Sei sulla pagina 1di 117

F o r t s c h r . H o c h p o l y m . - F o r s c h . , Bd. 3, S.

1 9 6 - 3 1 2 (1963)

Intrinsic Viscosities and Unperturbed


Dimensions of Long Chain Molecules 1
By
MICHIO K U R A T A 2 a n d W A L T E R H , STOCKMAYER 3

D e p a r t m e n t of C h e m i s t r y a n d L a b o r a t o r y for C h e m i c a l a n d S o l i d - S t a t e P h y s i c s
M a s s a c h u s e t t s I n s t i t u t e of T e c h n o l o g y , C a m b r i d g e , M a s s a c h u s e t t s
and
D e p a r t m e n t of C h e m i s t r y , D a r t m o u t h College, H a n o v e r , N e w H a m p s h i r e

W i t h 31 F i g u r e s

Table of Contents Page


I. I n t r o d u c t i o n . . . . . . . . . . . . . . . . . . . . . . . . . . 196
II. B a s i c E q u a t i o n s . . . . . . . . . . . . . . . . . . . . . . . . . 206
A. E x c l u d e d V o l u m e E f f e c t . . . . . . . . . . . . . . . . . . . . 206
B. E x c l u d e d V o l u m e E f f e c t ( T w o - d i m e n s i o n a l Chain) . . . . . . . . 212
C. I n t r i n s i c V i s c o s i t y . . . . . . . . . . . . . . . . . . . . . . 217
D. F r i c t i o n a l C o n s t a n t . . . . . . . . . . . . . . . . . . . . . . 229
E. H e t e r o g e n e i t y Corrections . . . . . . . . . . . . . . . . . . . 232
III. Chain Conformations . . . . . . . . . . . . . . . . . . . . . . . 234
A. A d d i t i o n P o l y m e r s . . . . . . . . . . . . . . . . . . . . . . 234
B. Cellulose, A m y l o s e a n d T h e i r D e r i v a t i v e s . . . . . . . . . . . . 247
C. C o n d e n s a t i o n P o l y m e r s . . . . . . . . . . . . . . . . . . . . 257
D. P o l y p e p t i d e s a n d P o l y n u c l e o t i d e s . . . . . . . . . . . . . . . . 264
IV. D i s c u s s i o n s o n R e l a t e d P r o b l e m s . . . . . . . . . . . . . . . . . 276
A. Second Virial Coefficient . . . . . . . . . . . . . . . . . . . . 276
B. S h o r t R a n g e I n t e r a c t i o n s . . . . . . . . . . . . . . . . . . . 288
A p p e n d i x (Table 13: V i s c o s i t y C o n s t a n t s of L i n e a r P o l y m e r s ) . . . . . . . 292
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302

I. I n t r o d u c t i o n
The dimensions of long chain molecules in solution are greatly
influenced by the interactions between chain elements. These inter-
actions are conveniently divided into two classes. The "short-range"
1 T h i s w o r k w a s s u p p o r t e d in p a r t a t M I T b y t h e U. S. A r m y R e s e a r c h Office
a n d t h e U. S. A r m y Q u a r t e r m a s t e r Corps a n d in p a r t a t b o t h i n s t i t u t i o n s b y t h e
N a t i o n a l Science F o u n d a t i o n .
2 P r e s e n t a d d r e s s : I n s t i t u t e for C h e m i c a l R e s e a r c h , K y o t o U n i v e r s i t y , K y o t o ,
Japan.
: Present address: Department of C h e m i s t r y , D a r t m o u t h College, H a n o v e r ,
New Hampshire, USA.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 197

interactions are those between atoms or groups separated by only a


small number of valence bonds, and they result in the effective con-
stancy of bond angles and in the torques hindering internal rotations.
The "long-range" interactions, on the other hand, are those between
non-bonded groups which are separated in the basic chain structure b y
many valence bonds; they are therefore identical in nature and in
magnitude to VAI~ DER WAALS interactions between the parts of two
different molecules [see FLORY (5b)]. The viscosities of dilute polymer
solutions are a principal source of experimental information on chain
dimensions, but their interpretation demands a separation of the effects
of the two types of interaction from each other. The purpose of the
present review is to discuss the current status of this problem and to
survey the existing body of experimental results. The discussion is
limited to unbranched chains, and the special effects in polyelectrolytes
are also ignored: The main practical result of our work, embodied in
Tables 7--11, is the accumulation of values of "unperturbed" chain
dimensions (see below) for a large number of natural and synthetic
macromolecules.
In the absence of both types of interaction, except for the covalent
binding forces which fix the lengths of the chain links, a long chain
molecule obeys Ganssian or random-flight statistics. Under these con-
ditions, the mean square value of the (spherical) radius of gyration, for
example, is given by
(S )oo = t?n/6, (1)

where n is the number of bonds, l is the bond length and the double
zero subscript denotes the lack of both kinds of interaction 1. Short-
range interactions introduce appreciable changes in the average chain
dimensions, but they do not alter the ,proportionality between the mean
square radius of gyration and n, nor do they destroy the Gaussian charac-
ter of the probability function for the distance between widely separated
points in very long chains 2. At this stage, the chain without long-range
interactions m a y be called the "unperturbed" chain [FLoRY (Sb)] and

1 Obviously if there are several different kinds of bonds in the chain skeleton,
the quantity nl ~ m u s t be replaced b y the sum ~v" nd~2.
2 This has been demonstrated quite generally by TCHEN (2d5"). The essential
condition is t h a t the probability function for the direction of a given chain bond
can~depeud on the directions of only afinite n u m b e r of t h e preceding bonds. If the
corr'elations between the directions of neighboring links are especially strong, as for
example in helix-forming polypeptides, the limiting Gaussian behavior may n o t
be attained in t h e experimentally accessible range of molecular weights, but it m u s t
exist mathematically. As remarked b y H. BENOIT, an infinitely long steel rod must
behave as a random coil.
198 M. KURATAand W. H. STOCKMAYER:

the mean square radius of gyration m a y be written in the form

<s~>0 = sZ~n/6, C2)


where s is a structural parameter independent of n. Long-range inter-
actions, on the contrary, give rise to the so-called "excluded volume
effect", which cannot be described b y Gaussian statistics I, and which
can be pictured as an osmotic swelling of the randomly coiled chain
b y the solvent-polymer interactions. As a result of the superposition
of both types of interaction, the mean square radius of gyration of a
real linear macromolecule in dilute solution is generally written as

<s~> = ~ < s o o = (~,s) ~,,16 , (s)

in which the linear expansion factor ~ depends on the number of bonds


n. The molecular-structural features of the polymer chain are primarily
related to the factor s, which is often called the "skeletal" factor, while
the effects of interactions with the solvent are mainly reflected b y the
expansion factor ~. I t is clear that separate determination of the factors s
and ~ is essential for a clear understanding of the conformational proper-
ties of chain molecules.
It is well established that the excluded volume effect vanishes under a
special condition of temperature or solvent, which is usually known as
the Flory " t h e t a " temperature or solvent. Thus, light scattering
measurements performed on solutions under theta conditions can furnish
direct knowledge of the unperturbed dimensions [see, for example,
OUTER, CARR and ZIMM (207); SHULTZ (23,3); and NOTLEY and DEBYE
(201)]. Viscosity measurements, though less directly, can also furnish
similar knowledge with the aid of the Flory-Fox equation (103, 109),
which m a y be written
E~] = 6'/, r <SO'/,/M, (4)
or in more explanatory form,
[~] = [~]o~ ~ , (5)
[~/]e = K MV, , (6)
K = 6"I, # ( < S = > o l M ) ' 1 , , (7)
where [~] and [~]o represent the intrinsic viscosities in ordinary and
theta solvents, respectively, M is the molecular weight of the polymer,
and # and K are constants independent of M. Since # is supposed to be
x Since the last chain element can always interact with all its predecessors,
a correlation of order n is required for a complete statistical description of a chain
with long-range interactions. The order, therefore, increases without limit as the
chain is lengthened.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 199

the same for all flexible linear polymers, the unperturbed mean square
radius of gyration ($2)0 can readily be calculated from K. The unperturb-
ed dimensions of various chain molecules thus determined are found in
very m a n y cases to be practically independent of the particular theta
solvent employed, and are thus characteristic of the chain under con-
sideration, apart from a usually slight downward trend with increasing
temperature [see, for example, OUTER,CARR and ZIMM (207) and FLORY
(5/)] 1. A clear exception to the last statement is found in the helical
conformation of various polypeptide chains, which are stabilized by
hydrogen bonds between chain elements not widely separated from each
other, or by other rather special short-range interactions. Such short-
range effects are influenced very strongly by the solvent [DUTY, HOLTZER,
BRADBURY and BLOOT (87)] and the temperature. In principle, examples
intermediate between these extreme cases are possible and are provided
for in the most general theories of short-range interactions [ZIMM and
:BRAGG (273), NAGAI (795, 795'), LIFSON (767, 168, 168'), PTITSYN and
SHAI~O~OV (216), HOEVE (124) and others]. The most complete and
systematic exposition of short-range interactions is to be found in the
book of VOLKENSTEIN (21'). See Section IV B.
It may be recalled that for non-polar molecules of low molecular
weight the energy difference between gauche and trans s t a t e s of an
internal rotation angle is virtually independent of environment; for
example, this difference is about 0.8 kcal/mole for n-butane in both
gaseous and liquid states [PITzER (273) and MIZUSHIMA and OKAZAKI
(186)]. However, this is no longer true for polar molecules such as
ethylene chloride [MIzuSHIMA, MORINO, WATANABE and SIMANOUTI
(788)]. The absence of significant solvent or temperature effects on the
unperturbed dimensions of many polymer chains therefore strongly sug-
gests that in these cases steric rather than dipolar forces are predominant
in controlling the value of the skeletal factor s.
Measurements of dipole moment offer another possibility for direct
estimation of the short-range interaction, since this quantity is almost
free from the perturbations caused b y long-range interactions even in
good solvents ~ [MARCHAL and BENOIT (774)]. The short-range inter-
1 According to the recent results of SCHULZ and KIRSTt~ (229"), the u n p e r t u r b e d
.dimensions of poly(methyl methacrylate) increase monotonically with temperature.
See also PETERLIN (16").
2 Exceptions to this s t a t e m e n t are found in polymers with asymmetric monomer
u n i t s coupled head-to-tail, such as cellulose derivatives [ScltERER, LEVI and
HAWKINS (224"); SCI~ERER, TANENBAU~t and LEvi (225); K~EI~ (137")], poly-
peptides [WADA (259')] and polypropylene oxides [BAuR (33b)]. I n general, the
dipole m o m e n t is free of the effects of long-range interactions only if t h e correlation
{ # - L } vanishes, where /~ is the dipole m o m e n t vector and L is the end-to-end
-v,3ctor.
200 M. I'{URATAand W. H. STOCKMAYER:

actions in conventionally prepared poly(methyl methacrylate) were


found in this way [MARCHAL and BENOIT (77zl) and MARCHALand LAPP
(175)] to be virtually independent of solvent, in essential agreement with
the results obtained from intrinsic viscosities in theta solvents. On the
other hand, the observed dipole moments clearly show the effects of
stereoregularity [POHL, BACSKAIand PURCELL (274'); SALOVEY(223a)],
and in the case of isotactic poly(methyl methacrylate) a small solvent
effect is, in fact, seen in the results obtained by the latter author.
Another method of obtaining directly the unperturbed dimensions
of chain polymers is based on measurements of diffuse small-angle
X-ray scattering. It has been developed principally by KRATKY and
POROD and their collaborators, and has recently been reviewed by
KRATKY (]42'). The method necessarily requires a considerable difference
in X-ray scattering power between the polymer and the solvent, but it is
not limited to theta solvents. It has given figures in reasonable agreement
with those based on light scattering or viscosity for polyvinyl bromide
[KRATKY and POROD (142)], for natural rubber [KRATKY and SAND
(]d2")] and for poly(methyl methacrylate) [KIRSTEand KRATKY(139')].
The most extensive investigations with this technique have been on
cellulose nitrate, which is discussed in Section III B. The results of the
X-ray studies have usually been expressed in terms of the "persistence
length", which is defined in Section IV B.
According to the statistical-mechanical theory of rubber elasticity,
it is possible to obtain the temperature coefficient of the unperturbed
dimensions, d Ins/d T, from measurements of elastic moduli as a function
of temperature for lightly cross-linked amorphous networks [VoLKEN-
STEIN and PTITSYN (258'); FLORY, I-IOEVE and CIFERRI (I03a)].
This possibility, which rests on the reasonable assumption that the chains
in undiluted amorphous polymer have essentially their unperturbed mean
dimensions [see FLORY (5)], has been realized experimentally for poly-
ethylene, polyisobutylene, natural rubber and poly(dimethylsiloxane)
[CIFERRI, HOEVE and FLORY (66") and CIFERRI (66')] and the results
have been confirmed by observations of intrinsic viscosities in athermal
(but not "theta") solvents for polyethylene and poly(dimethylsiloxane).
In all these cases, the derivative d Ins/d T is no greater than about 10-8
per degree, and is actually positive for natural rubber and for the siloxane
polymer.
After values of K or of <SZ>o/M have been obtained as described
above, and after it has been demonstrated that these values do not
greatly depend on the solvent, the expansion factor a may then readily
be evaluated in any given solvent from the ratio ([~]/[~]o) v, or from
(<S~>]<S~>o)'I,. Thermodynamic analyses of the polymer-solvent system
may then be accomplished with the aid of an appropriate equation for
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 201

the excluded volume effect, such as the Flory (100) equation


a s _ ~3 = 2 C 2 u ( 1 / 2 - Z) MV', (8)
where CM is essentially a numerical factor and X is a parameter related
to the local free energy of the polymer solvent interactions. It is well
known that the quantity (~--o~8)]M '1, calculated from oca= [~/]/[~]e
is not strictly constant for most systems but displays a maximum at
some value of M [see, for example, FLORY (5i) ; NOTLEY and DEBYE (201) ;
and CHINAI, MATLACK, RESNICK and SAMIJELS (59)]. Furthermore, the
originally defined CM is about twice as large as the corresponding constant
in an exact series expression for small ~ [see STOCKMAYER (238)] 1. For
these reasons, the absolute significance of the Z values obtained from
Eq. (8) is uncertain, bat the relative values give useful measures of the
thermodynamic interaction between polymer and solvent. In other
words, the minor defects of E q . (8) do not mar its utility for semi-
quantitative estimates of Z, provided that the unperturbed chain dimen-
sions have previously been determined in unambiguous fashion.
The method outlined above is now well established. Its application,
however, is often limited b y the difficulty of finding appropriate theta
solvents. This limitation becomes especially serious for the investigation
of crystalline polymers with high melting points, because for such poly-
mers the theta condition can rarely be attained at ordinary temperatures.
It is therefore highly desirable to develop a method for estimating the
unperturbed dimengions without the aid of theta-solvent experiments.
. Several methods have been proposed in the literature for this purpose.
The first method, which may be called the Flory-Fox-Schaefgen method,
is based on the combination of Flory and Fox's viscosity equation (4)
and F l o w ' s excluded volume equation (8) [see, FLORY and FOX (703)].
The substitution of Eqs. (5) and (6) into Eq. (8) yields
[~?]'h/M'h = g 'l" + 2 CM (112 -- Z) K'h M/[n]. (9)
This equation implies that, if viscosity data are available over a wide
range of M, the two quantities K and g can be determined simultane-
ously from the intercept and the slope of a straight line which should be
obtained b y plotting the ratio [71]'hlM'1" against another ratio Ml[rl].
The method is ingenious and attractive. Its application, however, often
leads to underestimation of K, and it cannot be recommended for general
use in good solvent systems ~. The left half of Fig. 1 illustrates this state-
ment, where the Flory-Fox-Schaefgen plot for poly(ethyl methacrylate)
Flory's definition of C~ is
C,~ = (2712'h ='l,) (~IN~ Vx) (6 (S*)oIM)-'i*, (8')
where ~ is the partial specific volume of polymer, Vx is the molecular volume of
solvent and N a is the Avogadro number.
z In fact, no explanation of this method is given in Flory's book (5).
202 M. KURATA and W. H. STOCKMAYER:

is reproduced from the original figure by CHINAI and SAMUELS (61). The
black circles represent theta-solvent data obtained in a mixture of methyl-
ethylketone (1 volume) and isopropanol (7 volumes) at 23 ~ C, while the
white circles represent data obtained in the good solvent methylethyl-
ketone at the same temperature. The latter plot crosses the former plot,
leading to a clearly underestimated value of K. In fact, the lower inter-
cept B in the figure gives K = 0.9 9 10-4 which is only about one fifth of
the correct value, 4.73.10 -a, corresponding to the upper intercept
0 g // 6 <9 10
T l l I T--T--I------

PEPIA o

o PIEK 2 d ~
o
,,~ 9 T/tO/a_Mfx/ ddoG / d

I /I ...'"" I
"-G z

__L_...~. l ~ 0
0 10 ~0 dO 0 0.6 10 /.6-
M.,/[,A 9 ct'~}/t'~}oj~z-/op c/-/
Fig. I. Solution properties of poly(ethyi methacrylate). Left figure, Flory-Fox-Schaefgen plot according to
Eq. (9). Right figure, the dimensionless ratio A2M][7] as a functioxx of expansiotx Tactor

A. This behavior has fundamental importance. To demonstrate this,


let us accept the K value corresponding to the point A in the figure.
It is then obvious from Eq. (9) that the slope of the lines connecting
each white circle with the point A represents the quantity 2CM(1/2--
- - Z) K~/" for each material. Since K is kept constant for all materials,
this is also proportional to (~5_ ~a)]Ml/, in Eq. (8). The slopes of these
lines, though not given in the figure, increase with increasing M, general-
ly, and display a maximum at the point corresponding to the dotted line
if we trace the experimental points honestly. In other words, there is a
very close connection between the underestimation of K by the Flory-
Fox-Schaefgen plot and the molecular weight dependence of the Flory
ratio ( ~ 5 ot~)/M'/,. As mentioned before, this dependence of ( ~ s _
- - o,3)]M 1/, on M scarcely affects our molecular picture of the polymer
chains if Eq. (8) is used with the established values of the unperturbed
chain dimensions. But this is no longer the case if we try to estimate the
unperturbed dimensions by using this same equation for intrinsic vis-
cosities in good solvents.
Intrinsic Viscosities and U n p e r t u r b e d Dimensions of L o n g Chain Molecules 203

According to present theories of the osmotic second virial coefficient


Az, the ratio A2M/E~] is expected to be a function of ~ alone [see, for
example, STOCKMAYER(20)],
A2M][~1] = / ( ~ ) , (10)
provided that the Flory-Fox relation (4) is valid and that triple and
higher multiple contacts among chain segments can be ignored 1. KRIG-
BAUN (143, 145) has proposed a semi-empirical relation

l(a) = 2 0 0 ( 1 - g-a), (11)


while the expression of OROFINO and FLORY (204), which is based
directly on the theory of FLORY and KRIGBAUM (5C) is 2
/(X) = 414 1oga0[1 + 0.885 (~2 - 1)]. (12)
However, as can readily be seen from Eq. (11) or (12), the function/(a)
becomes insensitive to ~, or even constant, as ~ increases. As a result,
experimental errors in A 2, and hence in A zM/[~/], which are unfortunately
not small, induce rather large fluctuations in the predicted values of x,
making almost impossible any reliable estimates of a in good solvents.
To illustrate this statement, we first show the experimental values of
AzM~/E~?] for the poly(ethyl methacrylate)-methylethylketone system
in the right half of Fig. 1, where the values calculated by Eqs. (11) and (12)
are also shown by the solid and the chain line, respectively. The scattering
of the A~ values is quite large in comparison with that of the viscosity
data for the same system by the same investigators (67) shown in the left
half of the figure, and the locations of the experimental points are also
inconvenient for evaluation of ,r from the theoretical curves. A more
extensive representation of some existing values of A aM/ [~/] is given in
Fig. 2, where the white circles refer to polyisobutylene, all types of
squares to polystyrene, all types of triangles to polyvinylaeetate, the
black circles to atactic polypropylene, and finally the three types of
crosses to linear polyethylene 3. The solid and chain lines again represent
Krigbaum's equation (11) and the Orofino-Flory equation (12), respective-
ly. The K ( = [tl]o/M `l') values employed for these plots are given in
Table 1. Since theta-solvent experiments on polyethylene have not yet
been reported, we show the data for this polymer separately in the right
a F o r a m o r e detailed a n d critical discussion of t h e second viral coefficient, t h e
reader is referred to Section IV A.
T h e numerical c o n s t a n t s in these expressions are appropriate to intrinsic
viscosities expressed in d l . g-X, b u t with A i M in m l . g-1. These are t h e u n i t s m o s t
c o m m o n l y employed a t present.
s T h e while d a t a points of various shapes represent t h e experimental d a t a
selected b y OROFI~O a n d FLORY (204) as specially reliable u p to 19S7.
Fortschr. Hochpolym.-Forseh.,Bd. 3 14
204 M. KURATA a n d W. H . STOCKMAV'~R:

quarter of the figure, using a somewhat different abscissa. The scattering


of the experimental data is surprisingly large, even for the standard
polymers such as polystyrene,, and it is almost obvious that the applica-
300 l I

v
9 9 9 •
9 v
/5-0 Av 9 ~ A j.~/./

9 29 :o 9 x x + . ++

X: o~ ~ D./i • + +
+
9 9 . / ~ ++

5"0
./// +
~-+
+t-i-
/ / , 9 .. 9 .

I I I
0 0.5" Zb- 0 5" 10
((~) / [,:)o/:~- : (,:)/M '/~ 9 :as
Fig. 2. Dimensionless ratio A~M/[,7] for various polymers. Open circles, polyisobutylene (/d8).Upright
triangles, poly(methyl methacrylate) (dO, 47", 51", 59). Inverted triangles, polyvittyl acetate (62, 233).
Squares, polystyrene (35, 71", 7,{, 748,206, 208). Filled circles, atactic polypropylene (738). Crosses, linear
polyethylene (26, 152, 255). The unfilled points were selected as specially reliable in 1957 by OROFINOand
FLORY(204)

tion of the A ~M/E~] method to polypropylene or polyethylene would not


necessarily lead to trustworthy estimates of ~.

Table 1. The values of K employed for the plots in Fig. 2


Polymer Temp. K. 104 Reference

Polyisobutylene 25 10.7 FLORY (St)


Polystyrene 34 8.2 FLORY (5/)
Polymethylmethacrylate 25 5.9 CHINAI e t al. (56)
Polyvinylacetate 25 9.3 SrlULTZ (233)
Polypropylene 34 16.9 DANUSSO e t al. (7if)

Ln addition to the aforementioned practical difficulty of the method


based on A2M][~I], it must be realized that its theoretical basis is also not
secure. The excluded volume problem of two entangled chains, which is a
fundamental part of the theory of the second virial coefficient, has not
been much advanced beyond the first-order perturbation stage, and as a
result the function /(a) of Eq. (10) is only imperfectly known. For these
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 205

reasons we are inclined to be generally suspicious of calculations of a


by this method.
In view of the foregoing analysis of the present status of the problem,
it seems worth while to refine the Flory theory of the excluded volume
effect of a single chain, and to develop a new method for estimating the
unperturbed dimensions of chain molecules from viscosity measurements
in good solvents. Such an investigation is reported here. In Chapter II,
we present the basic equations for the excluded volume effect and the
intrinsic viscosity, and we test their validity by comparisons with
computational data on the dimensions of lattice chains (Sections A and B)
and with viscosity data of some typical amorphous polymers whose chain
conformations have already been well established by means of theta-
solvent experiments (Section C). The new viscosity equation offers a
method of fairly wide applicability for separate estimation of the short-
range and long-range interactions. The remainder of this chapter is
devoted to discussions of the friction constant (Section D) and of hetero-
geneity corrections (Section E). In Chapter III, we illustrate the results
of application of the new methods to various polymer molecules and
investigate their chain conformations in some detail. We first describe
the mean-square dimensions for various types of freely-rotating chains,
and then estimate the steric hindrance to internal rotations for addition
polymers (Section A), for cellulosic and amylosic chains (Section B),
for condensation polymers (Section C), and for several polypeptides
and polynucleotides (Section D). In contrast to one prevailing view, it
will be shown that according to our method all cellulose derivatives can
be characterized by a rather normal degree of steric hindrance; in other
words, by a rather small unperturbed dimension and by a rather large
extent of swelling. In Chapter IV, we discuss the related problem of the
second osmotic virial coefficient (Section A) and make some observations on
various treatments of the influence of short-range interactions (Section B).
In the course of preparation of this review, we have had the opport-
unity to read a number of papers on viscometric studies of polymer
solutions, and hence to collect the constants of Mark's, Houwink's and
Sakurada's empirical relationship,
In] = K'M~, (13)
for various polymer-solvent systems. The most recent extensive tabula-
tion of the constants of this equation is that of PETERLIN (16), as far as
we are aware 1. Since these equations have great practical significance,
* Just as this article was being completed, a review b y MEYERttOFF (13 t')
appeared, containing a large but also incomplete tabulation of viscosity constants.
In general, the article of MEYERROFF and ours are complementary, ours being
somewhat more concerned with theoretical and his with experimental questions.
14"
206 M. KVRATA a n d W . H. SToCKMAYER:

we have compiled a list of Mark-Houwink-Sakurada constants, un-


doubtedly still very incomplete, in the Appendix (Table 13) of this
review.
Finally, we m a y note the appearance of a recent s u m m a r y article
b y PETERLIN (76'), which deals with the same questions as those of
this review, though less extensively and of course from a somewhat
different point of view.

II. Basic E q u a t i o n s
A. Excluded Volume Effect
The 10ng-range interactions between chain elements, i.e. inter-
actions between non-bonded segments of polymer chains, are represented
in sufficiently dilute solutions b y the binary cluster integral 1,

/3 = ~ ( 1 - - exp E-- w (r)lk T]} 4= r~dr , (14)


0
provided t h a t the pair potential of average force w(r) as function of
the intersegmental distance r is assumed to be of short-range nature,
as required b y the ineqnality
/3 (((L~) '/' . (15)

Here k T has its usual meaning and L is the displacement length of the
chain considered. The expansion factor ~ is then a function of a single
parameter which m a y be written

z = (3/2~) ']'/3a-~N '/' (16)

for a chain of N links of the effective length a. In the absence of this type
of interactions, i. e. at T = O where/3 = 0, the mean-square dimensions
of the chain are simply written

(L2)0 = 6($2)0 = a 2 N , (17)

which is identical to Eq. (2).


For small values of z, exact expressions for expansion factors can be
obtained b y means of perturbation treatments of the interactions [see,

x T h e opposite choice of sign for fl is m a d e b y o t h e r a u t h o r s [for example,


GRIML~Y (117)]. The p r e s e n t choice m a k e s t h e numerical value of fl positive in
almost all e x p e r i m e n t a l l y s t u d i e d p o l y m e r solutions. I t should also be r e m a r k e d
t h a t Eq. (14), a l t h o u g h )vritten for a central p o t e n t i a l w (r), is easily generalized to
t h e case t h a t w also d e p e n d s on t h e relative o r i e n t a t i o n s of t h e i n t e r a c t i n g s e g m e n t s
[see, for example, ZIMM (271)].
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 207

for example, TERAMOTO (246, 247), ZIMM, STOCKMAYER and FIXMAN


(276); FIXMAN (98); and YAMAKAWA and KURATA (267)]. These are,
for example,
a 2 = <S~>/<$2>o = 1 + (1341105)z . . . . , (18)
and 1
~2 = <L,>/<L2>o = 1 + (4]3)z--2.08z 2 + . . . . (19)

The difference between a and a~ is caused b y the non-Gaussian character


of the chain, but it is rather small and negligible for most purposes. These
equations can be effectively applied to systems in poor solvents near the
Flory t h e t a temperature, but not to systems in good solvents because of
slow convergence of the series [see, for example, KURATA, YAMAKAWA
and TERAMOTO (158)].
A t t e m p t s have been made to obtain a closed expression for a, or at
least a form amenable to numerical computation. The Flory expression,
Eq. (8), and the F i x m a n (98) differential equation,

zaz,(do~L/dz) = 1 - - a~ + 2(z/aL), (20)


are examples. Since Flory's interaction p a r a m e t e r Z is related to fl b y

fl = 2 V1(1/2 - - Z), (21)

the equation (8) can be written

0r ~3 ~ CF z , (22)
where V1 is the molecular volume of the solvent and the constant C~
has the value 2.60, or about twice t h a t needed to secure agreement
with the exact series (18) or (19) for small z [STOCKMAYER (238)] 2.
For extremely large z, this equation predicts the asymptotic relation,

o:5/z = C ~, . (23)

On the other hand, the F i x m a n differential equation (20), when solved


in the series form of z, gives the correct coefficient 4/3 for the linear term,
and --0.67 (instead of --2.08) for the quadratic term. For large z, it
predicts the asymptotic relation,

~3/z = 1.5o. (24)

x The coefficient of z' is exactly --(16/3) + (28~/27).


2 If CF is set equal to 4]3, thus forcing agreement with the linear term in
Eq. (19), the coefficient of z2 in the expansion of Eq. (22) becomes --2.67 in fair
agreement with the exact value --2.08 [see FIXMA~ (98)].
208 M. KURATA and W. H. STOCKMAYER:

Very recently, KURATA, STOCKMAYERand R o m (155) have suggested


another approximate equation 1,
.s_. = C g(.) z (25)
with
g(~) = 8 .~](3. ~ + 1)'1'. (26)
The constant C is set equal to 134/105 or 4/3, depending on the definition
of ~. This equation gives a value of about --0.2 for the coefficient of z~,
and predicts the asymptotic relation,
ocS/z= 1.85 or 2.05, (27)

for large z. This behavior has a remarkable resemblance to that of


Fixman's differential equation mentioned above. In fact, it has been
found that the equation (25) is also in good agreement with the numerical
solution of Eq. (20) for intermediate values of z, both showing the
quantity ( ~ -- o:3)/zto be an increasing function of z rather than constant
as expected from Eq. (22). This agreement between Eqs. (20) and (25)
is all the more remarkable in view of their apparently totally different
methods of derivation, and it is therefore tempting to regard this agree-
ment as additional support for their validity. I t should also be remarked
that the inferiority of these equations to Flory's in predicting the
numerical coefficient of z z in the expansion (19) need not be regarded as a
serious matter; the convergence of the series (19) is so slow that, for all
but the smallest values of z, terms well beyond the second play a domi-
nant role.
A somewhat similar equation has been given b y PTITSY~r (21.9) on
the basis of related arguments, and the deviations from the Flory
relation are in the same direction. Thus there seems to be general
agreement that some modification in the Flory expression for the long-
range interaction effects on chain dimensions is needed.
In spite of the above considerations, it nevertheless remains true
that there is no purely analytical way to assess the validity of Eqs. (20)

1 Eq. (13) and Eq. (32), and also the first equation in the abstract, in Kurata,
SrOCKMAYXR and l~tom's paper (155) are erroneously printed. The factors,
{(I + 1/3~-~)-'/,} -x and {[1 + (112~r -x,
in these equations should be read as
[1 + (l/3Gd)]-'h and [1 + (112~)]-'1,,
respectively. The footnote 13 is also erroneous; the number 0.151 in the second
equation should be replaced by 1.347, and the third equation should be read as
a * - - 1 = 0.875z . . . . .
This number, 0.875, is about 30% lower t h a n the value needed to secure agreement
with t h e exact series (18) or (19) for small z.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 209

and (25). For this purpose we must therefore turn to the results of
various numerical calculations of the dimensions of chains with excluded
volume on simple lattices. These calculations may be divided into two
classes: exact numerical calculation for all possible configurations
[TERAMOTO et al. (248, 249) ; and FISHER et al. (96, 97)] and statistical
or Monte Carlo calculation [WALL and coworkers (261,262)]. By using
an ingenious extrapolation method, FISHERand HILEXr(96) have recently
shown that for a simple cubic lattice chain, the exponent 7 in the
expression,
(L2) = A'n , (2S)
attains the rather large value 1.37 4-0.05 in the asymptotic limit of
extremely good solvents. This value of y is in good agreement with
the value 1.33 predicted from Eqs. (24) and (27), but higher than 1.20
predicted from Eq. (23)1.
The Monte Carlo calculations by WALL and ERPENBECK (261) also
give support to equation (25). This may be demonstrated as follows.
Substituting Eqs. (3) and (16) into Eq. (25), and putting
N=n/c and a 2 = c s l ~, (29)
we obtain
(,S~)l n = (S%I,, + 2.865. lO-a(fl/c ~) [g(a) n1(S2)'1.] . (30)
Here we assumed that C = 134/105. The quantity c is an arbitrary con-
stant which is needed to express the number and length of effective
segments, N and a, in terms of those of the real chain elements, n and l.
In other words, we cannot assign an unique value to c merely from the
present definition of the equivalent random chain, Eq. (17), since any
arbitrary choice of c satisfies the condition, Na2= s n l 2 [cf. Eq. (2)] 2.
The Eq. (30), Iike Eq. (9), gives the possibility of determining the
unperturbed dimension (SZ)o and the interaction parameter (r/c2) simul-
taneously from the intercept and the slope of the straight line which is
obtained by plotting the ratio (S~)/n against the quantity g(a) n/(S~) 'l'.
The existence of the factor g (a) complicates the procedure, but we may
1 I t is implied in the paper of FISHIER and HILEY (96) t h a t the asymptotic
value of ~ for large n might depend on the power of the solvent, as varied in their
calculations by the value assigned to the potential energy between a pair of chain
elements in nearest-neighbor contact. This situation, however, would be impossible
if the expansion factor m depended only on the single variable ~ of Eq. (16), and
there is no feature of their lattice model which could affect the x (z} relationship
in this way [see, for example, ZIMM, STOCXMXYER and FIXMAI~ (276)]. Of course,
we have no absolute guarantee t h a t the correct asymptotic relation should have
the form of Eq. (28).
With the introduction of any value of c other t h a n unity, Eq. (16) m u s t be
modified to read file 2 ~ V x (1 - - 2x), so t h a t only the ratio of file ~ is obtainable from
experimental data.
210 M. KORATAand W. H. STOCKMAYER:

first find the zeroth approximation for (S~)oin b y plotting (S2)/n simply
against hi(S2) 1/,, and so calculate a. Then, using this approximate value
of a, we m a y replot (S~ against g (a)nl(S2) '1., and find an improved
value for (S2)o/n. The convergence of this method is fairly rapid, and
we call generally obtain a sufficiently precise answer at the second
triM. Figure 3 illustrates this statement, where the Monte Carlo data of
WALL and ERPENBECK for a diamond lattice chain are plotted b y the

6:8

_ BJumondJ/a#/ce tchaln ~ 9 (~) ~,. ,..*'~ ~"~f*JJ1>'/~'~1 9


0U' -" "'"o.......~ -~-~o,0~ -

0.5

~ ~ - - ~ Pre~ent theonF

~ . 9 Flop//thsony

[ I I t r I r I
0 /0 ~0 30 ~'0
g ( e ~ ) n / <S~) ~/~" op nZ/'<S~') 3/~"
Fig. 3. Mean square radii of d i a m o n d lattice chains: Monte Carlo calculations of WALL a n d ]~RPENBECK (261),
plotted according to Eq. (30) (open circles) a n d Eq. (31) (filled circles)

small black circles in the first trial and b y the white circles in the second
trial. F r o m the latter, we obtain the values of (L2)o/n (~ 6 (52)o/n) and
/5/c2 which are listed as the "observed" values in the first row of Table 2.
Since l is unity b y the definition of the chain, {L2)oin is identical to the
skeletal factor s. The value of (Le)o/n thus predicted b y the use of Eq. (30)
is in good agreement with the rigorous value, which is equal to ( 1 -
--cos0)/(1 + cos0) and hence exactly two. According to STOCKMAYER
(20), the binary cluster integral for a pair of chain elements, t50, is equal
to the volume per lattice point T, which is 1.53 for the diamond lattice.
As the effective segment is naturally expected to be larger than the real
chain element, this value 1.53 should be referred to as the lower limit
for /5. I t turns out t h a t the constant c cannot be smaller than 2.1. In
other words, the usual but rather arbitrary choice c = 1 undoubtedly
leads to underestimation of/5, and cannot be recommended. The values
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 211

of z a n d cm~. are also given i n T a b l e 2. T h e F l o r y e q u a t i o n (22), on the


other h a n d , m a y be r e w r i t t e n

(S~)I n = (S~)o/,~ + 5.59- 10-~(/~/c ~) ((S~)ol .) [~l(S~}'l,] . (31)


T h e applicatioI~ of this relation to the Monte Carlo d a t a is also i l l u s t r a t e d
ill Fig. 3 b y the large black circles, which lead to t h e p a r a m e t e r values
(L~)o/n = 0.60 4- 0.15 a n d fl/c ~ = 2.64- 1.0, a n d finally c0~,,=0.8.
I t is seen t h a t (L~)o/n is seriously u n d e r e s t i m a t e d ~.

Table 2. Unperturbed dimensions (LZ)o and segmental interactions fl/c~ for various
lattice chains
Lattice chains rig.(L*}o~. r (/ff[~)obs Cmia

Three dimensional lattice:


diamond, 3-choice . . . . . . . 2 2.04-0.1 1.53 0.36dz0.03 2.1
simple cubic, 4-choice . . . . . . 1 1.0-4-0.1 1.00 0.35 :t:0.05 1.7
simple cubic, 5-choice . . . . . . 1.5 1.44-0.2 1.00 0.37+0.06 1.7
Two dimensional lattice:
square. 2-choice . . . . . . . . 1 1.04-0.05 1.00 O,81q-O.Ol 1.1
square. 3-choice . . . . . . . . 2 2.44-0.5 1.00 0,35+0.05 1.7

Similar tests of Eqs. (30) a n d (31) can be based on the older results
of WALL, HILLER a n d ATCHISON for various lattice chains, t h o u g h
over a m u c h more restricted range of n (262). The results o b t a i n e d for
two t y p e s of simple cubic lattice chains 2 are given i n T a b l e 2, together
w i t h the rigorous v a l u e of (L2)o/n a n d the values of ~ a n d cmi,. I t is
again concluded from t h e t a b l e t h a t the values of (L2}o[n predicted b y
the use of Eq. (30) are i n good a g r e e m e n t w i t h the rigorous values a n d
t h a t t h e u s u a l choice of c = 1 is too small to o b t a i n t h e correct v a l u e of ft.
I n a p p l y i n g the a b o v e m e t h o d to real p o l y m e r solutions, we m a y
rewrite Eq. (30) as

(S~)I M = (A2/6) + 2.865- 10-~ B [g(~r MI(S2)V,], (32)


w i t h the a b b r e v i a t i o n s
A S = ( L ~ ) o / M = s l~/m, (33)

B =fl/c 2 m ~ . (34)

1 If the Flory constant CF is replaced by the value C ~ 1341103, as suggested


earlier [STOCKMAYER(20, 238); FIXMA~ (98)], the numerical constant in the last
term of Eq. (31) is changed to 2.865 9 10-~. This cannot alter the value of (L2)a/n
obtained by the extrapolation but just doubles the value of fl[c~.
2 The bond angle is restricted to 900 in the four-choice simple cubic lattice
chain, but not in the five-choice chain. In other words, the transition probability
of two successive bonds in the latter chain is 4]5 for the angle of 90~ and 1]5 for the
angle of 180~ in the ideal case of no excluded volume.
212 M. KURATA a n d W . H . STOCKMAYER:

Here m(=M[n) is the molar weight of a chain element, or most fie-


quently the average molar weight per skeletal carbon atom. As an
example, we illustrate in Fig. 4 the application of Eq. (32) to the light
scattering data for polystyrene obtained by NOTLEY and DEBYE ( 2 0 1 ) ,
where the white circles represent the (S2)z/Mw vs. g (~)Mw] (S~), '/, plots
for cyclohexane solutions and the black circles represent those for
toluene solutions. Ignoring three white circles located at the left of the
J i I I I [ 11 I/,~ I

Polys/erene
0.2 o Oyclohaxune / .
9 Toluene J20 o~

~. ^

S4 ~ ~rheeapolnO

I r ! f f I I I
0 2 ~ 8 8 I0
g m) M/(s~) 'f~.io~(A.#.)
Fig. 4. Mean square radii of polystyrenes plotted according to Eq. {32). D a t a of It"OXLE'2"and DEEYE (201)

figurez, we can determine a set of straight lines with a common intercept,


as anticipated from Eq. (32). The existence of this common intercept
confirms the practical independence of the unperturbed dimensions on
solvent mentioned in the previous chapter and the possibility of deter-
mining the unperturbed dimensions from good-solvent measurements,
at least in principle. But, in practice, such scattering of experimental
points as that observed in Fig. 4 is almost inevitable in light-scattering
measurements in the low molecular weight range, and this seriously
reduces the accuracy of prediction of (L2)0 by Eq. (32).

B. Excluded Volume Effect (Two-dimensional Chain)


The investigation of the statistical properties of two-dimensional
systems, though not very important in a purely practical sense, often
plays a complementary role to that of three-dimensional systems in
x Similar s c a t t e r of t h e l i g h t - s c a t t e r i n g d a t a in a r a n g e of r e l a t i v e l y low m o l e -
c u l a r w e i g h t ( M = 2 N 3 9 10 5) of p o l y s t y r e n e is also o b s e r v e d i n m e t h y l e t h y l -
ketone by OUTER, CARR and ZI~H (207), and by OYAMA, KAWAHARA and UEDA
(13o).
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 213

clarifying the nature of the problem. This is true in the case of the exclud-
ed volume effect, for the exact treatment is confronted with formidable
mathematical difficulties and the numerical data needed to establish
approximate equations are still limited in number for three-dimensional
chains.
In the vicinity of the theta temperature, the perturbation treatment
can also be applied to two-dimensional chains, and the following equa-
tions can be derived for the expansion factors1:

~ = <S2)[<S~)o = 1 + (11/24)z* . . . . , (35)


and
o~ = <L~>/(L~>o = 1 + (1/2)z* . . . . , (36)
where the unperturbed dimensions, <$2)o and <L~>o, are defined as for
three-dimensional chains, and the parameter z* is defined as
z* = (/3*/~r a S) N . (37)

Here/3* is the two-dimensional binary cluster integral defined by

/3* = - exp [-w(,)/k dy. t3s)


o
The difference between cr and col is again insignificant, and it may be
ignore d for most purposes.
It is now well known that the essential assumption originally made
by FLORY in obtaining Eq. (22), is that of a spherical distribution of
segments about the molecular center of mass. The additional assumptions
are less crucial, merely affecting the value of the constant CM [see,
FIXMAN (98); CASASSA a n d OROFINO (5)'); and GRIMLEu (I/8)]. This
assumption of spherical symmetry, however, does not correspond to the
segmental distribution for a chain with fixed ends, which is needed for
calculation of the excluded volume effect. For this reason, I{URATA,
STOCKMAYER and ROlG (755) adopted a segmental distribution with
axial, rather than sphreical, symmetry about the end-to-end vector,
and derived Eq. (25).
Similar considerations if applied to two-dimensional chains, naturally
lead to an elliptical model, and finally to the following equation (155):
a3- a = C * g * ( a ) z* (39)
with
g* (a) = [3a2/(2~ ~ + 1)]'t,, (40)

where C* is set equal to 11]24 or 112, depending on the definition of ~.


In the asymptotic limit of extremely good solvents, this equation
1 To our knowledge, these equations have not previously been published.
214 M. I~URATAand W. H. STOCKMAYER:

predicts 1.67 for 7 in Eq. (28), while the circular model equation,

~4__ o~2 = C ' z * , (41)


predicts 1.50 for 7. The value obtained b y FISHER and HILEY (96)
for the three-choice square lattice chain is 1.57 • 0.06 and falls between
the two estimates. The substitution of Eqs. (3) and (29) into Eqs. (39)
and (41) yields

3 •qvape /o#/ce '/ ~ ~J


3- cho/'c8 chda 9f i : .
:/." ~ o~

B
j" oooO~o

9P o/z

9- - . gYrcular model

7
0 • 3 3 ~ 3-

[261)
Fig. 5. M e a n square radii of three-choice sqnare lattice chains: calculations of W A L L and E R V E N B E C K
and FISHER and HILEY (96), plotted according to Eq. (42) (open circles) and Eq. (43) (filled circles)

<S2>l~ = (<S2>ol~) + 2.43.10 - 2 (<S2)o/n) "l' (0"1c2) [g* (~) n'/'/<s2> v'] (42)
and
<S~>/n = (<S2>o/n) + 2.43. I0 -~ (<S2)o/n) (~*/c 2) [n2/<$2>], (43)

respectively. Figure 5 shows the comparison of these equations with


the Monte Carlo data of WALL and ERPENBECK for the three-choice
square lattice chain (261), where white circles represent the plots b y
Eq. (42) and black circles represent those by Eq. (43). There are also
shown four short-chain points of FISHER and HILEu (96) (small white
circles), corresponding to n = 4, 6, 8 and 10. The values of (L2)o/n
(=--6<$2>o/n) and (fl*/c 2) obtained from Eq. (42) are given in Table 2 in
comparison with the exact figures. The derived value of <L2)o/n agrees
fairly well with the exact one, but the circular model, equation (43),
gives an unreasonably small value for (L2)o/n.
I n t r i n s i c Viscosities a n d U n p e r t u r b e d D i m e n s i o n s of L o n g C h a i n Molecules 215

T a b l e 3. The mean-square end-to-end distance (L ~> of two-choice square lattice chain


and effect of solvent interactions 1
(L')
n !
p=l 0.2s I o.so I o.Ts 1.50 2,00 i 3.oo { s.oo ] io.oo

1 4 1.000
2 8 2.000
16 3.000 4.200 3.667 3.286 2.600 2.333 2,000 1,667 1.364
24 5.333 6.667 6.000 5.600 5.000 4.800 4.571 4.364 4.191
5 40 7.400 9.966 8.636 7.885 6.807 6.455 6.053 5.683 5.366
6 64 10.000 13.429 11.600 10.615 9.273 8.857 8.400 8.000 7.677
104 12.692 17.521 14.854 13.505 11.759 11,233 10.650 10.116 9.640
168 15.619 21.726 18.286 16.605 14.522 13.929 13.309 12.794 12.399
272 18.647 26.633 22.173 19.984 17.017 15.995 14.683 13.211 11.589
440 21.782 31.281 25.884 23.323 19.931 18.807 17.457 16.124 14.906
1 712 25.000 36.524 30.071 27.005 22.123 19.943 16.792 13.244 9.936
12 1128 28.879 41.794 34.324 30.964 25.988 23.746 20.203 15.559'10.451
13 I808 32.558 47.391 38.764 34.951 29.119 26.355 21.953 16.577, 11.865
14 2896 36.088 53.220 43.246 38.838 32.127 28.830 23.376 16.430 9.856
15 4600 39.98 I
16 7352 44.57
17 11712 48.81
18 18656 53,31
19 29752
20 47128
fl*/c 2 0.81 1.61 1.14 0.93 0.64 0.52 0.34 0.13 --0.10
fl*l~*(p = I) 1 . 0 0 1.99 1.41 1.15 0.79 0.64 0.42 0.16 ---0.12

4.U
YCuape la#ice

3.0
, o

l I I
0 ~- 10 15

Fig, 6, Mean square displacement lengths of two-choice square lattice chains with aearest-neighbor
interactions, front Table 3
1 T h e v a l u e s o f f a n d ( L 2) for p = 1 are t a k e n f r o m t h e t a b l e s g i v e n b y TERA-
MOTO a n d his coworkers (248, 249).
216 M. KURAT&and W. H. STOCKMkYER:

Another set of numerical data for <L2> has been presented b y TERA-
MOTO and his coworkers (248, 249) for the two-choice square lattice
chains 1. These data are shown in Table 3, together with the n u m b e r of
possible configurations [ and a set of new numerical d a t a concerning
the effect of solvent interaction. The variable p is defined as
p = exp (u/k T ) , (44)
where u is the energy of a nearest-neighbor contact between unbonded
segments, its sign chosen to be positive for poor solvent systems. The
application of Eq. (42)
to this chain is shown in
Fig. 6, where the points
Zqu~r~ /a#/~e chu&
corresponding to n below
7 or 10 are omitted for
the sake of clarity, ex-
cept for p = 1. In the
region of n higher than
o about 10, the linearity
of the plots d e m a n d e d b y
Eq. (42) is well satisfied 3,
-[ 1 E I
-2 -/ 0 / B d and we can determine a
In p = u / I c T group of lines having a
Fig. 7. Interaction parameters for two-choice ,~luaro inttlco chains, common intercept at a-
from Table 3
round <L~>/n = 1 which
corresponds to the rigorous value. The values of fl*[c2 obtained from the
slopes of these lines are given in Tables 2 and 3. To investigate the
energy effect on fl*, we eliminate the unknown parameter c b y taking
the ratio of r * and its athermal value r * (for which u --- 0 and p = 1),
and plot it against inp as shown in Fig. 7. Tlfis plot leads to
fl*/fl~' = 1 --0.55 (u/k T) (45)
for small u, which corresponds to

O ~ 0.55 u/k, or P o ~ 7.5, (46)


with a very large uncertainty which is inevitable because of the rather
b a d scattering of the points for the poor-solvent systems in Fig, 6 and
the lack of d a t a for longer chains. Indeed, WALL and MAZUR (262') on
the basis of Monte Carlo calculations have recently concluded t h a t the
two-choice square-lattice chain does not have a t h e t a point, and have
x The bond angle is restricted to 90 ~ in the two-choice square lattice chain,
but not in the three-choice chain.
The equation for (L 2> corresponding to Eq. (42) is
<L~)In = (<L~)o[n) + 0.159 (<L2)o/n)V,(fl*/c2) [g*(o~)n'l,l<L2>'&] . (42"}
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 217

given some qualitative arguments for this conclusion. In contrast,


FISHER and HILEY (96) find 0 - - ( 1 . 4 4-0.2)u/k or P o = 2.044-0.15
for the three-choice square-lattice chain. This striking difference suggests
that further investigations on the relationship between chain stiffness
and the theta condition would be worth while.
According to the definition, Eq. (38), the cluster integral t * may be
roughly evaluated as
t3" = (1 + ~) -- ( 2 w - ($) (u/k T), (47)
for the lattice chain, where ~ represents the correction to the coordination
number 2w required b y the occupation of two sites of the nearest-
neighbor shell b y the adjacent segments of a central segment [see, for
example, FLORY (5C); and HUGGINS (12)]. It is also well known that this
kind of correction is influenced b y chain flexibility [see, for example,
TOMeA (21)]. If we put ~ = 2, we obtain 0.67 for the coefficient of (u]k T)
in Eq. (45), in satisfactory agreement with the graphical estimate.
Now, returning to Table 2, we find the value of c to be always larger
than unity. Therefore, if we are concerned not only with the unperturbed
dimensions but also with the thermodynamic interactions, the usual
definition of the equivalent random chain, i. e. c = 1, can no longer be
recommended. On the other hand, a rough equality may be found be-
tween (L~)o/n and c,.~., except for the one case of the four-choice chain
on the simple cubic lattice. The equality of c to (L2)o/n corresponds to
the Kuhn condition, (153),
n l = N a, (48)
for the preferred statistical chain. However, the coordination number
correction 6 m a y destroy the apparent coincidence between (L~)o[n
and c,.i~, because in this case, the reference value for/3 becomes equal
to T(1 + 5) instead of z. A more rigorous investigation of the relation
between the real chain and the equivalent statistical chain is required.

C. Intrinsic Viscosity
As was pointed out in Section A in connection with Fig. 4, light-
scattering measurements of dimensions are attended b y large errors in
the range of low molecular weight, which is unfortunately the most
important range for achieving a rehable estimate of unperturbed dimen-
sions by extrapolation to zero molecular weight and hence to zero
excluded volume. A method of wider applicability is therefore needed to
investigate the conformations of various chain molecules, and for this
reason we have tried to extend the viscosity method.
According to perturbation calculations based on the Kirkwood-
Riseman model [Kn~KW00D and RISEMAN (139); and YAMAKAWA and
218 M. KURATA and W. H. STOCKRIAYER:

KURATA (268)], the intrinsic viscosity of chain molecules is expressed in


a power series of the excluded volume parameter z defined by Eq. (16):
i.e.
[~1] = [~1]o [1 + p (X) z . . . . ], (49)
[r/]o = (~l,Na/lO0) EXFo(X)] (<S2>o'l,IU), (50)
X = ~ N'/'/6'l'~'/'arlo, (51)
where N• is the Avogadro number, X is a parameter representing the
draining effect of the solvent molecules, ~]o is the solvent viscosity and
is the friction constant of a segment of the equivalent random chain.
The functions X F o (X) and/~ (X) both increase with X or M, and are
fairly close to their respective asymptotic limits, 1.259 and 1.55, when X
exceeds about ten [see AUER and GARDNER (27); and ZIMM (272)]. As
has been pointed out by FLORY and F o x (103), this limit for Xcorresponds
to about 50,000 for the molecular weight if normal values are assumed
for the length and the Stokes law radius of a segment, and consequently
the Kirkwood-Riseman equation (50) becomes practically identical to the
Flory-Fox equation, (6) and (7). However, the value of ~b thus obtained is
q~o = (~'/'N~i/6,f'" 100) [XFo(X)]x=o~= 2.87.10 ~1 , (52)
which is higher than the empirical value originally recommended by
FLORY and Fox, i. e. 2.1--2.2" 1021. On the other hand, the value of
p (X) in Eq. (49) is a little lower than the corresponding coefficient, 1.91,
in the power series for a a, indicating that the hydrodynamic radius of
polymer chains increases less rapidly than the statistical radius as the
excluded volume increases 1. In other words, the equation (49) is not
identical to Eq. (5), but has to be written as
[,fl = [n]o~., (53)
With 0%<: ~. Accordingly, if the Flory constant ~ is again defined by
Eq. (4), we obtain from Eqs. (50), (52) and (53)
g} = ~Po(%/.)3. (54)
This equation indicates that 9 is not strictly constant but decreases
with increasing molecular weight and solvent power. A similar conclusion
has also been advanced by PTITSYN and EISNER (217, 218) for essentially
the same reason, although their calculation differs in detail.
The value of p (X) obtained by the perturbation theory gives, of
course, only the initial departure of ~n from ~. An unlimited number of
1 An exact asymptotic value of p (X) is still unknown, b u t it is expected to be
somewhere between 1.55 and 1.72, the upper limit being suggested by STOCKMAYER
and ALBRECItT'S perturbation theory of the friction constant (239).
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 219

expressions in closed form could be constructed to agree with this result,


but intuitively only two seem attractive. An exponential type of relation z,

was originally suggested b y YAMAKAWA and KURATA (268, 757) and


invoked in the previous application [KuRATA, STOCKMAYER and R o m
/-2 L L t I [ I

/ / "',,~ M =/././0 ~
_\,%,~| + M = 4 . 6 . / 6 ' 6"
\,~x,r per
"'~.,,~. I 4 = ,z g , I0 b` '
LO

'~ 0..,o o ,,,,$. - , , , , - - - . - - - . ~ . _ -. o

+%,+ ~
ft.8 "- 9 o ~ 9 -

0. 7 I I I "'-. I t ~''1-~
d..8 /.0 /.2 /.# /.F 18 2.U

Fig. 8. Variations of viscosity p a r a m e t e r # with expansion factor. Experimental data of Sc~uz.z (229) (circles)
and KRmBAU~t and CA~tPENV~R(145) (crosses). Theoretical curves according to Eqs. (55) and (56) for homo-
geneous polymers {broken curve and chain curve) and for broad fractions (dotted curve and full curve)

(155)] of Eq. (25), b u t there are grounds, perhaps at worst only esthetic
or metaphysical, for preferring a modification of Eq. (25),
a ~ - - ~ = 1.10 g(a,~) z . (56)

Both equations predict the same initial dependence of a, on z near the


t h e t a temperature. I n good solvents, however, the former (55) predicts
an unlimited decrease of ~b, namely # = # o ~ -'t,, while the latter (56)
predicts an asymptotic limit of ~ as a approaches infinity:

g}oo= 2.87(1.10/1.276) 9 102z = 2.48. I0 ~ . (57)

This value is in excellent agreement with Flory's empirical value,


2.5 9 10al, for homogeneous polymers (5e). The values of gi/~b0 calculated
b y Eq. (54) with Eqs. (55) and (56) are shown in Fig. 8 b y the broken
line and the chain line, respectively. Heterogeneity in molecular weight,
of course, has m u c h influence on the quantitative value o f #. I f this
x The originally suggested exponent on the right side of Eq. (55) was 2143, but
the practical difference between this and the round value 5]2 is insignificant, and
the theoretical value of p (X) is uncertain by a larger fractional amount.
Fortschr. Hochpolym.-Forsch., Bd. 3 IS
220 M. KURATAand W. H. STOCKMAYER:

effect is taken into account, as described in Section E, we obtain the


dotted line instead of the broken line corresponding to Eq. (55), and the
solid line instead of the chain line corresponding to Eq. (56). These are
to be compared with the experimental values of q~ which were recently
obtained by SCHULZfor poly(methyl methacrylates) (229) and which are
shown by circles in Fig. 8. Though the observed upturn in the q) vs. a
plot in region of relatively large a cannot be explained by the present
theory, the general trend of the data seems to favor Eq. (56) over Eq. (55).
In the figure, the experimental data for polystyrene by KRIGBAtI~t and
CARPENTER (146) are also shown by crosses.
In view of the foregoing result, we here adopt the four equations,
i. e. Eqs. (6) and (7) with the modified constant q50, and Eq. (53) and
Eq. (56), as the basic equations for further investigation of the intrinsic
viscosity.
Now, substituting Eq. (6) and Eq. (53) into Eq. (56), we obtain after
simple rewriting

[,13"~,1M'/, = K "1, + 0.363 CoB [g (~z,~)M'/qErl]v. 3, (58)


where K is given by Eq. (7) or
K'/. = +o (59)

The parameters A and B are defined by Eqs. (33) and (34), respectively,
and the constant q9o is 2.87 9 1021for homogeneous polymers [see Eq. (52)].
The effect of molecular-weight heterogeneity is considered in Section E.
The above equation (58) offers a method, pelhaps of wide applicability,
for determining the two parameters A and B from intrinsic viscosity-
molecular weight relationships alone. We may first plot [,I]*/~/M '/,
against M'l,][~] v', ignoring g (an), and find an approximate value of K
which permits us to evaluate [~?]o, a,l and g (a~) from Eqs. (6), (53) and
(26). Then, using this result, we m a y replot [rl]fl'/M'l" against g (an)M'/'/
[rl~'l,, and find an improved value of K. The convergence of this procedure
is rapid enough so that the final value of K is attained in the second trial.
Figure 9 shows the application of this method to the viscosity data
of poly(ethyl methacrylate) obtained by CHINAI and SAMUELS (61),
where the black circles represent the theta-solvent data and the white
circles represent the good-solvent data. The lines have a common
intercept, leading to a single value of K independent of solvent, in
contrast to the Flory-Fox-Schaefgen plot for the same data which was
shown in Fig. 1. The value of K obtained from the intercept in Fig. 9 is
(49• 3). 10-5, which is a little higher than the simple mean of the
observed [fl]o/M~ 'l' values, i. e. 47.3.10 -5, reflecting the very small
negative slope of the best fit line for the theta solvent results. Such a
Intrinsic Viscosities and Unperturbed D i m e n s i o n s of L o n g C h a i n M o l e c u l e s 221

small difference as this is, of course, insignificant, but it illustrates the


accuracy of the present estimate of K.

LE PoIyethylmethaarl.,lafa ~

o~

t~
zo

f I I t
0 ,V 10 I,E ZO
gx~}M~'/{W ''~9 ~a- s
Fig. 9. Viscosity plot for poly(ethyl methacrylate) according to Eq. (58). Data of CHmAXand SAMUELS(67)

(alactic/) _

f / ' / <
I

o Folvsne
9 ~s

o Z~oamylacetste
a _______________L_
0 2.$ $.0 0 / a
M~/[,l) 9 -~
Fig. 10. Viscosity plots for atactic polypropylene according to Eq. (58) (left-hand figure) and Eq. (9)
(right-hand figure). Data of DANUSSOand MORAGLIO(7d)

Figure 10 displays a similar application to the recent results for


atactic polypropylene by DANUSSO and MORAGLIO (Td). The large black
circles represent the measurements in cyclohexane at 30 ~ C, the large
15.
white circles those in toluene at 30 ~ C, the small black circles those in
benzene at 30 ~ C, and finally the small white circles in theta-solvent
data in isoamylacetate at 34 ~ C. As shown in the left half of the figure,
the plots of [,I]':,/M 'h against g(o,,,)M'h/[,l] '/, give a set of lines with a
common intercept, indicating that the quantity K, and hence the un-
perturbed dimension of this polymer, is practically independent of
solvent. The value of K, obtained from this intercept is (160 -1- I0) 910-s,
which is included in Table 7 in the next chapter 1. In the right half of
Fig. 10, we illustrate the Flory-Fox-Schaefgen plot of the same data.
The four lines no longer
T a b l e 4. The apparent K values for polystyrene as display a common inter-
function of solvent power cept, but cross each
Tolueno-Methaaol Chloroform-Methanol other in a regular way,
xC~pp ,, K~pp the intercept progres-
sively decreasing as sol-
6.0" 10-* 0.76 5.5" 10-* vent power is increased.
0.75 5.8
0, i00: i
0.2 7~
7.6 0.68 6.3 This indicates that the
0.54 7.9. intersection of good-sol-
vent lines with poor-sol-
vent lines in the diagram is not accidental, but has its origin in
the approximate nature of the Flory equation (8).
In this connection, we m a y also refer to the work of OTH and DES-
I~EUX (206). They measured the intrinsic viscosity of polystyrene in
various mixed solvents and evaluated K from plots of [~]'h/M'[, against
M/[~/1. The results are given in Table 4, together with the exponents v
in the Mark-Houwink-Sakurada equation (13) as a rough measure of the
solvent power. Here r represents the volume fraction of methanol in the
mixed solvent employed. It is again clear from this table that the smaller
values of the apparent K are found in the better solvents. On the con-
trary, the present method, if applied to these data, yields the unique
Value K = 8 . 2 - 1 0 -~, regardless of solvent. There is, of course, no
absolute basis for believing the unperturbed dimensions to be independent
of solvent. However, at least in the case of non-polar polymers, the
existence of a common intercept to within certain limits, say :k5%,
m a y be regarded as lending support for the present viscosity theory.
Polystyrene prepared b y the Szwarc method (247'") is known to have
a very sharp distribution of molecular weight. In other words, this
polymer is almost free from the uncertainty due to heterogeneity which
affects both the intercept and slope of the correlation line between
[7] 'h /M ~], and g ( ~ ) M *h/[9] 'h , and it m a y b e adopted as an ideal example
z Similar a n a l y s i s of KINStNGER a n d HUGHES' d a t a for b e n z e n e a n d e y c l o h e x a n e
s o l u t i o n s of t h i s p o l y m e r (138) g i v e s a s l i g h t l y lower value, K = (150 :t= 20) 9 10 -5.
A n a v e r a g e v a l u e of t h e s e t w o e s t i m a t e s is g i v e n in T a b l e 7.
I n t r i n s i c Viscosities and U n p e r t u r b e d D i m e n s i o n s of L o n g Chain Molecules 223

for investigating the viscosity behavior of chain polymers. For this


polymer, three independent sets of measurements of viscosity and
molecular weight have been performed in the past several years [McCoR-
MICK (178); MUKHERJEAand REMPP (193); and MEYERHOFF (182)] 1.
These are plotted in Fig. 11. Superposition of the data from different
sources is excellent, and we obtain K = 8.2 9 10-4 from the intercept and
B = 1.05 9 10-27 (in toluene) from the slope. This value of K is in good

Z,.O i I l )
o

PolF s f F r e n e - To /uene

LE
of+ o 9

~/ + o 2~~ (Me CgRmCK]


1.0 +.,,~.o~"o~ * ) Z,O~ ['MEYERtlOFF,)
/ + 20 ~ {MtlKIIERJEA- REMt,'IP)

0.5 I I I l
0 s
g(~) M'/3/[,./),/~.
~ 6
~o-~
8 10

Fig. 11. Viscosityplot for polystyrene polymerized by anionic initiators

agreement with the previous estimate from the data of OTH and DESREVX
and also with the direct result based on measurements in theta solvents
[KI~IGBAUM and FLo~u (149)].
It is instructive to reverse the above calculation. With the values of
K and B just determined, the intrinsic viscosity of the polystyrene-
toluene system as a function of the molecular weight can easily be
calculated from Eq. (58). The parameter z m a y first be evaluated from
Eq. (16), which may be rewritten
z = 0.330 B A -a M '1, , (60)
and ana can then be found from Eq. (56), most easily b y the use of a
graph of as~ against z. The intrinsic viscosity is finally obtained a s
KMV, ga,. The result of this calculation is given in Fig. 12 b y the solid
curve, which is seen to be in very good agreement with the experimental
data, not only in the ordinary range of molecular weight [MCCORMZCK
z See also, COWlE, WORSFOLD a n d BYWATER (71"').
224 M. KUR&TA and W. H. STOCKMAYER:

( 1 7 8 ) ; ~V[EYERHOFF(182)] ~ but also in the short-chain region [MARZOLPH


and SCHULZ (175)]. The decrease in the stope of the double-logarithmic
plot at low molecular weights is naturally and faithfully reproduced. This
would be equally true of the F l o w - F o x theory, as represented by Eqs. (5)
to (8). There seems to be absolutely no evidence for partial draining of the
solvent through the polymer coils, such as would be predicted by the
theories of BRI~KMAN (45'), DEBYE and BUECHE (7Q), KIRKWOOD and

,ool~,~typen8 .~\'/ /
r +/ .
/
9,,~,, ~o§ ....7,;~oC.,
/ ~,~_~-'--""" ~'
%
---~ 0./
. / ~-7,.;.<S....
/~<::.':.-" o mRZoLP//-SU/o,.Z

#.0/
....~-
l/J"
I I 1 I
200 /0 3 /0 ~ /0 ~ /0 o /0 z
M~ op MSD
Fig. 12. Viscosity-molecularweight relations in polystyrene. The three dotted curves show the effect of
partial draining according to the Kirkwood-Risemantheory, Eqs. (S0), (51) and (62), with three different
choices of drainage parameter

RISEMAN (139) or KUHN and KUHN (153', 183"). For t h e sake of com-
parison, Fig. 12 also shows the theta-solvent intrinsic viscosities of poly-
styrene in cyclohexane [expermlental values of KRIGBAUMand FLORY (149),
small black points; theoretical values, broken line] and the theoretical
intrinsic viscosities of rigid ellipsoids with axial ratios p = M/500 (chain
curve). As a matter of course, the chain curve reduces to the Einstein
value of [~/] in the range of M below S00 [see, for example, PETERLIN (145)]2.
1 N o t e t h a t b o t h McCoRmiCK and MEYERFIOFF a c t u a l l y d e t e r m i n e d t h e mole-
cular w e i g h t M s ~ f r o m s e d i m e n t a t i o n velocities a n d diffusion coefficients, b u t for
S z w a r c - t y p e p o l y s t y r e n e s the difference b e t w e e n M s D a n d M , can safely be ignored
in a log-log r e p r e s e n t a t i o n such as Fig. I2.
2 T h e E i n s t e i n value of [7/] is 0.025 Q~ w h e r e Q, is the d e n s i t y of t h e solute
molecule, equal to 1.05 g./ml, for polystyrene. The value of 500 was r a t h e r arbitrarily
chosen for t h e p u r p o s e s of illustration, b u t it h a p p e n s to be quite close t o t h e molar
w e i g h t of t h e so-called preferred statistical s e g m e n t of p o l y s t y r e n e [see HURATA and
YAMAKAWA (157)],
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 225

To investigate further the absence of the draining effect, three


dotted curves are also given in Fig. 12. These correspond to the intrinsic
viscosities in the theta solvent predicted b y the Kirkwood-Riseman
theory (139) according to Eqs. (50) and (51) 1 with three different choices
of the drainage parameter X. I t is convenient to express the friction
constant $ of a chain segment in the form of the Stokes law,

= 3~,]o b , (61)
where b is the effective diameter of a segment. The drainage parameter
then becomes
X = 0.691 (b/a) N 1I,,
which m a y also be written
X = 0.691 (blAres) M '/' (62)

with the aid of Eqs. (17) and (33). In this last form, A is the parameter
of Eq. (33) and ms is the molar weight of a segment. I t m a y be pointed
out that although some latitude is available for choosing the size of a
segment, it is only the ratio b/ms which appears in the equation for X,
so that the latter is physically restricted to a rather narrow range for a
given polymer. The lowest of the dotted curves in Fig. 12 is obtained
with the choice X M -q, = 0.10, which corresponds to m8 = 104 (the
monomer molecular weight) and b = 10 Angstroms, a reasonable physical
value. The second curve is for X M -1[" = 0.32 and the third (highest) is for
X M -'l" ---- 1.0, which would correspond to an unreasonably large fric-
tion constant per segment. However, only this last curve remains
acceptably dose to the experimental points at low molecular weights;
and even here the latter pass smoothly over to the Einstein limit of
the ellipsoidal curve without any visible downward trend. We are,
therefore, led to conclude that the drainage parameter of the Kirkwood-
Riseman theory must be taken at least ten times larger than would be
expected from the physical dimensions of the polymer chain, in order
to avoid conflict with the experimental results for polystyrene. This
conclusion is not restricted to the data of one investigator, for recently
RossI, BIANCHI and BIANCHI (221) have found that [~/]oM -v, remains
constant down to a molecular weight of 700 for polystyrene in toluene-
methanol theta-solvent mixture. An entirely similar conclusion m a y be
reached for other flexible chain polymers, including poly(ethylene oxide),
I The function X F o (X) as tabulated in the original paper of KIRKWOODand
I~ISEMAN(139) is somewhat too large numerically. However, the correct calculation
has been made only for the limit of very large X by AVERand GARDNER(27) and
by ZIMM (272). We therefore assume, as did YAMAKAWAand KURATA(157), that
the ratio of X F o (X) to its limiting value is correctly given by the original tabulation ;
i. e., the original ordinate scale has been contracted by the factor 1.259/1.588. The
error of this step is greatest at low X.
226 M. ~UP.ATA a n d W . H . STOCKMAYER:

which has no large substituents. The calculations for this case, which
werebased on the data of SAI)I~ONand REMPP (223) will not be reproduced
here. Avoiding any discussion of the possible physical or mathematical
short-comings of the Kirkwood-Riseman theory 1, we shall simply accept
the experimental evidence that the draining effect is much smaller than
had originally been supposed, so that the molecular weight at which
it m a y manifest itself is some hundred times 2 lower than that correspond-
ing to predictions based on the physical dimensions of the polymer chain.
This point will become rather important in our later discussions of
cellulose derivatives in Section III B.
The effects produced b y our modifications of the Flory-Fox approach
are greatest in good solvents. In the limit of extremely good solvents,
we m a y neglect a~2 compared to unity, and then Eq. (56) simplifies to
a~ --- 1.69 z. Similarly, the first term in the fight-hand member of Eq. (58)
then m a y be dropped and g(~o) allowed to attain its limiting value. This
gives
[~] = 0.558 qboBM ~ (63)
or, with Eq. (52),
[~] = 1.60" 1021 B M . (63')
This is, of course, the historically famous Staudinger equation. It ]s
well known that the intrinsic viscosity of a solution of completely ]ree
draining Gaussian chains [KuHN (153) ; HUGGINS (728') ; KRAMERS (141') ;
and DEBYE (87)] also follows a Staudinger law:
[~] = (Na$/3600 ~oms)<L~>o, (64)

where m~ = (M/N), as before, is the molar weight of an effective segment.


If Eq. (61) is used for the segment friction constant, this equation
becomes
[V] = 1.57.1021 (ba2/m]) M . (65)

At first glance there is a remarkable resemblance between Eqs. (63')


and (65) in spite of their quite different origin. Closer scrutiny, however,
reveals two important differences. First, the coefficient of M in Eq. (65),
i. e., (ba2/m~), is essentially a geometric quantity and consequently
under some restriction. For instance, a and b m a y not both be as small
as 0.1 A. At the same time, the ratio a/b must be quite large, say 50 or
x C o m p a r i s o n s of t h e d a t a w i t h p r e d i c t i o n s b a s e d on t h e t h e o r y of BRINKMAN
(45") a n d DEBYE a n d BreECH~ (79) or o n t h e m o d e l e x p e r i m e n t s of KUHN a n d
IEUHIV (753") lead to v e r y s i m i l a r conclusions.
z S t r i c t l y t h i s figure is b a s e d o n t h e c a l c u l a t i o n of Fig. 12 a n d o n t h e pro-
p o r t i o n a l i t y b e t w e e n X a n d MI/=. F o r n o n - t h e t a c o n d i t i o n s t h e r e 'is n o real t h e o r y ,
b u t we m a y r e a s o n a b l y m o d i f y E q . (62) to
X = 0.691 (5/.4 m,~n) MV,. (62')
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain NIoleculos 227

100 at least, to secure the free-draining approximation in the ordinary


range of molecular weight, say 105 to l0 s [see Eq. (51)]. These two
restrictions are often incompatible and t h u s prevent wide application of
Eq. (65). On the other hand, the quantity B in Eq. (63') is an effective
excluded volume of a segment in a thermodynamic sense, and hence is
comparatively free of such restrictions Esee belowJ. As a result of this
difference, it is possible to interpret solvent effects on the Staudinger
coefficient K ' in t e r m s of B, but Eq. (65) offers no such possibility.
Second, the free-draining result has to be interpreted as an asymptotic
result for small values of the drainage parameter and hence for low
molecular weights. Thus, if a Staudinger law is known to hold in a certain
range of molecular weight and is interpreted by means of Eq. (65), then
deviations to a smaller viscosity index v would be expected one or two
decades higher in molecular weight. On the contrary, Eq. (63) is the
limiting law for large values of z, and so the deviations should occur only
at low molecular weights. In other words, if the Staudinger rule is
reached at a certain molecular weight, then according to Eq. (63) no
deviation is expected to occur at an 3, higher molecular weight.
According to the theory of FLORY and F o x (103), the exponent
in the Mark-Houwink-Sakurada equation (13) must lie between the limits
0.5 and 0.8 for linear flexible chains without draining effects. The lower
limit is for rather tightly coiled chains in theta solvents and the upper
limit for highly swollen polymers in very good solvents. A violation of
this upper limit must then be interpreted as evidence for the draining
effect. Furthermore, this violation is also usually regarded as evidence
for high "stiffness" of chains, since the draining effect should be greater
for more highly extended molecules. In fact, most synthetic linear
polymers have v values somewhere within the Flory-Fox limits, while
a number of cellulose derivatives show v values higher than 0.8. This
circumstance has long been regarded as support for the Flory-Fox
theory, because the idea that certain cellulosic chains are especially
stiff is commonly accepted. However, this view seems less firmly estab-
lished if we inspect the recent accumulation of Mark-Houwink-Sakurada
constants. Table S gives a partial list of the polymer-solvent systems in
which v values larger than 0.8 have been observed. Some cellulosic
materials have been omitted from this table, but they may be found in
the more complete table given in the Appendix. It m a y be seen from
Table 5 that violations of the upper Flory-Fox limit are by no means
confined exclusively to cellulose derivatives or other natural polymers,
but are also found on occasion for macromolecules conventionally and
justifiably regarded as flexible chains. It seems unlikely that the large
values in these cases can be interpreted in terms of abnormal chain
stiffness.
228 M. KURATA a n d W. H. STOCIKMAYER"

The present theory extends the upper limit of ~ for non-draining


molecules to 1.0, and thus at least in principle it permits any flexible
polymer to attain such a large value of v in very good solvents. However,
in the case of non-polar polymers there are limitations to the magnitude
of the binary cluster integral /3 of Eq. (14), and hence to that of the
parameter B [see Eqs. (34) and (58)3 which controls the departure of
from the Value 1/2. Since the London dispersion forces between unlike
non-polar molecules obey roughly a geometric-mean combining rule,
the heats of mixing of non-polar liquids are almost always positive
Table 5. Polymer-solvent systems characterized by a relatively large exponent v in the
Marh-Houwinh-Sahurada equation
Polymer Solvent Temp. K'- 10s. v M.Rangel0-* Investigator

Polyvinyl- Tetrahydro- 20 3.63 0.92 3 - - 17 ]3ATZER et al.


chloride furan (33)
Polymethyl- Chloroform 25 [ 3.40 0.83 ~1--326 CmNAI et al.
methacrylate 13.3 (59)
Cellulose Cupriethylene- 25 0.905 2 - - 54 IMMERGUT et al.
diamine I (732)
Cellulose Acetone 25 7.00 0.933 4 - - 49 HARLAND (9)
trinitrate
Amylose Dimethyl- 25 1.25 0.87 !2--305 CowIE (71)
sulfoxide 21
Poly-y-benzyl- Diehloroacetie 25 2.78 0.87 2 - - 34 DOTY et al. (85)
L-glutamate acid
Gelatin A q u e o u s salt 35 1.66 0.885 5-- POURADIER
et al. (215)

(endothermic), and the excess free energies of such mixtures are also
generally positive [see, for example, HILDEBRAND and SCOTT (70);
ROWLINSON (/9'), and also SMALL (234')1. This implies that the Flory
parameter Z which measures the local free energy of mixing in polymer
systems should also be positive and not too much smaller than 0.5. In
polar fiquid mixtures, on the contrary, a much wider range of excess free
energies is possible, and negative values are not uncommon. As a corol-
lary, Z should also display a wider range, with occasional negative
values, in polar polymer-solvent systems. This is in fact borne out by a
table of values of such interaction parameters [see HUGGINS (12),
pp. 48--49, who uses the symbol/D for the quantity which Flory calls Z],
and consequently according to Eq. (21) much larger /3 values should
be possible for polar polymers than for any non-polar ones. It is, there-
/ore, not surprising to find that the high v values, given in Table 5, are
all for quite polar systems. It may also be concluded that the temperature
dependence of B and of ~ will be more varied and extreme for polar
polymers.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 229

Another argument which has sometimes been used in favor of a


stiff-chain model concerns the ratio of intrinsic viscosities of the same
polymer in two different solvents. If this ratio does not depend on
molecular weight of the polymer, as is the case for certain cellulose
derivatives [HOLTZER, BENOIT and DOTY (125); FLORY, SPURR and
CARPENTER (/07)] the viscosity exponent v must be not far from unity,
and this has been taken b y several authors to mean t h a t the long-range
interaction factor ~ is not playing a dominant role in determining the
intrinsic viscosities. Again, the present theory permits an alternative
interpretation. If two very good solvents are involved, the ratio of
intrinsic viscosities is simply equal to the ratio of the two B values, and
invocation of chain stiffness due to short-range interactions is not required.
In summarizing the intrinsic viscosity relations presented in this
section, it must be admitted that they represent nothing more than
rather small semi-empirical refinements of the Flory excluded volume
theory and the Flory-Fox viscosity theory. For a large fraction of the
existing body of experimental data, they offer merely a slight improve-
ment in curve-fitting. But for polymers in good solvents it is believed
t h a t a more transcendental result has been achieved. The new equations
permit more reliable assessment of unperturbed chain dimensions in such
cases, and in some instances (e. g., certain cellulose derivatives; see
Section I I I B) they offer possible explanations of heretofore paradoxical
solution behavior.
D. Friction Constant
A t r e a t m e n t similar to the foregoing can be applied to the friction
constant [o, which is related to the sedimentation constant s and to the
translational diffusion constant D b y

s o = M(1 -- ~o)/Na[o (66)


and
O o = k T/I o , (67)
respectively, where ~ is the partial specific volume of the polymer,
e is the solution density, and the subscript zeros attached to s, [ and D
refer to infinite dilution [see, for example, BALDWIN and VAN HOLDE (1)].
Perturbation calculations of [o for flexible chains with excluded volume
have been carried out, b y KURATA and YAMAKAWA (157), and more
rigorously b y STOCKMAYER and ALBRECHT (239) ; the latter give

1o1~o = KIM'I'o~, (68)


with
K I = 6'I' Po ( ( S * ) o / M ) '1' , (69)

a l = 1 + 0.609 z . . . . (70)
230 M. KURATA and W. H. STocK~AV~R:

The theoretical value of P0 is 5.11 in the limit of the impermeable coil.


In contrast to the case of viscosity, the difference between a: and
[see Eq. (18)] is insignificant. In good solvent systems, we m a y again
assume an equation of the form of Eq. (25),

r162 ~: = 1.22 g(~:) z , (71)

which yields (o=:/o=)= 0.985 in the limit of very good solvent. The solvent
effect on the constant P, therefore, does not exceed two per cent, and
it is completely negligible.
As shown b y ]V~ANDELKERNand FLORY (171), the combination of
Eqs. (53) and (68) yields
M = {So [V]':"VoNa/fl(1 - ~e)}':' (72)
with
fl = q~o:"p~a ( o~/ o~:) . (73)
The new factor o:Jo~: is unity in theta solvents and 0.97 in the asymptotic
limit of very good solvents. This indicates that the quantity /q varies
only from 2.7. 106 to 2.6.108 with increasing solvent power. The
molecular weight obtained b y Eq. (72) is thus affected to only 5 or 6 %
9 b y the variation of fl, and the effect may be safely ignored for most
purposes.
Finally, the present theory predicts that
v: = 0.5 ~ 0.667 in [o-----K ~ M q , (74)
and
vs = 0.333 ~ 0.5 in s o = K ' ~ M ~, . (74')
If we neglect the small difference between x and a:, we have
n = (2 - ~)/s. (7s)

According to this equation, the lower limit of I/3 for ~s is reached in


very good solvents when v approaches unity. This is in marked contrast
to the free-draining theory of the Staudinger rule, according to which v~
must tend to zero as v tends to the Staudinger limit.
Intermediate situations between complete draining and total im-
permeability can be investigated if Eq. (62') is adopted for the draining
parameter X in good solvents. If the expansion factor is taken to v a r y
empirically with molecular weight according to an exponential law,

o~/ = K ' M ~c` , (76)


then a simple application of the Kirkwood-Riseman relations leads to
the results
v = e + (4 -- 2e) v~,, (77)
Intrinsic Viscosities-~.ndUnperturbed Dimensions of Long Chain Molecules 231

and
v, = (1 - 2 v=) e s , (78)

In these equations, ~ and es are the values of v and v,, respectively, in


the free-drainh~g case. The former quantity is tabulated b y KIRKwool)
and RISEMAN (#39) and the latter is easily found from their equation for
the friction constant.
Some of the foregoing relations are illustrated in Fig. 13, where v, is
plotted against v. The solid line is that of Eq. (75), which terminates at
~,~' J I t I t I ] i I I I l i i

3.5
65 i'"/-...
, .... ,',3

od ""-.. /0

kz Sd ~ bR "-\ "'"'..~. aO

\ x\
0.~ ~ ~" ~\ \ "''"
"~ \
...\
P ~-a~">,
/ 0 0 --",,

0 t
0,5.
I I [ 'LO
t
I
~ t
I
I
~ T!
t
I
I I
L5
I i I I "J
2.0

Fig. 13. Molecular-weight dependence of sedimentation constant (vt) and intrinsic viscosity (~), for various
degrees of draining and coil expansion. Full line is for coiled polymem without draining. Dotted curve is for
rigid ellipsoids of revolution at various axial ratios p. Experimental points: a, cellulose nitrate in ethyl
acetate {729); b, cellulose nitrate in acetone (/8/}; c, cellulose acetate in acetone (#26); d, ethyl cellulose in
ethyl acetate (22J'); e, ethyl hydroxyethyl cellulose in water (172)

the black point marked c in the limit of very good solvents according to
the present theory. The Flory-Fox theory follows the same line, but
only as far as the shaded barrier sketched at the point where v reaches
0.8. The left-hand broken curve follows the original relations of KIRK-
WOO1) and RISEMAN for chains without excluded volume at different
degrees of draining. The other two broken curves are calculated b y
Eq. (77) and Eq. (78), and correspond to the exponents v~ = 0.10 (good
solvent) and v== 1/6 (limit of extremely good solvent). The arrows drawn
on these curves show the direction of increasing molecular weight, and
heIp to emphasize the difference between the effects of excluded volume
and of partial draining. The two chain lines with arrows sloping down-
ward to the right connect points corresponding to two different constant
values of the draining parameter, X = 0.3 and 0.1. The circles marked a
232 M. KURATAand W. H. STOCKMAYER:

to e in the figure indicate published experimental results for several


cellulose derivatives. With the exception of point e, they can be fairly
well accommodated by a non-draining model with v~ = 1/6 (limit of the
present theory), or by a model with some draining and with v~ = 0.10
(limit of the' Flory excluded volume theory), or b y some intermediate
situation. A more complete discussion of these cases is reserved to
Section III B.
Finally, the dotted curve in Fig. 13 traces the relation between v
and vs for rigid prolate ellipsoids of revolution [see PETERLIN (16) or
FRISC~I and SIMHA (6') ] with axial ratio proportional to molecular weight.
This curve lies very far from those for flexible molecules except for very
low values of the axial ratio p. This seems to exhaust the available
information of the type represented by Fig. 13. In connection with the
behavior of D N A and perhaps other naturally occurring macromolecules,
it would be interesting to have calculations for rods with one or two or
at most a small number of flexible joints, such as might correspond to
almost completely helical structures [see Section III ol. In spite of the
absence of theories for this and possibly other relevant molecular models,
it is often possible to arrive at useful indications of conformation by
comparing the experimental data with Fig. 13.
It is an easy matter also to investigate the effects of partial draining
on the Mandelkern-Flory number ft. From Eqs. (72), (74), and (74')
we have
d In rid l n M = v, + (v - 2)/3, (79)
and it turns out that this quantity remains very small (less than 0.05
in absolute value) if X is greater than unity, whatever the excluded
volume. Thus, although /~ must increase indefinitely as free draining is
approached, this occurs only at such low values of X as to be experimen-
tally rarely, if ever, observable. Thus any appreciable increases of /~
above the value 2.6. 106 for flexible coils is to be regarded as rather good
evidence for a more rod-like conformation [see SCHERAGA and MANDEL-
KERN (224a)]. Conversely, molecular weights estimated b y Eq. (72) with
the conventional value of fl are unlikely to be erroneous because of the
draining effect.
E. Heterogeneity Corrections
The existence of a distribution of molecular weights in most polymer
samples must be taken into account in quantitative studies of solution
properties. The definition of the viscosity-average molecular weight and
its relation to other molecular weight averages has been described b y
m a n y authors; see, for example, FLOltu (5). We m a y also refer to the
papers by SHVLTZ (233) and NEWMAN, KRIGBAUM, LAUGIER and FLORY
(199) concerning the relation between the intrinsic viscosity, the radius
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 233

of g3,ration and the number 4 ; to the paper of GOUINLOCK, FLORY and


SCHERAGA (11Q) concerning the relation between the friction constant,
the radius of gyration and the number P ; to the paper of KRIGBAUM
and CARPENTER(ld6) for a treatment of the perturbation theory of (S ~)
and [2/]; and finally to the paper of YAMAKAWA and KURATA (269)
concerning the perturbation theory of the osmotic second virial coefficient
and the intrinsic viscosity for partially draining molecules. The calcula-
tions are rather trivial and are not reproduced here. Our choice of Eq.(56),
however, depends crucially on the quantitative behavior of 4 , and for
this reason we give here a brief account of the heterogeneity correction
made in Fig. 8. We describe also the effects of heterogeneity on our final
tabulations of unperturbed chain dimensions.
To treat the effects of heterogeneity on 4 , we take the familiar
exponential function [ScHuLz (228)] for the distribution of molecular
weights,
w ( M ) = [yh+I/F(h + I)] M ~ e x p ( - - y M ) ,
with
y = h i M , = (h + 1)lMw = (h + 2)IM , . (80)
Here the subsripts n, w and z have their usual meaning, and F represents
the gamma function. We now denote b y 40 w the value of the viscosity
constant 40 that would be obtained from measurements of M,~ and of
(S2)w under theta conditions, and by 4o, the value similarly obtained
from measurements of M w and of (S2)z. These are found to be
4o~ = q~tbo, q~, = / ' ( h + 1.5)/(h + 1)'h F ( h + 1), (81)
40z = q,4o; q, = [(h + 1)/(h + 2)]'I,q~. (sv)
For the "most probable" distribution specified by h --- 1, these equations
give q~o= 0.94 and q~ -----0.51 ; and for a typical, reasonably sharp fraction,
say h = 4, they give q~ = 0.98 and q, = 0.74. The correction factor q~
is obviously very close to unity and can be ignored for fractionated
materials. The correction to t5 (X) in Eq. (49) is almost equal to q~-Xif the
excluded volume parameter is expressed in terms of M~, and hence the
numerical coefficient in Eq. (56) for a n is replaced b y 1.10 q~l for het-
erogeneous polymers.
The correction to the observable radius of gyration, on the other
hand, is given by
(S~)0z = [(h + 2)/(h + 1)] (S2)0w = [(h + 2)/(h + 1)~ (AgMw/6), (82)
and
# = (S~),/($2)o,--- 1 + q,(134/105) z~ . . . . , (83)
with
q, = / ' ( h + 3.5)/(h + I) 'h (h + 2) F ( h + 1). (83')
234 M. KURATA a n d W. H. STOCKMAYER:

Accordingly, the coefficient C in Eq. (25) must be replaced b y 1.276 q~


for heterogeneous polymers. The above equation gives q~ = 1.37 for
h = 1 and q~ = 1.16 for h = 4; these are considerably larger than the
correction required for ~n. As a result of the difference between these
two corrections, the variation of ~b/r o with c~ is generally reinforced b y
heterogeneity. The solid and dotted lines in Fig. 8 correspond to the
choice h = 4.
The effect of heterogeneity on viscosity-molecular weight rela-
tionships, i. e., on the relation of the viscosity-average to other average
molecular weights, is a classical problem [see, for example, REHNER
(219") or FRANK a~qd ~REITENBAGH (713)], and it is now well recognized
that usually the viscosity-average molecular weight lies quite close to
the weight average. For example, even under t h e t a conditions when the
Mark-Houwink-Sakurada exponent is 1/2, the ratio M~/Mw is equal to
q~, which is not much smaller than unity even for the "most probable"
distribution, h = 1. In certain olefin polymers, however, the molecular
weight distribution appears to be extremely broad. The corrections in
such cases can be evaluated from the formulae given by CHIANG (53),
who used the revived logarithmic-normal distribution law of LANSING
and KRAEMER (764'), which was apparently first applied to Ziegler-type
polyolefins b y WESSLAU (263"). OccasionaUy we are obliged to work
with relations between intrinsic viscosities and number-average molecular
weights. Then extrapolation b y a graphical method such as that of FLORY,
F o x and SCI~AEF6EN or our modification based on Eq. (58) leads to the
quantity [~]o/M~, = Kq,~ = KI'(h + 1.5)/h 1/, F(h + 1), (84)

the last form being of course the special result for the Schulz distribution.
The correction here is quite large for poorly fractionated polymers; for
example, q~ is 1.33, 1.174 and 1.089 for h = 1, 2 and 4, respectively.
Fortunately, we are here eventually interested in the unperturbed
linear dimensions of chain molecules, as expressed b y the quantity A,
and this is seen in Eq. (59) to be proportional only to K 'l,. Therefore,
a fair estimate of chain dimensions can be based on viscosity relationships
which would be considered rather poor as a means of evaluating welt
defined average molecular weights.

III. Chain Conformations


A. Addition Polymers
In this chapter the viscosity method developed and illustrated in the
previous sections is systematically applied to the existing experimental
data, and the resulting values of the unperturbed chain dimensions are
tabulated for polymers of various kinds.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 238

I t is a familiar fact t h a t the skeletal factor s defined in Eq. (2) can


v e r y often be decomposed into two independent parts, depending
respectively on the restriction of b o n d angles and on the hindrance to
internal rotations. I n the simplest case of the p o l y m e t h y l e n e chain with
an internal rotational potential t h a t is s y m m e t r i c a b o u t the trans con-
formation, we have
s = [(1 + cos 0)/(1-- cos 0)] a ~ , (85)

where 0 is the supplement of the b o n d angle, and the factor a s is given


b y the simplest t h e o r y as
aS = (1 + (cos ~b))/(1 - (cos ~b)), (86)

in which ~b is the internal r o t a t i o n angle measured from the trans con-


formation [see, for example, FLORY (Sb)]. If the rotation is completely
unrestricted, (cos ~b) --- 0 and a becomes unity. If the restricting potential
is n o t everywhere zero b u t if it has threefold s y m m e t r y (i. e., three equal
m i n i m a at the trans and at the two gauche angles), the average value of
cos ~ is again zero, and the extension of the chain is the same as if the
rotation were free. Theories of the internal rotations in chain molecules
are now in a state of rapid further development, Esee, for example,
VOLKENSTEIN (21')] and are briefly discussed in Section I V B.
F o r more complicated chains such as cellulose derivatives, diene
polymers and polypeptides, the factor a m a y be defined in the more
general manner.
as= (LS)o/(L')o,, (87)
and m a y still be used as a measure of the hindrance to internal rotation
a b o u t the single b o n d s of the chain. T h e q u a n t i t y (L*)01 is the mean-

Table 6. Unperturbed dimension of freely-rotating chains


Chain type [(Lt)ot/M]'It. 10' (cm.mole*/*gram-t/t)

Polymethylene chain . . 3.08/M.q, 2.18/m11,


Amylosic chain . . . . 4.26[M,,'/~ 1.91/m'l,
Cellulosic chain . . . . 7.90[M,, ' , 3.83/mii,
Gutta-percha . . . . . 5.80/M,,'/, 2.90/m11,
Natural rubber . . . . 4.02/M,t/, 2.01/rntl,
Polypeptide . . . . . . 3.83]Mu'I, 2.21/ml/2

square displacement length corresponding to fixed valence angles b u t to


completely free internal rotations, and it has been calculated for various
types of chain b y BENOiT (37), b y MARKOVITZ (175a) a n d b y COWlE (71).
These calculations are summarized in Table 6, where M , represents the
molar weight of the repeating unit and m represents the average weight
per skeletal link; for example, m is t a k e n as Mu/S for a cellulosic chain
Fortschr. Hochpolym.-Forsch.,Bd. 3 16
23B M. KURATA and W. H. STOCKMAYER:

or as M,13 for a polypeptide. The figure in the last column of the table
thus represent the relative stiffness of the chain caused by such structural
features as rings, double bonds, etc.
Although the unperturbed dimensions of m a n y addition polymers
have been obtained b y means of theta-solvent experiments, there are
some important exceptions, for which the present method leads to useful
results. A notable member of this group is polyethylene. From the
~ < ' o
~ , .
, Polyethylene 9o

o ,/ 9 9

N J

N
o #hlopon~',ehlhale,Te 12~~
/ o ;o- X,vlene ,'00 ~
. Tetmel/n I3O~

a d 3
Jm M'l / ('1) I0"3
Fig, I I , Viscosity plot for polyethylene in several solvents

considerable body of data in the literature, we have selected three


independent sets of measurements, those in 0r at
125 ~ C b y ATKINS, Muus, SMITH and PIESKt (26), in p-xylene at 100 ~ C
b y KI~IGBAUM and TREMENTOZZI (152), and in tetralin at 130~ by
TUNG (256). Figure 14 shows the results of application of the present
method to these data. From this figure, we find that the common inter-
cept demanded by Eq. (58) is located somewhere between K '/" = 0.016
and 0.019, corresponding to K - - 9 . 0 to 2 6 . 1 0 -4. This estimate of K
leads to A = 0 . 9 5 . 1 0 -8 and thus to o = 1.63 4-0.08. This value of
is considerably smaller than that of polystyrene, as would be
expected from the difference between the sizes of the two side groups, H
and C,H 5.
Several previous estimates of the unperturbed dimensions of poly-
ethylene have been published. TREMENTOZZI (251) has given the very
large value A = 1.9. 10-8, corresponding to a = 3.2. He arrived at
this figure b y combining light-scattering measurements of radii of
gyration in the good solvent tetralin with values of the expansion
factor cr obtained from second virial coefficients and intrinsic viscosities
b y means of the Orofino-Flory relation, Eq. (12). More recently FLoRY,
Intrinsic Xriscosities and Unperturbed Dimensions of Long Chain Molecules 237

CIFERRI and CHIANG (102) have deduced A -- 1.07- 10 -s, in fairly good
agreement with our result, from intrinsic viscosities in cr
lene, using a normal value of # and also evaluating a b y the Orofino-
Flory method. The high figure of TRE~ENTOZZI can be accomodated
only b y abandoning the standard value of # , and we do not see how this
can be made physically reasonable for polyethylene. I t seems m u c h m o r e
likely that the well-known difficulties of clarifying polyethylene solutions
for light scattering measurements have persisted to some degree, in
spite of the strong efforts of even the most skilled workers.
Additional support for a value of A in the neighborhood of 1.0 9 10 -s
is furnished b y the results of our calculations of the unperturbed dimen-
sions of a n u m b e r of linear polyesters (see Section I I I C) which consist
mainly of methylene chain units and which yield similar values of A.
We should also mention an as yet unpublished calculation of HOEVE
[quoted in a paper b y CIFERRI, HOEVE and FLORY (66")], according to
which such a value of A is in excellent agreement with the observed
temperature coefficient, as deduced from intrinsic viscosities and from
stress-temperature coefficients for cross-linked polyethylene networks.
Finally we should add a word about the classically famous data of
MEYER and VAN DER WVK (779') on the intrinsic viscosities of pure nor-
mal paraffin hydrocarbons in carbon tetrachloride. If these d a t a are
included in our plot based on Eq. (58), the point for C34H~0 falls close
to the converging lines for high-molecular-weight polyethylenes in
several solvents, but the points for lower hydrocarbons, down to C17H36,
fall well below. In our opinion this does not argue for a still lower value
of A t h a n the one we have chosen, but is better interpreted as evidence
for an end effect, such as is also to be found in certain condensation
polymers. [See Section I I I C and I I I D where this effect is considered in
more detail.]
Our second example is polyacrylonitrile in dimethyl formamide.
Five different series of experimental d a t a (31, tiT, 68", 750, 202) are
shown in Fig. 15. I n spite of almost the same experimental conditions,
a relatively large scattering of data is observed, perhaps in part reflecting
different heterogeneity of materials x. However, if we carefully examine
each series of measurements separately, we find in each case the linear
correlation demanded b y Eq. (58), as shown in Fig. 15. The two intercepts
shown in the figure are as close as K '& = 0.016 and 0.017, so t h a t we
obtain a = 2.18 ~ 2.25, which is of the same order of magnitude as for
polystyrene and not as large as had previously been suggested [see

1 It is to be recMled t h a t effective fractionation is hard to achieve especially


in the case of crystalline polymers. However, we also recall t h a t special difficulties
with fluorescence occur in light scattering measurements with polyacrylonitrile
[see CLELAND and S'rOCKMAY~R {68'), and KRIGBAUM and I{OTLIAR (150)].
16"
238 M. KURATA a n d W. H. STOCKMAYER:

KRIGBAUM and KOTLIAR ( ~ 5 0 ) ~ , The slope of the correlation line (and


h e n c e t h e parameter B) is seen to decrease with increasing temperature,
indicating that this highly polar system has a negative heat of dilution
and that the molecules deswell upon heating [BIsscHoPs (d2)].
The values of the unperturbed dimensions, [(L2)o/M]'A= A, and
of the resulting steric factors a are listed for various addition polymers
in Table 7. The entries marked with an asterisk (*) are based directly
f

Polyacrylon#mTe
-DimethyI formamz'de / / i

IE

o CI,KZ,ANO ~ ,YTOCKHAYER 8 5 o~
o ONYON d5 ~
BAPIf'ORD e / . d 15~
I
9 KRIGS~UM 30~
9 ~l&~VgNOPX & 5 ~ 85~

0 I & 3 4t 5"

Fig. !5. Viscosityplot for polyacrylonltrile in dilxxethylforlxlalll~dB

on experiments at, or nearly at, theta-solvent conditions, and are there-


fore essentially independent of our revisions of the Flory-Fox viscosity
theory. The seventh column of the table gives values of a molar volume V,
measuring the bulk of the chain substituent. When possible, Vx is chosen
as the molar volume of a related liquid; for example, benzene for poly-
styrene, methyl formate for poly(methyl acrylate), and so on. When this
choice of Vx is not possible, a figure is derived from the van der Waals
radii of atoms and groups given by PAtlLING (15').
Some of the K values in Table 7 are not new, but are taken unchanged
from the tabulations of FLORY ( 5 / ) and CI~INAI (64). However, the a
values have all been recalculated with the preferred theoretical number,
2.87" I091, for q)o (of course, with proper allowance for the effects of
heterogeneity, as discussed in Section I I E ) , This revision has an interesting
consequence in the case of polyisobutylene, for which HoEvE (124) has
made a theoretical prediction of a using an internal rotation potential
derived from the known helical conformation of the crystalline polymer.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 239

]~he older value of a as tabulated by FLoRY (5g) was somewhat larger


than the theoretical figure, but our revised ~ is in almost perfect agree-
ment with the one predicted b y HOEVE.
Figure 16 shows plots of a against Vx, which serves as a rough
measure of the steric hindrance caused b y side groups. The white and
black circles represent molecules of the types,
\ t ,;i
11J
HX/p
t11-
\HX
J'
/p
respectively. Therefore, there are two points representing polypropylene;
one is the filled circle on the same abscissa as polyethylene, Vx= 8, and
the other is the open circle on the same abscissa as polyisobutylene,
Vx = 38. The chain line A in the figure connects five non-polar polymers,
i.e. polyethylene, polypropylene, polyisobutylene, polystyrene and
poly-~-methylstyrene. Two broken lines represent oxygen-containing

Z.5 // : -

i,,1'._q4" .4 T

/.S

LO 0 l l ~ I
5-0 /00 150 200 Z~CO
&
Fig. 16. Steric factors a of addition poiymers with various molar substituent volumes Vx

polymers, of which the branch B represents polyacrylic acid and related


polymers and the branch C represents polyvinyl alcohol and related
polymers. The dotted lines, D and E, represent halogen-containing
polymers, i. e. polyvinyl halides and polyhalogenostyrenes, respectively.
The point F represents polyacrylonitrile, the point G represents poly-4-
vinyl pyridine and finally the point H represents poly(chlorotrifluoro
ethylene). This figure, though somewhat complicated, reveals several
interesting relations between the steric factor and the molecular structure.
o o o o o o o o o o o o o o,o...o o o o o o o
"<t "<t '-~
0 ~ ' '
o o ~ o o
~~ ~" ~--, I~' 0 m 0 ,'I __~ -~"
v ~ v ~ 9
~o
, ~
o~ tn
~~ ~r162 ~ ' - ' o ~
l:l:x~
~ - ~ " to 9 ~~ . o _ .
~-~ ~ . ~ ~ g~" ~" 9 o~
9 ~-~ ~ ~ ~ , " ~. 9
~.~. ~.~ ,"t V c0 ~ . . . . . t-i
~. ~ ~ ~ ...... o"
~'" ~. o .....
. . . . . . . . . ~.~ . . . . . -.q
,W
~99999~9 oooooooooo
aaAv~xoo• H.3A. PUe v• OrB
T a b l e 7. (Continued)

P o l y p r o p y l e n e (isotactic) . . . . . . . . . 40 1204-20 765 4- 40 475 1.61 4- 0.08 38 (t~8, 209)


P o l y s t y r e n e (atactic) . . . . . . . . . . 30 8 2 4 - 5* 670 • 15 302 2.22 ! 0.O5 89 (;49, 17s, <
182, 206)
70 75* 650 302 2.15 89 (111, 4f)
P o l y s t y r e n e (isotaetic) . . . . . . . . . . ~~
30 9 0 4 - 10 695 i 25 302 2.30 4- 0.08 89 (24, 7s, 2s2)
P o l y ( v i n y l acetate) . . . . '. . . . . . . 25 934-10' 705 4- 30 332 2.12 4- 0.09 54 (62, 128,
177, 2as)
66 82* 670 332 2.02 54 (177') ~h
P o l y ( v i n y l alcohol) . . . . . . . . . . . 30 222=[=25 950 4- 40 464 2.04 ~: 0.10 25 (196, 82,
104, 236)
Poly(vinyl benzoate) . . . . . . . . . . 32 62 4- 8* 620 4- 25 252 2.46 :E 0.10 105 (223b) ~t
Poly(vinyl bromide) . . . . . . . . . . . 20 40 4- 5* 540 4- 20 298 1.82 4- 0.07 9 35 (67, 67')
Poly(vinyl n-butyrate) . . . . . . . . . . 30 80 -4- 10 670 4- 35 288 2.32 4- 0.12 86 (16o)
Poly(vinyl isobutyrate) . . . . . . . . . 30 80 ~ 10 670 4- 35 288 2.32 4- 0.12 86 (16o)
Poly(vinyl n-caproate) . . . . . . . . . . 30 91 4- 10 700 4- 30 258 2.71 4- 0.12 118 (16o)
Poly(vinyl chloride) . . . . . . . . . . . 25 100 4- 3O 720 + 60 393 1.83 4- 0.15 28 (s5, 45, 66, 5
76)
P o l y ( v i n y l m e t h y l ether) . . . . . . . . . 30 195 =k 30 900 j : 50 404 2.23 4- 0.13 40 (~73) O
P o l y ( 4 - v i n y l pyridine) . . . . . . . . . . 25 94~ 10 710 4- 30 300 2.37 :E 0.10 81 (s8, 44)
P o l y ( v i n y l pyrrolidone) . . . . . . . . . 25 100 4- 15 720 4- 40 292 2.48 :E 0.12 76 (113, 226) o
P o l y ( v i n y l sulfate) . . . . . . . . . . 20 25 ~ 15 460 5:80 278 1.65 4- 0.30 110 (2to)
Poly(butadiene) O
m
9 2 % cis . . . . . . . . . . . . . . . 32 1504-20 820 -4- 40 545 1.50 • 0.O8 (703 oq
70~/o traus . . . . . . . . . . . . . . 25 3004-4O 1030 :k 5o 702 1.45 4- 0.08 (214', 68) c~
7 9 % trans . . . . . . . . . . . . . . 30 2804-25 1010 4- 30 724 m"
1.40 4- 0.05 (219'")
Poly(chloroprene} 5"
8 5 % trans . . . . . . . . . . . . . . 25 115 + 20 750 4- 80 535 1.40 4- 0,15 (189"',189"") O
Poly(isoprene)
H e v e a (100% cis) . . . . . . . . . . . 20 13o 4- 20 810 • 45 485 1.67 4- 0.09 (260)
G u t t a (100% trans) . . . . . . . . . . 60 232 970 703 1.38 (260)
P o l y ( t r i f l u o r o n i t r o s o m e t h a n e / t e t r a f l u oro-
ethylene) . . . . . . . . . . . . . . . 35 38* 510 4- 25 304 1.68 4- 0.08 (t92')
* T h e s e K v a l u e s are essentially based on t h e t a - s o l v e n t e x p e r i m e n t s .
242 M. KVRATAand W, H. STOCKMAYER:

(i) There are four pairs of polymers displaying the effect of an


a-methyl group1; i. e., polyethylene and polypropylene, polypropylene
and polyisobutylene, polystyrene and poly-a-methyl styrene, and poly-
methylacrylate and polymethylmethacrylate. The effect on a is always
positive, perhaps reflecting the larger size of an a-methyl group. How-
ever, the effect is quite small. This is perhaps to be expected, because in
polymer chains a huge group of atoms corresponding to the remaining
part of chain is always attached to the carbon a t o m considered; so that,
if the second attached group X is larger than CH 3, the difference in the
third attached groups, H and CH 3, m a y contribute only a minor effect
to the hindering potential for rotation about the C - C bond.
(it) The difference between the a values of poly(chlorotrifluoro
ethylene) and polyvinyl chloride, 2.03 and 1.83, is larger than experi-
mental error, and m a y be interpreted in terms of the difference in the
van der Waals radii between fluorine and hydrogen atoms.
(iii) The difference between the a values of polypropylene and
polyvinyl bromide, 1.76 and 1.82 respectively, is less than the probable
experimental error. Since the van der Waals radius of the bromine atom
is almost the same as that of the methyl group, this m a y be taken as an
indication that steric repulsion is more important than electrostatic
interaction between bond dipoles in determining a. However, the small
difference we have found happens to be in the direction to be expected
from the observed conformations of n-butane and ethylene bromide
[MIzUSHIMA, MORiNO and SHIMANOUCHI (787)].
(iv) I t has been already pointed out b y CHINAI and coworkers
(55, 64), that the steric factor a of a series of polymethacrylates does not
increase monotonically as the side chain varies from methyl to n-lauryl,
but passes through a minimum at the ethyl or n-butyl esters. In other
words, the methyl ester has a relatively high value of o. This fact also
indicates t h a t steric repulsion between side groups cannot be the only
important factor in determining the hindering potential. For higher
members of this series, however, it m a y be naturally expected that the
stcric effect predominates over other interaction effects because of the
large alkyl groups attached to the ester group, and therefore that the
interaction between the side groups has a close resemblance to t h a t
between side alkyl groups. If this is the case, the representative points
of the higher methacrylates in Fig. 15 should lie on the same correlation

x Polyacrylonitrile and polymethacrylonitrile make a fifth pair of this kind,


but the viscosity relations for the latter are based on osmotic measurements for
only four unfractionated polymers with molecular weights lying between 180,000
and 500,000 [FrdI-IRMANand MESROBIAN(115a)], so that only a crude estimate of
can be made for this polymer. Nevertheless, the effect of the methyl group is about
the same as in the other, better established cases.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 243

line as polypropylene (black point) and polyisobutylene. This can be


shown b y the solid line, which we m a y regard as a standard correlation
presumably representing purely steric interactions. I t is remarkable that
all points for polar molecules show positive deviations from the standard
line, suggesting that electrostatic interactions between polar groups
generally stabilize the trans conformation relative to gauche conforma-
tions.
(v) Polystyrene has a considerably large value of a, 2.22, compared
with other non-polar polymers, and halogenation at the para-position
does not appreciabIy affect the value of o. This behavior is different
from t h a t of polymers with aliphatic side groups, and it would be inter-
esting to have data for additional polar derivatives of polystyrene.
If interactions of an electrostatic nature have some effect on the
relative stability of trans and gauche rotational isomers, it seems reasonab-
le to expect an effect of solvent, though perhaps small, on the factor o.
Such effects on the conformations of small molecules are well known
[see MlZt~SHIMA (ld, 788) and WAI)A (259)1. The relatively high value of a
for poly(methyl methacrylate), as discussed in paragraph (iv) above,
m a y be due to electrostatic interactions, and it is therefore appropriate
to search for a solvent effect here. No such effect was found b y MARCHAL
and LAPP (175) in their measurements of the apparent dipole moment of
the polymer in various solvents as compared with that of the model
monomeric compound, methyl isobutyrate. T h e y concluded that to
within experimental error (about 3%) the unperturbed dimensions were
independent of the solvent. More recently a very small solvent effect on
the dipole moment of the isotactic polymer has been reported b y SALOVEY
(223~)
The range of polarity of the solvents used in the above-mentioned
dielectric measurements was not large, and a much wider range is avail-
able in the viscosity data, some of which are reproduced in Fig. 17.
Here the large circles are the experimental results in chloroform, the
small circles those in acetone, and the crosses those in nitroethane.
All these measurements were made over a rather wide range of molecular
weights; for example, the data of SCHULZ, MEYERHOFF and CANTOW
(230, d7) 1 start at M = 2.5 9 10a and go up to M = 7.4 9 106. The measure-
ments of BlSCHOFI: and DESREUX (dO) and of CASASSA and STOCKMAYER
(51') cover almost two decades in M. I t is seen that the linear relation
demanded b y Eq. (58) is very well satisfied b y all the data over the
entire molecular weight range. However, the lines do not have a common
intercept, but give somewhat different values of K '/', in contrast to the
case of polypropylene (Fig. 10). The differences in a are probably outside

1 Shear-rate corrections are given in reference (47).


244 M. I~URATAand W. H. STOCKMAY]ER:

the error of the extrapolations. These results are collected in T a b l e 8,


which also includes t h e v a l u e for b e n z e n e solutions from t h e d a t a of
SCHULZ, CA~TOW a n d MEYERHOFF (230, d7). I t is e v i d e n t t h a t the
solvent effect on a, if it is r e a l 1, does n o t correlate with the dielectric

PolyrnefhylrnefhacpKle/e

...o
g

,2"
f ./++

o 9 Chlarofarm 2Y~
o 9 Acefone gS-~
+ IV//poelhane aS~
I I T
0 E , 10 IE 2,0

Fig. 17. Viscosity plot for poly(methyl methaeI3"late)

c o n s t a n t of t h e solvent, n o r w i t h the dipole m o m e n t of the solvent


molecule. There appears r a t h e r to be a correlation with the slope of the
viscosity plot, i. e., w i t h the solvent i n t e r a c t i o n p a r a m e t e r B. I t is also

Table 8. Unperturbed dimensions of polymethylmethacrylate in various solvents


Dielectric
Solvent Constant K - 10 i [(Ll)~M]l[s 9 1011 a

Acetone . . . . . . . 21.4 584-3 6004-15 1.95 4- 0.05


Benzene . . . . . . : 2.274 694-5 6354-20 2.06 4- 0.07
Chloroform . . . . . 4.722 864-6 6854-20 2.22 4- 0.07
Nitroethane . . . . . 30.0 704-5 638 4- 20 2.07 4- 0.07
Average . . . . . . . 704-20 , 640 4- 60 2.08 • 0.20

i n t e r e s t i n g to note t h a t the result in b e n z e n e does n o t fall a t the extreme,


b u t in the middle of the range. A n a n o m a l o u s solvent effect of b e n z e n e
has also b e e n r e p o r t e d for small molecules [see MIZUSHIMA (14)]. I t
1 The effect definitely depends on the use of Eq. (58) instead of the Flory-Fox
relation, Eq. (9). The value of K obtained in a mixed and quite polar theta-solvent
(see Table 13 in the Appendix) is very close to that for acetone.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 245

seems fair to conclude that no very large solvent effect exists for poly-
(methyl methacrylate), but that studies of the type of MAI~CHAL and
LAPP (175) should be extended to a wide range of solvent polarities.
Very recently IvlN and EI~DE (132") have found a very large solvent
effect on the unperturbed dimensions of another polar polymer, hexene-1
polysulfone. For example, a fraction with M~ about 7 9 105 has an intrin-
sic viscosity of 0.375 at 23.5 ~ in a theta solvent mixture, butanone/iso-
propanol (37/63), but an intrinsic viscosity of 0.559 in the pure theta
solvent n-hexyl chloride at 13 ~ This unequivocal difference in size is
confirmed b y light-scattering dissymmetry measurements. The authors
point out that the constituents of the mixed theta solvent are both more
polar than the polymer, while n-hexyl chloride is less polar. They suggest
t h a t in the latter case the chains will show a greater preference for con-
formations in which there is a certain cancellation of the internal dipoles
associated with the sulfone groups, and that such conformations must
give a more expanded coil than in the more polar mixed solvent. It would
be highly worth while to investigate this ad hoc suggestion by means of
dipole moment measurements, and also to study the "solvation" of the
polymer (selective adsorption of one constituent) in the mixed-solvent
system b y means of appropriate equilibrium measurements.
Another interesting problem, involving the interactions between
substituents separated by two links, is the effect of stereoregularity
on the unperturbed dimensions. It has repeatedly been reported by
m a n y investigators since NAT:rA, DANt~SSO and MORAGLIO (197, 73)
that the intrinsic viscosity of an isotactic polymer has essentially the
same value as the atactic species of the same molecular weight, but
that the second osmotic virial coefficient is somewhat lower for the
atactic material. The viscosity analysis presented here seems to indicate a
slight, but not unique, difference in the unperturbed dimensions of the
two types of polymer. Thus, the value of A for isotactic polypropylene is
slightly smaller than that of atactic polypropylene, which is in the
direction predicted b y the simplest type of theory [VOLKENSTEIN(258) ;
see Section IV B], but for polystyrene there is a small difference in the other
direction. A further discussion of this problem in connection with the
second virial coefficient is given in Section IVA. We m a y remark also that
the most useful quantitative experimental studies of this problem will
probably be made on poly(methyl methacrylate), which can be prepared
in both iso- and syndiotactic forms as well as essentially atactic varieties,
and in which the stereoregularity can be quantitatively described and
measured b y high resolution NMR spectroscopy [BovEy and TIERS
(4S', 43")].
Results for diene polymers, predominantly or completely 1,4-addition
polymers, are also included in Table 7. For each of the synthetic polymers
246 M. KURATA and Xu H. STOCKMAYER:

we list only the major structural unit; the exact composition of each
polymer is available in the original references, and was taken into account
in computing the unperturbed dimensions from the table of MARKOVlTZ
(175a). There is gratifying similarity among the several diene polymers.
The predominantly trans-l,4 polymers all have a values in the neigh-
borhood of 1.4, while those for the predominantly cis configurations are
over 1.S. Thus, although the trans isomers have larger values of A
because of their greater unperturbed dimensions in the freely rotating
state, this is somewhat compensated by smaller values of or, reflecting
fewer repulsive contacts between chain substituents.
It is interesting to remark that corrections for branching, although
affecting the intrinsic viscosity-molecular weight relations, are not
needed for the intercept of the viscosity plot if the distribution of branctl
points throughout the polymer is random, so that smaller molecules have
proportionately fewer branches. This was first pointed out by WALES,
MARSHALL, ROTI-IMAN and WEISSBERG (260') and was directly invoked
in the theta-solvent experiments on polybutadiene of POLLOCK,ELYASH
and DEWITT (214'). For the same reason, we do not have to worry
unduly about the effects of branching in the other synthetic diene poly-
mers. In support of this remark, we may also cite the tight-scattering
result of CLELAND (68) who evaluated A for predominantly trans 1,4-
polybutadiene by means of Benoit's relation (37') for the asymptotic
behavior of the reciprocal light-scattering angular distribution function
P(O) at large angles. The intercept of the asymptotic line, which is
independent of branching, gave A = (1.15 $- 0.10) 9 10-s, in satisfactory
accord with the viscosity values shown in Table 7.
It should also be pointed out that our calculation for polychloroprene
is based on the results of MOCI~EL, NICHOLS and MIGHTON (189'") for
Neoprene Type GN and of MOCHEL and NICHOLS (789") for Neoprene
Type W, both of which are polymerized at 40 ~ C, and the corresponding
isomeric composition is taken from the infra-red results of MAYNARD and
MOCHEL (177"). MOCHEL and NICHOLS (189") also reported results for a
third polymer, Neoprene Type CG, but the viscosity data display a
trend which we regard as spurious, as the composition of the polymer is
scarcely different from that of the other types. Inclusion of these data
would have produced only a minor change in an over-all average of A.
It may be worth calling attention to the recent results of MORNEAU,
ROTH and SBULTZ (192') on equimolal alternating copolymers of tri-
fluoronitrosomethane with tetrafluoroethylene. These chains contain no
hydrogen and have the repeating unit --CF2CF2N(CF~)O--. Light-
scattering dissymmetry measurements in a theta solvent (Freon 11a,
which is 1,1,2-trichloro-l,2,2-trifluorethane) indicated rather large chain
dimensions, corresponding to a ~-~2.8, b u t when these are combined with
the measured, intrinsic viscosity they give a value of # about six times
Intrinsic Viscosities and Unperturbed Dimensions o f Long Chain Molecules 247

too small. We prefer to retain the standard v a l u e of r neglecting the


dissymmetry data, and we then obtain the rather normal figure of
= 1.68 for this polymer, as recorded in Table 7.

B. Cellulose, Amylose and Their Derivatives


As already remarked in Section C, the theory presented here per-
mits the Mark-Houwink-Sakurada index v to attain a limiting value of
unity in extremely good solvents. Since v for Cellulosic chains is frequently
quite high, say between 0.8 and 1.0 Esee Table 13 in the Appendix], it
is clear that application of the new equations to these macromolecules
would lead to a large expansion factor and hence to a relatively small
unperturbed dimension. This is just contrary to a commonly held view,
according to which certain cellulose derivatives are supposed to have
abnormally extended unperturbed chains and a very small expansion
factor even in good solvents.
In a number of cases, the behavior of cellulosic materials in solution
is that normally expected of flexible chains. Thus ~/[ANDELKERN and
FLORV (770) found theta solvents for cellulose tributyrate and tricapry-
late at somewhat elevated temperatures and determined the unperturbed
dimensions from the intrinsic viscosities. The resulting steric factors a
were about two, and independent of solvent and temperature for the
longer-chain ester. However, it was observed that i n the Solvent tri-
butyrin, which was considered to be an athermal solvent, the intrinsic
viscosity of cellulose tributyrate showed an abnormally large negative
temperature coefficient ; and MANDELKERN and FLORY therefore conclud-
ed that the unperturbed dimensions of this polymer undergo a similar
behavior, starting with an abnormally extended conformation at room
temperature and arriving at a normally flexible state at high temperatu-
res, say (r = 1.8 at 130 ~ C, which is the figure determined in the theta-
solvent experiments. If tributyrin is really an athermal solvent for
cellulose tributyrate, this conclusion is inescapable, for the expansion
factor a must be very nearly constant in athermal solutions and therefore
cannot contribute appreciably to the temperature coefficient of intrinsic
viscosity under such conditions. However, the assumption that the
system involved in the present case is athermal m a y not be valid 1, and
1 This assumption is based on the fact t h a t the polymer-solvent interaction
parameter ;/[see Eq. (8)] of the tributyrin-cellulose tributyrate system, as evaluated
from melting-point depressions, is nearly zero at about 100 ~ C [MANDELKERNand
FLORY (169)]. I t does not follow, however, t h a t the system is athermal, for the
parameter X generally involves an entropy contribution. Furthermore, the heat
and entropy parts of this parameter vary with the concentration in a complicated
way, especially in polar systems [see, for example, TAKENAKA (243); ZIMM (22);
KURATA (154)]. Thus it is extremely hazardous to predict dilute solution properties
from concentrated solution properties such as the melting-point depression, at least
on a highly quantitative level as in the present problem.
248 M. K U R A T A a n d W . YI. STOCKMAYER:

it will be seen that an alternative interpretation of the observations is


possible.
The Mandelkern-Flory viscosity data for cellulose tributyrate are
plotted according to Eq. (58) in Fig. 18. It is seen that the line for the
viscosities in the good solvent methyl ethyl ketone at 30~ passes
without strain to the same intercept as the lines for the two high-tem-
perature theta solvents. The intercept gives K - - (9.7 4-1.5) 9 10-4 and
a = 1.79 ~k 0.10, as listed in Table 9. Although, as discussed in the

Oellulose fribulypofe
3

%
~

-~
~ J : 5

__x____

I 1~0 ~
Dadecane (0.75) + Tetr~lin (0.2E) /gO ~

0 I ~ 3 ~ 5"

Fig. 18. Main figure: viscosity plot for ceUulo~e tributyrate. Insert: ixlteraetion parameter B for cellulose
tributyrate in tributyrin

previous Section, a common intercept in such polar systems is not


necessarily to be expected, neither are large temperature coefficients of a
to be expected for normally flexible chains. Similar results are found for
cellulose tficaprylate, but these need not be reproduced here.
There is no line for tributyrin-ceUulose tributyrate in Fig. 18, because
the viscosity measurements were limited to only one sample. However,
if we a s s u m e that K maintains at all temperatures the value derived
above from data in other solvents, the interaction parameter B m a y be
evaluated from Eq. (58) at each temperature. The result of this calcula-
tion is shown in the insert to Fig. 18, which demonstrates that the linear
relation:
B = --0.015011 - (450/T)1 (88)
reproduces the temperature effect. This equation corresponds to a
negative heat of dilution which causes the polymer to deswell upon
heating, and it also yields a negative value of the entropy parameter iv
[ F L o R Y ( S d ) : see also KURATA ( 1 5 4 ) ~ , which may be related to orientation
~ntrinsic Viscosities a n d U n p e r t u r b e d D i m e n s i o n s of L o n g C h a i n Molecules 249

effects involving solvent molecules around polymer segmentst. These


signs are, of course, opposite to those found for most polymer-solvent
systems, but they are not surprising in highly polar systems. In this way
it is possible to account for an unusually rapid decrease of intrinsic
viscosity with rising temperature without requiring an abnormal tem-
perature coefficient of the unperturbed dimensions. Dilute solution
measurements of light scattering or osmotic pressure over a good range
5

Cellulo~e lpiaeetat8 o o /'* /

e~

2
///
///.. 9 .-Cresol

I " 8 3

Fig. 19. Viscosity plot for cellulose acetate

of temperature would aid in deciding between the two alternatives. It


may be noted that according to Eq. (88) there should be a lower consolute
temperature near 180~ C for the tributyrin-cellulose tributyrate system.
The viscosity data for cellulose acetate [ P H I L I P P and BJORK (2/2);
FLORY, SPURRand CARPENTER(]07)] for three solvents at ordinary tem-
peratures are plotted in Fig. 19. Again, although the viscosity exponents v
x B is p r o p o r t i o n a l to ( I / 2 - - ~ ) or to ( 1 - O T - 1 ) . F o r i l l u s t r a t i v e p u r p o s e s ,
a s s u m e t h a t t h e p o t e n t i a l of m e a n force w (r) h a s a h a r d core of d i a m e t e r b a n d t h a t
w (r) ~ k T for r :> b. T h e n E q . (14) m a y be w r i t t e n
03
fl = (4zr/3)b a - - (4~r[k T) f w(r) r 2 d r .
b
N o w since w (r) is a local free e n e r g y , i n v o l v i n g all possible c o n f i g u r a t i o n s of t h e
s o l v e n t molecules a r o u n d t h e p o l y m e r s e g m e n t s , it m u s t in general c o n s i s t of a n
e n e r g y p a r t u (r) a n d a n e n t r o p y p a r t T s (r). T h e r e f o r e t h e sign of t h e t e m p e r a t u r e -
i n d e p e n d e n t p a r t of fl is d e t e r m i n e d b y t h e sign of s (r) a n d its m a g n i t u d e relative
to t h e s e g m e n t size. S o m e e x a m p l e s of s u c h o r i e n t a t i o n effects are d i s c u s s e d b y
PRmOGINE (17) a n d b y MONSTER ( 1 5 ) .
230 M. KURATA and W. H. STOCKMAYER:

are quite high, there is no difficulty in drawing straight lines with a


common intercept which corresponds to a modest a value of 1.6.
The classic example of the extended cellulose chain is, of course,
cellulose trinitrate, to which we now turn. The most extensive viscosity
measurements include those of IMMERGUT, RANBY and MARK (~37) and
of HARLAND (Q). These data are plotted in Fig. 20. Although the molecular
weight ranges from 3 9 104 to about 106, the points scatter to a consider-
able extent, as the difficulty of working with this material is well known.

/4 ; E 9 ]C
/
Cellulose /rinlkrafe 9

o A ~ k o n e aS% (It<aCRaUT)
9 Kfhy/acela/e ZS~ fIM~)
I I I I
I . 3 ,g ~ 5-
" Z131{~II13 10-3
,g (~) ~ I
Fig. 20, Viscosity plot for cellulose nitratcs

Nevertheless, it is possible to draw three lines to a common intercept at


about K "/, = 0.01, which corresponds again to a ~ value of about 1.6.
Since the Mark-Houwink-Saknrada index v is at least 0.9 in acetone and
practically unity in ethyl acetate (actually 1.0S according to Table 13),
the present theory suggests t h a t very good solvents are involved. This
can also be appreciated from the magnitudes of the expansion factors
estimated with the aid of the K value-found above. T h e y are approxim-
ately 3.3, 2.7 and 2.3 in acetone at molecular weights of 106, S. 105, and
105, respectively; and in ethyl acetate at the same molecular weights we
find an = 3.7, 8.0 and 2.8. Since it is a~ which occurs in the expression for
[~/], there can be no disagreement about the statement t h a t these molec-
ules are enormously extended under the experimental conditions. The
question is whether this great extension is to be understood in terms of
unusual short-range interactions controlling the unperturbed dimensions
or in terms of unusually favorable polymer-solvent interactions resulting
in abnormally large expansion factors.
I n t r i n s i c Viscosities a n d U n p e r t u r b e d D i m e n s i o n s of L o n g Chain Molecules 251

From their light-scattering measurements HOLTZER, BEI~OIT, and


DoTY (726) concluded that the short-range interactions control the
dimensions of cellulose nitrate chains, and they discussed their results in
terms of the worm-like chain model of KRATKY and PoRoI) (742), obtain-
ing a persistence length of about 34.7 -~. In Fig. 21 these data are shown
as a plot of (S~)~/M,o against Mw. The open circles are the experimental
points and the broken curve is that calculated from the equations for
the worm-like chain model. The theoretical curve is claimed to reproduce
the data to within the probable experimental error in all but two cases.
L 1 I I I
oP
Ce//u/ose //V/7//r~/e
Acelone Z5 ~

. . . . . . . . "o-- . . . . .

1 q I l L
0 / 2 3
Mw. I0 -G
Fig. 21. Dimensionsof cellulosenitrates. Circles,data of H()LTZER,BENOITand DuTY(12~). Full curve, present
theory for flexible chain .with expansion factor; dashed curve, wormlike chain without expansion factor

However, at least an equally good fit1 is given b y the solid curve, which
has been calculated from Eq. (32) with the parameters required to give 2
the upper curve for acetone in Fig. 20. A similar result can be obtained
from the light-scattering measurements of HUNT, NEWMAN, SCHERAGA
and FLORY (725)) for cellulose nitrate in ethyl acetate. These authors also
favored the worm-like chain model. These data are given in the form of a
double-logarithmic plot in Fig. 22 (filled circles), where the data of
HOLTZER, BENOIT and DOTY are again shown (open circles). The essen-
tially straight lines through the points are based on Eq. (32). The para-
meters for the acetone data have already been given; for ethyl acetate
the value of A is, of course, the same, while in this solvent B ----4.35.10 -~e.
x N o t e t h a t t h e p o i n t P in o u r Fig. 21 w a s o m i t t e d in the original p l o t (Fig. 7
of reference 726). Also, t h e i r abscissa w a s M , while we h a v e preferred to use M . .
T h e l a t t e r c h a n g e does n o t affect o u r conclusions.
= Since the fractions used in t h e s e studies are n o t v e r y sharp, it is necessary t o
use t h e heterogeneity correction of Eq. (82) for t h e q u a n t i t y (S=>,[M'w, w h i c h
i n t r o d u c e s a f a c t o r (h + 2)](h + l) before t h e q u a n t i t y A *. T a k i n g A = 0.72 9 10 -8
(see Table 9) a n d choosing h ~- 1, w h i c h w a s r e c o m m e n d e d b y t h e original a u t h o r s
f r o m c o m p a r i s o n of t h e i r light-scattering a n d o s m o t i c - p r e s s u r e results, we o b t a i n
t h e solid curve of Fig. 21 w i t h B ---- 2.40 9 I0 -2~
Fortschr. Hochpolym.-Forsch.,Bd. 3 17
252 M. KURATA a n d W. H. STOCKMA.YER:

It is easily found that the limiting form of Eq. (32) in extremely good
solvents is
<S'> ~- 0.125 B%M % (89)
and the slopes of the lines in Fig. 22 are quite close to the limiting value
of 0.67 demanded b y this equation. Exactly this exponent has previously
been deduced for cellulose
nitrate from a consider-
ation of the viscosity and
o Acetone 2r176 // I sedimentation behavior
(see below) by 1V~EYERHOFF
IOa 9 ~thylacetatc2~~ J ] (187) and by PETERLIN
(211), but at that time
this exponent was not
regarded as possible for
flexible chains.
According to the Krat-
ky-Porod theory, or indeed
any theory of short-range
interactions, the chains
must attain Gaussian be-
I ~- ~ II 1 ~ iii I I havior as the molecular
104 105 IO s iOz weight is indefinitely in-
Mw
creased. The persistence
Fig. 22, Dimensions of cellulose nitrates
length obtained by HOLT-
ZER et el. and by HUNT et
6"ellulase fr/nlfrate " / al. is such that at molecular
o Acetone ea ~ / / /
9 Ethr,aceta,e ~Oo~ j . Z ~ . / / weights Mw above about
5 . 1 0 ~ the chains should
Cellulose /e/ace)ate ~z~" / /
. A c e t o n e 2,0~ j j . ~ l practically have attained
the Gaussian limit. On
the other hand, since the
viscosity index v remains
near unity at the highest
molecular weights studied,
about 106, we are then
forced to accept free-drain-
4403 /0 # /05- /0 9 ~./0 G ing hydrodynamics along
Mw with the worm-like chain.
Fig. 23. Moleeular weight-sedimentation ~oefflcient relations for
severer cellulosicpolymers
But then the sedimenta-
tion constant should be
essentially independent of M, as already pointed out in Section
II D and in Fig. 13. A glance at Fig. 23, which represents on a log-
log scale the sedimentation constants of cellulose nitrate in acetone
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 253

[MEYERHO~F (131)] and in ethyl acetate [Hu~T, NEWMAN, SCHERAGA


and FLORY (129)]; of cellulose triacetate in acetone [HOLMES and
SMITH ( ] 2 5 ) ] ; and of ethyl hydroxyethyl cellulose in water [MANLEY
(172)], shows that this prediction fails badly in every case. The lines in
Fig. 23 have in fact been drawn with a slope of 1/3, which represents the
limiting behavior in very good solvents according to our theory, and it is
seen that the data roughly agree with this value. Recently another
example of this type has been reported by SAMSONOVA and FRENI~EL
(223'), who found v = 1.0 and vs = 0.37 for ethylcellulose i n ethyl
acetate. Points representing the above five systems have already been
shown in Fig. 13, where the evidence against the free-draining theory is
very obvious. We thus arrive at the same conclusions as MEYERHOFF (781)
and PETERLIN (211), who used the Debye-Bueche theory of both viscosity
and friction constant and showed that the data for cellulose nitrate were
inconsistent with Gaussian chain statistics ~. More recently MARRINAN
and HERMANS (775') have also concluded that cellulose nitrate chains
must be strongly non-Gaussian at all molecular weights in order to
conform to the existing hydrodynamic theories of viscosity and sedimen-
tation.
From another point of view, it can be seen that approximate con-
formity of cellulose nitrates to both of the asymptot!c relations (63)
and (89) for [~] and <Sz>, respectively, implies that the number
should remain essentially constant at its asymptotic value qgoo, given
b y Eq. (57). Since it is difficult to assess the polydispersity of cellulose
nitrate fractions, the constancy of ~b, rather than its numerical value,
should be of prime importance in testing the consistency of our conclusions.
DOTY, SCHNEIDER a n d HOLTZER (89) and IX[EWMAN and FLORu (198)
both originally r e p o r t e d t h a t ~5 decreased systematically with decreasing
molecular weight. Later HOLTZER, BENOIT and DOTY (726) preferred to
regard # as essentially constant at about 1.9.1021, somewhat smaller

1 PETERLIN and I~EYERHOFF combine viscosities and sedimentation constants


into the quantity
s. [7]'~'M-'I' = H (X), (9O)
which is a k n o w n function of the draining parameter X. Inverting this relation
yields a value of X for each polymer, for which a radius of gyration can then be
calculated by Eqs. (50) and (5), with neglect of any difference between 0~ and ~ .
I t will be recognized t h a t Eq. (90) must reduce to the Mandelkern-Flory relation (72)
for very large X and from the discussion following Eq. (79) it is clear t h a t great
accuracy i n the evaluation of X cannot be attained by inverting Eq. (90) ; in turn,
however, the shielding correction to the viscosity can be determined quite accurately
even with a crude value of X. By this method, PETERLIN and MEYERHOFF calculated
t h a t for cellulose nitrate the mean square radius of gyration is proportional to the
4[3 power of M while X is independent of M. The latter result is, in effect, identical
to the condition of very strong hydrodynamic shielding.
17"
254 M. KURATAand W. H. STOCKMAYER:

t h a n for most synthetic polymers but subject to large uncertainties


because of polydispersity. On the other hand, HUNT, NEWMAN, SCHERAGA
and FLORY (129) reaffirmed their earlier conclusion and reported a
series of smoothed values ranging from 2.2 9 1021 at M = 890,000 down
to 1.0 9 10 ~1 at M = 48,000. Although the unsmoothed values seem to
v a r y less extremely, especially at high molecular weights, yet we m a y
regard these results as suggestive of a small draining effect on the intrinsic
viscosity.
To investigate the magnitude of the draining effect in cellulose
nitrate, we return to Eq. (62'), but we first recall the calculations made
for polystyrene and several other polymers, described in Section I I C.
There it was shown (recall Fig. 12) that the effective value of the draining
p a r a m e t e r X must be at least ten times as large as t h a t corresponding
to the geometrical thickness of the chain. Adopting a factor of ten for
cellulose nitrate, and taking b = 12 •, me = 297 and A = 0.72 9 10 -8,
we obtain
X = 0.40M '/' ~n 1 . (91)

Since ~r can be estimated from Fig. 20, each experimental intrinsic


Viscosity can now be "corrected" to non-draining conditions through
multiplication b y the factor [XFo(X)]~/[XFo(X)], which is easily
obtained from the table of KURATA and YAMAKAWA (157). The viscosity
plot can then be repeated with these corrected intrinsic viscosities to
yield a new value of the unperturbed dimensions. In principle, the
values of A and a n derived from the new plot should now be used again
in Eq. (91) to begin another cycle, but in the present case the first
approximation is sufficient, as the draining correction turns out to
be moderate if not altogether negligible. The modified plot leads to
A = 0.81 9 10 -s, or about ten per cent higher than the original uncorrected
value, and the corresponding value of a is now about 1.8.
Until very recently, the small-angle X - r a y scattering studies of
KRATKY and his coworkers (7,12') seemed to indicate that cellulose
nitrates were very stiff chains with a persistence length exceeding 100 A.
However, in the most recent investigation of this kind [-HEINE,KRATKY,
POROD and SCHMITZ (722b)] a persistence length of only 22/k was
obtained for cellulose nitrate. This figure is in fair agreement with our
result based on intrinsic viscosity, which Corresponds (see Section IV B)
to a persistence length of about 19/~. We therefore feel iustified in
regarding the cellulose nitrate chain as normally flexible in the unperturb-
ed state.
Draining corrections have not been applied to any of the data for
other cellulose derivatives, as they are in all cases smaller than for the
nitrate. The results of the viscosity calculations for these substances arc
I n t r i n s i c Viscosities a n d U n p e r t u r b e d D i m e n s i o n s of L o n g Chain Molecules 255

collected in Table 9, which also contains the molar weights M~ of the


repeating units in the chain. The estimates of A and a are subject to
relatively large uncertainties because of the rather long extrapolations
required in most cases. In Fig. 24, the resulting values of a are seen to
vary more or less symbaticalty with M~, which m a y be taken as a rough
measure of the relative sizes of the substituent groups. The points for
t t i I fo~ I /
t ./i
2.5

/ /,
2.0 / ../"

19 ~, j
9 If "f- ~ Cellulos6 series
~'I9 . I *. .-. . - *. mAmylase
y/ose series
ser/es
b.-...~ PolypeBNdes

/.60 I I I I I ,
200 4"00 6"00
Hu~
Fig. 24. Steric factors a for cellulosic and amylosic polymers and for polypeptides, plotted against molar
weight M~ of repeating unit

methyl cellulose and ethyl hydroxyethyl cellulose, however, lie well


above the others, and that for cellulose tricaproate is also somewhat
high 1.
Results for amylose and its derivatives are also shown in Table 9
and Fig. 24. There is an unfortunately large disagreement between the
results of EVERETT and FOSTER (94') and those of Cowm (77). Both
investigations included measurements in dimethyl sulfoxide and in
aqueous potassium hydroxide, while EVERETT and FOSTER also reported
results in a theta solvent. The viscosity plots from the two laboratories
are internally consistent, but they lead to quite different unperturbed
dimensions. In any case, it can be said that amylose and its derivatives
generally show larger a values than the cellulosic chains, suggesting that
the less stretched a-glucose ring accommodates substituents with greater
difficulty than does the fl-glucose ring. The a-glucose chain, however, is
smaller in the freely rotating state than the fl-glucose chain (Table 6).
z T h e d a t a for cellulose t r i c a p r o a t e [KRIGBAUM a n d SPXRLING (151)'] were too
scattered for t h e use of E q . (58), so t h a t we h a v e t a k e n a f r o m t h e e s t i m a t e s based
on light scattering m a d e b y t h e authors 9
256 M. KURATAand "~V.H. STOCKMAYER:

A similar complementarity
was already noticed in Sec-
tion I n A between the un-
perturbed dimensions of cis-
and t r a n s - d i e n e polymers.
Finally, it is perhaps in-
teresting to note that an
r.
equation recently proposed
b y MOORE and BROWN (700)
for solutions of cellulose deriv-
atives is in approximate ac-
c~
cord with the results obtained
here. Their equation, which
was based on rather different
considerations, can be written
[~] = Q ( M / T ) -- P M b , (92)
where T is the absolute tem-
perature and Q, P and b are
~ o 0 0 0 o 0 0 o o o 0 o o
independent of T and M. If
we take the asymptotic Eq.
(63) and assume that the in-
teraction parameter B has its
usual linear dependence on
1 / T , we obtain a result like
Eq. (92) with b = 1. However,
to make both Q and P posi-
tive numbers, as they gen-
erally appear to be for the
cellulose derivatives, it must
be assumed that both the
heat of dilution and the Flory
. . . . . . . . . . ~ " ~
entropy parameter ~v1 are
negative, as we have already
suggested for cellulose tri-
o~ b u t y r a t e in tributyrin, E q
(88). These tendencies m a y
be related to the rather special
behavior of cellulose deriv-
atives in respect to com-
patibility with other poly-
o o o o o,~o.~o o o o o ~ , . . . , ~
mers [DOBRY and BOYER-
~ AWENOKI (82t)].
I n t r i n s i c Viscosities a n d U n p e r t u r b e d D i m e n s i o n s of L o n g C h a i n Mol e c ul e s 257

Finally it m a y be remarked that the dynamic viscoelastic properties


of plasticized cellulose derivatives seem to give no evidence of any
unusual temperature dependence of the chain conformations. Thus,
LANDEL and FERRY (762, 163, 764) successfully applied the method of
reduced variables [see, for example, FERRY (6)] to various concentrated
solutions of cellulosic polymers, and found that the temperature reduc-
tion factors were quite similar to those for other flexible polymers such as
poly(isobutylene).

C. Condensation Polymers
As shown in Table 13 in the Appendix, the Mark-Houwink-Sakurada
index v of condensation polymers is always somewhere between the two
theoretical limits 0.5 and
I I
1.0; therefore, the appli-
cation of the present theo- Pely (ethylene axide)
ry to these substances does
not involve any difficulty,
at least in principle; we
2 - o -

can always find some 9

value of K by plotting
the ratio [~]'/~/M '1. against 9
g (~) M*/q[~/]'/'. However
the reliability of the K
values thus obtained de- o 8eRzeRe

pends on how far the Curbontetrachloride 9

method can be correctly + Methanol

applied down to low mole-


cular weights, because the I I
0 3"00 1000 IEO0
molecular weight of con-
densation polymers is usu-
Fig. 25. Viscosity plot for polyethylene oxides, showing large
ally limited to less than positive "end effect" in methanol
105. In this connection,
we m a y again refer to Fig. 12, where the present theory shows
complete agreement with the experimental data for polystyrene down
to molecular weights as low as 1000, suggesting that no difficulties are
likely to arise.
SADRON and REMPP (223) have recently investigated in various sol-
vents the viscosity behavior of poly(ethylene oxide) with low molecular
weights ranging from 62 (ethylene glycol) to about 2 . 1 0 4 . Fig. 25 shows
the result of application of the present method to these data. It is clear
from the figure that the data in benzene and in carbon tetrachloride fit
the present theory well and give a common intercept, from which K is
evaluated as (11 + 1). 10-4. Another series of data obtained in cyclohexane
258 M. KURATAand W. H. STOCKMAYE~:

also leads to the same result. BAILEY and his coworkers (28, 29)
have measured the intrinsic viscosities of the same polymer in a higher
molecular weight range, M w - - 2 . 1 0 4 to 5.106, in various solvents
including two nearly ideal solvents (0.45 M aqueous K2SO4 at 35 ~ C,
and 0.39 M aqueous MgS04 at 45 ~ C), and they obtained K = (10
13)-10 -4. Since the two estimates of K agree quite well, we feel
justified in using Eq. (58) for such short chains. Furthermore, MARCHAL
and BENOIT (/74) have found (cos r 0.30 for this polymer from
measurements of dipole moments. If we accept the estimate K = 11 9 10-4,
we obtain a = 1.38 as given in Table 10, and this leads to (cos r = 0.31
with the aid of Eq. (86). Since this is in perfect agreement with the
Marchal-Benoit value, we are even more confident that the present
viscosity method can lead to reliable information about the conforma-
tions of condensation polymers.
Returning to Fig. 25, we observe that the [21]'],/MV,vs. g ( ~ ) M ' / , / [ ~ l ] '!,
plot for the methanol solutions is divided into two regions at a molec-
ular weight of about 4000. In the higher molecular weight region,
the plot is quite normal and extrapolates to the common K value dis-
cussed above. However, in the lower molecular weight region, the plot
displays a sharp upturn, and appears to diverge as the molecular weight
approaches zero. Similar upturns are also observed in water and dioxane
solutions, in regions of molecular weight lower than about 4000 and 800,
respectively. Generally speaking, the larger this critical molecular weight
is, the sharper the upturn appears. This kind of phenomenon m a y be
interpreted in the following way.
In a purely mathematical sense, the lower limit of [7] predicted by
Eq. (58) or Eq. (6) is zero at M = 0. But this is physically meaningless
and the intrinsic viscosity curve should stop at a finite non-zero value
as the molecular weight approaches the monomer weight. As remarked
by SADRON and REMMP,this effect can be expressed empirically b y the
equati~ It/] = D + K ' M ~ , (93)
where D is a new constant representing the end effect. The existence of a
positive value of D naturally causes an upturn in the [21]*/'/M'/, vs.
g (~) M'lq [M]'/' diagram.
Now consider a binary mixture of non-polymeric liquids; and denote
the viscosity of liquid 0 (solvent) by */0, and that of liquid 1 (solute) b y 211.
Then, according to TAMURA and KURATA (244', 244"), the viscosity 21
of mixtures can be expressed by the semi-empirical equation,
~/= x0 ~o~/o + 2 (XoXlr ~bx)'/'~01 + xl ~ 7 ~ , (94)
x The empirical equation, [r/] = E + K'M, which was often used at the end
of the 1930's, may often Eo replaced by the Mark-Houwink-Sakurada equation (13).
But the D in Eq. (93) represents a true end effect.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 259

where x 0 and ~0 represent the mole fraction and the volume fraction of
solvent, x 1 and ~1 represent those of solute, and '1ol is a parameter called
t h e "mutual viscosity", representing the interaction between unlike
molecules. Very recently, this type of formula has been given a statistical
mechanical basis by BEARMAN and JoNEs (36). In the limit of dilute
solutions, this equation leads to
[ ~ o = (0.o2/a) [(~ol/~o) - 1~, (95)
where, for simplicity, we put xl = r where ~1 is the solute density.
This quantity must correspond to the constant D, at least semi-quanti-
tatively. It is well known that aqueous solutions of alcohols display
a marked maximum in the viscosity-concentration diagram. For example,
aqueous solution of ethanol can be specified by ~70= 0.467, ~h = 0.592
and ~7ol= 1.370 [centipoise] at 60 ~ C (2d4'), which leads to [~/]o = 0.05.
If an end effect of this size is used in Eq. (93), its influence is far from
negligible. It can easily be seen in Fig. 12, for example, that if the
intrinsic viscosities of polystyrene were all to be raised b y the constant
amount 0.05 the log-log-plot would flatten out at low M, and the plot
corresponding to Fig. 11 would show an upturn at about M = 5,000. As
polar polymers like poly(ethylene oxide) have the opportunity for
stronger interactions with polar solvents than with non-polar ones, the
upturn in the viscosity pIot should be more common with polar solvents.
These figures fit well to the observed behavior of the poly(ethylene
oxide)-methanol or -water system; and there is again no margin for the
draining effect between the ordinary molecular weight region and the
end region.
The viscosities of m a n y binary liquid systems display minima as
functions of composition at constant temperature, so that negative
values of D are also possible. YAJmK and his coworkers (265') long
ago observed that very frequently an extremum in the isothermal vapor
pressure-composition curve is accompanied b y an extremum of the
opposite sense in the viscosity-concentration curve. Data are apparently
not available for solutions of very low-molecular-weight paraffins in carbon
tetrachloride, but minima are found for the viscosities of solutions of
CCl~with ethyl iodide, ethyl acetate and acetone, so that a minimum
appears quite probable for mixtures of small aliphatic hydrocarbons with
carbon tetrachlofide. If this were true, the downward trend of the
Meyer-Van d e r W y k data on C1~-C84 paraffins, earlier discussed in
connection with the polyethylene plots of Fig. 14, would be understood.
It will be recognized that such a trend is also precisely what is to be
expected from the draining effect of the hydrodynamic theories of
DEBYE and BUECHE (79), BRINKMAN (45') and KIRKWOOD and RISEMAN
(~39). However, the absence of such a trend in the case of polyethylene
260 M. KURATA and W. H. STOCKMAYER:

oxide, which shows only positive or negligible end effects, argues against
the latter explanation.
The unperturbed dimensions of various condensation polymers
obtained by the present method are listed in Table 10. A polyelectrolyte
chain, sodium polyphosphate, has been included because theta-solvent
results are available. The freely-rotating chain dimension (L2)0i of
poly(dimethylsiloxane) in the table is due to FLORY and his coworkers
(105), that for the polyphosphate chains is taken directly from the paper
of STRAUSS and WINEMAN (241'), while most of the others have been
calculated in the standard manner with the convenient and only negligib-
ly incorrect assumption that all the aliphatic bond angles are tetrahedral.
The free-rotation values for the maleate and fumarate polyesters are
based on parameters consistent with those of Table 6 for diene polymers.
The steric factor ~ of poly(ethylene oxide) is certainly lower than
that of polyethylene (see Table 7). The principal reason for this difference
is probably the absence of hydrogen-atom substituents on the skeletal
oxygens, resulting in a reduction of the steric interactions. Electrostatic
effects between the skeletal C - O bonds seem to be minor, since the value
of a does not depend measurably on solvent. In this connection, the
Unperturbed dimensions of polyoxymethylene would be extremely
interesting, but we are unable to find any suitable intrinsic viscosity data
for this polymer 1. However, according to UCHIDA, I~URITA and KUBO
(257') the electric dipole moments of rather low oligomers with methyl
end groups indicate that the gauche conformations are rather more
stable than the tram conformation. They evaluated the energy difference
between these conformations in n-hexane as 1.74 kcal. mole -1, which
corresponds to a = 0.61. More recently STARKWEATHERand BOYD (235')
have shown that the above value of the energy difference is consistent
with the entropy of melting of polyoxymethylene, using a theory which
gives good results for polyethylene and poly(tetrafluoroethylene). It is
remarkable that no other polymer in Tables 7, 9 or 10 has a value of a
less than unity, and it seems almost certain that the strong electrostatic
interactions between the alternately directed highly polar C - O bonds are
responsible for the unique behavior of polyoxymethylene.
The value of a for poly(propylene oxide) was obtained from the
data of MOACANIN (789) for polyurethanes prepared b y condensation of
toluene-2,4-diisocyanate with atactic polypropylene glycols of molecular
weights about 1000 and 2000. The small quantity of diisocyanate present
in these compositions can make only a negligible effect on the chain
dimensions (less than one per cent on the freely rotating chain, as is
1 A relation between melt viscosity and end-group molecular weight for un-
tractiona*ced high-molecular-weight polyoxymethylenes has been given by K o c h
and LINDVIG (I,$0a).
Table 10. Unperturbed chain dimensions calculated from If values. III. Condensation polymers
I
Temp. /L=\ 111=
Polymer 9
.K 10~" Reference
~

Polyamides: 0
Poly:(&caproamide] . . . . . . . . . . . 25 1 9 0 4 - I0 890• 20 545 1.63 4- 0.04 (224)
P o l y - ( h e x a m e t h y l e n e adipamide) . . . . . . 25 1904-20 8 9 0 4 - 40 545 1.63 4- 0.08 (127, 2d5)
Poly-dimethylsiloxane) . . . . . . . . . . . ca. 25 804- 5* 6 7 0 4 - 20 456 1.47 4- 0.05 (32, lo5)
Polyesters :
"Polycarbonate" . . . . . . . . . . . . 25 1804-20 880 4- 40 796 I.I0 4- 0.05 (231)
Poly(cis-l,4-cyclohexane disebacate) . . . . 20 140+20 800 4- 30 495 1.62 4- 0.05 (32c) r
Poly(trans-l,4-cyclohexane disebacate) . . . 20 1604-20 840 4- 30 633 1.33 4- 0.05 (32c)
P o l y ( d e c a m e t h y l e n e adip~te) . . . . . . . ca. 25 1 0 0 4 - 10 720 4- 25 540 1.33 4- 0.05 (lo8)
P o l y ( d e c a m e t h y l e n e sebacate) . . . . . . . 25 220 4- 30 900 4- 50 549 1.65 4- 0 . I 0 (2ma)
Poly(ethylene terephthalate) ........ 25 1 6 0 4 - 15 840 • 25 687 1.22 4- 0.03 (70)
P o l y ( h e x a d e c a m e t h y l e n e sebacate) . . . . . 20 270 4- 40 1 0 0 0 4 - 50 555 1.80 4- 0.10 (sea)
Poly(hexamethylene aeetylenedicarboxylate) 20 180 4- 20 870 4- 30 627 1.39 4- 0.05 (Sac)
ol
Poly(hexamethylene e.e'-dibutylsebacate) . . 20 1554-25 835 4- 70 457 1.82 4- 0.15 (32b) O
Poly(hexamethylene fumarate) . . . . . . 20--50 180 4- 20 8 7 0 4 - 30 592 1.47 -~- 0.05 (322)
P o l y ( h e x a m e t h y l e n e maleate) . . . . . . . 20--50 1 3 5 4 - 15 790• 30 510 1.55 4- 0,05 (320
P o l y ( h e x a m e t h y l e n e sebacate) . . . . . . . 20 215 4- 60 910 4-100 540 1.7 4- 0.17 (32a, b)
P o l y ( h e x a m e t h y l e n e succinate) . . . . . . 20~60 165 4- 30 850 4- 60 522 1.62 + 0.14 (32e) O
Poly(oetamethylene cis-hexahydro- ffrt
terephthalate) . . . . . . . . . . . . . 20 140 4- 20 8 0 0 4 - 30 495 1.62 ~ 0.05 (32c)
P o l y ( o c t a m e t h y l e n e trans-hexahydro-
terephthalate) . . . . . . . . . . . . . 20 160 4- 20 8 4 0 4 - 30 633 1.33 4- 0.05 (sec)
P o l y (o)-oxyundecanoate) . . . . . . . . . 20 185 4- 60 8804-100 550 1,6 4- 0.16 (so, 16s") O
Polyethers:
P o l y ( e t h y l e n e oxide) . . . . . . . . . . . 20 110 ~ 1 0 ' 7 5 0 - t - 30 541 1.38 4- 0.06 (220, 22s)
P o l y ( p r o p y l e n e oxide) (atactic) . ..... 25 115 4- 10 7 5 0 + 25 472 1.59 4- 0.05 (1s9)
Poly(metaphosphate) . . . . . . . . . . . . 25 504- 3* 560 4- 20 370 1.51 4- 0.04 (241") I,O
* T h e s e K v a l u e s are e s s e n t i a l l y b a s e d on t h e t a - s o l v e n t ex )eriments.
262 M. KUI~ATAand W. H. STOCK~tAYER:

9 easily calculated), and this statement is confirmed by the fact that the
viscosities of the unmodified glycols give points which fall on the curve
for the polyurethane fractions. The increase in ~ caused by the additional
methyl group in tiffs polymer as compared to poly(ethylene oxide) is
somewhat larger than the difference in cr values between polyethylene
and polypropylene.
The great length9 of the C - # - O bond in "polyearbonate" and that
of the C - ~ b - C bond in poly(ethylene terephthalate) both cause marked
decreases in (r. These effects can be understood in terms of the diminution
of steric interactions between side groups. The a value for polycarbonate
is remarkably close to the limiting value of unity for a chain with free
internal rotations.
Table 10 also contains values of the unperturbed dimensions of
m a n y aliphatic polyesters. Most of these have been evaluated from
the extensive results of BATZER and his collaborators (32a, 32b, 32c,
32d, 32e, 33a, 168") which combine intrinsic viscosities with osmotic
molecular weights of fractions. Since almost all of these polymers have
rather high crystalline melting points, and since the fractionations were
generally performed at temperatures well below these melting points,
we have somewhat arbitrarily chosen a value of 1.5 for the molecular
weight ratio M,JMn, which corresponds to a figure of 1.22 for the correc-
tion factor q~ of Eq. (84). In other words, we believe that the fractions
in general had quite broad molecular weight distributions. Consequently,
the listed values of the viscosity constant K are quite uncertain, but the
linear unperturbed dimensions are proportional only to K ~h and there-
fore are almost surely within ten per cent of the correct ones. Larger
errors could arise only if there were a very pronounced trend in the
polydispersity of the fractions with molecular weight.
Examining the results for saturated linear polyesters, we note
that to a fair approximation the values of both A and ~ stay roughly
constant as the composition changes from that of poly(hexamethylene
succinate) to that of poly(hexadecamethylene sebacate), with the notable
exception of the Flory-Stickney (108) results for poly(decamethylene
adipate). These were based on end-group determinations of molecular
weight for unfractionated polymers with the "most probable" distribu-
tion, but so were those of RAICH (2~9a) for poly(decamethylene sebacate),
which fall agreeably close to the Batzer series. In any case, it is clear
that the two C - O bonds of an ester link are not greatly different from
two C - C bonds in their contributions to A 2, so that by any extrapolation
of the polyester results to pure paraffin chains we would arrive at a figure
close to 1 . 0 . 1 0 -s for A, which is very near our preferred value for linear
polyethylene (Table 7).
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 263

T u r n i n g n o w to the remaining polyesters, we r e m a r k t h a t the


structural effects on viscosity in good solvents (most commonly, chloro-
form) observed b y BATZER and his coworkers carry over to the u n p e r t u r b -
ed dimensions as well, since the variations in the viscosity exponent v
are after all n o t v e r y great. T h u s the A value for poly(hexamethylene
succinate) is lower t h a n t h a t of the f u m a r a t e b u t higher than the maleate
[BATZER a n d MOHR (32e)~; and the acetylenedicarboxylate closely
resembles the f u m a r a t e [BATZER and WEISSENBERGER (33a)1 , although a
is a bit smaller. I t is also reasonable t h a t poly(hexamethylene cr
dibutylsebacate) [BATZER (32b)] should have a larger a t h a n the un-
substituted sebacate. Finally, the u n p e r t u r b e d dimensions of polyesters
containing saturated six-membered tings in either acid or diol units
[BATZER and FRITZ (32C)] are larger in the trans-configuration t h a n in
the cis-eonfiguration. We have not taken the trouble to calculate A and a
for certain polyesters of oxalic, glutarie and pimelic acids [BATZER and
LA~G (32d)1.
There are relatively few d a t a available for the synthetic polyamides.
F r o m the studies of SCHAEFGEN and FLORY (224) on 6-Nylon in sulfuric
acid and those of HOWARD (127) on 66-Nylon in formic acid-sodium
formate we derive identical values of A and of a, as befits their nearly
identical structures. This great similarity of the isomeric polymers is also
shown in the work of BATZER and M6SCHLE (32]), who found t h a t b o t h of
these Nylons and their copolymers obeyed the same viscosity-molecular
weight relationship.
Fina]ly, it is interesting to note t h a t the value of a for the poly-
(metaphosphate) chain [STRAUSS and WINEMAN (24~')] has the same
order of magnitude as those of non-electrolyte chains with substituents
of comparable size. U n d e r the observed theta-solvent conditions (0.415 M
aqueous NaBr), the screened long range electrostatic interactions are just
balanced b y the various other contributions to the excluded volume
effect, b u t there is no reason to expect a simultaneous obliteration of
short-range electrostatic effects 1.

i A very crude calculation suggests, however, that the short-range electrostatic


effect will be small. The Debye-Hiickel radius 1]~ is 4.8 9 l018 cm in 0.415 M NaBr.
If we use the screening factor exp(--~r) to estimate the effective electrostatic
energy between the charged oxygens, placing half an electronic charge on each
side-chain oxygen atom, we find an energy, difference of only about 240 cal.mole-1
in favor of the cis conformations, i. e., in a direction opposite to the steric effect.
According to this calculation the residual electrostatic effect would reduce a by
about ten per cent in water. It would be extremely interesting to know the effect
of changing dielectric constant of the solvent on the unperturbed dimensions of this
chain.
264 M. KURATA and W. H. STOCKMAYER:

D. Polypeptides and Polynucleotides


Since this article is limited to linear polymers, and since poly-
electrolytes have been essentially excluded, there are relatively few
biochemically interesting materials which we can discuss. Several
polypeptides, however~ can be treated b y the present methods. It is
now well known that many polypeptides can exist in two widely different
conformations in solution, the random coil and the ~-helix. This was first

3 ~ I I I i 6

Po/y /"].,-be, Tzy/- u-gh~a-


o DCA " mafa) /"
9 DMF r
o3. - # l u t a m a h ) l
.ace I

oo 9 ~
o 1/ .
2

I I I I l
/ Z 3 5 8
. z/V (r
Fig. 26. Viscosityplot for poly-~,-benzylglutamates.Dashed curve through filled circles is for helical polymers

demonstrated for poly(y-benzyl-L-glutamate) b y DOTY and his co-


workers (85, 87) b y means of light scattering and viscosity measurements
in addition to optical rotation and rotatory dispersion [see also DOTY (4)].
The dimensions of the a-helix deduced from the intrinsic viscosities in
dimethyl formamide (and in other solvents permitting helix formation)
are in good agreement with those known from crystallography, and these
dimensions have also been confirmed b y measurements of dielectric
relaxation in dioxane [WADA (259')3.
Here we are mainly concerned with the polypeptides in their random-
coil form. Figure 26 shows several sets of viscosity data plotted according
to Eq. (58). The smaller white circles give the observations of DOTY,
BRADBURY and HOLTZER ( 8 5 ) and of MITCHELL, WOODWARD a n d D O T Y
(185) on solutions of poly(~-benzyl-L-glutamate) in dictfloroacetic acid.
The half-filled circles denote the measurements of SPACI~ (234") on
poly(~-benzyl-DL-glutamate) in the same solvent, while the larger white
Intrinsic Viscosities and U n p e r t u r b e d Dimensions of Long Chain Molecules 265

circles are Spach's results for the DL-polymer in dimethyl formamide 1.


In the same figure, the filled circles are the data (on a different ordinate
scale) of the D o t y group for the a-helLx form of poly(y-benzyl-z-glutamate)
in dimethylformamide. They show clearly enough t h a t polymers known
to be in the form of rigid rods or ellipsoids are in no great danger of being
transmuted into flexible coils b y our graphical method.
The viscosity data for the poly-DL-glutamate esters lie on good
lines, the values in dichloroacetic acid falling only slightly below those
of the pure L-polymers in the same solvent. Below a molecular weight
of 15,000 the latter show an upturn reminiscent of the ones discussed
earlier (Fig. 25) and of an order of magnitude consistent with the "end
effect" of Eq. (93). The necessarily strong solvation of the polypeptide
b y dichloroacetic acid lends support to the interpretation of a positive
end effect in the present case. The lines for the DL-polymers extrapolate
to a common intercept, and to within experimental error the points for
the L-glutamate above a molecular weight of 15,000 can be extrapolated
to the same intercept. The resulting value of K leads to ~ = 2.3, a fairly
high value which is, however, consistent with the rather large substituent
involved.
A polypeptide without the complications of helix formation is poly-
sarcosine, and the data of FESSLER and OGSTON (95) lead to the rather
small value of ca. 1.3 for o, again consistent with the small size of the
N-methyl substituent. A similar calculation can be made for gelatin.
Ignoring for the m o m e n t the possible complications due to the L-proline
residues of the chain, we find A = 0.54 9 10 -8 and ~ = 1.34 for gelatin
from the data of POORADIER (215). Finally, we have a very rough figure
for poly-DL-proline, which can safely be assumed to exist in the random-
coil form. STEINBERG, IHARRINGTON, BERGER, SELA a n d KATCHALSKI
(237a) report t h a t the intrinsic viscosity of a poly-DL-proline sample
of molecular weight 9700 in water is about 0.05, and that this viscosity is
unaffected b y addition of salt even up to 6M LiBr. This m a y be taken as
rough evidence t h a t water is not far from a theta-solvent for polyproline,
and this suggestion is also supported b y the very low osmotic second
virial coefficients found in dilute salt solutions. From this one point we
therefore find the crude estimate a = 1.5 for poly-DL-proline. The steric
factors of the four random-coil polypeptides are shown in Fig. 24 plotted
against the molar weights M~ of the peptide units, and it seems clear
t h a t steric interactions between the substituents are of maj or importance

x Only polypeptides prepared in benzene showed this behavior in dimethyl


formamide. Polypeptides prepared in dioxane or D M F gave a more complicated
curve indicating some helix formation, b u t inferior to t h a t of the pure L-glutamate
polymer. We have omitted these results from the figure.
266 M. KURATA and W. H. STOCKMAYER:

i n determining the unperturbed dimensions. Save for the possibility of


the end effects in very good solvents, which m a y be hard to predict
a priori, it appears that the relation between intrinsic viscosity and chain
conformation is fairly well understood for randomly coiling polypeptides.
We now return to poly-(7-benzyl-z-glutamate). In Fig. 27, the data
of DOTY and his collaborators (85, 785) are shown again, now in the form
of the standard double-logarithmic plot of intrinsic viscosity against

/0 , ~ i I

Pgly (,F-6enzyl-t. -qlu/amofe.) /- "- /


9 .
o O~chlopeacet/c gc/d
9 Oime/hylfapmgm/de
.

dS~ ,g///
25~
t
I"
/

/ ...--".....

al o

200 10 3 10 0 10 5 /0 g

HW
Fig. 27. Viscosity-molecular weight relation for poly-7-benzyl-L-glutamate. Solid curve, theory for random
coil in DCA, Eqs. (58) and (96). Chain curve, theory for rigid ellipsoids in DMF. Dashed curve, random coils
in theta solvent, K = 58 9 10 -a. Dotted curve, hypothetical curve for random coils (or interrupted helices)
with K = 34 9 10- i

molecular weight. The solid curve in the figure was calculated from the
line through the small open circles in Fig. 26, and corresponds to the
following parameters for the randomly coiled polymers in dichloroacetic
acid:
K = 5 . 8 - 10-~, A = 0 . 6 0 " 1 0 -s, B = 2 . 7 - 1 0 -37. (96)

The broken line gives the hypothetical theta-solvent intrinsic viscosity,


[7]0 -- KM'h, and the gap between the solid and broken curves, there-
fore, represents the expansion or swelling factor a~. The end effect is
again seen in the deviation of the four lowest open circles from the line.
It is interesting to note that if no measurements in dichloroacetic acid
existed above a molecular weight of about 50,000 the viscosity exponent v
would have to be chosen in the neighborhood of 0.5, even though it is
clear from the complete body of data that the system is very far from the
theta point !
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 267

Figure 27 also demonstrates the success of the rigid ellipsoidal


model in fitting the intrinsic viscosities of the helical poly-L-ghitamate
esters. The experimental results (filled circles) are correlated b y the chain
line, which is the theoretical result for prolate ellipsoids with axial ratio
p = 21//2650. This numerical factor corresponds to a minor axis of about
18 ~ for the ellipsoid [DOTY, BRADBURY, and HOLTZER (85)] ; and a rigid
rod of the same volume and length has a diameter of 15 ,~, in almost
perfect agreement with the crystallographic value. The rather large value
of the ratio M/p = 2650 m a y be held responsible for the crossing of the
two viscosity curves at a molecular weight as high as 66,000. I t is seen
t h a t the theoretical ellipsoid curve crosses even the theta-solvent
random-coil curve. To have the smooth transition from coil to ellipsoid
characteristic of polystyrene (Fig. 12) it would be necessary for K to
be as low as 34 9 10 -5, corresponding to the lowest, dotted line in Fig. 27.
This allows the essential elimination of the interrupted helix as an alter-
native model for the polymer in dichloroacetic acid, as was already
recognized b y DOTY et al. (85). Of course, the optical rotation scarcely
leaves room for doubt on this point.
NISHIHARA and DOTY (200) have found that the intrinsic viscosities
of a series of collagen samples degraded b y sonic irradiation are well
represented as those of prolate ellipsoids with M/p = 2020, which
corresponds to about 22 peptide residues. These figures resemble in
magnitude those of poly(y-benzyl-L-glutamate) helices, and have been
interpreted as confirmation of a three-stranded helical structure.
A less clear-cut case is t h a t of poly-L-proline, which can form two
alternative helices (Form I, right-handed, with positive optical rotation
and all peptide bonds in cis configuration; Form II, left-handed, with
large negative optical rotation and all peptide bonds in the trans con-
figuration). The data of STEINBERG et al. (237a) do not cover a wide
range of molecular weight, but are quite sufficient to show t h a t the rigid
ellipsoidal model is inadequate to explain the hydrodynamic behavior
of these polymers. The optical rotation then leaves no choice but the
interrupted helix as a model; that is, a chain composed of short helical
sections, interrupted b y flexible joints. In support of this model is the
fact t h a t the Mandelkern-Flory constant /~ is observed to have a value
which is probably equal to t h a t for non-draining random coils within
experimental error. Recalling that water is not far from a theta solvent
for polyproDnes, we find that over a molecular weight range from about
5,000 to 52,000 the intrinsic viscosities give a roughly constant value of
K--~ 3 9 10 -3. This corresponds to the unperturbed dimension A ~- 1.0 •
• 10 -s cm., which would have to be interpreted with the rather high a
value of about 2.6 if we believed the structure of the polymer to be that
of a conventional random coil.
Fortschr. Hochpolym.-Forsch., Bd. 3 18
268 M. KURATAand W. H. STOCKMAY]ER:

9 However, if we suppose the chain to be composed of helical sections,


wffh a negligible random-coil content between them, and if we further
assume, b y analogy with polycarbonate (Table 10), t h a t these helical
sections will display only feeble steric interactions with each other
(o'~_ 1), then we can compute their average size. Taking Equation (137) in
Section IV B and putting ~ = 1, we have
(L2)o[M = n 2 = nrb~lM --- rb~/Mu, (97)
where the translation distance per unit along the helical axis is bo and
the weight-average n u m b e r of units in a helical sequence is r. In deriving
(137) it was assumed t h a t successive helical sections can take random
directions with respect to each other. Since this is probably not possible
when the random-coil sections become very short, Eq. (97) should
probably contain an additional factor greater than unity, expressing
the limitations on the relative orientation of successive helical sections.
Ignoring this factor and taking b0 = 3 . 1 2 . 1 0 -s cm. [see STEINBERG
et al. (237a)], we obtain r ~ 10 residues. The correct answer could be as
small as five, which is adequate for essentially full development of the
characteristic optical rotation of a helix, according to ZlMM,DOTY, and
I s o (274).
The intrinsic viscosities of Form I I poly-L-proline in the good
solvents acetic or propionic acid are about twice as large as in water,
b u t they also follow roughly the power v = 0.5. Since there are un-
doubtedly large solvent effects on helix stability and on the average
length of helical sequences, it would probably be legitimate to assume
t h a t here also we have approximately theta conditions. However, there
is no supporting evidence for this view, and we must recognize the
alternative possibility t h a t there is an .end effect in these solvents,
similar in magnitude to that of poly(7-benzyl-L-glutamate ) in dichloro-
acetic acid, which could effectively reduce v to about 0.5 over the restrict-
ed molecular weight range thus far studied. F o r m I of poly-L-proline is
even less well characterized, and there is at present no adequate means
of assessing its conformation. The intriguing complexity of the system
during the mutarotation from one form to another Csee STEINBERG et al.
(237a) and also SASISEKHARAN (223")] is doubtless t o b e explained in
terms of three kinds of chain structure present in varying degrees:
Form I helix, Form I I helix, and random coil. F r o m the above inter-
pretation of the behavior of polyprolines, it is also clear that no difficulties
are to be expected from this source in treating the data for gelatin as
those for random coils.
We now turn to a brief discussion of polynucleotides, yielding to this
temptation in spite of their polyelectrolytic nature because of their great
importance. In this we are also encouraged b y the apparently innocent
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 269

behavior of the polyphosphate chain [STRAUSS and WINEMAN (241');


Table 10] under theta conditions. The simplest system is that of the
pure synthetic material, polyadenylic acid ("poly-A"). According to
FRESCO and DOTY (114), the Mark-Houwink-Sakurada index v of this
polymer in 0.15 M NaC1 at pH 7 is 0.65. Since they evaluated their
molecular weights from intrinsic viscosities and sedimentation constants
by means of the Mandelkern-Flory Equation (72), their sedimentation
constants are automatically proportional to the 0.45 power of M [see
Eq. (75)]. Analysis of the data leads to the parameters:
K= 11.5.10 -4 , A=0.76.10 -8, B=2.10.10 -27 , (98)

which are of normal magnitude for linear flexible chains; and the corre-
sponding value of a is about 2.3, which is also acceptable for such large
substituents. BEERS and 8TEINER (237') have reported light-scattering
measurements on poly(adenylic acid) at neutral pH, but at higher
molecular weights and mostly at much lower salt concentrations than those
used by FRESCO and DOTY. Essential agreement between these in-
vestigations can, however, be demonstrated. For example, the observed
(light-scattering) radius of gyration for a sample of molecular weight
3 . 1 0 s at pH 6.6 in 0.15 M salt is about 900/~. The unperturbed radius
of gyration for this polymer, from the A value of Eq. (98), is about 535 A,
and the viscosity expansion factor is r162= ([~]/[rl]o) '['-~ 1.5, so that the
radius of gyration predicted from the viscosity is about 800 A, in satis-
factory agreement with the measured value.
When the pH is lowered, poly(adenylic acid) undergoes a reversible
association to larger particles, which are believed to be interrupted two-
strand helices. The measurements of FRESCO and DOTy (lld), again
limited to viscosity and sedimentation, extend over a twenty-fold range
of molecular weight when interpreted in terms of the Mandelkern-Flory
equation as for the single-strand poly-A. The intrinsic viscosities of
these preparations lie well below those to be expected for rigid Watson-
Crick helices of the same molecular weight, and the v.~scosity index is
v ----0.92, which is still in the range available for coils in good solvents.
The data give a normal correlation diagram resulting in the parameters
K=2.3"10 -4 , A=0.44"10 -s, B=0.89.10 -27 . (99)

This K is considerably smaller than that of the single-strand polymer


at neutral pH, and suggests that the molecules have a rather thick
structure. Adopting the interrupted-helix model, we m a y apply Eq. (97)
as in the case of poly-L-proline. Noting that the unit translation distance
bo is 1.70/~ for the double-stranded Watson-Crick helix (263') [i. e.,
a distance of 3.40 A for two residties, one in each chain], and taking
M~ = 328 for poly(adenylic acid), we obtain r = 22 residues per helical
18"
270 M. KURATAand W. H. STOCKMAYER:

section. Again, this is an upper limit, since Eq. (97) assumes no restriction
on the angle between the axes of successive sections. According to this
calculation the geometrical length of the average sequence is about
37.5 A, or only a little more than one complete turn of the double helix.
The crystallographic diameter of the helix is about 21 A.
The foregoing calculation for the double-stranded poly-A might be
criticized for several reasons. First, it m a y be objected t h a t no allowance
has been made for the existence of sizeable untwisted non-helical portions
between the helical sequences. Within reasonable limits, the calculation
is not tremendously sensitive to this feature. For example, if we allow
as much as one-third of the polymer t o be in the non-helical form and
treat this portion as similar in flexibility to single-stranded poly-A, we
still recover an average helical sequence length of about 15 residues.
Second, no account has been taken of the draining effect, which m a y not
be negligible in this case. In applying Eq. (62'), there is some ambiguity
about the size of a segment in the present case. I t seems fair to choose a
portion of the helix as long as its thickness (21 A) for such a segment.
This makes b -- 21, ms --4000, and A = 0.44, and if we use the same
procedure as for cellulose nitrate (Section I I I B) we find X = 0.08 M'h/o~n,
which corresponds to a small but not altogether negligible draining effect
in the experimental range of molecular weights. The most probable value
of r is thereby raised to about 30 residues, or perhaps even as high as 40.
I t is rather interesting t o note that such values, though of course far
smaller t h a n an average helical sequence in any native DNA, are within
the range predicted b y ZIM~ (272') from his theory of the stability of the
double-stranded helix with reasonable values of the parameters.
The viscosity correlations for poly(adenylic acid) are shown in Fig. 28
together with those for several native materials, to which we now tun~.
I t is well established [see for example, DOTY (4) and RICH (18)J t h a t
the native molecule of deoxyribonucleic acid is a double-stranded helix
even in solution, and t h a t the helix can be separated or denatured into
two single strands b y heat or b y chemical agents. Clear evidence for the
complete separation has been reported b y DoTY, MARMUR,EIGNER and
SCHILDKRAUT (88), and they have given viscosity-molecular weight
relations i n H M P (hundredth molar phosphate) ~ for three series of
single-strand denatured D N A samples. These are derived from strains of
Diplococcus pneumoniae (R-36A), Escherichia colt (K-12) and Pseudo-
monas aeruginosa (NRRL-B23), respectively, and have different base
1 The solvent contains 0.0025 M NaaHPO4, 0.0050 M NaH,I~O4 and 0.00i0 M
disodium ethylenediaminetetraacetate as a chelating agent for divalent cations.
Such a low value of counterSion concentration as this is adequate for preventing
aggregation of denatured DNA molecules.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 271

compositions 1 which are reflected in the different viscosity constants


given in Table 13 (Appendix). From these results [the experimental
points are not in Reference (88) but were obtained from the thesis of
EIGNER (4')], we obtain a common K value for the three pol~uners
(Table 11), although the slopes in Fig. 28 are different. Thus the effect
of base composition on the unperturbed dimensions appears to be rather

~/ /001 ~
x+ flO/y -A / ~.~ /
9 ~7S
B o B'Paeumomue('ffg3"q~176 J x. ~ ~.
9 C, ca/z(Gc, sao/o) o7' o ~ eo[_'~...
* P. ~'cregiaasa(~'g~'%) / .~ ~ J "~'.
Y ./ ~zs I- "~..
s- /' /." :~ I } ",~.
o/ .~ /," 8.:o":A.u.:

9 :ajgregated) .........--'-"
'~ ."~.;'fr,a] P'6 law P .._._....-~.

~.~* gEC, DH = Z2
I I I I
o 5 1o 15 XO
.e : W t4"/3/ (,7) 'a . /a -s
Fig, 28. Main figure: Viscosity plot for polynucleotides. Insert: Percent recovery of extinction coefficient as
function of polymer-solvent interaction parameter B

small in comparison with the effect on polymer-solvent interactions


as expressed in the expansion factor. In support of this view, it is also
seen in Table 11 or Fig. 28 that the unperturbed dimensions of these
single-stranded DNA chains are nearly identical to those of single-
stranded poly(adenylic acid), as might indeed have been hoped from the
essential similarity of their chemical structures 3. In the figure are shown
three other experimental points corresponding to the single-stranded
chains in the solvent SSC (standard saline citrate) of higher counter-ion
concentration, which again extrapolate to the same unperturbed dimen-
z The guanine and cytosine content (expressed as the percentage of total base
groups) of these three types of DNA are 39, 50 and 660/0, respectively.
Note that the DNA polymers differ from poly(adenylic acid) not only in base
composition but also in the absence of the oxygen at the 2-position of the ribose
ring.
272 M. I(UR&TAand W. H. STOCKMAYER:

sion as above 1. Since the cor-


relation line is nearly parallel
to the abscissa axis, the sol-
vent SSC is regarded as a
theta solvent for the polymer
at 25 ~ C.
The effect of base com-
position on polymer-solvent in-
teractions deserves some com-
ment. I t has already been
shown b y MARMOR and DOTY
.~.~. ~ (175 b) that the "melting" tem-
peratures of the double-strand-
ed native molecules in a given
solvent increase linearly with
the guanine-cytosine (GC) con-
tent. Now the viscosity plot
reveals that the interaction pa-
rameter B is a decreasing linear
ooodo oo gV function of the GC content.
Since the constancy of the un-
perturbed dimensions suggests
that the entropies of melting are
the same for all DNA samples,
the dependence of melting tem-
perature on GC content is prob-
ably due to changes in the heat
of fusion. Stronger solvent-
polymer interactions would
then accompany lower melting
temperature, as is found ex-
perimentaUy. The effect m a y
be related to the fact that
guanine and cytosine can each
form three hydrogen bonds, but
adenine and thymine only two.
"9 ~~ " ~ o 1 Tho solvent SSC contains
" ~
" ~ ~ ~z 9 ~
0.15 M I~C1 and 0.015 M
Nas(C6HsOT), and it buffers at
~ ' ~ ~ o ~ pH 7.2. The single-stranded chains
. . . . .
in this solvent were obtained by
transferring the polymers from
HMP by dialysis at sufficiently
low polymer concentrations.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 273

The hypochromic effect in the 2600 A absorption band has recently


attracted considerable attention, because the effect is very sensitive to
conformational changes in DNA chains and also to the environment.
The optical density increases b y about 40 % for most DNA samples as the
sample passes from the native to the denatured state upon heating, and it
returns to lower values as the denatured sample is cooled down to room
temperature. The fractional recovery of the extinction coefficient can be
defined for the latter process as (D a - D ) / ( D a - D n ) , where D n is the
optical density of the native sample, and D a and D are those of the
denatured sample at high and room temperatures, respectively. For
three types of DNA in HMP, the recoveries were 27, 34 and 53 %, increas-
ing with the GC content; and for the samples in SSC, they were nearly
the same (about 70%) for all base compositions. The hypochromism in
the native samples is now well interpreted [TINoCO (249a)] in terms of
dipole-dipole interaction between bases which are coiled into the Watson-
Crick helix. If the same interpretation were extended to the single-strand
denatured chain, the hypochromism would be a measure of the fraction of
short intrachain regions possessing the double helix structure. In other
words, if the recovery of extinction coefficient is directly the recovery
of the Watson-Crick helix, it must accordingly be reflected in the un-
perturbed dimension of the chains, since the interaction responsible for
the helix formation is essentially of the short-range type. This type of
effect, however, has not been observed in the present viscosity analysis.
In the insert to Fig. 28, we plot the percent recovery of extinction
coefficient against the polymer-solvent interaction parameter B, finding
a linear correlation between the two quantities 1. This suggests that the
recovery may be closely related to the long-range type of interaction
between bases, and that the hypochromism could perhaps be interpreted
without postulating a n y helical regions in the single-strand denatured
chains.
At high counter-ion concentrations, according to EIGSER (if'), single-
strand denatured chains usually undergo non-specific interchain aggrega-
tion. Such denatured-aggregated samples also behave as typical random
coils, whose intrinsic viscosity, for example, is nearly proportional to
M ~ The unperturbed dimensions of these samples can readily be
evaluated from a viscosity plot (not reproduced in Fig. 28). The K value
is 85 9 10-5 which is slightly smaller than that of the single-strand chains
but considerably larger than that of double-stranded poly(adenylic
acid). This m a y indicate that the base-base contacts responsible for the
aggregation introduce a small fraction of helical regions possessing a
thick structure.
x The B value corresponding to zero recovery was evaluated from three viscosity
points obtained in HMP at 75~ and in HMP plus 8 M urea at 45~
274 M. KURATAand W. I-I. STOCKMAYER:

T h e hydrodynamic properties of solutions of native double-stranded


DNA have thus far eluded complete quantitative interpretation, in
spite of very extensive investigation. A synthesis of experimental data
has recently been furnished b y DOTY (84') ; some of the earlier experimen-
tal results m a y be found in the papers of DOTY, BUNCE-McGILL, and
RICE (86); DOTY, MARMUR, EIGNER, and SCHILDKRAUT (88); and
KAWADE and WATANABE (735'). It is easy to see that the double helices
are not perfectly inflexible, for the observed intrinsic viscosities are far
lower than those of rigid rods or ellipsoids with the Watson-Crick
dimensions, p = 3//4600. On the other hand, the customary flexible-coil
treatments also do not apply to these data. For example, if the correlation
plot of [~I]'/'M -'/' against g (~) M '/, [,/]-'/, is attempted, it is found that
no line intersecting the positive ordinate axis can be drawn without
ignoring the lower experimental points. In other words, the viscosity
index v is greater than unity; for example, over the molecular weight
range of about 3- 105 to 7- l0 s the value 1.12 was chosen b y DOTY,
BUNCE-MCGILL, and RIcE (86). As the molecular weight goes still higher,
there is definite indication that v decreases. In the absence of a reliable
'physical theory of intrinsic viscosity for rather stiff chains, with proper
allowance for both the draining effect and the rotatory brownian motion,
it is not possible to calculate reliable chain dimensions from these data.
V e l y recently it has b e e n demonstrated b y ROBEI~STEIN, THOMAS,
and HERSHEY (222') a s well as b y DAVlSON, FREI1;ELDER, HEDE, and
LEVlNTHAL (77') that each particle of T 2 bacteriophage virus very
probably contains just one molecule. To establish this result, extreme
care was necessary to avoid shear degradation of the polymer, or at least
to correct for it. The molecular weight of the DNA may then be estimated
from the phosphorus content of the virus particle, and turns out to lie
in the neighborhood of 1.5 9 108. A preliminary value of about 700 dl/g
has been obtained b y ZIMM (272") for the low-shear intrinsic viscosity
of this material in 0.2 M NaC1. The log-log plot given by DOTY (84')
extrapolates quite well to this point, but the viscosity index v still lies
above 0.5 even at this enormous molecular weight. Thus, although one
can scarcely imagine that these huge molecules are anything but non-
draining random coils (the contour length is about 0.1 ram), they are
unfortunately not in a theta-solvent. The best we can do is, therefore,
to find the upper limit of [<L2>o/M]'[', obtained by setting ~ at unity,
and this turns out to be about 2.8. 1078, which corresponds to a persis-
tence length of no greater than 750 Jk for the double-stranded DNA
molecule. It is, of course, not possible from this number alone to decide
whether the compliance of the molecule is to be ascribed mainly to
flexibility of the uninterrupted helix or to the existence of occasional very
flexible joints.
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 275

Light scattering measurements have also very frequently been made


on DNA, and they have been critically reviewed b y GEIDUSCHEK and
HOLTZER (6") and more recently by SADRON (19"). Because the per-
sistence length of the chain is not negligible compared to the wave
length of visible light, great caution must be exercised in interpreting the
results; however, the estimated radii of gyration, if the possible complica-
tions are ignored, are in rough accord with the viscosity estimate given
above. Some workers [HERMA~IS and HERMAIqS (122') and LETT and
STACE'r (165')~ have tried to work directly with the relation between
intrinsic viscosity and radius of gyration, but with no more conclusive
results.
The sedimentation coefficient of native DNA [see DOTY (84')] also
points to an intermediate condition of both chain stiffness and draining.
Since the theory of translational diffusion is much simpler and more
reliable than that of the rotational diffusion necessary for a rigorous
theory of intrinsic viscosity, it would seem quite possible to construct
adequate theories for the sedimentation and translational diffusion of
stiff chains. Some years ago PETERLIN (21 I') made such calculation for
chains with free rotation but with fixed valence angles up to and includ-
ing 180~ and very recently HEARST (122a) has worked out theoretical
expressions for the continuous worm-like chain and for any number of
rigid rods linked end-to-end b y universal joints. Both of the latter models
can be made to fit the experimental sedimentation constants of DNA,
but only the continuous worm-like chain does so with an acceptable
value (about 35 A) for the effective hydrodynamic thickness of the
molecule, and this model gives a persistence length of about 400 A.
Since the expansion factor should be very close to unity for such stiff
chains at the lower molecular weights considered here, the above figure
must be close to the unperturbed dimension. The corresponding value of
A = [(L~)o/M] '1" is about 2 . 0 . 1 0 -8. Although this result seems reason-
ably sound, it should still be regarded as preliminary until confirmed by
another independent method.
Still another indication of the upper limit given earlier m a y .be
found from another combination of sedimentation and viscosity data.
By combining Eqs. (4), (53), (66), (68), and (69) we easily find for non-
draining flexible chains:
[~] so/M = [q~/NAPo~ [(1 -v~)/*/o] [(L~)o/M~ [0csn/0cl] . (100)
Since the first two factors in square brackets in the right-hand member
are independent of molecular weight for any chain model, a log-log plot
of [~/] so/M against M should extrapolate to a constant limit at high M
in theta solvent, and it should attain a limiting slope of 1/3 at high M
in good solvents. If the data given b y DOTY (84') for DNA in 0.2 M NaC1
276 M. KURATAand W. H. STOCK,AVER:

are treated in this fashion, there is no constant limit for E~/]so/M, which
indicates that this medium is not a theta solvent. If we avoid unjustified
extrapolation and ignore the expansion factors in Eq. (100), we can get
an upper limit of about 800 A for the persistence length, in satisfactory
accord with the figure based only on Zimm's viscosity measurement for
the pristine undegraded T 2 bacteriophage DNA.
In contrast to DNA, the conformation of the ribonucleic acid chain
seems to be essentially that of a random coil. Four series of viscosity
data from different sources [BOEDTKER (43); HALL and DOTY (8);
KAWADE (13) ; and KURLAND (-161)] have been examined, and are some-
what discordant, but the general behavior is not very far from that of
denaturated-aggregated DNA, although the intercept of the viscosity
plot m a y be slightly lower. If we list K values, RNA has 65 9 10-~ which
is to be compared with 8 5 . 1 0 -5 of the denatured-aggregated DNA
samples and 2 3 . 1 0 -5 of the aggregated poly(adenylic acid). We thus
conclude that the unperturbed dimensions of RNA are fairly similar to
those of the partly paired polynucleotides. Once more, as for the single
stranded materials, it appears that the effects of varying base com-
position on the unperturbed dimensions are rather small, and that the
effects on the expansion factors are also small both in SSC and HMP. It is
only at much lower concentrations of counter-ion that RNA chains can
undergo large expansion. From a molecular point of view, such an effect
of the ribose oxygen at the 2-position on the stability of the intrachaiu
bonding is an intriguing question.

IV. D i s c u s s i o n s o n R e l a t e d P r o b l e m s
A. Second Virial Coefficient
Tile problem of the second virial coefficient of flexible chain polymers
has three phases. In the first place, there is the problem of interchain
interactions; in the second, the problem of the intraehMn interactions;
and finally, the problem of the coupling of the inter- and intra-chain
interactions. Different approximations are possible for each phase of the
problem, various combinations of such approximations are also possible,
and consequently we are confronted b y a wide variety of possible theories.
For example, the present development of a new excluded-volume equa-
tion for the intrachain problem almost doubles the variety of As-theories,
since most existing theories have been developed on the basis of the
Flory expression for the excluded volume effect and the new theory can
be combined, in principle, with all of these theories. On the other hand,
we have as yet no computational data for As; and furthermore, the
experimental accuracy of A s measurements is still rather poor, as il-
lustrated in Fig. 2. It is therefore unlikely that one particular combina-
tion of several approximations can be properly chosen from among the
Intrinsic Viscosities a n d U n p e r t u r b e d Dimensions of L o n g Chain Molecules 277

m a n y possibilities by means of experimental tests alone. For this


reason, we discuss here the logical influence of the new excluded volume
theory on the A 9. problem.
Two assumptions are essential for the excluded volume theory,
the pair-wise additivity and the short-range nature of the potentials
of average force between non-bonded segments. With these assumptions,
the expansion factor a for a single chain can be expressed in terms of the
parameter z defined by Eq. (16), and the real dimension of the chain can
be described by the two parameters, A and B of Eqs. (33) and (34). Since
the interactions between non-bonded segments would be the same whether
the participating segments belonged to the same molecule or not, and since
the second virial coefficient is essentially the problem of the excluded
volume effect of two entangled chains, the above assumptions m a y be
retained in the present problem, and A 2 m a y also be expressed in terms
of the same parameters, A and B or z [ZIMM, STOCKMAYER and FIXMAN
(276)]. ZI~tM (277) first developed a theory of this type z, and presented
the well-known equation,
A 2 = ( 1 / 2 ) N a B h (z), (101)
with
h (z) = 1 - 2.865 z + 0 (za) , (102)
where N~ is the Avogadro number. The coefficient of in Eq. (102) is
theoretically exact, and it also fits the experimental data in the vicinity
of the theta temperature, as has been shown by KURATA and YAMAKAWA
(157) and b y •UJITA, LINKLATER and WILLIAMS ( ~ ] 5 ' ) . Since the first
and second terms in Eq. (102) are determined b y interchain interactions
alone, the agreement of the parameter values, B and A, evaluated from
Eq. (102) with those from Eq. (18) is remarkable, and demonstrates the
essential correctness of the two-parameter perturbation theory [STocK-
MAYER (238)].
The next term in Eq. (102) depends on both the interchain and
intrachain interactions, and hence on the coupling effect. To avoid the
mathematical complexity of the coupling problem, a simplified model
called the uniformly expanded chain is often employed. This consists of
replacing the original chain b y an effective gaussian chain with a con-
ventional bond length s, a'-----a2a. As the expansion of the molecule
would be affected b y interchain interactions in the proximity of the
second molecule, ,r will not b e identical to x for an isolated molecule.
Assuming this simplification, ALBRECHT (23) evaluated exactly the
coefficient of z ~ in Eq. (102):
h (z) = I . - 2.865 (z/a,~) + 9.73 (zion) ~ . . . . . (103)
z Historically, t h e Z i m m t h e o r y of A ~ was developed before t h e t r e a t m e n t of
t h e excluded volume effect b y FLORY.
2 A similar model is used in P e t e r l i n ' s t r e a t m e n t (1~) of [7] a n d s 0.
278 M. KURATA and W. H. STOEKMAYER:

In this equation, however, c,2 remains unknown, though we m a y put


as ~ a in the first approximation 1.
YAMAKAWA (266) investigated the perturbation theory of a 2 and
derived
a~ = 1 + 2.043 z - 0(22). (104)
This result is still not exact because of the neglect of several complicated
b u t small interaction terms, but it represents the most rigorous among
the existing theories. Despite this neglect, the coefficient of z is already
m u c h larger than the corresponding coefficient, 1.276, in the expression
for 0r~. CASASSA (dQ) also obtained a large coefficient, 1:858, for the same
term b y using the cruciform model for a pair of interacting molecules.
His t r e a t m e n t gives a clear picture of the larger expansion in the inter-
acting state, but, in view of the approximations employed, we prefer
to use the larger value, 2.043, for further investigations. Very recently,
PTITSYS and EISNER (219) have also called attention to the importance
of the as problem. The combination of Eqs. (104) and (103) yields
h(z) = 1 - 2.865 z + 18.51 z~ . . . . . (105)
The application of this equation is, however, limited to the vicinity of
the theta temperature, where z is small.
To obtain a closed expression for A 2, suitable for all values of z,
two types of theories have been developed b y several authors in recent
years. The first type of theory is based on the uniformly expanded chain
model and on a spherically symmetrical distribution of segments about
the molecular center of mass. The segment distribution is taken to be a
spherical cloud of constant density in FLORY'S first theory (101), a Gauss-
tan function about the center of mass in FLORY and KRIGBAUM'S (103')
and in OROFINO and FLORY'S (204) theories, and a sum of N different
Gaussian functions in ISlHARA and KOYAMA'S theory (132'). All of these
theories m a y be summarized in the following type of equation given b y
OROFINO and FLORY,
hFx o (z) = In (1 + Coz~-3)/Cozo~ -~ , (106)
where C O is 2.30 according to the original authors. This equation, with
FLORY-FOX'S viscosity equation, leads to Eq. (12) given in Chapter I of
this paper. However, to secure agreement with the exact expression (102),
the constant C o must be set equal to 5.73, otherwise this equation cannot
be suitable for small z. Similarly, the a in Eq. (106) must be replaced b y
a2 in the Albrecht and Y a m a k a w a equations, (103) and (104). Thus, we
m a y rewrite Eq. (106) as
hFKo,~(z) ---=In (1 + 5 . 7 3 z ~ 3 ) / 5 . 7 3 z a ~ s , (107)
1 If the coupling between inter- and intra-chain interactions were rigorously
taken into account, the expression for h (z) would not be so simple. Different expan-
sion factors would be needed tot each term in Eq. (103).
Intrinsic Viscosities and Unperturbed Dimensions o1 Long Chain Molecules 279

a n d we m a y call this equation the modified Flory-Krigbaum-Orofino


(FKO) equation 1.
The second t y p e of t h e o r y is based on the uniformly e x p a n d e d chain
model and on a spherical distribution of segments about the locus of an
initial interchain contact, r a t h e r than about the molecular center of mass.
CASASSA a n d MARI~OVlTZ (50) developed this t y p e of t h e o r y and obtained

hcM(z ) ----[I -- exp(-- 5.68zo~2a)]/5.68zve[ a . (108)


This equation automatically gives the correct expression for small z.
To complete the equations (107) and (I08), we must express ~ in an
appropriate closed form, and we are faced by at least two choices of this
form, even if the initial z-dependence of cr2is set equal to that in Eq. (104).
These are
o ~ - oc~ = C2g(o~a) z (109)
and
a~ - - a~ = C a z , (110)

with C a = 2.043. However, as has been shown in Sections I I A to I I D, at


least four differently defined expansion factors do not obey the (~5 _ aa)-
t y p e equation for a wide range of z, b u t rather the (0ra - x)-type equation.
T h e quantities 0cL and a furnish the clearest examples, principally because
of the existence of extensive c o m p u t a t i o n a l data. Accordingly, it seems
somewhat unlikely t h a t au, which is essentially a modification of cr
should behave exceptionally a n d adhere more closely to Eq. (110) t h a n
to Eq. (109).
Let us investigate this problem more carefully. Combining Eq. (101)
with the viscosity equations (53), (6) and (7), we obtain

A 2MI ['7] = (915- 1021/(/)o) (ZlOty) h (Z). (1 1 1)


As to c%, we believe we have been able to establish the essential correct-
ness of the (aa _ ~)-type equation in the preceding chapters; therefore
we adopt Eq. (56). I n the region of small z, the q u a n t i t y AaM/E~I] is
doubtless an increasing function of z which starts with zero at z = 0.
I n the region of large z, the factor (z/a~) becomes practically c o n s t a n t :
lira (z/aS,) = i / C , g(~) = 0.59, (112)
Z ---~ o o

since g (c~)= (4/3)'1. and C, = 1. I0. A t this limit the behavior of A aMI [~]
would simply reflect t h a t of h (z). T h e essential feature of Eq. (110) is the

x ISIHARAand KOY'AMAhave reported another closed expression,


hl~(z) -----1.38(C~zcr-3) [ln(Cxztz-~)] '[, , (106")
with C~ = 8.75. This equation is valid only for large z, and gives almost the same
result as Eq. (106). Therefore, we may include this equation in Eq. (106). If we
adjust the parameters in the Isihara-Koyama theory to give the correct result
-for small z, we obtain Cs = 24.5 instead of 8.75.
280 1V[. K U R A T A a n d W . H . STOCKMAYI~R:

unlimited increase of the quantity (z/o~) with increasing ~2 or z, since


z/o~ = ( ~ - 1)/C~. (113)
This relation requires the function h (z) to decrease rather rapidly with
increasing z and finally to approach zero in the limit of infinite z, because
the numerators in Eqs. (107) and (108) are rather insensitive to the
variable (z/~) for large z. Consequently, if we adopted Eq. (110) for a 2,
the quantity A2M/[~I] would become a decreasing function o/ z in the
region of large z. In other words, A2M/[~] as a function of z would have
to display a maximum at some value of z. On the other hand, according
to Eq. (109), fhe variable (zloty) is bounded b y the lower limit
lim (z/~r = 0.65]C 2 = 0.315, (114)
z--~ oo

and the function h (z) becomes practically constant for large z. Conse-
quently, the quantity A 2M] Erl] would no longer display a maximum, b u t
would monotonously approach an asymptotic value, namely:
lim(AeM[[~l] ) = 120 ( F K O ) , or 98 (CM), (115)

w h e r e we have taken ~50 = 2.6 9 1021.


As is well known, most experimental values of A~M/[,1] for typical
amorphous polymers lie somewhere between 80 and 150 [see Fig. 2 in
Chapter I1, while those
/.a N ~ , , , , for cellulose derivatives
I\ 2"/r Eq. (mod/~)Jr//h are generally lower, say

/\
0.8' \~
for example, HOLTZER,
BENOIT and DOTY
'e'~ at! "'~.~~ (726); HUNT, NEWMAN,
SCHERAGA and FLORY
"~ "~, ~ a~'/S (129) ; and MANLEY
0.8 " ~ . \ . ~ I (172) ]. Sincethe cellulose
derivatives generally
a~- 'x,..1,~ correspond to very large
~ i i values of z, this situation
e 0./ e.~ o.8 o.d5 apparently favors the
(z/~) existence of a maximum
Fig. 29. The function b(*at-* ) related to th . . . . . ti . . . . . . d virial in Afll/l/[rl], and hence
coefficient. Full curve, modified Flory-Krigbaum-Orofmo theory,
Eq. (107); chain curve,.Casassa-Markovitz theory, Eq. (108) suggests the superiority
of Eq. (110). However,
we must also recall the experimental difficulties of A2-measurements for
cellulose derivatives or other crystalline polymers. We may obtain a general
idea of the experimental accuracies in viscosity and A~ measurements
from comparison of the left and right halves of Fig. 1, and then imagine
the expected accuracy in A 2 measurement for cellulose systems from the
Intrinsic "Viscosities and Unperturbed Dimensions of Long Chain Molecules 281

viscosity d a t a in Fig. 20. W e are also aware of the large scattering of


A~M/[~7] d a t a for polyethylene and of the clearly underestimated d a t a
for ataetic polypropylene, b o t h given in Fig. 2. I n view of these facts,
we are inclined, at least at the present stage, to reject the possible
existence of the m a x i m u m in A~M/[~] and to accept Eq. (109) r a t h e r
t h a n Eq. (110).

PMMA ' ..j/.. .-'"E .f-

i i j 9
I /
#I ;j ,,"
9 11 j /
/I / I
100 / ~ ,,"

I iI o~ ," /./'/"
|
/ ~ ,,
m" ~ " / /-'- " M~J'IO~
I /~ "/--'~'/" ~ 2/. 10~
uU H "B/y ~ "
I ,lo#;: 9 :/gO. I0 ~
I mj~./ -- f K O - m o d , fflI.

~ : - - - h(z,)~I

0 I 2 d c~
-,

Fig. 30. A~M[[*/]as a function of (ct~-- l). Points for poly(methyl mcthacrylates) in various solvents, data of
SCHOLZand I{msTr (229")~ Curves according to Eq. (111) with various expressions for h(z)

Figure 29 shows two functions, Eqs. (107) and (108), in the domain
of the variable, z]~. I t is remarkable t h a t the two functions differ from
each other only a b o u t 20% at most. This is an i m p o r t a n t consequence of
the present t r e a t m e n t 1.
V e r y recently, extensive unpublished d a t a on A 2, [7] and [~]o
for several p o l y ( m e t h y l methacrylates) were m a d e available to us b y
Professor G. V. SCHULZ. These are all plotted in Fig. 30. These plots are
m a d e b y the use of [M]o values obtained in each solvent ~. The theoretical
values obtained from Eq. (111) with Eqs. (107), (108) and (109) are
1 As a l r e a d y n o t e d , t h e Y a m a k a w a v a l u e , C~ -----2.043, is a l o w e r limit. H e n c e ,
E q . (114) m u s t g i v e a v a l u e l y i n g a b o v e t h e c o r r e c t u p p e r b o u n d of e/e~.
The employed theta solvents are: (i) isoamyl acetate at 57.5 ~ C, (ii) heptanone-4
at 40.4 ~ C, till) butyl acetate at --200 C, (iv) methyl isovalerate at --37 ~ C, (v)
2-methylpentanone-4 at --42 ~ C, (vi) several mixtures of methyl ethyl ketone and
isopropanol at 4, 12.8, 22.8 and 28.5 ~ C, and finally (vii) butyl chloride, whose theta
temperature shows a scattering from 31.8 to 35.4 ~ C.
282 M. KURATA a n d W . H. STOCKMAYBR:

shown by the solid and chain lines, together with the asymptotic values,
Eq. (115). There are also given two other lines; the broken line represent-
ing the combination of Eq. (111) and h(z)=--1, and the dotted line
representing the original Flory-Krigbaum-Orofino equation (12). In spite
of the wide variety of molecular weights, temperature and solvents, the
experimental points produce a fairly definite curve, especially in the
vicinity of the origin. In this region, the agreement between theory and
experiment seems good, confirming the foregoing statement. It may also
be noted that the original FKO equation gives a wrong initial slope as
expected. In the region of large expansion, the experimental points
appear to approach an asymptotic value, though the precise figure is
uncertain because of scattering of the data. However, as is easily shown,
the theoretical value of A~M/V~I] is 160 for an impermeable sphere [see
ALBRECI~T (23)~, and zero for a rigid rod. As the flexible chain is in a
sense intermediate between these two extremes, A~M/[~I] for chain
polymers should not greatly exceed the limit 160, which is indicated by
the shaded horizontal line in the right strip in Fig. 30. In fact, the broken
curve which should serve as the theoretical upper bound for A2M/[~]
approaches 210 in the limit of infinite ~,1. This value m a y be too large.
Although many experimental points exceed the broken line, the general
disposition of the data seems to support this conjecture. The shapes of
the two theoretical curves (solid and chain curves) are also tolerable.
In this connection, an interesting feature of the Fio W theory m a y
be noted. Multiplying both sides of Eq. (25) by ~ gives
~5_ ~a= C~,z (116)
with C~ = (134/105) g (~) a2. (117)
From this equation, C~ = 2.06, 2.52 and 3.04 for ~ = 1.4, 1.7 and 2.0,
respectively. FLORY'S original constant in Eq. (22), i.e. CF = 2.60,
agrees well with an average of these C~ values. This indicates that the
F l o w theory is appropriate for the use in this rather moderate range of ~.
The original FKO equation (12) (the dotted curve) also demonstrates
this behavior in Fig. 30. The so-called good-solvent systems of ordinary
amorphous polymers, such as polyisobutylene-cyclohexane, mainly fall in
this range if the molecular weight is of an ordinary magnitude, say l0 s
or 106 (see Fig. 2) ; therefore, the parameter values ~p and ~ obtained for
these systems by FLORY and his collaborators retain an approximate
validity. However, the parameter values obtained for poor-solvent (or
near-theta) systems should be corrected by a factor of about two, as
already pointed out by STOCKMAYER (238), and those obtained for
extremely good solvent systems should probably be discarded.
T h i s c u r v e is e s s e n t i a l l y identical to t h e K r i g b a u m e q u a t i o n (11) o v e r t h e
whole r a n g e of ~.
I n t r i n s i c Viscosities a n d U n p e r t u r b e d D i m e n s i o n s of L o n g C h a i n Molecules 283

Returning to the present theory, we must note several important


and necessary consequences. First, the adoption of Eq. (109) for a2 sets a
limit to the role of the function h(z) in the A 2 problem. This flmction
has been allowed to vary from unity to zero in previous theories, but
now it is a function decaying down to only about 0.6 or 0.5 instead of
zero. As this function is the only source of the molecular-weight depend-
ence of A v as readily seen from Eq. (101), A s can then no longer display
such a marked dependence on M as often given in the literature. A change
of about two-fold m a y be possible, in the present theory, from the highest
to the lowest molecular weight, say 107 to 104, but not more. If we
write A s in the usual empirical form,
A s = const. M-:', (118)
then the above condition that A2 (M = 104)]A~(M = 107) ~ 2 indicates
the largest possible value of ~ to be about 0.10. This type of equation,
of course, could not be valid for a very wide range of M ; therefore the
exponent ~ may have a larger value, say 0.15, in a limited range of M,
but it certainly cannot be as large as 0.2 or 0.3. This statement seems to
conflict with t h e accepted view. In view of importance of the problem,
we must examine this point further.
A rather typical example of a high 7-value may be found in the
results of CHINAI and SAMOELS (67) for poly(ethyl methacrylate) frac-
tions in methyl ethyl ketone. Their values of A 2 are reported to be well
fitted by Eq. (118) with an exponent of ) / = 0.30 over the molecular
weight range from about 0.5 to 2.6 million. In the right half of Fig. 1,
these results are shown as a plot of A~M/[~I] against (a~n- 1). The two
points of lowest molecular weight and highest ordinate must be regarded
with suspicion, and t h e y were in fact disregarded by the original authors
in applying Eq. (118). T h e remaining seven points give values of A~M[ [~/]
near 150, in good agreement with the now widely accepted figure for
high polymers in good solvents. However, if we inspect the points more
closely we see tha t they can only be connected b y a line with a negative
slope, which crosses the curves sketched in the diagram. This situation
is clearly at variance with any two-parameter theory, according to which
there should be a universal curve of A~M/[rl] against (c% ~ - 1). Because
the range of molecular weights involved is only five-fold, it seems very
likely that the high value of T should be attributed to a false trend in the
experimental measurements, or perhaps to an inadequate method for
handling the effect of the third virial ceofficient (see below); we m a y
also recall that light-scattering measurements at only three scattering
angles were made in this particular investigation. Conversely, if we
accept the alleged ~,Tvalue of 0.30, we must abandon the notion of a two-
parameter theory. This conclusion holds for all two-parameter theories.
Fortsehr. Hochpolym.-Forsch., Bd. 3 19
284 M. KURATAand W. H. STOCKMAYER:

In passing, it is interesting to note that a horizontal line through the


points in the right half of Fig. 1 would correspond to V = 0.20, the
Krigbaum equation (11) to V -----0.10, and the original F K O equation (12)
to 7 = 0.06.
Many other high values of 7 have been reported in the literature, but
invariably the points for the lower molecular weights fall into the upper
left area of the plot of AsM/[~]l against ( ~ 2 _ I). Some examples are
seen in Fig. 2. As with the data of Fig. I, we are obliged either to reiect
these trends in the data as spurious or else to abandon faith in the
a d e q u a c y of any two-parameter theory.We prefer the former alternative,
not only intuitively but also because the supremely careful recent
measurements of SCHULZ and KIRSTE, shown in Fig. 30, give strong
support to the two-parameter idea.
Next we examine certain measurements which have often been used
previously to test theories [KRIGBAUM and FLORY (IriS); ISIHARA and
KOYAMA (132') ; CASASSA a n d MARKOVITZ (50) ; CASASSA (d9) ; and KRIG-
BAUM, CARPENTER, KANEKO and ROIG (146')1. These are KRIGBAUM and
FLORY'S osmotic data for the polyisobutylene-cyclohexane and poly-
styrene-toluene system, for which the 7 values are reported to be 0.14
and 0.22, respectively. The former value is not very large, and it can be
accommodated b y an appropriate two-parameter theory. In fact, these
data when plotted in Fig. 2 display a rather normal behavior, as shown
b y the large white circles. The other value, 7 = 0.22, is more troublesome,
but if we plot these data in Fig. 2, we obtain the dotted squares, of which
the four lowest molecular weight points fall above the solid curve,
as expected from the large y value. OROFINO and FLORY (20~]), in
testing their equation (12), selected BAWN and WAJID'Sosmotic measure-
ments for the same polystyrene-toluene system (35) ; these are shown b y
the large white squares in Fig. 2. The y value for these data is 0.15,
instead of 0.22.
As is well known, the value of A s obtained from a set of osmotic
pressure measurements is much influenced b y the treatment of the third
virial coefficient A s in the series,

a l e R T = (l/M) + Asc + A n c s + 9 9 (119)

where a is the osmotic pressure, R is the gas constant, c is the concen-


tration in mass per unit volume. I t is customary to write

A s = gA~M, (120)

and if g is approximated b y 0.25, then Eq. (119) m a y be written

(re/cRT)'/, = (i/M)lh[1 + (A~M/2) c] + . . . (121)


Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 285

[see FLORY (ha)]. A n o t h e r m e t h o d is d u e to SAWN et al. (34). I f a n y t w o


values of c a n d vr/c are selected on t h e e x p e r i m e n t a l curve, t h e n ( w i t h
neglect of h i g h e r terms)

( z l c R T)I-- (zc/cR T ) 2 = A 2 ( c l -- c2) + A ~ ( c ~ -- c~) , (122)


and accordingly
(z~IcRT)'--(n/eRT)2-- A 2 + A 3 (q + c2). (123)
C 1 -- C2

B y t a k i n g such p a i r s of points, a n d p l o t t i n g t h e t e r m on t h e l e f t - h a n d
side a g a i n s t q + c2, a line of i n t e r c e p t A 2 a n d slope A s should be o b t a i n e d .
T a b l e 12 i l l u s t r a t e s t h e s e t w o m e t h o d s for t h e o s m o t i c d a t a of BAWN

Table 12. Second virial coefficient A 2 of polystyrene in toluene at 25 ~ C


At 910' [moleg.-t cm.I]
As 9 10*
Mn 9 10 -4 Florymethod [moleg.-Scm.S]
Bawn method I
g = 0.25 [ g ~ 518

158 2.89 2.19 1.83 1.86


87 3.01 2.70 2.13 1.70
32.0 3.28 3.19 1.58
28.8 3.09 3.58 3.03 1.42
24.8 3.32 3.31 1.54
16.2 3.36 4.03 3.86 1.70
9.40 3.89 4.72 4.47 1.55
7.20 4.35 5.13 4.58 1.78

0.15 0.26 0.33

(34, 35) ; also given are t h e A 2 v a l u e s o b t a i n e d b y t h e F o x - F l o r y - B u e c h e


m e t h o d , g = 5/8 (117'), a n d t h e A 3 values o b t a i n e d b y BAWN'S m e t h o d 1.
T h e A 2 v a l u e s f r o m Eq. (121) are in g o o d a g r e e m e n t w i t h t h e K r i g b a u m -
F l o r y d a t a o b t a i n e d b y t h e s a m e m e t h o d (748), i n d i c a t i n g t h a t t h e
m o l e c u l a r - w e i g h t d e p e n d e n c e of A 2 does reflect t h e m e t h o d of t r e a t m e n t
of t h e original osmotic pressure d a t a . T h e A 3 v a l u e s o b t a i n e d b y BAWN'S
m e t h o d m a y n o t b e free f r o m error due to neglect of t h e higher terms,
b u t t h e m e t h o d seems a t least as reliable as t h e g - m e t h o d . A c c o r d i n g
to STOCKMAYER a n d CASASSA (239') [see also STOCKMAYER (20)], t h e
n u m b e r g is clearly a f u n c t i o n of m o l e c u l a r weight, so t h a t t h e use of t h e
s a m e g over a wide r a n g e of m o l e c u l a r weights m a y be criticized, t h o u g h
g = 0.25 is a r e a s o n a b l e choice from the t h e o r e t i c a l p o i n t of view.
A m o n g l i g h t - s c a t t e r i n g e v a l u a t i o n s of A , , one of t h e m o s t e x t e n s i v e
sets of m e a s u r e m e n t s is t h a t of OUTER, CARR a n d ZIMM (207) for p o l y -
s t y r e n e in m e t h y l e t h y l k e t o n e a n d dichloroethane, These d i s p l a y a c h a n g e
t This table is essentially a reproduction of Table 1 in the paper of BAWN and
WAJID (35), although the results obtained by Eq. (121) are newly added.
19"
286 M. KURATA a n d W. H. NTOCKMAYER:

of a b o u t three fold from the highest to the lowest molecular weight.


However, there is an apparent break in a plot of A~ against M between
molecular weights of 230,000 and 138,000, which coincides with a change
in the method; for the higher molecular weights an extrapolation to zero
scattering angle (Zimm plot) was employed, while for the lower molecular
weights only data at 90 ~ were used. Since present knowledge of inter-
molecular correlation in light SCattering from polymer solutions is
relatively meager [see, for example, ALBRECHT (23')], we m a y with some
justification suggest that a lower value of y would also be possible in this
case. Thus we really find no experimental evidence in definite and
inviolable contradiction to our conjectures.
The problem of primary importance is after all the absolute magnitude
of A s, especially that of the agreement between the B value obtained
from A s and that obtained from intramolecular properties such as [~/] or
(S~). This is roughly demonstrated b y the approximate agreement
between the broken curve corresponding to h (z) ~ 1 and the experimental
points in Fig. 30. Another quantity, A~M/(S2) "h, could be used in
principle for the same purpose, but it cannot be recommended for general
use because of the present inferior precision of (S ~) measurements; the
combination of the two quantities A s and (S ~) is still dangerous to use
for critical discussions. The molecular-weight dependence of As is a
problem of secondary importance. Since this part of the problem involves
information about the quantity A, and since the existing theories are
still incomplete, as easily seen from Fig. 30, the problem is of course
interesting, both experimentally and theoretically. However, the relative
magnitude of this molecular-weight-dependent part of A s is probably
smaller than the past expectation, so that special care will be required
for future investigations of this problem.
The picture of the A s problem given above has a striking resemblance
to that obtained from the recent lattice theories [see, for example,
I{URATA (15d)] 1. This is encouraging, because the distribution-function
method and the lattice statistics should converge to the same conclusion,
at least on a semi-quantitative level.
Finally, we return to a practical problem, i. e., the chain conformation
of stereo-specific polymers. Since DANUSSO and MORAGLIO (73), it has
been repeatedly reported that the isotactic and atactic species of a given
polymer show the same viscosity-molecular weight relationship but dif-
ferent second virial coefficients. This statement m a y be readily confirmed
x Besides t h e earliest theories b y FLORY (98'), HUGGINS (128") a n d MILLER
(~83'), all improved theories are concerned with t h e effect of ring closure of chains on
t h e mixing e n t r o p y [STAvERMAN (237); GUGaENHEI~ and McGLAsI~AN (7); TO~tPA
(21) ; MONSTER (15) ; a n d KURATA, TAMURA a n d WATARI (1,{6)]. T h e p r o b l e m of ring
closure is essentially t h e excluded v o l u m e problem.
I n t r i n s i c Viscosities a n d U n p e r t u r b e d D i m e n s i o n s of L o n g C h a i n Mol e c ul e s 287

f r o m Fig. 31, for example, where DANUSSO and •ORAGLIO'S data for
atactic polystyrenes are shown b y the large black circles and those for
the isotactic polymers b y the large white circles. The solvent is toluene
in both cases. We can distinguish between these two polymers in the
figure for A s, but certainly not in the figure for [~]. For the atactic
polystyrene-toluene system, we have already shown a more complete set
of viscosity data and determined the parameter values A and B. These

3 I ~ I ] I I III I I I I I IIII 10 .3
~f'F

./o L .
d O ~ 9 e ~ oo 9 9
o 9
-z

1 /tT-~

1 ~I I lllll I I I I IIIII
E 10 10 5 10 r
er, v M 3/ (,71'/' 9 1. M,,
Fig. 31. Viscositiesand virial coefficientsof polystyrenes. Open circles, isotactic polymers in toluene (73, 252).
Triangles, isotactic polymers in benzene (24). Filled circles, ataetie polymers (73)

values correspond to the chain line in the left figure. Then, using these
values and the CM equation (108) tentatively, we obtain the calculated
values of A 2 which are also shown b y the chain line in the right figure.
For isotactic polystyrenes two other viscosity studies have been reported,
one for benzene solutions b y AI~G (24) (triangles), and the other again for
toluene solutions b y TROSSARELLI,CAMPI and SAIm (252) (small white
circles). ANa's data seem to give a larger intercept than that for the
atactic polymer. If we tentatively assign the two solid lines to two
series of toluene solution data, then we can evaluate the corresponding
A2-M relationships b y using Eq. (108). The result for DANUSSO and
MORAGLIO'S data is given in the right figure b y the solid line, We may
conclude from this demonstration that a difference in A 2 of this degree,
say 20 %, need not necessarily be followed by a notable difference in the
viscosity-molecular weight relationship over a limited range of molecular
weight. Possibly a more extensive viscometric study would reveal a
288 M. KURATA a n d W. H. STOCKMAYER:

difference between these two species. In Table 7, we have assigned a


slightly larger K value to isotactic polystyrene because of ANG'S data.
On the contrary, in the case of polypropylene, the existing viscosity
data suggests a slightly s m a l l e r K value for the isotactic species. This
conjecture is still uncertain, but it m a y have some sense if it is true. At
first thought, these two polymers both belong to the category of non-
polar polymers, so that their conformation might be expected to be very
similar. However, this m a y not be the case, for the value of a for poly-
styrene is abnormally high in relation to the substituent volume V~, as
remarked earlier. In the opinion of BIRSTEIN and PTITSYN (39'), this is
probably due to the existence of short helical sequences (as in the isotactic
polystyrene crystal), particularly for the isotactic polymer. In this
connection, it is also interesting that ARCUS and WEST (25') have found
presumptive evidence for helical conformations in certain optically active
methacrylate polymers.

B. Short Range Interactions


A complete discussion of the theory of short-range interactions in
chain molecules is far beyond the aim of this paper, and in this section
we confine ourselves to a few relevant definitions and remarks. The best
detailed treatment of such problems, although already incomplete, is the
book of VOLKENSTEIN (27').
First we summarize the several alternative ways that have been used
to express experimental results on unperturbed dimensions. In this
paper we first introduced a skeletal factor s b y the definition.
s - - (L~>o/nl ~ = 6 ( S ~ ) o / n l ~ , (124)
but before considering numerical values we transferred attention to the
steric factor a, most generally defined b y
a2~ (L2)o/(L2)o I . (125)
This is a useful maneuver, because (L2>0f for chains with free internal
rotation is easily computed if the chemical structure is known; an
example is the familiar formula
(L2>ot = n l 2 ( 1 + u)/(1 -- u); u---- cos0 (126)
for chains with only one kind of bond, 0 being the supplement of the
valence bond angle. In his recent review, PETERLIN (76') tabulated
experimental values of unperturbed dimensions in terms of a.
Other workers have preferred to characterize the chain b y an effective
link length a, defined b y the relation ( L * ) o = N a 2, in which N is a
number proportional to the actual number n of chain bonds. One possible
choice is 'to set N = n; the effective link lengths reported by OUTER,
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 289

CARR and ZIMM (207), for example, correspond to this choice. A more
common procedure is that of KUHN (153), whose statistical c h a i n e l e m e n t
is obtained by applying the condition 1
Na = Lma x ~ n l c o s ( 0 / 2 ) . (127)

Other investigators prefer to record the constant A ~ ( ( L Z ) o / M ) '/',


which has the advantage of quick reconversion to the dimensions of
a polymer of known molecular weight. In our tables and in those of
FLORY (5g) both A and ~ have been tabulated. Still another expression
of unperturbed dimensions is found in the p e r s i s t e n c e length, which
conceptually is defined as the average projection of an infinitely long
chain along the direction of its first link. For a chain with free internal
rotations, the persistence length a~ is easily evaluated as

a~ = ~ l cosk0 = l/(1 -- u) , (128)


k=0
while for real chains this must be replaced by
a~, = a ~ l/(1 -- u ) . (129)
The persistence length has been most frequently emp!oyed by KRATKY
and POl~OD (142, 142') in determining chain dimensions from small-
angle X-ray scattering measurements ~.
Theoretically the factor a differs from unity only because all the
possible sets of values of internal rotational coordinates are not equally
probable. As we have suggested in our main discussion, the most common
causes of energetic bias in rotational isomerism are probably steric,
although electrostatic forces involving polar groups are sometimes
obviously involved. In the simplest theories, it is assumed for con-
venience that each internal rotational angle is governed, independently
1 Still another possible choice is afforded b y writing
Na -~ L~ = nl , (127')
where Lr is called the contour length of the chain [see S e c t i o n l I 13].
s These authors have preferred to t r e a t the so-called worm-like chain, which is a
limiting form obtained from the customary chain with fixed valence angles by
allowing the bond length l and the bond-angle supplement 0 to become very small
and the number n of bonds to become very large while keeping the contour length
L , = n l and the persistence length a~ finite. I n this way, the chain becomes a
continuous thread or wire with a bending modulus proportional to the persistence
length. The virtue of this model is perhaps t h a t the resulting parameters are un-
prejudicedly phenomenological, and t h a t the rather neat formula
(L~)ol2a~ = x - - 1 + e-" ; x - - Lda~, (130)
describes the mean square displacement length for chains of any length; compare
the relatively complicated expression of BENOXT (37) for finite chains with fixed
valence angles and independent hindered internal rotations. A good discussion of
the relation between these models is found in Tolvtt,A's book ( 2 1 ) .
290 M. KLIRATA a n d W . H . STOCKMAYlgR:

of the other internal rotations, b y its own potential energy function.


Under these conditions, the average quantities
~/--= <cos~>; s~ (sinq0> (131)
are the same for all chain bonds, and correlations with neighboring bonds
are neglected. In general, and in particular for polymers with repeating
u n i t s of the types --CHtCHX-- and --CH~CXY-, the potential energy
governing an internal rotation is not symmetric about the trans conforma-
tion, i. e., the potential energy is not an even function of the internal
rotational angle 9- Then the stereoregularity of the chain also plays a
role. The formula of BIRSTEIN and PTITSYN [as given b y VOLKENSTEIN
(21', 258)] for this case is
0"2 = [1--~/~ - - e~J [1 + ( 2 p - - 1) (~/2 + e2)]
(1 - - ,1) 2 + ( 2 p - - 1) (~/-- ~12_ e=)~ , (132)
where p is the probability of a syndiotactic placement. The special cases
of completely isotactic or completely syndiotactic chains are obtained
b y setting p equal to zero or unity, respectively. If the rotational poten-
tim energy happens to be symmetric, as for chains with repeating units
like - C H ~ - or --CH2CX~-, it follows that e = 0, and then the above
equation reduces to the well-known result [BENOIT (37), TAYLOR (245a)
and others] :
~ = (1 + ~)/(1 - ,]), (la3)

which was already quoted in Eq. (86). I t should be noted that this simple
result can not be obtained from (la2) b y putting p = 1/2.
For actual chains, the assumption of independent internal rotations
cannot be valid, because of the steric interactions between substituent
groups on second-neighboring chain atoms and beyond. Various degrees
o f approximation exist, corresponding to the number of correlations
explicitly considered. As an example, we may give a formula of PW~TSrN
and SHARONOV[(216); see also VOLKENSTEIN(21', 258)] for chains of the
- C H 2 C X 2 - type:
a s = [1 -- ~]= + t 1 + - ( ~ 2 - - rh)]/(1 -- ~])2, 034)
where the new symbols are defined b y
~1 = (cos q0=,cos 93,+ 1) ;
(las)
q = (sin ~Y2~sin ~%,+ 1),
in which ~2, and 92,+, are the internal rotational angles about two suc-
cessive bonds between adjacent - C X = - groups. The quantity u, as
earlier, is the cosine of the bond angle supplement. If the correlations
were negligible, we could put */1= ~/~ and e l = 0, then recovering Eq. (138).
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 291

For vinyl polymers, the corresponding formulae contain an additional


correlation, as well as the stereoregnlarity p a r a m e t e r p. They will not be
displayed here. More recently, Yoo and KINSlNGER (270') have developed
equations for a in the case t h a t two parameters are required to describe
the stereoregularity of the chain. Other recent papers on short-range
interactions include those of HOEVE (72J), TOBOLSKY (249'), HIJ~tANS
and HOLLEMAN (123') and MIYAKE (185', 185"), but we cannot discuss
these here. The recognition of matrix formulations as particularly
economical and powerful for these problems is universal.
It remains for us to discuss the dimensions of polypeptides and other
helix-forming chain molecules. Most of the theoretical works on models
of such chains have been primarily concerned with the equilibrium of the
hehx-random coil transition and have not specifically treated the chain
dimensions. An exception is found in the work of NAGAI (195'), who
combined his theory of the transition with a very simple model of the
chain dimensions. I t amounts to the assumption t h a t each helical
sequence behaves like a rigid statistical chain element without correlation
in direction with the randomly coiled sections which are adjacent to it.
Then, if a fraction ~ of the monomer units are members of helical se-
quences, we m a y at once write
(L~)o = n(1 -- ~) 12s + b2oX, v~, (136)
where I is the bond length and s the skeletal factor of the random-coil
sections, b0 is the length per monomer unit of the helix, and v~ is the
number of units in the i th helical section. Since the total number of units
in helix form is n ~ = 2:~ vi, the above relation m a y be written
(L~)o/n = (1 - ~) 12s + 8b~or, (137)
where r ~ X v~/X,,~ is the weight-average length of a helical sequence.
If the helix content is large (~ ~ 1), we get Eq. (97).
To predict the dimensions of such molecules in the region of the helix-
coil transition, it is of course necessary to have a theory for the depend-
ence of the quantities ~ and r on the model parameters, which in turn
have to be related theoretically or empirically to variables such as
temperature, solvent composition and chain length. NAGAI did this for
typical polypeptides, using parameters appropriate to the data for poly-
~-benzyl-e-glutamate (cf. Section I I I D). He found, in agreement with
several observers, t h a t the initial stages of helix formation actually
produce a small reduction in chain dimensions (and hence in the intrinsic
viscosity) ; but this would not be the case for helices of much higher pitch.
Conversely, it would seem that the chain dimensions (or, as NAGAI pointed
out, equally well the dipole moment) could in principle be used to deduce
reliable values of the model parameters.
Appendix
T~,ble 13. L i s t o f the M a r k - H o u w i n k - S a k u r a d a constants, [r/] = K ' M ~ o f l i n e a r macromolecules

Polymer Solvent x Temp. l ~ Range Method I Reference


M 9 10 -~

Alginic acid (Na salt) Aqueous NaC1 25 7.97 1.0 5 - - 19 OS (84)


Amylose Aqueous KCI (0.33 M) 25 115 0,50 27--220 LS (943
Aqueous KOH (1 M) 25 1.18 0.89 31--310 LS (71)
Aqueous KOH (0.5 M) 25 8.50 0.76 27--220 LS (#4")
Dimethyl sulfoxide 25 1.25 0.87 22--310 LS (71)
Dimethyl sulfoxide 25 30.6 0.64 27--220 LS (943
Ethylene diamine 25 15.5 0.70 31--310 LS (71) >
,q
Amylose acetate Chloroform 30 1.06 0.92 12--480 LS (71) >
Methyl acetate 25 5.60 0.80 7 - - 19 SD (ss) t~
Nitromethane 30 1.I0 0.87 12---480 LS (71)
Amylose tricarbanilate Acetone 20 0.81 0.90 4--490 LS (46)
Dioxane 20 0.90 0.92 4--360 LS (40)
Pyridine 20 0.58 0.92 4--360 LS (,f6)
Toluene 30 16.5 0.78 3--35 OS (54) u~
Buna-S rubber
Cellulose Cuprammonium 20 105 0.66 2--25 OS (9) o
Cupriethylene diamine 25 13.3 0.90~ 1--54 OS (132)
Cellulose acetate-butyrate Acetic acid 25 14.6 0.83 1--21 OS (244)
Acetone 25 13.7 0.83 1--21 OS (244)
Cellulose triacetate Acetone 20 2.38 1.0 2-- 14 SD (125) . .

25 33.0 0.70 OS (9)


25 8.97 0.90 1 - - 18 OS (212)
Acetone (80)]Water (20 vol.) 25 21.0 0.80; OS (9)
Methylenechloride (80)] 25 13.9 0.83, OS (9)
Ethanol (20 vol.)
Cellulose tributyrate Methyl ethyl ketone 30 18.2' 0.80: 8 - - 22 OS (17o)
Dodecane (75)/Tetralin 130 82 0,50 11-- 21 OS (17o)
(25 vol.)
Cellulose tricaproate Dimethyl formamide 41 245 0.50 6---130 LS (151)
Dioxane 35 125 0.57 4--130 LS (151)
Table 13. (Continued)
Cellulose tricaprylate Dimethyl formamide 140 113 0.50 10-- 32 OS (170)
7-Phenylpropanol 48 129 0.50 8 - - 32 OS (17o) o
Toluene 30 17,3" 0.70* 8 - - 35 OS (17o)
Cellulose tricarbanilate Acetone 20 4.66 0.84 7--270 LS
Dioxane 20 4.20 0.88 7--270 LS (46)
Pyfidine 20 3.46 0.86 7--270 LS (~6)
Cellulose trinitrate Acetone 20 2.8 1.00 1--249 SD (181, 232,
1as)
25 7.00 0.933 5 ~ 50 OS (o)
25 1.65 1.00 8--265 LS (126)
25 11.0 0.91 3~100 OS (132)
Amyl methyl ketone 25 5.0 0.93 7 - - 26 OS (lO2)
n-Butyl acetate 25 5.68 0.969 5 - - 50 OS (o)
Cyclohexanone 25 2.24 0.810 7 - - 22 OS (7oi) e~
E t h y l acetate 25 3.8 1.03 3--100 OS (Ia2)
25 1.66 0.86 68--250 LS (1,3o)
30 2.50 1.01 4 ~ 57 LS (12o)
E t h y l butyrate 25 3.64 1.0 5 - - 50 OS (121)
Ethyl lactate 25 12.2 0.92 3 - - 65 OS (1a2)
Methyl acetate 25 18.3 0.835 7 - - 22 OS (ml)
Nitrobenzene 25 6.1 0.945 7 - - 22 OS (191)
Collagen (calf-skin) Citrate buffer 24 8 1.23 1.80 14-- 37 SV (200) 0
(0.15 M, pH 3.7) ~q
Deoxyribonucleic acid
(denatured: D. pneumoniae) Aqueous phosphate (0.01 M) 20 3.46 0.933 25--330 SV (88)
(denatured: E. eoli) Aqueous phosphate (0.01 M) 20 3.11 0.912 s--460 SV (88)
E t h y l cellulose Ethylacetate 25 10.7 0.89 4 - - 14 OS (100) N
Methanol 25 52.3 0.65 I O ~ 41 LS (255) 0
E t h y l hydroxyethyl cellulose Wafer 40 43.7* 0.78* 9 - - 28 OS (172)
Gelatin Aqueous KCNS (2 M) 25 29 0.62 2 - - 10 SE (265) g
Acetate buffer 0.17M, pH4.75) 35 1.66 0.885 5 - - 21 OS (215)
GR-S rubber Benzene 25 54 0.66 1--165 OS (270)
Gutta percha n-Propyl acetate 60 232 0.50 1 0 ~ 20 OS (26o) tO
~o
Table 13. (Continued) . g~

Solventt K' 910* Range Method= Reference


Polymer TempY~ M" 104

Natural rubber Benzene 30 18.5 0. 74 8--- 28 OS (200)


n-Propyl ketone 14.5 119 0. 50 8 - - 28 OS (260)
Toluene 25 50.2 0.667 7--100 OS (48)
Nylon 66 Aqueous formic acid 25 32.8 0. 74 EG (127)
25 110 0. 72 0 . 5 - - 2.5 EG (245)
Polyacrylamide Water 30 68 0. 66 I - - 20 PR (6g)
30 6.3 0. 80 2 - - 50 SD (227)
Poly(acrylic acid) Dioxane 30 76 0. 50 13--122 LS (199) >
P o l y (acrylomorpholide) Dimethyl formamide 25 18 0. 65 LS (20o3 H
>
Poly (acrylopiperidide) Dimethyl furmamide 25 32 0. 56 LS (2099
Polyacrylonitrile Dimethyl formamide 25 57.4 0.733 0.2-- 3 OS (Sl) f~
25 16.6 0. 81 5 - - 27 SD (41)
25 23.3 0. 75 3 - - 27 LS (os')
25 27.8 0. 76 3 - - 58 DV (14o)
25 39.2 0. 75 3--100 OS (202) u~
Poly(adenylic acid) 0.15 N[ NaCI, 0.015 M citrate, 25 38 0.65 I - - 13 SV (lt~) o
p H 7.1 ('1
Poly (y-ben zyl-DL-glutamate ) Dichloroacetic acid 2.81 0.85 1.5-- 10 LS (234")
Dimethyl formamide 37.7 0.55 1.5-- 10 LS (234")
Poly (7-benzyl-L-glutamate) Dichloroacetic acid 25 2.7~ 0.87 2--34 LS (ss) t~
Poly(p-bromostyrene) Benzene 20 95.5 0.53 3--30 OS (137)
Polybutadiene (71% trans, Cyclohexane 25 12 0.77 230--800 LS (683
4 % eis, 25% 1,2)
(79% trans, 0 % cis, 21%1,2) Cyclohexane 20 36 0.70 23--130 LS (2t9'")
(3% trans, 92% cis, 5% 1,2) Benzene 32 10 0.77 10--160 LS (709
Poly(1-butene) Ethylcyclohexane 70 7.3, 0.80 4--130 LS (147')
Poly(n-butyl methacrylate) Isopropanol 23.7 36.6 0.50 40--170 LS (57, 6,0
Methyl ethyl ketone 23 1.5~ 0.81 25--260 LS (57, 64)
Poly-p-(tert-butyl phenyl Acetone 20 5.7~ 0.68 6---352 LS (2s4)
methacrylate)
Table 13. (Continued)
Poly (e-caproamide) Sulfuric acid 25 63 0.76~ 0.4-- 3 EG (224) m

Polycarbonate Methylenechloride 25 11.i 0.82 1 - - 27 SD (231)


Tetrahydrofuran 25 39.9 0.70 1 - - 27 SD (231)
Polychloroprene
(Neoprene CG) Benzene 25 2.02 0.89 6---150 OS (1893 7.;
(Neoprene GN) Benzene 25 14.6 0.73 2--96 OS (189"')
(Neoprene W) Benzene 25 15.5 0.71 5--100 OS (189-)
Poly(p-chlorostyrene) Methyl ethyl ketone 25 29 0.59 3--140 LS (14o3
Toluene 25 13.2 0.64~ 1--244 LS (77)
30 13.0 0.64 3--140 LS (14o')
Poly (chlorotrifluoroethylene) 2,5-Dichlorobenzotrifluoride 130 6.15 0.74 7--51 OS (26s)
(KEL-F) r~
Poly (cis- 1,4 -cyclohexane Chloroform 20 27.8 0.78 2.1--4.6 OS (s2~)
disebacate)
Poly(trans-l,4-cyclohexane Chloroform 20 18.3 0.86 1.1 - - 3 . 7 OS (s2~)
disebacate)
Poly(decamethylene adipate) Chlorobenzene 25 11.7" 0.84' 1-- 3 MV (los)
Diethyl succinate 79 5.8* 0.86' 1-- 3 MV (1o8)
Poly (dichlorostyrene) Toluene 21 12.6 0.69 7 - - 67 LS (115)
Poly (2,5-dichlorostyrene) Ethyl acetate (15)[Ethanol 30.5 35.5 0.50 50--130 LS (93)
(1 wt)
Poly(n,n'-dimethyl Methanol 25 12.5 0.68 5--120 LS (25r) o
acrylamide) Water 4O 17.0 0.65 5--120 LS (25r) oq
Poly (dimethylsiloxane) Methyl ethyl ketene 2O 81 0.50 5--66 OS (1o5)
Phenetole 83 79 0.50 5--66 OS (lo5)
Toluene 25 20.0 0.66 0 . 3 - - 20 OS (32)
Poly(ethyl acrylate) Acetone 3O 4.19 0.66 14-- 71 OS (12o)
Poly(2-ethylbutyl Isopropanol 27.4 33.7 0.50 48--332 LS (s J)
methacrylate) Methyl ethyl ketone 25 2.21 0.77 48--332 LS (81) g
Polyethylene (low pressure) ~-Chloronaphthalene 125 43 0.67 5--100 LS (26)
129 27.1 0.71 LS (too")
Decalin 135 62 0.70 6 - - 34 LS (53)
135 46 0.73 3 - - 64 LS (12s) tO
r
r
tO
q~
Table 13.(Continued)

Polymer Solventt Temp.x ~ K; - l0t Range Method* Reference


M. 10-*

15olyethylene (low pressure) Tetralin 20 23.6 0.78 5--100 LS (90)


30 43.5 0.76 2 - - 30 OS (/35)
30 51 0.725 0 . 4 - - 10 OS (2S6)
p-Xylene O0 16.5 0.83 1 - - 20 LS (251)
Polyethylene (high pressure) Decalin 70 38.7 0.738 0 . 2 - - 3.5 OS (257)
p-Xylene 75 135 0.63 0 . 2 - - 7.6 OS (122)
81 105 0.63 1 - - 10 OS (250)
Poly(ethylene oxide) Aqueous K2SO 4 (0.45 M) 35 130 0.50 0.5--500 SV (28) a.
Aqueous MgSO4 (0.39 M) 45 100 0.50 0.5--500 SV (2s)
Benzene 20 48 0.68 0.01-- 2 EG (223)
Carbon tetrachloride 20 69 0.61 0.02-- 1 EG (223)
Cyclohexanone 20 35 0.69 0.2-- 1 EG (223)
Dimethyl formamide 25 24 0.73 1-- 3 OS (220)
Water 30 12.5 0.78 2--500 SV (29)
Poly(ethylenesulfonic acid) Aqueous NaCI (0.5 M) 20 21.5 0.65 0.3-- 3 SD (so)
Poly(ethylene terephthalate) Phenol (50)/Tetrachloro- 25 21 0.82 0.5-- 3 EG (70) o
t~
ethane (50 vol.)
Phenol (10)/2,4,6-Trichloro- 30 21 0.80 0.1-- 1 EG (116)
phenol (7 vol.)
Poly (ethyl methacrylate) Isopropanol 369 47.5 0.50 22--130 LS (64)
Methyl ethyl ketone 23 2.83 0.79 20--263 LS (61)
MEK (1)]Isopropanel (7 vol.) 23 47.3 0.50 20--263 LS (61)
Poly(hexadecamethylene Chloroform 20 74.7 0.70 2 - - 10 OS (32a)
sebacate)
Poly(hexamethylene acetylene Benzene 20 51 0.55 1.2-- 5 OS (33a)
dicarboxylate) Chloroform 20 91 0.61 1.2-- 5 OS (33a)
Poly(hexamethylene a,u'- Benzene 20 37.4 0.74 0.9-- 2.4 OS (32b)
dibutylsebaeate)
Poly(hexamethylene fumarate) Chloroform 20 27.1 0.80 2-- 4,3 OS (32~)
Table 13. (Continued)
Poly(hexamethylene maleate) Benzene 20 76.3 0.60 1.3-- 6.6 OS (32e)
Chloroform 20 36.2 0.73 1.3-- 6.6 OS (s2e)
Tetrahydrofuran 2O 43.7 0.66 1.3-- 6.6 OS (32e)
Poly(hexamethylene sebacate) Benzene 20 62.7 0.69 0 . 6 - - 1.8 OS (S2b)
Chloroform 2O 72.5 0.70 2 - - I0 OS (a2a) o
Poly(hexamethylene succinate Benzene 20 43.3 0.70 1.5-- 5 OS (32e)
Chloroform 2O 24.4 0.79 1.5-- 5 os (32e)
Tetrahydrofuran 20 44.3 0.69 1.5~ 5 os (32e)
Poly(n-hexyl methacrylate) Isopropanol 32.6 43.0 0.50 6 ~ 41 LS (55)
Methyl ethyl ketone 23 2.12 0.78 6~41 LS (55)
Polyisobutylene Anisole 105 91 0.50 IV (11o)
Benzene 24 107 0.50 18--188 IV (110)
40 43 0.60 O.1--130 OS, CR (lo9)
Carbon tetrachloride 30 29 0.68 O. 1--130 OS, CR (1o9) ~r
Cyclohexane 27.6 r
30 0.69 4 - - 71 OS (148)
Diisobutylene 20 36 0.64 1--130 OS (99)
Phenetole 91 0.50 IV (Jto)
Toluene 40 0.60 0.7--150 OS (109)
30 20 0.67 5--150 OS (to9)
60 13.5 0.71 11--150 OS (109) O
Poly(n-lauryl methacrylate) n-Amylalcohol 29.5 34.8 0.50 27--240 LS (165)
n-Butyl acetate 23 8.64 0.64 26---360 LS (ss) o
Isopropyl acetate 13 32.2 0.50 26--360 LS (sg)
Poly(metaphosphate) Aqueous NaBr (0.35 M) 25 6.5 0.69 1--125 LS (241')
Aqueous NaBr (0.415 M) 25 49.4 0.50 1--125 LS (2,tr) c~
Poly(methacrylic acid) Aqueous HCI (0.002 M) 30 6.6 0.50 10-- 90 IV (134)
Poly (methacrylonitrile) Acetone 20 95.5* 0.56* 35--100 OS (liSa)
Dimethyl formamide 20 220* 0.50* IV (207")
Poly (p-methoxystyrene) Methyl ethyl ketone 35 8.6 0.68 1--100 LS (too')
Toluene 30 18.0 0.62 1--100 LS ( 1~to')
Poly(methyl acrylate) Acetone 20 7.40* 0.76* 7 ~ 32 OS (236)
Toluene 35 50* 0.55* 5 - - 47 IV (2a5)
Poly(methyl methacrylate) Acetone 25 7.5 0.70 8--137 LS (,to)
25 7.5 0.70 2--740 SD (2s0) t tDo
25 5.3 0.73 15--780 LS C,tT,479 ,4
9 tO
Table 13. (Continued) m

Polymer Solventz Temp. x ~ K ' 9 10 5 v


Range Method* Reference
M- 104

Poly(methyl methacrylate) Benzene 20 5.5 0.76 15--780 SD (47, 232')


Chloroform 20 4.85 0.80 8--200 SD (185)
20 6.0 0.79 3--780 LS (47, 473
25 3.4 0.83 40--330 LS (59)
E t h y l acetate 20 6--110 SD (94)
Ethylene dichloride 25 17 0.68 3 - - 98 LS (59)
Methyl ethyl ketone 25 6.8 0.72 8--137 LS (40)
25 7.1 0.72 40--330 LS (59)
MEK (50)/Isopropanol 25 59.2 0,50 30--280 LS (56) >
(50 vol.) p~
Nitroethane 25 8.7 0.74 10--200 LS (51')
Toluene 25 7.1 0.73 4--330 LS (59)
Poly(~-methyl styrene) Benzene 30 24.9 0.647 14-- 91 OS (2Or)
Benzene (79.4)/Methanol 30 76.8 0.50 14-- 91 OS (201')
(20.6 vol.) o~
Toluene 25 7.81 0.73 3 - - 60 SE (/79) o
30 2.2 0.80 LS (14o3
30 10.8 0.71 3 - - 66 LS (234) >
Poly(2-methyl-5-vinyl Dimethyl form~mide 25 13.0 0.76 4 - - 40 OS (223"")
pyridine) Methanol 25 18.0 0.83 4 - - 40 OS (22a'")
Poly(octamethylene cis-hexa- Chloroform 20 22.9 0.79 3 . 3 - - 5.5 OS (see)
hydroterephthalate)
Poly(octamethylene trans- Chloroform 20 18.9 0.84 2.4-- 4.4 OS (32c)
hexahydroterephthalate)
Poly(n-octyl methacrylate) n-Butanol 16 8 26.8 0.50 33--1250 LS (6o)
Methyl ethyl ketone 23 4.47 0.69 33--1250 LS (6o)
Poly (o~-oxyundecanoate) Benzene 20 258 0.56 1.7-- 6 OS (16s")
Chloroform 25 36.3* 0.82* 0.5-- 1.2 EG (3o)
Poly(phenyl methacrylamide) Acetone 20 28.2 0.75 10-- 320 LS (253)
Table 13. (Continued)
Polypropylene (atactic) Benzene 25 27.0 0.71 6---31 0S (ts8) 2.
30 33.8 0.67 2 - - 34 OS (74)
Cyclohexane 25 16.0 0.80 6---31 OS (ts8) o

30 20.9 0.76 2 - - 34 OS (74)


Decalin 35 13.8 0.80 6 - - 31 OS (tss)
L35 15.8 0.77 2 - - 39 OS (7s) E
Isoamyl acetate 34 168.5 0.50 2 - - 34 OS (7~)
Toluene 30 21.8 0.725 2 - - 34 OS (7~)
Polypropylene (isotactic) ~-Chlorona )hthalene L39 21.5 0.67 10--170 LS (I~o'")
t45 4.9 0.80 5 - - 63 LS (2og)
Decalin 135 11.0 0.80 2 - - 62 LS (t38)
Tetralin 35 2.5 1.0 2--11 OS (68)
35 9.17 0.80 4 - - 54 OS (209)
Xylol 85 96 0.63 ? 0S (25)
Polysarcosine Water 20 56* 0.88* 0 . 7 - - 1.6 EG (95)
Polystyrene (atactic) Benzene 25 9.52 0.744 3--- 61 OS (1,i0)
25 11.3 0.73 7--180 OS (343
Chloroform 25 7.16 0.76 12--280 LS (206) =
25 11.2 0.73 7--150 OS (343
Cyclohexane 34 82 0.50 1 - - 70 OS 0
(I~9)
Dichloroethane 25 21.0 0.66 1--180 LS (207)
Ethylbenzene 25 17.6 0.68 7--150 OS (349
Ethyleyclohexane 70 75 0.50 36--127 IV (111) 0
Methylcyclohexane 70 76 0.50 (003
Methyl ethyl ketone 25 39 0.58 1--180 LS (207) c~
25 19.5 0.635 12--280 LS (206).:
25 30.5 0.60 7---150 OS (343
Toluene 25 13.4 0.71 7--150 OS (s4")
25 17 0.69 1--160 LS (2o7)
25 10.0 0.72 4--370 LS (208, tse)
30 9.2 0.72 4--150 LS (6s)
30 11.0 0.725 8 - - 85 OS (73)
Polystyrene (atactic, Toluene 20 11.2 0.72 3 - - 24 SD (182)
Szware-type) . 25 9.77 0.73 1--104 SD (178) r
25 - 34.5 0.62 1 - - 50 SD (19J) tD
Table 13. (Continued)

Range
Polymer Solventt Temp.t ~ K' 9i0s M 9 10 - a
Methodi Reference

Polystyrene (isotactic) Benzene 30 9.5 0.77 4-- 75 OS (2~)


30 10.6 0.735 3-- 37 OS (197, 73)
Toluene 30 11.0 0.725 3-- 37 OS (197, 73)
30 9.3 0.72 15-- 70 LS (252)
Poly (undecanoate) Chlorolorm 25 36.3 0.82 * 0.5-- 1.3 EG (zo)
Polyurethane (Polypropylene Benzene 25 41.3 0.64 1-- 8 LS (189)
oxide) Methanol 25 76.9 0.55 1-- 7 LS (189)
Toluene (5)/Iso-octane 39d 107.5 0.50 1-- 7 LS (1s9)
(7 vol.)
Poly(vinyl acetate) Acetonitr~e f 25 16.2 0.71 24--215 LS (38')
Acetone 20 15.8 0.69 19-- 72 LS (225)
30 8.6 0.73 8 - - 66 LS (02)
30 10.2 0.72 3--126 LS (177)
Benzene 30 22 0.65 34--102 LS (219")
Ethyl n-butyl ketone 29 92.9 0.50 ,5--83 LS (177")
Ethyl iso-amyl ketone 66 82.0 0.50 14-- 83 LS (177') oc~
Methanol 30 31.4 0.60 3--120 LS (177)
Methyl ethyl ketone 25 13.4 0.71 2--350 LS ( i zs, 233)
25 42 0.62 2--120 SD (91, 92)
30 10.7 0.71 3---120 LS (177) t~
Methyl isopropyl ketone 25 93 0.50 85---290 LS (233, 177)
(73.2)/n-Heptane (26.8 vol.)
Poly(vinyl alcohol) Water 30 66.6 0.64 3-- 12 OS (196)
Poly(vinyl benzoate) Xylene 32.! 62.0 0.50 10-- 24 OS (223b)
Poly(vinyl bromide) Cyclohexanone 20 32.8 0.55 2-- 10 LS (67)
Tetrahydrofuran 20 15.9 0.64 2-- 10 LS (67)
Tetrahydrofuran (83)/Metha- 20 38.8 0.50 2-- 10 LS (67')
nol (17)
Poly(vinyl n-butyrate) Benzene 30 11.15 0.735 3 - - 15 OS ( t 6o)
Poly(vinyl isobutyrate) Benzene 30 11.05 0.711 5 - - 20 OS (16o)
Table 13. (Continued)
e~
Poly(vinyl n-capmate) Benzene 30 15.47 0.689 3--126 OS (t6o) 2.
Poly(vinyl chloride) Cyclohexanone 20 11.6 0.85 3 - - 10 OS
25 24 0.77 3 - - 14 OS (76)
25 12.3" 0.83* 2 - - 14 OS (66)
Poly(vinyl chloride) Tetrahydrofuran 20 3.63 0.92 2 - - 17 OS (33) o
Poly(vinyl methyl ether) Benzene 30 76 0.60 t - - 45 LS (173) e~
Methyl ethyl ketone 30 92 0.58 I - - 45 LS (173)
Poly(4-vinyl pyridine) Ethanol 25 25.0 0.68 10--185 LS (as)
Ethanol (92)/Water (8 wt.) 25 12.0 0.73 7--224 LS
M E K (86)]Isopropanol (14 wt.) 25 38.0 0.57 7--224 LS (44)
Poly(vinyl pyrrolidone) Chloroform 25 19.4 0.64 2 - - 23 LS (166)
Methanol 30 23 0.65 2 - - 23 LS (166)
Water 25 67.6 0.55 0 , 7 - - 10 LS (t t3') e~
25 4.1 0.85 1-- 4 SD (184)
30 39.3 0.59 8---110 OS (52)
30 14 0.70 1--340 SD (226)
Poly(vinyl sulfate) Aqueous NaCI (0.5 M) 20 0.55* 1.06" 1-- 6 IV (2to)
Ribonucleic acid Aqueous phosphate (0.02 M) 20 62 0.53 2 - - 12 SV. (s) ~o
Styrene-Methyl methacrylate Methyl ethyl ketone 25 15.4 0,675 5--230 LS (241) O
copolymer (random, l / l )
o
* Newly calculated.
z Boldface letters and numbers indicate t h e t a solvents and t h e t a temperatures. O
= CR, cryoscopic method: DV, diffusion constant and intrinsic viscosity: EB, ebullioscopic method: EG, end-group titration: o~
IV, intrinsic viscosity-molecular weight relationship in other solvents: LS, light scattering: MV, melt viscosity-molecular weight
relationship : OS, osmotic pressure: PR, analysis of polymerization rate: SD, sedimentation and diffusion constants: SE, sedimentation P
equilibrium (ARCHIBALD'S method) : SV, sedimentation constant and intrinsic viscosity [see Eq. (72)].
o
8

03
p.
302 M. KURArA and W. H. STOCKMAVER:

Bibliography
:A. Monographs and Reviews
1. BALDWIN, R. L., a n d K. E. VAN HOLDE; Sedimentation of high polymers.
Fortschr. Hochpolym.-Forsch. 1, ~451--511 (1960).
2. CASASSA,E. F. : Polymer solutions. Ann. Rev, Phys, Chem. 11, 477--590 (1960).
3. CHANDRASEKHAR, S. : Stochastic problems in physics and astronomy. Revs.
Modern Phys. IS, 1 - - 1 1 0 (1943).
4. DoTY, P. : Configurations of biologically i m p o r t a n t macrom01eculcs in solution.
I n Biophysical Scienc8 pp. 107--117. edited b y J. L. 0NCLEY, New York:
J o h n Wiley & Sons, Inc. 1959.
4'. EmNER, J. : Native, denatured and renatured states of DNA. Doctoral Thesis,
D e p a r t m e n t of Chemistry. H a r v a r d University, Cambridge, Mass., April 1060.
5. FLORY, P. J. : Principles of polymer chemistry. Ithaca, N. Y., Cornell University
Press, 1953. (a) Chapter V I I : (b) Chapter X : (c) Chapter X I I : (d) Chapter X I I I :
(e) Chapter X I V : (f) p. 615, Table X X X V I I : (g) p. 618, Table X X K I X : (h)
p. 619, footnote: (i) p. 621, Table X L and Fig. 142: (j) p. 625, Table XLI.
6. FERRY, J. D. : Viscoelastic properties of polymers. Chapter II. New York:
J o h n Wiley & Sons, Inc. 1961.
6". F m s c s , H. L., and R. SIMHA: Viscosity of colloidal suspensions and macro-
molecular solutions. I n Rheology, Vol. I, pp. 5 2 5 ~ 6 1 3 . edited b y F. ElmC~,
New York: Academic Press 1956.
6". GEIDUSCHBK, ~-. P., and A. M. HOLTZER: Apphcation of l i g h t scattering to
biological systems. Advances in Biol. and Med. Phys. 6, 432 (1958).
7. GUGGENHI~IM, E. A.: Mixtures. London: Oxford University Press 1952.
8. HALL, B. D., and P. DoTY: Microsomal particles and protein synthesis. \Vashing-
ton Acad. Sci. p. 27 (1958).
9. HARLAND, W. G.: The degree of polymerization of cellulose and starch. I n
Recent advances in Chemistry of cellulose and starch, pp. 265--284, edited b y
HONEYMAN. New York: Interscience Pub; Inc. 1959.
10. HILDEBRAND, J. H., and R. L. SCOTT: The solubility of nonelectrolytes. Chap-
ters V I I and XX. 3rd ed. New York: Reinhold Pub. Corp. 1950.
11. HILL, T. L. : Statistical mechanics. New York: McGraw-Hill Book Co. 1956.
12. HUGGINS, M. L. : Physical chemistry of high polymers. New York: J o h n Wiley
& Sons, Inc. 1958.
13. •AWADE, Y.: Physicochemical s t u d y of ribonucleic acid. Ann. Rept. Inst.
Virus Res., Kyoto Univ. 2, 219--268 (1959).
13". MEYERHOFF, G.: Die Viscosimetrische Molekulargewlchtsbestimmung yon
Polymeren. Fortschr. Hochpolym.-Forsch. 3, 59 '(1961).
14. MIZUSHIMA, S.: Structure of molecules a n d internal rotation. New York:
Academic Press 1954.
15. MONSTER, A. : Statistische T h e r m o d y n a m i k hochmolekularer L6sungen. I n : Die
Physik der Hochpolymeren, edited b y H. A. STUART, Bd. II. Das Makromolekfil
in L6sungen. Berlin-G6ttingen-Heidelberg: Springer-Verlag 1953.
15'. PAULING, L.: The n a t u r e of t h e chemical bond. 3rd. ed. New York: Cornell
University Press, I t h a c a 1960.
16. PETERLIN, A. : Viskosit~t und Form. I n : Die Physik der Hochpolymeren, edited
b y H. A. STUART, Bd. II. Das Makromolekiil in L0sungen, Chapter 11. Berlin-
G6ttingen-Heidelberg: Springer-Verlag 1953
16". ~ Short and long range interaction in the isolated macromolecule. Ann. N. Y.
Acad. Sci. 89, 578 (1961).
Intrinsic Viscosities a n d U n p e r t u r b e d Dimensions of Long Chain Molecules 303

17. PRIGOGINE, I., 'A. BELLI~MANS a n d V. MATHOT: The molecular theory of solu-
tions. Chapters X I V and XV. Amsterdam: North-Holland Pub. Co. 1957.
18. Ric~, A.: Molecular configuration of synthetic and biological polymers. ~In:
Biophysical science, pp. 50--60, edited b y J. L. ONCLEY. New York: J o h n
Wiley & Sons, Inc. 1959.
19. ~RISEMAN, J., a n d J. G. KIRKWOOD: The statistical mechanical theory of irre-
versible processes in solutions of macromolecules. I n : Rheology, edited b y F. R.
]~iRlC~I, Vol. I, pp. 495--523, New York: Academic Press 1956.
19'. ROWLINSON, J. S. : Liquids and liquid mixtures. New York: Academic Press,
or London: B u t t e r w o r t h s Scientific Publications, pp. 111--190, 1959.
19". SAI)RON, C. L.: Deoxyribonucleic acids as macromolecules. I n : T h e nucleic
acids, Vol. III, pp. 1--37, edited b y E. CHARGAFF a n d J. Iq. DAVlDSON. New
York: Academic Press, Inc. 1960.
20. STOCKMAYER, W. H." Problems of t h e statistical Chermodynamics of dilute
polymer solutions. Makromol. Chem. 35, 54--74 (1960).
21. TOMPA, H.: Polymer solutions. London: B u t t e r w o r t h s Sci. Pub. 1956.
2 l'. VOLKENSTEIN,M. V. : Configurational statistics of polymeric chains (inRussian).
Moscow and Leningrad: Academy of Sciences, USSR, 1959. A n English trans-
lation b y S. TIMASHEFF is in preparation (Interscience Publishers),
22. ZIiM, B. H. : Concentrated maeromolecular solutions,: I n : Biophysical science,
pp. 123--129, edited b y J. L. ONCLEY. New York: J o h n Wiley & Sons, Inc.
1959.
B. R e f e r e n c e s
23. ALBRECHT, A. C.: J~ Chem. Phys. 27, 1003 (19.57).
2 3 ' . - - J. Chem. Phys. 27, 1014 (1957).
24. A~G, F.: J. Polymer. Sci. 25, 126 (1957).
25. P , and H. MARK: Monatsh. Chem. 88,427 (1957).
25'. ARCVS, C. L., and D. W. WXST: J. Chem. Soc. 1959j 2699.
26. A ~ i ~ s , J. T., L. T. Muvs, C. W. SmTH and E. T. PmSKI: J. Am. Chem. Soc.
79, 5089 (1957).
27. AUER, P. L., a n d C. S, GARDNER: J. Chem. Phys. 23, 1545 (L), 1546 (L) (1955).
28. BAILEY, F. E., jr., and R. W. CALLARD: J. Appl. Polymer Sci. 1, 56 (1959).
29. ~ , J. L. KUCERA and L. G. ]MHOF: J. Polymer Sci. 32, 517 (L) (1958).
30. BA~ER, W. 0., C. S. FULLER a n d J. H. HEms, jr.: J. Am. Chem. Soc. 63, 3316
(1941).
31. BAMFORD, C. H., A. D. JENXINS, R. JOHNSTON a n d E. F. T. WroTE: Trans.
F a r a d a y Soc. 55, 168 (1959).
32. BARRY, A. J.: J. Appl. Phys. 17, 1020 (1946).
32a. BATZER, H.: Makromol. Chem. 5, 5 (1950).
3 2 b . - - Makromol. Chem. 10. 13 (1953).
32c. - - , and G. FRITZ: Makromol. Chem. 14, 179 (1955).
3 2 d . - - , and H. LANG: Makromol. Chem. 15, 211 (1953).
32e. - - , and B. MOHR: Makromol. Chem. 8, 217 (1952).
32f. - - , and A. MOSCHLB: Makromol. Chem. 22, 195 (1957).
3 3 . . - - , and A. ~mCH: Makromol. Chem. 22, 131 (1957).
33a. - - , and G. WmSSE~B~RGXR: Makromol. Chem. 11, 83 (1953).
33b. BAUR, M. E.: Fh. D. Thesis, M.I.T. (1959).
34. BAWN, C. E. H., C. FREEMAN and A. KAMALIDDIN: Trans. F a r a d a y Soc. 46, 862
0950).
3 4 ' . - - - - - Trans. F a r a d a y Soc. 46, 1107 (1950).
35. - - , a n d M. A. WAJID: J. Polymer. Sci. 12, 109 (1954).
304 M. KURATA and W. H. STocKMAYER:

36. BEARMA~r R. J., and P. F. JoNEs: J. Chem. Phys. 33, 1432 (1960).
37. BENOIT, H. : J. Polymer Scl. 3, 376 (1948).
37'. BENOIT, H.: J. Polymer Sci. 11,507 (1953).
38. BERKOWITZ, J. B., M. YAMIN and R. M. F u o s s : J. Polymer Sci. 28, 69 (1958).
38'. BEVAK: Ph. D. Thesis, M.I.T. (1955).
39. BILLMEYER, F. W., jr., and C. B. DE THAW: J. Am. Chem. Soc. 77, 4763 (1955).
39'. BXRSTEIN, T. M., a n d O. B. PTITSYN: J. Tech. Phys. U.S.S.R. 29, 1048 (1959).
40. BmCHOFF, J., a n d V. DESREUX: Bull. soe. chim. Belges 61, 10 (1952).
41. BisscHoPs, J.: J. Polymer Sci. 17, 8I (1955).
42. - - J. Polymer sci. 17, 89 (1955).
43. BOEDTKER, H.: J. Mol. Biol. 2, 171 (1960),
43'. Bovv.Y, F. A., and G. V. D. TIERS : J. Polymer Sci. 44, 173 (1960).
43". - - J. Polymer Sci. 46, 59 (1960).
44. B o y , s , A. G., and U. P. STRAUSS: J. Polymer Sci. 22, 463 (1956).
45. BREITENBACH, J. W., E. L. FORSTER and A. J. RENNER: Kolloid-Z. 127, l
(1952).
45'. BRINKMAN, H. C. : Physica 13, 447 (1947).
46. BURCHARD, W., a n d E. I-IusEMANN: Makromol. Chem. 44---46, 356 (1961).
47. CANTOW, H.-J., a n d G. V. SCHULZ: Z. physik. Chem. (N. F.) 1, 365 (1954).
4 7 " . - - - - Z. physik. Chem. (N. F.) 2, 117 (1954).
9 48. CARTER, W. C., R. L, SCOTT a n d M. MA~AT: J. Am. Chem. Soc. 68, 1480 (1946).
49. CASASSA, E~ F.: J. Chem. Phys. 31,800 (1959).
50. - - , and H. MARKOVITZ" J. Chem. Phys. 29, 493 (1959).
51. - - , and T. A. OROFINO" J. Polymer Sci. 35, 553 (1959).
5 1 ' . - - , and W. H. STOCKMAYER: Polymer 3, 53 (1962).
52. CI~RNEY, L. C., T. E. HELMINIAK and J. F. M~IER: J . Polymer Sci. 44, 539
(1960).
53. CHIANG, R.: J. Polymer Scl. 36, 91 (1959).
54. CHIEN, J.-Y., \u CmNG and Y.-S. CHEN~: Kolloid. Zhur. 19, 515 (1957).
55. CHINAI, S. N.: J. Polymer Sci. 25, 413 (1957).
56. - - , and C. W. BONDURANT, jr.: J. Polymer Sci. 22, 555 (1956).
57. - - , and R. A. Guzzi: J. Polymer Scl. 21, 417 (1956).
58. f - - J. Polymer Sci. 41, 475 (1959).
59. f , J. D. MATLACK, A. L. R~SNICK a n d R. J. SAMUELS: J. Polymer Sci. 17, 391
(1955).
60. - - , A. L. RESNXCK a n d H. T. LEE: J. Polymer Sci. 33, 471 (1958).
61. - - , a n d R. J. SA~UELS: J. Polymer Sci. 19, 463 (1956).
62. ~ , P. C. SCH~RER a n d D. W. L ~ w : J. Polymer Sci. 17, I17 (1955).
63. - - - - , C. W. BONDURANT a n d D. W. LEVI: J. Polymer Scl. 22, 527 (1956).
64. ~ , a n d R. J. VALLES: J. Polymer Sci. 39, 363 (1959).
65. CIAMPA, G. : Chlm. e ind. (Milan) 38, 298 (1956).
66. - - , and tI. SCHWINDr: Makromol. Chem. 21, 169 (1956).
66'. CIFERRI, A. : Trans. F a r a d a y Soc. 57, 846, 853 (1961).
66". - - , C. A. J. H o ~ v ~ and P. J. FLORY: J. Am. Chem. Soc. 83, 1015 (1961).
67. ~ , M. KRYSZEWSKI a n d G. WmL: J. Polymer Sci. 27, 167 (1958).
67'. ~ , a n d M. LAUEETTI; Ann. Chim. (Rome) 48, 198 (1958).
68. CLELAND, R. L.: J. Polymer Sci. 27, 349 (1958).
68'. ~ , and W. H. STOCK~AY~R: J. Polymer Sci. 17, 473 (1955).
69. COLmNSON, E., F. S. DAINTON and G. S. McNAUGHTON: Trans. F a r a d a y Soc.
53, 489 (1957).
70. CoNIx, A.: Makromol. Chem. 26, 226 (1958).
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 305

70'. COOPER,W. G., G. VAUOHAN, D. E. EAVES and R. XV. MADDEN: J. Polymer


Sci. 50, 159 (1961).
71. CowiE, J. M. G.: Makromol. Chem. 42, 230 (1961).
7 ! ' . - - J, Polymer Sci. 49, 455 (1961).
71". - - , D. J. WORSFOLD and S. BYWATER: Trans. Faraday Soc. 57, 705 (1961).
72. CRAGG, L. H., T. E. DUMITRU and J. E. SIMKINS: J. Am. Chem, Sor 74, 1977
(1952).
73. DANUsso, F., and G. MORAGLIO; J. Polymer Sci. 24, 161 (1957).
74. - - - - Rend. accad, naz. Lincei 25, 509 (1958).
75. - - - - Makromol. Chem. 28, 250 (1958),
76. - - - - and S. GAZZERA; Chim. e ind. (Milan) 36, 883 (1954).
77. DAvis, J. E.: P h . D . Thesis, M.I.T. (May 1960).
77'. DAVISON, P. F., D. FREIFELDER, R. HEDE and C. LEVINTHAL: Proe. Natl. Acad.
Sci. U.S.A. 47, 1123 (1961).
78. DEBYE, P.: J. Chem. Phys. 14, 636 (1946).
79. - - , and A. M. BUECHE: J. Chem. Phys. 16, 573 (1948).
80. DIALER, K., and R. KERBER: Makromol. Chem. 17, 56 (1955).
81. DIDOT, F. E., S. N. CHINAI and D. W. LEv*: J. Polymer Sci. 43, 557
(1960).
82. Dx~u, H. A.: J. Polymer Sci. 12, 417 (1954).
82'. DoBRY, A., and F. BOYER-KAwENoKI: J. Polymer Sci. 2, 90 (1947).
83. DOMBROW, B. A., and C. O. BECKMANN: J. Phys. Colloid Chem. 51, 107
(1947).
84. DO,NAN, F. G., and R. C. RosE: Can. J. Research B 28, 105 (1950).
84'.DorY, P.: J. Polymer Sci. 55, 1 (1961).
85. - - , J. H. BRADBURV and A. M. HOLTZER: J. Am. Chem. Soe. 78, 947 (1956).
86. - - B. BU~cE-McGILL and S. A. RICE: Proc. Natl. Acad. Sci. U.S.A. 44, 432
(1958).
67. - - A.M. HoLTzER, J. H. BRADBURY and E. R. BLOUT : J. Am. Chem. Soc. 76,
4493 (1954).
88, - - J. MARMUR, J. EIGN~R and C. SCHILDKRAUT; Proc. Natl. Acad. Sci. U.S.A.
46, 461 (1960).
89. --, N. S. SCHNEIDER and A. M. HOL~Z~R: J. Am. Chem. Soc. 75, 754 (L)
(1953).
90. DUCH, E., and L. KOCHLER: Z. Elektrochem. 60, 218 (1956).
96'. DUMITRU, E. T., and L. H. CRAGG: Flory'$ book (5), p. 615, Table X X X V I I .
91. ELIAS, H.-G., and F. PATAT: Makromol. Chem. 25, 13 (1957).
92. ~ ~ J. Polymer Sci. 29, 141 (1958).
93. ESKIN, V. E., and K. Z. GUMARGALIEVA: Vysokomolekulyarnye Soedineniya
(High Molecular Weight Compounds) 2, 265 (1960).
94. ERmSSON, A. F. V.: Acta Chem. Scan& 10, 378 (1956).
94'. EVERETT, W. W., and J. F. FOSTER: J. Am. Chem. Soc. 81, 3464 (1959).
95. FESSLER, J. H., and A. G. OGSTON: Trans. Faraday Soc. 47, 667 (1951).
96. FISHER, M. E., and B. J. HILEY: J. Chem. Phys. 34, 1253 (1961).
97. - - , and M. F. SYKES: Phys. Rev. 114, 45 (1959).
98. FIXMAN, M.: J. Chem. Phys. 23, 1656 (1955).
98". FLORY, P. J.: J. Chem. Phys. 10, 51 (1942).
99. - - J. Am. Chem. Soc. 65, 372 (1943).
100. - - J. Chem. Phys. 17, 303 (1949).
101. - - J. Chem. Phys. 17, 1347 (L) (1949).
102. - - , A. CIF~RRi and R. CHIANG: J. Am. Chem. Soc. 83, 1023 (1961).
103. ~ , and T. G. F o x : J. Am. Chem. Soc. 73, 1904 (1951).
306 M. K U R A T A and W. H. STOCKMAYER:

103a. FLORY, P. J., C. A. J. HOEVE and A. CIFERRI: J. Polymer Sci. 34, 337 (1959).
103'. - - , and Vr R. KRmBAUM : J. Chem. Phys. 18, 1086 (1950).
104. - - , and F. S. LEUTNER: J. Polymer Sci. 3, 880 (1948).
105. - - , L. MANDELKERN, J. B. KINSlNGER a n d V~: B. SHULTZ: J. Am. Chem. Soc.
74, 3364 (1952).
106. - - , a n d J. E. OSTERHELD: .J, Phys. Chem. 58, 653 (1954).
107. - - , O. K. SPURR, jr., and D. K. CARPENT~.R: J. Polymer Sci. 27, 231 (1958).
108. - - , and P. B. STICKN•Y: J. Am; Chem. Soc. 62, 3032 (1940).
109. Fox, T. G., and P. J. FT.ORY: J. Phys. Colloid Chem. 53, 197 (1949).
110. - - --- J. Am. Chem. Soc. 73, 1909 0 9 5 t ) .
111. - - - - J. Am. Chem. Soc. 73, 1915 (1951).
1 1 1 ' . - - - and A. M. BI~ECHE: J. Am. Chem. Soc. 73, 285 (1951).
112. FRANCZS, P. S., R. COOKE, jr. and J. H. ELLIOTT: J. Polymer Sci. 31, 453
(1958).
113. FRANK, H. P., and J. W. BREITENBACH: J. Polymer Sci. 6, 609 (1951).
113'. FRANK, H. P., and G. B. LEVY: J. Polymer Sei. 10, 371 (1953).
114. FRESCO, J. R., and P. DOTY: J. Am. Chem. Soc. 79, 3928 (1957).
115. FRISMAN, E. F., a n d L. F. SHALAEVA: Doklady Akad. Nauk. S,S.S.R. 101, 907
(1955).
l15a. FUHRMAN, N., and R. B. MESROEIAN: J. Am. Chem. Soc. 76, 3281 (1954).
115'. FUJITA, tt., A. M. LINKLATER a n d J. W. WILLIAMS: J. Am. Chem. Soc. 82, 379
(1960).
116. GAYLORD, N. G., and S. ROSENBAUM: J. Polymer Sci. 40, 545 (L) (1959).
117. GRIMLEY, T. B.: Proc. Roy. Soc. (London) A 212, 339 (1,952).
118. - - Trans. F a r a d a y Soc. 55, 681 (1959).
119. GOUmLOCK, E. V., jr., P. J. FLORY a n d H. A. SCHERA6A: J. Polymer Sci. 15,
383 0955).
120. HACHIHAMA, Y., and H. SUMITOMO: Tcchnol. Repts. Osaka Univ. 3, 385
(1953).
121. HARLAND, W. G.: Nature 170, 667 (1952).
122..HARRIS, I . : J. Polymer Sci. 8, 353 (1952).
122a. HEARST, J. E., and W. It, STOCKMAYER: Private communication (1962).
122b. HEINE, S., O. KRhTKY, G..PoRoD a n d P. J. SCHmTZ: .Makromol. Chem. 47,
682 (1961).
122'. HERMANS, J., jr., and J. J. HERMANS: J. Phys. Chem. 63, 170 (1959).
123.1" /CI~RRY, P. M. : J. Polymer Sci. 36, 3 (1959).
123'. HIJMA~S, J., a n d TH. HOLL~MAN: J. Chem. Phys. 36, 47 (1962).
124. HOEVE, C. A. J.: J. Chem. Phys. 32~ 888 (1960).
125. ~HOLMES, F: l~., and D. I. SMITH: Trans. F a r a d a y Soc. 53, 67 (1957).
126. HOLTZER, A. M., H..BENOIT a n d P. DOTY: J. Phys. Chem. 58, 624 (1954).
127. HOWARD, G. J.: J.-Polj~mer Sci. 37, 310 (L) (1959).
128. HOWARD, R. O . : P h . D. Thesis, M.I.T. (Nov. 1952).
128'. HUGGI~S, NL L. : J. Phys. Chem. 43, 439 (1939).
128". ~ Ann. New York Acad. Sci. 43, 1 (1942).
129. HUNT, M. L., S. NEWMAN, H. A. SCHERAGA and P. J. FLORY: J. Phys. Chem.
60, 1278 (1956).
130. HU~UE, M. M., D. A. I. GORING and S. G. MASON: Can. J. Chem. 36, 952
0958).
131. IMAMURA, Y.: Nippon Kagaku Zassi (J. Chem. Soc. Japan, Pure Chem. Sect.)
76, 217 (1955).
132. IMMERGUT,]E. H., B. G. I~ANEY.Rnd H. MARK: Ind. Eng. Chem. 45, 2383 (1953).
132'. ISIHA~A, A., and R'fiKoYAMA: J. Chem. Phys. 25, 712 (1956).
Intrinsic Viscosities and U h p e r t u r b e d Dimensions of Long Chain Molecules 307

132". Ivll% K. J,, a n d H . A. ENDE: J. Polymer Sci. 54, S 17 (1961).


132"'. - - - - and,G. MEY~RHOFF: Polymer 3, 129 (1962).
133. JULLANDER, I.: Arkiv Kemi, Mineral Geol. A 21, No. 8 (1945).
134. KA•CHALSKY, A., and H. EISEI~BERG: J. Polymer Sci. 6, 145 (1951).
135. KAUFMA•N, H. S., and E. K. WALSlt: J. Polymer Sci. 26, 124 (L) (1957).
135'. KAWADE, Y., a n d I. WATANABE: Biochim. et Biophys. Acta 19, 513 (1956).
136. KAWAHARA, K., M. UEDA, Y. OKA and T. KUROIWA: Nippon Kagaku Zassi
(J. Chem. Soc. Japan, Pure Chem. Sect.) 79, 843 (1958}.
137. KERN, W., a n d D. BRAW~: Makromol. Chem. 27, 23 (1958).
137'. KHEI~, A. : Thesis, Leiden (1958).
138. KINSlNGER, J. B., and R. E. HUGHES: J. Phys. Chem. 63, 2002 (1959).
139. KIRKWOOD, J. G., and J. RISEMAN: J. Chem. Phys. 16, 565 (1948).
139'. IZlRSTE, R., and O. KRATKY: Z. physik. Chem. (Frankfurt) 31, 363 (1962).
140. KOBAYASHI, H.: J. Polymer Sci. 39, 369 (1959).
140a. KocH, T. A., and P. E. LINDRIG: J. Appl. Polymer Sci. 1, 164 (1959).
140". KOTXRA, A., T. SAITO, H. MATSUD& and R. KAMAT2t: Rept. Progr. Polymer
Phys. J a p a n 3, 51 (1960).
140". - - - - , K. TAKAMISAWA and Y. MIYAZAWA: Rept. Progr. Polymer Phys.
J a p a n 3, 58 (1960).
140'". ~ , K. TAKAMISAWA,T. KAMATA and H. KAWAGUCHI: Rept. Progr. Polymer
Phys. J a p a n 4, 131 (1961).
141. KOTLIAR, A. M.: J. App1. Polymer Sci. 2, 134 (1959).
141'. KRAMERS, H. A.: J. Chem. Phys. 14, 415 (1946).
142. KRATKY, O., a n d G. POROD: Rec. tray. ehim. 68, 1106 (1949).
1 4 2 ' . - - Angew. Chem. 72, 467 (1960).
142". - - , a n d H. SAND: Kolloid-Z. 172, 18 (1960).
143. KRIGBAUM, W. R.: J. Polymer Sci. 18, 315 (L) (1955).
144. - - J. Polymer Sci. 19, 159 (1956).
145. - - J. Polymer Sci. 28, 213 (1958).
146. - - , and D. K. CARPENTER: J. Phys. Chem. 59, 1166 (1955).
146".-- - - M. K A ~ K O and A. RoIG: J. Chem. Phys. 33, 921 (1960).
147. ~ ~ .and S. N~WMAN: J. Phys. Chem. 62, 1586 (1958).
147". ~ , and J. E. KuEz: J. Phys. Chem. 65, 1984 (1961}.
148. - - , a n d P. J. FLORY: J. Am. Chem. Soc. 75, 1775 (1953).
149. - - - - J. Polymer Sci. 11, 37 (1953).
150. - - , and A. M. KOTLIAR: J. Polymer Sci. 32, 323 (1958).
151. - - , and L. H. SPERLING: J. Phys. Chem. 64, 99 (1960).
152. - - , a n d Q. A. TREMENTOZZI: J. Polymer Sci. 28, 295 (1958).
153. KUHN, W.: Kolloid-Z. 68, 2 (1934).
153'. m , and H. KUHN: Helv. Chim. Acta 26, 1324 (1943).
153". ~ - - Helv. Chim. Acta 30, 78 (1947).
154. KURATA, M.: Ann. New York Acad. Sci. 89, 635 (1961}.
155. - - , XV. H. STOCKMAYER and A. RolG: J. Chem. Phys. 33, 151 (1960).
156. - - , M. TAMURA and T. WATARI: J. Chem. Phys. 23, 991 (1955).
157. - - , and H. YAMAKAWA: J. Chem. Phys. 29, 311 (1958).
158. ~ ~ a n d E. TERAMOTO: J. Chem. Phys. 28, 785 (1958).
159. - - - - and H. UTIYAMA: MakromoI. Chem. 34, 139 (1959).
160. KURIAN, C. J., and M. S. MUT~IANA: Makromol. Chem. 29, I (1959).
161. KURLAND, C. G.: J. Mol. Biology 2, 83 (1960).
162. LANDEL, R. F., J. W. BERGE and J . D . FERRY: J. Colloid Sci. 12, 400 (1957).
163. - - , and J. D. FERRY: J. Phys. Chem. 59, 658 (1955).
164. ~ ~ J. Phys. Chem. 60, 294 (1956).
308 M. KURATA and W. H. STOCKMAYER:

164'.LANsIN~, W. D., and E.:O. KRAEMER: J. Am. Chem. Soc. 57, 1369 (1935).
165. L~E, H. T., and D. W. LEVI: J. Polymer Sci. 47, 449 (1960).
165'. LETT, J. T., and K. A. STACEY: Makromol. Chem. 38, 204 (1960).
166. LEVY, G. B., and H. P. FRANK: J. Polymer Sci. 17, 247 (1955).
167. LIFSON, S.: J. Chem. Phys. 30, 964 (1959).
168. - - , and I. OPFENHEIM: J. Chem. Phys. 33, 109 (1960).
1 6 8 ' . - - , and A ROlG: J. Chem. Phys. 34, 1963 (1961).
168". LOMBARD, F.: Makromol. Chem. 8, 187 (1952).
169. MANDELKERrr L., and P. J. FLORY: J. Am. Chem. Soc. 73, 3206 (1951).
170. - - - - J. Am. Chem. Soc. 74, 2517 (1952).
171. - - ~ J. Chem. Phys. 20, 212 (1952).
172. MANL~Y, R. ST. J.: Arkiv Kemi 9, 519 (1956).
173. MANSON, J. A., and G. J. ARQU~.TT~: Makromol. Chem. 37, 187 (1960).
174. MARCHAL, J., and H. BENOXT: J. Polymer Sci. 23, 223 (1957).
175. - - , a n d C. LAPP: J. Polymer Sci. 27, 571 (L) (1958}.
175a. MARKOVITZ, H.: J. Chem. Phys. 20, 868 (1952).
175b. MARMUR, J., and P. D o r y : Nature 183, 1427 (1959).
175". MARRINAN, H. J., and J. J. HERMANS: J. Phys. Chem. 65, 385 (1961).
176. MARZOLPH, H., and G. V. SCHULZ: Makromol. Chem. 13, 120 (1954).
177. MATSUMOTO, M., and Y. O~YANA~I: J. Polymer Sci. 46, 441 (1960).
9 177'. ~ - - J. Polymer Sci. 50, S 1 (1961).
177". MAYNARD, J. T., and W. E. MocH•r.: J. Polymer Sci. 13, 251 (1954).
178. McCoRMmK, H. W.: J. Polymer Sci. 36, 341 (1959).
179. - - J. Polymer Sci. 41, 327 (1959).
179".M~YER, K. H., and A. VAN DER WYK: Helv. Chim. Acta 18, 1067 (1938).
180. MEY~RHOFF, G.: Z. physik. Chem. (N. F.) 4, 335 (1955).
181. - - J. Polymer Sci. 29, 399 (1958).
182. ~ z. physik. Chem. 23, 100 (1960).
183. - - , and G. V. SCHULZ: Makromol. Chem. 7, 294 (1952).
183'. MILLER,9A. R. : Proc. Cambridge Phil. Soc. 39, 54 (1943).
184. MILLER, L. E., a n d F. A. HAMM: J. Phys. Chem. 57, llO (1953).
185. MITCaP-LL, J. C., A. E. WOODWARD and P. D o r y : J. Am. Chem. Soc. 79, 3955
(1957).
185'. MXX'AKE, A., and R. CHUjo: J. Polymer Sci. 46, 163 (1960).
185". - - J. Polymer Sci. 46, 169 (1960).
186. MIZUSHIMA, S., and H. OKAZAKI: J. Am. Chem. Soc. 71, 3411 (1949).
187. ~ , Y. MORINO and T. SIMA~OUTI: J. Phys. Chem. 56, 324 (1952).
188. ~ - - , I. WATANABE a n d T. SIMANOUTI: J. Chem. Phys. 17, 663 (1949).
189. MOACANIN, J.: J. Appl. Polymer Sci. I, 272 (1959).
189a. - - Private communication.
189'. MOCHEL, W. E., and J. 13. NICHOLS: J. Am. Chem. Soc. 71, 3435 (1949).
189 "~. - - - - Ind. Eng. Chem. 43, 154 (1951).
189'". - - - - and C. J. MrGHTON: J. Am. Chem. Soc. 70, 2185 (1948).
190. MOORE, W. R., and A. M. BROWN: J. Colloid Sci. 14, 1,343 (1959).
191. ~ , a n d G. D. EDGE: J. Polymer Sci. 47, 469 (1960).
192. ~ , a n d J. A. ErsT~rN: J. Appl. Chem. (London) 6, 168 (1956).
192". MORNEAtr, G. A., P. I. ROTH and A. R. SHUr.Tz: J. Polymer Sci. 55, 609 (1961).
193. MUKnERJEA, R. N., and P. I~MMP: J. chim. phys. 56, 94 (1959).
194. MONSTER, A.: J. Polymer Sci. 8, 633 (1952).
1 9 4 ' . - J: ehim. phys. 49, 128 (1952).
Intrinsic Viscosities and Unperturbed Dimensions of Long Chain Molecules 309

195. NA~A~, K.: J. Chem. Phys. 31, 1169 (1959).


1 9 5 ' . - J. Chem. Phys. 34, 887 (1961).
196. NAKAJIMA, A., and K. FURUTATE: Kobunshi Kagaku (Chem. High Polymers
[Tokyo]) 6, 460 (1949).
197. NATTA, G., F. DANUSSO and G. MORAGLIO: Makromol. Chem. 20, 37 (1956).
198. NEWMAN, S., and P. J. FLOR~Z: J. Polymer Sci. 10, 121 (L) (1953).
199. - - , W. R. KRmBAUM, C. LAUGX~Rand P. J. FLORY: J. Polymer Sci. 14, 451
(1954).
200. NISHIHARA. T., and P. DoTY: Proc. Natl. Acad. Sci. U.S.A. 44, 411 (1958).
201. NOTLEY, N. T., and P. J. W. D~BYE: J. Polymer Sci. 17, 99 (1955),
201'. OKAMURA, S., T. HmASHIMURA and Y. IMANmHI: Chem. High Polymers 16,
244 (1959).
202. ONYo~, R. F.: J. Polymer Sci. 22, 13 (1956).
203. - - J. Polymer Sci. 37, 315 (1959).
204. OROFINo, T. A., and P. J. FLORY: J. Chem. Phys. 26, 1067 (1957).
205. OSAWA, F., N. IMAI and I. KAGAWA: J. Polymer Sci. 13, 93 (1954).
206. OTrI, J., and V. DESREUX: Bull. soc. chim. Belges 63, 285 (1954).
207. OUTER, P., C. I. CARR and 13. H. ZIMM: J. Chem. Phys. 18, 830 (1950).
207'. OV~RBERGRR, C. G., E. M. P~ARC~ and N. MAYES: J. Polymer Sci. 34, 109
(1959).
208. OYAMA, T., K. KAWAHARAand M. URDA: Nippon Kagaku Zassl (J. Chem.
Soc. Japan, Pure Chem. Sect.) 79, 727 (1958).
209. PARm~I, P., F. SRBASTIAr~Oand G. M~SSINA: Makromol. Chem. 38, 27 (1960).
209'. PARROD, J., and J. ELL~-S: J. Polymer Sci. 29, 411 (1958).
210. PATAr, F., and K: VO6LRR: Helv. Chim. Acta 35, 128 (1952).
211. I>ETERLI~r,A.: Makromol. Chem. 34, 89 (1959).
2 1 1 " . ~ J. Polymer Sci. 8, 173 (1952).
212. PHILIPP, H. J., and C. F. BJORK: J. Polymer Sci. 6, 549 (1951).
213. PITZ~R, K. S.: J. Chem. Phys. 8, 711 (1940).
214. PLAZEK, D. J., and J. D. F~.RRY: J. Phys. Chem. 60, 289 (1956).
214'. POI~L, H. A., R. BACSKAIand W. P. PURCELL: J. Phys. Chem. 64, 1701 (1960).
214". POLLOCK, D. J.i L. J. ELYASH and T. W. DRWITT: J. Polymer Sci. 15, 335
(1955).
214'". POLSON, A.: Kolloid-Z. 83, 172 (1938).
215. POURADIER, J., and A.-M. V~NRT: J. chim. phys. 47, 391 (1950).
216. PTITSYI% O. B., and J. A. SHARONOV; Zhur. Tekh. Fiz. (J. Tech. Phys.
U.S.S.R.) 27, 2744, 2762 (1957).
217. - - , and u Y. EIS~RR: Zhur. Fiz. Khim. (J. Phys. Chem. U.S.S.R.) 32, 2464
(1958).
218. - - ~ Zhur. Tekh. Fiz. (J. Tech. Phys. U.S.S.R.) 29, 1117 (1959).
2 1 8 ' . - Vysokomolekulyarnye Soedineniya (High Molecular Weight Compounds)
1, 715 (1959); an English translation appears in: Polymer Science USSR 1,
259 (1961).
219. - - , and Y. Y. EISlqRR: Vysokomolekulyarnye Soedineniya (High Molecular
Weight Compounds) 1, 1200 (1959).
219a. RAmH, W.: S. B. Thesis, M.I.T. (1948}.
219'. 1~o, M. R., and V. KAL~GnM: J. Polymer Sci. 49, 514 (1961).
219"'. REH~RR, J.: Ind. Eng. Chem. 36, 118 (1944).
219'". RIB~YROLI.~S, P. L., A. GuYoT and H. B~soI~: J. chim. Phys. 56, 377 (1959).
220. RITSCH~R, T. A., and H.-G. ELIAS: Makromol. Chem. 30, 48 (1959).
310 M. K U R A T A and Vq. H. STOCKMAYER:

221. RossI, C., U. BIANCHI and E. :BIA~CHI: Makromol. Chem. 41, 31 (1960).
222. ROUSE, P. E., jr. : J. Chem. Phys. 21, 1272 (1953).
222'. RUBENSTEIN, I., C. A. THOMAS, jr. and A. D. HERSHEY: Prec. Natl. Acad. Sci.
U.S.A. 47, 1113 (1961).
223. SADdeN, C., and P. REMPP: J. Polymer Sci. 29, 127 (1958).
223a. SALOVEY, R.: J. Polymer Sci. 50, 57 (1961).
223b. SAKURADA, I., Y. SAKAGUCHI a n d S. KOKURYO: Chem. High Polymers 17,
227 (1960).
223". SAMS0NOVA, T. I., and S. YA. FR~N~EL: Kolloid Zhur. 20, 67 (1958).
223". SASISEKHARA~, V. : J. Polymer Sci. 47, 373 (1960).
2 2 3 " . SATO, H., and T. YAMAMOTO: J. Chem. Soc. J a p a n (Pure Chem. Sec., Nipport
Kagaku Zasshi) 80, 1393 (1959).
224. SCHAEYGEN, J. R., and P. J. FLORV: J. Am. Chem. Soc. 70, 2709 (1948).
224a. SCnERAGA, H. A., and L. MANDEL~:~R~: J. Am. Chem. Soc. 75, 179 (1953).
224'. SCHERER, P. C., M. C. HAWKINS a n d D. W. L E w : J. Polymer Sci. 37, 369
(1959).
225. - - , A. TAN~NEAV~ and D. ~V. LEw: J. Polymer Sci. 43, 531 (1960).
226. SCHOLTAN, W.: Makromol. Chem. 71 209 (1951).
227. 2-- Makromol. Chem. 14, 169 (1954).
228. SCHULZ, G. V.: Z. physik. Chem. B 43, 25 (1939).
229. ~ Makromol. Chem. 35 A, 99 (1960).
229'. SeHULZ, G. V. : Private communication (1961).
230. - - , H.-J. CANTOW and G. MEYERHOFF: J. Polymer Sci. 10, 79 (1953).
231. - - , and A. HORBACH: Makromol. Chem. 29, 93 (1959).
232. ~ , and E. HUSEMA~N: Z. physik. Chem. B 52, 1 (1942).
232".--, and G. M ~ . Y ~ O F F : Z. Elektrochem. 56, 904 0952).
233. SHULTZ, A. R.: J. Am. Chem. Soc. 76, 3423 (1954).
234. SIRIANNL A. F., D. J. WORSFOLD and S. BYWAT~R: Trans. F a r a d a y Soc. 55,
2124 (1959).
234". SMALL, P. A.: J. Appl. Chem. 3, 71 (1953).
234". SPAcH, G.: Compt. rend. 249, 543(1959).
235. SRINIVASAN, N. T., a n d M. SA~TAPPA: Makromol. Chem. 27, 61 (1958).
235'. STARKW~ATH~R, H. W., jr., and R. H. BOYD: J. Phys. Chem. 64, 410 (1960).
236. STAUDINGER, H., and tI. WARTH: J. prakt. Chem. 155, 261 (1940).
237. STAVERMA~, A. J.: Rec. tray. chim. 69, 163 (1950).
237a. STEINBERG, I. Z., W. F. HARRINGTON, A. B~RG~R, M. SELA a n d E. KA~-
CHALSKt: J. Am. Chem. Soc. 82, 5263 (1960).
237'. STEINER. R. F., and R. F. B ~ R s , jr.: J. Polymer Sci. 30, 17 (1958).
238. S~OCKMAYER, W. H.: J. Polymer Sci. 15, 595 (L) (1955).
239. ~ , and A. C. ALBRECHT: J. Polymer Sci. 32, 215(1958).
239'. ~ , a n d E. F. CASASSA~ J. Chem. Phys. 20, 1560 (1952).
240. ~ , a n d M. FIXMAN: Ann. New York Acad. Sci. 57, 334 (1953).
241. - - , L. D. MooR~, jr., M. F~XMA~ a n d B. N. E~STEIN: J. Polymer Sci. 16, 517
(1955).
241". S~AUSS, U. P., and P. L. W ~ M A N : J. Am. Chem. Soc. 80, 2366 (1958).
241". $uzuKt, Y.: J. Chem. Phys. 34, 79 (1961).
241'". SZWARC, M.: Nature 178, 1168 (1956).
242. TAKED~, M., Y. IM~I~URA, S. OKAMURA ~nd T. H~G~.SH~IVJR,~: J. Chem. Phys.
33, 631 (1960).
243. TAK~AKA, H. : J. Polymer Sci. 24, 107 (1957).
244. T ~ B L Y ~ , J. W., D. R. MoR~Y and R. H. WA~N~R: Ind. Eng. Chem. 37, 573
(1945).,
Intrinsic Viscosities and U n p e r t u r b e d Dimensions of Long Chain Molecules 311

244'. TAMURA, M., and M. KURATA: Bull. Chem. Soc. J a p a n 25, 32 (1952).
244". - - - - and S. SATA: Bull. Chem. Soc. J a p a n 25, 124 (1952).
245. TAYLOR, G. B.: J. Am. Chem. Soc. 69, 635 (1947).
245a. TAYLOR, W. J.: J. Chem. Phys. 16, 257 (1948).
24S'. TCHEN, C. M.: J. Chem. Phys. 20, 214 (1952).
246. TERAMOTO, E. : Busseiron Kenkyu (Researches on Chemical Physics) 39, 1
(1951); 40, 18 (1951); 41, 14 (1951).
247. - - Proc. Intern. Conf. Theoret. Phys., Kyoto and Tokyo, J a p a n 1953, 410
(1954).
248. - - , and M. KURATA: Busseiron Kenkyu (Researches on Chemical Physics) 80,
11 (1955).
249. , R. CHUJO, C. SUZUKI, K. TANI and T. KAJIKAWA: J. Phys. Soc. J a p a n
10, 953 (1955).
249a. T1~oco, I., jr.: J. Am. Chem. Soc. 82, 4785 (1960).
249". TOBOLSKY, A. V. : J. Chem. Phys. 31, 387 (1959).
250. TREMENTOZZI,Q. A." J. Polymer Sci. 23, 887 (1957).
251. - - J. Polymer Sci. 36, 113 (1959).
251'. TROSSARELLI, L. : Paper given a t I U P A C Symposium on Macromol. Chem. in
Montreal, Canada (Aug. 1961).
252. - - , E. CAMPI a n d G. SAINI: J. Polymer Sci. 35, 208 (1959).
253. TSVETKOV, V. N., and V. G. ALDOS~IN: Zhur. fiz. Khim. (J. Phys. Chem.
U.S.S.R.) 33, 2767 (1959).
254. - - , and O. V. KALISTOV; Zhur. fiz. Khim. (J. Phys. Chem. U.S.S.R.) 33, 710
(1959).
255. - - , and S. YA KOTLYAR: Zhur. fiz. Khim. (J. Phys. Chem. U.S.S.R.) 30, 1100
(1956).
256. TUNG, L. H.: J. Polymer Sei. 24, 333 (1957).
257. UBEaRmTRR, K., H.-J. ORTHMAN and G. SoRoE: Makromol. Chem. 8, 21
(1952).
257'. UCHIDA, T., Y. KURITA a n d M. KUBO: J. Polymer Sci. 29, 365 (1956).
257". UDA, K., a n d G. MEYERHOFF: Makromol. Chem. 47, 168 (1961).
258. VOLKENSTEIN, M. V.: J. Polymer Sci. 29, 441 (19S8).
2 5 8 ' . - - , and O. B. PTITSYN" Zhur. Tekh. Piz. 25, 649, 662 (1955).
259. WADA, A.: J. Chem. Phys. 22, 198 (1954).
2 5 9 " . - J. Chem. Phys. 31, 495 (1959).
260. WAGNER, H. L., and P. J. FLORY: J. Am. Chem. Soc. 74, 195 (1952).
260". WALES, M., P. A. MARSHALL, S. ROTHMAN and S. G. WEISSBERG: Ann. N. Y.
Acad. Sci. 57, 353 (1953).
261. WALL, F. T., a n d J. J. ERPENEECK: J. Chem. Phys. 30, 634 (1959).
262.--, L. A. HILLER, jr., and W. F. ATCIIISOI~: J. Chem. Phys. 23, 2314
(19SS).
2 6 2 ' . - - , a n d J. MAZUR: Ann. N. Y. Acad. Sci. 89, 608 (1961).
263. WALSH, E. K., a n d I-I. S. KAUFMAN: J. Polymer Sci. 26, I (19S7).
263'. WATSON, J. D., and F. H. C. CRICK: Na1:ure 171, 737 (1953).
263". WESSLAU, H. : Makromol. Chem. 20, 111 (1956).
264. ~ Makromol. Chem. 26, 96 (1958).
265. WILLIAMS, J. W., W. M. SAUNDERS arid J. S. CICIRELLI: J. Phys. Chem. 58,
774 (1954).
265'. YAJNIK, N. A., M. D. BHALLA, R. C. TALWAR and M. A. SooFI: Z. physik.
Chem. 118, 305 (192S).
312 KURATA a n d STOCKMAYER."Intrinsic Viscosities a n d U n p e r t u r b e d Dimensions

266. YAMAKAWA, H . : J. P h y s . Soc. J a p a n 13, 87 (1958).


267. - - , a n d M. KURATA: J. P h y s . Soc. J a p a n 13, 78 (1958).
268. - - - - J. P h y s . Soc. J a p a n 13, 94 (1958).
269. - - - - j . Chem. P h y s . 32, 1852 (1960).
270. YANKO, J. A. : J. P o l y m e r Sci. 3, 576 (1948).
270'. Yoo, S. J., a n d J. B. KINSINGER: J. Chem. P h y s . 36, 1371 (1962).
271. ZZMM, B. H . : J. Chem. Phys. 14, 164 (1946).
272. - - J. Chem. P h y s . 24, 269 (1956).
272'. ~ J. Chem. P h y s . 33, 1349 (1960).
272". - - , a n d D. M. CROTH~RS: Proc. Natl. Acad. Scl. U.S.A. 48, 905 (1962); also
p r i v a t e c o m m u n i c a t i o n (1961).
273. - - , a n d J. I~. BRAGG: J. Chem. Phys. 28, 1246 (1958).
274. - - , P. DOTY a n d K. I s o : Proc. Natl. Acad. Sci. U.S.A. 45, 1601 (1959).
275. - - , G. M. RoE a n d L. F. EPSTEIN: J. Chem. P h y s . 24, 279 (1956).
276. - - , V~. H. STOCKMAYER a n d M. F I x M ^ ~ : J. Chem. P h y s . 21, 1716 (1953).

Potrebbero piacerti anche