites-defects-of the chemical structure: the allylic chlorine atom
of the internal unsaturation, the tertiary chlorine atom of the
butyl branches and the head-to-head units formed in the course of
synthesis. The head-to-head structures have an increased tendency
of splitting Cl, but do not propagate the DHC more readily than
PVC, as shown by studies on model compounds prepared by
chlorinating cis-l,4-polybutadiene.s The con- centration of the
above defects was evaluated as Oel-O-2/1000, 0*5-l/1000, and under
0.2/1000 vinyl chloride (VC) units, respectively.h Although the
rate of degradation has been clearly correlated with the
concentration of allylic and tertiary Cl, there is no definite
evidence on the
nature of the initiation act at low temperatures (SO-120C).
At the same time, it is possibile that physical defects
contribute to DHC. A certain confor- mation of syndiotactic
segments of chains in which the Cl atoms are tram relative to a
double bond makes the DHC very easy, while the presence of a Cl
atom gauche to the unsaturation favours the termination of polyene
growth. Syndiotactic chains of up to 13 VC units are present in
commercial suspension PVC samples, as part of very small imperfect
crystallites dispersed in the polymer matrix. Overall, the
conformation responsible has a concentration of the same order of
magnitude as the defects of the chemical structure and it is
likewise dispersed.
At high temperatures, the initiation of DHC takes place in the
whole mass of the polymer by the random scission of the secondary
chlorine atoms. At intermediate temperatures, chain ends can have
an important contribution in DHC, if proved to have labile
structure. It is largely accepted in the literature that chain ends
contribute to the formation of mononuclear hydrocarbons, as
reviewed below.
Benzene formation begins at low temperatures (soon after T,) as
soon as polyenes of suitable length are formed through DHC.
Polyenes are very active in crosslinking and molecular enlargement
has been observed at a very low degree of DHC (0.5% at lSOC).
Several mechanisms for benzene formation in the early stages of
thermal degradation have been advanced: the cyclisation of a triene
radical situated at a chain end and formed through DHC followed by
the scission of the macro- molecular chain, the thermal
condensation of a triene within the chain, or the reaction of a
triene in the triplet state easily achievable by thermal
excitation. The last hypothesis is still very appealing though it
did not gain experimen- tal support with the passing of the
years.
It is considered that at higher temperatures benzene and alkyl
benzenes are formed by an intramolecular process (cyclisation)
which starts at the chain ends of the macromolecules, while the
inner part of the macromolecules is bound into a crosslinked
network. The network is formed through: (i) Diels-Alder
condensation of double bonds belonging to different chains
generating cyclohexene rings and within the chain generating
cyclohexadiene rings, which then become aromatic: and (ii)
crosslinking
-
A study of the products of PVC thermal degradation 187
through free radical attack on unsaturation. The extent to which
either of these processes takes place is not known, but both
crosslinked and conjugated structures have been identified.13 The
Diels-Alder condensation, however, is considered a major process,
for the use of a dienophile strongly reduces PVC crosslinking.14
The present picture of PVC as it degrades at temperatures above
250C is that of a crosslinked internal core to which loose branches
are attached. The branches or chain ends produce benzene and other
aromatic hydrocarbons by the cyclisation of the free ends.
The breakdown of the crosslinked network produces condensed ring
aromatic hydrocarbons with or without aliphatic substituents and
aliphatic hydrocarbons.
4.2 The chain ends in suspension PVC
In the framework of the above picture and since we are concerned
with the degradation at all temperature levels it is worth
examining the structure of chain ends. The most probable structure
of the PVC macromolecule obtained by suspension has been the object
of an excellent review.6
In suspension polymerisation, the major termination path is by
chain transfer to the monomer. This is a complex process consisting
of several reactions by which one Cl atom is expulsed and chain
ends of the type a (1-chloro, 2-alkene) - Scheme 1 - are created.
Chain end a is referred to as pseudoterminal unsaturation. The Cl
atom reinitiates the polymerisation producing chain ends of type b
(1,2- dichloroalkane). This is a very efficient process,
reinitiation by Cl becoming the main way of initiation: four-five
polymer molecules per initiator residue are formed. Towards very
high conversions, under conditions of monomer starvation, the
termination is by chain transfer to
? CICHZ - y - CH2 - ~t+--C, - C - CH2 - CHCI - CH2 - C2C, Cl Cl
L Hz b L C
1
6
7 a CH2CI
Scheme 1. The structure of the suspension PVC macromolecule.
the polymer and polymer branching starts. The most frequent
branches are C, (type c, Scheme 1) which shows that backbiting is
the most probable transfer process. The termination at high
conversions can also take place by H abstraction from any chain,
hence, a number of normal (1,3-dichloroalkane) chain ends, -CH(
Cl)-CH,- CH,Cl, are formed.
Let us consider the C, branches among chain ends. On average, in
suspension PVC the concentration of chain ends is: 0.8-0.9 chain
ends of the type b, 0.7 chain ends of type a and O-2 normal chain
ends per molecule, 0.20-0.25 initiator fragments units and 0.5-l C,
branches per 1000 VC units. Considering the molecular weight of the
present PVC sample, 40,000, one can calculate the proportion of
chain ends (Table 5). As already known6 and presented here in Table
5, structures a and b are the most frequent chain ends. Another
observation is that C, branches (hence tertiary Cl atoms) are quite
numerous.
Considering the structure of the PVC molecule with three chain
ends as in Scheme 1 (including the C, branches), the MW of our PVC
sample and the amount of benzene being formed through pyrolysis
(5.6%), one can estimate that 9.5 molecules of benzene are formed
per chain end.
Little is known about the tendency to DHC of chain ends of a
particular structure. Table 6 shows the activation energy (E,) for
HCl elimination of model compounds. Long chains ending in the
structures shown in Table 6 are expected to have lower E, than the
mic- romolecular compounds in the same table. As one can see,
1-chloro, 2-alkene chain ends have the highest tendency to DHC,
followed by 1,2-dichloroalkane and finally by normal chain- ends.
For comparison, Table 6 also includes the E, for the elimination of
an allylic Cl associated with a terminal double bond (compound No.
4) which, if substituted at both ends (compound No. 5) is a model
for internal unsaturation and is
Table 5. The percentage of chain ends of different structures in
a typical suspension PVC sample
Type
b
StNCtUW %
CICHrCHCHr 40-32 Cl
a ClCHrCH=CHCH>- m-28 c Tertiary Cl at G branches 14-2s
lllfutor terminated 8-7 s NW& CICHrCHr 8-7 5
-
Table 6. Activation energies for HCI eliminations from model
compounds [Ref. 151
Model wmpound
H,C-CHJTiKI HC-CHCHKI
Cl H,CCH=CH-CH>CI H,C-CHCH=CHz
Cl -HK-CH-CH=CH-
Cl
E. (kalhole)
55 I 54 9(-ZHCI)
52 2 48 5
-
A study of the products of PVC thermal degradation 189
The macroradical I formed through tertiary Cl scission (reaction
1) would usually split in position d with the formation of
macroradical II and macromolecule III (reaction 2). Macroradical II
has the structure of the growing chain in the polymerisation and it
is not known to promote zip dehydrochlorination. Its main reaction
route in the absence of monomer is likely to be H atom abstraction
from another chain or from its own chain (backbiting at the fifth C
atom). Hence, it is improbable that radical II has a direct role in
benzene formation which has to be sought in another reaction.
The Cl atom expelled in reaction (1) will give a random attack
on any macromolecule. The affinity of Cl atom for the CH, group6
would lead to H abstraction through which radical IV is formed.
Radical IV can split (reaction 3) leading to radical V which has a
structure more favourable to zip DHC. The electron withdraw- ing
effect of Cl polarizes the bonds in radical V as shown in Scheme 2
favouring zip DHC by which radical VI results after three double
bonds have been fomed (reaction 4). Radical VI has a resonance form
VII. It is proposed that radical VII attacks the double bond at the
sixth C atom closing a cyclohexadiene ring and forming radical VIII
(reaction 6) which can then split the bond to the ring as this bond
is weakened by the withdrawing effect exerted by the Cl atom on the
radical. The scission is followed by aromatisation and H atom
rearrangement so that no H transfer from other molecules is needed.
Benzene formation has been shown to be intramolecular not only with
respect to C, but also to H. Aromatic ring formation takes place
after scission and does not infringe the orbital symmetry
interdiction which operates with the aromatisation of substituted
rings. Moreover, reaction (6) regenerates radical V which can
resume the cycle producing several benzene molecules per radical
chain end.
The alternative formation of toluene in reaction (6) from
radical VIII would imply the scission of the bond between the C
atom bearing the radical and that bearing the Cl atom. Though this
is possible, it is not highly probable due to the withdrawing
effect of the Cl which shortens and reinforces the bond. The
repeated formation of benzene at the radical chain-end apparently
prevents the formation of polyenes containing more than three
double bonds in this particular place. Consequently, the formation
of condensed
rings like naphthalene, indene etc., from radical chain ends is
of lower probability, a fact which has the support of our
experimental data since very little of these compounds are formed
in the same stage as benzene. Our findings do not agree with the
conclusions of a previous study in which it was conclued that
condensed ring aromatic hydrocarbons and alkyl aromatic
hydrocarbons are formed by the same mechanism as benzene.
The C, branch has a small, but finite probability of being
eliminated by the scission of bond e. The chlorobutene isomers
identified in Table 4 (Nos 19 and 20) in low concentration can be
formed in this process (reactions 7 and 8). It is not clear at this
stage whether the saturated homologue (chlorobutane, No. 21 in
Table 4) is formed in the same process or not.
The formation of macromolecule structure III with vinylidene
groups close to chain ends could explain the crosslinking observed
at very low degrees of DHC (O-3%)8 through Cl atom attack on this
pendant unsaturation.
The formation of benzene at higher tempera- tures through DHC of
loose chain ends followed by cyclisation starts with 1-chloro,
2-alkene type chain ends (pseudoterminal un- saturation). Scheme 3
explains the formation of benzene and other aromatic hydrocarbons
from chain-ends of this type.
The mechanism proposed in Scheme 3 accounts for benzene being
formed through repetitive scission of the chain ends of pseudoter-
minal unsaturation and for the main identified alkyl aromatics
being formed at smaller yields.
In Scheme 3, the 1-chloro, 2-alkene chain end is polarised as
shown, which makes the dissociation of Cl atom easier (reaction 9).
Radical XI is formed in which the bonds are polarised by the
electron withdrawing Cl atoms and the tendency to conjugation
between the double bond and the radical. This will render DHC
easier by which radical XII (reaction 10) is formed. Radical XII
has an alternative resonance form XIII which would be expected to
be somewhat more stable. Both radicals, XII and XIII, can explain
the formation of aromatic hydrocarbons. This process will be
illustrated here with radical XIII which can split in positions a,
b and c. Though route a is still most probable due to the maximum
gain in free energy for benzene formation, routes b (leading to
toluene) and c (leading to styrene, ethylbenzene and
o&o-xylene) have increased probability com-
-
I. C. McNeil1 et al.
- CH - CH* - CH - ;H* - CHtCH2-CH = CH - &I2 + a I I Cl Cl I
Cl
XI - 3HCI
-..-~y2- i - - CH CH CH-CH-CH.CH_-CH*
XII 1 -~-c*-cH-cH=CH-cH.CH-cH=CH~
XIII
c ba
__t
!I % $ t dH CH=CH-CH.CH-CHCCH*
a
(-: cCH - a$- CH = CH - iH2 + C&
Cl 1 tic,
XI
b - y - CH2 + CH3- C&
c,
Scheme 3. Aromatic hydrocarbon formation from pseudo-
unsaturated chain ends.
pared to the formation of the same compounds from macroradical
VIII in Scheme 2. Scheme 3 is also consistent with benzene being
generated repeatedly through the regeneration of radical XI
(reaction 12).
The formation of me&- and para- isomers of xylene cannot be
explained as arising from radicals formed at marginal, loose
branches of the structures in Scheme 1. Their formation can only be
explained by the scission of the crosslinked network containing
benzenoid str- uctures formed by Diels-Alder condensation.
5 CONCLUSIONS
PVC thermal degradation in vacuum up to 500C has been studied by
following the relative rate of volatile product evolution in
thermal volatilisa- tion analysis experiments and by monitoring the
formation of the main products by mass spectrometry. The material
balance after pyroly- sis combined with the GC-MS analysis of the
liquid fraction shows that HCl is the major fraction (53% of the
polymer), followed by tar (24%), char (9.5%), liquid fraction (7%)
and gas fraction (6.6%). 10% of the Cl atoms remain in the polymer
after HCl evolution ceases. The quantitative evaluation of the
products in the liquid fraction shows that benzene accounts for 80%
of the liquid fraction and for 5.6% of the
polymer. The chlorinated hydrocarbons (mainly chlorinated
alkenes and chlorinated aromatics) account for 1.75% of the liquid
fraction and for 0.14% of the polymer.
PVC shows two degradation stages. During the first one, between
200 and 360C HCl and benzene are evolved but very little alkyl
aromatic hydrocarbons (toluene, xylene isomers, ethylben- zene,
etc.) or condensed ring aromatics (naph- thalene, indene, indane,
etc.). The molar ratio of benzene and HCl shows that only 15% of
the double bonds produced through dehydroch- lorination generate
benzene. Most of the double bonds accumulate in the polymer and are
active in crosslinking through intramolecular Diels- Alder
condensation and through free radical attack. As a result, a
crosslinked network is created which contains cyclohexene and
cyclohe- xadiene rings embedded in an aliphatic matrix. In the
second degradation stage (360-5ooC), the aromatisation of the above
rings takes place with much scission. As already suggested in the
literature, the aromatisation of the substituted rings cannot occur
if it necessitates orbital overlapping of the substituted C atoms,
due to orbital symmetry interdictions. Hence scission is a
prerequisite for aromatisation. Scission gener- ates an important
amount of alkyl aromatic and condensed ring aromatic hydrocarbons,
as well as C,-C,, aliphatic hydrocarbons and a small amount of
hydrogen.
The mechanism of benzene formation thro- ughout the temperature
range has been con- sidered beginning with scission at the tertiary
Cl (the most abundant labile site) and continuing with the
formation of benzene through the cyclisation of chain ends, of
which the pseudoter- minal unsaturation is the most labile
structure to dehydrochlorination.
ACKNOWLEDGEMENT
Support from SERC for the work reported is acknowledged with
thanks.
REFERENCES
1. McNeill, I.C. & Memetea, L., Polym. Deg. Stab., 43
(1994).
2. McNeil& I.C., Europ. Polym. J, 6 (1970) 373.
-
A study of the products of PVC thermal degradation 191
3. McNeill, I.C., Ackerman, L, Gupta, S.N., Zulfiqar, M. &
Zulfiqar, S., J. Polyrn. Sci., 151 (1977) 2381.
4. Palma, G. & Carenza, M.J., J Appl Polym. Sci., 14 (1970)
1737.
5. Crawley, S. & McNeil& I.C., J. Polym. Sci., 16 (1978)
2593.
6. Hjeertberg, T. & Sorvik, E.M., In Degradation and
Stnbilisation of PVC, ed. E.D. Owen, Elsevier, London, 1984, p.21
and references therein.
7. Behnish, J. & Zimmermann, H., Makromol. Chem., 190 (1989)
2347.
8. Hjertberg, T., Martinsson, E. & Sorvik, E.,
Macromolecules, 21 (1988) 603 and references therein.
9. OMara, M.M., Pure Appl. Chem., 49 (1977) 649. 10. Starnes,
W.H., Jr & Edelson, D., Macromolecules 12
(1979) 797.
11. Rasuvaev, G.A., Troitskaya, L.S. & Troitskii, B.B., J
Polym. Sci., 9 (1971) 2673.
12. Winslow, F.H., Baker, W.O. & Yager, W.A., Proc. ConjI
Carbon, 1956, p.93.
13. Grassie, N. & Scott, G., Polymer Degradation and
Stubilisation, Cambridge University Press, Cambridge, 1985, and
references therein.
14. Kelen, T., Bela, I., Nagy, T.T., Turcsanyi, B., Tudos, F.
& Kennedy, J.P., Polym. Bull. (Berlin), 1 (1978) 79.
15. Esser, K.W. & Cocks, A.T., In The Chemistry of the
Carbon-Halogen Bond, ed. S. Patai, John Wiley, New York, 1973,
p.677.
16. Troitskaya, V.N., Myakov, B.B. & Razuvaev, G.A.,
Vysokomol. Soedin., 9A (1967) 2119.
17. Lattimer, R.P. & Kroenke, W.H., J. Appl. Polym. Sci., 27
(1982) 135.5.