Professional Documents
Culture Documents
-
distribution of polymer chain lengths describes the Initiation: ring opening 0
reaction product. For a product like nylon-6 that is ki II
typically processed by extrusion and melt spinning, HN-(CHn)j-C=O + HzO NH~(CHZ)~-C--OH
molecular weight distribution, through its effect on ki’
melt viscosity, is a very important property. The most (MI (w) (S1)
probable distribution of molecular weights is found
in most commercial nylon-6 products (18) and does Propagation: polycondensation
indeed give a desirable product of uniform quality kz
(20).
NH2 + HOOC---- + + HzO
kz’
With these factors in mind, we have developed a - - - -CONH- -- -
treatment of caprolactam polymerization which can
-
(Sn) (Srn) (Sn+m) (W)
predict not only conversion and number-average polyaddition
degree of polymerization, but also can describe k3
mathematically the molecular weight distribution as ----NHz + HN-(CH&-C=O s
k3’
a function of reaction time. Our objective has been to
(Sn) (MI
formulate a model adaptable enough to be applied to - - - -NHCO (CH2)sNHz
a wide variety of reactor conditions and configura- (Sn + 1)
tions. We are able to show in which reactor designs ~
a most probable distribution is indeed the expected Reaction with monofunctional species:
product and how it may be drastically different in a k4
CSTR or in other reactor configurations. In this same ---- NHz + HOOC--.---I
vein, the common practice of adding a monofunc- K
tional agent to control molecular weight is examined (Sn) (Am)
in some detail with regard to its effect on molecular - - -NHCO- - -I + Hz0
(An + rn) (W)
weight and molecular weight distribution. ~~
n=l
Sn
n=l
Sn+@ k3MSl -+ k;Sp
tiveness of this type process lies in the speed of the m m
Sn + Sm k2 Sn+m+ W
i= (2)
k2'
Aj 0
- k4)W(An- Al 6 <n - 1> + k:W 2 -
Sn + M*Sn+t
k,'
k3
(3) j=n+l j-1
(10)
where the function:
k4
Sn + Am
k41
An+, +W (4)
1This term is not exactly correct since an Si moleaule does not result
from every reaction between water and a polymer chain. The correct
m
term is . . . kz'W Z Sn/n
*k5a
- 1. Unfortunately, it cannot be easily
Batch Reactor transformed to give the correct term in the moment equations to h e
developed presently. However, it is apparent that the correct term
The molecular rate equations for the disappear- is some fraction of the term which has been used. Our term repre-
sents an upper bound. The small magnitude of the rate constant kz'
ance of the various species can be written as follows: (see Table 2) argues for the acceptance of this approximation.
?Solution of this set of equations including this term requires as-
dM =
- = - kIMW + kl'S1 suming that SI S2 which has been used previously ( 5 ) and is ac-
ceptable since both concentrations are quite small. Alternatively, it
dt could be ignored altogether which has also been done in some theo-
retical studies of reversible polymerization (12).
3This term gives the proper statistical weight to the hydrolysis r e a c
tion of water with the specific amide linkage in a molecule of size i
which will just give a molecule of size n.
POLYMER ENGINEERING AND SCIENCE, MAY, 1975, Vol. 15, No. 5 387
M. V.Tirrell, G. H . Pearson, R. A. Weiss and R. L. Laurence
dP,’
=O
dt
-k3) (P (s) - -1 P ( s) + Sl) + ks’Sls - k4P( s ) Po)
S
3aa POLYMER ENGINEERING AND SCIENCE, MAY, 1975, Vol. 15, No. 5
An Analysis of Caproluctam Polymerization
POLYMER ENGINEERING AND SCIENCE, MAY, 1975, Yo/. 15, No. 5 389
M . V. Tirrell, G . H . Paarson, R. A. Weiss and R. L. Laurence
-
I 0
I
C
b
0 5 10 15
time, hrr. a
Fig. 2. First moment of PCLD us time in a batch reactor at
three initial water concentrat'ons: (0)
Wo/Mo = 0.05; ( 0 )
W,/M, = 0.067; (0) Wo/M, = 0.1. p o is in mole kg-I
x 102; A = 0. Symbols are experimental data of Hermans, I
stants that provide the statistical best fit to a given Fig. 3. Predicted curves for conuersion x, number average
set of data over the entire conversion range. We did degree of polymerization z,and,
in u butch reactor at various
levels
polydispersity Z, vs time
monofunct'onal acid:
of
not attempt to optimize the fit to the data in this
(a) Ao/Mo = 0.00; (b) A,/M, = 0.0025; (c) A J M , = 0.01.
manner, however.
Two reactions not considered in the model may
contribute to the discrepancy between the predicted broaden the MWD somewhat, especially in the early
curves and the data. The first is a cyclization reaction stages of polymerization. As with any theoretical re-
which, although similar in mechanism to the reverse sult, the validity of the prediction is difficult to argue
reaction 3, produces not monomer but cyclic oligom- in the absence of some of the experimental results it
ers. Indeed, commercial products do contain these seeks to predict. This suggests that an experimental
species in minute amounts up to a cyclic nonomer investigation of MWD versus conversion would be
( 18). If kinetic data were available (namely, cycli- interesting.
zation constants ) this reaction, or an approximation
to it, could be treated by our model, similar to the The CSTR
treatment of Mochizuki and Ito (26). The other re- Figures 4 and 5 describe the polymerization in a
action not considered is the transamidation reaction. CSTR. We see that very broad distribution product
I t is difficult to assess a p r i o ~ ithe effect of this may be obtained in a CSTR, consistent with the par-
reaction. Some kinetic data are available ( 5 ) but tial step growth character of the polymerization (19,
inclusion of this reaction in a kinetic model would 23). No data on continuous caprolactam polymeriza-
be extremely difficult. tion are available for direct comparison. In Fig. 5
Figure 3 gives conversion versus time, as well as it is seen that control of product molecular weight
the development of degree of polymerization and and MWD can be obtained by addition of 0.25 to
MWD, for a typical initial water concentration with 1.0percent monofunctional acid.
and without monofunctional acid. Addition of mono- One final point with respect to the CSTR is that
functional acid somewhat decreases the time for at- we have assumed perfect mixing. If some degree of
tainment of the equilibrium conversion. The mo- segregation had been introduced, the prediction
lecular weight controlling effect is demonstrated by would probably be a narrower PCLD in keeping
the Dp, versus time curve. Considering now the with the general results for condensation polymeri-
molecular weight distribution, we note that a value zations (19, 2 3 ) .
of 2.0 is characteristic of a most probable distribu-
tion. Except for a plateau at about 1.75 in the early Series Configurations
stages of the reaction where polymerization is dom- A primary advantage of this deterministic model
inated by the chain growth mechanism polyaddition, is demonstrated in treatment of series reactor con-
the value of 2.0 is approached as the equilibrium figurations. Tubles 3-5 deal with various plug flow
conversion is neared. It is seen also that addition of reactor-CSTR schemes for producing nylon-6 con-
small amounts of monofunctional acid is predicted to tinuously. Equal reactor volumes are considered with
390 POLYMER ENGINEERING AND SCIENCE, MAY, 1975, Vol. 15, No. 5
An Analysis of Caprolactam Polymerization
I. r
1. I
,OOt
20 c 4
I
50 100 I50 200
Time, h r t .
Time, hn.
Fig. 5. Predicted curues for x,oPn and 2, us residence time
Fig. 4. Predicted curoes for x, so (m/kg x
102)=, and 2, in a CSTR at various leuels of monofunctional acid: (a) &/M,
os residence time in a CSTR at uarious initial water concen- = 0.00; (b) A , / M , = 0.0025; (c) A , / M , = 0.01.
trations: (a) W , / M , = 0.01; (b) W , / M , = 0.05; (c) W,/M,
= 0.1. it is seen that either series configuration can be used
to obtain a similar product at comparable residence
the space time, 7,required to give the specified prod- times. The key point is that it is not recommended to
uct after each stage r:hown. With no water
__
removal, try to obtain high conversion in the CSTR section.
using as acceptable product criterion a DP, of 150, The largest part of conversion and growth should
First reactor: PFR Water removal Second reactor: CSTR Total space
effluent specifications after first reactor effluent specifications time (hrs.)
- -
DP, X 2, TI (hrs.) DPn X 2, 72 (hrs.1
~~ ~ ~ ~ ~~
* Wo/Mo = 0.05.
First reactor: CSTR Water removal Second reactor: PFR Total space
effluent specifications after first reactor effluent specifications time (hrs.)
- -
D Pn X 2, TI (hrs.) DPn X 2, 72 (hrs.1
First reactor: PFR Water removal Second reactor: PFR Total space
effluent specifications after first reactor effluent specifications time (hrs.)
-
DPIl X 2, TI (hrs.1
-
DPIl X Z, ~~(hrs.1
Wo/Mo = 0.M.
occur in the PFR with the CSTR used as a starting Polym. Sci., 16,929 ( 1972).
or finishing reactor. These same general rules hold 3. H. K. Reimschuessel and K. Nagasubramanian, Chem.
Eng. Sci., 27, 1119 (1972).
for the cases of complete water removal after the 4. H. K. Reimschuessel and K. Nagasubramanian, Polym.
first stage. Water removal does give a considerable Eng. Sci., 12, 179 (1972).
process time advantage in agreement with results 5. H. K. Reimschuessel in “Ring Opening Polymerizations,”
obtained by other workers (2, 3 ) . This is further K. C. Frisch and S. L. Reegan, eds., Ch. 7, Marcel Dekker,
New York, N. Y. (1969).
demonstrated in Table 5 where two PFR’s in series 6. Ch. A. Kruissink, G. M. van der Want and A. J. Staver-
are run with water removal between stages. This man, J. Polym. Sci., 30, 67 ( 1958).
configuration has interest since it is the limiting case 7. P. H. Hermans, D. Heikens and P. F. Van Velden, J.
of an infinite series arrangement of CSTR’s. These re- Polym. Sci., 30, 81 ( 1958).
8. F. Wiloth, 2. Physik. Chem., 5, 66 (1955).
sults show that it is desirable to remove water from 9. F. Wiloth, 2.Physik. Chem., 11, 78 (1957).
the reaction as early in the process as possible. How- 10. H. K. Reimschuessel, J. Polym. Sci., 41,457 (1959).
ever, we would not expect it to be desirable to re- 11. R. W. Lenz, “Organic Chemistry of Synthetic High Poly-
move the water at a residence time lower than the mers,” Wiley-Interscience, New York, New York ( 1967).
point at which the maximum in the p,, curve occurs. 12. C.-R. Huang and H.-H. Wang, J. Polym. Sci., A-I, 10,
791 (1972).
The proper choice of series configuration can provide 13. J. Sebenda and V. Kouril, Europ. Polym. J., 7, 1637
a considerable improvement over a single CSTR. ( 1971).
In summary, the deterministic model developed 14. P. Biernacki, S. Chrzczonowicz and M. Wlodarayk, Erirop.
in this paper has been applied to various single and Polym. J., 7, 739 (1971).
15. S. Saunders, J. Polym. Sci., 30,479 (1958).
series reactor configurations. From this model pre- 16. Modern Plastics, p. 70, August 1969.
dictions about MWD as well as molecular weight 17. S. Smith, J. Polym. Sci., 30,459 (1958).
and reaction conversion can be made. Distributions 18. A. D. Bliss, Foster Grant Co., private communication.
other than most probable result in many cases. Some 19. R. L. Laurence, unpublished manuscript,
tuning of the model is indicated to improve the 20. S. I. Calouche, Firestone Synthetic Fibers Co., private
communication.
agreement between the model and the available 21. W. H. Abraham, Ind. Eng. Chem. Fund., 2,221 (1963).
data. A more systematic approach to determining 22. H. Kilkson, h d . Eng. Chem. Fund., 3,281 (1964).
the optimal reactor configuration through the use of 23. Z. Tadmor, Ph.D. dissertation, Stevens. Inst. Tech.,
this model is likely to prove quite fruitful. (1966).
24. D. Heikens, P. H. Hermans and G. M. van der Want,
NOMENCLATURE J. Polum. Sci., 44. 437 ( 1960).
25. H. Kiikson, lnd. hng. Chem. Fund., 7, 354 (1968).
A, = concentration of monofunctional species of 26. S. Mochizuki and N. Ito, Chem. Eng. Sci., 28, 1139
- chain length i (mole kg-’) (1973).
DP, = number average degree of polymerization
-
DP, = weight average degree of polymerization APPENDIX A
ki = forward rate constant of ith reaction It can be shown that
k; = reverse rate constant of ith reaction
Ki = equilibrium constant of ith reaction
M = concentration of caprolactam (mole kg-’ )
Si = concentration of difunctional species of
chain length i (mole kg-’)
W = concentration of water (mole kg-’ )
- -
Z,, = polydispersity = DPJDP,
pk = kth moment of the distribution of chain
lengths of difunctional species
-S s” -dsr}
P(S’) (A-1)
pk’ = kth moment of the distribution of chain O (s’)3
pk:= C nkSn
n=l
( A-2 1
m m n
01 m c 1
n-1
n2 2
j=n+t 1-1
LJ
6
.;
--=1(2fi-p1-S1) (A-5)
Sl, - SlO
= klMW - kl'S1- k2Sp4
e
POLYMER ENGINEERING AND SCIENCE, MAY, 1975, Vol. IS, No. 5 393