Professional Documents
Culture Documents
a r t i c l e
i n f o
Article history:
Received 29 March 2011
Received in revised form 5 July 2011
Accepted 20 July 2011
Keywords:
Glycolysis
PET
Granulometry
Surface area
Mathematical model
Glycolysis kinetic
a b s t r a c t
Polyethylene terephthalate post-consume (PET-pc) glycolysis was investigated by the use of ethylene
glycol (EG) and zinc acetate, as catalyst. It was focused the kinetic aspects through use of mathematical
model specially developed for application in this study. The grains-lot was sieved in different size ranges
and a relation between surface area and granulometry, surface area and temperature on the conversion
and depolymerization rate was proposed. At temperatures ranging 180190 C the depolymerization
rate is quite elevated and almost 100% of conversion is obtained up to 3 or 4 h reaction time. For lower
temperatures (170180 C), equilibrium is achieved and it becomes more important as the temperature
is decreased. The conversion prole showed an initial activation stage where the mass transfer between
the liquid and solid phases is minimal. The proposed mathematic model was based on these ndings and
on reaction mechanism that differentiates the reaction sites present in the PET surface. By that model the
value of rate constant (k) for each temperature, and the dependence of k with 1/T was calculated. Four our
best knowledge it is the rst time that a mathematical model considers the activation stage in the earlier
times of PET depolymerization reaction. The inputs yielding, time and temperature were included in the
used mathematical model that ts very well the experimental data obtained at temperatures higher than
180 C. This model helps to predict the necessary mass of PET for producing a desired amount of products.
2011 Elsevier B.V. All rights reserved.
1. Introduction
Technological advances in the manufacture of PET allowed it to
be produced at low cost. Allied to its mechanical properties, this
has led to a signicant increase in the production of this polymer. According to the Brazilian PET industry association, ABIPET
(Associaco Brasileira da Indstria do PET) [1], the quantity and
proportion of PET that is recycled have been growing year-by-year
and currently 263 k Tons (55.6% of production) are recycled.
There are three main types of PET-recycling [26]: (a) chemical
recycling (or depolymerization); (b) quaternary recycling (energy
recovery); and (c) mechanical recycling. Chemical recycling such
as depolymerization by hydrolysis [79], alcoholysis and glycolysis [1013] have been recently gained much attention. For instance,
terephthalic acid (TPA) and ethylene glycol (EG) can be recovered
and used as raw materials in many industrial processes, including polymer synthesis while the bis-hydroxyethyl terephthalate
(BHET) can be used in synthesis of new PET or other co-polymers.
Lorenzetti et al. [2] focused on the resulting products in their
review of PET degradation methods. Their study shows that gly-
Corresponding author. Tel.: +55 44 3261 3664; fax: +55 44 3261 4125.
E-mail address: gizilenecarvalho@gmail.com (G.M. Carvalho).
1385-8947/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.cej.2011.07.031
Nomenclature
a
ABET
A0
Ap
Aps
Au
c
d
Dp
f(%)
K
k
k1
L
m(t)
m0
ma
mBHET
MBHET
MEG
moligomer
mprod
mtheo
mu
v
M
r
s
s
S0
S1
S2
St
ti
Vu
X
yi
[]
red
MarkHouwinkSakurada parameters
surface area from BET method
initial contact area between the PET-EG phases
surface area of the actual particle
area of a perfect sphere with volume equal to that
of the a actual particle
unitary area of a particle
bigger size of PET grains to 3 parameters model
proportionality constant
mean mesh size of the sieves
mass fraction of sample retained in a specic mesh
size of the sieves.
MarkHouwink parameters
rate constant of the reaction
rate constant of the reaction
mean thickness of PET grains
residual mass of solid PET as function of time
mass of PET added to the reactor and m is the
mean mass values for each sample
BHET mass
molar mass of bis(hydroxiethyl)terephthalate
molar mass of ethylene glycol
mass of oligomers
mass of products retained on the lter paper
theoretical mass of BHET calculated based on the
weight of reacts PET
mean mass of a grain
average viscosimetric molar weight
PET grains radius from geometrical two-parameter
model
standard deviation of the mass of the samples
standard deviation of the thickness of PET grains
carbonyl groups that do not generate reaction products when cleaved (the middle of the chain)
carbonyl groups that generate reaction products
when cleaved (end of the chain)
carbonyl group does not change the polymer
the total number of sites at the beginning of the
reaction (t = 0)
induction time
mean volume of a grain
percentage conversion of solid PET
ratio between the number of sites of a determined
species on the surface of the polymer
intrinsic viscosity
reaction yield
reduced viscosity
sphericity
211
glycol solution. They compared the geometric area and the effective
surface area of PET, as measured by BET (BrunauerEmmettTeller)
analysis, and demonstrated that the effective area increases as the
reaction progresses. Based on this, the authors proposed a kinetic
model in which the effect provoked by the decrease in geometric
area, due to the reaction, on the reaction velocity is considered.
The glycolysis of PET has been object of study from different
point of view. Although the inuence of variables such as time,
temperature, EG:PET molar ratio, nature of catalyst, concentration,
particle size, stirring rate, reaction time on the glycolysis process
have been investigated [1517], available kinetic models do not
cover all aspects of depolymerization process. The method of catalyzed glycolytic depolymerization with optimization technique
was described by Goje and Mishra [16]. The authors pointed that
procedures and resulting kinetic parameters vary with assumed
kinetic model and applied data tting procedure. The different values of activation energy (Ea ) cited in literature for depolymerization
of PET was attributed to the changes in reaction parameters and to
different chemicals employed for PET depolymerization. According
to Paszun and Spychaj [6], the literature related to the glycolysis of
PET covers mainly the application of the resulting products, while
only few authors turn their attention to the reaction kinetics. When
glycolysis is carried out below the melting temperature of PET, the
reaction medium consists, initially, of a solid phase (pure PET) and
a liquid phase (EG + catalyst). However, as the reaction proceeds,
other phases appear: swollen PET, a solution of polyesters and oligoesters, until at the end of the reaction there is only a liquid phase
(solution of glycols and oligoesters) [12]. The depolymerization
changes from a heterogeneous reaction to a homogeneous reaction, as the reaction progresses. Lpez-Fonseca et al. [17] developed
a theoretical model to predict the time conversion of PET during
glycolytic depolymerization. The authors observed that at initial
stages the reaction occurred in a heterogeneous phase and only at
higher reaction times the reaction became a single homogeneous
phase. The kinetic model was developed according to a homogeneous reversible catalytic model and was found to be consistent
with experimental data in the range temperature of 150196 C.
If the depolymerization reaction occurs initially on the surface of
the PET particles, what is the inuence of contact-area between the
phases (that is, the surface area of the PET grains)? The progress of
the reaction depends on the EG diffusing onto the surface of the
PET and on the removal-rate of the depolymerized material from
the surface of the PET (dry and swollen) into the solution. Hence,
the diffusion process can be taken to control the reaction rate and,
accordingly, the rate at which the solution is stirred also becomes
an important parameter in the reaction mechanism.
In one of their studies, Pardal and Tersac [12] examined the inuence of temperature, the presence of a catalyst and the morphology
of the PET. The observation that reactivity is much greater at 220 C
suggests that the diffusion of diethylene glycol in the PET is favored
at this temperature, increasing the reaction rate as compared to
reactions at lower temperatures.
With the objective of contributing to the understanding of the
glycolysis reaction mechanism for PET-pc with EG, this study proposes a new kinetic model for this reaction. The proposed reaction
mechanism is based on the possibility that the PET chain is cleaved
at different sites by the EG and that the amount of cleavages
depends on the ratio of the number of sites on the surface of the PET
grains. The following steps were performed to achieve this objective (i) determination of the geometric surface area of the PET-pc
grains using geometric models with two and three parameters; (ii)
experimental determination of conversion (X) as a function of the
granulometry of the grains of PET-pc; (iii) study of the inuence
of time and temperature on the heterogeneous depolymerization
reaction; (iv) determination of the PET-pc conversion (X) using the
proposed model, for different time and temperature conditions, and
212
(1)
30
mu =
n (mu )i
i=1 i
30
n
i=1 i
30
i=1
30
n
i=1 i
(2)
mprod
mtheo
(3)
m0 m(t)
m0
(4)
where m0 is the mass of PET added to the reactor and m(t) is the
remaining solid PET mass as a function of time.
The relationship between conversion and the PET-pc geometric
area was evaluated for the glycolysis reaction at 180 C for 90 min.
A plot of conversion against time at the different temperatures for
A0 = 593.5 cm2 was produced and analyzed.
Thermal characterization of the samples was performed using a
Shimadzu DSC 50 calorimeter, with a heating rate of 10 C min1 in
an atmosphere of nitrogen at a ow rate of 20 mL min1 . The DSC
curve obtained for the product from second ltration was compared with the DSC curve for the oligomeric diols derived from
terephthalic acid [18].
3. Mathematical modeling
3.1. The reaction mechanism and denition of variables
In this study, a kinetic model with equations based on the different possibilities for polymer chain cleavage by EG is proposed.
The following hypotheses were considered for building the model:
(a) The reaction between an EG molecule and an ester group located
at the surface of the polymer causes a cleavage of polymer chain
at the point of the reaction. The reaction occurs at the interface
between the solid PET-pc and the diffused EG.
(b) If the reaction occurs between an EG molecule and an ester
group situated close to the end of PET chain, the each cleavage contributes to the reaction progressing and a leaving group
is formed consisting of a monomer (BHET) or an oligomer of low
molar mass.
(c) If the reaction occurs between an EG molecule and an ester
group situated far from the end of the chain, the cleavage does
not contribute to the reaction progress. In this case, the polymer molecule splits into two, forming two new chain-ends at
the point of cleavage. If this happens, the mass of the polymer is
increased by the accommodation of a molecule of EG relative
to the mass of polymer chain before the cleavage.
The sites where reaction may take place, that are the ester
groups located at the surface of the polymer, are classied as: S0 ,
S1 and S2 . The S0 sites indicate carbonyl groups that do not generate reaction products when cleaved (situated at the middle of the
chain). The S1 sites indicate carbonyl groups that generate reaction
products when cleaved (situated at the end of the chain). The S2
sites are inert, as a reaction involving these carbonyl groups do not
(5)
(6)
The terms for the rates of mass entering and leaving will be substituted by differential or algebraic equations, as required for each
specic case.
Mathematically, the proposed mechanism can be described by
the material balance, as given in Eq (6), where the solid PET granule
is considered as the dynamic system. According to this equation,
the change in the PET granule mass over time (dm/dt) equals the
difference between the mass entering in the granule and the mass
in ) rises as EG
leaving out the granule. The rate of mass entering (m
attacks the S0 sites. Taking this rate as being of rst order in relation
to yS0 and to mass of solid residual PET, gives the equation
in = kyS0 (t)m(t)
m
(7)
where k is a rate constant dependent on temperature and represents the cleavage rate.
out ) as EG attacks the S1 sites,
The rate of mass leaving rises (m
at which time the EG molecules condense onto the polymer and
simultaneously soluble BHET molecules leave out of PET chain. Taking the leaving rate to have a rst-order relationship with yS1 and
to mass of solid residual PET, gives the equation
out = k1 yS1 (t)m(t)
m
(8)
213
(9)
(10)
(11)
(12)
(13)
(14)
214
Fig. 2. Illustration of how S0 decreases and S1 increases during the conversion (X) of PET-pc.
Table 1
Estimates for mu and Vu and L as a functions of the granulometry from 15 samples of 30 grains each.
Dp (mm)
f (%)
ma (g)
s (g)
mu (mg)
Vu (mm3 )
L (mm)
s (mm)
7.18
5.56
3.56
2.03
1.44
0.89
5.43
25.15
60.62
6.16
1.80
0.77
1.1893
0.8212
0.6371
0.2140
0.1575
0.0921
0.1400
0.0243
0.0201
0.0176
0.0152
0.0117
39.64
27.37
12.74
3.567
2.230
1.149
28.83
19.91
9.267
2.594
1.622
0.835
0.5465
0.5014
0.4400
0.4027
0.3735
0.3401
0.0109
0.0093
0.0365
0.0200
0.0548
0.0346
Sample mass of 30 grains = ma ; Standard deviation of mass = s; Standard deviation of thickness (L) = s ; Mean mesh size of the shieves = Dp ; Unitary mass of grains = mu ; Unitary
volume of grains = Vu ; Percent of PET mass retained in the Tyler sieves f(%).
Aps
Ap
(15)
Fig. 3. Hypothetical models surface area. (a) Two-parameter model and (b) threeparameter model.
215
Table 2
Unitary area and sphericity as a function of granulometry.
Dp (mm)
7.18
5.56
3.56
2.03
1.44
0.89
2 parameters model
3 parameters model
r (mm)
Au (mm2 )
4.223
3.555
2.589
1.432
1.176
0.884
125.7
90.62
49.28
16.51
11.44
6.804
0.362
0.392
0.433
0.553
0.583
0.631
c (mm)
Au (mm2 )
7.804
7.142
5.916
3.173
3.015
2.761
127.5
92.26
50.46
17.07
12.01
7.397
ABET (mm2 )
0.357
0.385
0.423
0.535
0.556
0.580
52.00
36.57
20.48
A0 = nAu
(16)
90
Conversion (X, %)
model, while for the discoidal PET grains (p 2/3 and 2/3 < q 1.0),
the values of Au used were calculated using the two-parameter
model. To calculate the initial contact area between the PET-EG
phases (A0 ) for each size range, the values of the unit mass and the
total mass added to the reactor were used, as given in equation
60
30
400
The relationship between the conversion and the geometric
area of the PET-pc grains in the glycolysis reaction, (T = 180 C and
t = 90 min) were evaluated as a function of the initial contact area,
A0 , and are given in Fig. 5. As the reaction time was xed at 90 min,
it can be seen that the reaction velocity is largely dependent on the
contact area. As pointed by some authors [12,1517,24], stirring
rate inuence the conversion of PET. In conditions of high solution
stirring rate the mass transfer resistance can be eliminated in the
reaction medium. In all experiments the magnetic stirring is maintained constant at ca. 200 rpm and the mass transfer mechanism
was not eliminated neither considered in the mathematical model.
To estimate the PET mass required to produce a specic mass of
product (mprod ), the product yield was assessed. In accordance with
the stoichiometry of the reaction, 1.323 g of BHET (mtheo ) is formed
for every gram of PET reacted. However, due to inherent process
losses and to incomplete depolymerization, the mass of products
obtained on the lter paper, after drying, was smaller than the theoretical BHET mass, mtheo . The conversion (X) for the reactions was
600
800
1000
dened in Eq. (4). From this equation and take into account the stoichiometry of PET glycolysis to form BHET, the following equation
was obtained:
=
mprod
mtheo
mprod
1.323(m0 X)
(17)
216
Table 3
Values of p and q and the Zingg classication of the particles studied and the surface area for a 15 g sample of PET.
Dp (mm)
Class
Au (mm2 )
mu (mg)
A0 (cm2 )
7.18
5.56
3.56
2.03
1.44
0.89
0.072
0.090
0.124
0.198
0.259
0.382
0.920
0.779
0.602
0.640
0.478
0.322
discoidal
discoidal
lamelar
lamelar
lamelar
lamelar
125.7
90.62
50.46
17.07
12.01
7.397
39.64
27.37
12.74
3.567
2.230
1.149
376
553
1176
4258
6726
13055
473.3
501.6
593.5
727.0
808.0
965.7
25
100
80
75
%)
15
10
40
mprod
mtheo
0
0
20
40
60
80
20
X (%)
60
yield (
mass (g)
20
100
170 C
o
175 C
o
180 C
o
185 C
o
190 C
50
25
0
100
Conversion (X, %)
50
100
150
200
250
The left vertical axis on Fig. 6 gives the values of mprod and mtheo ;
the right vertical axis gives values of . The Grubbs test suggests that
the yield from the smallest conversion reaction should be excluded
when calculating the mean yield (as the conversion is small, any
loss results in large errors). Excluding the smallest conversion, the
mean reaction yield was 91.6%. Therefore, the mass of products
collected on the lter paper can be estimated, in terms of mean
value, using
area between the phases. Note that, for the used operating conditions, the induction period is greater at lower temperatures. This
behavior was not predicted by the proposed model and suggests
that diffusion may play an important role in the reaction mechanism allowing the reaction to take place more readily in the solid
phase at temperatures above 180 C.
mprod = 1.212m0 X
(18)
(19)
(20)
-4,4
y=-11977x+21.69
R2=0.990
-4,6
k=2.637.10 e-11977/T
-4,7
-4,8
lnk
-4,9
-5,0
-5,1
-5,2
-5,4
y=-5013.2x+6.346
2
R =0.994
-5013.2/T
k=572.5e
0,00218
80
40
0
170
-5,5
0,00216
120
-4,5
-5,3
217
0,00220
0,00222
0,00224
0,00226
1/T (K -1 )
180
190
o
temperature C
Fig. 10. Induction times (ti ) at different temperatures. The ti value was obtained, in
each case, from an intercept to the straight line in the major decomposition stage.
Fig. 8. Values of k as a function of 1/T. Fitting for lower temperatures (170 175 and
180 C, dash line) and higher temperatures (at 180, 185 and 190 C, solid line).
Fig. 9. Conversion of PET-pc as given by the experimental data (points) and the proposed model (line) as a function of temperature. (a) Temperatures of 180, 185 and 190 C
and (b) temperatures of 170 and 175 C.
218
BHET
Oligomers
PET-pc
100
200
Temperature /
300
o
Fig. 11. DSC thermogram for (a) PET-pc; products obtained in the second ltration:
(b) oligomeres retained on quantitative lter paper and (c) BHET obtained after
cooling the ltrated at 4 C.
An initial delay in the PET conversion can be seen at all temperatures. This induction period was attributed to the low initial
probability of the EG attacking the PET chain ends, which leads
principally to the formation of BHET.
Mathematical modeling and the reaction mechanism presented
here estimates the experimental data for temperatures of 180 C
and above. At these temperatures, the induction time observed can
be explained in terms of the types of sites (S0 ) on the polymer surface. For temperatures below 180 C, the diffusion process must
also be taken into consideration.
The product characterization showed that BHET is the main
reaction product, with small quantities of dimers and trimers also
being produced.
In the reaction yield study, the relationship established to calculate the mass of BHET formed as a function of the conversion,
showed that recovery was 91.6% of the theoretical value. This
percentage may be increased by implementing more efcient separation methods. However, the development of such methods was
not an objective of this study. Using this relationship and the solution from the mathematical model, it is possible to predict the
operating conditions and PET mass required to produce any quantity of products, thus providing a production planning tool for the
recycling of PET-pc and other polymers from a solidliquid reaction
such as the type studied here.
Acknowledgements
M.E.V. thanks to (CNPq, Brazil) for the master fellowship. The
authors thank CNPq, Brazil for the nancial support (proc. no.
309005/2009-4 and 481424/2010-5). All authors thank to COMCAP/UEM for access to DSC experiments.
References
[1] Associaco Brasileira da Indstria do PET (ABIPET)6 Censo da Reciclagem de
PET no Brasil, available http://www.abipet.org.br/, Accessed 01/16/2011.
[2] C. Lorenzetti, P. Manaresi, C. Berti, G. Barbiroli, Chemical recovery of useful
chemicals from polyester (PET) waste for resource conservation: a survey of
state of the art, Journal of Polymers and the Environment 14 (2006) 89101.
[3] V. Sinha, M.R. Patel, J.V. Patel, PET waste management by chemical recycling: a
review, Journal of Polymers and the Environment 18 (2008) 825.
[4] S.R. Shukla, A.M. Harad, L.S. Jawale, Recycling of waste PET into useful textile
auxiliaries, Waste Management 28 (2008) 5156.
[5] F. Awaja, D. Pavel, Review-recycling of PET, European Polymer Journal 41 (2005)
14531477.
[6] D. Paszun, T. Spychaj, Chemical recycling of poly(ethylene terephthalate),
Industrial & Engineering Chemistry Research 36 (1997) 13731383.
[7] G.M. Carvalho, E.C. Muniz, A.F. Rubira, Hydrolysis of post-consume
poly(ethylene terephthalate) with sulfuric acid and product characterization
by WAXD, NMR13C and DSC, Polymer Degradation and Stability 91 (6) (2006)
13261332.
[8] B.Z. Wan, C.Y. Kao, W.H. Cheng, Kinetics of depolymerization of poly(ethylene
terephthalate) in a potassium hydroxide solution, Industrial & Engineering
Chemistry Research 40 (2001) 509514.
[9] G. Gcl, T. Yalcnyuva, S. zgms, M. Orbay, Simultaneous glycolysis and
hydrolysis of polyethylene terephthalate and characterization of products by
differential scanning calorimetry, Polymer 44 (2003) 760976167.
[10] D. Navnath, Pingale glycolysis of postconsumer polyethylene terephthalate
waste, Journal of Applied Polymer Science 115 (1) (2010) 249254.
[11] F. Pardal, G. Tersac, Comparative reactivity of glycols in PET glycolysis, Polymer
Degradation and Stability 91 (2006) 25672578.
[12] F. Pardal, G. Tersac, Kinetics of poly(ethylene terephthalate) glycolysis by
diethylene glycol. I. Evolution of liquid and solid phases, Polymer Degradation
and Stability 91 (2006) 28402847.
[13] F. Pardal, G. Tersac, Kinetics of poly(ethylene terephthalate) glycolysis by
diethylene glycol. Part II: effect of temperature, catalyst and polymer morphology, Polymer Degradation and Stability 92 (2007) 611616.
[14] T. Yoshioka, T. Motoki, A. Okuwaki, Kinetics of hydrolysis of PET powder in nitric
acid by a modied shrinking-core model, Industrial & Engineering Chemistry
Research 37 (1998) 336340.
[15] A.C. Ruvolo-Filho, P.S. Curti, Chemical kinetic model and thermodynamic compensation effect of alkaline hydrolysis of waste poly(ethylene terephthalate)
in nonaqueous ethylene glycol solution, Industrial & Engineering Chemistry
Research 45 (2006) 79857996.
219
[21] L.E. Sissom, D.R. Pitts, Fenmenos de Transporte, rst ed., Guanabara Dois, Rio
de Janeiro, 1996.
[22] A. Botari, L.D. Bernardo, Modelaco matemtica macroscpica da perda de carga
e da remoco de slidos suspensos totais na ltraco direta ascendente, Revista
Engenharia Sanitria e Ambiental 12 (2) (2007) 149159.
[23] T. Demur, R.P.D. WALSH, Shape and size characteristics of bedload transported
during winter storm events in the Cwm Treweryn stream, brecon beacons, 14,
Turkish Journal of Earth Sciences, South Wales, 2005, 105121.
[24] C.Y. Kao, B.Z. Wan, W.H. Cheng, Kinetics of hydrolytic depolymerization of melt
poly(ethylene terephthalate), Industrial & Engineering Chemistry Research 37
(1998) 12281234.
[25] S. Baliga, W.T. Wong, Depolymerization of poly(ethylene terephthalate) recycled from post-consumer soft-drink bottles, Journal of Polymer Science Part A
Polymer Chemistry 27 (6) (1989) 20712082.