Intrinsic Viscosity of Polystyrene

Y. EINAGA, Y. MIYAKI, and H. FUJITA, Department of Polymer Science,

Osaka University, Toyonaka, Osaka 560, Japan

Synopsis

Zero-shear-rate intrinsic viscosities [77]0 in benzene a t 25OC and in cyclohexane a t 34.5OC were
measured for two series of polystyrene samples: six fractions ranging in Mw(weight-average mo-
lecular weight) from 8.8 X lo6 to 5.7 X lo7 and l l “monodisperse” samples ranging in Mu from 4
X lo3to 4.8 X lo6. The results, combined with typical literature data, yielded

[?lo = 7.8 X 10-5m0,75*0.0’ (4 X lo5 < Mw< 6 X lo7, benzene at 25OC)

[?lo = 8.8 X 10-4R0,50 (lo3 < Mu < 6 X lo7, cyclohexane a t 34.5OC)

These relations cover the broadest range of molecular weights reported so far in the literature on
synthetic polymers. In contrast to Slagowski et al., we observe no “anomaly” in the molecular weight
dependence of [ 9] 0for benzene solutions of very high molecular weight samples (>lo7).

INTRODUCTION
The recent success of Japan Synthetic Rubber Co. in the commercial pro-
duction of ultrahigh molecular weight samples of polystyrene has offered an
opportunity for investigating dilute solution properties of this typical linear
polymer at molecular weights above lo7. Thus, in our recent study,l six poly-
styrene fractions ranging in Zfw(weight-average molecular weight) from about
9 X lo6 to 6 X lo7 were prepared and their light scattering behavior in benzene
a t 25OC and in cyclohexane at 34.5”C (a theta solvent for polystyrene) was in-
vestigated in detail. The study reported in the present paper was undertaken
to determine the dependence of the zero-shear-rate intrinsic viscosity [17]0 on
molecular weight M for polystyrene in these two solvent systems with a series
of samples including our six ultrahigh molecular weight fractions. For these
samples M ranged from 4 X lo3to 6 X lo7. Thus the [17]0 vs. M relations deter-
mined covered the broadest range of M so far attained in the viscosity mea-
surements on synthetic polymers. Slagowski et aL2p3observed for polystyrene
in benzene that [17]0 for samples with M above lo7 increased in proportion to
M1.lf0.l. The exponent implies that the polystyrene molecule in benzene tends
to be rodlike as M exceeds a certain limit. One of the motivations to the present
study was to check whether this seemingly anomalous finding was correct.

Thomas G Fox Memorial Issue

Journal of Polymer Science: Polymer Physics Edition, Vol. 17,2103-2109 (1979)
0 1979 John Wiley & Sons, Inc. 0098-1273/79/0017-2103$01.00
2104 EINAGA, MIYAKI, AND FUJITA

EXPERIMENTAL
Polymer Samples
The “principal” samples of the present study were the six polystyrene fractions
of ultrahigh molecular weight which had been separated from JSR products
IK1500 and BK2500 for our previous light scattering measurements. Their
codes, weight-average molecular weights, and polydispersity indices are given
in ref. 1. In addition to these, 11“monodisperse” polystyrene samples of lower
molecular weights were chosen a t appropriate intervals of M down to oligomers.
They were Toyo Soda’s standard polystyrene F4, F4-b (prepared from F4 by
fractionation), F126, F288, and F450 and Pressure Chemical’s la, 3a, 4a, 2b, 8b,
and l l b , ranging in molecular weight from about 4.8 X 106 to 4 X lo3. The
supplier’s values of their Mu were accepted as reliable, though we confirmed some
of them by light scattering.
The solvents, benzene and cyclohexane, were purified according to standard
procedures.
Test solutions of the principal samples were carefully made up without vig-
orous mechanical stirring, according to the procedure described in ref. 1. So-
lutions of the lower molecular weight samples were prepared in the usual way
by stirring.

Viscome try
Suzuki e t al.4 have investigated in detail non-Newtonian flow of polystyrene
fractions in benzene at 25OC and in cyclohexane a t the theta temperature. Ac-
cording to their experimental results, [s]may be equated to the value at zero shear
rate if the measurement is made a t a reduced shear rate 0 = qs[ s ] & f G / R below
T
about 0.3, where ss is the solvent viscosity, (1710 is the zero-shear-rate intrinsic
viscosity, M is the molecular weight of the polymer solute, G is an apparent shear
rate defined according to the viscometer used, and R T has the usual meaning.
We used this information as a guide for choosing a viscometer useful for esti-
mating [ q ]for +
~ each polymer sample solvent pair.
Thus for the samples with M I1.34 X lo6 in benzene and those with M I4.8
X lo6 in cyclohexane, a conventional capillary viscometer of the suspended-level
type was used and no correction for shear-rate effects was applied to the viscosity
values obtained. For the samples of M 1 3.1 X lo6 in benzene, use was made

Fig. 1. Dependence of (In q r ) / c and qSp/con the apparent rate of shear C for fraction IK1500-1
in benzene at 25OC. Polymer concentrations of solutions A and B are 1.476 X lo-‘ and l.l& X lo-‘
g/cm3, respectively.
 INTRINSIC VISCOSITY OF POLYSTYRENE 2105

I---- IK 1500-1

- I

40 k BK2500-:

50 G / s e c ‘ 100 I
3

Fig. 2. Shear-rate dependence of ?,/c for the five highest-molecular-weightfractions in cyclo-

hexane at 34.5OC. The polymer concentrations in g/cm3 are 2.862 X for IK1500-1,4.641 X
for IK1500-2, 4.493 X for BK2500-1, 6.775 X for BK2500-2, and 6.472 X for
BK2500-3.

of a rotational viscometer of the Zimm-Crothers type constructed in our labo-

ratory and described el~ewhere.~ As will be shown below, it was found that even
for the highest-molecular-weight sample (M, = 5.7 X lo7) the value of [ q ]in~
benzene could be determined with this instrument if the measurement was made
a t a shear rate of about 1 sec-l. Therefore, the viscosities of all these high-
molecular-weight samples in benzene were measured at one or two shear rates
around 1sec-l and equated to those at zero shear rate.
The use of our rotational viscometer for the “principal” samples (with M 1
8.76 X lo6) in cyclohexane had to be given up, because rotor stability could not
be maintained in this solvent. Hence we adopted a four-bulb spiral capillary
viscometer,6 and the viscosities determined with the four bulbs were extrapolated
graphically to zero shear rate. This viscometer consisted of a 150 cm3 liquid
reservoir, a spiral capillary of 138 cm length and 0.43 mm inner radius, and four
bulbs of different capacity aligned one above the other. The lowest shear rate
attained in this instrument with cyclohexane at 34.5”C was about 20 sec-l.
Apparent shear rates G needed for extrapolating the data for the four bulbs to
zero shear rate were calculated according to the expression

where a and 1 are the inner radius and length of the capillary, respectively; K is
the height of the midpoint between the upper and lower marks for each bulb
above the meniscus of a liquid column that has ceased to descend; p is the density
of the liquid; and g is the acceleration of gravity. In our instrument, the values
of & were 8.67,5.94, 3.46, and 1.28 cm from the top to the bottom bulb.
2106 EINAGA, MIYAKI, AND FUJITA

RESULTS AND DISCUSSION

Shear Rate Effects

In Figure 1,the values of (lnq,)/c and qsp/c (7, is the relative viscosity, qSpis
the specific viscosity, and c is the mass concentration of the solute) at two con-
centrations of the highest-molecular-weight fraction IK1500-1 in benzene at 25°C
are plotted against the apparent shear rate G. A t each concentration the data
points are essentially independent of G in the region below 2 sec-l and permit
the value of either (lnq,)/c or qs,/c at zero shear rate to be extrapolated with
accuracy. On the basis of this finding, the measurements on other samples by
the rotational viscometer were made at one and sometimes two values of G
around 1sec-l and the viscosity values obtained were equated to those at zero
shear rate.
Figure 2 illustrates the shear rate dependence of vSp/cfor cyclohexane solutions
of the five fractions of highest molecular weight at the highest concentrations

c / 1 0 - ~~g r n - ~
Fig. 3. Determination of intrinsic viscosity [ d o for the principal polystyrene fractions in benzene
a t 25°C. From top to bottom, the lines for each fraction correspond to the Huggins plot, the Billmeyer
plot, and the Fuoss-Mead plot. Points are marked to indicate different series of measurements.

c / I ~ - ~ ~r n g- ~
Fig. 4. Huggins plots (open circles) and the Fuoss-Mead plots for the principal polystyrene
fractions in cyclohexane a t 34.5OC. The errors in the graphical extrapolation are less than fl%.
 INTRINSIC VISCOSITY OF POLYSTYRENE 2107

TABLE I
Viscosity Data for Polystyrene in Benzene at 25OC and in Cyclohexane at 34.5OC
Benzene, 25OC Cyclohexane, 34.5OC
Samples Mw x 10-6 [1]0/102cm3g-1 k' [q]0/1O2cm3g-* k'
IK1500-1 56.8 53.06 f 0.5 0.34 f 0.01 6.67 f 0.07 ...
IK1500-2 39.0 40.07 f 0.4 0.34 f 0.01 5.64 f 0.06 ...
BK2500-1 31.9 35.08 f 0.4 0.33 f 0.01 5.22 f 0.05 ...
BK2500-2 23.2 28.02 f 0.3 0.28 f 0.02 4.35 f 0.04 0.50 f 0.05
BK2500-3 15.0 20.04 f 0.2 0.34 f 0.01 3.50 f 0.03 0.50 f 0.05
BK2500-4 8.76 13.02 f0.1 0.35 f 0.01 2.59 f 0.03 0.50 f 0.04
F450 4.80 8.25 f 0.1 0.32 f 0.01 1.92 f 0.02 0.50 f 0.04
F288 3.08 6.00 f 0.05 0.30 f 0.01 1.55 f 0.02 0.50 f 0.04
F126 1.34 3.20 f 0.03 0.32 f 0.01 0.990 f 0.01 0.50 f 0.02
F4 0.0425a 0.266 f 0.003 0.38 f 0.01 . .. ...
F4-b 0.0470 ... ... 0.184 f 0.002 0.62 f 0.02

3a 0.411a 1.36 f 0 . 0 1 0.33 f 0.01 0.550 f 0.004 0.55 f 0.02

la 0.1608 0.68s f 0.007 0.35 f 0.01 0.347 f 0.003 0.52 f 0.02
4a 0.0972a 0.472 f 0.005 0.38 f 0.01 0.264 f 0.002 0.56 f 0.02
2b 0.0204a 0.159 f 0.002 0.46 f 0.02 0 . n 3 f 0.001 0.66 f 0.02
8b 0.0108 0.10~f 0.001 0.55 f 0.02 0.089 f 0.001 0.71 f 0.03
llb 0.00408 0.060 f 0.001 0.62 f 0.03 0.056 f 0.001 0.71 f 0.03
a Values given by the suppliers (others were determined directly by light scattering).

measured. For fraction BK2500-3the shear rate effect is virtually absent in the
range of G indicated. However, this is not the case for the higher-molecular-
weight fractions. The data for the three highest fractions exhibit no horizontal
approach to zero shear rate, but their graphical extrapolation to zero shear rate,
as indicated, should give the desired value of ( ~ , p / cwith
) ~ an error less than
fl%.

Viscosity Plots for the Principal Samples in Benzene and in

Cyclohexane
Figure 3 shows Huggins plots, Billmeyer plots, and Fuoss-Mead plots con-
structed from the zero-shear-rate viscosities of the six principal polystyrene
fractions in benzene at 25OC. Different symbols here indicate the results from
different series of dilution. For each fraction the common intercept of the three
plots, i.e., [&, can be determined with precision. It is seen that the Billmeyer
plot is particularly advantageous for the extrapolation to infinite dilution.
Except for fraction BK2500-2, the Billmeyer plots are essentially horizontal.
This means that the Huggins slope constant k' for polystyrene in benzene should
be close to 1/3, even in the range of ultrahigh molecular weights, as is usually
found at lower molecular weights.
In Figure 4, the corresponding viscosity plots for the cyclohexane solutions
are shown, with the Billmeyer plots omitted for the sake of simplicity. Unlike
the plots for the benzene solutions, the Huggins plots here are curved upward
rather than linear. The curvature becomes systematically greater as the mo-
lecular weight of the sample is increased. The corresponding Fuoss-Mead plots
are also nonlinear, though less markedly.
2108 EINAGA, MIYAKI, AND FUJITA

-2
3 4 5 6 7 8
log mw
Fig. 5. Relation between zero-shear-rate intrinsic viscosity [9]0and weight-average molecular
weight m,,,for polystyrene in benzene a t 25 or 30°C and in cyclohexane a t the theta temperature.
( 0 , Opresent
) data; (d,4)Altares e t al. (ref. 8); *Berry (ref. 9); (?,T)Fukuda e t al. (ref. 7); @,
+) Yamamoto et al. (ref. 10).

Viscosity vs. Molecular Weight Relations

Table I summarizes the values of 11710 and k’ for all the polystyrene samples
in the two solvents, together with their mu
values. Conventional log-log plots
for [17]0 as a function of Mu constructed from these data and from some literature
data7-10are illustrated in Figure 5. Different sets of data are distinguished by
different symbols.
All the plotted points for the cyclohexane solutions fall on a straight line with
a slope 0.50 and are best represented by the equation
[77]0 = 8.8 X 10-4%L50 (2)

I 3 5 7 9 II
as3
Fig. 6. Plots of the Flory viscosity factor + vs. expansion factor a,”for polystyrene in benzene.
0 present data; 9 Fukuda et al. (ref. 7); 0 Yamamoto et al. (ref. 10). The open (present data) and
closed (Yamamoto et al.) semicircles on the ordinate axis (a:= 1)represent the @values for cyclo-
hexane solutions a t the theta temperature.
 INTRINSIC VISCOSITY OF POLYSTYRENE 2109

The data points for the benzene solutions first follow a slightly bent curve and,
beyond an M w of about 4 X lo5, they follow a straight line with a slope 0.75
(fO.O1) as the molecular weight increases. This linear portion of the plot is
represented by the equation
[7110 = 7.8 X 10-5MO;75*0.0’ (3)
It is seen that our data for polystyrene in benzene exhibit no such “anomaly”
as was reported by Slagowski et al.,233who found an upswing of the log[q]o vs.
logx, plot a t Mu above lo7. It is hard to conceive that the straight line fitting
our benzene data might bend either upward or downward at still higher molecular
weights. Very recently, Daoud and Janninkl’ predicted from semitheoretical
considerations that the asymptotic slope of log[q]o vs. logMwplots in good sol-
vents would be 0.74. However, it is advisable to consider the agreement with
our experimental value as merely accidental.
If, as was assumed originally, the factor @ in the Fox-Flory viscosity equa-
tion12
[q]= 63/2CP(S2)3/2/M (4)
is a universal constant, the asymptotic slope of log[q]o vs. logMwfor flexible linear
polymers in good solvents should be 0.8. This is because ( S2) is proportional
to M1.2 at large M, as has been deduced from recent theoretical consider-
a t i o n ~ and
~ ~has
- ~also
~ been substantiated by our recent light scattering mea-
surements.l This prediction does not agree with the observed slope 0.75,
+
suggesting that @ for polystyrene benzene at 25°C varies as Md0.O5. In Figure
6, values of CP for this system are plotted against a:, where a, is the linear ex-
pansion factor defined as the root-mean-square radius of gyration in benzene
relative to that in cyclohexane at 34.5OC. The data of Fukuda et al.7 and ours
appear to be fitted by a single curve which fall monotonically as a: increases.
The two points on the ordinate represent the @ values computed from our ex-
perimental data and those of Yamamoto et for the theta solvent cyclohexane.
They are far below the theoretically predicted value of 2.87 X mole-l for
impermeable unperturbed polymer coils.16

References
1. Y. Miyaki, Y. Einaga, and H. Fujita, Macromolecules, 11,1180 (1978).
2. E. L. Slagowski, Ph. D. thesis, University of Akron, Akron, Ohio, 1972.
3. D. McIntyre, L. J. Fetters, and E. Slagowski, Science, 176,1041 (1972).
4. H. Suzuki, T. Kotaka, and H. Inagaki, J . Chem. Phys., 51,1279 (1969).
5. Y. Einaga, Y. Miyaki, and H. Fujita, J . SOC.Rheol. Jpn., 5,188 (1977).
6. J. H. Larkins, J. D. Perrings, G. R. Shepherd, and B. J. Noland, J. Chem. Education, 42,555
(1965).
7. M. Fukuda, M. Fukutomi, Y. Kato, and H. Hashimoto, J . Polym. Sci. Polym. Phys. Ed., 12,
871 (1974).
8. T. Altares Jr., D. P. Wyman, and V. R. Allen, J . Polym.,Sci., Part A, 2,4533 (1964).
9. G. C. Berry, J . Chem. Phys., 46,1338 (1967).
10. A. Yamamoto, M. Fujii, G . Tanaka, and H. Yamakawa,Polym. J., 2,799 (1971).
11. M. Daoud and G. Jannink, J. Phys., 39,331 (1978).
12. T. G Fox and P. J. Flory, J . Phys. Colloid Chem., 53,197 (1949).
13. S. F. Edwards, Proc. Phys. SOC.,85,613 (1965).
14. P. G. de Gennes, Phys. Lett. A, 38,339 (1972).
15. C. Domb and A. J. Barrett, Polymer, 17,179 (1976); see also references cited in ref. 1.
16. H. Yamakawa, Modern Theory of Polymer Solutions, Harper & Row, New York, 1971.

Received February 22,1979