Dependence of the Zero Shear Melt Viscosity and the Related Friction Coefficient

and Critical Chain Length on Measurable Characteristics of Chain Polymers

Thomas G Fox and Vernon R. Allen

Citation: The Journal of Chemical Physics 41, 344 (1964); doi: 10.1063/1.1725872
View online: http://dx.doi.org/10.1063/1.1725872
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/41/2?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Dependence of the viscosity on the chain end dynamics in polymer melts
J. Chem. Phys. 118, 10287 (2003); 10.1063/1.1572458

High-pressure cell for measuring the zero-shear viscosity of polymer melts

Rev. Sci. Instrum. 68, 2101 (1997); 10.1063/1.1148076

Chain length dependence of the polymer–solvent critical point parameters

J. Chem. Phys. 105, 802 (1996); 10.1063/1.471889

Concentration dependence of friction coefficients for polymer chains in solution

J. Chem. Phys. 78, 2051 (1983); 10.1063/1.444953

Rheological Properties of Molten Polymers. III. ShearDependent Melt Viscosity

Trans. Soc. Rheol. 13, 335 (1969); 10.1122/1.549143

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 THE JOURNAL OF CHEMICAL PHYSICS VOLUME 41, NUMBER 2 15 JULY 1964

Dependence of the Zero Shear Melt Viscosity and the Related Friction Coefficient and
Critical Chain Length on Measurable Characteristics of Chain Polymers*
THOMAS G Fox AND VERNON R. ALLEN
Mellon Institute, Pittsburgh, Pennsylvania
(Received 4 March 1964)

Available 1/-Z-T data for a wide variety of well defined linear and branched polymers and polymer-
diluent mixtures are well represented by the single equation 1/=4.8X1()8(X/4.7XlO-IG)ar, a=3.4 for
X~4.7X1O-16, a= 1.0 for X :54.7X 10-15. Here X = «SO')Ay/V a)cPl, where (SO'lAY is the unperturbed mean-
square radius of gyration of the macromolecule, Va the volume per chain atom, and cPl the volume fraction
of polymer; values of the friction factor per chain atom r, computed from viscosity data by this equation,
agree with independently determined values in the literature. This result confirms Bueche's theory for the
viscosity of short chains and supports his theory for the viscosity of long chain macromolecules. An attempt
to reduce roT data for different systems to a single curve by introduction into the WLF equation of the
term E/ RT, where E may be related to the rotational barriers within the chain, is also presented.

INTRODUCTION given temperature, expected to be constants charac-

teristic of the polymer type but independent of Z. The
S TRONG confirmation of the equation of Bueche
for the zero shear viscosity of short polymer chainsl dependence, if any, of Zo, Xc, and Ze on chain structure
and on the reduced chain dimensions is unknown.
7]=tNO[(S02)Ay/M) (Zw/v)Jt (1) Expressions virtually identical with Eqs. (2) may
was provided in the companion paper2 by the success- also be deduced by simultaneous solution of Eq. (1)
ful computation of this coefficient for short chain poly- and Bueche's theoretical equation for the zero shear
styrenes (Z=S;Zc) from the independently measured viscosity of long chains
values of the pertinent molecular parameters. Con- =! N02[( (so2 )AV)2.5 Z3.5 ]
(3a)
versely, the theoretical equation facilitates precise de- 7] 8 6t M (m a )!v2Zlf(X) t,
termination from viscosity data of the frictional coeffi-
cient per chain atom t and of the critical chain length Here
Z. above which the Bueche equation for short chains
fails.
Ze[j(X)]i=[6No2ma~~~M)3Zc7. (3b)
A general equation for 7] (obtained by simultaneous
solution of Eq. (1) and the empirical flow law, 7]= Evidence was presented in the preceding article2 sug-
KTZw3.4, for long chains) applicable to polymer homo- gesting the applicability of (3b) for polystyrene.
logs of any length was also proposed2 : The 7]- T -Z data for polystyrene fractions and blends2
are represented by Eq. (2a). The friction coefficient
7] = ":o[((S;iAV)~C](~:rt, is given approximately by a WLF type equationS
10gt=logtu-[cl(T- Tg)/(C2+ T- Tg) ] with4
a=3.4 for Zw"?,Zc, (4)
a= 1.0 for Zw=S;Zc. (2a) where To."", and Kg are constants characteristic of the
given homologous polymer series.
Alternatively, designating X= «so2)Av/M) (Zw/v) and To provide additional test of Bueche's theories, and
X.= «S02)AV/M) (Zclv) , Eq. (2a) may be written to obtain values of Zc (also Xc and Ze) and of t for
many different polymers, we now fit to the above
7]=tNoXc(X/Xc)"t,
equation available 7]-Z- T data for various well-defined
a=3.4 for X"?, Xc, polymeric systems. Simple equations of broad appli-
cability, relating Zc, t, and 1] to experimentally assess-
a=1.0 for X=S;Xc. (2b) able parameters, for long and short chain linear poly-
The values of Zc and Xc (and of Z. below) and, for mers, for branched polymers, and for polymer-diluent
sufficiently long chains, of (S02)Av/M, v, and t are, at a mixtures are established thereby.
GENERAL PLAN
* This work was supported in part by the Office of Naval Re-
search under Contract No. Nonr 2693(00), and by the Aeronau- Here we propose to fit Eq. (2) to available viscosity
tical Systems Division, Air Force Systems Command, Wright- data for different polymeric systems, and to evaluate
Patterson Air Force Base under Contract No. AF 33(657)-10661.
1 F. Bueche, J. Chem. Phys. 20, 1959 (1952); Physical Proper- 3 M. L. Williams, J. Phys. Chem. 59, 95 (1955); M. L. Williams,
ties of Polymers (Interscience Publishers, Inc., New York, 1962). R. F. Landel, and J. D. Ferry, J. Am. Chem. Soc. 77, 3701 (1955).
'V. R. Allen and T. G Fox, J. Chem. Phys. 41, 337 (1964). • T. G Fox and S. Loshaek, J. Polymer Sci. 15,371 (1955).
344

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 ZERO SHEAR MELT VISCOSITY 345

Zc and r from data on the measurable molecular char-

acteristics Zw, v, and (S02 /AV/ M. Tests of the applicability
of Eq. (2) will be the ability to reproduce the 7J-Z-T 4
data satisfactorily, and agreement between the values
of r derived here and the corresponding literature val-
3
ues obtained from viscoelastic relaxation spectra. Fi-
nally we seek expressions of general applicability relat- c
ing rand Zc (hence Xc and Ze) to other readily measured ~ 2
characteristics of the structure and composition of the ~
a.
macromolecular system. x
., 1
The computation of r by means of (1) from viscosity «If...
data for Z<Zc is straightforward. In this range, gen- 1.8
erally, r increases with Zn towards an asymptotic limit .,.~o
which in some instances (coincidentally) may be ap- .3
proached only in the vicinity of Zc. 2 Unfortunately, -I
positive deviations from Eqs. (2) are generally found
in this transition region, rendering difficult the precise
determination here of the limiting value of r for long -2
chains. Of course, it can be determined readily by
Eq. (2) from 71 at Z> Zc provided the value of Zc is
known. Hence, we undertake first, in the following
section, to evaluate Zc for different polymeric materials
of in teres t. FIG. 2. Reduced viscosity function IOg7}-AITn exp( -J3IZn)
versus logZ for polyisobutylene 0, polydimethyl siloxane 0,
polyvinyl acetate 8, anionic .... and free-radical i'::,. polystyrene.
CRITICAL PARAMETERS Zc AND Xc (Data sources given in Fig. 1).
Two methods (Figs. 1 and 2) are used to evaluate
Zc for the four homopolymers in Table I for which polymer of Z<Zc of viscosity 71 and rearranging, we
most extensive 7J-M data are available. have
The first method of determining Zc from viscosity
data is based on Eq. (2). Writing Eq. (2) for a refer- 3.410gZ*-log7J*+log7J/Z= 2.410gZc-log(r*/r). (5)
ence polymer of Z*> Zc and of viscosity 71*, and for a
A plot of the left side of Eq. (5) vs 10gZ, for a series
of polymers of Z <Z*, will be a horizontal line of
height equal to 2.4 10gZc, if Z<Zc and r=r*. For
Z> Zc, higher values of the ordinate increasing with
increasing Z will obtain; below a certain range r*/r
N
..... will be > 1, and lower values of the ordinate will
£10.0
result with decreasing Z. The value of Zc can be com-
.3'" puted from the magnitude (2.4 logZc) of the inter-
+
*1='" mediate horizontal line; as a further check, the line
.3'" representing the ordinate for Z>Zc, of slope 2.4, should
. 1
intersect the horizontal line at Z=Zc .
~ 8.0
Plots conforming to Eq. (5) are shown in Fig. 1 for
.3'" several polymers for which the appropriate data are
"It
r<l available. Generally, a value of Zc (Table I) can be
determined within ± 10% from the height of the hori-
6.0
zontal region of the plot, or, with less precision, from
the point of intersection of this line with the line of
positive slope obtained for higher molecular weights.
A second method is based on the use of empirical
expressions5 for the dependence of the viscosity-tem-
FIG. 1. Reduced viscosity function 3.4 logZw*-log'l*+log
perature coefficient on T and Zn to "correct" the ob-
('lIZ) versus JogZ for anionic polystyrene ... (Ref. 2), free-radical served viscosities for the effects of change in density
polystyrene i'::,. (Ref. 19), polyvinyl acetate 0 (Ref. 6), poly-
isobutylene °
1951)J, and polydimethyl siloxane °
[T. G Fox and P. J. Flory, J. Phys. Chern. 55,221
[A. Barry, J. App!. Phys.
with changes in these variables. Plots of the "corrected"
log7J vs logZ should then yield two straight lines of
17, 1020 (1946); E. L. Warrick, W. A. Piccoli, and F. O. Stark,
J. Am. Chern. Soc. 77, 5017 (1955); D. J. Plazek, W. Dannhauser, 6 T. G Fox, S. Gratch, and S. Loshaek, Rheology, edited by
and J. D. Ferry, J. Colloid Sci. 16, 101 (1961)]. (The arrows F. R. Eirich (Academic Press Inc., New York 1956), Vol. 1,
clesil9late shifts in the ordinate scale of the indicated magnitude.) Chap. 12.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 346 T. G FOX AND V. R. ALLEN

TABLE I. Values of Zc and of Xc for various polymers.

[d In( (S02)AV/MV)] Computation of Xc

-100
dT ( (S02)Av/M)
Polymer Zc' T(OC) %/l00ob XI0-1sc X c X1O-16d

Z. Values from Fig. 1

Polyisobutylene 460 25°-217° 5··! 10.5 4.3/25°'
Polydimethyl siloxane 630 25° 7- 8.9 5.4/20°4
Polystyrene 600 160°-217° 16 h 7.8 4.7/160°19
Polyvinyl acetate 570 120°-160° 30 i •i •k 6.8 4.4/1200n

Z. Values from log" vs log Z

Polyethylene 275 140° 17! 18.8 4.1/140°4
Polydecamethylene sebacate 290 109° 171 4.9/25°0
Poly-t-caproamide 324 253° 15.5 1 5.0/25°
Polyoxypropylene glycol 400 25° 8-111· m (3.2-4.4)/25°
Polydecamethylene adipate 280 109° 101 2.7/25°0
Polymethyl methacrylate 210 140° 6- 6.2 1. 5/140°4

a The first four values are from Fig.!. The remainder are from Ref. 7 except for the estimates made for polyoxypropylene glycol m and polyethylene [from A. K.
Doolittle, J. Appl. Phys.22, 1031 (1951); A. K. Doolittle and R. H. Peterson, J. Am. Chern. Soc. 73, 2145 (1951); H. L. Tung, J. Polymer Sci. 46, 409 (1960); and
W. L. Peticolas and J. M. Watkins, J. Am. Chern. Soc. 79, 5083 (1957)J.
b Values of the temperature dependence of «so' )Av/M) for polyisobutylene, polyethylene, and polydimethyl siloxane are based on relatively precise modulus-
temperature measurements;!" for the others "best" values were computed from (K/¢»"'/6 using ¢>=2.1XIO" and values of K at different temperatures taken, or
estimated from the indicated literature. The estimate for polyoxypropylene glycol is based, in part, on values l for polyethylene oxide and polypropylene oxide. The
v-T data were taken from the indicated sources in Column 6 (see d below).
C Computed at temperature noted in Column 6; source references given here or in Column 4.

d Computed from ({SQ')Av/M) ·Z./., with values of. from the indicated literature. For polY-E-caproamide and polyoxypropylene glycol v was taken as unity.
• T. G Fox, Polymer 3, 111 (1962).
fA. Ciferri, C. A. J. Hoeve, and P. J. Flory, J. Am. Chern. Soc. 83,1015 (1961).
• P. J. Flory (private communication).
h T. G Fox and P. J. Flory, J. Am. Cbem. Soc. 73, 1915 (1951).
i See Ref. 15.
i A. Shultz, J. Am. Chern. Soc. 76, 3422 (1954).
k M. Matsumoto and Y. Ohyanagi, J. Polymer Sci. 46, 441 (1960); SO, SI (1960).
1M. Kurata and W. H. Stockmayer, Advan. Polymer Sci. 3, 196 (1963).
m A. J. Havlik and J. Moacanin, Research Summary No. 36-10, Vol. I, p. 92, Jet Propulsion Laboratory, Pasadena, California, 1 September 1961.
n P. Meares, Trans. Faraday Soc. 53,31 (1957).
o P. J. Flory, J. Am. Chern. Soc. 62, 1057 (1940).

slope 1.0 and 3.4, intersecting at Z = Zc. For this pur- were employed to fit the data for 10g7] vs 10gZ in the
pose we employ an empirical expression of the form low molecular weight range; i.e., the point of inter-
section with the line for the high molecular weight
range was influenced by the rate of decrease of t with
decreasing Zn. We believe the present method, wherein
Here, A, n, and ~ are characteristic of the homologous compensation is made for this change of t with Z,
series. Data for various polymers treated in this way yields values of Zc of more theoretical significance.
are plotted in Fig. 2; in each case, the data could be Values of Zc for six other polymers are also listed
represented by straight lines of slopes 1 and 3.4, with in Table 1. Those numbers, determined from log-log
the point of intersection defining Zc. The numbers plots of 7] vs Z, would be little changed (except, per-
thus obtained agree, within the estimated experimental haps, for polY-E-caproamide) if the data were plotted
uncertainty (± 10%), with the corresponding values by the methods of Figs. 1 or 2, since here the coefficient
obtained in Fig. 1. (d In7]/dT) is relatively small and insensitive to Z, and
The values of Zc obtained in Fig. 1 are lower than the slope of 10g7] vs 10gZ for Z<Zc is not markedly
the corresponding numbers determined earlier from the different from unity. The data, however, are not suffi-
same data. 6 •7 Previously lines of slope greater than one ciently extensive to define Zc within the precision
(±1O%) cited above. For poly-(methylmethacrylate)
6 H. Nakayasu and T. G Fox, Colloid Chemistry Division the value of Zc is extrapolated from that found for 25%
Abstracts, 137th Meeting of The American Chemical Society,
Cleveland, Ohio, April 1960.
solutions in diethyl phthalate. s
7T. G Fox and S. Loshaek, J. App\. Phys. 26,1080 (1955). 8 F. Bueche, J. Appl. Phys. 26,738 (1955).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 ZERO SHEAR MELT VISCOSITY 347

Differences in excess of the above uncertainties exist

between values of Zc for different polymers in Table I.
For example, Zc values for polydimethyl siloxane, poly-
styrene, and polyvinyl acetate (~600) are more than
twice that (~280) found for polyethylene and for the
two paraffinlike polyesters.
Conceivably, differences in the values of Zc for dif-
ferent polymers are related to differences in some
measurable characteristics of their chain structures.
In seeking such a relation, we compute (Table I)
values of Xc= «so2)AvIM) (Zclv) for different polymers.
For the first seven polymers in Table I the values of 'o'---C5*o;---"'oo(n---';;t5o;;----;;2,;;;oo,---"'25~O---,3dooc=----.d350
Xc agree, within the estimated experimental uncertainty
(±lS% for the first four and ±20% for the next FIG. 3. -logi" versus (T- To) for polyisobutylene 0, poly-
three)9 and average (4.7±OA) X10-1s ; furthermore, vinyl acetate E:J, polystyrene 8, and polydimethyl siloxane
there is no discernible trend in the values of Xc with O. (Data sources given in Fig. 1). Corresponding values of logt
from relaxation spectralO are plotted as filled symbols.
decreasing (S02)Av/M, though values of the latter
vary by nearly a factor of 3. Lower values of Xc are
computed for the last three polymers; these values are, For most polymers the increase in v with temperature
however, less certain. 9 Although more precise studies is ~(S%-10%)/100°, and [d(ln(so2)Av/M)/dT] is
may establish variations of Xc with structure, for the negative. The latter coefficient, however, may be zero
present we hypothesize that Xc is, approximately at least, or even positive. Values of [-d In( (S02 )A,/Mv) /dT]
a universal constant for flexible chain macromolecular (Table I) correspond to an increase in Zc of ~(S%-
systems, i.e., 30%) /100°, i.e., to an "apparent heat of association"
of ~1Q2 to 103 caljmole. Precise determination (±1O%)
of Zc from 1]-Z plots at two temperatures have been
made in only a few cases and then, excepting for poly-
We can now rewrite (2b) as follows: isobutylene, for only a 50° interval. In no case do
1]= 4.8X 108 (X/4. 7X 10-1S ) "r,
such data indicate a temperature coefficient for Zc in
excess of the corresponding value indicated in I.
a=3.4 for X:2:4.7X10-IS, In the next section we consider the relation between
values of r deduced by application of Eq. (2) to data
a=l.O for X::::;4.7X10-Is. (2d) for various polymers with corresponding values deter-
mined from the observed relaxation spectra. We shall
It appears that for a homologous series 1]"""'X till X then return to further consideration of the influence
attains the critical value Xc, the same for all polymer of chain-coil dimensions and density on the values of
systems, and thereafter, abruptly for longer chains, 1]"-' X3.4. 1], Xc, and Z •.
Equation (2c) is useful for estimating Zc from mo-
lecular parameters when the necessary 1]-Z data for THE FRICTIONAL COEFFICIENT PER CHAIN ATOM, i"
its direct evaluation are not available. This also makes
possible, in such instances, the computation of r by We now compute by Eq. (2) values of r for various
(2d) from a measurement of 1] for a single long chain polymers from available 1]- T -Z data and values of
sample. (S02)Av/Mv (from the numbers in Table I). The results,
Assuming constancy of Xc, we obtain by differenti- represented as the open symbols in the plot of logr vs
ation of (2c) an equation for the temperature depend- T-Tg in Fig. 3, are in good agreement with the cor-
ence of Zc responding measures of r (dark symbols in Fig. 3)
determined independently1° from the observed visco-
d InZc d lnv d[ln ( (S02) A,I M) ] elastic relaxation spectra and application of the Rouse
(2c')
dT dT dT theory.l1 This agreement is again strong evidence for the
validity of Bueche's theoretical Eg. (1) for the zero shear
'Assuming a precision of ±1O% for Zc and for (s02)A,/M, the viscosity of short chains, and of the empirical extension
precision for X.is estimated as ",,±15% for the first four polymers Eg. (2), for long chains.
in Table I. Appreciable additional error may exist in the estimate
of Zc for polyethylene and poly-(methylmethacrylate), of v for In a later section we seek an empirical relationship
poly (E-caproamide) and for poly-(oxypropyleneglycol), and of for predicting r for different polymer systems from
(S02)A,/M for the latter polymer and for poly-(decamethylene-
adipate). Errors may be introduced in three instances by the
computation of Xc using values of (S02)Av/M determined at 25° 10 J. D. Ferry and R. F. Lande!, Kolloid Z. 148, 1 (1956); J. D.
and of Zc determined at higher temperatures; on the other hand, Ferry, Viscoelastic Properties of Polymers (John Wiley & Sons,
values of Zc insensitive to temperature were found for other Inc., New York, 1961).
polymers for which extensive 7J-Z-T data exist. 11 P. E. Rouse, Jr., J. Chern. Phys. 21,1272 (1953).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 348 T. G FOX AND V. R. ALLEN

3.5r---,--,..---,--,..---,-,..---,-r--,---,r---pi>-,
POLY-. -CAPROLACTAM
polymers I3 of functionality p, coincide (excepting at
oT
p the lowest chain lengths), in accord with the above
3.0
"4 suggestion. Here Y., Yw, and Yn are the "z"-, weight-,
"8
2.5
and number-average degrees of polymerization, respec-
tively.
2.0 Data for polystyrene-dibenzyl ether mixtures I4 are
Loq'1 plotted in Fig. 5 as 10g1/ vs 10gCPIZ. At each fixed value
1.5 of CPI, the values of 10g1/ are represented by two inter-
secting lines with the family of lines at different CPI,
1.0
displaced vertically from one anot.her, also in acc~rd
with the above equations. The shIft to lower 1/ WIth
decreasing CPI reflects the decrease in r on the addition
of diluent. More comprehensive evidence on the effect
2.6 2.8 * 3.0 3.2
Log Zv 9 of branching and of diluent on 1/ will be presented
elsewhere. 14 .I5
FIG. 4. Log'7 1400 versus the parameter 10gZ.g* for line~r and
branched polycaproamides. (Data of Schaefgen and Flory. )
CRITICAL CHAIN ENTANGLEMENT LENGTH, Z.

other measurable structural-dependent characteristics The observed dependence of 1/ on gZ for short chains
of the system. We now return to the influence of chain and on (gZ)8.4 for long chains for linear (g= 1) and
dimensions and density on 1/, Zc, Xc, and Z •. branched (g< 1) homologs is in accord with Bueche's
theoretical Eqs. (1) and (3a) , and with his recent
extension of these to branched chains,16 A more criti-
EFFECT OF BRANCHING AND OF DILUENT
cal test of Eq. (3a) requires an independent measure
of I(A), the "interchain slippage" parameter, and of
The term (S02)AV/M) (Z/v) designated here as X Z., the "average number of chain atoms betwe~n cou-
arises in the derivationl of Eq. (1) in the computation pling entanglements." Values of the latter (deSIgnated
of the product of the moment of inertia of the average as A) computed by rubber elasticity theory from the
polymer coil with the number of chains per unit vol- pseudo-equilibrium modulus or compliance or from the
ume. This suggests that for branched polymers, where maximum J mil of the loss compliance have been tabu-
the mean square radius of gyration is smaller by lated for a number of polymers.17 Comparison of these
a factor g than that, (S02 )AV/ M)z, of the correspond- values with those calculated by Eq. (3b) for corre-
ing linear polymer, and in the presence of a diluent, sponding polymers (taking a "universal" value of
where the volume fraction of polymer is CPl, and the
partial specific volume is VI, we should expect X =
[( (S02 )AV/M) I (ZgCPI/VI)]. Inserting this in (2) we get

1/=lNo( (So2)Av/M) 1[(gcptZ) /ihJr X::::;4.7Xlo-I6,

3.0 .,-0.50

(2e)
2.0

1/=~[ (S02)AV) ~r4(gcpIZ)3.4r X;:::4.7Xlo-I6, J

6Xc2.4 M IVI 8'
...s 1.0

(2f)
4.7 X lo-I5VI
(2g)
gCPIZc «S02)Av/M) I '

Of course, r may also be affected by branching or the

addition of diluent, but the effect should be separable
FIG. 5. Viscosity-concentration-molecular weight relations for
from that on X. fractions of polystyrene in benzyl ether (Ref. 14).
Data for linear and branched polY-E-caprolactam12
plotted in Fig. 4 as 10g1/ vs log (gZ) , where g=
13 T. A. Orofino, Polymer 2, 305 (1961).
(3Yw/PYn) - (2y./p 2Yn 2) for these star-type branched 14 V. R. Allen and T. G Fox (to be published).
16 H. Nakayasu and T. G Fox (to be published).
16 F. Bueche, J. Chern. Phys. 40, 484 (1964).
12 J. R. Schaefgen and P. J. Flory, J. Am. Chern. Soc. 70, 2709 17 H. Markovitz, J. D. Ferry, and T. G Fox, J. Phys. Chern.
(1948). 66, 1567 (1962).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 ZERO SHEAR MELT VISCOSITY 349
TABLE II. Comparison of calculated and observed Zc and Z•.

Zc Z,
Observed
Inflection of
Calcd. a from Observed Calcd.b Max. in viscoelastic
Repeating unit Xc=4.7X10-15 (Table 1) from Zc J" properties

n-Octyl methacrylate 1300/110° 630 1320/90°

n-Hexyl methacrylate 800/110° 550 720/105°
n-Butyl methacrylate 1000/110° 570 368/125°
Methyl acrylate 500/60° 350 400/80°
Ethylene oxide 430/25° 250 200/70°
Natural rubber (Hevea) 450/25° 270 272/-50° 120
296
480/-30°
Styrene 600/110° 600/160° 340 320, 610/100°
Vinyl acetate 520/50° 570/120° 340 480/90° 338/50°
Isobutylene 500/25° 460/25° 260 500/50° 240/-30°
320/90°
Dimethyl siloxane 550/25° 630/25° 390 320/25°
Methyl methacrylate 640/140° 210/140° 88 74
88/140°

a Calculated as Z,=4.7XIO-"/«S")Av/MV) with values of «S")Av/Mv) estimated as in Table I.

b Calculated by Eqs. (3-6) with}(A)=1.7, using the observed Zc (Column 3) or its computed values (Column 2) in the other instances.

f(A) = 1.7 as estimated by Bueche)! is afforded by the a broader temperature range, and which enable us
data in Table II. In four of five cases where observed to compute, from measurable material constants, re-
values of Zc are available, they agree well with those quiredlO specific values of its parameters (especially C2)
predicted by Eq. (2c). In these instances and for most for specific materials.
of the other polymers where values of Zc had not been Values of the frictional coefficient, plotted as logr vs
measured but were predicted by Eq. (2c), values of Ze T- Tg in Fig. 3, define characteristic curves for differ-
computed by Eq. (3b) agree with those from visco- ent polymers which lie relatively close to each other,
elasticity measurements within the uncertainty of the but are neither coincident nor "parallel."18 An attempt
latter evaluation. to fit the data to the WLF expression would require
More critical evaluation of Eq. (3) and of the rela- the assignment of a set of different values for each
tion between these measures of interchain interactions polymer for the four parameters-Tg , ry,
C1, and c2-in
in elastic and in viscous deformations awaits more that equation.
extensive and more precise measurements of Zc and A much simpler result would obtain, of course, if
of A for different polymers. the data had fallen on a single curve. It is, therefore,
useful to ask whether by suitable accounting of some
APPROXIMATE (MODIFIED WLF) EQUATION
RELATING I TO MEASURABLE
18 Uncertainties in the values of log I presented at a given value
STRUCTURE-DEPENDENT PARAMETERS
of T-To in Fig. 3 arise from experimental errors in the deter.
mination of 1/ and of Z (±0.02 in log I), from uncertainties in
We seek next an empirical relationship of general the values used for (so2hv/Mv (±0.15) and in its temperature
applicability for predicting !;, over a broad tempera- coefficient (±0.1O/1000), from possible individual variations in
ture range, for different polymer systems from other Xc from system to system (±0.1) or with temperature (0.15/
100°), and from errors in the assignment of To, especially for
measurable structure-dependent characteristics of the short chain polymers and for polyethylene. We estimate the un·
system. We are guided in this quest by experience lO certainty in log I at T- To= 160° to be ±0.2 with a maximum
uncertainty in d log r/dT to be ±0.3/100°. For most of the
with the successful WLF equation; we shall explore polymer pairs the difference in Fig. 3 in the heights and slopes of
modifications of it which extend its applicability to the curves, for T-T.<250° at least, exceed these limits.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 350 T. G FO,X AND V. R. ALLEN

2'-----------------------------------,
. ~ Fi9.6a
thought of as the familiar "fractional free volume at
the glass temperature", and !lar as the expansion co-
,~~ efficient of this fractional free volume. 3 •lo We have
.. 0
4
a • ®
found it expedient to take4.10.19.20 !la,=al-ag, where
Cil~\ \., al and ag are the experimentally measured thermal
expansion coefficients for the polymer system above
!l'"I 6 and below the glass temperature, respectively. The
insertion of a term E/ RT was explored previously.5
The results of attempts to reduce the data of Fig. 3
e 0 __ to a single curve corresponding to Eq. (7) are shown
.. ..,.0.20 ..,. graphically in Fig. 6, and may be expressed in analyti-
4 .. FiQ.6b
cal form as follows:

E [!lar(T-T g) ] (8)
logr= -1.2+ 2.3RT -11.4 0.024+1la,( T- To) .

In Fig. 6(a) a plot of logr vs !la, ( T- To) (open

, symbols) brings the data more nearly in coincidence
""co than the plot of logr vs (T- To) in Fig. 3. The further
';'e
reduction in Fig. 6(b) by the subtraction of E/ RT
from the ordinate, using the values of E in Table III,
brings the data (filled symbols) for several polymers
10 into approximate coincidence with the solid straight
line of that figure, corresponding to Eq. (8). Here
0.2 values of E to achieve the optimum coincidence for
all four polymers were arrived at by trial and error.
0.025 + Aap-Tq )
Thus the seemingly justifiable modification of the
WLF relationship attempted here is partially success-
FIG. 6. (a): -log! versus 1l00/(T-T.) for polyisobutylene 0,
polyvinyl acetate El, polystyrene 8, and polydimethyl siloxane ful. We are able with the introduction of one arbitrary
O. (References to viscosity data in Fig. 1.) Values of logt from parameter (E) to reduce the r-T data for several
relaxation spectra10 for corresponding polymers are plotted as polymers to a single curve. However, we found no
filled circles, and for other polymers as polyethyl methacrylate
(circled E), polyurethane (circled U), polymethyl methacrylate way to reduce the data for polydimethyl siloxane
(circled B), polyvinyl chloride (circled C), Hevea rubber (circled (Fig. 6) or (not shown) polyphenylene siloxane, poly-
H),andyolymethyl acrylate (circled M). (b): -[logt-E/2.3RT]
versus LAOI/(T-T.)]/[0.025+1l0l/(T-T.)J. The straight line ethylene, or various polymer-diluent mixtures to co-
corresponds to Eq. (8). incidence with the line representative for most of
the polymers in Fig. 6(b). In all these unsuccessful
instances viscosity data were available only at tem-
additional factors they may be brought into coinci- peratures relatively far above To. It is thus undeter-
dence. Modifications of the WLF equation for this mined whether such failure places an upper limit on
purpose have been suggested previously, based on the !la,(T-Tg) for the universal application of (8), or
observations first, that the expansion coefficient for whether a different form of the relation is required
the fractional free volume may be different for differ- for different "families" of liquids each with its charac-
ent liquids,6,lO and second, that the temperature coeffi- teristic structure.
cient of the viscosity of liquids at temperatures far Tentatively then, we advance Eq. (8) for further
above the glass temperature Tg (where the tempera- trial as an empirical representation of observed r- T
ture dependence of the WLF term is small) generally data for polymer systems generally, over a wide tem-
approaches adherence to an Arrhenius-type equation perature range above T g , requiring the assignment of
with a nonzero activation energy E.5.10 three parameters-T g, !la" and E-to each system.
In accord with this hypothesis, we were led to at- Two of these, Tg and !la,=al-ag, are assessable by
tempt to fit the data to a modified WLF equation of independent measurements. For the present E remains
the form a parameter which must be determined from the r-T
data themselves. If it is taken as ~4.4 kcal, the abso-
(7) lute magnitude of r for all curves in Fig. 3 would be

19 T. G Fox and P. J. Flory, J. Am. Chern. Soc. 70, 2384 (1948);

where E and ro are parameters to be evaluated, B is
J. App!. Phys. 21,581 (1950); J. Polymer Sci. 14, 315 (1954).
a constant of order of magnitude of unity, fg may be 20 R. Simha and R. F. Boyer, J. Chern. Phys. 37, 1003 (1962).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 ZERO SHEAR MELT VISCOSITY 351
TABLE III. Tentative values of E for various polymers.

E, kcal/mol, computed by Eq. (8)

Internal energy barriers (kcal/mole)
Computed from Computed from
Polymer t(TOC) in Ref. (10)- 1/-Z datab •c Molecule V

Polymethyl methacrylate 7 (125) CHa-CH2CN 3.3,5.2d

Polyethyl methacrylate 6 (100, 125)

Polyiso bu tylene 5.5 (100, 125) 5.5 (Table II) CH 3-C (CHa). 4.3-
CH3-CH(CH 3)z 3.6"
Polyethylene 5 (Table II) CH3CH..-CH2CH 3 3.3"
CH3-CH3 2.7-3.3 d

Polyvinyl acetate 3.5 (100, 125) 4.4 (Table II)

Polybutyl methacrylate 4 (100, 125)

Polyoxypropylene glycol 4 (25°)7

Polystyrene 2 (125) 3. 6 (Table II)

Hevea rubber 3.2 (25) CH3-CH2CH=CH2 3.6"

Polydimethyl siloxane 3 (Table II) CH"Si (CHa) 3 1. 3"
Polyvinyl chloride 1.8 (125) CH3-CH2CI 2.7,3.6,4.7
Polymethyl acrylate 1.7 (100, 125)

• Values for log t at T'C and of T. for each polymer taken from Ref. (10), p. 258. Values of t!..af were taken from Table II or from the data of Refs. (4), (20),
and L. A. Wood, J. Polymer Sci. 28,319 (1958).
b Computed by Eqs. (8) and (4) from the observed value of 1/ at T for a given Z, using X.=4.7X10- l6 and (so'/M.) from Table T and data of Refs. (4), e of
Table I, I of Table T, and (20); L. A. Wood, J. Polymer Sci. 28, 319 (1958); R. B. Beevers and E. F. T. White, Trans. Faraday Soc. 56, 744 (1960).
C The first two values of E for polymethyl methacrylate were computed from the pUblished 1/-Z data. taking T.=378'K and 10" Xc alternatively as 1.5 (Table

I) or 4.7. The third value was computed from viscosity data" at 217' on a polymethyl methacrylate of lower syndiotactic centent, taking X c=4.7XIO- l5 , T.=
4.7XIO-J6, T g=350'K and using the values of (s,'IM) for the conventional polymer.
d E. B. Wilson, Advan. Chern. Phys. 2, 370 (1959).
• G. J. Janz, Estimation oj Thermodynamic Properties oj Organic Compounds (Academic Press Inc., New York, 1958), p. 161.

given within a factor of 4. A closer estimate, or even computed if values of the characteristic parameters v,
its independent measurement, may be possible, per- (S02)Av/M, E, /:;.a" and To are known or estimated.
haps, since differences in values of E obtained (Table
III) may be related to differences in the internal barriers DISCUSSION
to rotation about the chain valence bonds. Thus high
values of E for f are observed for chains containing We regard the demonstration of the validity of the
two bulky side groups on alternate carbon chain atoms theoretical Eq. (1) and of the empirical Eqs. (2)
(polymethacrylate and polyisobutylene), intermediate expressing the dependence of the zero shear viscosity
values of f are found for chains containing only one for short and long chain polymers on pertinent mo-
bulky side group on alternate chain atoms (polysty- lecular characteristics as the most significant aspect
rene and polyvinyl acetate), and the lowest values of the preceding sections. It is now possible with these
of f obtain for the unsubstituted polyethylene and for to compute the viscosity from the independently meas-
the polysiloxane with the relatively long Si-O-Si bond. ured chain dimensions and friction coefficient. At pres-
Comparison of Eqs. (7) and (8) illustrate the useful ent, it is more likely that measurements of the viscosity
result that the values of logfo and of fg may be taken, and of chain dimensions will be used to compute f.
respectively, as -1.2 and 0.024 (a familiar result lO ), for A significant feature of Eq. (2) for the viscosity of
the different polymeric systems. The corresponding long chains is the dependence of 1/ on (S02)Av3 .4. A simi-
value obtained for B, 0.69, is somewhat lower than the lar result was suggested by Tobolsky,21 who observed
expected value of unity.lO proportionality between the maximum relaxation time
Equations (2), (7), (8), and (4) represent the com-
plete expression of the 1/- T -Z relations for linear flexi- 21 A. V. Tobolsky, Properties and Structures of Polymers (John
ble chain molecules. From them f, Zc, and 1/ can be Wiley & Sons, Inc., New York, 1960), p. 318.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07
 352 T. G FOX AND V. R. ALLEN

Tm and (S02)AV3.4 for long chain polymers, with the same pear more than ever to be of critical importance ill
value of the proportionality constant applicable to the further developments of molecular interpretation of
four different polymers for which data are available, the viscoelastic behavior of such materials.
where Tm for different polymers are measured in cOrre- The theories of Eyring, Rhee, and Hirai22 yield for
sponding states, i.e., at corresponding values of T- T g • short chains, a result different from the linear 1']-Z
In both the theoretical Eq. (1) and in the empiri- relation established here, and give no means for pre-
cal Eq. (2b) the parameter X plays a central role. dicting the dependence of 17 for long chains on their
Thus with increasing X, rr"X until X=4.7X1Q-15 chain dimensions.
and then 1']"-'X3.4. As mentioned, the parameter X = The present attempt to modify the WLF equation
«S02)Av/M) (Z/v) arises in the derivation of (1) as with the purpose of establishing a "universal" equa-
the moment of inertia of the average polymer coil tion for predicting r for various polymers over a wide
multiplied by the number of such coils per unit vol- temperature range from a few measurable parameters,
ume. Presumably X represents a measure of the hy- while not completely successful, may be useful. It sug-
drodynamic interaction or "entanglement" between gests that the dependence of the frictional coefficient
different coils, and both the viscosity and the appli- on intermolecular and on intramolecular factors may
cable interaction law are dependent on its magnitude. be separable, at least partially. For example, Eq. (8)
More precise data on 1'], Z, (S02)A,/M, and v at the may represent in primitive form, an expression of the
same temperature are required, of course, to determine influence of both intramolecular and intermolecular
whether Xc is actually the same constant for all poly- barriers to motion (E/ RT) and of the requirement
mer systems. The only serious existing discrepancy is that thermal fluctuations (against intramolecular and
that for conventional polymethyl methacrylate. Fur- intermolecular potentials) produce locally the "free
ther studies on both the conventional and the ideal volume" needed for motions of chain segments from
atactic polymer are in progress; more precise deter- one equilibrium position to another. Precise determina-
minations of the parameters for other linear and model tion of the temperature dependence of 1'], or of other vis-
branched polymers are planned as well. coelastic deformations, over wider temperature ranges
The observed dependence of 1'] on Z3.4 and on (Zg) 3.4 and for different types of polymers should yield further
for long chain linear and branched homologs, respec- insight into the question raised here.
tively, and the agreement between values of Z. com- More detailed and precise study of the 1']-Z-T re-
puted by Eq. (3b) and those from elasticity data lend lationships for the initial members of a homologous
support to Bueche's approximate theoretical treatment series would be also of interest. Here we would expect
for the viscosity of long chains. More extensive and (S02)AV/M to vary with Z, and both the values of r
careful theoretical and experimental examination of and of Tg may depart from those computed by (8)
his concepts of "interchain coupling entanglements" and (4).
and their role in determining the viscosity and time- 22 H. Eyring, T. Rhee, and N. Hirai, Proc. Natl. Acad. Sci.
dependent elasticity of macromolecular substances ap- (U.S.) 44, 1213 (1958).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.120.242.61 On: Wed, 26 Nov 2014 05:01:07