Vous êtes sur la page 1sur 23

Light Scattering by

Polymer Solutions

Light Waves A Brief Review


A traveling light wave can be described by the spatial and temporal
variations in its electric field vector, E:

(8.1.1)

where k is the wavevector. It has the


direction of the light wave propagation
and magnitude k = 2/, with being
the wavelength. Appearance of
kr in the cosine means that the phase of the wave on all the points lying
on a given plane perpendicular to k is the same at any given time. is
the angular frequency and related to the frequency , speed of light c
and wavelength in free space 0 by:

(8.1.2)

1
Light Waves A Brief Review
Notice that the wavelength in a medium is generally not equal to 0 .
They are related by

(8.1.6)

where n is the refractive index of the medium and v is the speed of


light in the medium. For visible light, 0 3500-7000 . So, = c/ 0
1015 Hz. At such high speeds, typical photodetectors can only
measure the light intensity (energy per unit area per unit time), I,
integrated over time, where I is proportional to |EE| = E2. Such
detectors are called square-law detectors. In experiment, we usually
normalize the detector signal by the incident beam signal. This
cancels out the unknown proportionality factor between the detector
signal and I or E2.
It is more convenient to work with complex notations for E(r,t) and write:

(8.1.4)
3

(8.1.5)

where E* is the complex conjugate of E and i = (-1)1/2.


As mentioned above, the wavelength and hence magnitude of k
depends on the refractive index, n. In turn, n depends on the
polarizability of the constituent molecules and their spatial
arrangement (density and orientation). Consequently, more
polarizable chemical moieties such as aromatic rings generally lead to
higher refractive indices. In the liquid state, all orientations of the
individual molecules are equally probable and we say that the liquid is
isotropic, and so do not need to worry about the fact that most
molecules are actually anisotropic. The magnitude of n usually lies
between 1.3 and 1.6 for polymers and aqueous or organic solvents.
According to the Lorentz-Lorenz equation for an ideal gas (or the
Clausius-Mosotti equation):
(8.1.7)

where ns is the refractive index of the pure solvent and 1/ is the


number of solute particles per unit volume. 4

2
The Basis What is Scattering?
Scattering is the re-radiation of a traveling wave due to a change in the
character of the medium in which the wave is traveling. For light,
scattering will be caused by local changes in refractive index or
polarizability, due to, e.g. a dust particle in the air or a polymer in the
solvent. Note that we assume nonabsorbing media here. So, the
incident light intensity will either be transmitted or scattered. If there is
no energy exchange between the medium and the wave, the incident
and scattered wave frequencies are the same and we call the
scattering process elastic. If very small differences in energy is
involved, the scattering is quasielastic. When there is large change in
the energy of the wave upon scattering, the process is inelastic.
Representative inelastic scattering in polymers include Raman and
Brillouin scattering. The former involves vibrational excitations and the
latter propagating density waves (i.e., sound waves or phonons) in the
medium. Lastly, scattering is incoherent if the intensity is independent
of scattering angle, and coherent if it depends on the scattering angle.
We will elaborate the concept of coherent scattering in this context
later. 5

Basis Scattering from Randomly Placed Objects


If the scattering is due to randomly placed, noninteracting objects, the total scattered
intensity, Is, should be proportional to the number of scattering objects:

Is ~ (number of scatters) x (scattering power of each object) (8.2.1)

This should apply to the case of molecules in solution. We also anticipate larger
molecules to scatter proportionally more than smaller ones.

N.B. Suppose we scatter light from objects that are connected to one another, e.g.,
monomers within one polymer. Now the phases of the waves scattered from different
monomers should be related, because the distance rjk between any two monomers j
and k has some preferred value or range of values. This will lead to some
interference between these waves. Recall that interference between two waves is
constructive only if the difference in distance or path length difference (PLD) that the
two waves travel to the detector is some integral number of wavelengths, m, and
destructive if the PLD is an integer plus one half of . Because the PLD is on the
order of the distance between scatterers, we can therefore expect that in order to see
significant destructive interference we need the distance between scatterers to be a
significant fraction of , say a few percent. This will tum out to be the case. For visible
light with o 5000 , only for polymers that are larger than at least 100
(corresponding to polymers with Mn > ~ 105 g/mol) do we have to worry about this 6
kind of interference.

3
Basis Scattering from a Perfect Crystal
PLD =L -L = (/2)(k -k )r
1 2 s i
Consider a homogeneous array of
identical scatterers (e.g. atoms) that are k s ks
placed every 1 apart and shine light
with = 5000 on it. Every scatter will L1
re-radiate in all directions with the same L2 ki
amplitude and wavelength. For any
particular scatter and particular !
scattering direction k, we can always r Fig. 8.*
find another scatterer that produces a
scattered wave in the direction of k with path length difference (PLD)
exactly equal to /2 w.r.t. the first scatterer. This is possible because the
separation between scatterers is much less than so the array is
essentially a continuum. However, the above argument does not apply in
the forward direction (ki // ks) because the phase shift of the scattered
waves between two atoms is exactly canceled by the phase shift
between the incident waves arriving at the two atoms. Thus, the light
beam propagates through a homogeneous medium without any
scattering. 7

Basis Origin of Coherent & Incoherent Scattering


There are two ways by which we can generate scattering from our
hypothetical array. One would be to remove a few scatterers or add a few
scatterers with different scattering strengths than the medium molecules at
random. Then the "pairing-off" argument fails because each atom we
remove used to cancel a scattered wave from some other atom, but now it
cannot. Thus we conclude that random fluctuations in all otherwise
homogeneous medium give rise to scattering. This scattering is incoherent:
because the fluctuations are random by construction, on average there can
be no phase relation between them.

The second way of obtaining scattering from the hypothetical array would
be to make close to the distance between scatterers. Using x-rays with a
typical of 1.54, the pairing-off argument fails at specific incident angles
where the PLD between beams reflected from adjacent planes are
separated by integral multiples of , so the scattering from different atoms
will be in phase, i.e., coherent. This process is called Bragg diffraction (see
Fig. 8.3). In general, there will be coherent scattering when there is spatial
correlation between scattering objects on a distance comparable to . The
more spatial correlation between scatterers, the sharper the scattered peak
8
(scattering as a function of the scattering angle , see Fig. 8.3a) will be.

4
From Fig. 8.3b, one may see that the condition for Bragg diffraction is
(8.2.2)

Central to Bragg diffraction is the scattering vector, q, defined by

(8.2.3)

(8.2.4) (Elastic scattering ki = ks)

It can be shown that q is always perpendicular to the scattering planes that


cause the Bragg diffraction along ks, and that q = 2/D, where D is the inter-
planar separation.
!
ks
/2

!
ki
/2

9
Fig. 8.3 Fig. 8.4

Basis Scattering by an Isolated Small Molecule


(Rayleigh Scattering)
An incident light wave Ei will induce a dipole moment in an atom or
molecule, where

(8.3.1)

An oscillating dipole involves an accelerating charge, and will therefore


radiate an oscillating electric field that we will call the scattered E field Es.
The magnitude of this re-radiated wave depends on the direction of
observation (through the angle of detection, , as measured from Ei), the
distance from the dipole (r), the electron charge (e), and the acceleration
of the charge in the dipole (a):

(8.3.2)

10

5
Basis Scattering by an Isolated Small Molecule
(Rayleigh Scattering)
This shows that |Es* Es| ~ r- -2. Notice also that /e is the
displacement a charge e undertakes from the equilibrium position in
producing a dipole moment of . So we can write:

(8.3.3)

(8.3.4)

11

Suppose the incident wave travels along x and is polarized along z. As


before, define to be the angle of detection measured from the polarization
direction of the incident wave or Ei. There is no angular dependence in the
x-y plane since sin = 1 along any direction in that plane. Along z, there is
no scattered wave since sin = 0 in this direction. Based on these, we
consider two cases.
Case 1: If the photodetector is in the horizontal (x-y) plane and is facing the
scatterer. At the same time, the incident Ei is polarized vertically (// z), we
expect

(8.3.5)

12
Fig. 8.5a Fig. 8.5b

6
Case 2: If the incident wave is unpolarized. We can view it as having
equal parts of vertically (// z) and horizontally (// y) polarized light.
Suppose again that the detector is in the x-y plane and is the
scattering angle as measured from the x direction (i.e., the propagation
direction of the EM wave). For the vertically polarized component (// z),
the scattered wave is the same for all . For the horizontally polarized
component (// y), is measured from the y-axis (see lower panel of
Fig. 8.5a). Clearly, sin can be written as sin(/2) = cos . So the
scattered field from the horizontally polarized part varies as cos , and
the scattered intensity as cos2. As a result, the total scattered
intensity varies as (1+cos2). Inserting this in (8.3.4), we get

(8.3.6)

13

14

7
15

Scattering from a Dilute Polymer Solution


Usually, the incident beam is smaller than the specimen. Its passage
through the specimen defines the illuminated or scattering volume.
Suppose we divide this volume into many cells of volume (n.b. this
volume is the same as the parameter in (8.1.7) if we consider each
cell as a scatterer particle) with the following properties:
1. 1/3 << , so each cell is effectively a point scatterer.
2. Each cell contains many monomers, with a concentration c that may
fluctuate statistically.
3. The cells are statistically independent, i.e., the fluctuations in one cell
is uncorrelated with that in other cells.
These assumptions require that Rg is smaller than 1/3. Each cell have
an instantaneous polarizability, containing an average value and a
fluctuation term:

(8.4.1)

= <()2>. (used <> = 0) (8.4.2)


16

8
Scattering from a Dilute Polymer Solution
Because Is/Io ~ <2> by (8.3.4), the scattering is determined entirely by
the mean-square fluctuations in polarizability,<()2>. Consider:

(8.4.3)

Assume that the scattering from the p and T fluctuations is the same in
the neat solvent as in the dilute solution. In the following discussions, we
will only consider the excess scattered intensity, Iex = Issolution - Issolvent. One
may perceive that only the fluctuation in c would affect Iex. Substituting
(8.4.2) in (8.3.5) gives Iex due to Rayleigh scattering from one cell. Since
there are 1/ number of cells per unit volume, the excess scattering due
to a unit-volume of specimen is:

per unit volume (8.4.4)

where Io is the incident intensity. To find /c, consider the Lorentz-Lorentz


equation (8.1.7):
(8.4.5) and (8.4.6), resp.
17

The key part is n/c, the refractive index increment (we drop the
reminder about constant T & P from now on), which can be measured or
looked up from tables. To find <(c)2>, write P(c) as the probability of a
given fluctuation c. So,

(8.4.7)

Since it should be equally probable to have positive and negative


fluctuations, P(c) is symmetric about c = 0. Moreover, P(c) should be
related to the fluctuation in the free energy, G:

(8.4.8)

Expand G as a Taylor series up to the second order term:

Substitute this expansion in (8.4.8) and use the resulting expression of


P(c) to evaluate <(c)2> by (8.4.7), one finds that 18

9
(8.4.9)

Sub. (8.4.6) & (8.4.9) in (8.4.4), we have

(8.4.11)

Recall that the condition for miscibility of a polymer solution is 2G/c2 > 0.
This can now be perceived from the Taylor expansion of G and the fact that
<c> = 0: For the solution to be stable, the energy change due to fluctuation
in c must be positive, so G > 0 2G/c2 > 0. With 2G/c2 > 0, the
quantity on the RHS of (8.4.11) is positive as required. Importantly, as the
spinodal is approached, 2G/c2 approaches zero. (8.4.11) implies that Iex
can get very large. Next, we relate 2G/c2 to a virial expansion. Suppose
that there are n1 moles of solvent and n2 moles of polymer in a solution that
has a volume of V (which we shall later set equal to the cell volume ):
(8.4.12a)

dV = 0 (8.4.12b)
19

Also, (M = molar mass of polymer) (8.4.13b)

At constant T & P, (8.4.14)

Subs. (8.4.13b) in (8.4.14): (8.4.15a)

(8.4.15b)
and

(8.4.16)
n1d1 + n2d2 = 0 at fixed T & P

So, (8.4.17)

But, (7.4.2)

So, (8.4.18)
20

10
and
(8.4.19)

In the above (from (8.4.12) to (8.4.19)), the choice of V is arbitrary so long


as it fulfills the three conditions listed on p.16. By choosing V = and
substituting (8.4.19) in (8.4.11), we get:

(8.4.20)

Introduce the Rayleigh ratio: (8.4.21)

and an optical factor: (8.4.22)

So, (8.4.23) 21

(8.4.24a)

or (8.4.24b)

Note:
1. According to (8.4.24b), in the dilute solution limit c 0, Iex ~ R cM
which is proportional to (number of scatterers) x (scattering power of a
scatterer), as stated in (8.2.1).
2. In the dilute solution limit, light scattering measures M, and so does the
osmotic pressure. The two measurements are actually related: Light
scattering is determined by c, and their amplitude is related to the
associated osmotic cost (= [ (c + c) 0 ] / V ). One can imagine
semipermeable membranes around each of the fictitious cells. Any
fluctuations in a cells concentration will cause to change in such a way
that if c > 0, solvent in the neighboring cells wants to migrate there, and
vice versa. Essentially, thermal energy drives random fluctuations in c,
but the osmotic compressibility resists them.

22

11
3. The virial coefficient is obtainable from the
c dependence of (8.4.24b). Evidently, in a
good solvent where B > 0, the intensity will

Iex, a.u.
grow sublinearly when c is increased
sufficiently to make the second term of
(8.4.24b) significant. Conversely, for a bad
solvent with B < 0, the intensity will grow
faster than linear at large enough c. (see
c, a.u.
Fig. 8.6.)
4. There is an important difference between Fig. 8.6
light scattering and osmotic pressure
measurement. Consider the polymer to be a collection of i-mers. So,

(8.4.25)

So, (8.4.26)

But osmotic pressure measures Mn. 23

24

12
Kc/R, mol/g

25
c, g/mL
Fig. 8.7

Form Factor of a Homopolymer Chain


Here, we consider the finite size of a polymer. As discussed above, if the
distance, rjk, between two monomers j and k on a chain is not << , then
the waves scattered from each monomer will have a phase difference,
leading to variation in the scattered intensity with . Compared to a point
scatterer, an extended scatterer with the same content can incur some
destructive interference and hence a net reduction in the scattered
intensity. Operationally, it is convenient to define a form factor for a single
polymer, P(), as:
(8.5.1)

where the (point-scatterer) Rayleigh scattering is given by (8.4.24) and 0


P() 1. Consider each (identical) monomer j to be a Rayleigh scatterer,
incurring scattered field Eo,s and a phase j in the scattered wave. The
electric field scattered by the chain, Es,tot, is given by the superposition of
the fields scattered by each monomer:

(8.5.2)
26

13
Form Factor of a Homopolymer Chain

(8.5.3)

For Rayleigh scattering, j = k. So,

(8.5.4)

The average in (8.5.4) is a time average or ensemble average.

From Fig. 8* on p. 6, one sees that

(8.5.5)
27

So,
n.b. q = ks ki, and q =
(8.5.6)
2kisin(/2), where is the
scattering angle, equal to
the angle measured from
with (8.5.7) ki to ks.

ki ks
where P(rjk) is the probability of monomers j and k

being separated by a vector rjk. For a chain in a
theta solvent, this function is Gaussian given by
(6.7.1). Assume that the solution is isotropic, which applies in most
cases. Then it is convenient to work in spherical coordinates (r,,)
because P(rjk) has no or dependence. Observe that

(8.5.8)

(8.5.9)

28

14
Then we can write

By looking
up tables
of integral:

Since , we have

! ! q2 q4
< exp[iq rjk ] >= 1 < rjk2 > + < rjk4 > ...
6 120

Substitute this in (8.5.6), we have:

(8.5.13)
29

But (6.5.8) (8.5.14)

So, independent of the shape of the particle. (8.5.15)

If the experimental condition is as such that q2Rg2 < 1, we may neglect


the higher order terms on the RHS of (8.5.15), and Rg can be
determined from a simple plot of the scattering result as discussed
shortly below.

30

15
Zimm Equation

We had derived (8.4.24a) for the (Rayleigh)


scattering of a dilute solution containing point scatterers. Here, we
attempt to modify it for the scattering of a dilute solution containing
extended, non-overlapping scatterers. The form factor, P(q) (or P())
we just derived provides the adjustment factor for the scattering power
of individual scatterers in going from being point-like to being
extended. As noted just beneath (8.4.24b), the scattering power of an
individual point-like scatterer is ~M. In a crude approach, we modify
(8.4.24a) by replacing M with P(q)M in (8.5.16) as shown below:

(8.5.16)

(8.5.17)
In the c 0 limit,

(8.5.18) Used for


So,
making Zimm
plots 31

Zimm Plot
Recast (8.5.16) in the form of (8.4.24b). We get:

(8.5.19)

This is the same as the equation Zimm derived by using a more


rigorous approach.
One common analysis of light-scattering data obtained from a
polymer solution is to make Zimm plots based on (8.5.18), as copied
below:

(8.5.18; notice that q2 = 4ki2sin2(/2))

Zimm plots are plots of Kc/R vs. sin2(/2) + c, where is a constant


(chosen as discussed below). In practice, the data are obtained by
measuring R as a function of at several fixed concentrations, c. To
gain insight about how Zimm plots work, perceive the data as two series
of plots: One obtained by varying at several fixed c; another by varying
32
c at several fixed . (8.5.18) suggests that

16
Zimm Plot
both series should converge to the same
y-intercept of 1/Mw when q 0 and c c 0 extrapolations,
0. For the first series, the slope should slope = (4k2Rg2)/(3Mw)
be the same and ~Rg . For the second
2

series, the common slope is 2B. The

Kc/R
choice of is to space out the curves in
the first series, which are separated by
c, where c is the incremental amount
of c employed in the experiment.
Accordingly, suitable choices of are 0
~1/c (with which the curves are extrapolations,
slope =
separated horizontally by amounts of 2B/3000
~1.) To analyze the plots, one first fits
the data, as functions of sin2(/2) at sin2(/2) + 3000c
various fixed c, to straight lines and find Fig. 8.8 Zimm plot for a sample of
the = 0 intercepts from the fits for each methyl cellulose in water.
c. Next, one fits the data as functions
of c at various to straight lines and find the c 0 intercepts for each .
The two procedures result in two sets of (extrapolated) points as shown in33
Fig. 8.8 by solid circles. These points would from two lines that intersect

Zimm Plot
the Kc/R axis at the same point. The value of that intercept is 1/Mw. It
can be shown that the slopes of the c = constant parallel lines are
(2kRg)2/(3Mw) and those of the = constant parallel lines are 2B/.

34

17
Example 8.3 on Zimm Plot

35

Example 8.3 on Zimm Plot

Fig. 8.9
36

18
Example 8.3 on Zimm Plot
For the c = 0 data in Fig. 8.9c, one finds that
c=0

=0
For the = 0 data in Fig. 8.9c, one finds that

Fig. 8.9 (contd)

Both data give 1/Mw = 3.03e-6 or Mw = 330 kg/mol. 37

Scattering Regimes
Based on (8.5.13), i.e.,

we can delineate four general regimes of behavior, depending on the value of qRg

1. If qRg << 1, meaning small molecules or small scattering angles, then


P(q) 1. This is the Rayleigh regime and eqns (8.4.24a,b) apply.
2. If qRg 1, then only the next term in the expansion matters: P(q) 1
(q2/3)Rg2. In this regime, one can obtain Rg without knowing the shape of
the molecule. Alternatively, P(q) exp((q2Rg2)/3), and a plot of ln(Iex) vs.
q2 should be linear with slope -(Rg2)/3. This latter plot is termed a Guinier
plot and this regime is the Guinier regime.
3. If 1 qRg 10, more terms in the expansion of P(q) becomes important,
and these depend on the specific shape of the molecule. So, in this
regime the mathematical form of P(q) could help distinguish different
molecular shapes. We shall discuss some specific functional forms
below.
4. If qRg >> 1, the scattering is dominated by the internal structure of the
molecule, and one can extract no information about Rg or the shape of
the molecule.
Physically, q-1 is essentially a ruler in the sample. The choice of q dictates
38
that the scattering experiment will explore fluctuations on the length scale q-1.

19
39

40

20
Form Factor of Several Shapes
Here, we state the form factor for three particularly important shapes: the Gaussian
coil, the rigid rod, and the hard sphere.

1. Gaussian coil: The form factor is known as the Debye function.

(8.6.1)

This function applies to chains in a theta solvent and in a melt. Note that it
doesnt apply when q-1 is small and comparable to the persistence length or
statistical segment length.

2. For a rigid rod of length L and zero width, the form factor is:

(8.6.2)

3. For a hard sphere of radius R, the form factor is:

(8.6.3) 41

Form Factor of Several Shapes

Fig. 8.11 Form factors P(q) for


Gaussian coils, hard spheres, and
very thin rods (a) as a funciton of
(qRg)2. (b) Inverse form factors as in
(a), showing the convergence to a
common slope of 1/3 in the small
qRg limit. (c) Form factors plotted on
a logarithmic scale, showing the
42
oscillations for the hard sphere.

21
Light Scattering Experiments
One common application of light scattering is in dynamic light scattering
(DLS) (to be discussed in the next chapter), which setup can also be
used to measure I(q). Another common application is in size-exclusion
chromatography as an absolute Mw detector, which is in routine use
nowadays.
Fig. 8.12 shows the schematic of a light scattering setup. The index
matching fluid (e.g. silicon oil and toluene whose refractive indices are
close to that of glass) serves to reduce reflection of the incident light from
the sample, which is enclosed in a glass container. The index matching
bath also provides an effective medium for controlling the sample
temperature.

Fig. 8.12 43

Light Scattering Experiments


The scattering volume is determined by the intersection of the incident
beam and the collected beam (see Fig. 8.13). Assuming that both beams
are comparable in width, one can see that as deviates from 90o, the
volume of intersection increases by a factor of 1/sin. So the signal
should be multiplied by sin to correct for this variation.

Fig. 8.13
44

22
Samples and Solutions

Two main issues: (1) Choice of solvent (2) Need to prepare dust-free
samples.

For issue (1), one would like to have |n/c| as large as possible. It
may also be helpful to choose a solvent with a relatively small R of its
own, so that the polymer contribution to the excess scattering is
larger. (For pure solvent, scattering comes from density fluctuations,
which are determined by the isothermal compressibility of the solvent,
; the intensity is ~kT.) For issue (2), some solvents are easier to
make dust-free than others; for example, more polar solvents such as
water and THF are often trickier to clean than toluene or cyclohexane.
It is essential to remove dust, as stray particles that are significantly
larger than the polymer molecules will scatter strongly.

45

Samples and Solutions


The two standard options are filtration and centrifugation, and the
former is usually preferred. Both are less than ideal, in that they may
change the concentration of the solution. There are two standard
diagnostics for the presence of dust in the sample. The first is to
examine the temporal fluctuations in the signal Ss (~Is). These should
be random, and have a root-mean-square amplitude close to
(<Ss>)1/2. If some dust is present, the signal is likely to increase
suddenly and then decrease suddenly some seconds later, as dust
particles drift in and out of the scattering volume. The second
diagnostic is to examine l0/I() versus sin2(/2) (or q2). According to
the Zimm equation this should produce a straight line. Dust will
increase I() selectively at low .

46

23

Vous aimerez peut-être aussi