144

CH9700079

SMALL ANGLE SCATTERING AND POLYMERS

J.P. COTTON
Laboratoire Leon Brillouin (CEA CNRS), CEA Saclay
F91191-Gif/Yvette France

ABSTRACT
The determination of polymer structure is a problem of interest for both statistical physics
and industrial applications. The average polymer structure is defined. Then, it is shown
why small angle scattering, associated with isotopic substitution, is very well suited to the
measurement of the chain conformation. The corresponding example is the old, but
pedagogic, measurement of the chain form factor in the polymer melt. The powerful
contrast variation method is illustrated by a recent determination of the concentration
profile of a polymer interface.

1. Introduction

Using a wavelength X of about 10A, small angle neutron scattering (SANS) is well
suited to the dimensions of polymers. Thus it is an excellent technique allowing [1] to
determine the conformation, i.e. the average structure, of polymers in different
environments. The parameters obtained are the molecular weight, the radius of gyration
and the persistence length of the polymer and more generally its form. They are obtained
from polymer solutions using similar procedures as those of light or X-ray techniques [2-
5]. But neutron is the scattering radiation that allows a labelling method keeping
unchanged the chemical properties of the labelled moieties: the isotopic substitution.
As a peculiar case, SANS is the only technique allowing to determine the
conformation of one chain inside different polymeric materials by using deuteration as a
labelling method. This is an important result since plastic material is essential element of
human life. The possibility to vary the coherent scattering lengths of the solvent
simplifies the determination of the partial structure functions of complex solutions. These
possibilities are at the origin of the success of SANS in the polymer field. The aim of this
lecture is to show how.
 145

2. Basis of the polymer conformation [6-8]

Fig. 1 - Schematic representation of polymer conformations on a lattice of step L Fig. la

represents a random walk, Fig. lb a self avoiding walk or a polymer with excluded
volume.

A polymer is a linear succession of N, chemically bound, identical units each of

length £. The easiest model to describe the polymer conformation is a random walk e.g.
there is no correlation between two successive monomers (see Fig.la). As a
consequence, the mean quadratic distance R between the polymer ends is:
R2 = N^2 = 6Rg2 (1)
where Rg is the radius of gyration of the chain (the inertial moment of the polymer
divided by its mass). Typical values of these parameters are: N ~ 5000 that corresponds
to a molecular weight M ~ 500000 and £~ 3 A. The size of a solvent molecule is similar
to £ . The length scale is given by £ and by the global size of the chain Rg ~
(5000* 10/6) 1/2~100A.
The number of configurations that a chain can adopt is zN. z is the number of
neighbours of each monomer extremity; z = 3 in the square lattice of Fig. 1 or for a
carbon-carbon bond. This number is usually greater than io1000. Then the polymer
entropy is very high and originates the basic random coil behaviour of the chain. The
conformation of polymers, in a given environment (a solution, a network, an interface, a
flow...), is submitted to the competition of the constraint of this environment and of the
entropy which strongly leads the polymer to adopt a random coil conformation. For a
real chain, the constraint is the impossibility o have two monomers far along the chain at
the same position whereas it was the case for an ideal chain. This steric phenomenon,
called the excluded volume effect, leads to an important swelling of the chain that clearly
appears in dilute polymer solutions (see Fig. lb). It occurs in good solvents where the
pair interaction between monomers is repulsive. Nevertheless, it is possible to find [6] a
temperature 0 corresponding to a compensation between interactions monomer-
monomer and monomer-solvent. At this Flory temperature, the ideal chain conformation
is observed. Thus the random walk model is not so far from reality.
The polymer conformation is well defined if the conditional probability gi(r) of
having two repetitive units at a distance r is known. It is convenient to write gi(r) with
the formalism of Van Hove [9], since this formalism is well suited to show how to
extract the polymer conformation from a scattering experiment. Following Van Hove,
the pair correlation function gi(r) is expressed as a function of the density p(r) of the
chain repetitive units of which positions are r, ( i = 1 ... N).
-ri) gl (r) = <p(o)p(r)> (2)

where 5(r) is the distribution function of Dirac, J 6(r)d 3 r = 1.

 146

The determination of gi(r) is a non trivial statistical problem which needs the
dichotomy between theoretical predictions and experimental measurements. Note, in
particular, that the second moment of gi(r) is proportional to Rg2,
<(IW >=
^"wlt ^ ^ ? <Sj2> (3)
where S; is the monomer position relatively to the centre of gravity G of the chain (see
Fig. 2), Z

Fig. 2 - Representation of a polymer where the statistical length t involves a few number
of monomers. G is the centre of gravity of the chain and Rg its radius of gyration.

In the intermediate range, (. « r « Rg, the variation of gi(r) gives information

concerning the form of the polymer. For instance, gi(r) ~ r"1 characterizes a gaussian coil
(a random walk), gi(r) ~ r"473 corresponds to a self avoiding walk whereas gi(r) ~ r'2
labels a rigid rod (de Gennes [7] p.37).

3. The SANS cross section and the contrast notion

Let us consider a solution of n polymers each having N monomers. Since the

domain of distances of interest was shown lying in between 3 and 500A, the scattering
vector range of Q needed for a SANS experiment varies from 0.1 to 10'3 A"1. At this
small angle scale, the structure factors of the solution molecules are reduced to unity.
Thus the monomers and the solvent molecules can be considered as scattering
points of which coherent scattering length bj is the sum of the coherent scattering
lengths of their atoms. Thus the intensity scattered by the solution writes.
I(Q) = 2 > i b ,< exp( iQOvrj)) > = I b . b ^ C Q ) (4)

where the scattering functions Sy(Q) are the partial structure factors. For the polymer
solution, I(Q) is:
I(Q) = bpXCQ) + 2bpbsSps(Q) + bs'SssCQ) (5)
where the indexes p and s characterize the monomers and the solvent molecules
respectively. Eq. 4 and 5 are general and valid even for wide Q values. Nevertheless it is
difficult to extract the polymer response, Spp(Q), from the measurement of I(Q).
Fortunately in the Q range of SANS, the incompressibility of the sample is a constraint
which simplifies [10] these equations by introducing the contrast notion.

3.1. Isothermal compressibility and the contrast notion

Let us consider first a system of identical scatterers, a solvent or a polymer melt for
instance, in a volume V. The average number of scatterers (a solvent molecule or a
monomer) is <n>, each has a partial volume vs = V/<n> and a scattering length bs. A
 147

general thermodynamical expression [11] relates the extrapolaton of I(Q) to zero Q value
to the isothermal compressibility XT :
1(0) = bs2S(0) = bs2(<n>/v,)kTxT (6)
Where k is the Boltzmann constant and T the temperature. In practice, the
thermodynamical limit is already reached [10] for Q values below Q*,
I(Q) = I(0) ; Q<Q* (7)
The value of Q* is rather large, about 0.6 A"1. Note that the Q range of interest for
polymer studies (Q^<1) lies also below Q*. Moreover, far from a critical temperature,
these systems are considered as incompressible (XT ~ 0) since the density fluctuations are
negligible relatively to the concentration fluctuations. Thus the intensity scattered by a
melt of polymers of one species is quasi-zero, in the range Q£ < 1. It is only due to the
very weak density fluctuations of monomers. It is the origin of the transparency of
materials made with amorphous (flexible) polymers such as polystyrene or Plexiglas
(polymethylmetacrylate). The incompressibility hypothesis is the basic point of the
notion of contrast in small angle scattering experiments.
Let us consider, now, a mixture of scatterers of a = l...m different chemical
species. The parameters are the mean number <na>, v a the partial molecular volume and
the scattering length b a . Here the sample scatters the following intensity:
(8)

The extension of the thermodynamical limit [11] writes:

; Q<Q* (9)

The combination of Eq. 8 and 9 allows to eliminate one component, m = 1 for instance.
Then, using XT = 0, Eq.8 becomes:
I
( Q ) = Z k a k p S a p ( Q ) . k a = b«-b 1 (v a /v 1 ) Q<Q* (10)

where ka is the contrast length of the species a relatively to the species 1. In the
following the condition, Q < Q*, of the incompressibility will be considered as fulfilled.
Applying this result to the polymer solution, the scattering intensity writes:
I(Q) = kp2Spp(Q) i kp-bp-b^vp/v.) (11)
It gives directly the monomer-monomer scattering function, the solvent molecules having
not to be taken into account.
Let us now give a few numerical examples of the kp value for polystyrene, (CgHgV,
in different solvents as the carbon disulphide, CS2, or the benzene, C ^ , deuterated or
not. The usual rules for the calculation of kp are: the partial molecular volume of a
molecule is deduced from the bulk density of the species; the isotopic substitution does
not change the volume of the labelled molecule. Table I gives these values that the reader
may verify using the more recent table of coherent scattering lengths [12].
Table I - Values of the contrast length for H and D polystyrene chains in different solvents
kp (10*"cm) CS2 C6H« C6D6
- CgHg - 0.319 0.396 -6.518
- C8D8 - 8.647 8.724 1.810

In Table I, the striking result derives from the intensity scattered by H chains in CS2 that
is proportional to 0.1 (k p 2 ). Whereas the same polymer, but deuterated, in the same
solvent scatters an intensity 700 times greater! This example shows why isotopic
 148

substitution is an excellent method of labelling since it keeps unchanged the electronic

shells of atoms and thus the chemical properties of the labelled molecules.
With polymers, using deuteration of the chains or of the solvent molecules, it is
possible to improve the contrast of SANS and to study the conformation of chains in
rather dilute solutions (10*3 gem'3). This conformation can also be determined as a
function of the concentration of the solution by using mixtures of H and D polymers
[13]. This until the polymer melt as described further.

3.2. The form factor and the distinct scattering function

Coming back to the polymer solution of volume V, Eq. 11 gives:
I(Q) = kp2ST(Q) = kp2 S Z < exp(iQ( i f - r? ))> (12)
<*,P i.j

where i"ja is the position of the i* monomer of the chain a (a = l,...,n), each with N
monomers. The total scattering fUnction ST(Q) can be split in an intra-polymer scattering
function Si(Q), the form factor, and an inter-polymer scattering function S2(Q):
= nS1(Q) + n2S2(Q)

Si(Q) is the Fourier transform of the pair correlation function gi(r) which
characterizes the polymer conformation as discussed in section 2. Even in the simple case
of a binary system, Si(Q) is not easily obtained: the experiment has to separate the
functions Si(Q) and S2(Q) in Eq. 13. Such a procedure has been early developed by
Zimm [14] for light scattering experiment. The determinations of the molecular weight
and of the radius of gyration of polymers in dilute solution are usually obtained from
extrapolations to zero concentration of data obtained with light, X-ray or neutron
experiments [1].

3.3. Conformation in the polymer melt: The 50/50 theorem.

Fig. 3 - A melt of polymer is a liquid-like structure.

A melt of polymers is a liquid like structure of monomers (see Fig. 3) of density

In this case, there is no concentration fluctuations and very weak density
fluctuations. The system, as already discussed in section 3, can be considered as
incompressible and does not scatter:

^ « 0 ; Q<Q' (14)
Therefore there is a compensation between the intra- and the inter-chain
fluctuations of monomer positions that is written from Eq. 13 and 14 :
 149

Si(Q,nm) = -n m S 2 (Q,n ni ) (15)

Thus a scattering experiment by a polymer melt cannot give the chain conformation.
Nevertheless, this problem can be solved using a labelling method (see Fig. 4)
which creates, artificially, concentration fluctuations due to the contrast of the polymer
with itself. For photon scattering experiments, as photons interact with the electronic
shells of atoms, the labelling of a polymer needs to change the atoms and therefore the
chemical species of monomers. But a mixture of polymers of different chemical species
leads, generally, to a phase separation since the threshold of miscibility of two polymers
varies as the inverse of the molecular weight or N (see ref [7] p. 106). Consequently to
label a polymer is very difficult with light or X-ray techniques. Whereas the neutrons,
which essentially interact [15,16] with the nuclei of atoms, allow to label the atoms using
the isotopic substitution which keeps unchanged the electronic shells. Moreover the
substitution proton (H) deuteron (D), rather easy to obtain, gives one of the highest
values of contrast [12]. These are, also, fundamental reasons of the use of small angle
scattering (SANS) in the polymer field.

Fig.4 - Schematic representation of a polymer melt with and without labelled chains.

Let us write the intensity of a solution of xn labelled polymers (noted D) of

contrast length ko and (l-x)n unlabelled ones (H, kn), both having the same number of
monomers. Assuming a totally random spatial distribution of labelled and unlabelled
polymers, Eq. 10 becomes:
I(Q) = kD2SDD(Q) + 2kD kHSDH(Q) + kH2SHH(Q) (16a)
where SDD(Q) = xnSi(Q,n) + xVs 2 (Q,n)
SDH(Q) = x(l-x)n2S2(Q,n)
2 2
SHH(Q) = (l-x)nS,(Q,n) + (l-x) n S2(Q,n) (16b)
Finally I(Q) is given by:
I(Q) =<k2 (x)>nS,(Q,n) + <k(x)>2 n2S2(Q,n) (17)
2 2
where <k (x)>= xkD + (l-x)kH and <k(x)> = xkD +(l-x)k H (18)
The total scattering function Sr(Q,n) can be introduced in Eq. 17; this leads to use the
mean square deviation of k(x):
<k2 (x)> - <k(x)>2 = (kD-kH)2 = (bD-bH)2
which only depends of the scattering lengths of the scatterers. Eq. 17 becomes:
I(Q) = x(l-x)(bD-bH)2 nS,(Q,n) + <k(x)>2 ST(Q,n) (19)
In the melt, using the incompressibility, Sr(Q,nm) ~0, Eq. 14, Eq. 19 becomes :
I(Q) = ( b D - bH f x(l - x)n. Si(Q) (20)
This result, early predicted by des Cloizeaux [17], shows that the scattering
intensity gives directly Si(Q) without any extrapolation of the concentration x of labelled
polymers. It is often called the 50-50 theorem since the scattering intensity is maximum
for x = 0.50. It has been experimentally verified [18] with amorphous polymers. For
these polymers, the quenched melt, sometimes, presents some voids of which contrast is
greater with D polymers than with H polymers. In order to minimize the spurious central
 150

scattering of such voids, it is often better [18b] to use a concentration of deuterated

chains smaller, x~0.20, than the optimum one (0.50). On the contrary, with liquid
crystalline polymers [19], probably due to their liquid character, such voids are not
important and all the experiments made used samples with one half of partially D
polymers.

4. From the form factor to the chain conformation

In the previous section, it is shown that SANS by a melt of H and D polymers

gives an intensity directly proportional to the form factor of one chain. Now, we want to
discuss how the chain conformation can be determined from this form factor.
The intensity i(Q), per unit volume of sample, writes (see Eq. 12 and 20):
KQ) = I(Q)/V =(A/N)S,(Q) = ANP(Q) (21)
2
where A = kp p for a polymer in solution and p is the monomer concentration (p=nN/V)
or A= (bo - bH ) 2 x(l-x)p m for a polymer melt containing x percent of deuterated chains;
pm is the number of monomers per unit volume. P(Q) is the form factor (normalized at
unity, P(0) = 1) which writes :

P(Q) = N- 2 2(e i Q ( r | " r > ) ) (22)

As a general principle, it can roughly be said that in a scattering experiment Q acts
as the magnification of a microscope. Depending on the Q values, are explored the scales
from the radius of gyration Rg of the chain to that of the statistical unit I.
For small Q values, QRg< 1, the Guinier range, the SANS experiment allows to
determine Rg or the global chain size. The form of the polymer is obtained in the
intermediate Q range, (4Rg)'1 < Q < f\ where the correlations between monomers
intervene. In this range the form of the polymer is deduced [7] from the characteristic
exponent a of the Q"0 dependence of P(Q), as it is developed, in the following, in the
cases of oriented and unoriented polymers.

4.1. Case of unoriented samples

If the polymer can adopt all the orientations in space as it is the case for isotropic
particles or for unoriented samples, the orientation average leads to write P(Q) as [2]:

(23)
I.J

where ry is the module of the vector rj-rj.

At small Q values, the expansion of P(Q) gives [2]:
i(Q) * AN( 1 - Q2Rg2/3 ) ; QRg< < ! (24)
where Rg is the radius of gyration of the polymer, already defined (see Eq. 3), which
gives its global size. The experimental value of Rg is deduced from the slope of the
straight line i"!(Q) versus Q2 in the so called Zimm representation [14]:
A 1.. _ , _ (25)
N *
This representation was adopted for polymers because it is valid until QRg ~ 1 and
greater if the polymers have a little polydispersity (see ref. 1 p. 189). From the linear
extrapolation to zero Q value, the number N of monomers (or the chain molecular
weight M = Nm, m is the molecular weight of the repetitive unit) can be measured. Such
 151

a determination of N needs absolute SANS measurements [10, 20] which are, always,
very useful.
The whole form factor allows to determine the chain conformation. The
procedure consists to fit the experimental data with models of P(Q). The more usual are
the following [1,2,21]:
The ideal chain: It is the basic conformation of linear polymers, that of a random walk of
N steps of length £. Its form factor is the Debye function [22]:

P(Q) = ™ ( e " X - 1 + X) ; X = Q2R2 = Q 2 ^ - (26)

In the intermediate Q range, (4Rg)''< Q < t \ the Q'2 variation of P(Q), i(Q) = \2AJQ2£2,
is a well-known characteristic of Gaussian polymers.
The self avoiding walk: There is no analytic function of the form factor of the chain with
excluded volume effect. Nevertheless the variation of Rg as a function of N and the
intermediate behaviour of P(Q) are known [7,8]:
Rg 2 ~N 2v ; P(Q)~Q"1/V (27)
The value of v is 0.588. The v change from 0.500 to 0.588 characterizes the swelling of
the chain due to the excluded volume effects.
The rod of length L , without thickness, has the following form factor [1,2]

P(Q) =l TJ ^ i i d u - ^2; X = ^ ; R 2 = ^ (28)

™ X o u X 2 * 12
In the intermediate Q range, 4QL>1, the Q 1 variation of P(Q), ~rc/QL, is another
signature of arigidrod.
The wormlike chain corresponds to a more realistic model of a polymer chain. It is a
Gaussian chain which evolves continuously to a rigid rod when decreasing the distance
below the persistence length Ip. Two parameters define the chain: its contour length L
and lp. The form factor of such a chain is not a simple function [23-25]. The simplest
expression is obtained from the amalgamation [26] of three functions, within an error
bar of 1.5%, thus much lower than the experimental ones [26]; it is valid only for
L> 101p. These functions are:

Q 2 l p LP 2 (Q) = 6 + 0.5470(Qlpp)2 -0.01569(Ql p ) 3 -0.002816(Ql p ) 4 ; 2< Qlp<4

2 , .« . . 2
;u=L/l p ;Ql p <2 (29)
x

Pi(Q) is the asymptotic form given by des Cloizeaux [23], P2(Q) is an approximation,
better than 4 10'3, of the tabulated data also given by des Cloizeaux in ref.23. Ps(Q) is the
approximation of Sharp and Bloomfield [24] for finite chains, valid only for small Q
values Qlp^2. The radius of gyration of this chain is [27]:

'••-l +^-"-(l-e-u)) (30)

When u=L/lp tends towards infinite, Rg2 tends to Llp/3, thus the statistical unit of the
wormlike chain is t - 21P.
 152

The sphere of radius R. This model can be useful to analysis the form factor of a
collapsed polymer.
= L ( s i n X - X c o s X ) 2 : X = QR; Rg2 = 3R2/5 (31)
In the intermediate range, if the polydispersity of the sphere sizes has a sufficient value to
level the oscillations of P(Q), P(Q) has a Q4 variation characteristic of the sharp
interfaces. This result is a particular case of the well known Porod law [28]:
i(Q) = 27C(k1/v,-k2/v2)2(S/V)Q-4 (32)
which allows to determine the area per unit volume S/V of a sharp interface between
media 1 and 2 .

4.2. Case of oriented samples

Let us consider the case of anisotropic polymers as stretched polymers or liquid
crystalline polymers. The preferential directions are those parallel and perpendicular to
the stretching direction or to the nematic director.
At small Q values, the expansion of the form factor P(QX), Eq. 22, in any direction
x, can be written:
P(QX)*1-Q2R2 ; Q2RX2«1 (33)
Rx is called [29] the inertial mean distance in the x direction and is defined by:

^ j
= jn(s)ds (34)
where n(s) is the number of monomers between two planes normal to this direction and
at distances x=s and x=s+ds of the centre of gravity G of the polymer. n(s) is also the
number of intersections of the polymer with such a plane, see Fig. 5. For an isotropic
model, the radius of gyration is given by:
^ * 2 2 = 3 R 2 (35)

Fig. 5 - The inertial quadratic mean distance Rx2 is the second moment of n(s) the number
of intersections of the polymer with the plane normal to the direction x at a distance s of
the centre of gravity G of the polymer. From ref. 19.

The whole form factor can also be used, here, to fit the data with a model in any
direction. A simple model is that of a cylinder (length 2H and radius R) of which form
factor is:
si"(QHcosa)2J 1 (QRsina) N | 2
QHcosa QRsina )
 153

where a is the angle between the cylinder axis and the scattering vector Q and Ji(x) the
Bessel function of the first order. The inertial mean distance in the a direction is:
4H2 cos2 a R 2 sin 2 a
K
a - Q(2H or R ) « 1 (37)
12 4
Other form factors of simple models are given in the book of Feigin and Svergun
(ref. 21 p. 93). The effects of the length distribution of the chain, the polydispersity, are
discussed elsewhere [ 1,19,30].

5. The polymer conformation in the melt

Before the SANS experiments, the conformation of polymers in their melt was
unknown. As the chains are rather globular in dilute solution, somebodies thought that
increasing the concentration has for effect to condense the chains on themselves, each
keeping its individuality. The supporters of the globular form underevaluated the
viscoelastic properties of polymers wich is an indication of the existence of
entanglements. The latter was taken into account by a Flory's prediction [6] assuming an
ideal conformation of chains. That corresponds to interpenetrating chains. Others
thought that local alignments of polymers occured, see Fig. 6.

Fig. 6 - Before SANS, the three possible conformations of polymer chain in the melt: a
globular form, a Gaussian conformation (the chains are interpenetrated) and the existence
of partial chain alignments.

This problem was solved with the development of the SANS technique, at the
beginning of seventies, when the possibilities of the deuteration in this field were
achieved. The first experiments were made in Europe using only a small amount of D
polymers dispersed in a matrix of H polymers. The 50-50 theorem was not yet
recognized! The results obtained with chains of polymethylmethacrylate (Plexiglas) [31],
polyethylene [32] and polystyrene [17,33] were consistent. They will be discussed now in
the light of the polystyrene case [17], a standard of flexible amorphous polymer.
Fig. 7a shows the raw data. The upper curve is obtained with an H polystyrene
containing 2 10'2 of deuterated chains of similar molecular weight (Mw -21 000). The
lower curve is the intensity scattered by a test sample made with only H polymers; it has
not to give a coherent signal. Both set of data shown in Fig. 7 are typical of the SANS
technique.
The central scattering of the lower curve results from defects of the collimation of
the incident beam. Its flat part is rather high; it is due to the incoherent background of
 154

the numerous protons of the sample. Notice that this background is proportional to the
solid angle AQ defined by the detector of the spectrometer. In SANS, AQ = s/D2 (s is
the detector area and D the detector-sample distance), AQ has a very small value,
typically 3 10'5. Thus the background due to protons is not a real problem as it is claimed
in the case of wide angle neutron experiments.
With a low D monomer concentration p and a weak number of monomers
(N-200), the coherent intensity, proportional to pN, is particularly weak in this example.
Nevertheless the intensity resulting from the difference between the two curves in Fig. 7a
is tractable as it appears clearly in Fig. 7b.

20 20
1
Q lOOA Q lOOA1

Fig. 7a - Raw data of the intensity as a function of the scattering vector Q. The lower curve
is that scattered by a -melt of H polystyrene chains, it is the background. The upper curve
corresponds to a similar sample where 2 percent of D polystyrene are dispersed. The
molecular weight of D chains (21 000) is similar to that of the H chains (20 000).
Fig. 7b - Form factor of the D chain as a function of Q. The data results directly from the
difference between those of Fig. 7a. In spite of the particularly weak coherent scattering of
these samples, the intensity is rather good. The full line is a fit with the Debye function
(Eq. 26). The data of Fig. 7a and 7b are those of ref. 17.

5.1. Conformation deduced from the radius of gyration

Following Eq. 20, the curve of Fig. 7b is proportional to the form factor of the
short polystyrene chain. From the slope of the Zimm representation, F!(Q) versus Q2,the
value of the radius of gyration, Rg= 38A, is obtained. Here the chain is too short to study
the intermediate Q range since this domain, QRg > 4, corresponds to dimensions well
inside that I of the monomer (I = 18A [26,35]). Thus the correlations between
monomers cannot be reach in this way. It is sad since the value of the radius of gyration
gives only a number that is the global size but nothing about the form. Nevertheless the
indication concerning the form can be obtained from the variation of Rg as a function of
N (or the molecular weight M) but it needs more samples. If the polymer is globular, its
size has to vary as N1/3 (Eq. 31), this variation becomes N1/2 for an ideal chain (Eq. 26)
and N 1 for a rod (Eq.28).
The experiments [17] were made using samples with eight different molecular
weights lying in between 20 000 and 1 100 000. The Rg values obtained in the melt were
 155

compared to those obtained in a good solvent CS2 and in a 0 solvent (H polystyrene in

D cyclohexane at 30°C). The results are shown in Fig. 8. In the good solvent the
exponent 0.6 is in good agreement with that of 0.588 predicted by modern theories [7,8]
for a self avoiding walk. In the 0 solvent and in the melt, the polymer sizes are similar
and the slope of 0.5 is in very good agreement with a Gaussian conformation of polymers
in their melts.

20 -

10
10'

Fig. 8 - Log log plot of the radius of gyration of a polystyrene chain versus the molecular
weight. Different environments were used: a good solvent CS2 (x), a © solvent and in its
melt (©). The data obtained in the melt and in the © solvent are very similar with a slope
of 0.5. These data are those of ref. 17.
Fig. 9 - Log log plot of the inverse scattered intensity as a function of the scattering vector
Q. The data are obtained from polymers in different environments: in a good solvent CS 2
(A), a © solvent (B) and in the melt (C). The data obtained in the melt and in the ©
solvent have the same slope of value 2.0. These data are those of ref. 35.

5.2. Conformation obtained in the intermediate range

In the intermediate Q range, see section 4, the chain conformation is deduced from
the exponent a of the Q"* dependence of the form factor P(Q), thus with only one
scattering curve. If the chain is globular a = 4, for a Gaussian chain a = 2 and for a
rigid rod a = 1. As the Q range obtained with SANS is rather limited, it is useful to vary
also the length of the polymer as a function of the searched parameters. If this parameter
is the radius of gyration, its determination is easiest with short chains. Whereas the best
polymer allowing to reach the intermediate range is the longest one. Thus the experiment
in the intermediate range was made [35] with polystyrene chains of molecular weights
greater than 1.1 106. The result is shown in Fig. 9.
The intensities determined in the limit of the large Q range, QRg > 4, correspond to
the form factor since the interactions are negligible in this domain [1,36]. Thus in good
solvent the Q"167±006 variation corresponds to a self avoiding walk with v = 0.59±0.02
(see Eq. 27), in good agreement with the modern predictions [7,8]. The Q exponent was
[35] 2.06±0.08 in the 0 solvent and 2.01 ±0.08 in the melt.
Thus in the melt a polymer has a Gaussian conformation. This was also been
verified by fitting the whole form factor with the Debye function, Eq.26. It is realized,
Fig. 7b, for the smaller chain, because both Q ranges (QRg >1 and <1) are reached. The
absence of excluded volume effects for a chain in its melt means [7] that there is a
 156

compensation of interactions between monomers belonging to the same chain and those
between monomers belonging to distinct chains (having the same length).

6. Polymer interfaces and the contrast variation method

The stability of colloids is often obtained by coating the particles with polymers,
see Fig. 10. Thus the repulsive interaction between polymers avoids the particle
aggregation. It is of current interest to determine the structure of the polymer layer.
SANS is a good technique of which capacities, in this field, are far to be totally
explored.

Fig. 10 - Schematic representation of polymers (p) coating a particle (g) in a mixture of

labelled and unlabelled molecules of solvent.

The SANS intensity by unit volume i(Q) of such a system of volume V writes.
I(Q) = V i(Q) = k2p Spp(Q) + 2 kpkg Spg(Q) + k2g S^Q) (38)
In order to separate the partial structure factors, it is very useful to vary continuously the
scattering length of the solvent and thus the contrast length values. This method,
describes now, was early introduced by Stuhrmann [37].

6.1. The contrast variation method

The contrast variation method consists in using for the solvent a mixture containing
a ratio y (volume per volume) of labelled molecules of scattering length bso and (1-y) of
unlabelled molecules (bSH) Under these conditions and because the isotopic labelling
does not change the partial molar volume of the molecules, the mean scattering length
bs(y) of the solvent molecule and the contrast length of the scatterers of species a are
written.
bs(y) = ybSD + (l-y)b sH ; ka(y) = ba-bs(y).\J\i.
This method is very useful for organic solvents, since the substitution of protons
(bsH^ -0.374 10*12 cm) by deuterons (bsD= + 0.667 10'12 cm) leads to changes of the
corresponding contrast lengths which can reach a range in between 1 to 9. The
corresponding scattered intensities have variations from 1 to 80, thus of sufficient
amplitudes to be clearly separated even if the variation of y is limited from 0 to 1.
Two specific values of y, the solvent proportion of labelled molecules, are of
interest: yp, for which kp(yp)=O, the solvent matches the polymer, and yg, kg(yg)=0, which
kills the grain: a contrast matching experiment can give directly one of the partial
structure factors. In the opposite case, three measurements made with three samples,
each prepared with a different y value, allow to determine Sgg(Q), Sgp(Q) and Spp(Q). It
will be shown in the following, how this method, first introduced by the Bristol group
[38] and developed by Auvray [39], was used to determine the concentration profile of
polymers adsorbed or grafted on particles. Until these works, only the specular reflection
of neutrons or X-rays was recognized to reach this aim [40].
 157

6.2. Determination of a layer concentration profile [39]

Let us consider the colloidal solution represented in Fig. 10. The solid grain g has a
sharp interface and a large radius of curvature R. The thickness of the polymer layer is
h. The Q range of interest for the SANS experiment lies in the neighbouring of Qh~l in
order to analyse the polymer layer. Now the basic assumption is that R is much greater
than h, thus at the observed Q scale the solid surface can be considered as a plane. Under
this condition, i(Q) is proportional to the form factor of the colloid and writes:
i(Q) = k2p pS'pp (Q) + 2 kpkg pS'pg (Q) + k2g pS'gg (Q) (39)
were p is the number of colloids by unit volume of solution. The index 1 recalls that the
interactions between colloids are negligible in the Q range QR » 1 .
The solid-solid partial form factor is given by the Porod law (Eq. 32):
QR»l
v
g
The cross term, S ^ Q ) is proportional to the Fourier transform < pg(0)pp(r) >. It
is written [39a] by considering the specific symmetry of the system i.e. the invariance of
the structure of the polymer layer by translation along the surface and its isotropy in the
directions parallel to it. Thus the local monomer concentration p p (r) is given by:
Pp(r) = < Pp(z) > + 8pp(r) = c(z) + 5pp(r)
where c(z) is the mean concentration of the polymer layer at a distance z of the solid
surface (see Fig. 10) and the cross term is:

^Q- 3 Jdz.c(z)sinQz ; QR»1 (40)

v Pp vv gg V
V
V
o
It is of peculiar interest since it gives directly the sine-Fourier transform of the monomer
concentration profile c(z).
The monomer-monomer scattering function is separated in two terms:
2

v 7dz.c(z)e iQz PSWQ); QR»1 (41)

i PP

where pSpp(Q) corresponds to the intensity scattered by the pair correlation function of
concentration fluctuations inside the layers. The latter is predominant at large Q values
Qh»l.
The striking result is the possibility to determine c(z) from the cross term by using
the contrast variation. Relatively to the specular reflection which gives only the density
profile involving the solvent molecules, the SANS determination of the concentration
profile is easier to interpret. Another advantage derives from the quantity of interface
that is greater in the bulk of SANS samples. It is well illustrated in the case of polymer
layers coated inside porous medium for instance.

6.3. Example of the determination of a polymer brushes [41]

The polymer brushes are formed by grafting a polymer extremity on a wall with a high
grafting density (see Fig. 11). The concentration profile was theoretically predicted to be
a step function [42] or a parabolic profile [43]. The step profile is: c(z)=Cs if 0<z<h and
c(z)=0 elsewhere. This leads [41] to the following scattering function for the cross term:
- — ^ Q - ^ O - cosQh) (42)
 158

Fig. 11 - Schematic representation of a brush of

polymers grafted on solid particle, h is the
thickness of the layer and z the distance to the
wall. The model is that of a rod of blobs with
excluded volume [42].

The parabolic profile is: c(z)=cs [l-(z/h) 2 ], the corresponding scattering function is:

ps l gp(Q)=- l— (l-cosQh) (43)

Qh
The experiment [41] was made using a deuterated polymer chain (Mw=69000)
grafted on a porous silica (mean pore diameter of about 3000A and 2.5 m2/cm3 of
specific area). The grafting density was 5.98 mg/m2. The solvent is a good solvent of
polystyrene, the dichloromethane CH2C12. A ratio yg=90% of CD2C12 in the solvent
matches the scattering length of the silica. The cross term pS 1 ^ was determined from
others measurement made with y=100 and 50%. The fit of experimental data with Eq 43
is shown in Fig. 12.

1 1
o 1 i 1 1 1

15 -
17 °-
o o.io
\
-
00
>~ 0.5
o
) 200 400 .500 soo
2 [A]
S CO
C * * *
In

c o 0.01 0.02 0.03

1
Q A"
Fig. 12 - Plot of the cross term polymer particle, Q'pkpkg S'K versus the scattering vector
Q. The full hne is a calculated curve using the parabolic profile c(z) shown in the inset as
a function of the distance to the wall z.

The quality of the fit proves that the profile is parabolic since the step profile gives a very
different function. This result was simultaneously obtained by neutron reflectometry [44]
The reader will find in ref. 41b a detailed discussion about the determination of c(z) from
the term S pp and the sensitivity of the SANS technique for such a determination.

7. Discussion

The aim of this introductory course was to demonstrate that SANS is the main
tool in order to determine the polymer conformations. It has been shown how to create
 159

a contrast by the deuteration of a polymer in its melt and how to simplify the study of a
binary system with the contrast variation method. It is difficult to discuss all the
numerous SANS possibilities in the polymer field. Nevertheless two methods, recently
developed to study the concentrated solutions, have to be mentioned. One is the ZAC
(zero average contrast) method [45] which allows to measure directly the chain form
factor from a mixture of H and D polymers in a mixture of H and D solvents. The other
is the TIS (Triple Isotopic Substitution) method [46] which gives the form factor of
labelled moieties of a macromolecule dispersed in a complex solution. The latter is
useful for block copolymers. A survey of these possibilities is given elsewhere [47].
At last, the difficulties of the isotopic substitution have to be raised. Until now,
the isotopic substitution was assumed to have no action on the labelled scatterers i. e.
there was no correlation between the position of an atom and its isotopic species. It is
true for the solvent molecules. For instance, preferential adsorption is not encountered
with a mixture of deuterated an undeuterated ones. It is not true with polymers for which
it is well known [1,3,7] that interaction between H and D monomers exists but is weak.
The formalism developed by de Gennes (section IV in ref. 7) for the polymer blends has
to be applied in the melt. If x is the Flory parameter of the interaction between
monomers of different species, assuming ND=NH, Eq. 20 becomes :
(bD-bH)2nm= 1 1
I(Q) xS1D(Q) (l-x)S 1 H (Q)
In the general case the spinodal decomposition occurs when 1(0) diverges :
_J_ + 1 2x = 0 (45)
xN D ( l - x ) N H
(this avoids negative intensity values for the larger x values). Thus the higher is the
molecular weight the higher is the probability to have a phase separation. For isotopic
blend of polystyrene, the % value is weak, about 2 10"*.
Another difficulty for the SANS workers is the transesterification [48]. It is a
random scission and recombination of carboxyl groups (C=O)O in each monomer of
polyester induced by heating. But the recombinations do not necessarily occur with the
arms of the originate polymer chain. Thus starting with a mixture of totally protonated
and deuterated polymers, then proceeding to a carefull annealing in order to eliminate
voids, as final sample a very homogeneous melt of D polymers, partially D polymers and
totally H polymers is obtained. The data scattered by such a sample are, to day, quite
impossible to analye.
Finally deuteration can introduce changes that can be important in the
neighbouring of a transition temperature. For instance the heavy ice occurs at 4°C and
not at 0°C as for H2O or the concentration from which surfactant molecules are
beginning to aggregate is different in H2O or in D2O. For polymers, it is known [1] that
the crystallisation temperatures or the Flory temperature 0 depend on the isotopic
species. The effects of these small temperature variations, a few degrees relatively to
about 300K, have to be studied before starting any SANS experiment.
The author wishes to thank A. Briilet for fruitful comments on the manuscript.
 160

References

1. J.S. Higgins, H.C. Benoit, Polymers and Neutron Scattering (Clarendon Press,
Oxford, 1994).
2. A. Guinier, G. Fournet, Small Angle Scattering of X-rays (Wiley Interscience, New-
York 1955).
3. G.D. Wignall, in Physical Properties of Polymers Handbook, ed. J.E. Mark (AIP,
New York, 1996) p. 299.
4. M.G. Huglin, Light Scattering from Polymer Solutions (Academic Press, London,
1982).
5. Neutron, X-Ray and Light Scattering, ed. P. Lindner and T. Zemb (North-Holland,
Amsterdam, 1991).
6. P.J. Flory, Principles of Polymer Chemistry (Cornell Univ. Press, New York, 1953).
7. P.G. de Gennes, Scaling Concepts in Polymer Physics (Cornell Univ. Press, New
York, 1979).
8. J. des Cloizeaux, G. Jannink, Les Polymeres en solution (Les Editions de Physique,
LesUlis, 1987).
9. L. Van Hove, Phys. Rev. 95(1954)249.
10. J.P. Cotton in ref. 5, p. 3.
11. PA. Egelstaff, An Introduction to the Liquid State (Clarendon Press, Oxford, 1975).
12. Table of neutron scattering lengths : V.F. Sears, Neutron News, 3(1992)26.
13. M. Daoud, J.P. Cotton, B. Farnoux, G. Jannink, G. Sarma; H. Benoit, R. Duplessix,
C. Picot, P.G. de Gennes, Macromolecules, 8(1975)804.
14. B.H. Zimm, J.Chem.Phys. 16(1948)1093.
15. See the lecture of R. Scherm in this journal.
16. Neutron Scattering, ed. K. Skoeld and D.L. Price ( Vol. 23 of the Series Methods of
Experimental Physics, Academic Press, New-York, 1986).
17. J.P. Cotton, D. Decker, H. Benoit, B. Farnoux, J.S. Higgins, G. Jannink, R. Ober, C.
Picot, J. des Cloizeaux, Macromolecules, 7(1974)863.
18. a A.Z. Akcasu, G.C. Summerfield, S.N. Jahsan, C.C. Han, C.Y. Kim, Y.J. Hyuk,
J.Polym.Sc, 18(1980)863.
b F. Boue, M. Nierlich, L. Leibler, Polymer, 23(1982)29.
19. J.P. Cotton, F. Hardouin, Progress in Polymer Science, in press (1996).
20. G.D. Wignall, F.S. Bates, J. Appl. Cryst, 15(1987)28.
21. LA. Feigin, D.I. Svergun, Structure Analysis by Small Angle X-ray and Neutron
Scattering (Plenum Press, New-York, 1969).
22. P. Debye, J. Chem. Phys., 55(1946)572.
23. J. des Cloizeaux, Macromolecules, 6(1973)403.
24. P. Sharp, V.A. Bloomfield, Biopolymers, 6(1968)1201.
25. T. Yoshizaki, H. Yamakawa, Macromolecules, 13(1980)1518.
26. A. Brulet, F. Boue, J,P. Cotton, J.Phys. II (France), 6(1996)885.
27. H. Benoit, P.M. Doty, J. Phys. Chem., 87(1953)958.
28. G. Porod , Kolloid-Z., 124(1951)83.
29. A. Guinier, Theorie et Technique de la Radiocristallographie (Dunod, Paris, 1964)
p. 645.
30. R.G. Kirste, R.C. Oberthur, in Small Angle X-ray Scattering, ed. O. Glatter, O.
Kratky (Academic Press, New York, 1982) p.387.
31. R.G. Kirste, W.A.Kruse, K. Ibel, Polymer, 16(1975)120.
32. G. Lieser, E.W. Fischer, K. Ibel, Polym. Letter, 13(1975)39.
 161

33. G.D. Wignall, D.H.G. Ballard, J. Schelten, Eur. Polym. J., 10,682(1974).
34. L. Fetters, D J . Lohse, D. Richter, T.A. Witten, A. Zirkel, Macromolecules,
27(1994); L. Fetters, N. Hadjichristidis, J.S. Lindner, J.W. Mays, J.Phys.Chem.
23(1994)619.
35. J.P. Cotton, D. Decker, В. Farnoux, G. Jannink, R. Ober, С Picot, Phys. Rev. Lett.
32(1974)1170.
36. G. Jannink, P. Pfeuty, A. Lapp, J.P. Cotton, Europhys. Lett., 27(1994)47.
37. H.B. Stuhrmann, J. Appl. Cryst. 7(1974)173.
38. T. Cosgrove, T.L. Crowley, B. Vincent, in Adsorbtion from Solutions R. Ottewill
ed., Academic Press (1983)
39. aL. Auvray, C.R. Acad. Sc. Paris, Ser.2, 302(1986)859.
b L. Auvray, P.G. de Germes, Eur. Phys. Lett., 2(1986)647.
с L. Auvray, J.P. Cotton, Macromolecules, 20(1987)202.
d L. Auvray, P. Auroy, in ref. 9 p. 199.
40. T.P. Rüssel, Materials Science Reports, 5(1990)173.
41. a P. Auroy, Y. Mir, L. Auvray, Phys. Rev. Lett., 69(1992)93.
b P. Auroy, L. Auvray, J. Phys. II France, 3(1993)227.
42. a S. Alexander, J. Phys. France, 38(1983)983.
b P.G. de Gennes, Macromolecules, 13(1980)1069.
43. a V.A. Pryamitsyn, O.V. Borisov, E.B. Zhulina, T.M. Birshtein, Modern Problems of
Physical Chemistry of Solutions p. 62, Donish and Leningrad University (1987).
b S. Milner, T. Witten, M.Cates, Europhys. Lett., 5(1990)413.
44. J.B. Field, С Toprakcioglu, R.C. Ball, H.B. Stanley, L. Dai, W. Barford, J. Penfold,
G. Smith, W. Hamilton, Macromolecules, 25(1992)434.
45. T. Csiba, G. Jannink, D. Durand, R. Papoular, A. Lapp, L. Auvray, F. Boue, J.P.
Cotton, R. Borsali, J.Phys.II (France), 1(1991)381. + F. Boue, J.P. Cotton, A. Lapp,
G Jannink, J.Chem. Phys., 101(1994)2562.
46. Y. Pavlov, I N . Serdyuk, J.Appl. Cryst., 20(1987)105. + I.N. Serdyuk, G. Zaccaï, J.
Mol. Structure, in press (1996).
47. J.P. Cotton, Submitted to the special issue Scattering Methods in Colloid Sciences -
Theory and Applications of Adv. in Colloid and Interface Sc. (1996).
48. K.P. Alea, J.M. Schultz, K.H. Gardner, G.D. Wignall, Polymer, 27(1986)1581. +
H. Benoît, E.W. Fischer, HG. Zachmann, Polymer, 30(1989)379. + M.H. Li, A.
Brûlet, P. Keller, С. Strazielle, J.P. Cotton, Macromolecules, 26(1993)119.