Polymer 42 (2001) 39093914

www.elsevier.nl/locate/polymer

In-line monitoring of weight average molecular weight in solution

polymerizations using intrinsic viscosity measurements
M.P. Vega, E.L. Lima, J.C. Pinto*
Programa de Engenharia Qumica / COPPE, Universidade Federal do Rio de Janeiro, Cidade Universitaria, CP: 68502, Rio de Janeiro 21945-970, Brazil
Received 21 August 2000; received in revised form 21 October 2000; accepted 25 October 2000

Abstract
The main objective of this work is developing a simple viscometrical method for in-line monitoring and control of weight average
 w in solution polymerizations. The method is based on the evaluation of the approximate intrinsic viscosity of the
molecular weight M
polymer solution, at each sampling time, using a single measurement of the ow time of a diluted polymer solution through a capillary tube.
Experiments were carried out for peroxide initiated styrene solution polymerization reactions in a tubular reactor. Results obtained for
approximate intrinsic viscosities are in agreement with the intrinsic viscosity values published in the literature and allow fast and fair in-line
 w : q 2001 Elsevier Science Ltd. All rights reserved.
values for M
Keywords: Intrinsic viscosity; Capillary viscometer; In-line weight average molecular weight

1. Introduction
The nonlinear behavior of polymerization systems and
the difculties of sensor development, which requires
knowledge and understanding of process, reactor design,
mathematical modeling, control techniques, instrumentation and multi-disciplinary elds, make sensor technology
a difcult task. This is why in-line monitoring of polymer
properties is complex in nature, in spite of the efforts to
develop density, viscosity, surface tension, refractive
index, composition, light scattering and chromatographic
methods for the evaluation of polymer characteristics [1,2].
There is much interest in the in-line monitoring and
control of molecular weight distributions (MWD) during
polymerization reactions as this may be regarded to be
among the most important molecular properties of polymer
resins. However, the analysis of polymer chain length using
gel permeation chromatography (GPC), size exclusion chromatography (SEC), and light scattering, requires very
expensive, sophisticated, time-consuming and unreliable
(at industrial environments) instruments. From a practical
point of view, most of the times, the whole MWD is not
needed and signicant amount of information about the enduse properties of the polymer resins may be provided by the
leading moments of the MWD, such as the weight average
* Corresponding author. Tel.: 155-021-280-8832; fax: 155-021-2906626.
E-mail address: pinto@peq.coppe.ufrj.br (J.C. Pinto).

 w [3]. Even in this case, however,

molecular weight M
 w are not well develtechniques for in-line monitoring of M
oped [1,2].
In-line viscometers are useful for monitoring the evolution of polymerization reactions and are widely used as an
indirect measurement of the average molecular weight of
polymer solutions [1,2,46]. However, as the individual
effects of conversion and molecular weight on viscosity
are not readily separable, it is very difcult to use viscosity
techniques at middle and high conversions [1,2]. In principle, the intrinsic viscosity [h ] of a polymer solution does not
depend on conversion and may be very useful for in-line
monitoring of average molecular weight of polymer resins.
In spite of the huge amount of information available about
[h ] in the literature, attempts to monitor [h ] in-line and use
this value to predict the average molecular weight of the
polymer resin have not been made.
By measuring the intrinsic viscosity of polymer solutions,
the polymer average molecular weight can be predicted
through the MarkHouwink empirical equation, as
described by
 av
h K M

where [h ] is the intrinsic viscosity, K and a are the para v the

meters that depend on the solvent/polymer pair and M
viscosityaverage molecular weight of the polymer. If the
shape of the molecular weight distribution does not change
signicantly, as usually observed in continuous free-radical

0032-3861/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0032-386 1(00)00780-1

3910

M.P. Vega et al. / Polymer 42 (2001) 39093914

 v usually is a multiple of
solution polymerizations, then M

 w in Eq. (1)

M w ; so that M v can be normally replaced by M
[7,8].
The intrinsic viscosity can be determined experimentally
through measurements of the specic viscosity, dened in
the following equation

hsp

t 2 t0
t0

where h sp is the specic viscosity, t the ow time of a

polymer solution through a capillary tube of known
diameter and length and t0 the ow time of the pure solvent
through the same capillary tube.
The intrinsic viscosity is described as the limit of the ratio
between the specic viscosity (h sp) and the polymer concentration (c), as the polymer concentration approaches zero,
hsp
3
h lim
c!0
c
Experimentally, the specic viscosity is measured for at
least three different polymer concentrations and a plot of
hsp =c versus c is extrapolated for zero concentration, leading to high precision intrinsic viscosity measurements.
However, this method is not appropriate for in-line monitor w because of the large lag-time required. Besides, it
ing of M
is difcult to precisely control the polymer concentration of
very diluted solutions at plant site.
In the literature one may nd faster methods for obtaining
the intrinsic viscosity based on a single point determination
of the specic viscosity. Kraemer [9] and Huggins [10] were
the rst to propose empirical equations for describing how
the intrinsic viscosity varies with the polymer solution
concentration, Eqs. (4)(6). Therefore, assuming that the
polymer concentration c is known, it is possible to evaluate
[h ] based on a single measurement of the specic (relative)
viscosity. In both cases, however, it is necessary to know an
additional parameter (k1 and k2) that is usually unavailable.
Besides, these additional parameters may depend on the
polymer characteristics and on the polymer/solvent pair,
which means that they have to be estimated simultaneously
with [h ]. Therefore, in practice three or more values of the
specic (relative) viscosity for different polymer concentrations are required for evaluation of [h ] with Eqs. (4)(6)
[11], unless additional assumptions are introduced into the
problem formulation.
hsp
h 1 k1 h2 c
4
c
ln

hrel
h 1 k2 h2 c
c

hrel

t
t0

5
6

Gomes and coworkers [11] showed, however, that quadratic

approximations of the function that relates the specic viscosity of polymer solutions with the polymer concentration

may lead to values of intrinsic viscosity that are much more

uncertain than the values obtained through simpler linear
ts. Schultz and Blaschke [12] developed an alternative
rst-order approximation, as shown in Eq. (7), that also
depends on an additional model parameter k. Solomon and
Ciuta [13] obtained an equation that may lead to the intrinsic
viscosity value with a single point viscosity measurement
and that does not depend on any additional parameter
describing the solventpolymer characteristics, Eq. (8).
Baruah and Laskar [14] determined the parameters of the
MarkHouwink equation for the polydispersed poly(ndocosylacrylate) solutions of 0.5% w/v, using intrinsic viscosity data calculated by the single point method described
by Solomon and Ciuta [13]. It was observed that Eq. (8)
yielded accurate intrinsic viscosities and high reproducibility when measurements were made for dilute polymer solutions, with concentrations below 0.2% w/v. However, such
low concentrations are inadequate for most practical industrial applications.
hsp
h
7
c1 1 khsp
h

p q
2
hsp 2 ln hrel
c

2. Intrinsic viscosity determination

 w ; h sp is assumed
For in-line monitoring and control of M
here to be a simple linear function of the polymer concentration (c), as shown in Eq. (9). The linear coefcient, the
approximate intrinsic viscosity [h ap], is assumed here to
follow the MarkHouwink empirical equation, Eq. (10).
This may provide a simple and fast procedure for monitor w ; requiring the preparation of a single sample of
ing of M
the polymer solution. Therefore, a single concentrationviscosity measurement must be performed for a dilute polymer solution in a capillary viscometer at a xed tempera w information obtained from a
ture, yielding real-time M
MarkHouwink calibration.
hap

t 2 t0 1
t0 c

hap KMwa

9
10

h sp may be measured experimentally with numerous

commercial in-line capillary viscometers available in the
market. Viscotek provides the Semi Automated Sample
Preparation (SASP), which comprises a balance, a syringe
pump and a computer [15]. From Viscologic, the instrument
named Visiologic runs automatically all the operations of
dilution, stirring, cleaning, draining and sequential measurements of ows [16]. Lauda commercializes the PVS viscometry measuring system with a program-controlled burette
for solvent addition, automatic cleaning and ow-time

M.P. Vega et al. / Polymer 42 (2001) 39093914

measurement of viscosities at different concentrations.

Parallel operation permits a sample throughput of eight
samples per hour [17]. Besides, Fig. 1 shows how the specic viscosity depends on the polymer solution concentrations
for typical polystyrene/solvent systems. It may be observed
that the linear behavior is approximately valid up to polymer
concentrations around 1.0 g/dl.
In order to validate the strategy presented here, the peroxide initiated solution styrene polymerization reaction is
carried out in a 12-m long tubular reactor. The feed solution
contains a 4060 w/w% monomersolvent solution and
0.05 kmol/m 3 initiator concentration. Description of the
experimental unit can be found elsewhere [18]. A digital
densitometer, Anton Paar mPDS-2000, is used to monitor
monomer conversion in-line. Samples of the polystyrene
solutions were withdrawn every 1800 s for dilution and
feeding into a capillary viscometer. Diluted solutions were
prepared at room temperature by diluting the original

mer solutions were measured with a CannonOstwald

Fensk capillary viscometer at a temperature of 30 ^ 0:18C
maintained with a thermostatic controlled bath (Haake). The
viscometer was selected in order to lead to a ow time above
100 s. The lag-time between sampling and viscosity
measurements includes dilution and homogenizing (100 s),
ow-time measurements (up to 600 s) and cleaning procedures (100 s). No attempt was made to minimize the lagtime in this work, but some operation parameters may be
used to minimize the measurement time, such as the
diameter of the feeder and mixer tubing, the overall ow
rates and the amount of reaction solution sampled. Polymer
samples, prepared at different reaction conditions, were also
characterized through SEC measurements carried out in
 w (obtained from
solutions of tetrahydrofuran (THF). M
SEC measurements) and viscosity data were used in order
to build a MarkHouwink empirical equation for estimating
polymer quality in-line.
100 X

X=dP 1 1 2 X=dS 1 MT0 =MS0 1=dT 1

sample with pure toluene in order to reach a nal polymer

concentration in the range from 0.1 to 0.5 g/dl. No sort of
purication of the nal polymer solutions was performed.
The polymer concentration (c) was calculated as shown in
Eq. (11) with the help of the process densitometer, which
provides in-line monomer conversion (X) data of the tubular
reactor outlet stream. Remaining variables are the measured
volumes of the reactor sample (VOUTLET) and of the pure
toluene employed for dilution (VADD), initial toluene and
styrene mass concentrations of the feeding tank (MT0, MS0,
respectively) and polystyrene (dP), styrene (dS) and toluene
(dT) density values. The rheological properties of the poly-

Fig. 1. Specic viscosities as a function of polymer concentration for

typical polystyrene/toluene solutions.

3911

VADD

1
M
1
VOUTLET =
1 T0
MS0
dS
dT


11

3. Results and discussion

Three runs named A, B and C were carried out in the
tubular reactor. During 4 h, the system operation conditions
(temperature, ow, feed concentration) were unchanged for
reaction A. After this period, a step perturbation of the feed
modier concentration was implemented. The tert-dodecylmercaptan (TDM) feed concentration was varied from zero
to 0.001 kmol/m 3, keeping the reactor temperature at 798C
through out the experiment. In reaction B, after attaining
steady-state conditions, simultaneous step perturbations of
the temperature and TDM concentration were implemented,
from 65 to 808C and from zero to 0.0004 kmol/m 3, respectively. Reaction C was similar to reaction B, but serial step
temperature perturbations from 50 to 928C and a TDM
concentration step perturbation from zero to 0.05 kmol/m 3
were performed. Samples were withdrawn every sampling
time from the outlet stream of the tubular reactor, containing
polystyrene, unreacted monomer, modier, initiator and
solvent, and were diluted using toluene for introduction in
to the capillary viscometer.
Table 1 illustrates the small intrinsic viscosity differences
[Dh ] obtained when Eqs. (8) and (9) are compared for the
diluted samples of reactions A, B and C.
Fig. 2 presents results obtained with solutions prepared
with the pure and dry polymer samples and with the original
diluted reactor samples. The pure polymer sample
comprises toluene and styrene with the same proportion as
the original diluted reactor sample, but with puried (dry)

3912

M.P. Vega et al. / Polymer 42 (2001) 39093914

Table 1
 w data
Experimental hsp and M

Table 1 (continued)
 w ; SEC
M
measurements
(g/gmol)

Reaction

[h ], [13]
(dl/g)

c (g/dl)

[h ap],
Eq. (9)
(dl/g)

Dh
(dl/g)

0.12
0.15
0.14
0.14
0.15
0.14
0.14
0.14
0.13
0.14
0.14
0.15
0.15
0.14
0.15
0.14
0.13
0.14
0.12
0.12
0.13

0.2785
0.2896
0.3112
0.2868
0.2604
0.3080
0.2764
0.2888
0.2992
0.3472
0.3184
0.3232
0.2660
0.2916
0.3416
0.3420
0.3052
0.2884
0.2956
0.2900
0.3188

0.12
0.15
0.15
0.14
0.15
0.14
0.14
0.14
0.13
0.14
0.14
0.15
0.15
0.14
0.15
0.14
0.14
0.14
0.13
0.12
0.13

0.00
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.00

17
18
17
18
17
18
17
17
16
18
16
17
16
16
18
16
17
17
15
15
16

840
740
412
054
833
555
540
366
375
268
922
490
925
591
140
774
464
453
781
732
116

0.23
0.23
0.20
0.19
0.18
0.18
0.17
0.16
0.17
0.17
0.17
0.18
0.18
0.16
0.15
0.14
0.14
0.13
0.13
0.14
0.13
0.13
0.13

0.2933
0.2600
0.2860
0.3044
0.3016
0.2776
0.2880
0.2828
0.3076
0.274
0.3228
0.3092
0.3196
0.3032
0.3072
0.3164
0.3020
0.3044
0.3068
0.2904
0.2924
0.3024
0.2804

0.24
0.23
0.20
0.19
0.19
0.18
0.18
0.16
0.17
0.18
0.17
0.18
0.19
0.17
0.15
0.14
0.14
0.13
0.13
0.14
0.13
0.13
0.14

0.01
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.00
0.01
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01

29
28
23
22
21
20
19
20
20
20
19
19
22
22
18
15
12
12
12
12
12
13
12

094
955
140
007
088
646
605
556
363
815
556
777
035
195
926
620
313
245
763
961
968
059
925

0.49
0.35
0.30
0.29
0.26
0.25
0.24
0.23
0.22
0.22
0.20

0.2732
0.2748
0.2544
0.2608
0.2528
0.2412
0.2832
0.2632
0.2544
0.2636
0.2496

0.51
0.36
0.31
0.30
0.27
0.26
0.24
0.24
0.23
0.22
0.20

0.02
0.01
0.01
0.01
0.01
0.01
0.00
0.01
0.01
0.00
0.00

146
96
76
65
62
53
49
47
44
39
37

574
172
518
877
763
980
149
078
514
559
565

Reaction

[h ], [13]
(dl/g)

c (g/dl)

[h ap],
Eq. (9)
(dl/g)

Dh
(dl/g)

0.12
0.07
0.07
0.08
0.06
0.06
0.06
0.06
0.07
0.06
0.06
0.05
0.05
0.05
0.06

0.2432
0.2444
0.2420
0.2476
0.2620
0.2472
0.1368
0.2392
0.2432
0.2472
0.3076
0.2580
0.2556
0.2480
0.2696

0.12
0.07
0.07
0.08
0.06
0.06
0.06
0.06
0.07
0.06
0.06
0.05
0.05
0.05
0.06

0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00

 w ; SEC
M
measurements
(g/gmol)
18 097
7968
5522
4635
3993
4110
4501
4225
3896
4162
4309
4203
4156
4140
4388

polystyrene. It can be observed that results are almost the

same. Therefore, the proposed viscometrical method does
not require the tedious and time-consuming sampling of the
reaction mixture, followed by separation, washing and
drying of the polymer product, steps which have been
frequently employed, prohibiting in-line application of the
method.
Table 2 shows results for two polystyrene samples with
 w : It can be observed that the accuracy of
very different M
the viscometer is about 0.01 dl/g.
Differentiating Eq. (12) with respect to polymer concentration (Eqs. (13) and (14)) and the MarkHouwink Equa w (Eqs. (16)(18)), it can be
tion (Eq. (15)) with respect to M
observed that the relative error of the intrinsic viscosity is of
the same order of magnitude of the relative error of the
concentration detector, but that the relative error of the
 w is larger than the relative error of the polymer concenM
tration, as a is smaller than 1. Therefore, good accuracy is
required for polymer concentration measurements.
hsp
12
hap
c


hsp
Dhap 2 2 Dc
13
c
Dhap
Dc
2
hap
c

14

bhap c KMwa

15

Dhap aKMwa

DMw
Mw

Dhap
DMw
a
ap
h
Mw

16

17

M.P. Vega et al. / Polymer 42 (2001) 39093914

Fig. 2. Intrinsic viscosities of polystyrene samples in reaction C measured

rigorously from dry ( p ) and pure (K) polymer and approximately from the
original reactor samples (W).

DMw
1 Dc
2
a c
Mw

18

 w of polymer samples isolated from reactions A, B, and C

M
were determined by SEC, as can be seen in Fig. 3. These
data were used for estimating the parameters of the Mark
Houwink equation, using intrinsic viscosity information
from Table 1. Eq. (19), for polystyrene in toluene, is the
 w when a quasiequation obtained between h ap and M
Newton non-linear regression method is used [19]. The
parameters of the MarkHouwink equation for polystyrene
w
in solution of toluene at 308C were estimated using M
values ranging from 3800 to 150 000 g/gmol. Results may
be regarded as excellent and adequate for control purposes.
 w values estimated with Eq. (19) are almost always within
M
the error bounds of the SEC method, even when uncertainties of the h sp measuring technique are neglected. Fig. 3
 w evaluation may be based upon
shows very clearly that M
single measurements of h sp, from samples collected directly
from the reactor, without any further purication.
bhap c 68:7 1025 Mw0:55 ; in dl=g

19

According to data provided by Brandrup and Immergut [20],

the intrinsic viscosity of polystyrene in solutions of toluene
Table 2
Accuracy of the measurements
c (g/dl)

[h ap] (dl/g)

0.3987

0.11
0.12
0.12
0.10
0.11

0.4926

1.04
1.07
1.06
1.06
1.07

hap (dl/g)

D[h ap] (dl/g)

Repeatability (dl/g)

0.11

0.00
20.01
20.01
0.01
0.00

0.01

1.06

0.02
20.01
0.00
0.00
20.01

0.02

3913

Fig. 3. Chain length obtained both from SEC (K) and in-line intrinsic
viscosity measurements (W).

at 308C is given by Eq. (20). Although Eq. (20) seems to be

very different from Eq. (19), Fig. 4 shows that Eqs. (19) and
(20) actually lead to very similar values of the intrinsic
viscosity, which means that Eq. (19) provides very good
estimates of [h ] in the range analyzed, in spite of the simplications and simplicity of the single point method. Actually,
as shown by Gomes [11], it is very difcult to estimate K
and a independently in Eq. (10), given the huge correlation
between these two parameters. This is why Eqs. (19) and
(20) lead to very similar results of [h ] in spite of the different parameter sets considered. For most practical reasons,
though, Figs. 3 and 4 show that the single point method
presented here may be used with condence for in-line
monitoring and control of average molecular weights in
solution polymerizations.
h 11 1025 Mw0:71 ; in dl=g:

20

4. Conclusions
It is shown here that the in-line viscometrical method
 w during the
proposed allows the accurate monitoring of M
course of a solution polymerization reaction. The method
depends on the sampling of the polymer solutions and on
independent evaluations of monomer conversion. In spite of
the fact that the concentration of the outlet stream changed
during the reactions performed, the ow time of diluted
polymer samples was dominated by the change in the
chain length. Besides, the linear approximation for calculating intrinsic viscosity using a single concentration value of a
reactor sample provided almost the same results provided by
the single point relation method of Solomon and Ciuta [13].
The in-line viscometrical method was calibrated with the
MarkHouwink equation to allow real time evaluation of
the weight average molecular weight. The simplicity of the
method and short time required for analysis make this technique an extremely practical tool for monitoring and control
of polymer quality in homogeneous reactions.

3914

M.P. Vega et al. / Polymer 42 (2001) 39093914

Fig. 4. Intrinsic viscosity of polystyrene in toluene solutions.

Acknowledgements
We are indebted to CNPq (Conselho Nacional de Desenvolvimento Cientco e Tecnologico, Brazil) for supporting
this research and providing scholarships for authors. We
also thank CAPES (Fundacao de Coordenacao e Aperfeicoamento de Pessoal de Nvel Superior) and FINEP (Financiadora de Estudos e Projetos) for supporting this research.
References
[1] Penlidis A, Chien DCH. JMS-Rev Macromol Chem Phys
1990;C30(1):142.
[2] Kammona O, Chatzi EG, Kiparissides C. JMS-Rev Macromol Chem
Phys 1999;C39(1):57134.
[3] Ray WHJ. J Macromol Sci Revs Macromol Chem 1972;C8(1):156.
[4] Ponnuswamy SR, Saha SL. J Appl Polym Sci 1986;32:3239.
[5] Ponnuswamy SR, Saha SL, Kiparissides C. Ind Engng Chem Res
1987;26:229.

[6] Zisenis M. Eur Polym J 1997;33:77380.

[7] Billmeyer FW. Textbook of polymer science. 3rd ed. New York:
Wiley, 1984 (p. 20813).
[8] Rosen SL. Fundamental principles of polymeric materials. 2nd ed.
New York: Wiley, 1993 (p. 65).
[9] Kraemer EO. Ind Engng Chem 1938;30:1200.
[10] Huggins ML. J Am Chem Soc 1942;64:2716.
[11] Gomes DF. MSc thesis, PEQ/COPPE/UFRJ, Rio de Janeiro, 1999.
[12] Schulz GV, Blaschke F. J Prakt Chem 1941;158:130.
[13] Solomon OF, Ciuta IZ. J Appl Polym Sci 1962;VI-24:6836.
[14] Baruah SB, Laskar NC. Polym J 1996;28:8935.
[15] Viscotek Corporation 2000. http://www.viscotek.com/sasp.html.
[16] Visiologic Corporation 2000. http://www.l-i-d.com/lidpag/instv.html.
[17] Lauda R. Wobser GMBH and CO.KG. 2000. http://www.lauda.de/e/
all-m2-e.html.
[18] Vega MP, Lima EL, Pinto JC. Comput Chem Engng.
1997;21:S104954.
[19] StatSoft Inc. 2325 East 13th Street, Tulsa, OK, USA, 1995.
[20] Brandrup J, Immergut E. Polymer handbook. 2nd ed. New York:
Wiley, 1989 (p. IV-12).