(Sici) 1097 4628 (19990808) 73 6 1069 Aid App26 3.0

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Kinetic Analysis of the Thermal Degradation of

PVC Plastisols

A. JIMENEZ,1 J. LOPEZ,2 L. TORRE,3 J. M. KENNY3


1
University of Alicante, Department of Analytical Chemistry, Alicante, Spain

2
Institute of Toys Research (AIJU), Ibi, Alicante, Spain

3
University of Perugia, Institute for Chemical Technologies, Terni, Italy

Received 4 September 1998; accepted 12 January 1999

ABSTRACT: The thermal degradation of plasticized polyvinyl chloride (plastisol) is


reported here. Plastisols used in the present work were prepared with the plasticizer
diethylhexyl phthalate in different proportions. Thermogravimetric analysis has been
applied to study the behavior of plastisols at high temperatures and to evaluate their
degradation kinetics. Several tests were carried out at different heating rates and the
variation of the degree of reaction with time and temperature was calculated. The
influence of the heating rate in dynamic measurements (5– 40°C/min) on kinetic pa-
rameters, such as activation energies and reaction orders, has also been studied. These
parameters were calculated from dynamic thermogravimetric analysis tests using
Friedman analysis and a kinetic model for the degradation of poly(vinyl chloride) and
plastisols has been then developed. The obtained model was able to simulate the
thermal degradation process of plastisols in dynamic conditions and was used to
evaluate the effects of additives in the degradation. The results of this study can be used
to optimize the concentration of plasticizers and stabilizers in poly(vinyl chloride)
formulations. © 1999 John Wiley & Sons, Inc. J Appl Polym Sci 73: 1069 –1079, 1999

Key words: PVC; plastisols; kinetic analysis; thermal degradation; thermogravim-


etry

INTRODUCTION mately associated with the resin molecules and it


has been shown that these additives not only
The mechanism of poly(vinyl chloride) (PVC) deg- modify the mechanical behavior, but can also
radation and stabilization remains one of the still have large influence on many of the polymer prop-
unsolved problems in polymer science. Further- erties.1 For instance, some recent studies on the
more, since in many PVC applications the poly- influence of different plasticizers in the final prod-
mer is mixed with plasticizers, the study of the ucts show that some additives change substan-
degradation must take into account the effects of tially the plastisol mechanical and electrical prop-
such additives. In plasticized PVC (plastisol), in erties2; moreover they can modify the resin ther-
fact, the molecules of the plasticizer are inti- mal degradation making it faster or slower.3
However, despite of the broad interest in such
studies, systematic investigations on the effect of
Correspondence to: J. M. Kenny.
such additives in PVC thermal degradation have
Journal of Applied Polymer Science, Vol. 73, 1069 –1079 (1999)
© 1999 John Wiley & Sons, Inc. CCC 0021-8995/99/061069-11 only recently started.

1069
1070 JIMENEZ ET AL.

The degradation of polymeric materials has gravimetric analysis (TGA). Then, the degrada-
been studied mainly in relation with the repro- tion kinetics can be studied correlating the weight
cessability of thermoplastics after several pro- loss to time, temperature and a generic degree of
cessing cycles. The particular mechanisms of deg- degradation. Although TGA data may not give
radation are a function of the polymer structure enough information about complex degradation
and can follow different steps because of the com- mechanisms, it can be a very useful tool to ana-
plexity of the polymeric chains. Functional group lyze the overall behavior of the polymer in ex-
transfer, chain unzipping with elimination of treme conditions.
volatiles and the formation of intermediate com- Isothermal or dynamic (constant heating rate)
pounds are the main decomposition mechanisms tests can be used to study the thermal decompo-
that contribute to complicate the development of sition of polymeric materials, such as PVC and
a mechanistic kinetic model of the polymer deg- plastisols, by TGA. With these techniques kinetic
radation process. parameters, such as activation energy and reac-
In particular, the thermal degradation of PVC tion order as well as others such as the tempera-
in a broad range of temperatures is substantially ture at the maximum decomposition rate can be
a two-step process.1 The first step (up to 330°C) calculated. However, these two modes, dynamic
mainly involves the zip-elimination of HCl and and isothermal, present some advantages and
the simultaneous formation of conjugated double drawbacks. The advantages of determining ki-
bond containing sequences (polyenes) in the poly- netic parameters by a dynamic method are
mer chain. This process yields a corrosive gas, mainly based on its fastness and reproducibility.
HCl, and leads to highly reactive polyenes absorb- There is no need of long runs to get the results
ing light in the ultraviolet and visible region. This and it is possible to obtain a lot of information
absorption results in undesired coloration from from a single run. Mathematical models of ther-
yellow to nearly black in some cases and general mal decomposition reactions make possible the
loss of physical properties. Some evaporation of understanding of the whole process and the ob-
plasticizer may be also observed in this first step. tainment of quantitative conclusions from appar-
At higher temperatures, the product also under- ent kinetic parameters.3–9 This is the main rea-
goes a partial carbonization with the release of son why a mathematical model would be helpful
free carbon among other products, such as ben- to get a clear vision of the mechanism of plastisols
zene and other aromatics. thermal degradation, a subject which has raised
On the other hand, the thermal degradation of some interest during the last years.2,10 –14
phthalate plasticizers starts at 200°C. It has been The aims of the present work are the evalua-
suggested that their thermal degradation lead tion of the kinetic parameters as well as the char-
first to the formation of a monoester and olefins acteristic temperatures of the thermal decompo-
by a cis-elimination mechanism, and secondly to sition of plastisols based on PVC resins and plas-
the cleaving of the monoester into phthalic anhy- ticizers under different operating conditions. This
dride and alcohol by a scission mechanism.3 The evaluation will lead to the proposal of a mathe-
study of the kinetic of plastisols degradation is matical model able to describe the thermal deg-
further complicated by the fact that the two indi- radation of plastisols and to get an insight for the
cated processes happen simultaneously. theoretical approach to the reactions involved in
Although degradation reactions can be, as in the degradation mechanism.
this case, very complex, the utilization of a mate-
rial under critical temperature conditions re-
quires, at least, the knowledge of its degradation Degradation Kinetic Models
kinetics that is related both to morphological
changes and to chemical decomposition reactions. In the present work the following symbols will be
Polymer degradation is normally accompanied by used to describe the degradation kinetic models:
different changes in polymer properties, such as E a : apparent activation energy (kJ mol21); A:
increase of viscosity, crosslinking and weight loss. preexponential factor (min21); n and m: apparent
Experimentally, thermal degradation is typically order of reaction; R: gas constant (8.31 J mol21
detected by monitoring weight changes occurring °C21); T: temperature (°C); t: time (min); b: heat-
to a polymer exposed to heat as a function of time ing rate (°C min21); k: rate constant; a: degree of
and temperature, which is the basis of thermo- conversion or fractional mass loss.
THERMAL DEGRADATION OF PVC PLASTISOLS 1071

In degradation studies the degree of conversion to this method, applied to a general kinetic equa-
is typically calculated as tion is as follows:

a5
W0 2 W
W0 2 Wf
(1) lnS Dda
dt
5 ln A 1 ln~f~ a !! 2
Ea
RT
(6)

Where W is the actual weight, W 0 is the initial Thus, a plot of ln(d a /dt) versus 1/T for a con-
weight, and W f is the weight at the end of the TG stant a allows to obtain the activation energy,
test. The basic rate equation generally utilized to reaction order and preexponential factor. It
represent the reaction kinetics is expressed in the should be noticed that the use of the Friedman
following way: method does not require the specification of the
kinetic model (like in eqs. 3 or 4). Any suitable
da kinetic model can be applied with the only condi-
5 kf~ a ! (2) tion that the contributions of temperature and
dt
degree of reaction may be uncoupled as in the
kinetic models described above. Furthermore, in
This equation defines the rate of conversion, this work, the information of the Friedman anal-
d a /dt, as a function of the reactant concentration ysis will be used to calculate the first guess of the
loss and of the temperature dependent rate con- other kinetic parameters of the degradation equa-
stant. tion for plastisols.
The form of the function f( a ) is dependent on An additional problem when long term data
the mechanism of the degradation reaction. Dis- are needed is the test duration. This is the reason
cussions of the form and of the validity of this why some accelerated degradation experiments
approach have been the concern of many investi- at high temperatures can be performed and the
gators.15 In the case of polymer degradation, the long-term behavior can be extrapolated at lower
simplest form of the function f( a ) assumes that temperatures. This procedure may be applied as-
the rate of conversion is proportional to the nth suming a simplified degradation process.18 Some
order of the material concentration: works have been previously carried out in our
research group with the obtainment of results for
f~ a ! 5 ~1 2 a ! n (3) several TGA tests, performed under different con-
ditions. These data are further used to analyze
A more complex form of the kinetic equation the thermal stability and to develop a kinetic
assumes an ,,autocatalytic.. behavior as re- model of the degradation process. The model can
ported for the thermal degradation of PEEK16,17 be also applied to extrapolate the degradation
behavior of the materials studied.18
f~ a ! 5 ~1 2 a ! na m (4)

EXPERIMENTAL
The combination of these equations and the
Arrhenius expression for the temperature depen-
dence of the rate constant gives the following Dynamic thermal degradation measurements
relationship have been performed using a Mettler TG-50 An-
alyzer coupled to a Mettler TC11-TA microproces-

S D
sor. Tests were performed in dynamic mode at
da A Ea different heating rates in order to cover a wide
5 exp 2 dT (5)
f~ a ! b RT range of thermal conditions. Temperature pro-
grams were run from 30 to 900°C at different
This is the fundamental expression for the de- heating rates between 5 and 40°C/min in 5°C/min
velopment of the analytical methods used to cal- intervals. Data obtained from TGA experiments
culate kinetic parameters based on TG data. One were transformed in ASCII format and analyzed
the most used methods to determine the activa- using a statistics computer program. TGA tests
tion energy from dynamic measurements was were carried out in aluminum crucibles (5.9 3 4.7
proposed by Friedman.16 The expression referred mm) where samples were placed without any pre-
1072 JIMENEZ ET AL.

Figure 1 TG and DTG curves at 10°C/min on pure PVC powder.

vious treatment and experiments were run imme- bender P-600 mixer during 20 min and 23 6 2°C.
diately. Nitrogen flow was used in order to re- The paste obtained was then put in a vacuum
move all corrosive gases evolved in the degrada- dryer (P , 7 3 10 2 Pa) during 5 min. Once air
tion and to avoid thermoxidative degradation. was eliminated from pastes, they were ready for
As a PVC raw material, Vestolit 7021 (HULS, tests.
Germany) was used. This is a suspension-pro- Three different formulations were analyzed in
duced polymer nonprestabilized and capable of this work: Pure PVC and two different PVC/
forming low viscosity paste easily. This resin, nor- DEHP blends, i.e., 50/50% wt and 70/30 % wt.
mally used in rotational molding, was character-
ized by determining some properties mainly re-
lated to the molecular weight. The molecular
weight obtained by means of Gel Permeation RESULTS AND DISCUSSION
Chromatography (Shimadzu LC-10AD) was 1.35
3 105. A K value equal to 72.0, related to the Dynamic thermogravimetric curves obtained on
viscosity of the resin and plasticizer mixture was neat PVC at 10°C/min are shown in Figure 1.
obtained by application of the DIN 53-726 stan- Weight loss derivatives are also reported in the
dard. same plot; it can be noticed that the polymer is
Industrial grade DEHP was used. This plasti- thermally stable up to 200 –220°C. Above this
cizer has been historically the most used in the temperature interval, the polymer undergoes the
toy industry. Purity of DEHP was checked by first step of the thermal degradation, which lasts
using gas chromatography (Shimadzu GC-9A) up to the temperature of 380°C when the second
with the obtainment of a single peak. degradation step starts. The exact values of the
Plastisols were prepared by mixture of the ad- onset temperatures of the degradation processes
equate amounts of resin and plasticizer in a Bra- depend on the previous thermal history of the
THERMAL DEGRADATION OF PVC PLASTISOLS 1073

Figure 2 Degree of reaction versus temperature at 10°C/min of different plastisols.

polymer and on the heating rate of the DSC test. neously and therefore it is very difficult to observe
These two steps are evident in the derivative a sharp distinction of these two phenomena from
curves shown in the same figure, which present the thermograms and their derivatives.2 It is very
two different peaks associated to each of the two difficult, then, to separate the effects of each re-
steps, respectively. Because the working temper- action on the overall mechanism of the thermal
atures of PVC and plastisols barely overcome degradation of plastisols. Since the goal of this
200°C and never reach 380°C, this work will focus work was to develop a phenomenological model to
on the study the first reaction step, which mainly be used to evaluate the effect of the plasticizer on
corresponds to the dehydrochlorination.1 There- thermal stability of plastisols, a complete elucida-
fore, in the calculation of the degree of degrada- tion of the mechanism of the reaction was not
tion a from eq. 1 the term of the final weight (W f ) considered.
corresponds to the weight at the end of the first The apparent activation energy was calculated
step, i.e., in correspondence of the baseline of the using the Friedman analysis applied to the gen-
first peak of the derivative curve. eral kinetic model (eq. 2) without supposing any
Curves of a versus T, for a dynamic test at form for the function f( a ). According to eq. 6, a
10°C/min, calculated according to these assump-
tions, are shown if Figure 2 where data corre- S D
plot of ln
da
dt
versus 1/T at fixed a, should give a
sponding to three different samples have been straight line whose slope gives the apparent acti-
plotted. The effects of the plasticizer on the ther- vation energy. The results of such analysis are
mal degradation can be observed in this figure. presented in Figure 3 for one of the samples an-
The curves relative to plastisols are in fact shifted alyzed (DEHP 30%) where the curves correspond-
to lower temperatures. This behavior is in agree- ing to five different values of a, ranging from 0.1
ment with the two different reactions that char- to 0.9, are shown. Each data point on a single
acterize the first step of the degradation of plas- curve was extracted from a single dynamic run at
tisols. It was, in fact, previously indicated that a different constant heating rate. Straight lines
during this step the evaporation of plasticizer and fitting the data of Figure 3 are practically parallel
dehydrochlorination of PVC proceed simulta- indicating constant activation energy in the range
1074 JIMENEZ ET AL.

Figure 3 Application of Friedman method to the degradation of the sample DEHP


30% in correspondence of different degrees of reaction.

of a analyzed and confirming the validity of the This specific selection and the elimination of eq. 3
approach applied. It should be reminded that this model are supported by the particular results ob-
analysis corresponds only to the first degradation tained in isothermal tests. It can be noticed in the
process. A different and independent value can be isothermal thermogram reported in Figure 4 that
expected for the second degradation process not the maximum of the curve does not correspond to
analyzed in this work. The same procedure was the starting point (a 5 0) as predicted by eq. 3. A
used to compute the apparent activation energy maximum of the curve after the beginning of the
relative to the other two samples. The obtained reaction is typical of autocatalytic models as the
values, shown in Table I, were used as starting one represented by eq. 4. The full model adopted
points for the regression analysis of the full was then the following:
model.

S D
In order to obtain a phenomenological model
da Ea
for the thermal degradation of plastisols, an ex- 5 A exp 2 ~1 2 a ! n~ a ! m (7)
plicit expression for f( a ) in the kinetic equation is dt RT
needed. Following the same approach already ap-
plied to the degradation of other polymers18 a The kinetic parameters of eq. 7 were calculated
behavior represented by eq. 4 was considered. by regression analysis using as starting point the
results of the Friedman method previously ob-
tained.
Table I Results of the Kinetic Analysis While the activation energy is a parameter
that can be directly calculated from dynamic tests
E a (kJ/ ln(A) as described above, reaction orders n, m, and the
Sample mol) (min21) m n pre-exponential factor A typically require infor-
mation given by isothermal tests. In this research
PVC 249 6 7 52 0.25 2.7 an alternative method to determine such param-
DEHP 50% 174 6 6 44 0.25 2
eters using only dynamic experiments has been
DEHP 30% 169 6 7 35 0.2 1.8
developed and applied.
THERMAL DEGRADATION OF PVC PLASTISOLS 1075

Figure 4 Degree of reaction and its derivative, as a function of time, during an


isothermal test on pure PVC at 210°C.

In the Friedman analysis, the first two terms of The results of the complete procedure are re-
the right hand side of eq. 6 are not function of the ported in Table I where an overall reaction order
heating rate and temperature. The value of the 2 is reported for the plastisols while is close to 3
sum S, of these two terms, represents the inter- for the pure PVC. Also the activation energy and
cept with the y-axis of each isoconversion curve the preexponential factor decrease with the plas-
shown in Figure 3. Then, using as function f( a ) ticizer content, compared with results for pure
the one described by eq. 4, the following yields: PVC. The values of the apparent activation en-
ergy of PVC and plastisols obtained here cannot
S 5 ln A 1 n ln~1 2 a ! 1 m ln~ a ! (8) be fully compared with those reported in litera-
ture,2,12 as ours refer to eq. 7 model not consid-
Plotting the term S, calculated at different val- ered in the previous investigations. In any case,
ues of a, versus a and performing a nonlinear an apparent activation energy of the same order
regression analysis on the data, the values of A, has been obtained.20
n, and m can be easily calculated. The results of The decrease of the apparent activation energy
such analysis in terms of S versus a are shown in obtained in plastisols, with respect to the values
Figure 5 for the formulation DEHP 50%. The observed in the PVC resin, is coherent with the
agreement with the model proposed is shown in decrease of the temperature of maximum weight
the same figure; a correlation coefficient of 0.98 loss observed in the TGA tests. These two phe-
was obtained giving a good fit of the data. nomena, which result in a faster start of the deg-
Once the values of the kinetic constants of the radation reaction, can be explained taking into
eq. 7 model were calculated according with the account the interactions between plasticizer and
procedure described above, plots of a versus tem- PVC during mixing and cooling. Thus, one of the
perature were obtained for different heating rates consequences of the mixture between the plasti-
by numerical integration of the model. Parameter cizer and the resin may be the separation of
values were subsequently improved by using a chains in the 3-dimensional structure of PVC.
regression analysis on the integrated model. Because of this, the interactions between macro-
1076 JIMENEZ ET AL.

Figure 5 Representation of eq. 8 versus degree of degradation for the sample DEHP
50%.

molecules become less strong and therefore the additional point to be noticed is related to the
beginning of the degradation reaction is more fa- influence of the heating rate on the experimental
vorable energetically. Furthermore, a small in- results. As it is indicated in Figure 6, results
crease in apparent activation energy is observed obtained at 5 and 10°C/min fit well with the ex-
for a DEHP 50 plastisol, compared to DEHP 30 in perimental data, but this is not true at 50°C/min.
agreement with previously reported results which This may be caused by the thermal lag between
indicate that a PVC/DEHP (70 : 30) combination the furnace nominal temperature and the actual
has the lower activation energy.2 In fact, it has sample temperature. It has been also indicated
been previously suggested that the higher ther- that some problems related to heat and/or mass
mal degradation resistance of the PVC/DEHP (70 transfer may be important at high heating rates,
: 30) formulation can be attributed to the maxi- affecting the temperature control and the unifor-
mized interaction between the PVC and the plas- mity of temperature inside the sample.8 More-
ticizer, who acts as a protection coating of the over, some additional errors can be caused by the
PVC macromolecules. This explanation is in intrinsic thermal conductivity of the sample,
agreement with the extensive industrial use of mainly PVC, because the sample used was a fine
this compound taking also advantage of its higher powder respect the use of a continuous fluid as in
processability and workability. the case of plastisols. Therefore, the use of heat-
The final results obtained by the integration of ing rates lower than 10 to 15°C/min is highly
eq. 7 model, using the kinetic parameters re- recommended in order to avoid such problems.
ported in Table I, are shown in Figures 6, 7, and Another cause of the discrepancy between the
8; model results are well compared with experi- experimental data and the model observed for
mental data at different heating rates for PVC, pure PVC may be attributed to the effect of
DEHP 50% and DEHP 30% respectively. As it can DEHP. While it is evident the effect of DEHP on
be observed, results obtained from the application the apparent activation energy, which results in a
of the proposed model fit with the experimental clear decrease for every sample, the presence of
for every sample. This is a sign of the good agree- this plasticizer could also influence the form of
ment between the model proposed here and the the kinetic curve resulting in a different kinetic
results obtained from the TGA experiments. An expression. The values of the reaction orders re-
THERMAL DEGRADATION OF PVC PLASTISOLS 1077

Figure 6 Comparison between model results and experimental data on pure PVC.

ported in Table I for the degradation of pure PVC fact that the evaporation of plasticizer and the
are in fact different from the ones of the two dehydrochlorination may not happen simulta-
plastisols. This behavior may be attributed to the neously, but at different stages that superimpose

Figure 7 Comparison between model results and experimental data on DEHP 30%.
1078 JIMENEZ ET AL.

Figure 8 Comparison between model results and experimental data on DEHP 50%.

in correspondence of a certain temperature and reactions involved in the mechanism is the evap-
conversion range. Therefore it can be considered oration of plasticizer, which influences the kinet-
that the first reaction, i.e. evaporation of DEHP, ics of the other reaction, i.e., dehydrochlorination
influences the second process, changing the over- of PVC macromolecules. The apparent activation
all mechanism of the reaction. Nevertheless the energy decreased with the DEHP content accord-
approach used in this work keeps its validity and ing to the theory. The proposed model can be used
allows the comparison of the effects of plasticizer as a tool to evaluate the thermal stability and to
and other additives on the thermal degradation of understand better the effect of concurrent reac-
plastisols. tions during thermal degradation. The procedure
here developed can be suitable for application to
the study of degradation kinetics of other kind of
CONCLUSIONS plastisols containing more than one additive,
mainly thermal stabilizers.
A new phenomenological model was developed in
order to explain the kinetics of the thermal deg-
radation of DEHP based plastisols. This model REFERENCES
has shown a good agreement with experimental
data. The kinetic parameters—activation energy 1. Minsker, K. S.; Kolesov, S. V.; Zaikov, G. E. Deg-
(E a 5 179 kJ/mol) and reaction orders (n 5 2, m radation and Stabilization of Vinylchloride Based
5 0.2)—were calculated using the Friedman Polymers, Oxford: Pergamon Press, 1988.
method and a modification of it. The model was 2. Jiménez, A.; Berenguer, V.; López, J.; Vilaplana, J.
J Appl Polym Sci 1996, 60, 2041.
then integrated numerically to obtain the conver-
3. Wu, C. H.; Chang, C. Y.; Hor, J. L.; Shih, S. M.;
sion curves. Some clear differences in parameters
Chen, L. W.; Chang, F. W. Can J Chem Eng 1994,
have been observed when the model was applied 72, 644.
to thermal degradation of pure PVC compared to 4. Tanaka, H. Thermochim Acta 1995, 267, 29.
plastisols. These discrepancies were attributed to 5. Simon, P. Polym Degrad & Stab 1995, 47, 265.
the existence of two simultaneous reactions in the 6. McNeill, I. C.; Memetea, L.; Cole, W. J. Polym
thermal degradation of plastisols. One of the two Degrad & Stab 1995, 49, 181.
THERMAL DEGRADATION OF PVC PLASTISOLS 1079

7. Bockhorn, H.; Hornung, A.; Hornung, U.; Teepe, S.; 14. Hollande, S.; Laurent, J. L. Polym Degrad & Stab
Weichmann, J. Combust Sci & Tech 1996, 116 – 1997, 55, 141.
117, 129. 15. Schneider, H. A. Polym Eng Sci 1992, 32, 1309.
8. Carrasco, F.; Pagès, P. J Appl Polym Sci 1996, 61, 16. Friedman, H. L. J Polym Sci C 1964, 6, 183.
187. 17. Kenny, J. M.; Torre, L. in High-Temperature Prop-
9. Pielichowski, K.; Hamerton, I. Polymer 1998, 39, 241. erties and Applications of Polymeric Materials,
10. Biggin, I. S.; Gerrard, D. L.; Williams, G. E. J Vinyl Tant, M. R.; Conell, J. W.; McManus, H. L. N., Ed.,
Technol 1982, 4, 150. Am Chem Soc, 1995, 140.
11. Bublick, A. T.; Bogomolova, L. K.; Paufner, O. E.; 18. Kenny, J. M.; Torre, L.; Nicolais, L. Thermochimica
Tsirkun, R. R. Plast Massy 1985, 11, 17. Acta 1993, 227, 97.
12. Jiménez, A.; Berenguer, V.; López, J.; Sànchez, A. 19. Hawkins, W. L. Polymer Degradation and Stabili-
J Appl Polym Sci 1993, 50, 1565. zation; Springer–Verlag: Hamburg, 1983.
13. Jiménez, A.; López, J.; Vilaplana, J.; Dussel, H. J. J 20. Carberry, J. J. Chemical and Catalytic Reaction
Anal & Appl Pyrolysis 1997, 40 – 41, 201. Engineering; McGraw Hill: New York, 1976.

You might also like