Solid state polymerization

S.N. Vouyiouka, E.K. Karakatsani, C.D. Papaspyrides

*
Laboratory of Polymer Technology, School of Chemical Engineering, National Technical
University of Athens, Zographou, Athens 157 80, Greece
Received 23 November 2004; revised 24 November 2004; accepted 24 November 2004
Available online 7 January 2005
Abstract
Polyesters and polyamides are commercially important polymers prepared by polycondensation. The conventional
solution to melt polymerization techniques stop at a low or medium molecular weight product, due to problems arising
from severe increase of the melt viscosity and operating temperatures. Higher molecular weights may be reached by Solid
State Polymerization (SSP) at temperatures between the glass transition and the onset of melting. Polycondensation
progresses through chain end reactions in the amorphous phase of the semicrystalline polymer, which in most cases is in
the form of akes (mean diameterO1.0 mm) or powder (mean diameter!100 mm); reaction by-products are removed by
application of vacuum or through convection caused by passing an inert gas. The advantages of SSP include low operating
temperatures, which restrain side reactions and thermal degradation of the product, while requiring inexpensive equipment,
and uncomplicated and environmentally sound procedures. Disadvantages of SSP focus on low reaction rates, compared to
melt phase polymerization, and possible solid particle processability problems arising from sintering. The review begins
with a theoretical background regarding SSP, the fundamentals of techniques and equipment, including the use of inert gas
in reactive systems. Further, it is explained how SSP progress involves both chemical and physical steps, since it is
controlled by reaction kinetics, reactive chain-end mobility in the amorphous phase, and condensate removal through
diffusion. The reaction temperature emerges as the most important parameter of SSP rate variation, due to its interaction
with all aspects of the process. High prepolymer molecular weight affects positively the SSP rate, since it is accompanied
by elevated degrees of crystallinity; this implies more effective connement of the amorphous phase and, therefore high
concentration and homogeneous distribution of reactive chain ends in the non-crystalline regions. Similarly, renement in
reacting particle size distribution and morphology, in conjunction with high gas ow rates, increases the interfacial area per
unit volume and the effectiveness of convective by-product elimination. Finally, the SSP rate increases principally by the
use of phosphorous catalysts, which also reduce agglomeration. In the following sections of this review, emphasis is on the
progress in experimentally determining the intrinsic rate constants of principal relevant chemical reactions. The
corresponding kinetic models are either based on the Flory theory, where the rate expressions are in terms of end-group
concentrations, or on a power-law description of the rate with respect to reaction time. Recent advances in modeling and
large scale simulation of the various physical and chemical processes occurring within a SSP reactor are reported. The goal
is to describe the dynamic evolution of all chemical species within the particle and its surroundings, and assess its
dependence on basic process variables. Finally, special focus is given on methods of manipulating the molecular weight
Prog. Polym. Sci. 30 (2005) 1037
www.elsevier.com/locate/ppolysci
0079-6700/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.progpolymsci.2004.11.001
* Corresponding author. Tel.: C30210 7723179; fax: C30210 7723180.
E-mail address: [email protected] (C.D. Papaspyrides).
distribution of the SSP product, by varying prepolymer particle size distribution, initial stoichiometry and condensate
content in the surroundings.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Solid state polymerization; SSP; Polyamides; Polyesters; SSP rate; Kinetics; Simulation; Molecular weight distribution
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2. SSP techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1. Types of reactor and main characteristics for batch and continuous processes . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2. Use of inert gas in SSP systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3. Parameters affecting the SSP rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.1. Reaction temperature effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2. Initial end group concentration effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3. Particles geometry and gas ow rate effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.4. Crystallinity effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.5. Use of catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4. Kinetics and simulation of SSP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.1. Reaction kinetics of SSP processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2. Modeling and simulation of SSP processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5. SSP and molecular weight distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1. Introduction
Polyesters and polyamides (PAs) are important
commercial polycondensation polymers, widely used
in a variety of applications. In addition to their
commanding positions as textile bers, PAs can be
readily processed through extrusion or injection
molding to yield mechanical and electrical parts,
while polyesters play an important role in the
automotive tire cord and bottle market [1]. Both are
mostly prepared by a stoichiometric stepwise reaction
between difunctional reactants, which is accompanied
by the formation of a lowmolecular weight condensate
[2,3]. In such condensation polymers, there is an
equilibrium such as the one described in Eq. (1) [4]
P
1
CP
2
4P
3
CB (1)
where P
1
and P
2
are monomer molecules or polymer
chains which combine to form P
3
(a longer chain), and
Bis a small molecule by-product, e.g. water in the case
of PAs and glycols and water in the case of polyesters.
The reaction schemes for the formation of typical
polycondensation polymers are presented in Fig. 1.
The prerequisites to obtain a polymer product of
high molecular weight (MW) consist of maintaining
proper end-group stoichiometry, removing the
condensate to prevent depolymerization, thereby
shifting the reaction equilibrium to the right [46].
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 11
Nomenclature
a
0
, a
1
, a
2
constants
BPA-PC poly(bisphenol A carbonate)
[C
1
]
t
concentration of the stoichiometrically
decient end groups at any given time
[C
1
]
0
initial concentration of the stoichiometri-
cally decient end groups
[C
2
]
t
concentration of the end groups in excess
at any given time
[C
2
]
0
initial concentration of the end groups in
excess
[C*] concentration of end groups in the non-
crystalline phase of polymer
[C] total end group concentration
[C]
t
total end group concentration at any given
time
[C]
0
initial total end group concentration
[C
ai
] apparent inactive end group concentration
[C
amorphous
] concentration of a component in the
amorphous phase of the polymer
C
condens.
concentration of the by-product in the
semicrystalline polymer
CO
2
carbon dioxide
KCOOH carboxyl end groups
[COOH] concentration of carboxyl end groups
[COOH]
t
concentration of carboxyl end groups at
any given time
[COOH]
0
initial concentration of carboxyl end
groups
[C
overall
] concentration of a component in the total
mass of the polymer
D diffusivity
D
condens.
diffusivity of the by-product in the
semicrystalline polymer
D
0,condens.
diffusivity of the by-product in the
completely amorphous polymer
D
EG
diffusivity of ethylene glycol in the
semicrystalline polymer
D
H2O
diffusivity of water in the semicrystalline
polymer
d
mean
mean diameter of the reacting particle
DP degree of polymerization
E
a
reaction activation energy
EG ethylene glycol
He helium
k reaction rate constant
k
a
apparent rate constant
K
eq
equilibrium constant
k
2
2nd order kinetics rate constant
k
f
forward rate constant
k
transest.
transesterication rate constant
k
ester.
esterication rate constant

M
n
mean number-average molecular weight

M
n
0
initial mean number-average molecular
weight

M
v
viscosity average molecular weight at any
given time

M
v0
initial viscosity average molecular weight
MW molecular weight
MWD molecular weight distribution
n reaction order
N
2
nitrogen
KNH
2
amine end groups
[NH
2
] concentration of amine end groups
[NH
2
]
t
concentration of amine end groups at any
given time
[NH
2
]
0
initial concentration of amine end groups
OH hydroxyl end groups
[OH] concentration of hydroxyl end groups
p
t
fractional polymerization conversion at
any given time
p
0
initial fractional polymerization
conversion
P product of the concentrations of the end
groups
P
t
product of the concentrations of the end
groups at any given time
P
0
product of the initial concentrations of the
end groups
PAs polyamides
PA-6,6 polyhexamethyleneadipamide
PA-6 polycaproamide
PA-4,6 polytetramethyleneadipamide
PA-12 polylaurolactam
PBT poly(butylene terephthalate)
PDEs partial differential equations
PDI polydispersity index
PEN poly(ethylene naphthalate)
PET poly(ethylene terephthalate)
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 12
A signicant difference between polyamides and
polyesters is that the equilibrium constant for
polyamides is hundreds times larger than that for
polyesters, and thus a much higher by-product
concentration can be tolerated in the rst case,
where the removal requirements are much less
severe [1]. For example, in polyamides, the
equilibrium constant varies between 100 and 750
depending on the water content in the reacting
system [2,79], meanwhile the relevant values for
transesterication and esterication widely used in
poly(ethylene terephthalate) (PET) melt polycon-
densation models are 0.51 and 1.25, respectively
[1013].
The predominant industrial process for polyhex-
amethyleneadipamide (PA-6,6) production involves
rst solution polymerization starting from an
aqueous PA-6,6 salt solution, removal of the
water from the reactor and nally polymerization
in the melt state (250270 8C) [14]. In the case of
poly(ethylene terephthalate) (PET), the common
commercial practice of production involves melt
polycondensation of the reactants (dimethyl tereph-
thalateethylene glycol or terephthalic acidethyl-
ene glycol), usually carried out around at 285 8C
[15]. An example of polymerization operating
conditions is presented in Table 1 [1], where a
degree of polymerization (DP) of 80 was reached
for both polymers considered. The high tempera-
tures used in these processes, combined with high
residence times, encourage thermal degradation
and gel formation, which can drastically impair
the quality of the end product [16]. For example, in
the case of PA-6,6, an undesirable reaction is the
dimerization of the diamine to a triamine, which
results in three-dimensional network formation and
nally in gelling [2,17,18]. The problems are
enhanced when dealing with more thermo-sensitive
polymers (i.e. polytetramethyleneadipamide, PA-
4,6), due to the high processing temperatures
required (260320 8C) [1821]. In addition, since
the melt viscosity increases very rapidly with the
polymerization progress, various problems arise
regarding the stirring of the reacting system, the
removal of the by-product and the temperature
control. Therefore, melt-based techniques are
usually not carried to high conversion, and result
mainly in the production of resins useful in
applications which do not require a high macro-
molecular chain length, e.g. a typical reached value
of the mean number-average molecular weight is
15,00025,000 g/mol, suitable for textile appli-
cations, whereas for injection or blow molding
applications, the molecular weight should be more
than 30,000 g/mol [14,15,2224].
With step-growth polymers, such as polyamides
and polyesters, one route to high molecular weight
products has been through Solid State Polymeriz-
ation (SSP) [25,26]. Accordingly, starting materials
are heated to a temperature higher than the glass
transition temperature (T
g
), but lower than the
onset of melting (T
m
) so as to make the end groups
mobile enough to react [5,15,27] and the by-
products are removed by passing inert gas through
PEPA 2-(2
0
pyridyl) ethyl phosphonic acid
RV relative viscosity
R the universal gas constant
R
ester.
esterication rate expression
R
transest.
transesterication rate expression
SHP sodium hypophospite
SMT solid-melt transition
SSP solid state polymerization
S/V particle surface area per unit volume
scCO
2
supercritical carbon dioxide
t reaction time
T absolute temperature
T
m
melting point
T
g
glass transition point
x the distance in the direction of diffusion
x
c
the mass fraction crystallinity
Greek letters
D[C] the difference between the total end group
concentration of the prepolymer and nal
product
D[COOH] the difference between the carboxyl end
group concentration of the prepolymer and
nal product
3 fractional molar excess of end groups
q
t
migration time of functional end groups
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 13
Table 1
Operating polymerization conditions resulting in DP equal to 80 [1]
Feed line Stages
Esterication Prepolycondensation Finishing stage
PET Terephthalic acid T: 360 8C T: 270 8C T: 290 8C
Ethylene glycol t: 3 h t: 2.2 h t: 1.5 h
P: 4 atm P: 50 Torr P: 5 Torr
Prepolycondensation Melt polycondensation
PA 6,6 Adipic acid T: 254 8C T: 267 8C
Hexamethylenediamine t: 1 h t: 1.1 h
Water P: 18 atm P: 1 atm
Fig. 1. Reaction schemes for the formation of typical polycondensation polymers.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 14
the reacting mass or by maintaining reduced
pressure [15,22,28].
SSP has certain advantages which render its
application attractive. SSP polymers often have
improved properties, because monomer cyclization
and other side reactions are limited, or even
avoided, due to the low SSP operating temperatures
[16]. Only linear chains seem to be formed [29],
and usually SSP products show greater heat
stability in the molten state than samples prepared
in the melt [30]; on the other hand, their monomer
and oligomers content is so low, that there is no
essential need for it to be removed [31]. Further-
more, the increase in the molecular weight during
SSP is accompanied by increased crystallinity and
crystal perfection [32] while drying the polymer,
which is important because moisture content may
inuence processability in the manufacture of yarns
[33]. In addition, there is practically no environ-
mental pollution, because no solvent is required,
and the process can be a continuous operation [34].
However, at SSP low temperatures, the chain
building reactions are slow compared to polym-
erization in the melt [30,32,35], because of the
reduced mobility of the reacting species, and the
slow diffusion of the by-products. Agglomeration of
the reacting particles has also been reported during
SSP, especially at high reaction temperatures, and
is related to a low softening point of the reacting
mass and to condensate retention in the system
[3640].
Various aspects of SSP have been reviewed during
the last decade [4042]. The relevant papers suggest
that two categories of SSP may be considered,
according to whether the starting materials are
crystalline monomers or semicrystalline prepolymers.
In the rst case, the monomer is transformed into a
polymer at a temperature lower than the melting
points of either monomer or polymer, by a reaction
which rarely takes place in a real solid state.
Depending on the reaction conditions, SSP is
accompanied, by a distinct transition of the process
from the solid to the melt state, as thoroughly
explained in Section 3.3 [3840,4346]. In the
second case, the polymerization is carried out on
low or medium molecular weight semicrystalline
prepolymers at a temperature below their melting
points [4,47,48]; Zimmerman [49,50] suggested
a two-phase model, according to which polymeriz-
ation proceeds in the amorphous regions, where end
groups and low molecular weight substances (con-
densate, oligomers) are excluded from crystalline
regions (Fig. 2). The equilibrium in the amorphous
regions is the same as for a completely amorphous or
molten polymer at the same temperature. Although
the main interest in SSP of monomers has been
conned up to now to laboratory studies, SSP of
prepolymers is already integrated into industrial
production processes, usually in the nishing stage.
For example, to have PET with DP equal to 145, SSP
of PET prepolymer (DPZ80) is performed at
atmospheric pressure, at 235 8C for 7 h. By compari-
son, for PA-6,6 of the same initial DP, the SSP is
carried out at 200 8C, and the residence time is 2 h to
achieve the same DP increase [1].
The SSP starting materials may take on various
physical shapes or geometry. Preextrusion SSP of
prepolymers uses starting materials in the form
of akes (or chips, mean diameter O1.0 mm) or
powders (mean diameter !100 mm), and has become
a common feature in the production of polymers for
industrial bers or molded products. If SSP is carried
out after the prepolymer shaping operation, e.g. in
bers or thin lms, it is termed as postextrusion
SSP. The latter, still at the development stage,
potentially offers several advantages over the tra-
ditional preextrusion SSP, because the geometries
used in postextrusion SSP (bers or thin lms) have
at least one dimension much smaller than chips,
and the condensate can be more effectively removed
[28,32,48,52,53].
Fig. 2. Scheme of the two-phase model. o: by-products or
oligomers. Reproduced from Mallon et al. [51] by permission of
John Wiley and Sons, Inc., New York.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 15
2. SSP techniques
2.1. Types of reactor and main characteristics
for batch and continuous processes
SSP may be carried out in glass tubes (Fig. 3a) [33,
34,54,55], uidized (Fig. 3b) and xed bed reactors
[36,56,57], rotating asks [20], tumbler dryers [19,22,
32,5860], a liquid inert medium (Fig. 3c) [3840,45,
46,61], vertical reactors with stirring blades [17],
rotating blenders [21], etc. Mechanical agitation is
usually provided in SSP systems, in order to facilitate
good heat and mass transfer, and to prevent
agglomeration, especially for SSP temperatures
above 200 8C [30,36,62]. The use of microwave
energy has been also applied in order to increase the
SSP rate, as a result of exciting and heating the
condensate in the polymer, and increasing diffusion
rates. For example, in the case of PA-6,6 SSP at
202 8C the diffusion coefcient of water is increased
from 1.09!10
K6
to 2.22!10
K6
cm
2
/s when using
microwaves [51,63]. In the past, the progress of an
SSP reaction was often estimated by continuously
weighing the reacting mass [4345,64] in order to
determine the amount of water formed. In current
technology, progress is often followed by an end
group analysis [65]. In the rst case, regarding PAs,
false DP may be assessed because of the separation of
volatile diamine along with water [44].
During SSP procedures, by-products are removed
by application of vacuum or through convection
caused by passing an inert gas, usually at atmospheric
pressure. Oligomers formed during the reaction may
pass by sublimation into the gas phase, along with the
condensate. Therefore, if the inert gas is recycled, the
system should contain apparatus to remove vaporous
reaction by-products and any atmospheric oxygen
(e.g. bag lters, gas washing, catalytic gas cleaning)
Fig. 3. Typical solid state polymerization reactors. (a) Glass tube. Reproduced from Li et al. [34] by permission of John Wiley and Sons, Inc.,
New York, (b) uidized bed reactor [36], (c) in liquid medium [38].
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 16
[22]. If a high vacuum is used in an SSP process, the
subsequent return to atmospheric pressure may result
in oxidation and discoloration of the polymer [16,58].
On the other hand, Ma et al. [66] showed that for PET,
SSP proceeds faster in vacuum than under a nitrogen
sweep, as a result of the faster removal of the by-
products and the formation of cyclic oligomers
(primarily trimers), the sublimation of which provides
an additional pathway for the progress of the reaction.
Finally, in the case of polyamide monomers, high
pressure SSP (196490 MPa) has been also applied to
provide well oriented polymers. However, such a
technique results in low polymerization rates, due to
the slow diffusion of water [67].
Among SSP systems, a uidized bed consumes a
great deal of energy and is not economically feasible,
vacuum process is preferred for small capacity, and a
xed bed is ideal for a continuous process. The xed
bed provides for large capacity, but it is difcult to
minimize the temperature gradient across the reactor,
and the method calls for a large amount of hot
nitrogen, which is an energy consuming process.
However, increased capacity averages out the invest-
ment and energy cost [62].
2.2. Use of inert gas in SSP systems
The use of the inert gas in SSP systems serves three
principal objectives: to remove the condensate, to
inhibit polymer oxidation by excluding oxygen from
the reactor atmosphere, and to heat the reacting mass.
In solid state monomer polyamidation, the use of the
inert gas is also associated with a suggested
mechanism, according to which the volatile com-
ponent of the PA-6,6 salt (i.e. hexamethylenediamine)
partially volatilizes at the reaction temperature,
thereby creating a vacancy defect, proper for the
nucleation of the generated polymer phase. The inert
gas contributes to this volatilization process and as a
result the vacancy defects and the number of
nucleation sites increase and so does the conversion
rate [68].
The inert gases used most often in SSP processes
are nitrogen (N
2
), carbon dioxide (CO
2
), helium (He)
[60,69], superheated steam [19] and supercritical
carbon dioxide (scCO
2
) [23,27,70,71]. In many
cases, the characteristics of the inert gas must be
taken into consideration, since they may inuence
the SSP process. For example, the use of ultra dry gas
characterized by a dew point below 30 8C may lead to
a signicant increase in the SSP rate [14], whereas it is
observed that steam in the carrier gas reduces the
overall rate, but also suppresses the degradation which
leads to colored products [19,22]. A dependence of
the nal molecular weight achieved on the inert gas in
PET prepared by SSP has been reported in some
studies. Devotta and Mashelkar [72] relate the
dependence of this inuence to the different rates of
absorption and solubilization of each gas in the
polymer mass, and thus there is a dependence on the
inert gas identity of the free volume available for
diffusion of by-products and reaction of end groups.
For example, due to its smaller molecular size, He has
a relatively higher diffusivity than either N
2
and CO
2
,
resulting in a higher free-volume growth rate and a
higher DP with use of He. Similarly, CO
2
can be
advantageously used instead of N
2
, because of its
higher solubility in the polymer, which can induce a
plasticization effect, and considerably enhance the
local mobility. This is also conrmed in the case in the
use of SSP to prepare poly(bisphenol A carbonate)
(BPA-PC) [23,27,73], where the use of scCO
2
resulted in higher reaction rates in comparison to N
2
and higher nal molecular weights, with a reaction
activation energy about half of the value with N
2
. This
effect, in conjunction with the positive impact of the
scCO
2
pressure on the SSP rate, appears to be a
consequence of the increased end group mobility due
to plasticization of the polymer by dissolved CO
2
and
of the easier condensate (phenol, Fig. 1c) removal due
to its solubility in the sweep uid. By contrast, Mallon
and Ray [69] found that using the right experimental
conditions, the type of inert gas apparently has no
signicant effect on the preparation of PET by SSP,
since equivalent molecular weights are achieved with
N
2
, CO
2
and He. Finally, if air is used for the removal
of by-products formed in SSP in the preparation of
PET, degradation offsets polymerization on exposure
to oxygen and moisture at 210 8C and above, resulting
in reduced or negative gains in molecular weight;
moreover, in such case, the carboxyl content in PET
product increases [62].
SSP systems may involve heating under continu-
ous inert gas ow (open system), where the by-
product removal is favored [19,20,27,32,34,55], or
under a stagnant inert gas atmosphere (closed system),
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 17
where the loss of monomers and oligomers is hindered
[18,45]this system was mostly applied in the
previous decades in monomer SSP. Alternatively, a
combination of these two systems may be used, with
heating rst carried out in an inert atmosphere,
followed by ow of an inert gas [74,75]. This
combination offers the advantages of both systems,
and provides maintenance of the monomers in the
reactor and satisfactory removal of the by-product in
order to favor the polymerization reaction.
3. Parameters affecting the SSP rate
As it involves both chemical and physical attri-
butes, SSP presents a complex reaction process. Based
on Eq. (1) and on the restrictions set by SSP nature,
one can identify several rate-determining steps [4,5,
19]:
The intrinsic kinetics of the chemical reaction
The diffusion of the reactive end groups
The diffusion of the condensate in the solid
reacting mass (interior diffusion)
The diffusion of the condensate from the reacting
mass surface to the inert gas (surface diffusion)
A large number of parameters are reported to affect
the SSP overall rate and dene the relevant rate-
determining step. In Section 3.1, an effort is made to
present these parameters and the relevant controlling
mechanisms, beginning with the ones considered to be
the most important. Accordingly, the discussion
presents reaction temperature, initial end group
concentration, particle geometry, gas ow rate and
crystallinity effects on the SSP rate; the use of
catalysts is also included. The dependence of the
most important operating variables on the controlling
mechanism in the absence of gas phase resistance, i.e.
when the surface by-product diffusion is not a limiting
factor, is presented briey in Table 2 [15,24].
3.1. Reaction temperature effect
The reaction temperature is probably the most
important factor in SSP, due to its interaction with
almost all other aspects of the process [76]. The role
of the SSP temperature is signicant, as it may
inuence the limiting step of the process and may
cause changes in the controlling mechanism [5,77,
78], with a resulting change of the reaction activation
energy [24]. Walas has pointed out that when the
effect of the reaction temperature change is intense,
the chemical reaction controls [79].
The dependence of the reaction temperature on the
intrinsic SSP rate constant is indicated by the values
of the SSP activation energy (E
a
), reported to be
between 10.581.5 kcal/mol in the case of PAs, and
15.042.5 kcal/mol for polyesters (Table 3). In
general, the reported SSP values are slightly higher
than those for melt processes [1,7,10,20,80]. With
PET, the polycondensation equilibrium constant (K
eq
)
used in SSP studies is similar with that for the melt
process [11]. For polyamides, the K
eq
is a function of
the reaction temperature. More specically, in the
case of SSP to prepare polycaproamide (PA-6) [31], at
given water content and for reaction temperatures in
the range of 170210 8C, the amide equilibrium
constant is described by Eq. (2), where the enthalpy
of reaction is K7.6 kcal/mol, revealing an exothermic
Table 2
Dependence of the most important variables on the controlling mechanism in the absence of gas phase resistance [15,24]
Controlling mechanism Parameters
Reaction temperature Particle size Prepolymer
MW and
crystallinity
Catalyst
concen-
tration
At low reaction tem-
peratures
At high reaction tem-
peratures
Chemical reaction Yes (strong inuence) Yes (weak inuence) No No Yes
End group diffusion Yes (weak inuence) Yes (strong inuence) No Yes No
Interior by-product diffu-
sion
Yes (weak inuence) Yes (strong inuence) Yes (strong inuence) Yes Yes
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 18
Table 3
Reaction kinetic data related to the irreversible SSP of polymers
Starting material Operating conditions Rate constant (k) Activation
energy
(kcal/mol)
Khripkov et al. [44] PA 6,6 salt 180188 8C,
08 h, N
2
6.332.5!10
K3
81.5
Catalytic reaction,
k: (g/mol) min
K1
9.243.8!10
K3
57.872.8
Oya et al. [45] 3-Aminocaproic acid 140185 8C, Two stages vs. reaction time, k: s
K1
030 h, vacuum First stage: 0.8536.40!10
K5
38.1
in liquid medium Second stage: 2.0310.1!10
K5
32.1
Griskey et al. [47] PA 6,6
(d
mean
Z0.18 cm)
90135 8C,
0K10 h, N
2
kZ1.53!10
10
exp(K12,960/RT)
k: h
K0.51
13.0
Chen et al. [4] PA 6,6
(d
mean
Z0.350.20 cm)
120180 8C,
520 h, N
2
kZ1.39!10
4
exp(K10,500/RT)
k: h
K0.5
10.5
PA 6,10
(d
mean
Z0.220.34 cm)
120180 8C,
520 h, N
2
kZ1.68!10
4
exp(K13,200/RT)
k: h
K1
13.2
Fujimoto et al. [33] PA 6,6
(d
mean
Z0.3 cm)
160210 8C,
080 h, N
2
log kZ13.8(5.90!10
3
/T),
k: h
K1
26.0
Srinivasan et al. [32] PA 6,6 bers 220250 8C,
04 h, N
2
kZ6.29!10
40
exp(K76,000/RT)
k: (g/mol)
2
s
K1
76.0
Srinivasan et al. [48] PA 6,6 bers 220250 8C,
04 h, N
2
2nd order: kZ3.06!10
18
exp(K42,000/RT) k: (g/mol) s
K1
42.0
3rd order: kZ1.18!10
31
exp(K61,000/RT) k: (g/mol)
2
s
K1
61.0
Gaymans et al. [55] PA 6
(d
mean
Z0.020.05 cm)
110205 8C,
124 h, N
2
For conversionsO30%:
kZ0.28
Chen et al. [4] PET
(d
mean
Z0.100.21 cm)
160200 8C,
520 h, N
2
kZ6.6!10
17
exp(K42,500/RT) k: h
K1
42.5
Duh [94] PET
(d
mean
Z106160 mm)
200230 8C,
020 h, N
2
kZ1.0287!10
9
exp(K0.0068[C]
0
23,565/RT) k: (kg/meq) h
K1
23.56
Duh [88] PET 190220 8C,
030 h, N
2
kZ653.044 exp(19,326/RT)
k: (kg/meq) h
K1
19.33
Duh [11] PET
(d
mean
Z425600 mm)
230 8C,
750 h, N
2
k
transest.
Z1.031.63!10
K3
(kg/meq) h
K1
k
ester.
Z1.071.38!10
K3
(kg/meq) h
K1
Jabarin et al. [116] Commercial PET (Good-
year VFR-6014, Firestobe
A, Eastman 7328)
200250 8C,
016 h, N
2
Catalytic reaction,
k: (g/mol) min
K0.5
18.423.2
Goodyear VFR-6014
kZ624!10
10
exp(K22,800/RT)
Firestobe A kZ828!10
10
exp(K23,200/RT)
Eastman 7328 kZ5.70!10
10
exp(K18,400/RT)
Kim et al. [24] Commercial PET (d
mean
Z
0.250.28 cm)
160230 8C,
012 h, N
2
160200 8C:
1935
200230 8C:
1516
Ma et al. [66] PET lms 250 8C, 6 h, Catalytic reaction
N
2
, vacuum k
transest.
Z0.0111 (kg/meq) h
K1
k
esterif.
Z0.0264 (kg/meq) h
K1
(continued on next page)
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 19
reaction. Another pertinent equation (Eq. (3)) has
been suggested by Kulkarni et al. [81]
log K
eq
Z
1761:4
T
K0:6614 (2)
K
eq
Zexp
3:9842 C
2:4877!10
4
T
R
(3)
where K
eq
is the equilibrium constant for polycon-
densation of PA-6, T is the absolute temperature and R
is the universal gas constant.
The temperature of SSP for monomers must be
high enough to facilitate chain growth but not so high
that it leads to partial melting with simultaneous
sticking, cyclization or other side reactions. In
general, nucleation and growth behavior is observed.
Of course, increased temperature, shortens the rst
stage of the SSP reaction, beyond which the
polymerization rate increases more intensively [45,
68,82]. In the case of SSP involving aminoacids [43]
and PA salts [68], the rate doubles with every 2 8C
increment of the reaction temperature. It is reported
that this high SSP temperature coefcient may be
related to the molecular mobility of the monomer and
to the number of active sites, which increase
signicantly with increasing temperature [83]. Fur-
thermore, it is observed that when the melting point
(T
m
) of the monomer is high, the temperature range of
SSP becomes larger, the SSP temperature coefcient
decreases and thus the impact of the reaction
temperature on the SSP rate is diminished [43].
Turning to prepolymers, an increase of the SSP
temperature accelerates the overall rate of the process
(Fig. 4) as a result of speeding up the chemical
reaction, the mobility of the functional end groups, the
by-product diffusion and the relevant mass transport
rates [5,32,33,47,48,57,76,8487]. According to Duh
Fig. 4. Effect of different reaction temperatures on

M
n
during PA 6
SSP in a xed bed reactor: (1) 220, (2) 200 and (3) 190 8C.
Reproduced from Xie [84] by permission of John Wiley and Sons,
Inc., New York.
Table 3 (continued)
Starting material Operating conditions Rate constant (k) Activation
energy
(kcal/mol)
Karayannidis et al.
[115]
PET lms 210240 8C,
08 h, vacuum
kZ1.83!10
10
exp(K99,600/RT)
k: (g/mol) s
K1
23.80
Shi et al. [71] PBA-PC
(d
mean
Z75125 mm)
120165 8C,
010 h, N
2
Catalytic reaction k
f
: h
K1
k
f
Z[C
1
]
0
kZ3.36!10
13
exp(K99,600/RT)
23.8
Shi et al. [27] PBA-PC
(d
mean
Z75125 mm)
90135 8C,
04 h, ScCO
2
,
(138345 bar)
Catalytic reaction k
f
: h
K1
138 bar: k
f
Z1.31!10
9
exp(K64,900/RT)
15.5
207 bar: k
f
Z1.21!10
7
exp(K48,400/RT)
11.6
345 bar: k
f
Z1.15!10
7
exp(K47,800/RT)
11.4
Sun et al. [100] PEN 200245 8C,
015 h, N
2
kZ1.78!10
6
exp(K7922/RT)
k: (g/mol) min
K0.5
7.9
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 20
[88], increased reaction temperature leads to a
decrease of the inactive end group concentration,
since some of these are rejected into the amorphous
phase as a result of the polymer crystallization at a
higher temperature. Although there are many dis-
crepancies on the optimum SSP temperature, it is
generally accepted that such a temperature should be
not too close to the T
m
to prevent particle agglomera-
tion [30]. Proposed SSP temperatures vary between
20 and 160 8C below the nal T
m
, with the usual
preferred temperature just below T
m
[3,19,51,77,84,
8991].
In many cases, SSP is carried out successively at
different temperatures. By this temperature-staging,
the problem of polymer grains agglomeration during
SSP may be overcome, because the prepolymer
softening temperature is thus increased and sticking
at higher reaction temperature is prevented [34,36,37,
51,62,63]. In addition, problems related to oligomer
formation [22] and to initial moisture and impurities
in the prepolymer [14,21,92] are also diminished.
3.2. Initial end group concentration effect
Based on the aforementioned two-phase model
proposed by Zimmerman [49,50] (Fig. 2), the rate of
SSP is expected to be higher than the one extrapolated
from melt equations at the same temperature, because
of the increased concentration of the chain end in the
amorphous regions of the polymer. However, this
does not happen, mainly because of end group
diffusion limitations, which do not exist in melt,
where the activation energy for diffusion is much
lower [50]. Thus, in the beginning of SSP, the active
end group distribution in the reacting sites would be
homogenous as in the case of melt process and
consequently the reaction kinetics and mechanisms
will be similar. As SSP proceeds further, the reacting
species close to each other are mostly reacted and
their concentration and distribution change locally
[50,55]. Then, end group diffusion limitations appear
and result, in many cases, in reducing the apparent
reaction rate constant [24,71,93] and reaching an
asymptotic value of molecular weight [27,66,71,89] at
long reaction times.
The diffusion of end groups comprises in many
cases the rate-controlling step, which is often
concluded experimentally by the effect of the initial
end group concentrations on the reaction rate.
Accordingly, it is generally observed that the lower
the concentration of the end groups (higher initial
number-average molecular weight M
n
0
), the higher is
the mean number-average molecular weight M
n
at the
end of the SSP reaction (Fig. 5) [19,55,86,94].
According to Gaymans et al. [55] and regarding SSP
for PA-6, this may be due to a more effective
connement of the amorphous phase and, therefore
high concentration and homogeneous distribution of
reactive chain end in the non-crystalline regions of the
higher M
n
0
prepolymers. The same author, in a
previous paper [19], attributes this effect also to the
fact that the higher M
n
0
prepolymers show less
tendency to crystallize during SSP and thus the end-
group mobility is less inhibited. A similar explanation
is given by Duh [88,94], who indicates that in a lower
M
n
0
prepolymer it is easier for polymer chains to t
into the crystal lattices and to form rigid crystals; as a
result, a greater number of end groups will be trapped
and become inactive. However, these researchers
have considered the SSP as a one-stage process and
studied the effect of M
n
0
on the SSP rate. A different
approach was made by Li et al. [95], who divided SSP
for PA-6 into two stages. In the rst reaction stage, the
end groups with the smallest end-to-end distances
react easily without needing to diffuse; thus, in this
initial stage, the lower the M
n
0
of the prepolymers, the
faster is the rate of the reaction because of the higher
end-group concentration. In the second reaction stage,
Fig. 5. Effect of prepolymer intrinsic viscosity (IV) on the progress
of PET SSP at 210 8C. Reproduced from Duh [94] by permission of
John Wiley and Sons, Inc., New York.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 21
the diffusion of the polymer chains (segmental
diffusion) starts to be the limiting step and faster
molecular weight growth is observed for samples with
higher M
n
0
for the reasons discussed above. On the
other hand, in the case of SSP of BPA-PC, it was
reported that the kinetics are independent of the initial
molecular weight with supercritical CO
2
as the sweep
uid, as long as the reaction temperature is much
higher than T
g
[27].
The effect of remelting on the SSP rate is also
associated with the end-group diffusion and may be
explained based on the previous theory: the procedure
of remelting, following some time after starting SSP, is
found to increase the polymerization rate [11,55,81,86,
95,96]. This effect is considered to be a result of the
homogenization/redistribution of the reactive end-
group separation and of the decrease of the prepolymer
water content, which take place during the remelting,
and which facilitate end group diffusion [55,93,95].
Besides the inuence of the initial molecular
weight of the prepolymer, the initial ratio of the end
groups also plays an important role. In the case of SSP
for PAs, and especially when the starting material is
salt, the volatile diamine escapes along with the
polycondensation water and therefore the amine end
groups should be in excess in order to compensate for
the loss and to maintain near stoichiometric equival-
ence [43,45]. Proposed ratios of initial amine to
carboxyl end-group concentrations ([NH
2
]/[COOH])
are in the range of 1.22 and 0.61.1 [17,58,97]. In the
case of PA prepolymers, the observed autocatalytic
effect of the carboxyl end-group in SSP [55]
implies that polymers with a large excess of
COOH over NH
2
polymerize much faster than in
the opposite case [50]. In the case of PET, it has been
supported by many researchers that there must be an
optimized ratio of initial carboxyl to hydroxyl end-
group concentrations ([COOH]/[OH]), because the
main reactions of esterication (reaction between
hydroxyl and carboxyl ends) and transesterication
(reaction between two hydroxyl ends) (Fig. 1b) are in
competition and the by-products formed (water and
ethylene glycol (EG), respectively) exhibit different
diffusivities D in the solid polymer, e.g. at 230 8C
D
EG
Z3.1!10
K6
and D
H2O
Z5.7!10
K6
cm
2
/s
[11,24]. Thus, Chang et al. [62] support that the
optimized ratio [COOH]/[OH] depends on the geo-
metry of the reacting particles and should vary
relatively. In the case of powdered PET, a high
hydroxyl end concentration is preferred, because this
way the diffusion of the main by-product (EG)
becomes easy; in the case of larger chip size, a high
carboxyl concentration is preferred in order to favor
esterication, since water diffusion is easier in
comparison to EG. The same conclusions were
deduced in a recent paper by Duh [11], who analyzed
many detailed data found in patent literature, and set
optimum values for carboxyl end group concen-
trations, depending on the controlling mechanism of
the reaction: in SSP of PET pellets, the lower
diffusivity of EG retards transesterication and, after
a certain reaction time, higher rates are observed in
higher carboxyl content prepolymers. Finally, the
[OH]/[COOH] ratio was found to play a very
important role in the case of the SSP of poly(butylene
terephthalate) (PBT). The higher this ratio (O0.5), the
more M
n
increases, while for small ratios, the process
can lead to only minor increases in the M
n
[98].
The role of the end group diffusion in the SSP
process has been also emphasized in simulation
models with a molecular basis [34,81,86,96,99],
through the denition of the characteristic migration
time of the functional end groups (q
t
), referred to a
specic reaction. In general, a low value of this
parameter means higher micro-level diffusivity of the
polymer molecule in the reacting mass, thereby
causing an increase in the rate constant of the relevant
reaction [34]. According to Kulkarni and Gupta [81],
three different values of q
t
must be used during the
modeling of SSP for PA-6, since three major
reversible reactions are involved in its kinetic scheme,
namely ring opening, polycondensation and polyaddi-
tion; the inuence of the change of these three q
t
on
the reaction rate is examined. The same approach was
made by Li et al. [34], in a study on the SSP of PA-6,6.
3.3. Particles geometry and gas ow rate effect
The presence of the polycondensation by-product
in the reacting mass may cause degradation reactions,
which explain the fact that in many SSP cases there is
a maximum value M
n
(for long reaction times), after
which the molecular weight reaches an asymptotic
value, or even starts decreasing [72,90,98]. In the case
of polyamides, it is anticipated that the initial moisture
content of the prepolymer pellets will affect SSP: an
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 22
increase of moisture content results in a decrease of
the DP, because higher initial water content promotes
the reversible polycondensation reaction towards the
left [99]. On the other hand, it is reported that the rate
is not greatly inuenced by a high initial amount of
water, because the latter is decreased to zero at the
rst stages of SSP due to the simultaneous drying
[3335].
Turning to reaction mechanism for melt polyami-
dation, the effect of the by-product concentration in
the reacting mass is also explained through the
reaction scheme presented in Eq. (4) [7]:
-COOH
C
H
2
N-4-COO
K
/
C
H3N-
4-CONH- CH
2
O
(4)
The rst step is similar to the salting reaction and the
second reaction is the rate-controlling step. The free
water greatly affects the dielectric constant of the
reaction mixture due to its high polarity, and a higher
dielectric constant makes the lefthand side of the
salting reaction more stable, and reduces the amount
of the intermediate involved in the rate-controlling
step. Therefore, the reaction rate is decreased with
increasing water concentration.
The effect of the by-product diffusion on the SSP rate
is generally more intense at high operating tempera-
tures, where the chemical reaction is no longer the
controlling step [5,24,93]. Aby-product diffusion effect
may be detected by varying the reacting particle size
and the owrate of the inert gas. If SSP is reaction-rate-
controlled, the DP is independent of particle size [27].
By contrast, particle size strongly inuences the overall
ratewhendiffusionof the by-product withinthe polymer
particle controls, with this inuence weakening when
the process is controlled by both diffusion and reaction
[15,86].
In the case of monomers SSP, the size of the
crystals of the reacting mass is disregarded by most
researchers; its effect is not signicant for grain sizes
below 2025 mesh [41].
In the case of prepolymers, it is generally accepted
that a smaller size of prepolymer particles can lead to
an increased SSP rate (Fig. 6), and consequently to a
decrease of the residence time, due to the shorter
diffusion distance and the larger particle surface area
per unit volume (ratio S/V) [1,94]. Under certain
reaction temperature conditions, the SSP rate is found
to increase when the diameter of polymer particles
decreases, indicating that diffusion of the by-product
through the solid polymer (interior mass transfer) is
rate-controlling [4,5,15,71,86,94,100]. More speci-
cally, particle size decrease may lead to a change of
the mechanism, as Huang and Walsh [78] observed in
the SSP of PET, where a shift from interior diffusion
control to surface diffusion control occurred on
reducing the polymer size. On the other hand, by-
product diffusion limitations may be totally neglected
when the half thickness (x) of the reacting particle
follows Eq. (5) [66]
x!

D
condens:
kC
condens:

(5)
where k is the reaction rate constant, x is the diffusion
distance, and D
condens.
and C
condens.
are the diffusivity
and concentration of the by-product in the semicrys-
talline polymer, respectively.
The effect of particle size on SSP rate is more
intense in the case of polyester prepolymers, while in
polyamides the reaction rate does not increase so
Fig. 6. Effect of particle size on SSP rate of PET at 230 8C.
Reproduced from Duh [94] by permission of John Wiley and Sons,
Inc., New York.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 23
dramatically with decreasing particle size. For
example, in the case of SSP for PET, a decrease of
the particle diameter from 0.266 to 0.14 cm results in
reducing the residence time by 56%, whereas the
relative decrease in SSP for PA-6,6 is only 3%. This is
attributed to the larger equilibrium constant for
polyamidation, which tolerates much higher conden-
sate concentrations in the particle without serious rate
reduction [1].
In addition, the surface by-product diffusion is
inuenced by the ow of the inert gas. Acceleration in
the gas ow can increase the mass and heat transfer
rates in the gas-solid system and decrease the
resistance to the diffusion of the by-product from
the particle surface into the bulk of the gas phase
[1,11,27,35,71,101]. It is reported that, at a given
reaction temperature, increasing the gas ow velocity
in the SSP of small-sized PET sample results in
changing the limiting step from surface diffusion
control to chemical reaction control [78]. On the other
hand, the effect of gas ow rate is not very signicant
in cases where the diffusion of condensate inside the
particles is the controlling step [27,57,66,84].
The role of the by-product diffusion in the SSP
process has been also emphasized in simulation
models, where three possible geometric situations
are considered. To the extent that polymer akes
resemble plane sheets or cylinders, and polymer
powders resemble spheres, the main by-product
diffusion equations are expressed by Eqs. (6)(8) [4,
5,102]
Plane sheets :
vC
condens:
vt
ZD
condens:
v
2
C
condens:
vx
2
(6)
Cylinders :
vC
condens:
vt
ZD
condens:
v
2
C
condens:
vx
2
C
1
x
vC
condens:
vx

(7)
Spheres :
vC
condens:
vt
ZD
condens:
v
2
C
condens:
vx
2
C
2
x
vC
condens:
vx

(8)
where x the distance in the direction of diffusion, and
D
condens.
and C
condens.
are the diffusivity and
concentration of the by-product in the semicrystalline
polymer, respectively.
It is important to emphasize that, during SSP, the
effect of the chemical nature of the reacting mass on by-
product diffusion is not greatly considered. However,
Papaspyrides et al. [3840,46,103] underlined the effect
of the hygroscopic nature of the reacting mass on SSP
for PAs, as derived from the water sorption mechanism
of polyamides. Thus, the role of polycondensationwater
becomes very important, since it was suspected that the
water formed had no tendency to diffuse out of the
reacting mass. More specically, the SSP of different
PAs salts was investigated by dispersing the monomer
particles in an inert non-solvent and using a glassware
assembly, which provided continuous monitoring of the
physical form of the reacting mass (Fig. 3c). It was
observed that the SSP was accompanied, depending on
the reaction conditions, by a distinct transition of the
process from the solid to the melt state (Solid-Melt
Transition, SMT), where a very fast agglomeration of
the reacting grains took place. The phenomenon was
readily seen macroscopically, since stirring failed to
keep the particles in suspension, while microscopically
the transformation of sharp-edged crystals to nearly
spherical particles was evident. Taking into account
these experimental ndings and the hygroscopic nature
of the polyamides, a generalized mechanism for the
effect of polycondensation water on reaction behavior
has been proposed (Fig. 7). The reaction begins at the
defective sites of the monomer crystalline structure,
being the active centers of the reaction (Fig. 7a). For
active centers up to or very near to the grain surface, the
water formed can be easily removed to the surrounding
heating medium, without affecting the reacting mass.
On the contrary, in the inner grain, the water cannot be
easily removed and hydrates the polar hydrophilic
groups of the salt structure. In the case of low reaction
rates (i.e. low rates of water formation) an organized
accommodation of the by-product within the crystal
structure is performed. As the accumulated amount of
water increases, a highly hydrated area of monomer
surrounds the active centers. This highly hydrated area
has a lower melting point and soon falls into the melt
state (Fig. 7b). After the formation of these melt areas,
the reaction proceeds mainly in the melt state, the rate is
considerablyincreased, andwater accumulationleads to
an increase of the total melt area (Fig. 7c). Eventually an
overlapping of these melt areas occurs, which explains
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 24
the observed transition of the reaction from the solid to
the melt state (Fig. 7d). As the reaction proceeds
further, the molecular weight increases, the hygrosco-
picity of the reacting system decreases, and nally the
solidcharacter of the systemis restored. At highreaction
rates, it is proposed that the time available for a
controlled accommodation of the water formed is very
limited, resulting in a more rapid breakdown of the salt
structure and subsequently in a faster appearance of the
SMT phenomenon. This proposed model of water
accumulationhydration-transition to the melt was
found topredominate inPAs salts of moderate structural
organization, while deviation from this model may
occur when the network of the coordinating polar
groups becomes more rigid (e.g. ethylenediammonium
fumarate).
3.4. Crystallinity effect
Contradictory opinions have been expressed con-
cerning the change of crystallinity during SSP.
According to Wu et al. [86], crystallinity can be
assumed constant in SSP. By contrast, Li et al. [34]
and Mallon and Ray [51] observed that the crystal
perfection and/or size gradually increase during the
SSP of PA-6,6 and PET. Kim et al. [24] observed that
PET crystallinity increases signicantly within the
rst few hours of SSP and then stabilizes, showing
only small variations with increased reaction time.
Crystallinity is considered to inuence the SSP
rate, because its role is essential in controlling critical
parameters of the reaction, such as the end-group and
the by-product diffusion. It is assumed that the size
and perfection of the crystalline lamellae and their
packing inuence the molecular mobility of the chain
end groups. For example, in well-crystallized semi-
crystalline polymers, the motion of chain segments is
restrained by the fact that almost all of them are
anchored in crystals [51,95]. In fact, the effect of
crystallinity on the SSP rate is two-sided. Again,
accepting the two-phase model [49,50], in which the
end groups are assumed to be expelled in the
amorphous phase (Fig. 2), an increase of crystallinity
leads to higher concentration of end groups rejected in
the amorphous phase and thus to an increase of the
reaction rate [15,51,86]. On the other hand, as SSP
proceeds, the mobility of the polymer chains is
believed to decrease because of the crystallinity
increase [28], which also hinders the escape of by-
products from the reacting mass; therefore diffusivity
decreases with increasing crystallinity [34]. More-
over, when the degree of crystallinity is high, a
considerable fraction of macrochains are immobilized
and protected against attack of low molecular weight
by-products formed during condensation, while inter-
action between end groups is favored [87]. Summar-
izing the above, the contrary effects of crystallinity
can be understood with respect to the SSP kinetic
mechanisms: in by-product diffusion limited reac-
tions, high crystallinity reduces the SSP rate by
imposing higher resistance to mass transfer, whereas
in chemical reaction-controlled process, high crystal-
linity results in increasing SSP rate because of the
concentration of end groups in the amorphous regions
[24]. It has been suggested that for optimum behavior,
the reacting particles should have a sufciently high
crystallinity to prohibit particle agglomeration [1,37];
specically, Wu et al. proposed a value of 40% [86].
The effects of crystallinity are included in the SSP
kinetics models and simulation mainly through
equations based on the concentrations of
Fig. 7. Schematic diagram of the solid-melt transition phenomenon
(SMT). ($), Defects of the monomer crystalline structure; Dark area,
Polymer nuclei insoluble in water; Shaded area, Highly hydrated
and eventually melt area. Reproduced from Kampouris et al. [38] by
permission of Elsevier Science LTD, Oxford, UK.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 25
the noncrystallizing components (i.e. end-groups, by-
products (Eq. (9)) and the by-product diffusion (Eq.
(10)) [48,51,90,95]
C
amorphous
Z
C
overall

1 Kx
c
(9)
D
condens:
Z1 Kx
c
D
0;condens:
(10)
where x
c
is the mass fraction crystallinity, [C
amorphous
]
and [C
overall
] are the concentrations of a component in
the amorphous phase of polymer and the total mass of
the polymer, respectively, and D
condens.
and D
0,condens
are the diffusivities of the by-product in the semi-
crystalline polymer and the completely amorphous
polymer, respectively.
In addition to the effects of the crystallinity of the
prepolymers, the method by which crystallization
develops may also inuence the SSP process. Crystal-
lized prepolymers are produced either by solvent
evaporation or by crystallization from the melt [104].
In the second technique, the structures of thermally
crystallized polymers are strongly affected by the
crystallization process and present a memory of the
thermal history, i.e. temperatures and times of
crystallization, cooling rates, etc [105]. Therefore, in
cases of SSP with thermally crystallized prepolymers,
the reaction progress may be inuenced by the
crystallization history, which may effect the structure
characteristics of the SSP products.
Turning to the solvent induced crystallization
processes, it seems that the latter are preferable
when particulate characteristics are of highest
importance. Thus, in cases of by-product diffusion-
controlled SSP processes, for which the reacting
particles size plays a signicant role, solvent
induced technique seems preferable, since it can
give a specic crystal size distribution, crystal
shapes and structures of the prepolymer. The latter
cannot be fully achieved in crystallizing by cooling
from the melt. Here, the polymer molecules exhibit
strong competition with each other and add to a
specic crystal surface simultaneously, rendering the
situation much more chaotic [106]. On the other
hand, when purity of the SSP product is essential,
crystallization from the melt would be preferable, in
order to avoid any contamination of the product
by he solvent. Finally, it should be mentioned that
the environmental benets of SSP are evident for
thermally crystallized prepolymers, e.g. PET, but the
advantage is lost with solvent induced prepolymers,
like polycarbonates. In fact, the solvents known to
induce crystallinity, e.g. acetone, are difcult to
handle in a large scale commercial plant, and their
removal is very difcult. However, the use of the
proper gas, e.g. scCO
2
, can compensate the use of
organic solvents, inducing crystallization in the
polycarbonate prepolymer [27,70,71,73].
3.5. Use of catalysts
The presence of catalysts in the preparation of
polyamides and polyesters has an accelerating effect
on the rate of polymerization in both solid and melt
phase reactions. The disadvantages of SSP are mainly
overcome with the use of catalysts, which aim to
increase the reaction rate and avoid particles agglom-
eration [103]. Therefore, acidic, basic and neutral
compounds have been examined for their catalytic
action. In the case of acid catalyzed polyamidation,
the mechanism proposed involves the following
equilibria [2,107]:
-COOHCH
C
4-COOH
C
2
(11)
COOH
C
2
CH
2
N-4-CONH- CH
C
CH
2
O (12)
However, there is not a unique proposed mechan-
ism to explain the catalytic action in SSP processes.
For example, Khripkov et al. [108] conclude that
noncatalytic SSP of PA-6,6 salt, involves reactions
between the end-groups of monomers and the
propagating polymer chains, with initiation and
propagation reactions carried out in the defective
parts of small salt crystals. In catalytic processes, the
presence of linear oligomers is reported after a short
time of SSP reaction; thus, they assume that the
growth of polymer chain is achieved not only with the
reaction between the monomer and the propagating
polymer chain, but also between the oligomers
themselves. On the other hand, Papaspyrides et al.
[46,103] correlated the effect of catalysts to the
proposed mechanism of Solid-Melt Transition in the
PAs salts. It was proposed that the presence of a good
catalyst in the reacting structure contributes to an
easier removal of the water formed in the reaction
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 26
(Fig. 8). In other words, hydration seems restricted
and the diffusion of the water is favored so that the
right-hand reaction is encouraged. It should be
mentioned that the use of SSP catalysts currently
constitutes a signicant research area, since a uniform
catalysis mechanism has not yet been suggested, and
most catalysts are used empirically.
The catalysts most often used in the SSP of
monomers are presented below. For the SSP of
aminoacids, the effectiveness of the catalytic action
of several compounds follows the order: 1% H
3
BO
3
,
0.2% MgOO0.5% (COONH
4
)
2
O0.5% (CH
3
COO)
2-
ZnO0.2% Na
2
CO
3
O0.6% CH
3
COOHO0.5%
(NH
4
)
2
SO
4
O1% SnCl
2
[43]. For the SSP of PA-6,6
salt, reported catalysts include: H
3
BO
3
O(COOH)
2
O
H
3
PO
4
OMgO, and Na
2
CO
3
, NaHSO
4
and (SiO
2
)
n
are
proved to be inactive [44]. For polyoxamidation,
compounds belonging to groups IVb and Vb of the
periodic table of elements were found to act as
catalysts: SbF
3
wAs
2
O
3
[GeO
2
OSb
2
O
3
OBi
2
O
3-
wPbO [30].
In the SSP of prepolymers, the addition of easily
diffusing acidic compounds (e.g. H
3
PO
4
, HBO
3
,
H
2
SO
4
) leads to higher reaction rate, while in the
absence of them the reaction rate is limited by the
diffusion of the autocatalyzing acid chain end groups
[55]. In the SSP of PAs, the catalysts used are mainly
phosphorous compounds, such as 2-(2
0
pyridyl) ethyl
phosphonic acid (PEPA), sodium (SHP) and manga-
nous hypophosphite [14,37]. Moisture in the pellets is
considered to deactivate the catalyst, therefore
evaporation of water from the surface should be
encouraged by the use of a low dew point inert gas
[14].
The use of thermoplastic polyurethane [109] and of
sterically hindered hydroxylphenylalkylphosphonic
ester or monoester (such as Irganox
w
1222 and
1425) [110] has been also proposed so as to catalyze
SSP reactions.
Reported techniques to incorporate the catalyst in
the SSP starting materials may vary, and their
effectiveness depends on whether the catalyst is
thoroughly dispersed in the SSP reacting mass.
Thus, the catalyst may be added in a PA salt solution
during the prepolymer production [14,58], or alter-
natively into the prepolymer melt, by injection into a
low relative viscosity melt prior to pelletizing [14].
Other techniques suggest impregnation of the pre-
polymer in a catalyst solution [111] or melt-blending
the prepolymer with a masterbatch containing the
catalyst [16] before carrying out SSP. Finally, in the
case of the SSP of PA salts, a nucleationcoprecipi-
tation technique is proposed, in which the catalyst is
introduced in the solution of the reactants and
precipitates together with the PA salt [103,112,113].
4. Kinetics and simulation of SSP
One can roughly classify the literature of SSP
kinetics and simulation into two areas. In the rst,
emphasis is given primarily on the determination of
intrinsic or apparent rate constants of the various main
and side reactions, assuming no diffusion limitations.
The second places emphasis on the modeling and
simulation of the various physical processes that
occur simultaneously with chemical reaction in large-
scale reactors. Both areas are rapidly growing, and
future activity is expected, since the major task in the
kinetic study of solid state polymerization is to
determine the optimum operating conditions and
establish a kinetic model useful for reactor design.
4.1. Reaction kinetics of SSP processes
Despite the fact that the chemical kinetics of SSP
in polyesters and polyamides has been the subject of
numerous studies, there is no universal agreement on
the relevant chemical kinetic expressions. In general,
the kinetic models developed to date are classied
Fig. 8. Solid state polyamidation of hexamethylenediammonium
adipate at 142 8C. , pure salt; , salt containing 1.60% w/w
boric acid, (,), conversion; (:), water accumulation parameter.
Drawn after data from Ref. [46].
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 27
into two groups: the Flory theory-based models and
the power-law models. According to Florys theory
[114], the kinetics can be described as either second-
(Eq. (13)) or third- (Eq. (14)) order, assuming equal
reactivity of the functional end groups. This principle
is most accurate for the nal stages of polymerization,
for which the molecular weight is high. Thus, it
becomes evident that in the SSP of prepolymers,
which excludes the very early stages of polymeriz-
ation, the kinetic expressions can be written in terms
of functional group concentrations [102]. In the case
of polyamides, the rate equations are
2nd order : K
dCOOH
dt
Zk
2
COOHNH
2
(13)
3rd order : K
dCOOH
dt
Zk
3
COOH
2
NH
2
(14)
where k
2
(kg/meq h) and k
3
(kg
2
/meq
2
h) are the 2nd
and 3rd order rate constants, respectively, and
[COOH] and [NH
2
] are the concentrations (meq/kg)
of carboxyl and amine end groups, respectively.
Obviously, third order kinetics indicate that one of
the functional groups exhibits catalytic behavior, and
thus its effect on polymerization must be included in
the rate equation. The kinetics during the melt
polyesterication, studied by Flory, showed that 2nd
order kinetics refer to polymerizations catalyzed by a
small amount of a strong acid catalyst, the concen-
tration of which is included in the rate constant k
2
. In
these cases, the 2nd order course continues up to a
degree of polymerization of 90%, while in uncata-
lyzed polymerizations, 3rd order is followed only for
conversions higher than 80% (8093%) [114].
Many different but almost equivalent sets of
integrated equations have been developed to describe
Florys theory, and are presented below for the
irreversible SSP reaction, since the hydrolysis rate
may be assumed negligible due to the low reaction
temperatures (Table 3). In addition, in many studies of
intrinsic SSP kinetics, only data for short reaction
times are used, to exclude limitations regarding end
group diffusion [27,66,71].
In particular, Gaymans et al. [19] studied the SSP
of unbalanced PA-4,6 and they proposed a kinetic
expression in terms of the product of the concen-
trations of the end groups (P) and of the reaction order
(n) (Eq. (15)). In case of a balanced starting material,
the apparent reaction order is assessed by the increase
of the number-average molecular weight versus
reaction time. Under the specic experimental limits,
the SSP reaction did not follow 3rd order kinetics, but
the apparent order varied between 3.5 and 5.2,
revealing that with decreasing reaction temperature,
the apparent order increases
K
dCOOH
dt
Zk

COOH$NH
2

n
Zk

P
p

n
0
1

P
t
p

nK1
K
1

P
0
p

nK1
Zn K1kt (15)
where k is the rate constant (kg
nK1
/meq
nK1
h),
[COOH]
t
and [NH
2
]
t
are the concentrations (meq/kg)
of the end groups at any given time t, [COOH]
0
and
[NH
2
]
0
are the initial concentrations (meq/kg) of the
end groups, P
0
Z[COOH]
0
$[NH
2
]
0
and P
t
Z
[COOH]
t
$[NH
2
]
t
, (meq/kg)
2
.
In the study of the SSP of PA-6,6 bers [48],
assuming a balanced prepolymer and a polydispersity
index (PDI) equal to 2 during SSP, a rate expression in
terms of the reaction order (n), of the viscosity
average molecular weight (M
v
) and of the degree of
crystallinity (x
c
) has been developed (Eq. (16)) based
on the two-phase model and Eq. (9). The relevant
data gave equally good ts with 2nd and 3rd order
reaction kinetics
K
dC

dt
ZkC

n
0M
v

nK1
KM
v
0

nK1
Z
n K12
nK1
1 Kx
c

nK1
kt (16)
where k is the reaction rate constant (g
nK1
/mol
nK1
h),
[C*] is the concentration (mol/g) of end groups in the
noncrystalline phase of polymer, M
v0
is the initial
viscosity average molecular weight, and M
v
is the
viscosity average molecular weight at any given time
t, g/mol.
Turning to PET, Duh [11] determined the intrinsic
rate constants for esterication and transesterication
in the case of PET powder SSP. The relevant rate
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 28
expressions are
R
transest:
Z2k
transest:
OH
2
02k
transest:

t
0
OH
2
ZKDC K2DCOOH (17)
R
ester:
Z2k
ester:
OHCOOH02k
ester:

t
0
OH
!COOH
ZK2DCOOH (18)
where k
ester.
and k
transest.
are the esterication and
transesterication rate constants ((kg/meq h), respect-
ively, [OH] and [COOH] are the hydroxyl and
carboxyl end group concentrations (meq/kg), respect-
ively, D[C] and D[COOH] are the differences
between the total and carboxyl end group concen-
trations (meq/kg), respectively, of the prepolymer and
nal product.
The product of the end-group concentrations was
expressed as exponential function of reaction time t
(Eq. (19)), and the relevant constants (a
0
, a
1
, a
2
) were
determined by curve tting. Eq. (19) was used in the
integration of the relevant rate expressions (Eq. (17)
and (18)), to calculate the rate constants
OH
2
or OHCOOH Za
0
Ca
1
exp K
t
a
2

(19)
Duh [88,94] also developed a simple semiempi-
rical rate equation (Eq. (20)), assuming that SSP is
reaction-controlled and that only transesterication
occurs. According to his studies, two categories of end
groups exist: active and inactive groups. The inactive
end groups include chemically dead end groups and
functional end groups that are rmly trapped in the
crystalline structure and cannot participate in the
reaction. It was suggested that the overall SSP
followed a 2nd order kinetics. The proposed rate
equation contains two parameters: the apparent
reaction rate constant (k
a
) and the constant apparent
inactive end group concentration [C
ai
], found to
decrease linearly with the SSP temperature. Finally, in
the Arrhenius equation of the rate constant (k
a
), the
frequency factor was found to be an exponential
function of the initial total end group concentration
[C]
0
.
K
dC
dt
Z2k
a
C KC
ai

2
0
C
0
KC
t
t
Z2k
a
C
0
KC
ai
C
t
K2k
a
C
0
KC
ai
C
ai
(20)
where k
a
is the apparent 2nd order rate constant
(kg/meq h), [C]
0
and [C]
t
are the total end group
concentrations (meq/kg) initially and at time t,
respectively, and [C
ai
] is the constant apparent
inactive end group concentration (meq/kg).
Karayannidis et al. [115] considered only transes-
terication during the SSP of amorphous and
unoriented PET lms and veried the rate equation,
for which M
n
increases linearly with time. On the
other hand, during PET SSP, Ma et al. [66] considered
both the transesterication and esterication reactions
and using relevant software, they solved the kinetic
expressions in Eqs. (21)(23). In the same paper, they
also modied the above equations according to the
aforementioned Duh theory [94] and estimated new
rate constants based on inactive end group concen-
trations
dOH
dt
ZK4k
transest:
OH
2
Kk
ester:
COOH
!OH (21)
dCOOH
dt
ZKk
ester:
COOHOH (22)
M
n
Z
2!10
6
COOH COH
(23)
where k
transest.
and k
ester.
are the rate constants
(kg/meq h) for transesterication and esterication,
respectively, [OH] and [COOH] are the concen-
trations of hydroxyl and carboxyl end groups
(meq/kg), and M
n
is the mean number-average
molecular weight (g/mol).
Shi et al. [27,71] investigated the kinetics of the
SSP of BPA-PC based on the Flory theory, using only
the data at short reaction times, and assuming that the
rate is controlled by transesterication. They devel-
oped a 2nd order rate expression (Eq. (24)), written in
terms of end group concentrations, fractional conver-
sion (p) of the stoichiometrically decient end group
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 29
and fractional molar excess of the other group (3)
K
dC
1

dt
ZkC
1
C
2

0
1
3
ln
1 C3 Kp
t
1 Kp
t
Kln
1 C3 Kp
0
1 Kp
0

ZC
1

0
k
2
t
(24)
where k
2
is the 2nd order kinetics rate constant,
(kg/meq h), [C
1
]
t
and [C
2
]
t
are the concentrations
(meq/kg) of the stoichiometrically decient end-
groups and the end groups in excess at time t,
respectively, with initial values [C
1
]
0
and [C
2
]
0
,
respectively, p
0
the fractional conversion in the
prepolymerization stage, and
3 Z
C
2

0
KC
1

0
C
1

0
(25)
p
t
Z
C
1

0
KC
1

t
C
1

0
(26)
As far as the power-law models are concerned, a
widely used equation is that of Walas [79], who
pointed out that the rate of a process in a solid
material, which is controlled by the chemical reaction
and diffusion, usually varies as some power of the
time (n) (Eq. (27))
rate Zkt
n
(27)
Griskey and Lee [47] used a modied form of
Walas equation, which assumes that the number-
average molecular weight (M
n
) in solid state polym-
erization varies as a power of the time equal to K0.49
for the SSP of PA-6,6. For the analysis of their kinetic
data on the SSP of PA-6, Gaymans et al. [55]
suggested that the process is limited by the diffusion
rate of the autocatalyzing acid end and that this is not
only dependent on the concentration and the tem-
perature, but also on the changing pair-wise distance
distribution among end-groups. Therefore, they
developed a kinetic expression which associates the
rate of reaction with the concentration of the
catalyzing end groups [COOH], and to a power n of
time (Eq. (28)), combining thus the Flory theory with
the Walas equation. It was found that the SSP kinetics
had more than one region, and for conversions higher
than 30% the reaction rate was 1st order in the
carboxyl end-group concentration, and reciprocal to
the reaction time (nZK1)
K
dCOOH
dt
Zk COOH t
n
0ln K
dCOOH
dt
=COOH
t

Zln k Cn ln t (28)
where k the reaction rate constant (1/h
nC1
) and
[COOH]
t
is the carboxyl end group concentration at
time t (meq/g).
Chen and Chen [93] also used both Flory theory-
based rate equations and power-law equations to
determine the effective rate constants of esterication
and ester interchange reaction, which take place
during the SSP of PET. The Flory theory-based rate
equations they used were 2nd and 3rd order for the
esterication and interchange reaction, respectively.
The order of the power-law equations used was K2 in
both cases. According to Jabarin and Lofgren [116],
who investigated the kinetics of SSP of PET for a
variety of conditions, the mean number-average
molecular weight at any time of SSP (M
n
) is related
to the square root of the reaction time (

t
p
), and is
given through a simple empirical equation. The same
kinetic approach was also used by Kim at al. [24] in
PET SSP and by Sun at al. [100] in the case of
poly(ethylene naphthalate) (PEN) SSP.
Finally, Fujimoto et al. [33] formulated a power-
law rate equation during their study of the SSP of PA-
6,6 based on the relative viscosity (RV), found
experimentally from the ratio of the viscosity of a
solution of 8.4% (by weight) polymer in a solution of
90% formic acid to the viscosity of the formic acid
solution. It was found that the RV of their samples
increased linearly with heating time during SSP.
However, it should be emphasized that such an
expression serves only as a tool to get some idea of
the rheological behavior of the polymer during SSP.
According to Zimmerman [50], if the course of the
reaction was followed by RV measurements rather
than by end-group concentrations, the results could
become confusing because of the possibility of
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 30
branching reactions, since the viscosity will not
usually increase as fast with increasing molecular
weight for a branched chain as for a linear one.
4.2. Modeling and simulation of SSP processes
Because of the industrial importance of SSP,
mathematical modeling and process simulation have
been employed to gain a better understanding of the
relevant mechanisms and to predict the inuence of
different parameters on the SSP rate. The majority of
these modeling studies involve the solution of the full
system of partial differential equations (PDEs), which
describe the change with time and position of all
chemical species within the particle. In these models,
each species balance contains the rates of all reactions
in which the species participates, and each condensate
molecule balance contains an additional term for
diffusion of the condensate within the particle. These
models contain a signicant number of physicochem-
ical parameters (rate constants, diffusion coefcients,
etc.), one or more of which must be adjusted to t the
experimental data to the model. Such models have
been developed for SSP of a variety of polymers,
belonging to the families of both polyesters and
polyamides, including PET, PBT, BPA-PC, PA-6 and
PA-6,6.
Regarding PA-6, Kaushik and Gupta [96]
explained some limited experimental data quite well
and predicted qualitative trends observed experimen-
tally in the SSP of PA-6 chips with intermediate
remelting. More specically, they set critical values
for the water diffusivity inside the reacting particles,
for the particles surface water concentration and for
the particle radius for which no considerable increase
of

M
n
occurs. Kulkarni and Gupta [81] later devel-
oped an improved model, using the Vrentas-Duda
theory for diffusion coefcients. The effects of
changing the important operating conditions on SSP,
e.g. intermediate remelting of PA-6 powder, water
concentration in the vapor phase, minimizing the
monomer and water contents before SSP, size and
degree of crystallinity of polymer particles, etc. have
been studied. The same effects are also considered in
the comprehensive model for the SSP of PA-6
developed by Li et al. [95]. Xie [99] also developed
a model for SSP of PA-6, which explores the effect of
different parameters on number-average chain length
and polydispersity index.
Yao et al. [35,117] developed a reactor model to
describe the SSP of PA-6,6 in a moving bed and to
simulate process start-up, shut-down and different
disturbances during operation. A model of the SSP of
PA-6,6 has also been developed by Li et al. [34], in
which the variations of molecular weight and water
content at different positions inside the chip indicated
that the major polymerization occurs in a thin shell
near the periphery of the chip, with the molecular
weight at the core increasing more slowly because of
the limited diffusion of water. Furthermore, in
agreement with experimental data, based on their
simulation, Yao and Ray [1] found that the residence
time in the SSP reactor increases rapidly with a
decrease of the DP in the feed line, and that the size of
PA particles does not have a considerable effect on the
SSP rate.
Turning to PET, a general model to describe
continuous SSP reactors has also been applied to a
moving packed bed reactor case study. Reactor
temperature and the condensate (ethylene glycol and
water) concentration in the inert gas both affected the
nal DP, while variable crystallinity and gas phase
mass transfer effects were also studied [51,57]. Wang
and Deng [85] proposed a comprehensive model for
the SSP of PET to account for the inuence of reactive
chain mobility on reaction rate, the diffusion of volatile
by-products (ethylene glycol, water and acetaldehyde)
and the effect of crystallinity. The effect of crystallinity
was also brought out by a mathematical model for the
SSP of PET, developed by Devotta and Mashelkar
[72]. This work was a continuation of the work of
Ravindranath and Mashelkar [15], who used a
mathematical model for simulation of industrial SSP
process of PET to predict the inuence of particle
shape, size, temperature, etc. on the polycondensation
process for all the operational regimes (reaction rate-,
diffusion- or diffusion and reaction rate-controlled
process). In order to analyze the mechanismof the SSP
process for PET, Gao et al. [101] proposed a semi-
analytical method, which showed that the overall
reaction rate for a single PET pellet can be appro-
priately simulated by a model controlled jointly by
diffusion and reaction rates. The same conclusion was
reached by Tang et al. [118] who proposed a single
dimensional model to seek a solution for the SSP
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 31
process and analyze its mechanism. They also found
that fromthe core to the surface, the SSP rate increases
monotonically due to a gradual reduction of the
concentration of by-products. Another model, devel-
oped by Goodner et al. [77] assumes that diffusion of
the reaction condensate in the solid polymer is the rate-
limiting step in the overall polymerization kinetics.
This equilibrium model provides an upper bound on
molecular weight and its rate of increase, as well as a
useful tool for understanding the effects of temperature
and particle size. Aquantitative prediction in the effect
of temperature, particle size, starting molecular weight
and ratio of end-groups on the DP is also possible using
a comprehensive model developed by Wu et al. [86],
based on an analysis of the similarities and differences
between solid state and melt polymerizations. More-
over, the effects of temperature and chain entangle-
ment on chain mobility were considered by Kang [119]
to estimate the rate constants of nine chemical
reactions occurring during the SSP of PET. Kang
applied the free volume theory to diffusion of the
volatile by-products, and formulated an ordinary
differential equation to calculate the mass transfer
rate of these. Finally, Kim et al. [120], based on Kang
simulation [119], suggested a model, to predict the
concentrations of hydroxyl, carboxyl, vinyl ester
end groups and terephthalic acid monomer in the
resulting PET.
Gostoli et al. [89] modeled the SSP of PBT, taking
into account ve chemical reactions, and the diffusion
of three volatile species. They found that the
molecular weight increases and reaches a maximum,
which strongly depends on the water diffusivity. At
longer times an asymptotic value is reached due to the
acidolysis reaction, in competition with thermal
degradation. A few years later, based on the data
obtained through their rst model, the same research-
ers [98] reported a quantitative mathematical descrip-
tion for the SSP of PBT, which helped them to analyze
the inuence of the initial number-average molecular
weight, the initial ratio between hydroxyl and
carboxyl end group concentrations and the sample
thickness. The models of Wang et al. [85], Devotta et
al. [72], Ravindranath et al. [15] and the model of
Kulkarni et al. [81] can also be used to analyze the
SSP of PBT.
The models developed by Goodner et al. [77] and
by Devotta et al. [72] can be used for analyzing
the SSP process for BPA-PC. In addition, Goodner et
al. [90] developed a model describing the melt phase
polycondensation kinetics of polycarbonate, which
was then applied to the SSP of an individual spherical
particle. The model includes the diffusion of the by-
product (phenol) in the particle, and tracks the
evolution of the molecular weight as a function of
both the process conditions (time, temperature, and
initial crystallinity and molecular weight) and the
properties of the system (rate constants, equilibrium
constant and phenol diffusion coefcient).
5. SSP and molecular weight distribution
Using a statistical method and assuming equal
reactivity of the functional groups, and the absence of
intramolecular reaction or other side reactions, Flory
[114,121] came to the conclusion that during a typical
linear step-polymerization the molecular weight
distribution (MWD) widens and the polydispersity
index (PDI) reaches a value of 2 when the fractional
conversion (p) is close to 1. On the other hand,
Korshak [3] claims narrowing of the MWD due to
degradation reactions, to which the longest molecules
are especially liable. In particular, for SSP reactions,
the interaction between the reaction kinetics and the
MWD should be emphasized. At very low values of
the forward kinetic constant (k
f
), the reaction is
kinetically limited, the condensate produced quickly
diffuses out of the polymer particle and the overall
polydispersity barely increases. If the forward reac-
tion kinetics is increased, diffusion limitations start to
manifest, causing a radial condensate concentration
gradient in the particle. This gradient causes a
gradient in the local MW, which in turns leads to a
broadened MWD over the particle as a whole, as
evidenced by the drastic increase of the overall
polydispersity [77].
In agreement with the aforementioned approaches,
different polymerization conditions resulting in a
variety of MWD data are presented in the following.
Feldmann and Feinauer [122] investigated the SSP of
polylaurolactam (PA-12) and reported an unstable,
broad MWD. In their study on SSP of PA-4,6 in a
uidized bed reactor, Gaymans and Schuijer [19]
concluded that the SSP process is not susceptible to
MWD broadening. On the other hand, Fakirov [28]
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 32
found that SSP of PA-6 bers under vacuum resulted
in narrowing of the molecular weight distribution.
Furthermore, Kulkarni and Gupta [81] found that
during the SSP of PA-6 the average PDI rises rapidly
at rst, and then increases more gradually to values
above 2.0. In the SSP of PET, Cha [123] found a
broader than most probable MWD. Mallon and Ray
[51] also reported the variability of molecular weight
in a PBT particle after SSP, which causes the overall
polydispersity to increase dramatically, and men-
tioned that similar results should be expected for PET.
It must be emphasized at this point that the shape of a
distribution function usually depends on the polym-
erization conditions as well, which differed in all of
the examples above.
Meyer reported that broadening of the MWD may
be due to the fact that equilibrium is not achieved in a
typical SSP process and, as a result, reactions of
additional polycondensation keep occurring [124].
Zimmerman [49] attributes the same effect to a
microscopic nonuniformity of the solid reagent,
where the polymer chains present a distribution of
crystallite sizes. Furthermore, the observed variation
of M
n
with location in the solid pellet of PA-6 leads to
values of PDI above the normal value of 2.0 [81].
Especially for polyesters, Mallon and Ray [51]
mention that the diffusion of the by-product is
sufciently slow and a particles outer regions have
lower local concentrations of the diffusant than do the
interior points. This feature for reactions close to
equilibrium means that the effective polymerization
rate at the exterior is greater, and the nonuniformity in
average M
n
across the particle radius leads to
increased polydispersity. This is also the case for
BPA-PC [73], for which the broadening of MWD is
attributed to the slow diffusion of the by-product
(phenol) inside the polymer particle.
Another explanation for the broadening of MWD,
based on the operating problems of the different kinds
of reactors used during SSP, is given by Mallon and
Ray [57], as follows: In the case of a moving packed
bed reactor, the uneven ow of polymer particles
leading to a variation in residence times, the uneven
ow of purge gas resulting in differing mass transfer
rates, and the uneven radial temperature distribution
across the reactor giving a range of particle reaction/
diffusion rates and gas-solid mass transfer rates, lead
to spread of molecular weights. In the case of
a uidized bed reactor, the very broad residence
time distribution results in a very broad MWD under
reaction-controlled conditions. However, with appro-
priate operation changes, these problems may be
overcome somewhat. For example, in the case of PET,
one can use a purge gas that contains increased
condensate level, decreasing in this way the variation
on the molecular weight inside a particle, since areas
of high MW will be inhibited from further polym-
erization while areas of low MW will still polymerize.
An explanation for the reported MWD narrowing
is based on the fact that certain prepolymers contain
relatively large amounts of monomer and oligomers.
During SSP, these volatile residues are eliminated
along with the reaction by-product, leading to
narrowing of the MWD [42,99].
Based on the considerations given above, the more
important parameters mentioned to affect the evol-
ution of MWD during SSP may be highlighted. First
of all is the particles size, a decrease of which leads to
narrowing of the MWD [73,90,99], as a result of
enhanced by-product diffusion. For example, in the
case of BPA-PC SSP [73], the PDI of powder
remained just below 2, meanwhile the value for
polymer beads (d
mean
Z3.6 mm) was as high as 2.6.
Secondly, the stoichiometry is a signicant factor:
according to Goodner et al. [125], as the rate of
polymerization decreases due to the nonstoichio-
metric ratio of end groups, more time is available
for condensate removal from the polymer particles,
and its concentration gradient inside the particles
becomes less severe. So, if a less polydisperse
material is needed and reducing the particle size is
not an option, stoichiometry may be used to minimize
the adverse effects of slow mass transfer. Xie [99] also
reported that, in the case of PA-6, the PDI decreases
with an increase of the reaction temperature, and/or an
increase of the initial water concentration, and/or a
decrease of the initial polymerization degree of the
polymer, and/or an increase of the concentration of
the monomer.
6. Conclusions
Solid State Polymerization (SSP) may be applied
with polycondensation polymers, e.g. polyamides and
polyesters, in order to increase the degree of
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 33
polymerization and to improve the quality of the end
product. The most important commercial advantages
of SSP focus on the use of easy and inexpensive
equipment, and on avoiding some of the drawbacks of
conventional polymerization processes.
A signicant amount of literature on SSP has been
accumulated during the last 50 years and may be
classied to experimental and theoretical data includ-
ing simulation aspects: in both cases the scope is to
assess the importance of each possible process factor,
to suggest a relative mechanism, to predict SSP
behavior and nally to optimize the overall procedure.
Because of the complexity of the process, different
experimental conditions lead to different conclusions,
often contradictory, which reveal that the chemical
and physical process involves a variety of reaction
parameters. This article provides a background to
determine the impact of each parameter and to
correlate it to the reaction mechanism. Among these
parameters, the reaction temperature may be high-
lighted as the most important, since the temperature
coefcient of the SSP reaction is rather high. The
particles size, crystallinity, initial molecular weight of
the SSP starting material and the polycondensation
by-product formed greatly inuence the reaction rate
and determine the controlling step of the process.
On the other hand, it seems that the central issue
today involves means to overcome the main drawback
of SSP, i.e. the slow rate compared to melt
polymerization. Therefore, the use of catalysts in
SSP comprises an important research topic, with
breakthroughs needed to devise methods to obtain a
uniform catalysis mechanism. This will result in
further industrial application,and boost the use of SSP
based processes. Of course, research concerning all
other aforementioned aspects of SSP should be
continued and enhanced in the future to fully under-
stand and establish this promising technology.
Acknowledgements
The authors wish to thank E. Barabouti and
S. Christou for their signicant contribution to the
literature survey and Christos Tsenoglou for his help
and support. We would like to thank also Joe Weber
and David Marks for useful discussions within the
frame of an on-going research interaction between
NTUA and INVISTA, Inc. (former DuPont Textiles
and Intermediates, Inc.).
References
[1] Yao K, Ray W. Modeling and analysis of new processes for
polyester and nylon production. AIChE J 2001;47(2):40112.
[2] Gaymans R, Sikkema D. Aliphatic polyamides. In: The
comprehensive polymer science, vol. 5. Oxford: Pergamon
Press; 1989. p. 35773.
[3] Korshak V, Frunze T. Synthetic hetero-chain polyamides.
Jerusalem: IPST; 1964 p. 87,442.
[4] Chen F, Griskey R, Beyer G. Thermally induced solid state
polycondensation of nylon 66, nylon 610 and polyethylene
terephthalate. AIChE J 1969;15(5):6805.
[5] Chang T. Kinetics of thermally induced solid state poly-
condensation of poly(ethylene sterephthalate). Polym Eng
Sci 1970;10(6):3648.
[6] Samant K, Ng K. Synthesis of prepolymerization stage in
polycondensation processes. AIChE J 1999;45(8):180827.
[7] Mallon F, Ray H. A comprehensive model for nylon melt
equilibria and kinetics. J Appl Polym Sci 1998;69:121331.
[8] Steppan D, Doherty M, Malone M. A kinetic and equilibrium
model for nylon 6,6 polymerization. J Appl Polym Sci 1987;
33:233344.
[9] Ogata N. Studies on polycondensation reactions of nylon salt.
I. The equilibrium in the system of polyhexamethylene
adipamide and water. Makromol Chem 1960;42:5265.
[10] Ravindranath K, Mashelkar R. Finishing stages of PET
synthesis: a comprehensive model. AIChE J 1984;30(3):
41522.
[11] Duh B. Effects of the carboxyl concentration on the solid-
state polymerization of poly(ethylene terephthalate). J Appl
Polym Sci 2002;83:1288304.
[12] Peli T, Davis T. Diffusion and reaction in polyesters melts.
J Polym Sci: Polym Phys Ed 1973;11:167183.
[13] Kosky P, Guggenheim E. Solid-state polymerization in
poly(butylene terephthalate): the equilibrium constant.
Polym Eng Sci 1985;25(18):11457.
[14] Dujari R, Cramer G, Marks D. Method for solid phase
polymerisation (E.I. du Pont de Nemours and Company) WO
Patent 98/23666; 1998.
[15] Ravindranath K, Mashelkar R. Modeling of poly(ethylene
terephthalate) reactors. IX. Solid state polycondensation
process. J Appl Polym Sci 1990;39:132545.
[16] Heinz H, Schulte H, Buysch H. Process for the manufacture
of high molecular weight polyamides (Bayer AG) EP Patent
410,230/91 A2; 1991.
[17] Shimizu K, Ise S. Polyhexamethyleneadipamide with
restricted three-dimensional formation and process for the
manufacture (Asahi Chemical Industry Ltd) JP Patent 4-
93323; 1992.
[18] Gaymans R, Van Utteren T, Van den Berg J, Schuyer J.
Preparation and some properties of nylon 46. J Polym Sci:
Polym Chem Ed 1977;15:53745.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 34
[19] Gaymans R. Polyamidation in the solid phase. In:
Henderson JN, Bouton CT, editors. Polymerization reactors
and processes. ACS Symp Ser; 1979.
[20] Roerdink E, Warnier J. Preparation and properties of high
molar mass nylon-4,6: a new development in nylon
polymers. Polymer 1985;26:15828.
[21] Sheetz H. Sold phase polymerisation of nylon (DSM N.V)
US Patent 5,461,141; 1995.
[22] Weger F, Hagen R. Method and apparatus for the production
of polyamides (Karl Fischer Industrieanlagen Gmbh) US
Patent 5,773,555; 1998.
[23] Gross S, Roberts G, Kiserow D, DeSimone J. Crystallization
and solid state polymerization of poly(bisphenol A carbon-
ate) facilitated by supercritical CO
2
. Macromolecules 2000;
33:405.
[24] Kim T, Lofgren E, Jabarin S. Solid-state polymerization of
poly(ethylene terephthalate). I. Experimental study of the
reaction kinetics and properties. J Appl Polym Sci 2003;89:
197212.
[25] Flory P. Polymerization process (E.I. du Pont de Nemours
and Company) US Patent 2,172,374; 1939.
[26] Monroe G. Solid phase polymerisation of polyamides (E.I. du
Pont de Nemours and Company) US Patent 3,031,433; 1962.
[27] Shi C, DeSimone J, Kiserow D, Roberts G. Reaction kinetics
of the solid-state polymerization of poly(bisphenol A
carbonate) facilitated by supercritical carbon dioxide.
Macromolecules 2001;34:774450.
[28] Fakirov S, Avramova N. Inuence of thermal treatment,
molecular weight and orientation on the mechanical proper-
ties of polyamide-6. Acta Polym 1982;33:2715.
[29] Morawetz H. Polymerization in the solid state. J Polym Sci
1966;Part C(12):7988.
[30] Bruck S. New polyoxamidation catalysts. Ind Eng Chem
Prod Res Dev 1963;2(2):11921.
[31] Mizerovskii L, Kuznetsov A, Bazarov Y, Bykov A. Equili-
brium in the system polycaproamidecaprolactamwater
below the melting point of the polymer. Polym Sci USSR
1982;24(6):131026.
[32] Srinivasan R, Desai P, Abhiraman A, Knorr R. Solid-state
polymerisation vis-a`-vis ber formation of step-growth
polymers. I. Results from a study of nylon 66. J Appl
Polym Sci 1994;53:173143.
[33] Fujimoto A, Mori T, Hiruta S. Polymerization of nylon-6,6 in
solid state. Nippon Kagaku Kaishi 1988;(3):33742.
[34] Li L, Huang N, Liu Z, Tang Z, Yung W. Simulation of solid-
state polycondensation of nylon-66. Polym Adv Technol
2000;11:2429.
[35] Yao K, McAuley K, Berg D, Marchildon E. A dynamic
mathematical model for continuous solid-phase polymeris-
ation of nylon 6,6. Chem Eng Sci 2001;56:480114.
[36] Beaton D. Continuous, solid-phase polymerization of poly-
amide granules (E.I. du Pont de Nemours and Company) US
Patent 3,821,171; 1974.
[37] Blanchard E, Cohen J, Iwasyk J, Marks D, Stouffer J, Aslop
A, Lin C. Process for preparing polyamides (E.I. du Pont de
Nemours and Company) WO Patent 99/10408; 1999.
[38] Kampouris E, Papaspyrides C. Solid state polyamidation of
nylon salts: possible mechanism for the transition solid-melt.
Polymer 1985;26:4137.
[39] Papaspyrides C. Solid-state polyamidation of nylon salts.
Polymer 1988;29:1147.
[40] Papaspyrides C. Solid state polyamidation. In: Salamone JC,
editor. The polymeric materials encyclopedia. Boca Raton,
FL: CRC Press; 1996. p. 781931.
[41] Pilati F. Solid-state polymerization. Comprehensive polymer
science. vol. 5. New York: Pergamon Press; 1989 p. 20116.
[42] Fakirov S. Solid state reactions in linear polycondensates. In:
Solid state behavior of linear polyesters and polyamides.
Englewood Cliffs, NJ: Prentice Hall; 1990 [Chapter 1].
[43] Volokhina A, Kudryavtsev G, Skuratov S, Bonetskaya A.
The polyamidation process in the solid state. J Polym Sci
1961;53:28994.
[44] Khripkov E, Kharitonov V, Kudryavtsev G. Some features of
the polycondensation of hexamethylene diammonum adipi-
nate. Khim Volokna 1970;6:635.
[45] Oya S, Tomioka M, Araki T. Studies on polyamides prepared
in the solid state. Part I. Polymerisation mechanism.
Kobunshi Kagaku 1966;23(254):41521.
[46] Papaspyrides C. Solid state polyamidation processes. Polym
Int 1992;29:2938.
[47] Griskey R, Lee B. Thermally induced solid-state polymeris-
ation in nylon 66. J Appl Polym Sci 1966;10:10511.
[48] Srinivasan R, Almonacil C, Narayan S, Desai P,
Abhiraman A. Mechanism, kinetics and potential morpho-
logical consequences of solid-state polymerization. Macro-
molecules 1998;31:681321.
[49] Zimmerman J. Equilibria in solid phase polyamidation.
J Polym Lett 1964;2:9558.
[50] Zimmermasn J, Kohan M. Nylon-selected topics. J Polym
Sci: Part A: Polym Chem 2001;39:256570.
[51] Mallon F, Ray W. Modeling of solid-state polycondensation.
I. Particle models. J Appl Polym Sci 1998;69:123350.
[52] Knorr R. High tenacity nylon yarn (Monsanto Company) US
Patent 5,073,453; 1991.
[53] Almonacil C, Desai P, Abhiraman A. Morphological
consequences of interchange reactions during solid state
polymerization in oriented polymers. Macromolecules 2001;
34:418699.
[54] Fortunato B, Pilati F, Manaresi P. Solid state polycondensa-
tion of poly(butylene terephthalate). Polymer 1981;22:
6557.
[55] Gaymans R, Amirtharaj J, Kamp H. Nylon 6 polymerization
in the solid state. J Appl Polym Sci 1982;27:251326.
[56] Gaymans R, Venkatraman V, Schuijer J. Preparation and
some properties of nylon-4,2. J Polym Sci: Polym Chem
1984;22:137382.
[57] Mallon F, Ray W. Modeling of solid state polycondensation.
II. Reactor design issues. J Appl Polym Sci 1998;69:
177588.
[58] Hosomi H, Kitamura K. Ultra-high-molecular-weight poly-
hexamethyleneadipamides (Asahi Kasei Kogyo K.K.) JP
Patent 1-284525; 1989.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 35
[59] Fujimoto A. Drying of nylon pellets [12]. Nippon Kagaku
Kaishi 1988;(5):8026.
[60] Muller K, Eugel F, Gude A. Process for the preparation of
polyamide powders processing high viscosities (Chemische
Werke Huls Aktiengesellschaft) US Patent 3,476,711; 1969.
[61] Dolden J, Harris G, Studholme B. Catalytic suspension/dis-
persion preparation of polyamides from polyamide forming
salt (BP Chemicals Ltd) US Patent 4,925,914; 1990.
[62] Chang S, Sheu M, Chen S. Solid-state polymerization of
poly(ethylene terephthalate). J Appl Polym Sci 1983;28:
3289300.
[63] Mallon F, Ray W. Enhancement of solid-state polymeriz-
ation with microwave energy. J Appl Polym Sci 1998;69:
120312.
[64] Winslow F, Matreyek W. Pyrolysis of crosslinked styrene
polymers. J Polym Sci 1956;22:31524.
[65] Walz J, Taylor G. Determination of the molecular weight of
nylon. Anal Chem 1947;19:448.
[66] Ma Y, Agarwal U, Sikkema P, Lemstra P. Solid-state
polymerisation of PET: inuence of nitrogen sweep and high
vacuum. Polymer 2003;44:408596.
[67] Ikawa T. Nylons (by high pressure solid-state polycondensa-
tion). In: Salamone JC, editor. The polymeric materials
encyclopedia, vol. 6. Boca Raton, FL: CRC Press; 1996. p.
468994.
[68] Zeng H, Feng L. Study of solid-state polycondensation of
nylon 66 salt. Gaofenzi Tongxun 1983;5(5):3217.
[69] Mallon F, Beers K, Ives A, Ray W. The effect of the type of
purge gas on the solid-state polymerisation of polyethylene
terephthalate. J Appl Polym Sci 1998;69:178991.
[70] Gross S, Flowers D, Roberts G, Kiserow D, DeSimone J.
Solid-state polymerisation of polycarbonates using super-
critical CO
2
. Macromolecules 1999;32:31679.
[71] Shi C, Gross S, DeSimone J, Kiserow D, Roberts G. Reaction
kinetics of the solid state polymerisation of poly(bisphenol A
carbonate). Macromolecules 2001;34:20624.
[72] Devotta I, Mashelkar R. Modelling of polyethylene tereph-
thalate reactorsX. A comprehensive model for solid-state
polycondensation process. Chem Eng Sci 1993;48(10):
185967.
[73] Gross S, Roberts G, Kiserow D, DeSimone J. Synthesis of
high molecular weight polycarbonate by solid state polym-
erisation. Macromolecules 2001;34:391620.
[74] Wlloth F. Solid state preparation of polyamides (Vereinigte
Glasstoff-Fabriken A.G.) US Patent 3,379,696; 1968.
[75] Papaspyrides C, Vouyiouka S, Bletsos I. Preparation of
polyhexamethyleneadipamide prepolymer by a low tempera-
ture process. J Appl Polym Sci 2004;92:3016.
[76] Yoon K, Kwon M, Jeon M, Park O. Diffusion of ethylene
glycol in solid state poly(ethylene terephthalate). Polym J
1993;25(3):21926.
[77] Goodner M, DeSimone J, Kiserow D, Roberts G. An
equilibrium model for diffusion-limited solid-state polycon-
densation. Ind Eng Chem Res 2000;39:2797806.
[78] Huang B, Walsh J. Solid-phase polymerization mechanism of
poly(ethylene terephthalate) affected by gas ow velocity
and particle size. Polymer 1998;39(26):69919.
[79] Walas S. Uncatalysed heterogenous reactions. In: Reaction
kinetics for chemical engineers. New York: Mc Graw Hill;
1959. p. 12630.
[80] Plazl I. Mathematical model for industrial continuous
polymerisation of nylon 6. Ind Eng Chem Res 1998;37:
92935.
[81] Kulkarni M, Gupta S. Molecular model for solid-state
polymerization of nylon 6. II. An improved model. J Appl
Polym Sci 1994;53:85103.
[82] Yamazaki T, Kaji K, Kitamaru R. Polymerization kinetics on
the thermo induced solid state polycondensation of 3-
aminocaproic acid and nylon 66 salt. Bull Kyoto Univ
Educ Ser B 1983;63:5363.
[83] Bagramayants B, Bonetskaya A, Yenikolopyan N. The
reason for the high temperature coefcient in solid state
polycondensation of u-aminoacids. Vysokomol Soyed 1966;
8(9):15948.
[84] Xie J. Kinetics of the solid-state polymerization of nylon-6.
J Appl Polym Sci 2002;84:61621.
[85] Wang X, Deng D. A comprehensive model for solid-state
polycondensation of poly(ethylene terephthalate): combining
kinetics with crystallization and diffusion of acetaldehyde.
J Appl Polym Sci 2002;83:313344.
[86] Wu D, Chen F, Li R, Shi Y. Reaction kinetics and
simulations for solid-state polymerisation of poly(ethylene
terephthalate). Macromolecules 1997;30:673742.
[87] Kuran W, Debek C, Wielgosz Z, Kuczynska L, Sobczak M.
Application of a solid-state postpolycondensation method for
synthesis of high molecular weight polycarbonates. J Appl
Polym Sci 2000;77:216571.
[88] Duh B. Semiempirical rate equation for solid state polym-
erization of poly(ethylene terephthalate). J Appl Polym Sci
2002;84:85770.
[89] Gostoli C, Pilati F, Sarti G, Di Giacomo B. Chemical kinetics
and diffusion in poly(butylene terephthalate) solid-state
polycondensation: experiments and theory. J Appl Polym
Sci 1984;29:287387.
[90] Goodner M, Gross S, DeSimone J, Roberts G, Kiserow D.
Modeling and experimental studies of the solid state
polymerization of polycarbonate facilitated by supercritical
carbon dioxide. Polym Prepr (Am Chem Soc, Div Polym
Chem) 1999;40(1):978.
[91] Kumar A, Gupta S. Simulation and design of nylon 6
reactors. J Macromol Sci Rev Macromol Chem Phys 1986;
26(2):183247.
[92] Kosaka M, Muranaka Y, Wakatsuru K. Process for preparing
aromatic polyamides (Mitsui Chemicals Inc.) EP Patent
0916687 A1; 1999.
[93] Chen S, Chen F. Kinetics of polyesterication III: Solid-state
polymerization of polyethylene terephthalate. J Polym Sci:
Part A: Polym Chem 1987;25:53349.
[94] Duh B. Reaction kinetics for solid-state polymerisation of
poly(ethylene terephthalate). J Appl Polym Sci 2001;81:
174861.
[95] Li L, Huang N, Tang Z, Hagen R. Reaction kinetics and
simulation for the solid-state polycondensation of nylon 6.
Macromol Theory Simul 2001;10:50717.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 36
[96] Kaushik A, Gupta S. A molecular model for solid-state
polymerisation of nylon 6. J Appl Polym Sci 1992;45:50720.
[97] Silverman B, Raleigh N, Stewart L. Process for producing
ultrahigh molecular weight polyamides (Monsanto Com-
pany) US Patent 3,562,206; 1971.
[98] Pilati F, Gostoli C, Sarti G. A model description for
poly(butylene terephthalate) solid-state polycondensation.
Polym Process Eng 1986;4(2-4):30319.
[99] Xie J. Kinetics and simulation of solid state polymerisation
for nylon 6. Ind Eng Chem Res 2001;40:31527.
[100] Sun Y, Shieh J. Kinetic and property parameters of
poly(ethylene naphthalate) synthesized by solid-state poly-
condensation. J Appl Polym Sci 2001;81:205561.
[101] Gao Q, Nan-Xun H, Zhi-Lian T, Gerking L. Modelling of
solid state polycondensation of poly(ethylene terephthalate).
Chem Eng Sci 1997;52(3):3716.
[102] Secor R. The kinetics of condensation polymerisation.
AIChE J 1969;15(6):8615.
[103] Katsikopoulos P, Papaspyrides C. Solid-state polyamidation of
hexamethylenediammonium adipate. II. The inuence of acid
catalysts. J Polym Sci: Part A: Polym Chem 1994;32:4516.
[104] Davey R, Garside J. From molecules to crystallizers. An
introduction to crystallization. New York: Oxford Science
Publications; 2002 p. 12-14.
[105] Strobl G. The physics of polymers. Concepts for under-
standing their structures and behavior. 2nd ed. Berlin:
Springer; 1997 p. 14445.
[106] Gedde U. Polymer physics. London: Chapman & Hall; 1995
p. 14751.
[107] Hiemenz P. Polymer chemistry. The basic concepts. New
York: Marcel Dekker; 1984 p. 306.
[108] Khripkov E, Lavrov B, Kharitinov V, Kudryavtsev G. Some
problems in solid-phase polycondensation of hexamethyle-
nediammonium adipate. Vysokomol Soedin Ser B 1976;
18(2):825.
[109] Brink A, Owens J. Thermoplastic polyurethane additives for
enhancing solid state polymerisation rates (Eastman Chemi-
cal Company) WO 99/11711; 1999.
[110] Pfaendner R, Hoffman K, Herbst H. Increasing the molecular
weight of polycondensates (Ciba-Geigy AG) WO 96/11978;
1996.
[111] Endo Y, Ibata J, Fujimoto T. Manufacturing method of
polyamide (Asahi Chemicals Industries Co) JP Patent 48-
23199; 1973.
[112] Papaspyrides C. Inuence of metal catalysts on solid state
polyamidation of nylon salts. Polymer 1986;27:143740.
[113] Pipper G, Mueller W, Dauns H. Process for the production of
linear polyamides (BASF Aktiengesellschaft) EP Patent
0,455,066 A1; 1991.
[114] Flory P. Principles of polymer chemistry. Ithaca, NY: Cornell
University Press; 1975 p. 7583, see also pages 31725.
[115] Karayannidis G, Sideridou I, Zamboulis D, Stalidis G,
Bikiaris D, Lazaridis N, et al. Solid-state polycondensation of
poly(ethylene terephthalate) lms. Die Angew Makromolek
Chemie 1991;192:15568.
[116] Jabarin S, Lofgren E. Solid state polymerization of poly(-
ethylene terephthalate): Kinetic and property parameters.
J Appl Polym Sci 1986;32:531535.
[117] Yao K, McAuley K. Simulation of continuous solid-phase
polymerisation of nylon 6,6 (II): Processes with moving bed
level and changing particle properties. Chem Eng Sci 2001;
56:532742.
[118] Tang Z, Gao Q, Huang N-X, Sironi C. Solid-state
polycondensation of poly(ethylene terephthalate): kinetics
and mechanism. J Appl Polym Sci 1995;57:47385.
[119] Kang C. Modeling of solid-state polymerization of poly(-
ethylene terephthalate). J Appl Polym Sci 1998;68:83746.
[120] Kim T, Jabarin S. Polymerization of poly(ethylene tereph-
thalate). II. Modeling study of the reaction kinetics and
properties. J Appl Polym Sci 2003;89:21327.
[121] Munari A, Manaresi P. Molecular weight distributions. In:
Bevingtion JC, Allen GC, editors. The comprehensive
polymer science, vol. 4. Oxford: Pergamon Press; 1989. p.
4762.
[122] Feldmann R, Feinauer R. Postcondensation of polylaurolac-
tam in solid phase. Angew Makromol Chem 1973;34(460):
17.
[123] Cha C. The molecular-weight distribution of poly(ethylene
terephthalate) from solid-phase polycondensation. Polym
Prepr (Am Chem Soc, Div Polym Chem) 1965;6:84.
[124] Meyer K. Post condensation of polyamides in the partially
crystalline state. Angew Makromol Chem 1973;34(519):
16575.
[125] Goodner M, Gross S, DeSimone J, Roberts G, Kiserow D.
Broadening of molecular-weight distribution in solid-state
polymerisation resulting from condensate diffusion. J Appl
Polym Sci 2001;79:92843.
S.N. Vouyiouka et al. / Prog. Polym. Sci. 30 (2005) 1037 37