The Design of Continuous Polycondensation Reactors

University of Massachusetts Amherst
ScholarWorks@UMass Amherst
Doctoral Dissertations 1896 - February 2014
1-1-1989
The design of continuous polycondensation reactors/

David Daniel Steppan
University of Massachusetts Amherst
Follow this and additional works at: https://scholarworks.umass.edu/dissertations_1
Recommended Citation
Steppan, David Daniel, "The design of continuous polycondensation reactors/" (1989). Doctoral
Dissertations 1896 - February 2014. 752.
https://doi.org/10.7275/eyxq-fp50 https://scholarworks.umass.edu/dissertations_1/752
This Open Access Dissertation is brought to you for free and open access by ScholarWorks@UMass Amherst. It
has been accepted for inclusion in Doctoral Dissertations 1896 - February 2014 by an authorized administrator of
ScholarWorks@UMass Amherst. For more information, please contact scholarworks@library.umass.edu.
THE DESIGN OF CONTINUOUS
POLYCONDENSATION REACTORS
A Dissertation Presented
By
DAVID DANIEL STEPPAN
Submitted to the Graduate School of the

University of Massachusetts in partial fulfillment
of the requirements for the degree of
DOCTOR OF PHILOSOPHY
February 1989
Polymer Science and Engineering

(c) Copyright by David Daniel Steppan 1989
All Rights Reserved

THE DESIGN OF CONTINUOUS
POLYCONDENSATION REACTORS
A Dissertation Presented
By
DAVID DANIEL STEPPAN
Approved as to style and content by:
Michael F. Doherty, Co-Chairman of Committee
Michael F. Malone, Co-Chairman of Committee
Robert L . Laurence, Member
David A. Hoagland, Member
William tf. MacKnighfe, Department Head

Polymer Science and Engineering
To my parents
iv
ACKNOWLEDGEMENT
I would like to express my sincerest gratitude to Dr. Michael F.
Doherty and Dr. Michael F. Malone. They have taught me all facets
of research and I am very grateful. I can only hope that this
research has been as satisfying for them. I wish them both the best
of luck in all their personal endeavors because they do not need it
professionally.
I also extend my thanks to my thesis committee for their help
and suggestions. In addition, I thank all the people who helped me
in the course of this work including the members of the Design and
Control Center and fellow group members. The financial support of
the Dupont Company is also acknowledged.
Lastly, I would like to express my thanks to my family and
friends. They are the most important people in my life and this
thesis is also their accomplishment.
v
ABSTRACT
THE DESIGN OF CONTINUOUS POLYCONDENSATION

REACTORS
FEBRUARY 1989
DAVID STEPPAN, B. S., CASE WESTERN RESERVE UNIVERSITY
Ph. D., UNIVERSITY OF MASSACHUSETTS
Directed by: Professor Michael F. Doherty
and Professor Michael F. Malone
A simplified approach to polycondensation kinetics has been used
to develop engineering models for the design of film-forming
polycondensation reactor-separators
This approach has been applied to a nylon 6,6 prepolymer in an
idealized flowing film device. The model is characterized by two
dimensionless groups; the Damkohler number and the Thiele modulus.
A realistic kinetic and equilibrium description was incorporated
with an activity-based kinetic model with composition-dependent
apparent rate and equilibrium constants. The model is suitable for
evaluating the effect (s) of changes in throughput, flow, film
thickness, temperature and catalyst reactivity on the device
performance.
Since nylon 6,6 is typically used for fibers, molecular weight
is not the only important product property. It must also have the
vi
proper balance of end-groups for good dyeability and very small
amounts of gelled material for fiber spinning. A simplified thermal
degradation model that is consistent with available data has been
formulated to predict the amount of degraded material in the final
product. The importance of the residence time distribution on the

amount of degradation in the flowing film device has demonstrated.
Finally, a realistic wiped film model for the design and

performance analysis of continuously mixed thin-film nylon 6,6
"finishing" reactors has been constructed. The model incorporates
composition dependent rate and equilibrium constants, a realistic
degradation scheme and a finite gas-phase concentration. Even small
amounts of mixing have been shown to yield very large improvements
in both the mass transfer and molecular weight generation in such
film reactors.
vii
TABLE OF CONTENTS
ACKNOWLEDGEMENT
v
ABSTRACT
VI
LIST OF TABLES
XI
LIST OF FIGURES ..
NOMENCLATURE **• •••••••••••••••••••••••»••••••• XIX.

•
Chapters
1 INTRODUCTION !
1.1 General 1
1.2 Previous Work 2
1.3 A Simplified Approach 5
1.4 Research Objectives 7
2 A SIMPLIFIED KINETIC, EQUILIBRIUM AND DEGRADATION MODEL

FOR NYLON 6,6 10
2.1 Kinetic and Equilibrium Model 10
2.1.1 Introduction 10
2.1.2 A Nonideal Liquid Model 14
2.1.3 Equilibrium Correlation 15
2.1.4 Amidation Reaction Rate Correlation 23
2.1.5 Discussion 27
2.2 Degradation Model 31
2.2.1 Background and Degradation Schemes 31
2.2.2 Experimental Degradation Studies 36
2.2.3 Degradation Model 45
viii
DIFFUSION EFFECTS IN NYLON
6,6 POLYMERIZATION 54
3 .
1 Introduction. . .
54
3.2 Preliminaries...
- _ „ . 56
3.3 Constitutive Relations
3.4 Stationary Film Model
3.5 Model Predictions..
65
3.6 Conclusions
86
4 A FLOWING FILM MODEL FOR NYLON

6,6 POLYMERIZATION 87
4.1 Introduction
87
4.2 Flowing Film Model
9Q
4.3 Model Predictions
93
4.4 A Design Example -
11
4.5 Conclusions „,
115
5 DYNAMIC BEHAVIOR OF THE FLASHER AT HIGH TEMPERATURE

118
5.1 Introduction
11 g
5.2 Incorporation of the Degradation Scheme
119
5.3 Model Predictions for Changes in Throughput
121
5.4 Dynamic Predictions
12 5
5.5 Discussion and Conclusions t 129
6 A WIPED FILM MODEL FOR NYLON 6,6 POLYMERIZATION 131
6.1 Introduction 131

6.2 Wiped Film Polymerization Reactor Model 133
6.3 Model Predictions 139
6.4 Conclusions 156
7 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK 157
ix
REFERENCES CITED
LIST OF TABLES
Table 2-1. Nylon 6,6 degradation reactions

33
Table 2-2. Nylon 6,6 crosslinking reactions

37
Table 2-3. Nylon 6,6 degradation and crosslinking model

47
Table 2-4. Kinetic parameters for nylon 6,6 degradation

51
Table 3-1. Typical values of the Thiele modulus 65
Table 4-1. Typical values of the Damkohler number 93
Table 6-1. End-group concentration vs. molecular weight for

nylon 6,6 polymerization without degradation 147
xi
LIST OF FIGURES
Figure 1-1. Continuous process for nylon 6,6 (U.S.

Patent; 3, 900,450)
3
Figure 2-1. Apparent enthalpy of reaction vs. mole

fraction water
^
Figure 2-2. Apparent equilibrium constant at 200 °C vs.
mole fraction water
20
Figure 2-3. Apparent equilibrium constant vs. mole

fraction water 21

fraction water 22

fraction water 24
Figure 2-6. Apparent rate constant at 200 °C vs. mole

fraction water 28
Figure 2-7. Comparison of the simulation and the data

of Ogata (1961) 30
Figure 2-8. Nylon 6,6 degradation data (293 °C and 1
atm steam) of Meacock (1954) and model fit 40
Figure 2-9. Nylon 6,6 degradation data (305 °C under

nitrogen) of Kamerbeek, Kroes and Grolle
(1961) and model fit 41
xii
Figure 2-10. Nylon 6,6 end-group degradation data
of
Wiloth (1971) and model fit
43
Figure 2-11. Nylon 6,6 volatile evolution data of

Wiloth
(1971) and model fit
44
Figure 3-1. Predicted melting point of nylon 6,6

and water
6g
Figure 3-2. Water concentration profiles for the nylon

6,6 stationary film model (f 2 = 0) .69
Figure 3-3. Water and polymer concentration profiles

for the nylon 6, 6 stationary film model
(* - 0.5) 71

for the nylon 6,6 stationary film model
2
( - 5) 72
Figure 3-5. Film average M evolution for the nylon 6,6

r
stationary film model at various Thiele moduli 74

r
stationary film model at various Thiele moduli 75

n
2
stationary film model for + = 5 and 100 77
Figure 3-8. Molecular weight profiles for the nylon 6,6

2
stationary film model (+ = 5) 78
Figure 3-9. Molecular weight profiles for the nylon 6,6

2
stationary film model (* = 100) 79
xiii
n
and ideal stationary film models
for
= 0.2 and 5 81

r
and ideal stationary film models for 2 =
0.2 <(>
82
Figure 3-12. Polydispersity evolution for the nylon

6,6
stationary film model
34
Figure 3-13. Polydispersity evolution for the ideal

stationary film model 35
Figure 4-1. Flowing film model a) Flasher cross-section

and b) flowing film geometry 89
Figure 4-2. Molecular weight profile for the nylon 6,6

flowing film model with a parabolic velocity
2
profile a) Da = 1.0, <(»
= 1.0, b) Da - 1.0,
2 2
<t>
= 5.0, c) Da = 5.0, <|>
= 5.0) 96
Figure 4-3. Number average molecular weight for the

effluent of the nylon 6,6 flowing film model
with a parabolic velocity profile vs. the
Thiele modulus and Damkohler number (Contour
Interval - 500 g/mol) 97
Figure 4-4. Average water concentration for the effluent

of the nylon 6,6 flowing film model with a
parabolic velocity profile vs. the Thiele
modulus and Damkohler number (Contour
Interval = 0.075) 99
xiv
Figure 4-5. Number average molecular weight for
the
effluent of the nylon 6,6 flowing film
model
with a
parabolic velocity profile vs. the
Thiele modulus and 4 /Da (Contour
Interval
= 1000 g/mol)
101
Figure 4-6. Average water concentration for the

effluent
of the nylon 6,6 flowing film
model with a
parabolic velocity profile vs. the Thiele
2
modulus and 4 /Da (Contour Interval = 0.075) 102
Figure 4-7. Number average molecular weight (a) and

average water concentration (b) for the
with a linear velocity profile vs. the Thiele
modulus and Damkohler number a) Contour (
Interval = 500 g/mol, b) Contour Interval

= 0.075) 104

average water concentration (4-8b) for the
with a plug flow velocity profile vs. the
Thiele modulus and Damkohler number ( a)
Contour Interval = 500 g/mol, b) Contour

Interval = 0.075) 106

average water concentration (b) for the
effluent of the ideal flowing film model with

a parabolic velocity profile vs. the Thiele
modulus and Damkohler Number ( a) Contour
Interval = 500 g/mol, b) Contour Interval =
0.075) 108
xv
Figure 4-10. Outlet water concentration vs. required film
area for the nylon 6,6 flowing film model
with
a parabolic velocity profile with
various outlet
molecular weights 3
(Q = 1 m /hr) 113
Figure 4-11. Average film velocity and width vs. aspect

ratio for the nylon 6,6 flowing film model
(A = 83 m M = 7, 500g/mol, Q = 1 m 3 /hr)
,
116
n
Figure 5-1. The time evolution of the effluent average

a) number average molecular weight, amine and
carboxyl end-group concentrations and b) water,
stabilized end-group, Schiff base end-groups
and crosslink, concentrations for the flowing
film model as the Damkohler number is changed
from 2.5 to 3.0 at a Thiele modulus of 1.0 and
270 °C 122

and crosslink concentrations for the flowing
from 2.5 to 3.0 at a Thiele modulus of 1.0 and
280 °C 124


and crosslink concentrations for the flowing
from 2.5 to 3.0 and the temperature reduced from
270 to 262 °C at a Thiele modulus of 1.0 126
xvi
Figure 5-4 The time evolution of the effluent
average
a) number average molecular
weight, amine and
carboxyl end-group concentrations and b)
water,
and crosslink concentrations for the
flowing
film model as the Damkohler number is
changed
from 2.5 to 3.0 and the temperature reduced
from
280 to 268 °C at a Thiele modulus of 1.0
128
Figure 6-1. Wiped film polymerization reactor schematic

(a) and wiped film polymerizer model
(b) . . 134
Figure 6-2. Number average molecular weight predictions

for the nylon 6, 6 wiped film reactor model
effluent at 280 °C and 300 mm Hg steam vapor
pressure; a) = 1, b) = 10,
C) N 50, d) N . = 300 141

mix mix

for the nylon 6,6 wiped film reactor model
effluent at 280 °C and N = 50; a) P =
mix
100 mm Hg, b) P = 760 mm Hg 143

for the nylon 6, 6 wiped film reactor model
effluent N = 50 and a steam vapor
r pressure
r
mix
300 mm Hg; a) 270 °C, b) 290 °C, c) 300 °C 145
Figure 6-5 Concentration of cyclized end-groups

(mmol/113g) for the nylon 6,6 wiped film
reactor model effluent at a steam vapor

pressure
r of 300 mm Hg
3 and N . =50; a) 270 °C,
mix
b) 300 °C 148
xvii
Figure 6-6. Difference in amine and carboxyl end-group
concentrations (mmol/113g) for the nylon
6,6 wiped film reactor model effluent at
a steam vapor pressure of 300 mm
Hg and
N 50; a) 280 ° c' b 300 »C
mix* )
150

for the nylon 6,6 wiped film reactor
model
effluent neglecting degradation reactions
at 280 °C and 300 mm Hg Steam Vapor Pressure
151
Figure 6-8. Ultimate number average molecular weight

predictions for the nylon 6,6 wiped film
reactor model effluent for N . = 50;
mix
a) T=280 °C, P = 300 mm Hg, b) T =280 °C,
P = 100 mm Hg, c) T = 300 °C, P = 300 mm Hg 153

for the nylon 6,6 wiped film reactor model
effluent at 280 °C and 300 mm Hg Steam Vapor

Pressure; a) N^ = 50, b) N^ = 1 155
xviii
NOMENCLATURE
a activity of component i, dimensionless

i
A amine end-group
C carboxyl end-group
A
C total molar density of mobile species
...
dimensionless
(C
W
+ C
P
)
'
C concentration of component i, dimensionless

i
C total molar density for nylon 6,6 amidation (C =1 C),

T
dimensionless
o
C initial water concentration, mol/1
w
d reactor diameter, cm
2
binary system diffusivity, cm /s
0
Da Damkohler number (k L/<v >) , dimensionless
app y
E activation energy, cal/mol

a
E apparent activation energy, cal/mol

app
H film height, cm
AH apparent enthalpy of reaction, cal/mol

app
AH^ partial molar heat of mixing of component i, cal/mol
xix
molar flux of component i relative to the molar average
2
velocity, mol/cm hr
thermodynamic forward and reverse

reaction rate
constants, gmol /total mol x hr
reference apparent forward rate

constant, gmol/total mol
hr
apparent forward reaction rate constant, 1

hr"
reference apparent forward reaction rate 1

constant, hr"
apparent equilibrium constant, dimensionless
reference apparent equilibrium constant, dimensionless
Y
A Y
c
/ Y
L Y
w t dimensionless
apparent equilibrium constant (X^ / X X

& ) , dimensionles
amide linkage or reactor length, cm
th "\
the k moment of the distribution, mol/cm
number average molecular weight, g/mol
molar flux of component i relative to fixed coordinates,

2
mol/cm hr
molar flux of water in a spacially fixed reference frame

2
(Eulerian frame), mol/cm x hr
the conversion of end-groups, dimensionless
the polydispersity of the system, dimensionless
xx
Q volumetric flow rate, cm /hr
R. dimensionless reaction rate of component

l i (R./k° C °),
1 app W
dimensionless
r reaction rate per unit volume of

1 component i, mol/1 x hr
t time, dimensionless
T temperature, °K
T reference temperature, 473 °K

0
v velocity, dimensionless
y
w water molecule or reactor width, cm
X binary mole fraction of water (C. /C) , dimensionless
X. mole fraction of component i (C./CJ, dimensionless

l
<x > number average degree of polymerization

n
<x > weight average degree of polymerisation

w
Z length, dimensionless
GREEKS
0
2 2
Thiele modulus (H k / D) , dimensionless
app
Y activity coefficient (a^ = Yj_ x j_)/ dimensionless

i
xxi
CHAPTER 1
INTRODUCTION
1.1 General
There are many reversible polycondensation systems in which the
equilibrium constant and the condensate diffusivity are small enough
that mass transfer effects become an important factor in determining
the molecular weight of the product. Since the molecular weight is
sensitive to the conversion of the end-groups virtually all
polycondensations with finite equilibrium constants fall into this
category. The desired product molecular weight is above the
equilibrium molecular weight at the concentrations of condensate
normally encountered in reactors without condensate transfer.
Therefore, it is important to accurately model both the reaction and
mass transfer which occur simultaneously in these condensation
reactors. Major industrial examples include nylon 6,6, nylon 6 and
polyethylene terepthalate
In many applications it is important to use the kinetics for a
specific polycondensation rather than an idealized kinetic model where
the kinetic constants depend only on temperature. This is because

2
most polycondensations are known to have unique degradation

reactions
and apparent rate and equilibrium constants which can be highly
composition dependent. A common example is nylon 6, 6 which will be
used as a representative polycondensation
in this work. The three
reactors in the continuous commercial nylon 6,6 process are the
prepolymerizer, the flasher and the finisher
(U. S. Patent; 3,900,450)
shown in Figure 1-1. Since the problem of condensate removal and the
growth of viscosity is common to many polycondensations, these are

typical reactors. Thus, many of the results of this work will be
applicable to other polycondensation systems with proper

modification (s) of the kinetic description.
1.2 Previous Work
A polycondensation reaction can be represented as
P + P = P + w (l.l)
n m n+m
where P represents a chain of length n and W is a condensate

n
molecule. An infinite set of rate equations may be written and
integrated directly for the irreversible, isothermal case
(Biesenberger; 1965) if the forward reaction rate constant is
considered to be independent of chain length (equal reactivity
hypothesis) . Generating functions (Scanlan; 1956) and the Z-transform
(Abraham; 1963, Kilkson; 1964, Laurence and Tirrell; in preparation)
have also been used to determine the molecular weight distribution
(MWD) for the irreversible condensation reaction.

Figure 1-1. Continuous process for nylon 6,6 (U.S. Patent;
3, 900, 450) .
4
The polydispersity of the molecular weight distribution in an

equilibrium polycondensation generated
in a well-mixed, isothermal
batch reactor with a fixed forward reaction rate constant and
no side
reactions is two. Kilkson (1964) studied an irreversible
polycondensation in plug flow reactors with recycle and showed that
the polydispersity of the polymer product

may have values considerably
in excess of two. Tadmor and Biesenberger (1966) showed that
irreversible polycondensation in segregated,
continuous flow, stirred
tank reactors (CSTR) has a significantly higher polydispersity
than in
a homogeneous CSTR. However, most studies of reversible
polycondensations have shown that the polydispersity deviates only
slightly from two (Gupta, Kumar, and Ghosh; 1983)
In all but the simplest ideal case the rate equations for the
polymerization must be integrated numerically; this is inefficient for
process models that incorporate the entire MWD. More recent work has
focused on determining moments of the distribution rather than the

fc
complete MWD . The k moment of the distribution is defined as
(lc) k
M = I n C (1.2)
n=l
where C is the molar concentration of species of chain length n. The

n
number and weight average degrees of polymerization can be written in
terms of the moments as
< x > = ^
M
M
(1)
(0)
(1.3)
5
M (2)
w (l) (1.4)
M
The infinite set of rate equations

may be rewritten in terms of
an infinite set of moment equations.
However, the first three or four
moments are usually sufficient to adequately describe the product
polymer. For the reversible case, the k th
moment depends on the k+l st
moment and some sort of closure condition
must be assumed to decouple
the hierarchy of moments (Tai; 1979, Kumar; 1986, Laurence and
Tirrell) Fortunately, the results are not usually very
.
sensitive to
the particular choice of closure approximation.
More recently, Mills (1986) and Costa and Villermaux

(1986) have
used a fast fourier transform technique to numerically invert the
Z-
transform to find the exact MWD . The method has been successfully
applied to the reversible case and the MWD found without any closure
approximations. This method is much faster than integratinq the
individual species equations, but is considerably slower than the
moment approach.
1 . 3 A Simplified Approach
A much simpler alternative approach is often possible and
sufficient for much of the process modeling. The polycondensation
reaction can be treated as a second order reversible reaction between
functional end-qroups if the rate and equilibrium constants are
independent of chain length. For example, the formation of amide

links (L) and water (W) from amine (A) and carboxyl (C) end groups in
nylon 6,6 polymerization can be
represented by
A + C = L + W
(1.5)
It was Flory's (1946) statistical analysis of reaction (1.1) that

provided the first insight into the gross features of
polycondensations. He considered the reaction to
be irreversible and
derived his famous results for the number and weight
average degrees
of polymerization < x > and < x
(
n
> ) and the polydispersity P = <
w (
X > 1 < x > for a stoichiometrically balanced feed,

w n
)
W - l A '/
1 p '
P = 1 + P (1.8)
where p is the fractional conversion of functional groups (i.e., A or
C in reaction 1.5) . From eqs. 1.6 and 1.7 it is obvious that high
conversions are required for high degrees of polymerization and a
large molecular weight . Also, the polydispersity asymptotically
approaches a value of 2 with increasing conversion.
Flory (1944) also showed that eqs. (1.6-1.8) are also valid for
reversible polycondensations at equilibrium. Recently, Kumar et al.
(1982) and Gupta et al. (1982) have shown that eqs. (1.6-1.8) apply at
all times for the equal reactivity, reversible case. However, the
reversibility of the reaction
significantly alters the time evolution
of the extent of reaction. For our purposes, it is important t 0
realize that eqs. (1.6-1.8) give no information about the t lme
evolution of conversion or MWD. The forward reaction rate constant
determines how fast a given conversion can be reached.
For reversible
polycondensations the equilibrium constant places an upper bound on
the conversion at a given condensate
composition.
There are relatively few studies of
simultaneous polycondensation
and diffusion of condensate for reversible
polycondensations despite
its commercial relevance; this research
effort will be concentrated in
this area. A moment analysis will not be used, but Flory's most
probable distribution (eqs. 1.6-1.8) will be assumed to hold
instantaneously, which is a a minor restriction on our results as
there are very few reversible polycondensation systems in which the
polydispersities which deviate markedly from two (Gupta, Kumar, and

Ghosh; 1983)
1 . 4 Research Objectives
The critical design and control variables for the flasher and
finisher units in the continuous nylon 6,6 process will be identified
through process modeling. This will require development of simplified
process models for the flasher and finisher. In addition, a realistic
kinetic and equilibrium model for nylon 6,6, complete with composition
dependent rate and equilibrium constants will be formulated. For
realistic high temperature modeling, a the simplest chemically

3
reasonable degradation scheme that

is consistent with all available
degradation data will complete the
kinetic model.
We will investigate film diffusion
effects in a stationary film
of nylon 6,6 to determine under
what film thicknesses molecular weight
generation is under diffusion control and reaction control. The

effect of compositional gradients across the film on the
polydispersity will also be studied.
A flowing film approximates the operating conditions in the

flasher. The effect of a residence time distribution in a flowing
film will be determined and compared to the stationary film case.
Particular attention will be paid to the longer residence time

material to determine at what throughputs and
temperatures degradation
and gel reactions have important effects in a distributed residence
time model. Dynamic studies to determine the response of the flasher
to a change in throughput will be performed. Various policies to
respond to these disturbances will also be tested.
The finisher is a wiped-film reactor and separator. Periodic
mixing of a flowing film will form the basis of our simplified model
of the finisher. The degradation model will allow prediction of the
optimum operating conditions (temperature and pressure) in the
finisher (which is operated at high temperature, 280-300 °C) . The
objective is optimize the trade off between fast molecular weight
generation and gel formation and end-group imbalance.
It is the goal of this research to enhance our knowledge of the
design and control of continuous polycondensation reactors. This will
enable us to have better control of molecular weight and product
uniformity in these systems. This approach is useful for the

evaluation or modification of existing designs as well as for the
generation of future process designs. The ultimate goal for
application of these models is the
production of a less costly, higher
quality product polymer in a more efficient process.
CHAPTER 2
A SIMPLIFIED KINETIC, EQUILIBRIUM

AND
DEGRADATION MODEL FOR NYLON 6,6
2.1 Kinetic and Equilibrium Model
2.1.1 Introduction
One of the most striking features of the nylon 6,6
polycondensation is that the observed rate and equilibrium constants
are functions of composition. Traditionally, these quantities are
often treated only as functions of temperature. In this section we
show why one can expect the apparent reaction rate and equilibrium
constants to show a composition dependence in highly nonideal liquid
mixtures. In addition, a nonideal liquid model is used to develop
correlational forms in order to regress the apparent reaction rate and
equilibrium constants as functions of temperature and composition.
A nylon-6,6 polyamidation can be treated as a second order
reversible reaction
11
A + C = L + W
( 2.1)
of an amine (A) and carboxyl (C) end-group to form an amide linkage

(L) with the elimination of a water molecule
, (W) . The thermodynamic
rate and equilibrium constants are
defined in terms of the activities
of the four components (Hougen and Watson; 1947)
r = k a, a„ "- k '
a a a
A C L w (2-2)
k aT a
L W
where r is the intrinsic reaction rate and k and k' are the forward
and reverse reaction rate constants. These true thermodynamic rate
and equilibrium constants only depend on temperature and a reference
state
An apparent equilibrium constant, K , and forward reaction rate

app
constant, k , are defined as
app
X.
K
app " «
»eq (*•«>
and
k
app
=
(
X X
^
" X )
(2,5)
A C K L *W
app
12
where X. is the mole fraction of the th

i constituent. Mole fractions
have been used in eq. (2.5) rather than the usual concentrations
because this eliminates the need for
density correlations and because
they are more fundamentally related to activities. The
apparent and
thermodynamic rate and equilibrium
constants are related as follows.
k = k
app Y Y (2.6)
A c
where y is the activity coefficient of

A component A ( a_ - y x ) and
A A A
K K K
app " /
T
(2.7)
where K
In
y
- ( y
& 1(, /
^ T> ^
an ideal solution all the activity coefficients are unity and
the apparent and thermodynamic rate and equilibrium constants are
identical. However, in nonideal solutions the activity coefficients
may deviate considerably from unity and the apparent rate and
equilibrium constants can exhibit a strong compositional dependence.
In principle, the composition dependence of the apparent rate and
equilibrium constants could be resolved by carrying out enough
experiments to determine the activity of all the relevant species,
including those produced by side reactions, as functions of
temperature, pressure, and composition. Although this approach is
usually impractical, it is advantageous to regress the available data
in a thermodynamically consistent way. We show here that when only
the rate and equilibrium constants are required, the composition
dependence of the activities may be lumped into two independent

13
groups, which depend only on the reactant and product activities.

Hence, it is not necessary to
determine the composition dependence
of
each component's activity separately.
The reaction equilibrium in many

polyamidations is such that the
reaction must be driven by removal of the water in order
to obtain
high molecular weights. This removal of water during the course of
the polymerization, along with the disappearance of polar
end-groups
by reaction, causes the polarity of
the reaction medium to change
drastically. Even if one assumes that the equal
reactivity hypothesis
is valid, the apparent rate and the apparent
equilibrium constants can
depend on the instantaneous solution composition.
In other words, the
assumption that the end-group reactivity is independent

of molecular
size does not preclude its dependence on solution
environment.
A fundamental molecular model that would be suitable for
describing the behavior of nylon polycondensations over a wide range
of variables is not currently available. This is presumably due to
the complex nature of the problem arising from a highly nonideal
liquid solution, the complex equilibria and the role of acid catalysis
during polycondensation. In this section we will show how a simple
activity-based model can be used to determine the proper form of the
apparent reaction rate and equilibrium correlations for nonideal
polycondensations. This model is applied to the experimental data for
the nylon-6,6 polyamidation.

14
2.1.2 A Nonideal Liq uid MoHpI
During
polyamidation, the polar nature of
the end-groups and the
condensate causes the system to
deviate considerably from ideal
solution behavior, with activity coefficients that deviate
substantially from unity. The temperature
dependence of the activity
coefficient is given by the standard thermodynamic relation
(e.g.
Smith and Van Ness; 1975)
-AH~
dlnY i = dT
"zl
RT
(2.8)
where AH. is the partial molar heat of mixing of component i; eq.
(2.8) is strictly correct at constant pressure and composition. If
the true thermodynamic rate and equilibrium constants

are written in
terms of activities and eq. (2.8) is integrated assuming temperature
independent heats of mixing, the following expressions for the
apparent rate and equilibrium constants may be derived.
AH
K = K ex P
app o
[ "
HP* <
5 "
*
o
>1 (2.9)
E
k = k exp
K [
L
- -SEE (
± - ± )] (2.10)
app
r o R T x v '
where AH = - AH + AH_ + AH, 7 - AH,, - AH n , E = E - AH - AH- and

app L w A C app a A C
is an arbitrary reference temperature.

15
The quantities
mole fractions, X.; only in an

% ,
v ^ 4H
ideal solution are they all

f and ^ may depend on ^ ^
independent
of composition. For a stolchiometrically
balanced polycondensation,
the composition dependence of
these quantities can be a function
of at
most two of the independent mole
fractions.
Also, one may develop the identical
expression for k if both
app
the forward and reverse reaction
in eq. (2.1) are assumed to be acid-
catalyzed. The apparent activation energy is then E = E -
app a
(AH^ + 2ITC )
and k = V Xq Thus, the second and
Q
.
third order
kinetic models in these nonideal liquid mixtures can not be
distinguished based upon the available kinetic data without
determining the activities of the individual
constituents.
2-1.3 Equilibrium Correlation
A decrease in the apparent equilibrium constant for nylon-6,6
polymerization with increasing water mole fraction (x^ > 0.2) was
reported by Wiloth (1955) and Ogata (1961) . Giori and Hayes (1970)
found that the apparent equilibrium constant for nylon-6
polymerization also increased with water content at mole fractions
less than 10% and that it decreased at higher water contents in a
manner consistent with the previous findings of Wiloth (1955) and
Ogata (1961) for nylon-6,6. This increase in the apparent equilibrium
constant at very low water content was also reported by Wiloth (1971)
for nylon 6,6. Since nylon-6 and nylon-6,6 have identical ratios of
methylene, carboxyl, amide, and amine groups at any conversion, it is

16
expected that the ionic character

of the reacting medium will be
similar at a given composition (i.e., the same number of amide
linkages, end-groups, and water
molecules) This suggests that
.
the
nylon-6,6 system might also exhibit similar reaction rate and
equilibrium behavior. In fact, an apparent equilibrium constant of
250 for nylon-6,6 at very low water content
(approximately 1 mole %)
reported by Jacobs and Zimmerman (1977)

and a value of 225 calculated
from data presented by Jones and White (1972) are remarkably
consistent with the value of 250 obtained
by extrapolating the nylon-6
data of Giori and Hayes (1970). Based on this observation, we include
these data of Giori and Hayes (1970) in our equilibrium regression.

This is also consistent with the commonly accepted "long-chain
hypothesis," which has been verified in many systems
(Rabinowitz and
Wood; 1936 and Flory; 1953)
Ogata (1960) reported high temperature nylon-6,6 equilibrium

studies at low water content. However, these were not included in the
equilibrium regression because we suspect that degradation reactions
was present during the prolonged (6 hr) experiments. These data also
disagree with those of Jacobs and Zimmerman (1977) and Jones and White
(1972). For the thermally sensitive nylon-6,6, the determination of
the intrinsic polycondensation kinetics may actually be more
accurately made based on data from the relatively stable nylon-6
system which is less complicated by side reactions during extended
high temperature studies. Wiloth (1971) showed that nylon-6 reaches
equilibrium after about 80 hours at 240 °C at a water content of 0.4
wt %.
17
Kumar et al. (1981) correlated the equilibrium data of Ogata

(1961) using the form K -
app 6 xp [a - b (W°> ] where (W°> is the
dimensionless initial water content
(moles water/mole initial nylon
salt), corresponding to mole fractions of water in the range
0.5 to
0.8; the temperatures were in the range 200 -
220 'C. The values of a
and b were determined as functions of T
and (W°) as shown in Table 1
of Kumar (1981) However, the dependence of the

.
correlation on the
initial condition renders it unacceptable for
the study of continuous
polyamidation reactions.
Although the apparent equilibrium constant can be a

function of
two of the independent component mole fractions,
the data of Wiloth
(1955), Ogata (1961), and Giori and Hayes (1970) clearly show that the
mole fraction of water produced the predominant variation in the
observed equilibrium constant. Thus, we regressed the apparent
equilibrium correlation solely in terms of the mole fraction of water.
The apparent heat of reaction was determined from the combined data of
Ogata (1961) and Giori and Hayes (1970) by evaluating d (In K ) / dT

app
as a function of X . The data are shown in Figure 2-1 along with the
?f
empirical correlation,
AH
app "
7, 650 tanh t 6 - 5
^ " 0,52)]
+ 6,500 exp [-J^ /0.065] - 800 (2.11)
Care was taken to fit the data at lower mole fractions of water data
most accurately and not to greatly exceed the accepted heat of
amidation for the reaction, which is 6 to 7 kcal/mol (Jacobs and
Zimmerman; 1977, Jones and White; 1972, FuJcumoto; 1956, Tobolsky and
Eisenberg; 1959, Zimmerman; 1964)

Giori and Hayes (1970)
Ogata (1961)
Eq.(2.11)
0.2 0.4 0.6 0.8
Mole Fraction Water
Apparent enthalpy of reaction vs. mole fraction water.

19
The reaction appears to

be endothermic at water compositions
greater than 52 mole », and exothermic at lower
concentrations. This
is consistent with the observation
that nonideal liquid

of Ogata (1961)
explanation of this inversion of the
unnecessary
heat
phase
to account for any -hydrolytic

behavior could
of
. However, we
provide
reaction.
g
It
^
may
note
be
degradation of nylon-6,6,-
as suggested by Ogata (1961), although such a hydrolysis is not
necessarily ruled out.
The prefactor K is easily determined via eqs.

q (2.9), (2.11) and
the data. A reference temperature of 200 «C was
chosen arbitrarily.
In this regression the data of Wiloth
(1955) and the single points of
Jacobs and Zimmerman (1977) and Jones and White (1972) were
added to
the data of Ogata (1961) and Giori and Hayes (1970) . The K
0
correlation is shown in Figure 2-2 and is given by the empirical
expression
1/2
K
0
- exp [(1 - 0.47 expl-X^'70.2]) (8.45-4.2 y ] (2.12)
The fact that all of the data can be described by a single curve,
indicates that the nylon-6 equilibrium data is representative of the
nylon-6,6 system and attests to the fact that the apparent equilibrium
constant may be described solely by the mole fraction of water and the
temperature. This is surprising since two mole fractions are
necessary to specify the stoichiometrically balanced polyamidation.

20
« Wiloth (1971) ^\ 0
- Ogata (1961)
5 O
+ 0
Gioriand Hayes (1970)
x
Jacobs and Zimmerman (1977) & ^\ 0
Jones and White (1972)

Eq. (2.12)
1
1
' I
! 1 ,
0.2 0.4 0.6 0.8
Mole Fraction Water
Figure 2-2. Apparent equilibrium constant at 200 e

C vs. mole
fraction water.
21
800
— r 1
p-
0.6 0.8
Mole Fraction Water
Figure 2-3. Apparent equilibrium consta nt vs. mole fraction water

22
900
Mole Fraction Water
+ Giori and Hayes (1970) 260 C

Giori and Hayes (1970) 240 C
o Jacobs and Zimmerman (1977) 280 C &
Jones and White (1972) 290 C
Figure 2-4. Apparent equilibrium constant vs. mole fraction water.

The final equilibrium correlation
is given by eqs. (2.9), (2.11),
and (2.12). Figures 2-3 and 2-4 compare the resulting
correlation
against the available published
data. A more comprehensive view of
the correlation is shown in Figure
2-5.
2.1.4 Amidation Reaction Rate Correlat

ion
Ogata (1961) published rate data and

a correlation for the nylon-
6,6 polymerization, which was correlated with the initial water
content. Unfortunately this dependence prevents its use
in the study
of continuous polycondensation. Furthermore, the regression was done
under the assumption that K is constant,
app which is incorrect except
over small ranges of temperature and composition. Ogata (1961)
reported the concentrations of end-groups, water, and amide linkages
at time intervals from 0.5 to 11 hr, water mole fractions of 0.1 to
0.8, and at temperatures from 200 to 220 °C. We regressed these data
using an integral least squares analysis, with a value of K from

app
our correlation. The total number of moles in the system was chosen
as the extensive variable since eq. (2.1) conserves this quantity in
closed batch experiments. Using a reference temperature of 200 °C,
eq. (2.10) reduces to k = k . Our calculations showed that Ogata's

app O
experiments at 200 °C and at mole fractions of water less than 0.5
could be described by a single constant k , whereas those with >
0.6 could by described by another smaller constant. Thus, k changes

Q
rapidly at about X.. = 0.55 and was regressed with the empirical
expression
24
Mole Fraction Water
Figure 2-5. Apparent equilibrium constant vs. mole fraction water.

25
k = exp (2.55 - 0.45 tanh

0 [25 - 0.55)]) ( 2 .1 3 )
We find that the rate constant

changes rapidly in the same region
Wh6re AH chan ? es si ^- The results were not very sensitive
app to the
value of E over the range of the
app data and a constant value of
21,000 cal/mol fit all the data satisfactorily.
The validity of the equal reactivity

assumption (Taylor; 1947)
along with the slowing of the amidation rate at low water
content,
accompanied by a sensitivity to carboxyl concentration have
been well-
documented. Studies which follow apparent second order kinetics
(Ogata; 1961, Giori and Hayes; 1970) have been at high water
contents.
At lower water contents an acid-catalyzed mechanism is increasingly
important (Wiloth; 1955a, 1955b, 1955c, 1958, Heikens, Hermans and Van
der Want; 1960, Reimschuessel and Nagasubramanian; 1972) . Although

virtually all of the low water content kinetic studies have been done
on nylon-6, it is expected that the nylon-6,6 rate constant will be
very close to that of nylon-6 at low water concentrations, just as
the equilibrium was found to be. We have therefore merged the high
water content correlation of Ogata's (1961) data with that of
Reimschuessel and Nagasubramanian (1972) to obtain a comprehensive
reaction rate correlation.
The transition from a second order to an acid-catalyzed
description is expected to be smooth. However, it is clear that the
reaction rate appears to be second order at high water content but has
both second and third order character at low acid and low water
content. Based on the lowest water content study of Ogata (1961) at
11 mole % and the highest water content study of Reimschuessel and

26
Nagasubramanian (1972) at mole

8 %, we can expect a transition near
10
mole % water. This also corresponds to the point where AH
app
increases rapidly as shown in Figure
2-1.
It is important to realize
that the majority of data for nylon-6
have been taken at water mole fractions
less than 0.08 because the
reaction proceeds with a net consumption of water due to the
caprolactam ring opening. Conversely, the nylon-6, 6 reaction rapidly
generates water so that the majority
of data were taken at water
content greater than 50%. It is expected that the correlation
of
Ogata's data is most accurate at
^ > 0.3 and Reimschuessel' s and
Nagasubramanian' s correlation at X < 0.05. A gradual transition near
w
0.1 represents the best available estimate, but could be easily
adjusted with additional kinetic data.
Reimschuessel and Nagasubramanian (1972) regressed the kinetic

rate constant as a linear combination of second and third order rate
models. The rates of the two contributions were characterized by

different activation energies. The constants of Reimschuessel and
Nagasubramanian (1972) were converted to the same units as eq. (2.13)
by assuming a value of the ratio (g polymer/ mol amide linkages) for
dry nylon-6 of infinite molecular weight, which is a good
approximation to the experimental conditions. The final kinetic
expression in eq. (2.10) requires a single activation energy and a
constant of 21, 400 cal/mol was used. This is the mean of the two
values for the composite model of Reimschuessel and Nagasubramanian
and the value of 21,000 cal/mol from Ogata's data. The final
empirical expression is
27
Jc
Q = exp (2.55 - 0.45 tanh [25(X -
W 0.55)])
+ 8.58 (tanh [50(3^-0.10)] - -

1) (1 30.05 X ) (2 .14)
Q
The resulting k at 200

app °C is shown as a function of water
content and carboxyl concentration
in Figure 2-6, along with the
correlation of Reimschuessel and Nagasubramanian
(1972) . Simulations
using the final rate and equilibrium
correlations are compared with
some of the data of Ogata (1961) in Figure 2-7. The data of
Reimschuessel and Nagasubramanian are valid
up to approximately 265
•C. Therefore the combined kinetic and equilibrium model has been
tested up to 265 °C but the equilibrium

correlation is valid up to 290
2.1.5 Discussion
The final equilibrium and kinetic correlations are good

descriptions of the available data. Their complex mathematical form
reflects both the nonideality of the liquid mixture and complex
chemical and ionic interactions in the reacting medium. Some of the
equilibria which are considered to be significant in the polyamidation
system include (Jones and White; 1972 and Wyness; 1958)
-NH + -COOH = -NH* + -COO~ (2.15)

2
NH + -COOH = -CONH- + H 0 (2.16)

2 2
+
COOH + H 0 = -COO~ + H 0 (2.17)
2 3
28
Mole Fraction Water
Figure 2-6. Apparent rate constant at 200 °C vs. mole fracti

water
29
+ _
H 0 = H + 0H
2 (2.18)
+ +
-COOH + H = -C (OH)
2 (2-19)
+
C (OH) -Nh = +
2 +
2
-coNH- + H 0 (2.20)
It is not surprising that the rate

constant of eq. (2.16) should vary
significantly with the acid and water
content in the system. The
ionic character of the system changes
drastically as it goes from a 50
% salt solution to a dry, high molecular

weight polymer and all the
various equilibria are affected. A more accurate simulation would
have to consider some or all of these equilibria and estimate the
activities of all species, including ionic moieties, as functions of
the solution composition and temperature.
The reason that the apparent equilibrium constant should vary so
markedly with the mole fraction of water may be qualitatively

explained by Flory-Huggins theory (Flory; 1953) . If water is
considered as a low molecular weight solvent for nylon-6,6, the theory
predicts that the activity of the water will vary non-linearly with
its volume fraction. Since the thermodynamic equilibrium constant is
fundamentally a ratio of activities (which Flory has established are
not directly proportional to mole fractions) the apparent equilibrium
constant is not expected to be concentration independent.
The model from which the correlations were derived is not
restricted to nylons or polymers in general. The model may be applied
to any second order reversible reaction. It is particularly useful
for nylons because the polyamidation takes place over such a wide
30
. 030
. 024 - 200 C, WO = 0.50

210 C, WO = 3.05
-e-
± +
018
220 C, WO = 10.0
-
012
-
. 006
.000 4
0 . 0 2.2 4 . 4 6.6 8 . 8 11 . 0
TIME (HRS)
Figure 2-7. Comparison of the simulation and the data of Ogata

(1961)
31
range of composition in a highly nonideal liquid mixture. Another

class of polymers where we expect the model to be useful is
polyesters; like nylons, there is
some controversy as to the reaction
order of the polymerization
(Hiemenz; 1984)
2.2 Degradation Model
2.2.1 Background and Degradation s c heme
We seek to develop a degradation model

that is both simple enough
to be readily incorporated into our kinetic scheme and subsequently
into a reactor model and that also accurately

reflects the effect of
degradation on the nylon 6,6 product. It is not intended to reveal
all degradation reactions which are occurring nor is it a detailed

mechanistic study of their underlying chemistry. Our goal is to
develop a simple, chemically reasonable degradation model which
accurately describes the available degradation data in order to
develop realistic process models for nylon 6,6 reactors.
It is well-known that given sufficient time nylon 6,6 degrades
and gels at high temperature even in the absence of oxygen. It is
also well-known that the degradation products include carbon dioxide,
ammonia and some cyclopentanone (e.g., Sweeney and Zimmerman; 1964).
In this section we summarize a minimal set of reactions that can
account for the formation of these products and for gelation. We have
ignored degradation schemes which require oxygen so our task is

32
considerably simplified. Previous investigators have postulated a

number of
primary and secondary
degradation reactions to account
for
the thermal instability of
nylon 6,6. We sus»arize the more
chemically feasible reactions in
the following brief review of the
literature.
At high temperature and under basic conditions, adipic

acid is
known to decompose into cyclopentanone,
water and carbon dioxide as
shown by reaction (2-1.1) in Table 2-1 (Morrison and Boyd; 1971)
Sweeney and Zimmerman (1964) proposed an

analog of reaction (2-1.1)
that may take place both on the end of a nylon
6, 6 chain and in the
interior as shown in reactions (2-1.2) and (2-1.3). Achhammer,
Reinhart and Kline (1951) have also suggested that the cyclopentyl
chain-end produced by reaction (2-1.3) may also continue

to degrade as
shown in reaction (2-1.4) to yield an amine end-grou
P/ cyclopentanone
and carbon dioxide. An amine end-group can subsequently react with
cyclopentanone to yield a Schiff base and water as shown in reaction
(2-1.5) (Ballistreri et al.; 1987) . The cyclopentyl end-group can
also react with an amine end-group to form a Schiff base and water as
shown in reaction (2-1.6) (Twilley; 1961).
Another possible degradation step is shown in reactions (2-1.7)
and (2-1.8) (Kamerbeek, Kroes and Grolle; 1961, Sweeney and Zimmerman;
1964). In reaction (2-1.7), the weakest bond in the polyamide is
broken at high temperature to form an olefinic end and a primary amide
end. The latter is not stable and rapidly decomposes into a nit rile
with the elimination of water by reaction (2-1.8). Wiloth and
Schindler (1967) have shown that carbon dioxide and a Schiff base may
be produced by a unimolecular mechanism shown in (2-1.9) from the
cyclopentyl end-group produced by reactions (2-1.2) and (2-1.3).

33
Table 2-1. Nylon 6,6 degradation reactions
O
o o I
II / \
HO-C-(CH 2 ) 4 -C-OH « CH 2 CH 2 + CO, + H.O (2-1.1)
CH 2 -CH 2
O
o o o c
II II
II / \
»»NH-C-(CH 2 4 -C-OH
) « --NH-C- CH CH 2 + H20
\ /
CH 2 -CH 2
H (2-1.2)
O
0 0 O C
II II || / \
NH-C-(CH 2 ) 4 -C-NH <-> NH-C- CH CH

\ /
CH2-C H2
H 2N (2-1.3)
Continued Next Page

Table 2-1. (conti nued)
O O
o c C
II / \
/ \
NH-C- CH CH 2 + H2 0 C0 2 + CH 2 CH
\ /
\ /
CH 2 -CH 2 CH 2 -CH 2
NH (2-1.4)
O N—
II
C
/ \
/ \
CH 2 CH 2 NH ^ CH CH + H2 0 (2-1.5)
\ / \ /
CH 2 -CH 2 CH 2 -CH 2
O N.
II
O C O C
II / \ \
II /
NH-C-CH CH 2 NH 2 f-> NH-C-CH CH 2 + H?0

\ / \ /
CH 2 -CH 2 CH 2' CH 2
(2-1.6)
Continued Next Page

35
Table 2-1. (continued)
° O
» II
C-NH-(CH 2 ) NH <-> C . NH
6
CH 2 =CH-(CH -NH (2-1.7)

2)4
O
II
C NH 2
"
<-» C=N + H20 (2-1.8)
O N
II II
o c c
II / \ / \
NH-C- CH CH 2 <-> CH 2 CH 2 + C0 2 (2-1 9)

\ / \ /
CH 2 -CH 2 CH 2 -CH 2
36
There is very little information

on the origin of the crosslinks
which are known to form in nylon
6,6 since there are so few and there
chemical structure has not been unambiguously
determined. In
addition, it is very difficult to measure
the number of crosslinks in
any given sample because their relationship
to measureable quantities
is unknown (solution viscosity, titrable end-groups) . Nevertheless,
three candidate crosslinking routes are shown in Table 2-2. If a
significant amount of water is present, there
will be an appreciable
amount of free amine and carboxyl groups from the reversible main
amidation reaction. The amine end-groups may combine and

subsequently
react with a carboxyl end-group to form a crosslink as shown in
reactions (2-2. la and b) in Table 2-2 (Ravens and Sisley; 1964).

Alternatively, two carboxyl end-groups can combine and subsequently
react with an amine end to form a crosslink (2-2. 2a and b) . Wiloth
(1971) reports that the Schiff base formed via reactions (2-1.5) and
(2-1.9) may subsequently react to form a variety of products including
a crosslink as shown in (2-2. 3a and b) .
2.2.2 Experimental Degradation Studies
Although there have been numerous studies on the degradation
behavior of nylon 6,6, many do not contain enough quantitative data
about the time evolution of end-groups and volatile gases for the
regression of kinetic constants necessary for the development of a
degradation model. In the following section we summarize literature
studies of nylon 6, 6 degradation which have merit for the regression
of kinetic constants in a degradation model.

Table 2-2. Nylon 6,6 crosslinking reactions
2 NH 2 ~ N + NH. (2-2. la)
O
" N HO-C-— <-» N + H? 0
- (2-2. lb)
I
H 0=C
O O
II II
C -° H <-> C + C0 2 + H2 0 (2-2. 2a)
+ H 2 N-(CH 2 )
6
-NH *->
C + H2 0 (2-2. 2b)
II
N
I
NH-(CH 2 )
6
Continued Next Page

N-
N—
II
C II
C
/ \
/ \
CH 2 CH 2 NH 2 CH 2
<-»
CH + NH
\ /
\ /
CH 2 -CH 2 CH 2 -CH 2 (2-2. 3a)
N— N-
II
C II
C
/ \
/ \
CH CH NH CH CH + NH
\ /
\ /
CH2~C H2 CH 2 -CH 2 (2-2. 3b)
39
Meacock (1954) studied how

the number of amine and
oarboxyl end-
groups in nylon 6,6 changed
with time at 293 -C under 1
atm of steam;
these data are shown in Figure
2-8. The number of amine end-groups
increases with time whereas the
number of carboxyl end-groups
decreases. Kamerbeek, Kroes
and Grolle (1961) also studied the
thermal degradation of nylon
6,6 under nitrogen. Their data for the
cumulative evolution of C0 and NH3
2
from nylon 6,6 at 305 >C are shown
in Figure 2-9. We see that the polymer changes
drastically over the
course of the experiment since more than half of the nitrogen
initially bound in amide linkages escapes as ammonia. Kamerbeek,
Kroes and Grolle emphasize the difficulty
explaining the source of
water required to drive reactions (1.4) and (2.1a) if
these are the
source of the carbon dioxide and ammonia.
in addition, they found
very little cyclopentanone and water in the
evolving gases.
Edel and Etienne (1961) followed the amine
end-group, ammonia and
carbon dioxide evolution during the thermal degradation of nylon
6,6
under nitrogen. However, since they did not follow the carboxyl end-
group evolution it is difficult to access what role the main amidation
reaction played in the amine end-group evolution. This greatly
reduces the utility of their data for our modeling purposes.
Peebles and Huffman (1971) found that the degradation behavior of
nylon 6,6 was drastically altered if the degradation took place in a
sealed reactor where the gaseous products were maintained over the
melt compared with an open reactor (nitrogen gas maintained over the
melt) In open reactors, the polyamide quickly discolored and
subsequently crosslinked, becoming insoluble after 5-8 hours at 282

40
20.
cn
n
4 -° 8.0 1 2.0 1 6.0 20.0
Time (hrs)
Figure 2-8. Nylon 6,6 degradation data (293 °C and 1 atm steam) of
Meacock (1954) and model fit.
41
600.
0. 20. 40. 60. 80. 100. 1 20.
Time (hr)
Figure 2-9. Nylon 6,6 degradation data (305 °C under nitrogen) of

Kamerbeek, Kroes and Grolle (1961) and model fit.
42
•C. However, in closed reactors the polymer remained white and

crosslinked after about 50 hours. The drastic difference between the
two reactors was lessened when
ammonia instead of nitrogen was passed
through the reacting melt in the
open reactor. The results of Peebles
and Huffman suggest that the reaction that produces ammonia is
reversible and/or that the ammonia reacts further when
it remains in
the melt.
In fact, Wiloth (1971) suggests a mechanism where the ammonia

reacts further in a closed reactor. wiloth (1971) performed an
extensive study on the thermal degradation in open and closed
reactors. We will only analyze his data on open reactors
(see Figures
2-10 and 2-11) since our purpose is to describe commercial thin film
reactors which correspond to this mode of operation. Furthermore, the
potential problem with the reversibility of the reaction which

produces ammonia is eliminated since the gaseous products (ammonia,
carbon dioxide and water) are rapidly removed from open reactors.
More recently, Ballistreri et al. (1987), studied the degradation
of nylon 6,6 by direct pyrolysis into a mass spectrometer. By
analyzing the degradation products they concluded that the primary
step in the thermal degradation was given by reaction (2-1.3) rather
than (2-1.7). They found species with the cyclized end-groups
produced by (2-1.2) and (2-1.3), Schiff base species produced by (2-
1.5) and (2-1.9) and the linkage produced by (2-2. 3a).

43
30.0-
+ 0 A Amine: 273, 285 <fc 305 C

VXD Carboxyl: 273, 285 it 305 C
cn
n
24.0-
\
o
E
E 1 8.0—
o
c
o
u 1 2
0-
D
O
o
c
UJ
0.0 2.0 4.0 6.0 8.0
Time (hrs)
Figure 2-10. Nylon 6,6 end-group degradation data of Wiloth (1971)

and model fit
44
40.0-
+ 0 A NH3: 273, 285 ic 305 C
cn
VXD C02: 273, 285 k 305 C
n
32.0—
\
o
E
E 24.0-
o
c
o
u 1 6.0-
o.
o
o
"O
c
Time (hrs)
Figure 2-11. Nylon 6,6 volatile evolution data of Wiloth (1971) and
model fit.
45
2 -2.3 Degradation Model
We now describe a model which is consistent with the known

reaction products and the minimum essential
reactions from Tables 2-1
and 2-2. From the data of Meacock and
Wiloth we know that reaction (s)
that produce amine end-groups and consume carboxyl end-groups are
necessary. Reactions (2-1.1), (2 -i. 2 and (2-2. lb) all
)
consume
carboxyl ends. However, reaction (2-1.1) is eliminated since the
experiments all begin with high molecular

weight starting materials
(high conversion > 98%) and the Flory distribution shows that
negligible amounts of adipic acid are present.
Reaction (2-1.2) is
included in our model as the step which consumes
carboxyl end-groups.
We have not included (2-2. lb) because, as discussed below, reaction
(2-2. la) is ruled out as a major step in the formation
of ammonia.
From the data of Kamerbeek, Kroes and Grolle (1961) it is clear

that not all of the C0 and NH^ produced can be accounted for from
2
reactions of the chain ends. For example, after 100 hours,
approximately 550 mmol/113g of NH^ and 330 mmol/113g of C0 have been

2
evolved, but at the start of the experiment only 10 mmol/113g each of
amine end-groups and carboxyl end-groups were present (113g/mol is the
molecular weight of the nylon 6,6 repeat unit). Clearly, the amide
linkage must be the source of much of the necessary oxygen and
nitrogen. Reactions (2-1.1), (2-1.4), and (2-1.8) all produce C0 -
Reaction (2-1.1) is excluded for reasons mentioned previously. The
reverse of the main amidation reaction produces a carboxyl end-group
but we are realistically ignoring this reversibility for the analysis

46
of the open reactor data where

water is rapidly removed. Reaction (2-
1.4) requires water reaction (2-1.9)
does not. Since we are assuming
that the gaseous products are
rapidly removed, reaction (2-1.9) will
be used in the degradation model.
Therefore, reaction (2-1.3) must be
included as a source of stabilized
end-groups for consistency.
Ammonia is produced by reactions (2-2. la) and (2-2.3a,b).
Looking more closely at the data of
KamerbeeJc, Kroes and Grolle (1961)
one sees that the final ammonia

concentration of about 550 mmol/113g
cannot be generated solely by reaction (2.1a) since the initial
concentration of amide linkages is 1,000 mmol/113g which
can produce
at most 1,000 amine end-groups and subsequently 500 mmol/113g of
ammonia. Therefore, it is not sufficient to include only
reaction (2-
2.1a) in the model; however, it is sufficient to
include only reaction
(2-2.3) which we have done in keeping with our goal developing the
simplest realistic degradation scheme. In addition, these reactions
give rise to a crosslink which is consistent with the data of both
Kamerbeek, Kroes and Grolle (1961) and Wiloth (1971)
The degradation model is summarized in Table 2-3 along with the
main amidation reaction. They are abbreviated as
2-3.1 C -» SE + W
2-3.2 L -» SE + A
2-3.3 SE C02 + SB
2-3.4 SB + 2A -» X + 2NH
3
2-3.5 Polyamidation A + C -» L + W
where SE refers to a stabilized (or cyclized) end-group, C02 refers to
carbon dioxide, SB to a Schiff base and X to a crosslink.

Table 2-3 Nylon 6,6 degradation
and crosslinking model.
O
O O O C
11
II
I / \
NH-C-(CH 2 4 .C-OH
) « NH-C CH CH, + H,0
\ /
CH 2 -CH 2
(2-3.1)
? ? o c
II II
II / \
NH-C-(CH 2 )
4
-C-NH « NH-C- CH CH 2 +
\ /
CH 2 -CH 2
H 2N (2-3.2)
O N
II
o C C
/ \ / \
NH-C- CH CH 2 <-> CH 2 CH 2 + C0 2 (2-3.3)

\ / \ /
CH 2 -CH 2 CH 2 -CH 2
Continued Next Page

48
N
C II
/ \
C
/ \
CH 2 CH 2 + 2 NH CH CH + 2 NH
\ /
\ /
CH 2 -CH 2 CH 2 -CH 2
(2-3.4)
O O
II
-C-OH + C-NH + H2 0 (2-3.5)

49
The rates (per unit volume) of the reaction steps

(2-3.1 to 2-
3.5) are assumed to be given
by
R * CT k X
l l C (2.21)
R
2
- CT X
L
(k
2
+ k
2C V (2.22)
R " C k X X
3 T 3 A SE (2.23)
R = C k X
4 T 4 A (2.24)
R
5
" CT k
app
(X
A
X
C
' ~\ X
L
) (2.25)
app
where {Cf - C + C + C + + C + C + cy, - C /C

A x etc., and
Q L SE Sfi ft A r
reaction (2-3.2) has been treated as an end-group
catalyzed reaction
as recommended by Twilley (1961) . We have also assumed that reaction
(2-2. 3b) is very rapid compared to (2-2. 3a). Therefore, the rate
limiting step for reaction (2-3.4) is given by reaction (2-2. 3a) . The
rate of (2-2. 3a) is given by C k X in eq

T 2t2a ^B A ' ( 2 -24). The rate
and equilibrium constants for the amidation reaction are known
functions of temperature and composition (see eqs. 2.9 - 2.14).
The mass balances for the various species are given by
dC
A
- R - 2 R - R (2.26)
2 4 5
dC
dT " " R l " R5 (2 ' 27)

50
R " R
dt l 5 (2.27)
R + R
dt 2 5 (2.28)
dC n
W
R + R
dt l 5 (2.29)
5*
dt
,
R + R ' R
l 2 3 (2.30)
dt ' ^4 (2.31)
dC
C02
dt * R3 (2.32)
"'SB B _
R R
dt~~ " 3 "
4 (2-33)
dc
x =
R 2 -34)
dt~ 4 (
Since we have assumed that the rate of removal of the volatile
components in the reacting mixture is very rapid eqs . (2.29), (2.31)
and (2.32) really reflect the rate at which these products are evolved
as gas per unit volume of the reacting mixture. The concentration of
water, carbon dioxide and ammonia is assumed to be negligible in the
analysis of the data in Figures 2-8 through 2-11. The sole exception
to this is for the closed reactor data of Meacock where we estimate
the water concentration as a constant from the the correlation of

Ogata ,1960, for the equilibrium
mole traction of water in the
poller
as a function of partial
pressure of water in the gas phase.
The temperature dependence
of the rate constants was fit
with the
usual Arrhenius form (see ea ? mi u
isee eq. 2.10) by optimizing an integral
. . .
analysis. The resulting set of

kinetic parameters is shown in
Table
Table 2-4. Kinetic parameters for nylon

6,6 degradation
Model Rate v
E T
u app 0
Reaction Constant (l/hr) (cal/mol) (°C)
1 k
! 0.06 30,000 293
2 k °- 005 30,000 305
2
2 k
2c
0-32 30,000 305
3 k 0.35 "305
3 10,000 a = 0.1
4 k
4
10-0 50,000 305 b = 0.3
The predicted end-group concentrations are compared to the data
of Wiloth in Figure 2-10. The fit is reasonably good at all

temperatures for the amine end-groups. However, the carboxyl end-
group predictions are slightly high, especially for the highest
temperature data. This is because the assumption of a zero water
concentration is not exactly correct and the carboxyl end-groups are
consumed by the main amidation reaction rather than degradation
reaction (2-3.1) in this simulation. Consequently, it is difficult to
estimate the rate of reaction (2-3.1) from the open reactor
experiments because it is masked by the main amidation reaction.

52
Therefore, we have estimated it from the closed reactor data of

Meacock, which is under a constant finite water concentrate
that so
the nidation reaction is
near equilibria throughout
the experiment.
The parameters obtained
from open reactor data can
accurately
describe the amine end-group
evolution in open reactors but
cannot do
the same for the closed reactor,
m
order to properly account for
the
effect of the reversible main
amidation reaction on the
concentration
of carboxyl end-groups we have
fit the amine end-group evolution in
the closed reactor with an
empirical function of the amide linkage
concentration and the amine end-group
concentration. This allows us
to accurately determine the rate of reaction (2-3
. 1) . These results
are shown in Figure 2-8. We are unable to estimate an activation
energy for reaction (2-3.1) since Meacock presents data only at a
single temperature. However, since (2-3.1) is so similar to (2-3.2)
we have set the two activation energies
equal.
The data of wiloth on the evolution of NH^

and C0 are shown in
2
Figure 2-11 along with the model predictions which seem to be as
accurate as justified by the data at the three temperatures

investigated. The fit of the long time data of Kamerbeek, Kroes and
Grolle for NH^ and C0 evolution is shown in Figure 2-9. Again, the
2
model provides an accurate quantitative estimate for the amount of
each of the components except, perhaps, at very long time.

53
2.2.4 Discussion
Many possible degradation mechanisms are postulated in the

literature for nylon 6,6. It is not feasible or necessary
to include
all the side reactions and/or
to determine their kinetic
constants in
order to describe the performance
of typical industrial reactors.
We
have hypothesized a minimal
subset of reactions that can
be used to
estimate the degradation and we have regressed
kinetic constants for
them that are consistent with
available data. We expect that this is
a representative degradation mechanism that would be sufficient for
estimating the low levels of byproducts
that can be permitted in the
typical industrial product. It has also been demonstrated that this
simplified degradation scheme can be readily
used in conjunction with
the kinetic and equilibrium model for the main amidation (see eqs.
2.27-2.35). This will allow realistic, high temperature process

modeling to be performed.
CHAPTER 3
FILM DIFFUSION EFFECTS IN NYLON

6,6 POLYMERIZATION
3.1 Introduction
Most esterifications and amidations are reversible and the

commercial production of polyesters and nylons requires significant
condensate removal in order to reach a useful molecular weight
(M -
20,000 g/mol) .
Therefore, proper understanding of the simultaneous
reaction-diffusion problem is necessary as a fundamental basis for
process models that incorporate mass transfer effects in
polycondensation reactors.
Other studies have examined the problem of simultaneous reaction
and diffusion in reversible polycondensations . Secor (1969) analyzed
the problem by numerically solving the penetration theory equations
but did not investigate the effect of film thickness on the reaction.
Hoftyzer and van Krevelen (1971) also used a penetration theory model
and linearized the second order polymerization rate expression; their
results are limited to small changes in bulk concentration. Gupta,
Agarwalla and Kumar (1982) studied the effect of mass transfer on
polycondensations in simple geometries (film, solid and hollow

55
spheres) at very low molecular weights.

Ault and Mellichamp (1972)
and Amon and Denson (1980)
have also formulated diffusion
models for
wiped-film reactors for polycondensations . In all of these studies,
the effect of mass transfer was
analyzed for systems with composition
independent reaction rate and equilibrium constants
and no gas-phase
diffusional resistance to mass transfer.
It is well-known that many

polyamidations and polyesterif ications
appear to change reaction order with
conversion. This apparent change
has been quantified for nylon
6,6 with an activity-based kinetic
model
described in Chapter 2 (see eqs. 2.9-2.14). We will investigate the
effect of these nonideal kinetics on a polymerization in a stationary
liquid film. We will not include the degradation reactions in the
kinetic scheme at this point. This is because we want to separate the
film diffusion effects from degradation effects

and also because these
reactions typically take place at low temperature (250°C) where

degradation is practically negligible (see eqs. 2.21-2.24 and
Table 2-
4) .
In the vapor space above the film, there is a large volume of
steam vapor. Mass transport by diffusion is assumed to take place
between the film and the vapor in a direction perpendicular to the
film. Initially, the film contains a nylon 6,6 prepolymer at
equilibrium. In contrast to previous studies, the water transport is
described by a constitutive relation which uses the gradient of the
mole fraction of water as the driving force for diffusion, thereby
leading to a model that requires only a single mutual diffusivity.
One of the goals of this study is to describe the time evolution
of the molecular weight and the polydispersity (averaged across the

film) for a range of representative film thicknesses and reaction
conditions. Although we will assume that
a Flory distribution exists
locally throughout the film,
the polydispersity of the
entire film may
differ from (This happens whenever
2. two of more fractions of the
same polydispersity but different
nuêr average molecular weights are
mixed.) A dimensionless group, the
Thiele modulus, which is the ratio
of a characteristic diffusion
to a characteristic reaction
time, will
be important for our modeling and for
interpreting the model results.
For example, it is expected that at low Thiele
moduli, the
polymerization will proceed under reaction control, while at high
Thiele moduli it will be limited by
diffusion. By simulation we will
determine the range of values of the Thiele modulus that lead to
diffusion control, or to reaction control
along with the range of
values where both are important.
3.2 Preliminaries
The conservation of species mass in an n component mixture can be
expressed by (e.g., Bird, Stewart, and Lightfoot; 1960)
3C.
JT +
^î = h (3.1)
where ft. is the flux of component i relative to fixed spatial
coordinates. This flux is defined as
= C.
if.
i li
v. (3.2)
57
where C. is the molar concentration and ?. is the veiocity of

component i relative to fixed spatial
coordinates. Most constitutive
relationships for diffusion are written, not in terms of i?., but in
terns of a flux relative to
some average velocity. The molar' average
flux of component i relative to the molar average velocity,
3. , is
a • • •
defined as
J - C. ( v. - v
A )
(3<3)
where the molar average velocity, v\ is defined as
n
Z C. v.
V "
— =
.
x
i
v
i
(3.4)
1 1
z c.
1
1-1
The relative molar fluxes of the species obey the

relation (see eqs
3 . 3 and 3.4)
1
(3.5)
i-i
It is important to relate the two fluxes in order to incorporate
an appropriate constitutive relation into the mass balance. The
necessary relationship is
n
Ff. = 3. + x, Z if. (3.6)
1 1 1 3
j-1
58
where x. is the mole fraction of

component i which is given by x. =
A _ 1
VC The total molar density,
C I c. is related to the s
the fluxes by the following identity that is used in deriving

eq
(3.6) .
n n
I N. - C v -
1
i=l 1 1 [
'
'
i-i
3.3 Constitutive Relations
For multicomponent diffusion, a common generalization of Fick's

Law is
n-1
~ J. = I d. V* C i=l, .,n-l
-ij
.
'
-j
.
•
(3.8)
j-1
In this relation, the driving force for diffusion is taken to be the
gradient of molar concentration, v*C Another class of constitutive
relations may be obtained by using the gradient of mole fraction, Vx

]
as the driving force (Bird, Stewart and Lightfoot; 1960), Young and
Stewart; 1986, Krishna and Taylor; 1986).
n-1 A
- J " * C D. . V* x. i = l,..,n-l (3.9)
A t
j-1
These two constitutive relations are identical if and only if the

A
total molar density (C) is constant. In general, the elements of the
two matrices of diffusion coefficients D. and d. . will not be the

59
same, although they are interrelated.

In addition, these matrices are
often composition dependent,
if the Onsager reciprocity
relations are
valid (Onsager; 1945) then these matrices possess certain strong
"positiveness" properties (Miller,
Vitigliano and Sartorio; 1986).
However, the validity of the Onsager
relations has been challenged in
numerous publications e.g., Truesdell
(1969).
For a binary mixture of polymer molecules (P) and water (W) the
diffusivity matrix reduces to a scalar.
This diffusivity describes
the relative mobility of the two species and may be composition
dependent. From eqs. (3.5) and (3.9) and the definition of mole
fractions (x
p
+ ^ = 1) we see that (Krishna and Taylor; 1986)
Therefore, if eq. (3.9) is used as a constitutive relation, it is
inconsistent (and, in fact, unnecessary) to assume anything but a
single diffusivity for a binary system. This is true even if one of
the two components is essentially stationary relative to fixed
laboratory coordinates (e.g., an entangled polymer in a solvent).
This further emphasizes the fact that the single diffusivity describes
only the relative motion of the two components.
From eq. (3.8) and the definition of the total molar density (C =
C_ + C T7 ) it follows that
r W
= " d * c + d * c (3.11)
<
<*w P >
w P
60
This relation shows that the diffusivity

of the polymer and the water
are equal only when f £ . cf,
a
species of very different sizes.

very unlikely occurrence fQr
^
If the composition dependence
of the diffusivity for the nylon
6,6-water system were known for each
constitutive law, there would be
no difference in using either
of them. Since we do not have any data
on the composition dependence, and do not know the relative
sensitivity of the dif fusivities in
the two constitutive relations,
the choice becomes one of convenience
and we use eq. (3.9) since it
leads to a single diffusivity for a
binary system. In addition, the
Vrentas and Duda free volume theory (1976, 1977a
and 1977b) predicts
that the diffusivity in eq. (3.9) should have a weak dependence on
molecular weight and on solvent concentration under the conditions
being studied (T > T + 200 °C and less than 10 wt % solvent),

g
especially for a solvent molecule as small as water. Limited
experimental evidence on a crystalline polymer in a solvent
(polyethylene in o-xylene) is in agreement with this prediction
(Vrentas, Duda, and Lau; 1982) . This is important as we will treat
the diffusivity as independent of concentration in our model.
3.4 Stationary Film Model
The polyamidation reaction may be represented as
A + C = L + W (3.12)
61
where an amine ,A) and carboxyl , C , end-group react to form an amide

linkage (I) and a water molecule (W) . Ihe re aotion rates for the
individual species are related
by
R =
(3.13)
The reaction rate is given by
R = k C X X
app T
(
A " X (3-14)
C | L *w >
app
where X. = c./C and C - C + C + C

T T + C are the mole fractions
ft c L w
and total molar concentration for the
reacting mixture. The apparent
rate and equilibrium constants in eq.
(3.14) are complicated but known
functions of temperature and composition (see

eqs. 2.9-2.14).
There are two molar concentrations and

(C c ) and two
corresponding mole fractions (x., X.) for the reacting mixture. For
diffusion, the mixture is treated as a binary mixture of polymer and
water since the end-groups of a polymer molecule can not diffuse
independently from its amide linkages. Therefore, the binary total

A A
molar density and mole fractions are C - C n + C r7 , xr = C /C and x
P W W W P
A
Cp/C. This implies that the polymer concentration is equal to the
concentration of either end-group for a stoichiometrically balanced
feed (i.e. C = C = C ) . However, to describe the reaction, the

p ft Q
mixture is treated as one containing four components; amine and
carboxyl end-groups, water and amide linkages since the reactivity of
these species depends only on the local environment. The total molar
62
concentration for the quaternary

mixture and the mole fractions are
C
= C + C + C + T
A C and = C./C
c L M *i It should be noted that for a
.
T
polyrcer-solvent system the approximation that the
total binary molar
A
concentration (C) rnnofanf is
is constant ^„ a
poor one. We will not assume that
C or C is constant.
T
The motion of all the chains

of various lengths in the reacting
prepolymer is described by a single
diffusivity. Otherwise one would
be forced to treat every species
of differing chain length as an
independent component giving rise
to many additional dif fusivities
and
conservation equations. We will treat
the diffusivity as composition
independent and the polymer will
be assumed to be immobile with
respect to fixed coordinates =
(i?
p
tf> . Then, fromeqs. (3.7) and
(3.9)
A
C D
~ ' v
T^l^ *w (3.15)
where mole fractions denoted by lower case x are based on the

pseudo-
binary representation of the liquid solution, i.e. x, = C /C and x =
,
A V? w p
Cp/C. The assumption that the polymer does not move is frequently
employed (e.g., Ault and Mellichamp; 1972, Amon and Denson; 1980) and
is a good approximation when the polymer chains are entangled, but may
be less accurate in the early stages of the polymerization.
Combining eq. (3.15) with the continuity eq. (3.1) leads to the
following evolution equation for the water concentration.
ac ax
w a c w 2
w
63
Similarly for the polymer,
P 2
dt " R
(3.17)
The variables in the .ode! have been scaled with respect to a

characteristic time, length,
concentration and reaction rate
constant.
All further references to
quantities appearing in egs.
(3.16) and
(3.17) are dimensionless We have chosen the characteristic
.
time as
2
(H /D) where the diffusivity is taken at the temperature of the
system. The length was chosen as the film height (H) and the
concentration as the initial value for water (C°) . For the
characteristic reaction rate constant we use the value at 200 °C as
both the water and the carboxyl end-group
mole fractions approach zero
O
(k = 2 9 26/hr) in
app
-
. chemical reaction engineering, the
dimensionless group that appears in eqs. (3.16) and (3.17) is the
Thiele modulus (e.g., Smith; 1970) which may be thought of as the
ratio of a characteristic diffusion time to a characteristic reaction

time
2
H D t..«
0
A ^ _
1
/
k
— -
i
diffusion
reaction
_
(3.18)
/
app
The Thiele modulus will be important in interpreting the simulation
results
From eqs . (3.16) and (3.17) we can see that two boundary
conditions on the water concentration together with initial conditions

64
for both polymer and water are required.

The initial concentrations
(t < 0) of amide linkages, amine
and carboxyl ends and water
molecules
is assumed to be uniform throughout the film. For t > 0 the water
mole fraction at the gas-film
interface is set to a fixed value
depending on the steam temperature and
the pressure; Henry's Law can
be used to determine this constant
=
*W yW P 1 * P
(3-19)
W w
where y is the water mole fraction in

w the gas phase, y the activity
w
coefficient of water in the liquid phase, p'
the vapor pressure of
water and P is the total pressure. The activity coefficient of water
is equal e 1+x at infinite dilution is the Flory-Huggins interaction
( x
parameter) . Note that the mole fraction of water in the liquid

phase
in eq. (3.19) is a quaternary mole fraction which is consistent
with
the fact that the activities of the solution should be described by
these mole fractions (see Chapter 2) . Since the incoming feed is
about 90 mole % amide linkages one may easily show that the
A
concentration of water (C - C) can only change by about 3% at the
w
interface (Z = 1) even if the reaction proceeds to completion. Thus,
C can safely be taken as a constant at the gas-film interface. The

w
final equilibrium molecular weight can change by only 1.5% if C T7
w
changes by 3%.
At the wall (Z = 0) we assume that the flux of water is zero (tfT7 =

w
(?) or equivalently dx /dZ = 0 from eq. (3.9).
65
The method of lines (Press

et al.; 1986) was used to
integrate the
partial differential equations
(3.16) and (3.17). since „e are
neglecting side reactions, assumng
that the polymer dees not diffuse
and that an equimolar feed is
used, the amine and carboxyl end-group
concentrations are equal to each
other and to the concentration of
polymer molecules. From the change in the amine end-group
concentration and the
tne In-iHai
initial ^r^'*-'
condition, the
.
amide
.
linkage
concentration was calculated.
A comprehensive picture of the relative

importance of reaction and
diffusion can be found by examining values
of the Thiele modulus range
from 0.1 to 100 (see Table 3-1), which correspond to film thicknesses
of 0.2 to 5.5 cm and a value for the diffusivity

of water in a nylon 6
melt at 265 °C of 2.5x10 cm /s (Nagasubramanian and Reimschuessel;
1973) The values of the characteristic diffusion time

2
.
(H /D) are
also shown in Table 3-1. The characteristic reaction time (l/k° ) is

apP
0.34 hr.
Table 3-1. Typical values of the Thiele modulus.
2 2
H H /D
(cm) (min)
0.2 2 0.1
0.4 10 0.5
0.75 40 1.8
1.2 100 5.0
1.7 200 10.
5.5 2000 100.
66
The value of the Thiele

modulus depends directly on the
value of
the reference reaction
rate constant (characterise
reaction time)
and on the system temperature
through the temperature
dependence of
the diffusivity. we have chosen
rate constant as the water

^
to be 2 926/hr , the value Qf
and amine end-group concentrations

.
^
approach
zero at 200 *C. If we had chosen 240
<C as our reference temperature,
the Thiele modulus would range
from 0.6-600 for the same
values of
2
H /D in Table 3-1. Uncertainty in the diffusivity
would also directly
affect the value of the Thiele
modulus.
Initially, the film contains a nylon 6,6 prepolymer (M = 3,500
n
g/mol) at equilibrium (250 °C) containing 7.5% (wt) water. This is
below the melting temperature of dry nylon 6, 6 (265 °C, Sweeny and
Zimmerman; 1964) although it is a typical low molecular weight
processing temperature in certain stages of the continuous
polymerization (U. S. Patent 2,361,717). Flory's (1953) thermodynamic
theory for binary polymer-diluent solutions predicts the following
melting point depression relationship.
11 R V
—
~
T
mm
" -o -
T AH
u
(—
V,
1
) ( v-
ill - x, v1 ) (3.20)
Here T^ and T are the melting point of the polymer in the polymer-
m
diluent system and of the pure polymer, R is the gas constant, AH is
u
the enthalpy of melting per repeat unit of the polymer, is the
molar volume of the repeat unit of the polymer, V- is the molar volume
of the diluent, v^ is the volume fraction of the solvent and %^ is
Flory's interaction parameter. The volume fraction of solvent is

67
n
defined as - _____ l
where ^ and ^ are fche number Qf moles Qf
polymer and solvent, respectively and x is the degree of

polymerization of the polymer (which is assumed
to be monodisperse) .
The interaction parameter is defined by AH = R t

Xi n, v 0 where AH
is the enthalpy of mixing and
^= 1^ is the volume fraction of
polymer. Flory (1953) has shown that the interaction parameter
must
be less than 0.5 for spontaneous mixing to occur.
For nylon 6,6 T° is 265°C, AH^ is 5288.4 cal/mol, (Sweeny and

3
Zimmerman; 1964), V is 121 cm /mol at 1 atm and 265°C
u (Van Krevelen
and Hoftyzer; 3
1976) and V is 23 cm /mol for water in equilibrium
1
with saturated steam at 265 °C (ASME Steam Tables; 1967) . in Figure

3-1 we show the predictions of eq. (3.20) for the melting point of
nylon 6,6 vs. the volume fraction of water for several values of the
interaction parameter (0.5, 0, -0.5). The melting point is only
slightly affected by the value of the interaction parameter,
particularly at low volume fractions. These calculations are in
agreement with the qualitative observation of large reductions of the
melting point (40-50°C) for thermodynamically good solvents in dilute
solutions (Billmeyer; 1984) . In the stationary film model the inlet
feed has a water content of 7.5 wt % and if allowed to come to
equilibrium has a final water content of 0.4 wt %; these correspond
to v^ values of 0.33 and 0.025. Thus, the melting point of the
flasher effluent is expected to be 251°C.
The water concentration profile in the film as the Thiele modulus
approaches zero is shown in Figure 3-2. The base of the film is at
2
Z=0 and the gas-film interface is at Z=l . At = 0 the reaction is
68
270
Volume Fraction of Water
Figure 3-1. Predicted melting point of nylon 6,6 and water.

69
0.2-
0.1 -
0.0
0.2 0.4 0.6 0.8 1.0
Position (I)
Figure 3-2. Water concentration profiles for the nylon 6/6

2
stationary film model ( = 0) .
70
zero and the usual solution to the diffusion equation

is obtained.
The simulation results for
a Thiele modulus of 0.5 are
shown in Figure
3-3. in this case, the reaction proceeds
at a fairly constant rate
through the film as shown by
the flat amine end-group
concentration
profile; however, the water profile varies
significantly. This i s
because the reaction is proceeds
slowly enough so that the mixt
ure
remains far from equilibrium and
nearly approximates an irreversible
polymerization. An irreversible reaction would
not be affected by the
water concentration and would proceed with
the same reaction velocity
through the entire film giving rise
to a completely flat amine end-
group profile. As expected, the water concentration profiles are
similar to Figure 3-2.
The results for a Thiele modulus of 5 are shown in Figure 3-4

where the curvature of the water concentration
profiles has changed.
The amine end-group profiles also vary significantly through
the film
and reach equilibrium at the gas-film interface by t=2,
as indicated
by the amine profiles merging into a single point at Z=l . At long
times the entire film will reach equilibrium. The kinks and bumps in
the amine end-group profile are due to the fact that both the reaction
rate constant and the equilibrium constant change significantly with
composition across the film; if these parameters are taken as
constants, these features disappear.

2
Eventually, the entire film will reach equilibrium if $ > 0. For
this example, the equilibrium molecular weight (number average) is
15,400 g/mol and the water concentration is 0.4 wt % (X=

w
0.025). The
0.2-
0.0
0.0 0.2 0.4 0.6
1
—
0.8 1.0
Position (Z)

for the nylon
2
6,6 stationary film model ($ = 0.5) .
72
0.0 l
r
0.0 0.2 0.4 0.6 0.8 1.0
Position (Z)
Figure 3-4. Water and polymer concentration profiles for the nylon
2
6,6 stationary film model ($ = 5).
73
evolution c£ the number average
^^^^
molecular weight, averaged over the
film thickness, is 3hown in Figure ftt very
Thiele modulus, the reaction

time is much smaller than the
diffusion
time, local equilibrium is attained quickly
throughout the film and
the process is diffusion controlled. The upper limit is always set
by reaction equilibrium, but
it is diffusion that controls
the rate of
approach in this case.
Figure 3-5 is most useful when one wants to anticipate the

benefits of changing the reaction
rate (and hence the Thiele modulus)
e.g., via a catalyst, while maintaining a constant film
height and
temperature. This keeps the characteristic time the same for the
various curves and allows a direct comparison between them.
At low
Thiele moduli, it takes a long time to reach
the equilibrium molecular
weight, and there is a big incentive to increase the reaction rate
(via catalyst) 2
thereby increasing and greatly reducing
«(,
the
transient time. However, at Thiele moduli greater than 5 very little
enhancement is possible since diffusion of water out of the film
controls the molecular weight growth. High reaction rates rapidly
produce local equilibrium in the film and increasing the reaction rate
does not speed the net growth of molecular weight.
To analyze the benefits of decreasing the film thickness
(decreasing Thiele modulus) , Figure 3-6 is more useful because the

2
abscissa, t = t<|> , is a new dimensionless time, scaled by the
characteristic time for reaction, that is independent of the film
thickness (see eq. 3.18). Under fixed reaction conditions and for
Thiele moduli greater than 2, Figure 3-6 shows that decreasing the
film height always leads to significantly shorter transient times.

74
1 6000
1 4375.-
1 2750.-
Ll_
1 1 1 25.-
0)
9500.-
7875.-
o
6250.-
c
4625.-
3000.
0.0 3.0 6.0 9.0 12.0 1 5.0
Time (t)
Figure 3-5. Film average M evolution for the nylon 6,6 stationary
n
film model at various Thiele moduli.
75
1 6000
1 4375.-^
v
1 2750.-^
1 1 1 25.-
CD
9500.-
7875.-
o
6250.
4 6 2 5.-
3000
0.0 4.0 8.0 12.0 16.0 20.0
n
film model at various Thiele moduli.
76
However, very little 2

benefit is realized below = 1;
, at low Thiele
moduli diffusion is fast and
the moleoular weight growth
is reaction
controlled. Decreasing the film height
further is not product lve
The film-average molecular 2
weight for *
= 5 and 100 is shown in
Figure 3-5 and in more detail in Figure
3-7. The two profiles cross
twice. These curves can only be compared
if one considers a system
with a fixed film height (characteristic time) in which the Thiele
modulus is changed via catalyst
addition. The corresponding molecular
weight profiles are shown in Figures
3-8 and 3-9. At short times, the
higher Thiele modulus corresponds to a more rapid generation of
molecular weight at the gas-film interface. However, water is
replenished more rapidly at the interface
(due to the higher reaction
rate) thus keeping the concentration of water in the interior
of the
film higher relative to the lower Thiele modulus film. Since the
process 2
is diffusion limited (high <(, ), and begins at eguilibrium, a
certain amount of water must diffuse out of the interior

of the film
before the reaction can proceed. Eventually, enough water diffuses
out of the film so that the higher reaction rate does indeed increase
the rate of molecular weight generation. This anomaly does not occur
at low Thiele moduli as the diffusion is rapid relative to the
reaction.
The predictions of this nylon 6,6 kinetic model were compared to
those of an idealized model based on constant rate and equilibrium
constants in the kinetic description; forward reaction rate constants
of 177/hr and 58/hr were used. These correspond to the initial and
the equilibrium (t -> <*>) values of ^ in the nylon 6,6 stationary

a pp
film simulation. An equilibrium constant of 470 was used, which is

77
16000.
14373.-
12730.-
11123.-
5
2 • 300.-
0»
> 7873.-.
<
9
A 6250.-
E
3
4623.-
3000
Time (t)
n
2
film model for * = 5 and 100.
78
0.0 0.2 0.4 0.6 0.8 1 .0
Position (Z)
Figure 3-8. Molecular weight profiles for the nylon 6,6 stationary
2
film model (+ » 5) .
79
1 6000.
1 4400.
^ 1 2800.-
O 1 1 200.
9600.
^ 8000.-
6400.
4800.1
3200.
1 600.
0
°-0 0.2 0.4 0.6 0.8 1 .0
Position (Z)
Figure 3-9. Molecular weight profiles for the nylon 6,6 stationary
2
film model (+ = 100).
80
the equilibrium value in

the nylon 6,6 simulation.
The results f rom
the idealized model are compared to those of
the nylon 6, 6 model in
2
Figure 3-10 for ,
= 5 and, 2 =0 .2. At the larger Thiele modulus,
the ideal model does not have muoh error at
long times because the
effect of variations in the forward reaction rate constant with
composition are diminished as
equilibrium is reached in the film.
The
non-ideal model predicts a faster
molecular weight evolution because
the equilibrium constant is greater than
the final value of 470 (K
a PP
increases with water content in this
composition regime) during the
course of the simulation.
The molecular weight evolution for 2

=
* 0.2 is also shown in
Figure 3-10. This Thiele modulus corresponds to
a film thickness of
about 0.25 cm. The ideal model has significant error

because it does
not properly reflect the changes in the reaction
rate constant. If a
mean value for the forward reaction rate constant
of 117/hr is used
the agreement is only slightly better. The characteristic reaction
time is 0.34 hr, and Figure 3-11 shows the prediction of the time
required to reach a given molecular weight can easily be from 0.5 to 1
hr in error (20-40% difference) . Similarly, the prediction of the
molecular after a given reaction time can be 500 to 2,000 g/mol in
error (about 30% difference) . Thus, it is necessary to properly
describe the composition dependence of both the rate and equilibrium
constants for an accurate model. In addition, even if the composition
dependence of the rate constants is known, there is no obvious way to
choose an average value so that the ideal description may be used.
For systems with a significant composition dependence the ideal model

81
0.0 3.0 6.0 9.0 12.0 15.0
Time (1)
Figure 3-10. Film average M evolution for the nylon 6,6 and ideal
2
stationary film models for + =0.2 and 5.
82
1 6000
1 4375.-
k V
- 12750.- app
Ll
177/hr
1 1 1 25.-
CD
1 17/hr
9500.-
Mode
7875.-
O
58/h r
6250.-
4625.-^
5000
0.0 4-0 8.0 12.0 16.0 20.0
T
Figure 3-11. Film average M evolution for the nylon 6,6 and ideal
n
2
stationary film models for + =0.2.
83
results will have large errors,

especially at low values of the
Thiele
modulus (i.e., under reaction
control).
We have assumed the
polydispersity to be 2 at all points wxthin
the film which is the classic result of
Flory (1953) for the equal
reactivity case. However, the film is really a mixture Qf many
fractions with the same polydispersity
(2) but with different number
average molecular weights. Thus, the polydispersity of the enti re
film will deviate from 2 if the number average
molecular weight varie s
from point to point. The film averaged number average molecula r
weight, weight average molecular weight and polydispersity can be
calculated from the number average molecular
weight profile. The time
evolution of the film averaged polydispersity
at various Thiele moduli
is shown in Figure 3-12. The polydispersity deviates by as much as

20% from a value of 2 for Thiele moduli in the range of 0.2 to
100.
All of the deviations due to mass transfer act to increase the
polydispersity, particularly at short times. The deviations increase
with Thiele modulus and a maximum is seen near a time of 0.5 to 1. As

the film approaches equilibrium the polydispersity deviates negligibly
from 2. If one considers the apparent rate and equilibrium constants
to be composition independent (Figure 3-13) the polydispersity shows
the same overall pattern but the variations are smaller in magnitude.
The maxima also occur at longer times with decreasing Thiele modulus.
Mass transfer causes relatively small deviations in the
polydispersity from a value of 2. Thus, if large deviations from 2
are observed they can be attributed to changes in end-group reactivity
with chain length (a failure of the equal reactivity hypothesis)
and/or additional side reactions rather than to mass transfer effects.

84
2.6
2.5- A
2.4- 1 00
if)
1 0
CD 5
a 2
if)
• — 1 I
0.5
o 0.2
Q_
1.9-
1 .8
°-0 1.2 2.4 3.6 4.8 6.0

Time (t)
Figure 3-12. Polydispersity evolution for the nylon

6,6 stationary
1
film model.
85
1 .9-
1
1
1 1
i ' 1
I
°-° 1-2 2.4 3.6 4.8 6.0

Time (t)
Figure 3-13. Polydispersity evolution for the ideal stationary film

model.
86
3.6 Conclusions
It is necessary to have accurate rate and kinetic

constants to
accurately describe the molecular
weight generation of nylon 6,6 in
a
film. At a fixed film thickness, it is beneficial
to increase the
reaction rate via catalyst addition
until a Thiele modulus of 5,
because molecular weight generation is diffusion
limited beyond this
point. At fixed reaction conditions,
it is beneficial to decrease the
film thickness until a Thiele modulus of 1, since molecular weight

generation is reaction limited below this
value. Mass transfer causes
the polydispersity averaged across the
film to deviate somewhat from a
value of 2 at short times.
CHAPTER 4
A FLOWING FILM MODEL FOR

NYLON 6,6 POLYMERIZATION
4.1 Introduction
Process equipment which serves the dual purposes of condensate
removal (separator) and molecular weight generation (reactor) are

common in continuous polycondensations. An example of such a device
is the "flasher" (U. S. Patents; 2,361,717; 3,900,450; 3,960,820)
which is used in the continuous production of
nylon 6,6. This unit is
essentially a pipe in which a prepolymer flows along the wall
driven by a large quantity of flowing steam; this complicated two-
phase flow may result in instabilities that limit the range

of steady-
state operation (U. S. Patent; 3,900,450). By introducing a flow
field into a stationary film model (see Chapter 3) we will study the
effect of distributed residence times on the performance of continuous
nylon 6,6 film devices and determine the key variables which affect
their design and performance.
Previous studies have considered the problem of simultaneous
reaction and diffusion for reversible polycondensations in simple
geometries (Secor; 1969, Hoftyzer and van Krevelen; 1971, Gupta,
Agarwalla and Kumar; 1982) . In all of these studies, the effect of

88
-as transfer was analyzed for

systems with composition independent
reaction rate and equilibrium constants. However, it is well known
that many polyamidations
and polyesterif ications appear to change
reaction order with conversion. This apparent change has been
quantified for nylon
6,6 with an activity-based kinetic model (see
Chapter and the effect of these
2)
nonideal kinetics on the molecular
weight evolution in a stationary
flla nas been analyzed (gee
3) .
in this work we will assume

that the volume of steam is much
greater than the polymer volume (see Figure 4-la) , enabling us to
approximate the thin tubular film as
a flat film (see Figure 4-lb) .
The accuracy of this assumption

depends on the ratio of the prepolymer
film thickness to the radius. If this ratio is small, as
we expect,
the flat film is a good approximation.
Diffusion of water is assumed
to take place from the film to the
vapor in a direction perpendicular
to the flow.
One of the goals of this study is to describe

the steady state
number average molecular weight and average water
concentration of the
film effluent for a range of film thicknesses and reaction conditions.
Two dimensionless groups, the Thiele modulus and the Damkohler

number
will characterize the results. The Thiele modulus is the ratio of a
characteristic diffusion time to a characteristic reaction time and
the Damkohler number is the ratio of a characteristic residence time
to a characteristic reaction time. We will also determine the effect
of the flow field on the molecular weight profile in the film. The
results for nylon 6,6 will be compared with an idealized kinetic model
which employs composition independent rate and equilibrium constants

89
Figure 4-1. Flowing film model a) Flasher cross-section and b)

flowing film geometry.
90
in order tostudy the effects of

apparent change in reaction order
in addition, the effect
of three flow fields,
a plug, a parabolic and
a linear velocity
profile, will be examined.
4-2 Flowing Film Model
The differential equations

for the flowing fil m mod el
are similar
to those for the stationary
f Um model but they
CQntain an addUional
convective term. In dimensionless form they
are
at az < TÎ az >

+ R " v
y ay
(4.1)
aCp 2
^
at" --
.2 _ ,*
R " v (4.2)
y ay
where Y is the flow direction and v is the fluid velocity. The

y
characteristic length in the flow direction is the reactor
length, L.
The characteristic velocity is the film average

velocity <v > The
y
other variables and their corresponding characteristic scales are
discussed in Chapter 3. In deriving eqs. (4.1) and (4.2) we have
assumed that the reacting mixture can be treated as a binary solution
for diffusion (pseudo-binary approximation) , the mutual dif fusivity
depends only on temperature but not on composition, that the polymer
does not diffuse in the Z direction, that water diffuses in the Z
direction and that diffusion is negligible compared to convection in
the Y direction.
The Thiele modulus 2
(* ) is defined in Chapter 3 (see eq. 3.18)
The Damkohler number (Da) is give n by (e.g., Lin and Van Ness; 1973)
L/<v > t
Da = 5L - reside nce
l/k°
17
t " (4.3)
app reaction
where <v > and L are the average

y velocity and the axial length of the
O
film k is the reference apparent
'
a PP reaction rate constant for nylon
6,6 polyamidation (see Chapter 2). This Damkohler number is

independent of temperature since we have chosen a fixed reference
state since our reaction rate constant
which is both temperature and
composition dependent.
The model is useful for design or

steady-state control studies.
For example, the Damkohler number would reflect
a change in residence
time under fixed reaction conditions. The Thiele modulus would

reflect a change in the film height (diffusion time) under constant
reaction conditions. Two limiting conditions are obtained at very
high and very low Damkohler numbers. As Da-* °° the stationary film
result is recovered, and as Da-+ 0 the residence time approaches zero
thus effectively suppressing all reaction.
The velocity field in the flasher is complex and probably not
adequately described by any simple profile. However, we will
investigate two simple flow fields which provide a distribution of
residence times and look at the sensitivity of the model to the
details of the flow. The first is parabolic (0 < Z < 1)

92
V 2
= (2Z - Z
y l )
(4.4)
and the second is linear
v = 22
y (4.5)
Strictly speaking, the parabolic

flow field is actually quadratic
although we shall refer to it as parabolic. These two flow fields
will be compared to a a single
residence time plug flow field.
V =1 (4.6)
y
A plug flow profile, like a stationary film, has a single residence

time rather than a distribution of residence times. Eqs. (4.4-4.6)
have been scaled by the average film velocity <v > and are
y
dimensionless
The volumetric throughput Q is
Q = W <v > H (4.7)

y
where W is the film width (for a thin tubular film W is the
circumference) . For a given throughput Q, <v > and H (and therefore,

2
<j> and Da) can be treated as independent parameters.
The boundary conditions are identical to those used in the
stationary film model. Thus, C is constant at the gas-film
interface, and at the wall (Z=0) we assume that the flux of water is
zero (Ff = (?) or equivalently dx/dZ = 0.

93
The Thiele modulus is

proportional to the value of the
reference
reaction rate constant
(characteristic reaction time, and
also depends
on the system temperature
(through the diffusivity)
Typical values .
of the Thiele modulus range

from 0.1 to 10, which correspond
to film
thicknesses of 0.2 to 1.7 cm (see
Table 3-1).
The value of the Damkohler number

depends on the characteristic
reaction time and on the mean
residence time (L/<v » The
.
Damkohler
y
number is independent of the system
temperature and typical values are
given in Table 4-1.
4-1. Typical values of the Damkohler number.
L/<v > Da
y
(min)
1. 0.05
10. 0.5
50. 2.5
100. 5.0
1000. 50.
The model equations were integrated by using the method of lines
(Press et al.; 1986). At the inlet the film contains a nylon 6,6
prepolymer (M « 3,500 g/mol) with a composition corresponding to the

n
reaction equilibrium at 250 °C and 7.5% (wt.) water. The film
temperature was estimated from the patent (U. S. Patent; 3, 900, 450)
and all simulations were run isothermally at 250 °C. The steady-state
number average molecular weight for Damkohler numbers in the range of
0.04-5.0 and Thiele moduli of 0.2-5.0 were obtained.

94
A typical model result for nylon 6, 6 with a parabolic velocity

profile, a Thiele modulus
of 1.0 and a Damkohler number of 1.0 i,
shown in
Figure 4-2a; this
corresponds to a film height of
about 0.6
cm and a residence time
of 20 min. The wall-film interface is at
>0
and the gas-film interface
Iace 1S
is at z7 -l; i-u
at-
the inlet is at Y=0 and the
1 •
,
outlet is at Y=l, as noted in «-h* *

the figure. m The molecular weight
increases from the inlet
to the outlet as expected.
Under these
operating conditions, the highest
molecular weight material at
the
outlet is found at the wall-film
interface (~
j-nterrace ?nn ~/™«-n
(= 8,200 u
g/mol) whereas ft
the
molecular weight of material
at the gas-film interface is
significantly lower (. 6,100 g/mol,.
This is because mass transfer is
fast at this Thiele modulus and
the process proceeds under
reaction
control. Therefore, the material with the
longest residence time has
the highest molecular weight.
Figure 4-2b shows the corresponding results for a parabolic

velocity profile, a Thiele modulus of 5.0 and a Damkohler number of
1.0 (corresponding to a film thickness of 1.2 cm and a residence
time
of 20 min) Under these conditions, the
. highest molecular weight
material at the outlet is found near the gas-film interface
(- 6,200
g/mol) whereas the molecular weight at the wall-film interface is
significantly lower (- 4,300 g/mol). At this Thiele modulus, the mass
transfer is slow and molecular weight generation is greatest in the
material near the surface of the film, since this is where water can
be most easily removed.
The model results for a parabolic velocity profile, a Thiele
modulus of 5.0 and a Damkohler number of 5.0 (corresponding to a film
height of 1.2 cm and a residence time of 100 min) are shown in Figure
95
4-2o. At the outlet, the

moleeular weight i. smallest in
the oenter
of the film „i th higher molecular weight
material at the wall-fil m
interface and at the rr
t-hp 30 «ji
gas-fUm interface. The region of higher
molecular werght near the gas fil m
interface is due to betfcer magg
transfer, while the material
at the wall also yields
a high value by
virtue of a larger residence
time. Thus, by manipulation of the
film
thickness and residence time
the molecular werght
profile can be
changed significantly. This is in sharp contrast
to the stationary
film (or equivalents, plug flow)
case (see Chapter 3) where highest
molecular weights are always found at the gas-film interface.
However, a flat molecular weight
profile is never achieved in either
case, although this may be desirable
because variations is the profile
cause broadening of the film-averaged
molecular weight distribution.
The integrated "cup-mixing" average outlet molecular weight
is a
quantity of major interest, and it can be obtained by averaging of the

outlet molecular weight profiles (e.g. Figures 4-2a
, -» 4-2c) across
the film. The steady state cup-mixed number average outlet
molecular
weight for the nylon 6,6 flowing film model with a
parabolic velocity
profile is shown in Figure 4-3a as a function of the Thiele modulus
and the Damkohler number. The largest molecular weight is obtained
for thin films (small Thiele moduli) and long residence times (large
Damkohler numbers) . Under these conditions, mass transfer is fast and
the molecular weight evolution proceeds under reaction control. In
fact, the outlet molecular weight always increases with residence time
(Damkohler number) at constant film height (Thiele moduli) .
Conversely, molecular weight decreases with increasing film height

Coi-Film
Interface
4,300 g/mol
0 1
Molecular weight profile for the nylon 6,6 flowing film

model with a parabolic velocity profile a) Da 1.0,
2 2
= 1.0, b) Da = 1.0, = 5.0, c) Da - 5.0, $ - 5.0).
97
1
I
-23
i
i
2.52
—
r— r-
3.76
Damkohler Number
Figure 4-3. Number average molecular weight for the effluent of the
nylon 6,6 flowing film model with a parabolic velocity
'profile vs. the Thiele modulus and Damkohler number
(Contour Interval - 500 g/mol)
98
(Thiele modulus) at
at- all fK,
all
3, the residence times
•
,
(Damkohler numbers) in the

investigated range. However
owever, t-ho^
there is a much greater
increase in the
outlet molecular weight
gnt when the
fh a r OB ^
residence time (Damkohler number)
is
increased than when the film u, •
height v,
(Thiele modulus) is reduced
These observations are
more easily seen in
Figure 4-3b.
The steady state
with film
oup-mixed average outlet
the nylon 6,6 flowing film
shown in Figures 4-4a and

model „ ith .
4-4b
4b.
water conoentration for
mu
The water concentration
^ ^
increases
height (Thiele modulus) at constant
constant- r ..'H
residence time
(Damkohler number) and decreases with increasing residence time
(Damkohler nuêr) at constant film height
(Thiele modulus) . However
this decrease is very slight
beyond a Damkohler number
of 2.0. Figure
4-4 also reveals that for thin
films (low Thiele moduli)
most of the
water is removed after a
relatively short residence
time (small
Damkohler numbers). For thick films it takes a very long residence
time to remove most of the water.
Figures 4-3 and 4-4 show that for
a
thin film (0.25 cm corresponding to a Thiele
modulus of 0.2) with a
short residence time (12 min corresponding to a Damkohler number of
0.6) it is possible to remove most
of the initial water from the film
and realize very little increase in the

molecular weight (viscosity)
This can be important if a large viscosity
increase due to increasing
molecular weight has to be avoided because of the
different processing
requirements of very viscous fluids.
Figures 4-3 and 4-4 represent the performance of flowing film

nylon 6,6 model in a very compact form. They can be used to design a
film device (determine necessary length and diameter) to achieve a
desired molecular weight at a specified throughput. Alternatively,

99
5.0
4.2 -
in
~ 3.4
XD
O
2.6 -
Pi 1.8
1 .0 -
0.2
0.04 1.28 2.52 3.76 5.00
Damkohler Number
5.0
0.04
Thiele
0.2 Modulus
5.00
Damkohler Number
Figure 4-4 Average water concentration for the effluent

of the
nylon 6,6 flowing film model with a parabolic
velocity
profile vs. the Thiele modulus and Damkohler
number
(Contour Interval = 0.075).
100
they may be used to predict

how a ohange in operating conditions
win
effect the outlet .olecular weight for an existing unit, which
is
useful for control studip<?
studies. These points will be illustrated later
with an example.
The characteristic reaction time appears in both the Thiele

modulus and Dam*ohler number (see eqs. 3.18 and 4.3) and it is not
possible to readily show the
effect of a change in k
*
(catalyst
reactivity) on the outlet molecular weight and
water concentration.
This problem has been eliminated
by plotting the results as
a function
of the Thiele modulus and 2
/Da in Figureg 4 _ 5 and 4 _ g
<,
2 t
L_ = H /D .
diffusio n
Da L/<v > t (4.8)
z residence
2
The quantity * /Da is a Peclet number (see e.g., Bird, Stewart and
Lightfoot; 1960) and is independent of the reaction time. From
Figures 4-5a and 4-5b we see that at constant 2
<> /Da (film height and
residence time), increasing the rate constant via a more active
catalyst (increasing the Thiele modulus) increases the outlet
molecular weight; this effect is largest at low Thiele moduli and
low
2
/Da. Beyond a Thiele modulus of 2.6 almost no benefits of
increasing catalyst activity are realized. That is, for any value of
2
4> /Da, increasing the catalyst activity increases the outlet molecular
2
weight up to <f>
= 2.6/ beyond this value, no further increase in
molecular weight is possible by increasing the catalyst activity and
the reaction proceeds under diffusion control. In Figures 4-6a and 4-
6b we see that the outlet water concentration hardly changes with

101
5.0
°- 2 1.0 1.8 2.6 3.4 4.2 5.0
<t>
2
/Da
5.0
0.2 Th ; e e
I
Modulus
0.2
5.0
<t>
2
/Da
Figure 4-5. Number average molecular weight for the effluent of the
2
profile vs. the Thiele modulus and f /Da (Contour
Interval - 1000 g/mol)
Figure 4-6. Average water concentration for the effluent of the
profile vs. the Thiele modulus and $ 2 /Da (Contour
. Interval = 0.075)
103
increasing catalyst activity

(Thiele modulus, at constant
fUm height
and residence time
(^/Da) This is because even if the
.
reaction
proceeds to completion with
no mass transfer (i.e., if lt „ ere
irreversible), the concentration of
water would be 1.06 vs. the value
of 1.0 at the inlet. Thus, the water formed via reaction is small
compared to the total water in the
system. Therefore, changes in the
reaction rate through catalyst reactivity cause only very small
changes in the water concentration.
The predictions of the nylon

H
6,6 flowing film model with a
linear
velocity profile for the outlet number average
molecular weight are
shown in Figure 4-7a. This should be compared with
the corresponding
predictions in Figure 4-3a for the parabolic velocity
profile. The
molecular weight predictions for both velocity profiles are very
similar for Damkohler numbers of less than
1.0 for the entire range of
Thiele moduli. However, the model with a linear velocity profile

predicts a slightly lower outlet molecular weight (less
than 4%) than
for the parabolic velocity profile at low Thiele
moduli (0.2-1.0) and
Damkohler numbers greater than 2.0. At high Thiele moduli (4.0-5.0)
and Damkohler numbers greater than 2.0 the linear velocity

results in
a slightly greater outlet number average molecular weight (less than
5%) than the parabolic velocity profile.
The outlet water concentration for the linear velocity profile are
shown in Figure 4-7b and should be compared to Figure 4-4a. The water
concentration predictions of the flowing film model with a linear
velocity profile are somewhat lower (up to 5% of the initial water
concentration) than those for the parabolic velocity profile over the
entire range of Thiele moduli and Damkohler numbers investigated.

104
1
1
I t *]
h £_
1-28 2.52 3.76 5.
Damkohler Number
0.04 1.28 2.52 3.76 5.00
Damkohler Number
Figure 4-7. Number average molecular weight (a) and average water
concentration (b) for the effluent of the nylon 6,6
flowing film model with a linear velocity profile vs.
the Thiele modulus and Damkohler number a) Contour
(
Interval - 500 g/mol, b) Contour Interval = 0.075).

105
This is consistent with the greater outlet Secular weight

predictions for the mo del with a linear velocity profile at high
Thiele moduli ,4.0-5.0) and
Damkohler numbers greater than
2.0 since
the polyamidation is
reversible.
in order to explain why

the linear velocity profile
results in a
slightly lower outlet number average molecular weight than the
parabolic velocity profile at low Thiele moduli (0.2-1.0) and
Damkohler numbers greater than 2 . 0, one must remember that in this
region the film is under kinetic control. The highest molecular
weight material is found at the wall-film
interface (e.g., Figure 4-
2a) .
In addition, the molecular weight increase
is not linear with
conversion (p) but goes as 1/1-p. Therefore, although both films have
identical average residence times, the
parabolic velocity has more
long-residence-time material (with a higher conversion and molecular
weight), resulting in a slightly higher cup-mixed average molecular

weight even though it contains slightly more
water.
The predictions of the nylon 6,6 flowing film

model with a plug
flow velocity profile for the outlet number average molecular weight
are shown in Figure 4-8a. This should be compared with the

corresponding parabolic and linear velocity profile predictions in
Figures 4-3a and 4-7a. For films thicker than 1 cm (Thiele modulus
greater than 3.25) the molecular weight generation is reduced (up to
20%) for the entire range of Thiele moduli for the plug flow profile
compared with either the parabolic or linear profiles. However, for
film thicknesses less than 0.6 cm (Thiele moduli less than 1.0) the
molecular weight generation is increased (up to 10%) compared with
either the parabolic or linear profiles.

106
3.0
0.04 1.28 2.52 3? 6 5.00
Damkohler Number
5.0
Damkohler Number
Figure 4-8. Number average molecular weight (a) and average

water
concentration (4-8b) for the effluent of the nylon
6 6
flowing film model with a plug flow velocity
profile vs
the Thiele modulus and Damkohler number a) Contour
(
Interval - 500 g/mol, b) Contour Interval = 0.075).

107
The predictions of the nylon

6,6 flowing film model with a plug
flow velocity profile for the outlet
water concentration are shown in
Figure 4-8b and should be compared
with the corresponding parabolic
and linear velocity profile
predictions in Figures 4-4a and 4-7b. The
mass transfer is considerably slower
for the plug flow profile (up to
15% of the initial water concentration)
than in either the parabolic
or linear velocity profiles for the
entire range of Thiele moduli and
Damkohler numbers. These results indicate that mass
transfer becomes
much more efficient as one goes from a plug flow velocity profile to a
parabolic profile and only slightly more efficient as
one goes from a
parabolic profile to a linear profile. The reduced molecular weight

production for the plug flow profile for thicker films
(Thiele moduli
greater than 3.25) can be explained from the reduced
mass transfer and
the fact that the main amidation reaction is reversible. For thin
films (Thiele moduli less than 1.0) the polyamidation proceeds under
kinetic control and the residence time distribution determines which
profile has the largest molecular weight. The plug flow profile
contains less low residence time material near the gas-film interface
which results in a higher cup-mixed average molecular weight than
either the linear or parabolic flow profiles.
The predictions of this nylon 6,6 kinetic model were compared to
those of an ideal model based on constant rate and equilibrium
constants in the kinetic description. A value of the forward reaction
rate constant of 117/hr and an equilibrium constant of 470 were used
(the rate constant is the mean between the initial value and the long
time (equilibrium) value and the equilibrium constant (470) is the
long time value) . The outlet molecular weight for the ideal model
108
0.04 1.28 2.52 3.76 5.00
Damkohler Number
Figure 4-9. Number average molecular weight (a) and average water
concentration (b) for the effluent of the ideal
flowing film model with a parabolic velocity profile
vs. the Thiele Modulus and Damkohler Number ( a)
Contour Interval = 500 g/mol, b) Contour Interval =
0.075) .
109
with a parabolic ve locity

profile as a £unction
^^^^
of the Thiele modulus
an d the Oaêr nuBb . t ia sh ow„ in
Flgure
compared with Figure
4-3a.
At Damkohler numbers
below
Deiow 11.5r fK
thea ideal ,
model predicts molecular
weights that are too
low at *n all Thiele moduli in i
the investigated
range. For example, at a
Damkohler number of 1.3
and a Thiele modulus
of
2.5 the ideal -del predicts a molecular
weight of about 5,000
g/mol and the nylon
6,6 model predicts about 6,000 g/mol. At
Damkohler numbers greater than 3 and Thiele moduli less
than 2 the
ideal model predicts molecular
weights that are too high.
At a Thiele
modulus of 0.2 and a Damkohler
nuêr of 5.0 the ideal model predicts
an outlet molecular weight of
about 13,600 g/mol and the
nylon 6,6
model predicts about 12,000
g/mol. The fact that the ideal model
predicts a molecular weight that is
too low at low Damkohler numbers
and too high at high Damkohler
numbers is to be expected. The
apparent reaction rate constant for nylon
6,6 is known to decrease
with decreasing water content. Thus, for long reaction times (large
Damkohler numbers) the average value used in
the ideal model is too
large, leading to the prediction of too fast a buildup of the
molecular weight. For short reaction times the mean value of the
reaction rate constant is too small (and hence the reaction rate),
thereby predicting the molecular weight evolution to be

too slow.
For thin films at short reaction times, the errors in the rate
constant of the ideal model cause only slightly different molecular

weight predictions. These errors become more significant with
increasing film thickness (Thiele modulus) because the mass transfer
gets progressively slower (see Figure 4-4) resulting in the ideal

110
model having progressively

j larger
yer errors in fh*
the equilibrium constant
or reverse reaction
rate. mese errors also increase as
These i
the residence
time (Damfcohler nuêr,
<«e eq. 3.14,

increases because an
is integrated over
a longer time.
^
The predictions of
the idealized model had
their largest differences
from the nylon 6 6
model when the flowing fiim
is operating under
reaction control (i.e.
for thin films (low Thiele
neie moduli)
moduli* and long residence ^
times
i
(high
Damkohler numbers)). Generally,
the ideal model is most
accurate for
thin films with short residence
times (low Thiele moduli and low
Damkohler numbers) and least accurate for
thick films and long
residence times (low Thiele moduli
and high Damkohler numbers)
The following examples give a
feel for the magnitude of the
errors
incurred using the idealized
kinetic model. Suppose we had an
existing film device (i.e., a fixed L and W) which was operated at a
constant residence time (i.e., Damkohler
number). For a Damkohler
number of 2.5 and an outlet molecular
weight of 7,500 g/mol, the nylon
6,6 model predicts a Thiele modulus of 3.46 is required whereas the

idealized model predicts one of 2.32, corresponding to film
thicknesses of 1.1 cm and 0.8 cm respectively.
If instead, we consider operation at a constant

film height (i.e.,
Thiele modulus) the two models can be used to predict the required
residence time. For a Thiele modulus of 2.5 and an outlet molecular
weight of 7,500 g/mol, the nylon 6,6 model predicts a Damkohler number
of 2.14 is required and the idealized model predicts one of 2.7
corresponding to residence times of 44 min and 55 min respectively.
For a Thiele modulus of 1.0 and an outlet molecular weight of 11,000
g/mol, the nylon 6,6 model predicts a Damkohler number of 4.23 is

required and the idealized model
predicts one of 3.57, corresponding
to residence times of 87 min
and 73 min respectively. The important
point to realize is that a device designed
from the idealized model
would not have the correct
outlet molecular weight. For a set of
operating conditions calculated from the idealized model, the nylon
6,6 product would actually have molecular weights of 8,200 g/mol,

8,500 g/mol and 10,250 g/mol instead of the originally stipulated
7,500 g/mol, 7,500 g/mol and 11,000
g/mol.
The water concentration predictions of
the idealized model with a
parabolic velocity profile are shown in
Figure 4-9b and those of the
nylon 6,6 kinetic model in Figure 4-4. The predictions of the
idealized kinetic model for water concentration are in better
agreement with the nylon 6, 6 kinetic model than the predictions of
molecular weight. In fact, except for small discrepancies at high

Damkohler numbers (>2) the two plots are essentially identical. This
is because the water formed via reaction is small compared to the
total water in the system. Therefore, errors in the kinetic model
affect the water concentration predictions much less than the
molecular weight predictions.
4 . 4 A Design Example
Suppose we wish to design a flasher for a throughput (Q) of 1
3
m /hr with an outlet molecular weight of 7,500 g/mol. Using Figure 4-
3 we can determine the combinations of Thiele moduli and Damkohler
numbers which give the desired molecular weight. The Damkohler number
must be in the range 1.3 < Da < 3.28 and the Thiele modulus in the
112
range 0.2 M 2
s 5 .0. Thus, a wide range o£ designs (whose steady
state operation correspond to different Thiele moduli
and Cashier
numbers, can produce the desired molecular weight. However, each
operating point has a different outlet water
concentration (see Figure
4-4) .
in order to determine the film area we begin by choosing a

Damkohler number from the
appropriate range ,1.3 < Da <
3.28,. The
DamRohler number fixes the
Thiele modulus (see Figure 4-3)
for an
outlet molecular weight of 7,500 g/mol and the Thiele modulus
specifies the film height for
nylon 6,6 at 250 °c (see eg.
3.18). The
area (A) may now be calculated
directly from the volumetric flowrate,
film height, Damkohler number and characteristic
reaction rate
constant
A - t w Da Q
1
(4.9)
k H
app
To derive eq. (4.9) one starts with the definition of the Damkohler
number (eq. 4.8) substitutes Q/WH for < Vy > (see eq. 4.7) and
rearranges for the area (LW) Thus, the required area increases
.
with
both the Damkohler number (residence time) and throughput and
decreases with the film height (Thiele modulus) . For a tubular
geometry, the area in eq. (4.9) is related to (jidL) where d is the
diameter.
The outlet water concentration is plotted against the required
area for outlet molecular weights of 4,000, 5,000 and 7,500 g/mol in
Figure 4-10. The water concentrations are dimensionless so a value of

113
0.5
4,000 g/mol
0.45
c 0.4
o
Amin - 35
CO
0.35
c
o 0.3
c
o
o 0.25 5,000 g/mol
00
7,500 g/mol
0.2
Amin = 83
0) 0.15
o 0.1
0.05
0 —
'
1 i i i
i
1 1 1 1 1 1 1 1 1 1
10 30 50 70 90 110 1 30 1 50 1 70
Area (mxm)
Figure 4-10 Outlet water concentration vs. required film area for
the nylon 6,6 flowing film model with a parabolic
velocity profile with various outlet molecular weights
3
(Q = 1 m /hr)
114
0.1 means that the outlet

water concentration is
10% of its initial
value. The required area
has a mini™ for each
of the three outlet
molecular weiaht-
"ghts. tv^
The range of outlet
<?
water concentrations is
constrained by the inlet
concentration, the reaction
eguilibrium, the
mass transfer rate and the
reaction rate. Therefore,
the three outlet
molecular weights can not
always be generated at
the same outlet water
the same outlet mo lecular weight and two different water

concentrations is not an indication
of multiple steady-state
operating
points since the film thickness
is different in each case.
Figure 4-10 also shows that the
outlet molecular weight and water
concentration are not completely
independent. For example, for an
outlet molecular weight of 4,000
g/mol the water concentration
can not
fall below 0.36 whereas for an
outlet molecular weight of 5,000
g/mol
the outlet water concentration can
be below 0.1. Beyond an outlet
molecular weight of 5,000 the
required area increases sharply,
especially for low outlet water concentrations.
The reason for this
can be seen in Figures 4-3 and 4-4. We can consider that increasing
the Damkohler number is eguivalent to

increasing the length (and thus
the area) at a fixed average velocity. For small Thiele moduli (less
than 1.0) and Damkohler numbers greater than 0.6, the water
concentration changes very little with increasing length, but the
molecular weight changes steadily. Therefore, devices with outlet

molecular weights of much greater than 5,000 g/mol (those with high
Damkohler numbers) have large portions dedicated to molecular weight
generation where very little mass transfer is occurring.

115
Although the area has been determined, the aspect ratio (L/W) has
not. Once an aspect ratio is chosen the length and width can be found
from the area, and the film velocity is determined from eq. (4.3). in
Figure 4-11 the average film velocity and width are shown as functions
of the aspect ratio for an outlet molecular weight of 7,500 g/mol and
2
an area of 83 m . It is clear that the greater the aspect ratio (L/W)
the greater the average film velocity must be. Certainly there is an
upper limit on the aspect ratio for which the required film velocity
can not be attained. The value of this critical aspect ratio will
depend on how the flow is generated.
4 . 5 Conclusions
The flowing film model is useful for the design and control of
continuous, thin-film polymerizers . The addition of a flow field to
the stationary film reaction-diffusion problem has been shown to move
the region of highest molecular weight from the gas-film interface
toward the wall-film interface for thin films. An idealized kinetic
model leads to significant errors for molecular weight predictions
although it is sufficient for predicting the outlet water
concentration. Linear and parabolic velocity profiles have been shown
to yield similar predictions for the outlet average water
concentration and molecular weight whereas the plug flow profile gives
notably different predictions. Operating a flowing film device as a
molecular weight building device is inefficient because very large
sizes are required. Most of the water is removed at the beginning and
in the remaining portion very little additional water is removed

116
0 2 4 6 8 10
Length to Width Ratio (L/W)
Figure 4-11. Average film velocity and width vs. aspect ratio for the
nylon 6,6 flowing film model (A = 83 m M = 7,500
,
3
g/mol, Q - 1 m/hr)
117
because the driving force fnr

lorce f
y ror rt-i
diffusion •
i
is low and molecular weight
generation is reaction controlled.
CHAPTER 5
DYNAMIC BEHAVIOR OF THE

FLASHER AT HIGH TEMPERATURE
5.1 Introduction
A comprehensive kinetic model for nylon 6, 6 polymerization that

accounts for side reactions that are
significant at high temperatures
and/or long residence times has been
developed (see Chapter 2) . This
degradation description will be incorporated
into the dynamic model
for the flasher (see Chapter 4) . A velocity profile across the film
gives rise to a distribution of residence

times. Therefore, there is
some material near the wall with residence
times long enough to result
in significant degradation at the higher temperatures (270-300 °C)
typically encountered in the continuous production of nylon

6,6 (U. S.
Patent; 3,900,450). This thermal degradation can have a major impact

on the guality of nylon 6,6 polymer, even though the concentration
of
degraded ends in the typical industrial product is very low and can
only be measured accurately with very sensitive technigues.
In this chapter, we will evaluate extent of degradation for a
range of operating conditions in the flasher along with the relative
time scales of the number average molecular weight to come to steady-
state as compared with that of the degradation products. The effects

of reaction temperature and residence time cn the amount c£
degradation are also studied in order to determine operating
conditions that will minimize
degradation.
5 ' 2 ^corporation of the Degrad ation

Scheme
We have incorporated this

degradation model into our flowing
fUm
polymerizer model. The flowing £il m geometry
is shown in Figure 4-1.
The degradation scheme is
summarized in Table 2-3 along with
the main
amidation reaction. The reaction rates (R^
of these reactions
% )
are given by eqs. (2.21 - 2.25)

and the kinetic constants in Table 2-
4. The evolution equations for the

for the non-volatile components
and water, in dimensionless form, are
-2r 4 -r 5 _A
at"* (R
2
)
- Vy (51)
dC 2 dc
C 2 a
c
V
,
R V
at""* <" i
" "
fc) yaT (5.2)
*T = * ( " R
2
+
V " V
yaT (5.3)
9 p 2 a r a
W
-J? =
at
a
2
/r
(R
1
+ d
+
V v _
fe }
w
V
y W +
3
9Z
,
(
rÊ
C dx
dZ
.
> (5.4)
—
9C
SE 2
(R
1
+ R
2
-R 3 )
"
(fe)
a
2
v
y ^
aC <;r
(5.5)
^
dt
= *
2
(R,
3
- RJ
4'
- (i!)
v
Da'
V.
y dY
^5 (5<6)
120
^
H - *
2
R. .
4 W
(|f> V
v
y
!f§B
dY~ (5.7)
Only the water balance

contains a diffusion term since it is
the only
are part of the large

poller molecules which we have
assumed do not
diffuse. The boundary
condition for the water balance
are that the
concentration of water is
constant at the gas-film
interface and the
flux of water is zero
(»«/„ . 0 at the solid wall-fil m ,
interface.
All of the approximations
made in deriving (5.1-5.7) and the
characteristic scales for
nondimensionalizing them have been
discussed
previously in Chapters 3 and 4. 2
The Thiele modulus is defined by
(, )
eq. (3.18) and typical values given in Table 3-1. The Damkohler
number (Da) is defined in eg. (4.3) and typical values given in Table
4-1.
We have estimated the total

concentration of polymer molecules
(C D )
by
+ + C + C C
-
c
L - °A °C SE SB X
P 2 '
(5.8)
in these modeling studies. This expression is approximate, although
it becomes exact if there is at most one crosslink in any polymer

molecule. Therefore, it is valid for small amounts of crosslinking.
The balance equations have been integrated by the method of lines
(Press et al. ; 1986)

we have evaluated
extent of degradation
for some typical operating
condrtions in order to
industrial produot
estimate the amount of
In addition
scales of the nuêr average
as compared with that

.
have
degraded material in the
molecular weight to come to
of the degradation
steady-state
, „ ^ ^^
products for a reasonable
change in reactor throughput
to demonstrate why
certain product
properties may continue to
change even after the
number average
molecular weight has reached
steady-state. The complex interplay of
the effects of reaction
temperature and residence time
on the amount
of degradation are also
studied
BU in ororrlpr
aer to determine operating
conditions which will minimize
degradation.
From Table 3-1 we can see that typical values of the Thiele
modulus range from 0 to 10 which correspond to film thickness
.
1 . 0
of
0.2 to 1.7 cm whereas from Table 4-1 we see
that typical values of the
the Damkohler number range from 0 . 05 to 5 corresponding to residence
times from 1 to 100 min. We have evaluated the effect of changing

the
Damkohler number from 2 . 5 to 3 . 0 at a constant Thiele modulus of 1 .
(film thickness of 0.5 cm) and reactor temperature

of 270 °C. This
corresponds to increasing the mean residence time in the reactor
from
50 to 60 min, an increase of 20%. The evolution of the cup-mixed

average of different species as a function of intervals of the mean
residence time is shown in Figure 5-1. Figure 5-la shows that the
cup-mixed number average molecular weight of the effluent increases
from 10,800 to 11,440 g/mol, the concentration of amine ends decreases
from 9.7 to 9.1 mmol/113g and the concentration of carboxylic acid

122
o
E
N 12000
MOLECULAR WEIGHT
rM o n
AMINE ENDS
.c - - CARBOXYL ENDS
1 1 600 10.5 O
0 E
E
L
1 0 0
o " 200.
3
U
o
o H 95 L
10800 c
V
0
- 9.0 c
o 0
L o
1 0400. -
> a
< 8 5 D
o
L L
© o
D
E
3
'0000
°0 0.8 1.8 2 4
——
1 I
3.2 4 0
8.0 D
u
2
t/t
re3
Figure 5-1. The time evolution of the effluent average a) number

average molecular weight, amine and carboxyl end-group
concentrations and b) water, stabilized end-group,
Schiff base end-groups and crosslink concentrations for
.the flowing film model as the Damkohler number is
changed from 2.5 to 3.0 at a Thiele modulus of 1.0 and
270 °C.
123
ends decreases fro, 9.3 to 8.7 mol/113g

,
t^t water concentration

decreases fro, 25.1 to 24.3
„ol/ll3g, the
stabilized end concentration
increases fro, 0.53 to 0.64
„ol/113g,
the crosslink concentration
increases fro, 0.6 to 0.7
mmol/HSg and
the Schiff base end
concentration (which has been
multiplied by 1.5
for visibility in the Figure)
increases from 0.4 to 0.48 mmol/113g.
in Figure 5-2 we show the effect of
increasing the reactor
temperature to 280 °C for the same
case as in Figure 5-1. Figure 5-2a
shows that the cup-mixed
number average molecular weight
of the
effluent increases from 11,300 to
11,900 g/mol, the concentration of
amine ends decreases from 9.2 to
8.7 mmol/113g and the concentration
of carboxylic acid ends decreases

from 8.9 to 8.3 mmol/113g. In
Figure 5-2b we can see that water concentration
decreases from 25.9 to
24.8 mmol/113g, the stabilized-end concentration
increases fro, 1.25
to 1.5 mmol/113g, the crosslink concentration increases
from 1.15 to
1.3 mmol/113g and the Schiff base-end concentration
(which has been
multiplied by five in the Figure) increases from 0.19 to 0.21
mmol/113g.
Comparing Figures 5-1 and 5-2 reveals that the steady state
concentration of stabilized ends and crosslinks approximately doubles
with the increase in temperature while the concentration of Schiff
base ends decreases by a factor of two. In addition, the molecular
weight, amine-end, carboxyl-end and water concentration come to steady
state in about 1.2 residence times (1.2 hours) at both temperatures.
However, the system is not at steady state. The stabilized end,
Schiff base end and crosslink concentrations continue to increase
slowly, for much longer times. Therefore, the molecular weight and
124
a
11.0 n
N
-10.5 0
E
UOLECULAR WEIGHT
E
AUINE ENDS
- CARBOXYl ENOS 10.0
o
9.5 ^
o *->
t 1 100 C
V
u
©
c
9.0
o 0
L. U
10800.-
> a
< 8.5 D
0
L L
V o
D
E
10500.-
0.0
I'll
08 1
«
i
2 4
i
—— i
3.3 4.0
8.0 "D
u
c
z
t/t
res
2.00-r-
CP -p27.0
n WATER
STABILIZED ENDS
N 1.70-
SCHIFF BASE ENDS . 5
CROSSLINKS
o
E r—
40-
-26 0 s
1
c 0
o E
E
o
L
1.10-
c c
o 0
u -25 0 +-I
c 0
o t
u 0 80- c
L (J
1)
c
o 0
U
5 0 50--
24 0
res

the flowing film model as the Damkohler number is
changed from 2.5 to 3.0 at a Thiele modulus of 1.0 and
280 °C.
125
water concentration
eventration
scales.
o£ degraded
„ay appear
products
t0 be at steady state
changes over
,
^
^^^
^
This is because the slow m0 ving material with i onger
residence ti m es near the
statuary reactor wall has not reached
steady state and contributes
to changing the
«u amounts of degraded
product but does not affect
the Secular weight. Therefore, the
product guality înability, dyeability, may continue to change
although the nor^lly me asured
product properties Molecular weight,
end-group balance) would
suggest that the product was
not changing.
5.4 Dynamic Predictions
It is useful to try to counterbalance the

increase of molecular
weight resulting from the increased
mean residence time by decreasing
the reaction temperature. Figure 5-3a shows the effect of
decreasing
the reaction temperature from 270
to 262 °C for the same increase in
residence time as in Figure 5-la. The molecular weight goes through a
minimum after one-half a residence time and then regains its initial
value. This minimum occurs because the material in the reactor
undergoes an instantaneous decrease in temperature and concomitant

instantaneous decrease in amidation rate whereas the increase in the
mean residence time is not instantly realized by any of

the material
already in the reactor at the time the average velocity (mean
residence time) is reduced since it has already traversed part of the
reactor at a faster velocity (shorter residence time) . After slightly
more than one residence time this effect has disappeared and the
molecular weight regains its initial value. Similar behavior is

126
o
E
X 10900
CP 1- 11. n
MOLECULAR WEICHT
AUINE ENDS
r CARBOXYL ENDS X
10800 o
10.8
E
E
L.
0 10700.-
3 1 0.2
U
V
o
10600.- c
- 9.8 4)
u
0)
c
C7>
o
O
i- u
9 10500.
> - 9.4 a
< 3
O
L L.
QJ o
0400
E
1
^ I '
I
1
1 1
| 1 9.0 X)
00 c
10 2.0 3.0 4 0 5 Ld
z
res
1 00
t-28.0
CP
X n
o
E 0 80-
WATER 27.0 _
E O
STABILIZED ENDS
E
SCHIFF BASE ENDS • 1.5
c E
0 CROSSLINKS
0 60
*-/
26 0
o
c
0)
u 0.40 ~
25.0
c c
0 V
u u
c
0
a 0.20 24.0 O
O
— — ——
I
0 00
^ i
I i |
-23.0 ^
C I
" I
u 0.0 1.0 2.0 3.0 4.0 5,0
t/t
res

changed from 2.5 to 3.0 and the temperature reduced from
270 to 262 °C at a Thiele modulus of 1.0.
displayed by the amine ends.
to the longer residence

The carboxyl ends decrease
somewhat ^
time and the relatively
low activate energy
of 3.1.
The evolution of water

and degradation produot
evolution for the
same s im ultaneous increase
in residence time and
decrease in operating
temperature are shown
own in £pimira
igU re 5-3b
q u- u u
which should be compared with the
isothermal case in figure
Fiaurp 5s-ih
lb. tk. ~
The outlet water concentration shows
similar overall behavior,
decreasing in both cases.
However, the
higher operating temperature
yields a slightly higher outlet
water
concentration because water is a
product of both the main amidation
(3.5) and stabilized end-group

formation (3.1) reactions which are
accelerated as the temperature is
increased. The concentration of
Schiff base end-groups also shows
similar behavior, increasing in both
cases. However, the Schiff base concentration
is higher (about 20%)
at the lower temperature. However, the concentration of stabilized
end-groups and crosslinks show very different

behavior under the two
operating conditions. In the isothermal case (Figure 5-lb) they both
increase significantly (by about 15%) whereas they
both decrease (by
about 35%) at the lower temperature (Figure 5-3b) . Therefore it may

be possible to adopt a response policy to maintain the Schiff base
ends at some constant level but the molecular weight and end-group
balance would change. Therefore, it is important to decide which is
the most important for maintaining the desired product properties.
The effect of decreasing the reactor temperature from 280 to 268
°C as the mean residence time is increased is shown in Figure 5-4
which should be compared to the isothermal case in Figure 5-2. Very
similar conclusions may be drawn. The molecular weight and amine end
128
C7»
a rn
2.00-
26.5
WATER
N STABILIZED ENOS n
-
o ~ ~ SCHIPF BASE ENDS . 1,5
E 1.80-
E
CROSSLINKS
-26 .0 \
o
r
_ b
c
0 E
1.20-
+->
-25 5
D r
L
+-i 0
C
V n
0 0.80- L
c -25 0
0 C
u 1)
0
c
a 0.40- 0
D -24 o
5
O
L
O D
I
T?
C
UJ
0.00
00 1
2
1
I
2.4
— I
1
3.8
1
— |
r -24, 0
0
4.8 6 0
t/t
res
o
E
S 1 1600
10.5 n
CP
U01ECULAR WEIGHT
AUINE ENDS
CARBOXYL ENDS
N
en 1 1 400 -10.2 o
E
V
$ E
L 9 8
o c
1 1 200
0
3
u o
V - 9.5 L
o c
1 1000.- V
u
© 9.2
c
CP o
o u
L
V 1 0800 Q.
>
8 8
< o
L o
0)
ID
E
10600 1
0.0
— 12 2.4
1 1 '
1
r 8 5 V
c
3 6 4 8 6 0 UJ
z
res
Figure 5-4 The time evolution of the effluent average

a) number
average molecular weight, amine and carboxyl
end-group
Schiff base end-groups and crosslink concentrations
for
changed from 2.5 to 3.0 and the temperature reduced from
280 to 268 °C at a Thiele modulus of 1.0.
129
concentration regain their

initial values after one
residence time and
the carboxyl end concentration
decrease to a slightly l ower
value
The water and degradation
products all change significantly
fro, their
initial values with the
water and Schiff b ase
end-groups showing
qualitatively similar behavicr
under the two operating
conditions.
The stabilized end-group and crosslink concentrations show
qualitatively different
evolution, decreasing at the
lower operating
temperature and increasing in
the isothermal case.
5.5 Discussion and Conclusions
Figures 5-1 through 5-4 illustrate

two important points. Even
though the flowing film reactor may be operated under
two different
operating conditions to produce a product with essentially identical
molecular weights, the final products can contain very different
amounts of degraded products and may therefore
display significantly
different end-use properties in some applications. Therefore, the
optimal operating policy will depend on the desired final product
characteristics (molecular weight, end-group balance, etc.).
Secondly, this model is suitable for control studies. It may be used
to determine how the operating temperature and pressure (the pressure
in the gas phase affects the value of the constant water concentration
at the gas-film interface) can be manipulated to counteract changes in
throughput
We have shown that the flowing film model predicts that the outlet
average molecular weight and water concentration come to steady-state
in about one mean residence time after a change in flasher operating

130
conditions whereas the

degradation products oontinue
to ohange slowly
even after four residenoe
times. I» addition, „e have shown
that it
i. Possible to choose
the operating conditions
of the flasher so the
s*ne molecular weight
is produced at two
different temperatures but
with very different amounts
of degradation products
giving rise to two
products of significantly
different quality.
CHAPTER 6
A WIPED FILM MODEL

FOR NYLON 6,6 POLYMERIZATION
6 .
1 Introduction
There are several previous studies of

wiped film polymerization
reactors. For example, Ault and Mellichamp (1972) developed a
"periodically mixed" film model in which they assumed that a
stationary polymer film was laid down,
subsequent polymerization and
mass transfer occurred for a certain exposure time, after which the
film was instantaneously well-mixed
and the process repeated. They
used a simplified kinetic model that included no side reactions or
degradation reactions and assumed that the condensate concentration

in the gas phase was zero.
Another model for wiped film polymerization reactors was

developed by Amon and Denson (1980). They assumed that only a
fraction of material was laid down in the well-mixed film while the
majority remained in a bulk pool adjacent to the moving wiper blade.
The reaction was assumed to occur only in the bulk and condensate
removal only in the film. They approximated the film thickness as
infinite as far as mass transfer was concerned. Consequently, they
did not investigate the effect of film thickness. In addition, they

132
also used a simplified
degradation reactions.
kinetic model which did
not include any ^^
Gupta, Kuâr and Ghosh
,1963, .edified the model o£ Amon and
Denaon to include
-del o, Ault and Mellichamp

the effect of film thickness
. They CQmpared ^
to that of Amon and
Denson and found the
two „odels gave nearly
identical predictions.
More recently, Kuâr et
al. (1984) studied the
finishing stage of polyethylene
terephthalate
(PET) polymerization with this modified
Amon and Denson model and
a
kinetic scheme that included
reactions with monofunctional
compounds
as well as redistribution and cyclization
reactions. The finishing
stages of PET polymerization
has also been studied by
Ravindranath and
Mashelkar ,1984, with the mixing film model of Ault
and Mellichamp.
However, they included degradation
reactions in their kinetic scheme
as well as an "effective flashing"
technique to account for a changing
interfacial concentration.
Recently, realistic kinetic and equilibrium correlations

as well
as a degradation scheme have become available for nylon 6,6 (see
Chapter 2). PET and nylon 6,6 have very different equilibrium
constants (about unity vs. 250, respectively) and degradation
reactions. It is interesting to compare the results of Kumar
et al.
(1984) and Ravindranath and Mashelkar (1984) for PET to those of

nylon
6,6. We will use a mixing film model similar to Ault and Mellichamp.
However, we will assume a plug flow velocity profile in the film which
enables us to relate a reaction time to a residence time (and
ultimately a reactor size) . We will not employ the effective flashing
technique of Ravindranath and Malshelkar primarily because it requires
knowledge of the compositional variation of the Flory (1953) chi

133
parameter or the assertion that it is constant. There is st rong

evidence that the latter
i3 unrealistic in these polar condensati on
systems (Costa; 1987)
6 ' 2 ^Ped Film Polyrn^^, React or Model
The
main amidation reaction
and degradation model are
summarized
in Table 2-3. These reactions are
abbreviated as
C -> SE + W
(6.1)
Degradation L -+ SE + A
(6.2)
SE -» C02 + SB
(6.3)
SB + 2A -» X + 2NH
3 (6.4)
Polyamidation A + c -> L + w
(6.5)
where SE refers to a cyclic or stabilized

end-group, C02 refers to
carbon dioxide, SB to a Schiff base and X to a crosslink. The
reaction rates of the degradation model
(6.1-6.5) are given by (2.21-
2.25). The kinetic parameters for reactions 6.1-6.4 are shown in
Table 2-4. The apparent rate and equilibrium constants in
eq. (6.5)
are complicated but known functions of temperature and composition
(see eqs. 2. 9-2.14)
A schematic of the wiped film reactor is shown in Figure 6-la
along with the an idealized film model in 6-lb. In practice the
rotating blades are pitched slightly in the axial direction to force

134
Polymer Film
Rotating
Blades
Gas-Film Interface
(Z=H)
Direction of
Mixing Blade
Outlet
Inlet
Figure 6-1. Wiped film polymerization reactor schematic (a) and

wiped film polymerizer model (b).
135
the pol^r down through the device Bhile providing

radiai
water diffuses out of
the fil» toward the
ga3 . fUm ^^ z
direction) and reacting
palymei flows thrQugh
^ ^ ^ ^
direction. The water balance
for the flowing f ilm
model is ^ eq
2 *
2 ac f,
at 1
i - x az ) +
* W
*w (6.6)
Y,t Da y aY
Y,z, t z,t
The assumptions made in deriving eq. (6.6) have been discussed

previously (see Chapters and
3 4) and the dimensionless groups
defined
(see eqs. 3.18 and 4.3)
Since we are assuming a plug flow velocity profile the reaction

time of a material element and its
position are not independent and it
is simple to show that
t = —Da
9
Y 0 < Y < 1 (6 . 7)
It can be seen from eq. (6.6) that C = C (Y,Z,t). The chain rule
w w
yields
dC dC
= w w w
dc dY + dz + dt
w ay az (6.8)
Z,t Y,t Y/Z
136
Subsequent substitut i0 -
" —
£° r dYb
-J. uxnerent
^ differentiating eq. (6.7) , taking
the partial derivative
Ve with
wit"h r£>or^„4-
respect l_ _
t0 ime at constant z and
substituting eq. (6.6) î?

for yields
w
at 1
- ]
1 x Jl (6.9)
f t Y,Z,t
Equation ,6.9, is identical to the stationary fUa mo del equation

(3.16, Thus, if „ e move with the
.
velocity of the plug flow the
water
evolution is identical to the stationary fi lm oase.
However, at a
given fil ra thickness (Thiele
modulus) any value of til corresponds to
a single residence time or Damkohler number
(see eq. 6 .
7, . This is
important since we are solving
the evolution equations by
the method
of lines (Press, et al.; 1986) and eq.
(6.6) requires integration in
two spatial dimensions and eq.
(6.9) only one.
The evolution equations for the various components, in

dimensionless terms, are
SC
A .2
" (R " 2 R
IT 2 4
" R
5 » (6.10)
SC
C .2
jt - R
i
-
y (6.1D
SC
L .2
R + R
JT " * <*
2 5> (6-12)
137
dx
ii. L ,
c 2
at az 1
} + R + R
az <
i 5 ) (6.13)
—
ac
sE
=
*
2
(R
1
+ R
2
" R
3
)
(6.14)
~
aC
SB = 2
(
R
3
" R
4
)
(6.15)
^ ~*
9t
2
R
4 (6.16)
Da Y
where t = The boundary conditions for
the water concentration
are identical to those used in the

stationary film model. Thus, C is
w
constant at the gas-film interface, and
at the wall (Z=0) we assume
that the flux of water is zero =

(^ ff) or equivalent^ dx/az = 0.
We assume that the film is mixed after a

certain exposure time
and represent the amount of mixing as the number of times
the film is
mixed in one residence time. The residence time (dimensionless) is
given by
= Da —
res (6.17)
The number of times the film is mixed per residence time, N .

, is
mix
given by
138
t
N = -1*1
mix (6 18 >
texp -
w^re t is the exposure time

exp (dimensionless) between consecutive
mixing of the film -
(note, N i
'
mix ~ 1 corresponds to non-mixing
operation) .
m a given wiped film device
the residence time and the
number of mixings per residence
time are not independent, in order to
keep the mean residence time
constant while xncreasing or
decreasing
the amount of mixing requires
changing the axial pitch of the
rotating
blades. It is this pitch which
determines how far the polymer
is
pushed down the device for each
blade rotation (mixing)
Ogata (1960) has shown that Henry's
law is valid at the low water
concentrations in the wiped film reactor.
We will use his correlation
to relate the equilibrium concentration
of water at the gas-film
interface to the steam vapor pressure in
the gas phase.
= 22 r 3 °50/T - 10.09)
C P in
P <
C 10
Sf,int T (6.19)
In this expression temperature is in degrees Kelvin

and pressure in mm
Hg. This differs from the effective flashing technique of
Ravindranath and Mashelkar (1984) because we do not allow the
interfacial concentration to change during the condensation. It is
fixed by the partial pressure of steam in the reactor.

139
6.3 Model Prediction.,
The
Thiele modulus is
proportional to the value of the
reference
reaction rate conatant
(characteristic reaction time,
and also depends
on the system temperature
(through the diffusivity,
We have chosen .
< pp to be 2.926/hr, the value

of the rate constant as
the water and
amine end-group concentrations approach zero at 200 «C, and
„e have
estimated the diffusivity of
water in molten nylon 4
6,6 as 2.5xl0"
2
cm la at 265 °C (Nagasubramanian and Reimschuessel;
1973) . Typical
values of the Thiele modulus
for nylon 6,6 range from 0.1
to 10, which
correspond to film thicknesses
of 0.2 to 1.7 cm (see Table
3-1). 1£
we had chosen 240 °c as our

reference temperature for k° , the Thiele
modulus would range from 0.6-60 for the same

values of H and H 2 /D in
Table 3-1. Furthermore, if we assume that the diffusivity has a
typical activation energy of 10 kcal/mol (e.g. , Newitt and Weale;
1948) the film thicknesses in Table 3-1 would
increase by one third
(i.e., they would range from 0.27 to 7.3 cm for the same range of
Thiele moduli) at 300 °C. The value of the Damkohler number depends
directly on the characteristic reaction time and on the mean
residence
time (L/<v >) The Damkohler number is independent of
y
.
the system
temperature and typical values are given in Table 4-1.
We have assumed that the feed to the wiped film reactor is an
equilibrium polymer-water mixture with a number average degree of
polymerization of 55. The feed composition and operating conditions
of the wiped film reactor were estimated from the patent (U.S. Patent;
140
3,900,450, and the (lowing

filmm0 del results ,see Chapter 4, We
have approximated the
polymer oonoentration by
P
2 (6.20)
This expression is approximate,

since it is only exact provided
there
is at most one crosslink
per polymer molecule.
Therefore, it is
accurate for small amounts of
crosslinking.
The number average molecular
weight predictions of the model at
280 % 300 mm Hg of steam and of unity (non-mixing case) are
shown in Figure 6-2a. The molecular weight is a

strong function of
the film thickness (Thiele modulus) . Higher molecular weights are can
only be achieved with thin films (low
Thiele moduli) in the range of
operating conditions investigated. If is increased to 10 at the
same operating conditions, the effect

of film thickness is diminished
although it remains significant as shown in Figure 6-2b. The
molecular weight production, particularly for thicker films (higher
Thiele moduli), is enhanced considerably. This trend continues as we
increase N to 50 and 300 as shown in Figures 6-2c and

mix 6-2d. As
N becomes very large the effect of film

mix thickness become s
negligible (as it should for such well-stirred systems), and the
molecular weight depends only on the residence time. It is
interesting to note that the point where the molecular weight contours
intersect the abscissa in all these plots remains virtually constant.
Therefore a film thickness of about 0.2 cm (a Thiele modulus of 0.1)

141
i oo
o.» —
m q
7.8 -
01
3
8.7 -
D
O
O 3.6 -
5
4.5 -
«j
V
3.4 -
c
H
2 3-
1 2 -
0. 1
0 10 o ea i
25 1.17*2.42 3 00
Damkohler Number 3 00
Damkohler Number
10.0
89 -
7.8 -
n
3
8.7 -
3
D
O 5.6 -
2
4.5 -
«j
3.4 -
c
h
2 3 -
1 .2 -
° 10 0 « « 1 84 2 42 3.00
0. 1
0 1 0 OOfl 125
Damkohler Number 3 00
Damkohler Number
Figure 6-2. Number average molecular weight predictions for the

nylon 6,6 wiped film reactor model effluent at 280
°C and 300 mm Hg steam vapor pressure; a) N =1
mix '
b) N 10 < c> N 50 <

mix ' mi x -
d »
Vx ' 300 •
142
truly represents the reaction controlled limit for

molecular weight
production of nylon 6,6 at
this temperature.
We may conclude that for very thin films it does not matter
whether the film is mixed
or not since mass transfer
is fast and the
process proceeds under
reaction control. However, as the film
thickness (Thiele modulus) is increased, the diffusion time increases
slowing the mass transfer and the process no longer
proceeds under
reaction control. By mixing the film, we
speed up the mass transfer
and themixing rate is high enough
the process can approach the
rate
limiting, reaction-controlled,
thin film limit. These effects have
been seen previously by both Ault and
Mellichamp (1972) and
Ravindranath and Mashelkar (1984).
Kumar et al. (1984) also found
that increasing the amount of mixing in the reactor increases the
outlet molecular weight. However, they found the molecular weight
increases as the film thickness is increased, which is exactly
opposite of our results. This is a consequence the different model
for the process (they use the model of Amon and Denson
(1980) where
the polymer film and bulk pool are separate) and the operating
conditions they investigated.
We have also investigated the effect of pressure, which

linearly
affects the interfacial concentration through eq. (6.19). The model
predictions for a steam pressure of 100 mm Hg are shown in Figure 6-3a
and should be compared to Figure 6-2c which is at 300 mm Hg and
identical operating conditions. The molecular weight predictions are
similar for short residence times (Damkohler numbers less than 1) .
However, as the residence time is increased, the lower operating
pressure (100 mm Hg) predictions are significantly higher (by about

0 10 0.88 1.25 1.84 2.42 3.00
Damkohler Number

°C and N = 50; a) P = 100 mm Hg, b) P = 760 mm Hg.
mix
144
2-ooo g / mo i> If th6 operating

^
.
pressure is increaaed atmospheric

pressure (see Figure
6-3b, . the molecular weight ^
reduced very significantly
reason for this (pointed

compared tQ ^m^^ ^
out by Ravindranath
and Mashelkar; 1984) is
that the fita is reaching
eguilibrium for the main amidation.
Higher
operating pressures (and higher interfacial
concentrations, reach
equilibrium more quickly and
therefore have reduced molecular
weight
predictions at longer residence
times (Damkohler numbers)
than lower
operating pressures. Clearly, an assumption of an
interfacial
concentration of zero is
unrealistic, particularly for longer
residence times, because this
equilibrium limit can never be reached
and the molecular weight will
climb indefinitely toward infinity
if
degradation reactions are ignored.
Reactor temperature is another

important operating variable. One
has to keep in mind that it not only affects the rate
of the main
amidation, but also the diffusion time through the temperature
dependence of the diffusivity. In addition, the melt viscosity (and
therefore the energy required to mix the film)
and the product quality
through the temperature dependence of the degradation reactions.
Figure 6-4a shows our molecular weight predictions for a reactor
temperature of 270 °C and should be compared to Figure 6-2c which is
at 280 °C and identical operating conditions. The higher temperature

leads to the greater molecular weight over the entire range of
operating conditions. At longer residence times (higher Damkohler
numbers) an increase in molecular weight of about 1,000 g/mol is
typical
Darnkohler Number
0 10 0 08 I
•
25 1
i
84 2
——
42
i
3.00
Domkohler Number

nylon 6,6 wiped film reactor model effluent N = 50 .
mix
and a steam vapor pressure 300 mm Hg; a) 270 °C, b)
290 °C, c) 300 °C.
146
If the reactor temperature is increased to 290 °C

as shown in
Figure 6-4b we can see that
at low Damxohler numbers or residence
times the urease in molecular weight
270 to 280
Thiele
o
moduli)
c .
and longer residence
is small (the upper right

is about the same as
However , the improvemenfc
times (higher Damkohler nuêrs)

going from
thicker
^
hand corners of the Figures
6-2c and 6-4b
are nearly identical).
As the operating temperature
is increased
still further to 300 <C the
molecular weight beg.ns to
decrease for a
large portion of the
operating conditions investigated
as shown in
Figure 6-4c. Furthermore, if one interprets
increasing the Damkohler
number as increasing the
residence time it may be seen that
the
molecular weight would go through a maximum if the reactor were
operated at 300 °C and a Thiele modulus of less than 2.5. This
behavior was observed by Ravindranath
and Mashelkar (1984) . Kumar et
al. (1984) found that the molecular weight always increased
as the
reactor temperature was increased. However, they did find that the
amount of cyclized material increased dramatically.
Molecular weight is not the only important product

property for
industrial quality nylon 6,6. For use in fibers it must also often
have the proper balance of end groups for good dyeability, and very
small amounts of gelled material for fiber spinning. In the absence
of degradation reactions, the number of end-groups at a given
molecular weight can be calculated as in Table 6-1.
In a typical industrial nylon 6,6 molecular weight of 20,000
g/mol there are about 6 mmol/113g of amine and carboxyl end-groups.
The difference in amine and carboxyl end-groups for typical operating
conditions is shown in Figure 6-5a. Even at this temperature (280 °C)

the aide reactions are
present in appreciable
amounts giving rise to a
difference of up to 1.0
mmol/113g. As the temperature is
increased to
300 °C the maximum diffprpnô 4 n ^k^*
airrerence in chain ends
increases to 3 7 Our
calculations have shown that the difference in end-groups
approximately doubles with
every 10 - C increase in
temperature from
270-300°C.
Table 6-1. End-group concentration vs.

molecular weight for nylon
6,6
polymerization without degradation.
< M > Total End-Groups

n Amine or Carboxyl Ends
g/mol mmol/113 g mmol/113g
1000 230.0 115.0

5000 45.4 22.7
10000 22.6 11.3
15000 15.1 7.5
20000 1 5.7
25000 9.0 4.5
The most striking feature of these Figures

(6-5a,b) other than the
sharp increase in the difference in chain ends is that

they are nearly
independent of the rate of mass transfer (e.g., film thickness or
Thiele modulus). Kumar et al. (1984) report a similar result for the
amount of cyclic material found in PET. It is also possible to
understand why the molecular weight begins to level off at 300 °C (see
Figure 6-4c) at longer residence times. The difference in chain-ends
is nearly one-half the total number of chain-ends (see Table 6-1)

148
1 0.0
8.9 -
7.8 -
tn
=> 3-
B.7 o
D
O 5.6 -I
4.5 -I
3.4
I-
2 3 -
1 .2 -
0. 1
1
010
'"I—
0.68
i ^ n r
1 2 6 1.84 2.42 3.00

Damkohler Number
i o.o
0.10 0.68 1 .26 1 .84 2.42 3.00

Damkohler Number
Figure 6-5 Concentration of cyclized end-groups

(mmol/113g) for
the nylon 6,6 wiped film reactor
model effluent at a
steam vapor pressure of 300 mm Hg
and N =50; a) 270 .
miX
°C, b) 300 °C
This large stoichiometric imbalance is limiting the amount o£
amidation which can take
place.
The majority of
stabilized chain-ends which
are inactive and can
not undergo nidation are
cyclized ends. The concentration of
cyclized ends for the same
operating conditions as
Figures 6-5a,b are
shown in Figures
6-6a,b. Similar overall trends
emerge for the
cyclized end-group concentrate;
they are present at low
temperature
although there production
increases very sharply with
increasing
temperature and is insensitive
to the rate of mass transfer.
However,
the concentration of cyclized
ends is always higher than
the
difference in amine and carboxyl
end-group concentrations.
Since a
comprehensive kinetic model including
degradation is not
always available it is important
to examine the predictions
of the
wiped film reactor model in the absence of degradation
reactions.
Figure 6-7 shows the model predictions
for a typical set of operating
conditions. This should be compared to the predictions
of the model
with degradation for the same operating
conditions in Figure 6-2c. At
smaller residence times (Damkohler numbers less than 1.5) the two
models give similar molecular weight predictions;
however, at larger
residence times, the predictions of the model without degradation
are
significantly larger. The highest molecular weight predicted without
degradation is over 19,000 g/mol whereas the prediction with
degradation is just over 17, 000 g/mol. This is consistent with the
increase in degradation products shown in Figures 6-5b and 6-6b. It
should be noted, however, that these results are in contrast to those
of Ravindranath and Mashelkar (1984) who found that PET degradation
reactions increased the rate of molecular weight generation. This is

150
a 1 0.0
8.9 -
7.8 -
tn
D —
—)
8.7 -
D
T—
O 5.6 -
4.5 -
3.4 -
c
H
2.3 -
1 .2 -
0. 1 -i
0. 1 0 0 68 1.84 2.42 3.00
Damkohler Number
o. i o 0 68 1.26 1 S4 2.42 3 00
Damkohler Number
Figure 6-6. Difference in amine and carboxyl end-group

concentrations (mmol/113g) for the nylon 6,6 wiped
film reactor model effluent at a steam vapor
pressure of 300 mm Hg and N =50; a) 280 °C, b) 300
.
mi.x
°c.
151
0 68
f
'
i
2B
i
r
1.84
i
i
—
2.4^
4
3.
Domkohler Number

nylon 6,6 wiped film reactor model effluent
neglecting degradation reactions at 280 °C and 300
mm Hg Steam Vapor Pressure
152
because degradation reactions in the

cne PET nni „
polycondensation form
reactive carboxylic „,
y acid end-groups
fl
which participate in the
condensation reaction
action r+-K«,
(they react with hydroxyl en( ,_ _
(
Polycondensation via t-fc-i«
this . ,
mechanism is much faster than the usual

ester interchange route.
ê wiped
continues to polymerize
film polymerizer effluent ig ^ ^ equiubrium ^ d
with further processing (e.g., in transfer
lines and processing equipment). Peebles and Huffman (1971)
showed
that the degradation
behavior of nylon 6,6 was
drastically altered if
the degradation too,
p laC e in a closed reactor
where the gaseous
products were maintained over
the melt compared with an
open reactor
The degradation scheme
we are employing is only
applicable to open
reactors. We may turn off the
degradation reactions and calculate
the
equilibrium molecular weight which
may be reached in a closed
reactor
(transfer line) This has been done in Figure
.
6-8a which should be

compared with Figure 6-2c which is
the initial condition for Figure 6-
8a. There is a significant increase in
molecular weight which will
occur if the reactor effluent is
kept in a heated transfer line long
enough (approximately 20-40 min) without

significant degradation. We
expect that Figure 6-8a represents an upper bound on the attainable
molecular weight as it has been calculated by ignoring

the degradation
which would probably consume end-groups and lead to a lowering of the
molecular weight
A similar calculation for the same operating temperature and
degree of mixing but for a lower pressure are shown in Figure 6-8b
which should be compared to Figure 6-3a. Even more drastic increases
in the molecular weight are possible. A striking feature of the

0 10 0 85 I
I
26
I
1
I
84
-1 1
2 42
I
— 3 00
Domkohler Number
Figure 6-8. Ultimate number average molecular weight

predictions
for the nylon 6,6 wiped film reactor
model effluent
f0r = 50; a) T=280 °
Vlx c P = 300 mm Hg, b) T =280
'
°C, P = 100 mm Hg, c) T = 300 °C, P = 300 mm Hg

154
ultimate
the reactor outlet

molecular
molecular weight lots

B8ight ^ (Figures ^^^ ^
p (Figures 6-2o and 6 - 3a
is
^^ ^ ^
,
the fact that the
former are .ore
highly dependent on the
film
thinness (Threle modulus,
temperature
than the
^ ^
i, increased t0 30 „
^ >t , ^ ^^
and N
m x of 50 the increase in ultimate molecular
.
weight is reduced as
shown in Figure
increase of the ultimate

6-8c ,c. £ ., Figure 6 . 4c) . The reductiQn in ^
molecular weight can he
attributed to the
difference in reactive
chain ends £ or the
nidation reaction and the
prolxferation of unreactive
chain ends (see Figures
6-5d and 6-6d, .
Since the ratio of the

Thiele modulus to the Damkohler
nuêr is
independent of the reaction time
(see egs. 3.18 and 4.3, we may
interpret an increase of the
Thiele modulus at a constant
value of the
ratio as a larger reactivity
of the main amidation
reaction, e.g.,
increasing the catalyst activity. The effect of a catalyst on
the
effluent molecular weight is shown in Figure 6-9a for a reactor
temperature of 280 «C, steam pressure of 300 mm Hg and N . of 50.
inix
The molecular weight is increased

significantly at all values of the
2
ratio of <)> /Da. At high values of this ratio, corresponding to short
reactors (small residence times) and thicker films (longer diffusion
times) the increase in molecular weight due to a more
active catalyst
is significant. However, at low values of the ratio which correspond
to long reactors (long residence times) and thinner films (shorter
diffusion times) the increase in molecular weight is much greater.

Thus, when the reactor is operating under reaction control the effect
of a more active catalyst is much greater than when the reactor

0. 1
2 A 57 10.0
Thiele Modulus/Da

°C and 300 mm Hg Steam Vapor Pressure; a) N = 50,
.
mix '
b) N .
=1
mix
156
operates closer to diffusion control

ro1, Thl. *fa ct iis emphasized
Thls
in
e-s b „ hich ...
and therefore thi

calculated under a no _
reactor ep . m„ ^
mixing
^
operaUn9 oonditiM
6.4 Conclusions
A wiped film raodel ( „ the design pertormanM ^ ^ of

continuously
developed.
mixed
The model
thln . £Um „ ylon
incorporates
^
composition
^ ^
dependent rate and
equilibria constants, a realistic degradation scheme
and finite gas-
Phase concentration. E ven small amounts of mixing
yle i d very large
rmprovements in both the mass
transfer and molecular weight
generate
In a wiped film reactor. Nylon
6,6 degradation reactions
reduce the
moiecular weight generation by
about 15% under typical operating
conditions. These wiped film reactors
should be operated between 280-
285 «c and at low pressure S 300 m
Hg) in order
(
to achieve maximum
molecular weight and good product
quality with the smallest reactor.
Higher temperatures severely affect product quality through
degradation, and increasing the pressure
lowers the maximum attainable
molecular weight. A catalyst for the main amidation reaction could
yield improved reactor performance at all
but the lowest mixing rates.
CHAPTER 7
CONCLUSIONS AND RECOMMENDATIONS

FOR FUTURE WORK
A simple, activity-based
solution model has been used
to develop
thermodynamic^ p r0 per correlation forms for the
description of the
rate and equilibrium behavior of nonideal polyamidations. The
approach is especially useful
in systems for which
limited equilibrium
and kinetic data are available
such as nylon 6,6.
Many possible degradation mechanisms are postulated in the
literature for nylon 6,6. It is not feasible or necessary
to include
all the side reactions and/or to measure their
kinetic constants in
order to describe the performance
of typical industrial reactors.
We
have hypothesized a minimal subset of reactions that can
be used to
estimate the degradation and we have
regressed kinetic constants for
them that are consistent with available
data. We have shown that such
a simplified degradation scheme can be used to
predict the behavior of
continuous industrial reactors.
An unsteady reaction-diffusion model has been developed to

represent the polymerization of a nylon 6,6 prepolymer in a stationary
film. The realistic kinetic and equilibrium description for nylon

6,6
has been incorporated by the use of apparent rate and equilibrium
constants that are composition dependent. The model is suitable for

158
evaluating the effect

tect(s)
fi> nf
of -i.
changes in throughput
(film thickness)
temperature and catalvst
talyst ar-t-i,,;,.
activrty on reactor
performance. The model
- characterised hy a ,
modulus. The composition

ingle d i men3lonless
^
dependent kinetic
constants are important
for accurate
prediction of molecular
weight evolution in the
film
Below a threshold
value of the Thiele
modulus, molecular weight
generation is reaction
controlled, while aoove
another threshold value
it is diffusion
^ ^
controlled.
^ flowing film model ig ugeful ^^ ^

continuous, thin-film
polarizers. The addition of a flow
field to
the stationary f Um reaotion-dif fusion problem
has been shown to move
the region of highest
moleoular weight from the gas-film interface
toward the wall-fil m interface for thin
fil ms . An idea ii Z ed kinetic
model leads to significant
errors for molecular we.ght predictions
although it is sufficient for predicting the outlet water
concentration. Linear and parabolic velocity
profiles have been shown
to yield similar predictions for the outlet average water
concentration and molecular weight
whereas the plug flow profile gives
notably different predictions. Running
the flowing film device as a
reactor is inefficient because very large
sizes are required. Most of
the water is removed at the beginning and in the remaining
portion
very little additional water is removed because
the driving force for
diffusion is low and molecular weight generation is reaction
controlled.
By performing dynamic modeling we have shown that the flowing
film model predicts that the outlet average molecular weight and
water
concentration come to steady-state in about one mean residence time

159
<t.r a c h an g e in fl
Product. co„ tinU e t0
r"
condition,
0
"' ^
a S „e r operatin,
haTC
change
co„ ditlo „s wher eas the
th3t U
^
even
P ° S3lble
^^
* —^ «"
Ration
^
of the £lash e r
30 the sam e mol ec ular
wei ght is produced at
different textures but „ ith very
degradation
di£ferent ^
products givinq
ing rise fo
y to f
two products of significantly
different quality.
By using
both our simplified
kinetic and degradation
models for
nylon 6,6 we have developed
a realistic wiped film model
for t he
design and performance
analysis of continuously
mixed thin-film nyl on
6,6 polymerias. Even small amounts
of mixing yield very
large
improvements in both the mass
transfer and molecular weight
generation
in a wiped film reactor.
Nylon 6,6 degradation
reactions reduce the
molecular weight generation
by about 15* under typical operating
conditions. These wiped film reactors
should be operated between 280-
285 °C and at low pressure (.300 mm Hg) in order to achieve maximum
molecular weight and good product
quality with the smallest reactor.
Higher temperatures severely affect product quality through
degradation, and increasing the pressure
lowers the maximum attainable
molecular weight. A catalyst for the main amidation reaction
could
yield improved reactor performance at all
but the lowest mixing rates.
Future research in the area of continuous polycondensation

reactor design could be benefited most by performing experimental
studies on different (ideal or well characterized) reactors. These
could verify the reactor models developed in this thesis.
Also, they
would provide a base of published data for the testing of future
models. Lastly, they could provide a hueristic basis for deciding

160
— som e o£ the typical

^
-activity) made in po ly
co„ densation reaotor modeling _
REFERENCES CITED
Abraham' w
^,^^0.,
- ind
F.
Ji
„.
n _ch M^
Reinhart and s
2,
. H
221 (1963)
Kline>
^
Sm° n
ASME
'
^ °- ~< Inl^H^ûnda^ 19, 415 ,1980,
«r?
(1980)
wîVri
p. e-18.
t^Hi^^
^ ' CRC Press, Boca Raton, FL
Ault, J. w. andD. A.
Mel li champ, CJêm^n^ci, 27, 1441 (1972).
Balli er
^ 2 991
D
il987)
ZZ
°' ^ GiUffrida and G
-
M -taudo, Macromol,
Biesenberger, J. A. AIChE
, J. U, 3 69 (1965).
^^-^-i£^_^^
^'^scLnc:- (1984) p
wiley .
^^'pje^; wil^f ^ ^rt (1960)

E
-
Lightf ° 0t
'
C ° Sta
°-
Dissertation Natl. Poly. Inst
F^-(l983)-: ' . Lorraine
Edel, G. and H. Ettiene, Bull.

Inst. Text. Fr .
23(142) 343
, (1969).
Flory, P. J J. Chem. Phys. 12.
. , 425 (1944).
Flory, P. j., Chem. Revs. 39. 137

(1946).
Flory, P. J., Principles of Polyme r

Chemistry Cornell University
Press, Ithaca, NY (1953) y .
Fukumoto, 0., J. Polymer Sci. 22, 263

(1956).
Giori, C. and B. T. Hayes, J. Polymer Sci. A-l 8, 355 (1970).
Gupta, S. K. , N. L. Agarwalla, P. Rajora and A. Kumar,

Sci., Polym. Phys. Ed. 20, 933 (1982) .
J. Polym —
Gupta, S. K., N. L. Agarwalla, and A. Kumar,
27, 1217 (1982) .
J.
™
App. Polym. Sci
1 1
Gupta, S. K., A. Kumar and A.

1063 (1983) .
K. Ghosh, J. "
App. Polym.
1
Sci 28,
162
Gupta, S. K., a. K.
Ghosh S K
?olym. Sci^ 29, 3217
rw
PU/ an ^' Ku™r, J. App.
(1984)".
Heikens, D. p w u Qy^
,
Herma "3
Mi 437 ,i 9 60) '
and G - « der Want, J. p„ lvBer Sr.i
NY (1984)': -^^^^ Dekker,
Ja ° 0b
Ni U977)
and J '
Zi ~erman - Polymerization Prn...... Mlley>
K "Sr
M«3 Ifuiuoes,
Kilkson, H., Ind. Eng. Ch em.

Fundam.
^rsoc
3 G
^
l
in soc. Chem. ind.
Degradation of
?
Monograph 13, 357 (1961)
'% ^Sal
.
3, (1964) 281.
Kirkwood J G ., R. L. Baldwin, P. j. Dunlop , L< j Gogti ' and
b0Stina
G- Kegeles, J. Chem. Phys. 33, 1505 (I960).
Kri3h and R
;i^
Publishing,
-
Ta V 10
Houston, TX (1986)
^
Handbook of Heat and Mass Tranfer
Gulf
er bUir .
'
Kumar, A. S K. Gupta, S. Madan, N.

, G. Shah and S. K. Gupta,
Polym. Eng. Sci. 24. 194 (1984). P '
KUmar
'(1984)
S
'
N
*
Shama
^ S '
K '
GUPU '
Po1 ^- Eng. Sci. 24. 1205
KumarÂ.^S. Juruville, A.R. Raman and S.K.Gupta, Polymer
Kumar, A., P. Rajora, N. L. Agarwalla, and

S. K. Gupta
Polymer 23, 222 (1982)
Kumar, A., S K. Gupta, S. Madan, N.

Eng. Sci. 24, 194 (1984).
G. Shah and S. K. Gupta, Polym.
^
Laurence, R. L. and M. V. Tirrell Polymerization Reaction
"
Engineering in preparation ,
"
Lin, K-H and H. C. Van Ness, Reaction Kinetics, Reactor Design, and
Thermodynamics Ch. 4 in "Chemical Engineers' Handbook'* (R. H.
,
Perry and C. H. Chilton, eds.) McGraw-Hill, NY (1973).

163
Meacoc *' G
'' i_*EEii_£ha5j. 4, ,1954, 172.
Mill3
<
'-568^1^ 41, 1045 ,19 86) .
MU1S P
' '
^^..Dev^ 25, 575 ,1986).
MUls '
p ' 1- ComâemÊ^ io, 3 99 (1986).
Morrison, R. T anri r m «
B ° yd
Bacon, ,1966) ' Siaanic^hemistrîed, Allyn and
Nagasubramanian, K and n « D
*' Reims ûessel,
12, 1663 (1973) J^Jôl^m^Sci,
Newitt, D. m. and
Weale
»eaie, K f
K. E. t
J. ,u
Chem.
, Soc. 1541 (1948).
0gata '
M./ Makromol. ChPm 49 52 (1960) ,
° gata ' N - Makromol.

Chem 43 117
; (1961) _
Peebles, L. H. and M. w. Huffman,

J^m^Sci, 9, 1807 (1971).
PreSS
Nu^rlci/^
P
and W. T. Vetterling
Scientlfl ^ ConjE utin Cambridge
"m-rmt^ r, ,. ! 11! gju!) g
RabinowitÊ. and W. c. Wood, Tr^nâda^ 32, 1381
RaVen S a d
R;aotions of po lêrs
Interscience, (1964??'
Ttlo f^^F^' tSS '
in
ed ° Wiley "
Ravindranath, K. and R. A. Mashelkar, AIChE J. 30, 415 (1984) .
Ray W H - Advanced Pro cess Conrol

' -
. McGraw-Hill, NY (1981).
Reimschuessel^H. K. and K. Nagasubramanian, Chem. Eng. Sci. 27,
Secor, R. m. and J. A. Beutler, AIChE J. 13, 365 (1967).

Secor, R. M., AIChE J. 15, 961 (1969).
Smith ChemiCal Enqin eerinq Kinetics nd

2 ed., McGraw-Hill, NY
U^'/u;
Smith, J. M. and H. C. Van Ness Introduction

to Chemical
Engineering Thermodynamics McGraw-Hill (T575) ,
-
Snider, 0. E. and R. J. Richardson, Polyamide

Fibers in The
Encyclopedia of Polymer Science and Technology, Vol.
10, pp
347-460, Wiley-Interscience, New York, (1964)
164
Steppan, D. D M r D0hert
n u
111 ihl'anj; ^ «* «• r. Malone, J. Appl.
gola^cj,
Tadmor, Z anrt n*
336 '(1966)
t a
Blesenb êr, InoÊnc^^^
Tadmor, z anH r n n
G ° g0S
NY, ,19^°- ' SHH^Mesôla^Proce^, wiley ,
Tai
Taylor
'
li: m'l?^' i^l^^ Arai »d Tagawa,
' J^^hemôc, 69, 642 ,1947,

T ° b0l
af59?:
V
'
0^^^ A "
81, 23 02
3 ° f "l£gal Degradation
'""lîSei/L'soc"
ooc. ?^^
Chem. Ind. Monograph
of Some
13, (1961) .
U
"
"posies 1 ' 717

<«' "44, f "Process for Making
U. S Patent, 3,900,450 (Dupont,

1975)
by Continuous Polymerization-
" Prftna
Pre P a "tion
^ of Polyamides
U 8
"
»U?'p!i512 LUP n
in a
3 Cont
1976) "îng the Flow of
Contln ûs Process for *
(
'
Thereof". the Preparation
^ D
STl2,
B
472
M
il983r
erty
'
* *' I^-M-
Wiloth, F., Kolloid Z. 3, 129 (1955) and 3, 138 (1955)

Wiloth, F Z. Physik. Chem. 4.
. , 66 (1955).
Wiloth, F, , Makromol. Chem. 27. 37 (1958).
Wiloth, F., Makromol. Chem. 15. 98 (1955).
Wiloth, F. and E. Schindler, Chem. Ber. 100 , 2373 (1967

Wiloth, F., Makromol. Chem. 144, 283
(1971).
Wiloth, F., Makromol. Chem. 144, 329 (1971).
Wyness, K. G., J. Chem. Soc. 2934 (1958).

165
Z
~-' Pol^s^Letters 2, i 95 5 ,1964).

The Design of Continuous Polycondensation Reactors

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

The Design of Continuous Polycondensation Reactors

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Design of Continuous Polycondensation Reactors

Uploaded by

Copyright:

Available Formats

University of Massachusetts Amherst

Doctoral Dissertations 1896 - February 2014

The design of continuous polycondensation reactors/

Follow this and additional works at: https://scholarworks.umass.edu/dissertations_1

DAVID DANIEL STEPPAN

Submitted to the Graduate School of the

Polymer Science and Engineering

All Rights Reserved

DAVID DANIEL STEPPAN

Approved as to style and content by:

Michael F. Doherty, Co-Chairman of Committee

Michael F. Malone, Co-Chairman of Committee

Robert L . Laurence, Member

David A. Hoagland, Member

William tf. MacKnighfe, Department Head

I would like to express my sincerest gratitude to Dr. Michael F.

of luck in all their personal endeavors because they do not need it

I also extend my thanks to my thesis committee for their help

and suggestions. In addition, I thank all the people who helped me

Control Center and fellow group members. The financial support of

the Dupont Company is also acknowledged.

Lastly, I would like to express my thanks to my family and

thesis is also their accomplishment.

THE DESIGN OF CONTINUOUS POLYCONDENSATION

DAVID STEPPAN, B. S., CASE WESTERN RESERVE UNIVERSITY

Ph. D., UNIVERSITY OF MASSACHUSETTS

Directed by: Professor Michael F. Doherty

and Professor Michael F. Malone

A simplified approach to polycondensation kinetics has been used

to develop engineering models for the design of film-forming

This approach has been applied to a nylon 6,6 prepolymer in an

idealized flowing film device. The model is characterized by two

dimensionless groups; the Damkohler number and the Thiele modulus.

A realistic kinetic and equilibrium description was incorporated

with an activity-based kinetic model with composition-dependent

apparent rate and equilibrium constants. The model is suitable for

evaluating the effect (s) of changes in throughput, flow, film

thickness, temperature and catalyst reactivity on the device

Since nylon 6,6 is typically used for fibers, molecular weight

product. The importance of the residence time distribution on the

Finally, a realistic wiped film model for the design and

composition dependent rate and equilibrium constants, a realistic

degradation scheme and a finite gas-phase concentration. Even small

amounts of mixing have been shown to yield very large improvements

in both the mass transfer and molecular weight generation in such

NOMENCLATURE **• •••••••••••••••••••••••»••••••• XIX.

1.2 Previous Work 2

1.3 A Simplified Approach 5

1.4 Research Objectives 7

2 A SIMPLIFIED KINETIC, EQUILIBRIUM AND DEGRADATION MODEL

2.1 Kinetic and Equilibrium Model 10

2.1.2 A Nonideal Liquid Model 14

2.1.3 Equilibrium Correlation 15

2.1.4 Amidation Reaction Rate Correlation 23

2.2 Degradation Model 31

2.2.1 Background and Degradation Schemes 31

2.2.2 Experimental Degradation Studies 36