You are on page 1of 27

12 LIGHTSCATTERING

FROM POLYMER SYSTEMS


R, W. RICHARDS
Department of Chemistry, University of Durham, Durham DHl 3LE, UK

12.1 INTRODUCTION
Light scattering from dilute polymer solutions has long been used by polymer
scientists to determine the molecular weights, the radii of gyration and the second
virial coefficients of polymers [1,2]. Such information has been instrumental in
the development of two-parameter theories [3] of polymer solutions, and
classical intensity light scattering has been exhaustively discussed. In addition to
two of the parameters mentioned (mean molecular weight and second virial
coefficient), classical intensity light scattering has also been used to obtain
information on the compositional dispersity in copolymers, and depolarised light
scattering can provide information on molecular anisotropy in rod-like poly-
merss such as the main chain liquid crystal polymers [4,5]. Both of these
applications have attendant difficulties. For copolymers, light scattering
measurements have to be made in at least three solvents with different refractive
indices to obtain compositional heterogeneity, although the judicious choice of
solvents can considerably simplify the calculation process [6]. Depolarised light
scattering intensity is often low and the signal-to-noise ratio can also be low.
These aspects of light scattering have been plentifully discussed and will not be
reviewed here.
A more recent development has been quasi-elastic light scattering [7,8], which
is often used to obtain diffusion coefficients of polymers in dilute solution.
Quasi-elastic light scattering has been applied to solvent-swollen cross-linked
networks and to semi-dilute solutions of polymers in an effort to investigate
scaling theories and reptation theories. Additionally, the possibility of obtaining
the viscoelastic properties of spread films of polymers at the air-water interface
by quasi-elastic light scattering has also been discussed. Both of these aspects will
be reviewed here.
Light scattering from solid polymer films and melts was first reported some 40
years ago [9]. The experimental difficulties have been considerably simplified by
newer technology, and the increased power of computing has eased the process of
Polymer Spectroscopy. Edited by Allan H. Fawcett
© 1996 John Wiley & Sons Ltd
data analysis. Much simpler experiments than implied in the original theory are
capable of producing information on the kinetics of phase-separating blends and
the thermodynamics of the systems. An overview of small angle light scattering
applied to semi-crystalline polymers and phase-separating blends will also be
given here.
All of these different types of experiment are united by a common source for the
light scattering observed, that is, the existence of fluctuations in polarisability,
and hence refractive index, at microscopic length scales in the material upon
which the light is incident. The cause of these fluctuations may differ markedly
from thermal fluctuations (surface quasi-elastic light scattering), concentration
gradients (quasi-elastic light scattering), density variations due to packing (cry-
stallinity) etc., but such fluctuations scatter light efficiently, so that light scattering
provides a convenient non-perturbative probe of the structure and dynamics of
polymer systems.

112 SMALL ANGLE LIGHT SCATTERING (SALS)

12.2.1 SEMI-CRYSTALLINEPOLYMERS
The pioneering work in SALS was done by Stein and his collaborators [10] some
30-40 years ago, and it is a relatively simple technique for investigating spherulite
growth and size in nearly transparent polymers. The original equations derived
by Stein and Rhodes [10] sought to explain the 'four leaf clover' pattern of
scattered light intensity observed in the Hv scattering experimental set-up shown
schematically in Figure 12.1. (Hv implies vertically polarised incident light,
horizontally polarised scattered light.) Spherulites are optically anisotropic, and
originally Stein and Rhodes explained the disposition of the scattered light as
lobes along the azimuthal angle of n/4 as being due to the orientation of dipoles in
the spherulite with respect to the plane of polarisation of the incident light. Much
interesting work has been analysed on the basis of the original equations (e.g.
influence of deformation [12], size distribution of spherulites [13], influence of
chain branching [14]). However, this explanation has been shown to be funda-
mentally incorrect by Meeten and co-workers [15-17], although the extraction
of parameters such as spherulite size etc. from SALS data using the original
equations is not altered. Meeten and Navard [16] used Mie theory, Rayleigh-
Gans-Debye theory and anomalous diffraction theory on isotropic spheres
and the latter two theories on anisotropic spherical scatterers. For all theories,
both types of particles produced 'four leaf clover' scattering patterns for Hv
scattering. The dependence of scattered intensity on azimuthal angle (f> and
scattering angle 0 can be calculated from the dimensionless angular gaing (which
is approximately the ratio of the intensity of the scattered light to the incident
Laser Potariser

Sample

Analyser

Detector Plane

Figure 12.1 Schematic of small angle light scattering experiment: scattering angle = 0;
the azimuthal angle <j> is measured from the plane of polarisation of the incident light

light intensity).
G^ = ( I A V ) I S 1 - S 2 I 2 sin 2 20
GVv = (4AV)IS 1 Sm 2 <t> + S2COs2 <f> \2
and
k = 2nlk

(Vv, vertically polarised incident light, vertically polarised scattered light).


The Rayleigh-Gans-Debye approximations are the most easily handled expres-
sions for S1 and S2.

Isotropic sphere
2/fcV
5 1 = — — Ox— l)(sinu —MCOSM)

5 2 = S1COsO
V = wm/ns

with nm and ns being the refractive index of the matrix and the sphere respectively.
Anisotropic sphere
2ik3r3
51 = 3 {3(/x— l)(sinw —MCOSW) + A / * [ u c o s K - 4 s i n u + 3Si(u)]}

2ifc3r3
52 =-^~3-{3(/I- I)(sinu-Mcosw)cos0- A^(I + cos2(0/2))(wcosw-4sinwSi(u))}
where jx = (nr + 2nt)/3nm, Afi = (nr — nt)/nm, nT is the radial refractive index, nt the
tangential refractive index, and Si(u) is the sine integral of u. For both cases r is the
radius of the sphere and u = An/X r sin(Q/2), with A being the wavelength of light
in the scattering polymer film. Figure 12.2 shows the form of the scattering
for both isotropic and anisotropic spheres and Hv scattering. Note that both
show a maximum scattering disposed in lobes at azimuthal angles of n/4.
Figure 12.3 shows the intensity variation with u along one such lobe; again both
have the same qualitative features, i.e. a maximum at a defined value of u. For

Figure 12.2 (Continued)


Figure 12.2 Contour plots of Hv scattered light intensity from (a) optically anisotropic
spheres (b) optically isotropic spheres; x = r sin <£, y = r cos <£, where r is the radius of the
sphere

anisotropic spheres wmax = 4.09, for isotropic spheres wmax = 2.74 for the first
order maximum.
Over the years the method of detecting the scattered light has improved;
originally photographic methods were used, followed by high speed cameras,
vidicons and optical multichannel analysers. Nowadays CCD cameras are able
to record scattering patterns digitally, and fast shutter speeds mean that the data
can be recorded in real time. A typical apparatus as constructed at Durham [18]
is shown in Figure 12.4; a description of similar equipment has recently appeared
[19]. Typically, the fastest shutter speed may be «20/xs and the time needed to
refresh the detector area and store 4K pixels each with 18 bit dynamic range is
« 2 s. One system on which this apparatus has been used is the crystallisation
Intensity

10 Intensity
4

Figure 12.3 Variation of Hv scattered light intensity at <j> = 45° for (a) optically anisot-
ropic spheres, (b) optically isotropic spheres; in each case u = (4nnr/Ao)sin(0/2). (a)
Reprinted with permission from Macromolecules, 1982, 15, 1004; (b) reprinted with
permission from [53]. Copyright 1982 and 1993 American Chemical Society

kinetics of linear diblock copolymer of methyl methacrylate and ethylene oxide


[20]. A copolymer with 76 mol % of ethylene oxide was quenched to a tempera-
ture of 308 K from 400 K and the Hv SALS pattern was recorded as a function of
time; the variation of intensity obtained is shown in Figure 12.5. From the
maxima in such curves, the radius of the spherulite was obtained, and was used in
an Avrami analysis of the crystallisation kinetics [21]. This provides a parameter
proportional to the rate of crystallisation and the Avrami exponent. From the
latter parameter it is sometimes possible to infer something about the growth
mechanism and geometry (Table 12.1). For this block copolymer the Avrami
Fibre optic link

CCD
Camera

Marata plate

analysor

hot stage+
sample

polariser

ND
filters

beam
expander

Figure 12.4 Block diagram of a small angle light scattering instrument

exponent obtained was 1.8; the equivalent homopolymer blend had an Avrami
exponent of 1.5. A word of caution is appropriate here. Meeten's analysis shows
that the value OfU1114x depends on the anisotropy of the spherulite; thus in the early
stages the observed growth rate may appear to be smaller than the actual growth
rate.
PEO-b-PMMA (76% w/w PEO) Block Copolymer, 7C=35 0C
Intensity

Polar scattering angle

Figure 123 Variation of Hv scattered light intensity at <j> = 45° for a linear diblock
copolymer of methyl methacrylate and ethylene oxide. The copolymer was quenched from
1000C to a crystallisation temperature of 350C and the time lapse between each curve is
2.1s

Table 12.1 Avrami exponents predicted for a variety of growth geometries and growth
control mechanisms
Exponent Nucleation" Growth geometry Growth control*
0.5 Instantaneous Rod Diffusion
1 Instantaneous Rod Interface
1 Instantaneous Disc Diffusion
1.5 Instantaneous Sphere Diffusion
1.5 Homogeneous Rod Diffusion
2 Homogeneous Disc Interface
2 Homogeneous Disc Diffusion
2 Homogeneous Rod Interface
2.5 Homogeneous Sphere Diffusion
3 Instantaneous Sphere Interface
3 Homogeneous Disc Interface
4 Homogeneous Sphere Interface
'Instantaneous: nuclcation on existing heterogeneities. Nuclei form simultaneously at beginning.
Homogeneous: sporadic formation of nuclei. Nucleation continuous in the untransformed material.
^Diffusion controlled: kinetics are controlled by the rate of diffusion of molecules to the nuclei.
Interface controlled: kinetics controlled by rate of attachment of molecules to the nuclei.
12.2.2 PHASE-SEPARATING POLYMER MIXTURES

Generally, compatible polymer mixtures display a lower critical consolute (or


coexistence curve), and the variation of the Gibbs free energy change on mixing
with composition shows a pair of inflection points above a particular tempera-
ture defined by the thermodynamics of the system and the mixture composition.
At these inflection points (d2AGJd<j>2)TtP = 0, and they define the locus of
the spinodal curve. This curve is the limit of stability of the mixture, i.e. within
the curve the mixture is unstable to any fluctuation and demixing (spino-
dal decomposition) takes place spontaneously. Between the coexistence curve
and the spinodal curve there is a metastable region wherein large fluctuations
are necessary to initiate demixing, usually via a nucleation and growth pro-
cess (Figure 12.6). Both curves meet at the critical temperature where
(d2AG/d<t>2)Tp = (d3AG/d<t>3)Tp = 0. If a compatible blend is quenched into the
spinodal region, phase separation takes place and the kinetics of the demixing are
describable by the linearised Cahn-Hilliard theory of spinodal decomposition,
which gives the time dependence of the composition variation as [22]
(d<t>/dt)TtP = M(d2Ald<t>2)T%pV24> - 2 M X V 4 0
where M is the mobility of the polymer and KV2<j> is the free energy density

binodal
spinodal
Temperature (K)

Figure 12.6 Schematic phase diagram for polymer-polymer mixtures


gradient due to composition gradients. The solution to this equation is a Fourier
series describing the compositional fluctuations in the system, i.e. the local
compositional deviations from the average value, and these fluctuations are the
source of the scattered light intensity. Since light scattering is described in Fourier
space terms, the solution to the Cahn-Hilliard equation is also needed in Fourier
space, Le. in terms of a wave vector, where the wave vector is (In/X) and X is the
concentration fluctuation wavelength. The intensity of light scattered from the
phase-separating mixture is proportional to the square of the fluctuation ampli-
tudes and is given by:
/ ( a 0 = /(Q^ = 0)exp[2R(Q)r]
where Q is the scattering vector = Ann sin(0)/Ao, with n the refractive index of the
sample, 20 the scattering angle and X0 the wavelength of light in vacuo. The term
R(Q) is known as the amplification factor and
R(Q) = - M(d2A/d<t>2)TtPQ2 - 2MKQ*
In the phase-separating system, there will be a most probable composition

10- 3 Q(Cm- 1 )
Figure 12.7 Scattered light intensity (Vv conditions) as a function of angle for different
times for a phase-separating mixture of polystyrene and polyvinyl methyl ether. Times
after start of phase separation (seconds)
A 2 +20 D40 O60
V 10 x 30 O 50 • 70
10ln(I(Q max )

Time (S)

Figure 12.8 Exponential dependence of scattered light intensity for a phase separating
mixture of polystyrene and poly vinyl methyl ether at Qmmx

fluctuation wavelength which will grow preferentially as phase separation pro-


ceeds. This leads to a maximum in the observed scattered intensity at a finite
value of Q. The position of this intensity maximum does not alter as in the early
stages of phase separation but increases in intensity as phase separation proceeds.
At late stages in the phase separation, there is a coalescence of particles via
Ostwald ripening and the maximum will shift to lower Q values.
During the early stages, at a fixed value of Q, the scattered light intensity from
a spinodally decomposing system should increase exponentially with time, and
from this relationship the amplification factor can be obtained. A set of light
scattering data for a demixing polystyrene/polyvinyl methyl ether (PVME)
mixture collected at discrete time intervals is shown in Figure 12.7 [23]; the
dependence of the scattered intensity on time at the Q value (Qmax) where the
maximum intensity is seen is shown in Figure 12.8. From values of R(Qn^x)
the effective diffusion coefficient De, can be obtained, as Dt = 2R(QmAX)/Qliax. At
the spinodal curve Dt = 0, and thus if Dc is obtained for a series of composition
and over a range of temperatures, the spinodal curve can be obtained. Figure 12.9
shows values of Dc as a function of temperature obtained for the mixture of
polystyrene and PVME referred to earlier, and the spinodal curve predicted from
these data is given in Figure 12.10. Light scattering investigations of other
demixing polymers have been reported elsewhere [24, 25] and recently a very
sophisticated instrument for such studies has been described [26].
PS(2)/PVME
1O10C-D9)Cm2S'1

Temperature (K)
Figure 12,9 Effective diffusion coefficient as a function of temperature

B PS(2)/PVME

cloud point curve


TEMPERATURE (K)

WEIGHT FRACTION PS

Figure 12.10 Spinodal curve (•) predicted from temperature dependence of Dc for
polystyrene/polyvinyl methyl ether mixtures
123 QUASI-ELASTIC LIGHT SCATTERING (QELS)
12.3.1 DILUTE POLYMER SOLUTIONS

Light scattering by polymers in solution is not a perfectly elastic process, small


amounts of energy being transferred between molecules and photons. This energy
transfer leads to a broadening of the frequency of the scattered light relative to the
incident light, and the intensity variation of the scattered light over a frequency
range from — oo to + oo is the spectral density or power spectrum, which is given
by

I((o) = 1/2« I °° < E*{t)E{t + T) > exp iojTdt

where <£*(r)£(t+T)> is the electric field autocorrelation function gt(t). In


quasi-elastic light scattering (QELS) what is actually obtained as the output from
the photomultiplier tube is the unnormalised intensity autocorrelation function
G2(t\ and
G2(t) = A + [Bg 1 (O] 2 (homodyne)
where A is a constant background intensity to which the correlation function
decays after a suitably long delay time f, and B is a constant close to unity. If we
have a single species in the solution, e.g. a monodisperse polymer, and there are
only concentration gradient relaxation processes, then
^1(O = exp(-Ff)
and F l is the relaxation time of the diffusive process of the polymer down the
concentration gradients; F = DQ2 with Q = (4nn/Xo)sin(0/2) and D is the transla-
tional diffusion coefficient. For polymer solutions, D is concentration dependent
D = D 0 (l + fcDc)
where D 0 is the infinite dilution value of D and c is the concentration of polymer.
The term kD is composed of thermodynamic and factional parameters for the
polymer in the particular solvent conditions investigated.
Polymers are not often monodisperse, and each different relaxation time will
make a contribution to the observed average F. A popular method of obtaining
the diffusion coefficient is to use the cumulants approach outlined by Koppel
[27] and the algorithm of Pusey et al. [28]
In Q1(Z) = - T11 + (F2/2!)r2 - (F3/3!)r3 + • • •
Generally only the first two cumulants can be extracted from the correlation
functions with any confidence, and TJQ2 = D29 the z-average diffusion coeffi-
cient. About 12 years ago, Burchard et al. [29] showed that
r JQ2 = D(I+ CR2Q2)
Hq 2 XiO 8 (Cm 2 S 1 )

Cj 2 XiO- 10 ^kC(Cm 2 )

Figure 12.11 Dynamic Zimm plot for polystyrene in toluene. Reproduced with per-
mission of the American Chemical Society from ref. [29]

where R9 is the radius of gyration of the polymer molecule and C is a parameter


related to the molecular architecture and the thermodynamic environment.
Incorporating the concentration dependence of D,
TJQ2 = D0(I + kDc)(l + CRlQ2)
Thus, as c->0 and Q->0, TJQ2 = D 0 and D 0 , kD and C can be obtained from
a 'dynamic' Zimm plot (Figure 12.11). The slope of the line dependent on Q2 alone
is CR2, whereas the slope of the line dependent on c only is DokD. Thus method
has not been widely used; however, it has been applied to naturally occurring
polymers to extract C and thus to enable something to be said about their
structure.
We noted earlier that each relaxation time will contribute to T and hence
influence the shape of the correlation function. Consequently, all the information
on polymer polydispersity is contained within the intensity correlation function
because D is proportional to (molecular weight)"0. The extraction of the molecu-
lar weight distribution from the correlation function is an 'ill posed problem', as
there are an infinite number of solutions to the Laplace inversion of the data that
is required to obtain the distribution. Several attempts have been made at
developing suitable computational methods to derive a distribution from a corre-
lation function. Perhaps the most widely known and used is the constrained
regularisation programme CONTIN [30,31]. In many cases the programme works
well, but care has to be taken in choosing the right range of D to explore for
a solution, and the original data must be of high quality, as 'noisy' data can lead to
artefacts in the analysis. A comparison of CONTIN with maximum entropy
methods has recently been published [32].
Relative Contribution

Diffusion Coefficient (cm 2 s'x) x1


°

Figure 12.12 Distribution in diffusion coefficients for an aromatic terpolyester in a mixed


solvent of trifluoroacetic acid and dichloromethane obtained by CONTIN analysis of
quasi-elastic light scattering data

Obtaining molecular weight distributions by this means has two benefits.


Firstly, with a high power laser light source on the correlator and with fast data
links to a work station, a full molecular weight distribution can be obtained in
«2min. The second benefit is when only ferocious solvents are available, ones
which would destroy size-exclusion chromatography (SEC) column packings;
quasi-elastic light scattering then becomes a highly suitable method to obtain
a molecular weight distribution. An example of this is the aromatic terpolyester
prepared from hydroxybenzoic acid, isophthalic acid and hydroquinone [33],
which is soluble in a mixture of trifluoroacetic acid and methylene chloride. The
low refractive index of the solvents and the high refractive index of the polymer
make the solutions extremely strong scatterers of light and ideal for CONTIN
analysis, even with only a modest laser. An example of the distribution in
diffusion coefficients (and hence molecular weight) is shown in Figure 12.12.

12.3.2 GELS

A cross-linked polymer swollen by a solvent constitutes a gel, and if swollen


sufficiently the concentration of polymer in the gel is that of a semi-dilute
solution, i.e. it is between c* and c** as defined by de Gennes. The gel has
continual local fluctuations in the degree of swelling (equivalent to polymer
concentration) which lead to variations in the local osmotic pressure. The
analysis of the intensity correlation function obtained from the scattering of light
by these fluctuations produces a co-operative diffusion coefficient. The first
QELS experiments on gels and the theoretical analysis of the data were reported
over 20 years ago by Tanaka et al. [34]. They showed that at a delay time of zero
(i.e. extrapolating the correlation functiion to t = 0), the scattered light intensity
above the background was equal to the osmotic moldulus Mos (= Kos + 4Gos/3
where Kos is the bulk osmotic modulus and Gos is the shear osmotic modulus),
also known as the longitudinal modulus. The co-operative diffusion coefficient is
given by
DC = (KOS + 4GOS/3)(1 -<t>p)/f
where cf)p is the volume fraction of polymer in the gel and / is the total friction of
the polymer against the solvent per unit volume
/ = CeNAc/m
where c is the polymer concentration in gml" 1 , m is the monomer molecular
weight, and £c is the monomeric friction coefficient at concentration c.
Since the first report there have been many papers published on light scattering
from polymer gels, the work of Geissler and Hecht on polyacrylamide gels
[35-39] being noteworthy.
Measurements [40, 41] obtained on radiation cross-linked polystyrene gels
subsequently swollen in cyclohexane at different temperatures exemplify the type
of results obtained. A typical correlation function is shown (Figure 12.13) in
which the ordinate axis was calibrated directly in terms of osmotic modulus using
data obtained by Scholte [42] from ultracentrifugation analysis of polystyrene
solutions. Scaling relationships can be used to interpret the dependence of Mos
and Dc o n <t>p.
The dependence of Dc on the volume fraction of the polymer in the gel varied
markedly with the temperature (Figure 12.14), whereas the osmotic modulus for
these same gels could be fitted by the same scaling relationship
Mos = 4.7 x 1 0 6 ^ 6 N m - 2
Scaling laws predict that the exponent of 0 p for Mos should vary from 2.25 in good
solvents to 3 in theta solvent conditions; for the concentration dependence of
diffusion coefficients the same exponents are 0.75 and 1 respectively. At 308 K an
exponent of 1.17 was observed, which within the experimental error agreed with
predictions. However, at 333 K, the exponent was 0.46, much lower than theory
predicts. This observation and the high exponent observed for Mos were at-
tributed to the presence of dangling chains in the network, since the correlation
functions were observed to become more non-exponential as the temperature
T = 308K
Solvent C6H12
Intensity (a.u.)

Baseline

Time (jus)
Figure 12.13 Intensity autocorrelation function obtained for a randomly cross-linked
network of polystyrene swollen in cyclohexane at 308 K

was increased. Increased non-exponential behaviour has been identified with the
overlapping of molecules and appears to be possible only when there are many
loose dangling chains.

12.3.3 SEMI-DILUTE S O L U T I O N S A N D TRAPPED CHAINS

The broad outlines of reptation theory are well known, and the detailed theory is
available elsewhere [43,44]. Essentially, a polymer molecule in a melt is confined
to a tube which is defined by the surrounding molecules, and can only move along
the tube axis. The time dependence of the various dynamic modes of the molecule
in the tube has been discussed by Doi and Edwards [45]. Additionally, de Gennes
[46] has set out equations which relate the translational diffusion coefficient of
a probe polymer to its molecular weight (Mp), the entanglement molecular weight
of the matrix (MJ and the molecular weight between cross-links (AfJ. Three
regimes are predicted:
1. Free draining (A/p < Afc, Afp > AfJ, D = D0M; K
DJm2S'1)

Figure 12.14 Co-operative diffusion coefficient as a function of volume fraction of


polymer in cyclohexane swollen polystyrene networks; (o) 308 K, (o) 318 K, (•) 333 K

2. Simple reptation (M p > Me, M c > Mc), D = D0M tM; 2.


2
3. 'Strangulation' regime (Me > Mc, M p > Mc), Dt = D0M0M; .
Attempts have been made at observing these regimes using semi-dilute sol-
utions of a matrix polymer with a chemically identical probe of a different
molecular weight incorporated in the solution. The conclusion of these experi-
ments was that the reptation theory was inappropriate for such semi-dilute
solutions [47,48]. A possible explanation for the failure of reptation theory may
be in the recent analysis of Wang [49-51]. He shows that the quasi-elastic light
scattering from a semi-dilute solution has contributions from both concentration
fluctuations and density (pressure) fluctuations, and consequently the long time
viscoelastic relaxation spectrum, usually observed by dynamic mechanical
means, will also contribute to the autocorrelation function. The extent to which
both contributions are seen depends on the frequency distribution of the stress
relaxation modulus and a coupling parameter j8 (proportional to the partial
log[D t <M x >/M x ]

log M

Figure 12.15 Diffusion coefficient of polystyrene tracer in polyvinyl methyl ether gels as
a function of tracer molecular weight. Diffusion coefficients normalised by ratio of
molecular weight between crosslinks of gels. Reprinted with permission from [52].
Copyright 1992 American Chemical Society

specific volume of the polymer minus the partial specific volume of the solvent).
Very recently, QELS investigation of reptation predictions has been made using
randomly cross-linked networks containing chemically distinct trapped chains.
Rotstein and Lodge [52] prepared polyvinyl methyl ether gels containing
trapped polystyrene chains, and obtain tracer diffusion coefficients for the
toluene-swollen gels. Values of M c were calculated from swelling data, and
4 x 103 ^ Mc ^ 14 x 103. Figure 12.15 shows the diffusion coefficient data nor-
malised by the ratio of the M c values for the three networks involved. There
appears to be little or no influence of Mc even when M p » Me; furthermore, the
probe molecular weight dependence of D (DocM~2S) is much stronger than
predicted by reptation theory. Pajevik et al. [53] prepared randomly cross-linked
polymethyl meth'acrylate gels containing polystyrene probe molecules. Their
results are shown in Figure 12.16. When M p < Mc («80000) then D scales as
Mp ° 6; above this molecular weight the influence of M p is marked and D scales as
M~ l'*±°-29 i.e. almost exactly in agreement with reptation theories, CONTIN or an
equivalent program was used in both investigations, and the isorefractivity of
toluene with polyvinyl methyl ether and polymethyl methacrylate aids the
D*/D0

Mp

Figure 12.16 Ratio of polystyrene tracer diffusion coefficient (D1) in toluene swollen
PMMA gel to diffusion coefficient of polystyrene in dilute toluene solution (•); (A) values
for PS tracer in PMMA solutions. Reproduced with permission from the American
Chemical Society from Ref. [53]

process of extracting the probe diffusion coefficient. However, about 14 years


[54] ago it was noted that, when polystyrene was dissolved in a semi-dilute
benzene solution of polymethyl methacrylate, the value of D decreased as the
polymethyl methacrylate concentration increased, i.e. rather similar to the
molecular weight dependence seen by Pajevik et al, and this may be due to
polymer-polymer interactions.
To overcome these possible complications, polystyrene networks with trapped
polystyrene molecules have been prepared [55] and are currently being inves-
tigated.

12.3.4 SURFACE QUASI-ELASTIC LIGHT SCATTERING


(SQELS)

A liquid surface is continually roughened by thermal excitations, which give rise


to the hydrodynamic modes known as capillary waves. The r.m.s. amplitudes of
the waves are small ( « 2A) but they are efficient light scatterers. The displacement
of the liquid surface from its equilibrium position by a wave propagating in the
x direction is:
C(x,r) = C 0 exp(/ex-ho)0
where Q is the surface wavenumber or the scattering vector parallel to the liquid
surface. The wave frequency o is a complex quantity given by a>0 + iT, where co0
is the capillary wave frequency and F is the decay rate of the waves. A dispersion
equation relates co and Q, and for pure liquids the controlling factors (for fixed Q)
are the kinematic viscosity and the surface tension [56]. For most instruments the
accessible range of Q is 100-2000Cm"1 and hence the wavelengths probed are
« 600-30 /mi. If a polymer film is spread on the surface of the liquid, additional
hydrodynamic modes modify the dispersion equation. Only the transverse modes
(capillary waves) scatter light, but there is coupling with the longitudinal or
dilational modes, and hence in principle some information is obtainable on both
modes from the power spectrum of the scattered light. The parameters obtainable
are the surface tension y and the dilational modulus e; both of these are
viscoelastic properties, as energy dissipation takes place in the relaxation pro-
cesses, and thus
y = y0 + icoy'
e = 6 0 + icoe'

where y0 and £0 are the static surface tension and dilational modulus I — I,
\ A aA J
y' is the transverse shear viscosity and e' is the in-plane dilational viscosity.
Although direct measurement of the frequency broadening of the scattered
light by the capillary waves has been used, the frequency shifts are rather small,
and a more direct means of observing the frequency of the capillary waves is to
use heterodyne quasi-elastic light scattering [57,58]. The experimental arrange-
ment to collect such data is shown in Figure 12.17; the diffracted beams produced

Laser

rough

PM Tube

Figure 12.17 Schematic diagram of surface quasi-elastic light scattering apparatus. Ll,
L2 = lenses, T = transmission grating, F = neutral density filter, Ml, M2, M3, M4 =
mirrors
Normalised correlation function

Time (us)
Figure 12.18 Heterodyne correlation function for syndiotactic polymethyl methacrylate
spread on water at a surface concentration of 1.7mgm~2

by the transmission grating act as the reference beam of zero frequency shift, and
this beats with the scattered light at the photocathode to produce the typical
correlation function shown in Figure 12.18. From these data the capillary wave
frequency co and the decay constant F can be obtained. By assuming that y and e!
are zero, y0 and e0 can be obtained from these values by solving the dispersion
equation. Extracting the viscous moduli requires a non-linear least squares fit of
the Fourier transform of the power spectrum equation to the data. A computa-
tional method for this process has been developed by Earnshaw et al. [59] and
exhaustively justified [60]. Wider aspects of light scattering from liquid surfaces
are discussed in the book edited by Langevin [61].
To date much of the work published on SQELS from spread polymers has
emanated from Yu and colleagues [62-65], but assumed that the viscous moduli
are zero.
We have reported [66] a limited study of spread polymethyl methacrylates and
polyethylene oxide. Figure 12.19 shows the variation in surface tension, shear
viscosity and dilational modulus obtained from SQELS data as a function of
surface concentration. The viscoelastic moduli both show maximum values at
finite values of the surface concentration. As the capillary waves generate
oscillatory stress and strain, these are related via the complex dynamic modulus
of the surface

a* =y*[G'(co) +iG"(co)]
Surface tension (mN nrr1)

Surface concentration (mg nrr2)


Shear viscosity (mN s rrr1)

Surface concentration (mg m*2)

Figure 12.19 (Continued)


Dilational modulus (mN nrr1)

Surface concentration (mg nrr2)


Figure 12.19 Derived parameters from surface quasi-elastic light scattering as a function
of concentration of polymethyl methacrylate spread on water: (a) surface tension; (b)
surface shear viscosity; (c) dilational modulus

where <x* is stress, y* is strain, G'(co) is the storage modulus (surface tension) and
G"(a>) is the loss modulus (a>yf). Using volume fraction composition data obtained
from neutron reflectometry on the spread polymer films, it is evident that the
surface film loss modulus is linearly dependent on the volume fraction of polymer
in the film. If we presume that the relaxation process in the surface film is
described by a Maxwell model, then
G'(co) = Ge + GCO2T2/(1 + CO2T2)
where Ge is the elastic modulus at co = O, i.e. the static surface tension. Further, if
there is only one relaxation process in the spread film, then
T = A7c/co2y'
where An is the difference in the surface tensions measured by SQELS and from
static (Wilhelmy) plate methods. The dependence of relaxation time on the
volume fraction of the polymer shows an exponential increase, Figure 12.20. To
obtain further insight into the relaxation mechanism requires the frequency
dependence (i.e. different Q values) of the transverse shear viscosity to be known.
SYN PMMA SQELS DATA
Relaxation time (s)

VoI fraction of polymer

Figure 12.20 Relaxation time for spread polymethyl methacrylate as a function of


volume fraction of polymer in the spread film

12.4 CONCLUSIONS
An overview of some of the areas where light scattering has made contributions to
polymer science has been given. The emphasis has been on dynamics, either by
using light to follow a process (crystallisation or phase separation) or using
dynamic light scattering per se. A broad range of polymer types and situations has
been covered and the discussion has by no means been exhaustive. Evidently,
despite its maturity as a laboratory technique, light scattering is still capable of
providing much information on polymer systems. Furthermore, the development
of newer applications such as surface quasi-elastic light scattering will enable
investigations of surface gelation and surface ordering in polymer solutions, areas
which have yet to be investigated.

12.5 REFERENCES
[1] M.B. Huglin (Ed.), Light Scattering from Dilute Polymer Solutions, Academic Press,
London, 1972.
[2] P. Kratochvil, Classical Light Scattering from Polymer Solutions, Elsevier, Amster-
dam, 1987.
[3] H. Yamakawa, Modern Theory of Polymer Solutions, Harper, New York, 1971.
[4] GC. Berry, J. Polym. ScL, Polym. Symp., 1978,65,143.
[5] W.R. Krigbaum and G. Brelsford, Macromolecules, 1988, 21, 2502.
[6] Z. Tuzar, P. Kratochvil and D. Strakova, Eur. Polym. J., 1970,6,1113.
[7] BJ. Berne and R. Pecora, Dynamic Light Scattering, Wiley, New York, 1976.
[8] K.S. Schmitz, An Introduction to Dynamic Light Scattering by Macromolecules,
Academic Press, San Diego, 1990.
[9] R.S. Stein and JJ. Keane, J. Polym. ScL, 1955,17, 21.
[10] R.S. Stein and M.B. Rhodes, J. Appl. Phys., 1960, 31,1873.
[11] R.S. Stein, P. Erhardt, JJ. van Aartsen, S. Clough and M. Rhodes, J. Polym. Sci. C,
1965,13,1.
[12] RJ. Samuels, J. Polym. Sci. C, 1965,13, 37.
[13] G.E. Wissler and B. Crist, J. Polym. Sci., Polym. Phys. Ed., 1985, 23, 2395.
[14] M. Ree, T. Kyu and R.S. Stein, J. Polym. ScL, Polym. Phys., 1987, 25,105.
[15] J.V.Champion,A.KilleyandG.H.Meeten,J.Polym.ScL,Polym.Phys.Ed., 1985,23,
1467.
[16] G.H. Meeten and P. Navard, J. Polym. ScL, Polym. Phys., 1989, 27, 2023.
[17] M. Desbordes, G.H. Meeten and P. Navard, J. Polym. ScL, Polym. Phys., 1989, 27,
2037.
[18] P.H. Richardson and R.W. Richards, unpublished work.
[19] WT. Culberson and M.R. Tant, J. Appl. Polym. ScL, 1993,47, 395.
[20] P.H. Richardson, ubpublished results.
[21] J.C. Schultz, Polymer Materials Science, Prentice-Hall, New Jersey, 1974.
[22] J.W. Cahn and J.E. Hilliard, J. Chem. Phys., 1958,28, 258.
[23] J.G. Connell, Ph.D. Thesis, University of Strathclyde, 1989.
[24] H.L. Snyder and P. Meakin, Macromolecules, 1983,16, 757.
[25] T. Hashimoto, M. Itakura and N. Shimidzu, J. Chem. Phys., 1986,85,6773.
[26] A. Cumming, P. Wiltzius, F.S. Bates and J.H. Rosedale, Phys. Rev. A, 1992, 45,
885.
[27] D.E. Koppel, J. Chem. Phys., 1972,57,4814.
[28] P.N. Pusey, D.E. Koppel, D.W. Schaefer, R.D. Camerini Otero and S.H. Koenig,
Biochemistry, 1974,13, 952.
[29] W. Burchard, M. Schmidt and W.H. Stockmayer, Macromolecules, 1980,13,1265.
[30] S.W. Provencher, Comput. Phys. Commun., 1982,27, 213.
[31] S.W. Provencher, in E.O. Schulz-DuBois (Ed.), Photon Correlation Techniques in
Fluid Mechanics, Springer, Berlin, 1983.
[32] S.W. Provencher, in S.E. Harding, D.B. Sattele and V.A. Bloomfield (Eds.), Laser
Light Scattering in Biochemistry, Royal Society of Chemistry, Cambridge, 1992.
[33] A.D.W. McLenaghan, Ph.D. Thesis, University of Strathclyde, 1990.
[34] T. Tanaka, L.O. Hocker and G.B. Benedek, J. Chem. Phys., 1973,59, 5151.
[35] E. Gleissler and A.M. Hecht, J. Phys. (Paris) Lett., 1979,40, L173.
[36] A.M. Hecht and E. Geissler, J. Phys. (Paris), 1978, 39, 631.
[37] A.M. Hecht, E. Geissler and A. Chosson, Polymer, 1981,22, 877.
[38] E. Geissler and A.M. Hecht, J. Chem. Phys., 1982, 77,1548.
[39] EJ. Amis, P.A. Janney, J.D. Fery and H. Yu, Macromolecules, 1983,16,441.
[40] N.S. Davidson, Ph.D. Thesis, University of Strathclyde, 1984.
[41] N.S. Davidson, R.W. Richards and E. Geissler, Polymer, 1985, 26,1643.
[42] T.G. Scholte, J. Polym. ScL A2, 1970,8, 841.
[43] W.W. Merrill and M. Tirrell, in G.R. Freeman (Ed.), Kinetics of Nonhomogeneous
Processes, Wiley, New York, 1987.
[44] T.P. Lodge, N.A. Rotstein and S. Prager, Adv. Chem. Phys., 1990,79,1.
[45] M. Doi and S.F. Edwards, The Theory of Polymer Dynamics, Oxford University
Press, Oxford, 1986.
[46] P.G. de Gennes, Macromolecules, 1986,19,1245.
[47] W. Brown and P. Zhou, Macromolecules, 1989, 22, 3508.
[48] T. Nicolai, W. Brown, S. Hvidt and K. Heller, Macromolecules, 1990,23, 5088.
[49] CH. Wang, J. Chem. Phys., 1991,95, 3788.
[50] CH. Wang, Macromolecules, 1992, 25, 1524.
[51] CH. Wang and X.Q. Zhang, Macromolecules, 1993,26, 707.
[52] N.A. Rotstein and T.P. Lodge, Macromolecules, 1992, 25,1316.
[53] S. Pajevic, R. Bansil and C Konak, Macromolecules, 1993, 26, 305.
[54] AJ. Hyde, J. Hadgraft and R.W. Richards, J. Chem. Soc. Faraday Trans II, 1979,75,
1495.
[55] D.A. Davison, University of Durham, work in progress.
[56] J.C Earnshaw and R.C McGivera, J. Phys. D, 1987,20, 82.
[57] S. Hard and R.D. Neuman, J. Colloid Interface ScL, 1981,83, 315.
[58] J.C. Earnshaw, in CA. Croxton (Ed.), Fluid Interfacial Phenomena, Wiley, New
York, 1986.
[59] J.C Earnshaw, R.C McGivern, A.C McLaughlin and P. J. Winch, Langmuir, 1990,
649.
[60] J.C. Earnshaw and R.C. McGivern, J. Colloid Interface ScL, 1988,123, 36.
[61] D. Langevin (Ed.), Light Scattering by Liquid Surfaces and Complementary Tech-
niques, Dekker, Basel, 1992.
[62] M. Kawaguchi, M. Sano, Y.-L. Chen, G. Zografi and H. Yu, Macromolecules, 1986,
19, 2606.
[63] M. Kawaguchi, B.B. Sauer and H. Yu, Macromolecules, 1989, 22,1735.
[64] B.B. Sauer, M. Kawaguchi and H. Yu, Macromolecules, 1987, 20, 2732.
[65] K.-H. Yoo and H. Yu, Macromolecules, 1989, 22,1989.
[66] J.A. Henderson, R.W. Richards, J. Penfold and R.K. Thomas, Macromolecules, 1993,
26,65.

You might also like