Characterization of Poly(viny1 Alcohol)

DANIELE C. BUGADA and ALFRED RUDIN,* Guelph-Waterloo Centre

for Graduate Work in Chemistry, University of Waterloo, Waterloo,
Ontario, Canada N2L 3Gl

Synopsis
Molecular weight distributions, long chain branching frequency, and solution viscosities of
samples of commercial poly(viny1alcohol) (PVA) are reported. The PVA was fully reacetylated
to poly(viny1 acetate) (PVAc) for characterizations by size exclusion chromatography using a
low angle light scattering detector. The Mark-Houwink constants for PVAc in toluene were
determined to be K = 0.106 cm3 g-' and a = 0.59,at 25°C.Long chain branching frequency
in the commercial PVAs studied was small and was little affected by polymer molecular
weight. Some 95% or more of the branches in these species were short. Aqueous solutions at
10% (w/v) of PVA were Newtonian. The polymers examined differed in chemical composition,
molecular weight distributions, and mean block lengths of vinyl acetate residues. Variations
in a single characteristic, like a solution or intrinsic viscosity, cannot be used to deduce
structural differences between PVAs.

INTRODUCTION
The principal characteristics that determine the physical and technolog-
ical properties of polymers are chemical composition, molecular weight
distribution, and branching character. Poly(viny1 acetate) (PVAc) and its
derivative poly(viny1 alcohol) (PVA) are not exceptions. Fine differences in
PVA microstructure reflect structural variations in the precursor PVAc
and in the reaction system used to convert the PVAc to PVA. These dif-
ferences are often not reflected in the results of commercial quality control
tests, but they can be very important in the performance of PVA as a
protective colloid in emulsion and suspension polymerizations..
We have earlier' reported on the characterization of the molecular ar-
chitecture of commercial PVAs. Polymer samples were analyzed by 13C
nuclear magnetic resonance spectroscopy and by differential scanning ca-
lorimetry to measure tacticity, chemical composition, branch frequency,
and mean run lengths of vinyl acetate and alcohol units.
Further characterization procedures are reported and discussed in this
article. Parameters measured here include molecular weight distribution,
long chain branching, intrinsic viscosities, and aqueous solution viscosities
and unsaturation in the polymer backbone.
Mark-Houwink constants for linear PVAc in toluene are also reported.
This report and its predecessor' provide a critical survey of procedures
that are thought to give reliable and fairly comprehensive characterizations
of the chemical structure of commercial PVAs.

* Address communications to this author.

Journal of Applied Polymer Science, Vol. 30,4137-4147 (1985)

@ 1985 John Wiley & Sons, Inc. CCC OO21-8*5/85/ 104137-11$04.OO
4138 BUGADA AND RUDIN

EXPERIMENTAL
Six commercial partially hydrolyzed PVA samples, referred to as A to F
inclusive, were used in this study. These materials have been described
earlier.'

Acetylation Procedure
The following reacetylation procedure was used for all of the six partially
hydrolyzed PVA samples under investigation. Five grams of the polymer
was placed in a round bottom flask. The flask was purged with nitrogen
gas for about 10 mins. Fifty milliliters of a 2:l pyridine/acetic anhydride
solution were added to the reaction vessel, and the contents were heated
on a steam bath at 93-94"C, with stirring, under a nitrogen atmosphere.
After 4 h, the steam bath was removed, but stirring was continued for
another 30 min. The contents of the reaction vessel were then diluted with
acetone, and subsequently precipitated out in ice water. The poly(viny1
acetate) produced was collected and dried under vacuum at 60°C for about
20 h.

Gel Permeation Chromatography

Gel permeation chromatographic (GPC) measurements of the poly(viny1
acetate) samples were made in toluene solutions with 500, lo4, and lo5 A
(nominal porosity) porous polystyrene colums; at a flow rate of 0.5 mL/min
using injection volumes of 300-500 pL. Attempts to use silanized porous
silica columns were unsuccessful because of adsorption of the poly(viny1
acetate) onto these columns. The polymer samples were analyzed at 25"C,
at solution concentrations of about 3.5-4.5 mg/mL, in toluene containing
0.1% (w/v) of 2,6-di-tert-butyl-p-cresolantioxidant.
The polymer concentration in the eluant was monitored with a differ-
ential refractive index detector and KMX-6 (Chromatix) low angle laser
light scattering (LALLS) photometer, at a 6-7" angle. The specific refractive
index increment (dn/dc) measured with a Chromatix KMX-16 differential
refractometer at 25°C in toluene, was found to be -0.027 mL/g for samples
A and B, and -0.018 mL/g for the rest. Both the light scattering photometer
and differential refractometer employ a n He-Ne laser source at 6328 A
wavelength. All solutions were filtered through a 0.5 pm polytetrafluoroe-
thylene filter (FHUP, Millipore Corp.) before analysis.

Intrinsic Viscosities
The intrinsic viscosities of poly(viny1 acetate) solutions were measured
in toluene, at 25"C, with a n Ubbelohde viscometer that had a pure solvent
time of 174.3 s. The intrinsic viscosities were determined by a double ex-
trapolation of Huggins and Kraemer equations. Intrinsic viscosities of the
poly(viny1 alcoho1)s were measured in dimethylsulfoxide at 23.5"C.

Ultraviolet Spectra
Ultraviolet (UV) spectra of the partially hydrolyzed PVAs were run on
a Varian Cary 118 spectrometer, with 1 cm quartz cells (Hellma), in 0.5
 CHARACTERIZATION OF PVAs 4139

(w/v) aqueous solutions. Samples E and F were not characterized because

their high acetate content made them effectively insoluble in water.

Rheological Measurements
Rheological measurements were made with a Contraves Rheomat-30 cup-
and-bob viscometer, and recorded over a shear range of about 20-662 s-l.
The lower shear rate was limited by the instrument response towards the
10% (w/v) aqueous PVA solutions used, while the upper shear rate maxi-
mum of 662 s-' is determined by instrumental design. The sample tem-
perature was held constant at 25 k 0.1"C.
The procedure for the rheological measurements was essentially that
described by Rudin and Baas,2 and normally used for characterization of
water-based paints. The method measures thixotropy and shear thinning
behavior. In the present study all PVA solutions were found to be New-
tonian and showed no evidence of thixotropy.
Samples 11.-D, inclusive, were characterized as 10% (w/v) solutions in
water. Samp!es E and F were not studied because of their low water sol-
ubility.

BRANCHING IN PVAc AND PVA

Background
The formation of branches during vinyl acetate polymerization is ex-
pected because of the nature of the system. However, the actual detection,
and more importantly, the quantitative determination of the amount of
branching is not a trivial problem.
Branches in PVAc (which is, of course, the precursor to all commercial
PVA) can be long or short and hydrolyzable or nonhydrolyzable. Their
nature depends on the following different processes that are responsible
for branch formation:
(i)chain transfer to p ~ l y m e r , ~
(ii) chain transfer to monomer, followed by terminal double-bond co-
polymerization~,~
(iii) 1,5-backbiting m e ~ h a n i s m , ~
(iv) double-backbiting m e c h a n i ~ m . ~
Chain transfer to polymer can occur at both the methine carbon of the
main chain, resulting in nonhydrolyzable branches, and at the acetoxy
methyl group on the polymer, giving a hydrolyzable branch point.
Chain transfer to monomer at the acetate group results in the formation
of a polymer with a terminal double bond, which can copolymerize with
monomer to give a hydrolyzable branch point. Chain transfer to monomer
at the a- or &positions, followed by terminal double-bond copolymerization,
yields a nonhydrolyzable branch point. This sequence is unlikely, h ~ w e v e r . ~
Nonhydrolyzable branches may occur via a 1,Ei-backbiting mechanism,
followed by chain propagation (similar to ethylene polymerizations5), re-
sulting in short branches two monomer units in length. Furthermore, short-
4140 BUGADA AND RUDIN

er branches only two carbons (one monomer unit) long could result from a
double backbiting mechanism, again by analogy to polyethylene polymer-
ization.6
Quantitative determination of branch content in PVAc and PVA has,
obviously, important consequences since branching densities can affect
physical properties and technological applications of the polymer. For ex-
ample, short branching may influence protective colloid ability, which can
affect the latex viscosity of emulsions polymerized in the presence of PVA,7
and ~ r y s t a l l i n i t y , ~
which
.~ affects tenacity and water resistance of PVA
fibres.1° It is to be expected that the extent of branching in PVA will differ
with polymerization conditions" and reactor type.12
McDowell and Kenyon13 were among the first to address the problem of
branching in PVAc, by experimentally determining the presence of hydro-
lyzable branches in PVAc. This was done by comparing degrees of poly-
merization determined by viscosity measurements. Other studies have
appeared since, with different approaches to the problem. Most of the earlier
work was based on determinations of kinetic rate constants for transfer
reactions.14-16Nagai and Sagane17 estimated the amount of short chain
branching in PVA by using the infrared absorptions of the 1030 and 1093
cm-l bands in PVA. Graessley and ~ O - w o r k e r s have ~ ~ J ~shown that broad-
ening of molecular weight distribution, decrease in the intrinsic viscosity,
and decrease in molecular weight with saponification and reacetylation are
attributable to long-chain hydrolyzable branches present in the original
PVAc.
A kinetic study of the polymerization of radioactively labelled 14Cvinyl
acetate, in the presence of a crosslinked PVAc network which can be se-
lectively and quantitatively decrosslinked,20 indicated that chain transfer
to the main chain of the polymer was favored over transfer to the acetoxy
group,21but that a large number of hydrolyzable branches in PVAc are
formed by monomer transfer followed by terminal double bond copolymer-
i~ation.~
The detection and quantitative determination of short, nonhydrolyzable
branches was shown to be possible by using both the 13C NMR spectra of
PVA and the 'H NMR spectra of the acetylated PVA in both commercially
available1s7and model compound^.^^-^^ Estimation of branch content was
based on the ratios of the intensities of methine and methylene carbon
resonances as well as ratios of the resonance peaks assigned to the &protons
of the P-acetoxyethyl group (-CH2-CH,-OCOCH3) - compared to an in-
ternal standard.
In a more recent article,25the extent of long chain branching in PVAc
was measured by gel permeation chromatography. The authors concluded
that most of the branching in PVAc occurs through the acetate group, but
they also showed the existence of some chain branching in saponified and
reacetylated polymers.
The general conclusion from these reports seems to be that the original
PVAc contains a large concentration of hydrolyzable long chains, while the
PVA derived from it contains fewer nonhydrolyzable short branches and
only a small amount of long branches.
 CHARACTERIZATION OF PVAs 4141
GPGLALLS Studies
The amount of branching present in the polymers under investigation
has been reported from NMR results.' However, the length of these branches
was not clear. We have here attempted to determine the frequency of long
chain branching by application of GPGLALLS and the method of Rudin,
Grinshpun and O'Driscoll.%
The procedure generates a n estimate of the long branches per weight
average molecule with specified radius of gyration. The measured values
are presented for convenience as number of long branches per lo00 carbon
atoms vs. molecular weight. The analysis used is valid only if the number
of long branches does not increase with molecular weight.n This appears
generally to be true for low density polyethylene. The present results in-
dicate also that this criterion is valid for the reacetylated PVAs studied in
this research.
We note parenthetically that the minimum length for a long branch, as
defined by this analysis, lies between 6 and 16 carbon atoms in polyeth-
ylenes.%
The procedure used compares the molecular weight of the unknown poly-
mer with that of unbranched species that would elute with the same re-
tention time. That is to say, both the branched and unbranched polymers
have the same solvodynamic volumes and presumably the same radii of
gyration. The molecular weight of the branched polymer (M,) is measured
directly with the LALLS detector, while the molecular weight of the cor-
responding linear macromolecule is derived from the universal calibration
curve. The latter operation in this instance requires information as to the
Mark-Houwink constants of unbranched PVAc in toluene.
The Mark-Houwink constants were calculated from the GPC chroma-
tograms and the intrinsic viscosities of two samples of unbranched PVAc
in toluene.% Samples A and D were previously shown to be unbranched
from high resolution '3c spectroscopy.' These polymers were used to estab-
lish the constants in the Mark-Houwink equation:

as K = 0.106 cm3 g-' and a = 0.59, at 25°C in toluene.

Long chain branching results obtained by GPC-LALLS analyses of the
polymer samples are shown in Figure 1. Three polymers (B; E, and F)
contained detectable long branches. Samples A, C, and D had no long
branches. Samples A and D contained no short branches either, as men-
tioned, while polymer D was relatively lightly short branched (10 branches
per 1000 main chain carbons'). Sample F, which had the highest total
branching, from NMR analyses,' also had most long chain branching. Poly-
mers E and B were almost equal in total branching, but E had double the
number of long branches. In all cases, the great majority of branches are
short. Long branches constitute only 2-3% of the total.
Agarwal and co-workersZ5measured long branching in reacetylated PVA
by another technique. Their result is consistent with ours, in that the long
4142 BUGADA AND RUDIN

In
I
0
l-
a
u
0
0
0
LL
W
n
In
W
X
V
z
a
LT
m
W
z
0
-I
0
10 lo5 lo6
MOLECULAR WEIGHT
Fig. 1. Long chain branching frequency as a function of molecular weight for samples B,
E, and F, in toluene, at 25°C from GPGLALLS analyses.
chain branch density is small and little affected by molecular weight of the
reacetylated PVA.
The uniformity in the long chain branching density throughout the mo-
lecular weight range shown in Figure 1 is different from long chain branch-
ing observed in low density polyethylene (LDPE) by the same technique.%
In LDPE the long chain branching frequency is generally highest at low
molecular weights, and decreases continuously with increasing molecular
weight.
Variations in the branching distributions may reflect differences in the
syntheses of LDPE and PVAc. Commercial PVA is derived from PVAc made
by suspension or solution polymerization. In either system most of the
polymer is produced under reasonably isothermal conditions. LDPE is made,
however, in a nonisothermal reactor with no back mixing. The lower mo-
lecular weight polymer is produced at higher temperatures which favor
branching reactions and thus long chain branching and molecular weight
are inversely correlated.

MOLECULAR WEIGHT
The molecular weight of PVA is believed to be important when this
polymer is used as a dispersant and stabilizer. It is reported mthat surface
activity increases, i.e., interfacial tension decreases, with decreasing mo-
lecular weight, whereas protective colloid ability increases with higher de-
grees of polymerization. Particle size and latex viscosity are also thought
to depend on PVA molecular weight when this polymer is used as a pro-
tective colloid in emulsion polymerizations.
The usual control method for molecular size of partially hydrolyzed PVA
is by measuring viscosities of 4% aqueous solutions at 20°C. It is obvious,
however, that the measured viscosity can also be affected by aggregation
in solution, long chain branching, acetate content, and blockiness of residual
vinyl acetate units.
 CHARACTERIZATION OF PVAs 4143

Since commercial PVA is a copolymer, its molecular weight distribution

cannot be determined at present by GPC unless the material is converted
to a homopolymer by complete hydrolysis or by reacetylation. The latter
is the more common approach since the product PVAc is soluble in organic
solvents and can be analyzed by conventional GPC techniques. This pro-
cedure was followed in this work.
Table I lists the different molecular weight averages for the six PVAs
under investigation for GPGLALLS analysis of the reacetylated polymer
samples. Also listed are the number average molecular weights of the orig-
inal, partially hydrolyzed PVAs, as calculated from the number average
molecular weight of the reacetylated samples and the degree of.hydrolysis
determined by NMR analyses.'

RHEOLOGICAL MEASUREMENTS
It is well known that aqueous solutions of fully hydrolyzed PVA exhibit
non-Newtonian behavior,1° especially for high molecular weight samples at
higher concentrations. This is to be expected because of intermolecular
interactions resulting from entanglements and hydrogen bonding.
The situation is very different for partly hydrolyzed PVA, however. Hy-
drophobic residual acetate groups, especially when in blocklike distribu-
tions, reduce hydrogen bonding between molecular chains, thus reducing
aggregation and increasing water solubility .lo
Rheological measurements of samples A-D inclusive were performed on
a cup-and-bob viscometer as described in the Experimental section.
Solutions were measured at 10% concentration rather than the conven-
tional 4% used in quality control procedures in order to fit better into the
sensitivity range of the measuring instrument.
All 10% (w/v) aqueous solutions were Newtonian and not thixotropic.
Table I1 lists the viscosities of the aqueous solutions along with other
pertinent characteristics of the polymers. Intrinsic viscosities of the original
PVAs are also recorded in Table 11.
These polymers differ somewhat in chemical composition, molecular
weight distributions, and mean block lengths. Differences in solution vis-
cosity or intrinsic viscosity can be due to variations in any of these param-
eters. Samples A and B are similar in acetate content and mean block
character and the higher viscosities of polymer B solutions presumably
reflect the more pronounced high molecular weight tail in the latter sample.
Sample C has much lower viscosity in aqueous and DMSO solutions. Its
molecular weight parameters are not different enough from those of A and
B to account for this behavior. A more likely explanation is the higher
acetate content of polymer C.
In any event, it is clear that variations in a single characteristic, like a
solution viscosity or intrinsic viscosity, cannot be used to deduce structural
differences between PVAs.
The Newtonian behavior of water solutions of these PVAs is consistent
with their molecular architecture. We have shown' that these polymers
have blocklike residual vinyl acetate contents greater than 20 mol% and
relatively low crystallinity. These characteristics and the moderate molec-
 TABLE I
Molecular Weights of Reacetylated PVAs
- - - - - _
Sample M, M na M, M, MJM, SD (Nib SD (WIb SKEW (N). SKEW(WF
t0!!
A 39,800 24,400 104,600 1,269,000 2.6 50,800 349,000 37.5 26.0
B 44,900 27,800 148,600 4,556,000 3.3 68,300 809,400 93.4 54.9
C 25,000 15,800 90,900 3,078,000 3.6 40,600 521,100 102.5 48.8
**&
D 23,500 14,500 78,200 2,972,000 3.3 35,900 475,700 116.2 51.5 Z
E 11,200 7,900 29,300 478,000 2.6 14,300 114,800 51.5 18.5 U
F 3,510 2,500 76,400 725,000 21.7 16,000 222,700 46.8 11.3
a Number average molecular weight of original partially hydrolyzed PVA estimated from M,of reacetylated PVA and chemical composition from Ref. 1.
z
Standard deviation of number (N) and weight (W) distributions.30 Z
Skewness of number (N) and weight (W) distributions.m
 TABLE I1 %
Dynamic and Intrinsic Viscosities of Commercial PVAs
Composition' Long branches Dynamicb Intrinsid
G
vinyl acetate Block' per 1000 - - viscosity viscosity 5
5 p l e (mol%) character C atoms M wa M *a (mPa s) (mL/g) M
A 20.7 0.41 0 104,600 1,269,000 238 117.0 El
B 21.8 0.41 0 148,600 4,556,000 1019 176.8
C 25.0 0.44 0 90,900 3,078,000 84 63.6 E
D 21.6 0.46 1.o 78,200 2,972,000 66 69.0 5
E 40.0 0.72 2.0 29,300 478,000 - 18.8 Z
F 41.9 0.60 2.5 76,400 725,000 - 20.8
a Molecular weight averages of completely acetylated samples.
8
b 10% (w/v) aqueous solutions at 25'C.
2
c Dimethyl sulfoxide solutions at 23.5"C. 8
4146 BUGADA AND RUDIN

ular weights of the particular samples minimize intermolecular interactions

and result in "well-behaved" solutions even at 10% (w/v) concentration.
Aqueous mixtures of PVAs at the concentrations used in emulsion and
suspension polymerizations are certainly Newtonian, also.

ULTRAVIOLET SPECTRA
PVA is susceptible to degradation even at temperatures below lOO"C,
especially in the presence of oxygen and catalytic amounts of acid. The
initial decomposition products of partially hydrolyzed PVA are water and
acetic acid. Dehydration of the polymer results in the formation of unsat-
urated ketonic structures, whose presence and concentration can be deter-
mined from the ultraviolet spectrum of the polymer sample. Three
absorption bands have been observed in the spectra of aqueous PVA so-
lutions, at wavelengths of approximately 225, 280, and 325 nm. These ab-
sorption maxima have been attributed to the presence of carbonyl groups
conjugated to one, two, and three double bonds, re~pectively.~'
It is quite possible that partial degradation during the drying process of
PVA could be responsible for the conjugated carbonyl structures, especially
in commercial samples. In fact, some commercial samples are not white
powders as expected, but may very well be yellow, orange or even brown,
reflecting increased amounts of dehydration.
Figure 2 illustrates UV spectra of four of the samples under investigation.
Samples A and B (white colored) show the presence of two rather weak
absorption maxima at 205 and 270 nm. These two bands correspond to the
monoene and diene ketones. Samples C and D (yellow powders from a
different manufacturer) differ significantly in their ultraviolet spectra, not
only in the presence of an additional band at 320 nm, but obviously in the
magnitude of the absorbance for the bands at 280 and 320 nm.

200 250 300 350

x (rim)
Fig.2. Ultraviolet spectra of partly hydrolyzed samples in 0.5% (w/v) aqueous solutions:
(- - -)A, -
- * -)B(-)C; ( * * )D.
(*
 CHARACTERIZATION OF PVAs 4147

Differences in the ultraviolet spectra and thus in the concentrations of

unsaturated ketonic structures, are significant when we consider the use
of PVA as a surfactant and emulsifier. It is known that the presence of
these structures retards the rates of when PVA is used
as a protective colloid in emulsion and suspension polymerizations. Also,
grafting during free radical polymerization will be facilitated at carbons
alpha to unsaturation in the PVA backbone.

The authors thank Imperial Oil Limited and the Natural Sciences and Engineering Research
Council of Canada for their support of this research and V. Grinshpun for molecular weight
analyses.
References
1. D. C. Bugada and A. Rudin, Polymer (London), 25, 1759 (1984).
2.A. Rudin and J. Baas, Journal of the Oil and &bur Chemists Association, 66, (4),116
(1983).
3. S.Nozakura, Y.Morishima and S. Murahashi, J. Polym. Sci. Polym Chem Ed, 10,2853
(1972).
4.R. L. Adelman and R. C. Ferguson, J. Polym. Ski. Polym. Chem Ed, 13,891 (1975).
5.M. J. Roedel, J. Am. Chem. Soc., 75, 6110 (1953).
6.A. H.Willbourn, J. Polym. Sci., 34, 569 (1959).
7. A. S. Dunn and R. S. Naravane, Br. Polym. J., 12,75 (1980).
8. K.Kikukawa, S. Nozakura, and S. Murahashi, Polym J., 3,52 (1972).
9.Y.Morishima, Y.Irie, H. Iimuro, and S. Nozakura, J. Polym Sci., Polym. Chem. Ed.,14,
1267 (1976).
10.C. A. Finch, Polyvinyl Alcohol hperties and Applications, Wiley, London, 1973.
11. Y. Morishima and S. Nozakura, J. Polym Sci., Polym. Chem. Ed.,14,1277 (1976).
12.K. Magasubramanian and W. W. Graessley, Chem. Eng. Sci., 25, 1559 (1970).
13.W. H. McDowell and W. 0. Kenyon, J. Am. Chem. Soc., 62,415 (1940).
14.0.L. Wheeler, E. Lavin and R. N. Crozier, J. Polym. Sci., 9,157 (1952).
15. K. Mataumoto and M. Maeda, Kobunshi Kagaku, 14,582(1957).
16. S. Imoto, I. Ukida and T. Kominami, Kobunshi Kagaku, 14,101 (1957).
17.E.Nagai and N. Sagane, Kogyo Kagaku h h i , 59, 794 (1956).
18. W. W. Graessley and H. M. Mittelhauser, J. Polym. Sci., A-2, 5, 431 (1967).
19.W. W. Graessley, R. D. Hartung, and W. C. Uy, J. Polym. Sci., A-2, 7, 1919 (1969).
20.S. Nozakura, Y.Morishima, and S. Murahashi, J. Polym Sci, A-1, 10,2767 (1972).
21.S. Nozakura, Y.Morishima, and S. Murahashi, J. Polym. Sci., A-1, 10,2781 (1972).
22. Y.Inoue, R. Chajb and A. Nishioka, Poly. J., 4,244 (1973).
23. Y.Morishima, H.Iimuro, Y. hie, and S. Nozakura, J. Polym. Sci., Polym Lett. Ed,13,
157 (1975).
24. S.Nozakura, Y.Morishima, H. Iimuro, and Y. Irie, J. Polym Sci., Polym Chem Ed,14,
759 (1976).
25. S.H. Agarwal, R. F. Jenkins, and R. S. Porter, J. Appl. Polym. Sci,27,113 (1982).
26. A. Rudin, V. Grinshpun, and K. F. O'Driscoll, J. Liq. Chmmatogr., (1984),to appear.
27. A. Rudin, V. Grinshpun, and D. Potter, Polymer Bull., 13,71 (1985).
28. C. J. B. Dobbin, A. Rudin and M. F. Tchir, J. AppL Polym Sci,26,2985(1980);27,1081
(1982).
29. J. G. F'ritchard, Poly(viny1 Alcohol) Basic properties and Uses, Gordon and Breach,
London, 1970.
30. A. Rudin, J. Chem Ed,46,595(1969).
31. D.G. Lloyd, J: Appl. Polym Sci, 1, 70 (1959).
32. A. S.Dunn, Chem I d ,20,801 (1980).
Received September 6,1984
Accepted December 12,1984