Vous êtes sur la page 1sur 10

Analysis of small angle neutron scattering spectra from polydisperse

interacting colloids
Michael Kotlarchyk and SowHsin Chen

Citation: J. Chem. Phys. 79, 2461 (1983); doi: 10.1063/1.446055
View online: http://dx.doi.org/10.1063/1.446055
View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v79/i5
Published by the American Institute of Physics.

Related Articles
Effect of shell thickness on two-photon absorption and refraction of colloidal CdSe/CdS core/shell nanocrystals
Appl. Phys. Lett. 99, 231903 (2011)
Measuring the ordering of closely packed particles
Appl. Phys. Lett. 99, 221910 (2011)
Structure and transport properties of polymer grafted nanoparticles
J. Chem. Phys. 135, 184902 (2011)
Short-time self-diffusion coefficient of a particle in a colloidal suspension bounded by a microchannel: Virial
expansions and simulation
J. Chem. Phys. 135, 164104 (2011)
Communication: A dynamical theory of homogeneous nucleation for colloids and macromolecules
J. Chem. Phys. 135, 161101 (2011)

Additional information on J. Chem. Phys.
Journal Homepage: http://jcp.aip.org/
Journal Information: http://jcp.aip.org/about/about_the_journal
Top downloads: http://jcp.aip.org/features/most_downloaded
Information for Authors: http://jcp.aip.org/authors
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
Analysis of small angle neutron scattering spectra from
polydisperse interacting colloids
Michael Kotlarchyk and Sow-Hsin Chen
Nuclear Engineering Department, Massachusetts Institute o/Technology, Cambridge, Massachusetts 02139
(Received 4 April 1983; accepted 15 April 1983)
In this paper, we outline a simple procedure for analyzing small angle neutron scattering (SANS) spectra from
interacting colloids containing a continuous distribution of particle sizes. In particular, the effects of
polydispersity on the apparent interparticle structure factor S'(Q) observed by SANS is investigated. We find
that the oscillations in S'(Q) are significantly damped in comparison to those of the true structure factor S(Q).
When our procedure is extended to the analysis of SANS spectra from nonspherical particles, we find that
orientational averaging affects S'(Q) in a qualitatively similar way. The applicability of the procedure to the
analysis of real data is demonstrated with spectra taken from water-in-oil microemulsions, ionic micelles, and
interacting proteins.
I. INTRODUCTION
Small angle neutron scattering (SANS) has been
demonstrated to be a useful tool for characterizing the
structure of interacting colloid systems. To analyze
SANS data, one generally assumes that interactions oc-
cur between particles which are monodisperse, although
it is often desirable to consider a spectrum of particle
sizes. In the literature there are extensive examples
of incorporating polydispersity into the analysis of small
angle scattering measurements from systems dilute
enough to neglect interparticle interactions. 1-5 How-
ever, when Significant interactions are present, an exact
polydispersity analysis can only be performed for cer-
tain interparticle pair potentials. In particular, scatter-
ing functions for polydisperse fluids of hard and perme-
able spheres have been obtained in rather complicated
closed form. 6-8 Aside from this case, the authors are
unaware of the determination of exact scattering func-
tions for any other interaction potentials incorporating
polydispersity.
It is desirable to develop a tractable procedure for
incorporating polydispersity considerations into the
analysis of SANS data for two reasons. First, one is
often interested in an estimate of the width of the par-
ticle size spectrum. Secondly, by ignoring the existence
of a nonzero width, one can grossly distort the outcome
of determinations on other parameters of interest. In
this paper, we demonstrate the applicability of a rather
simple procedure for analyzing SANS data from polydis-
perse interacting colloids. In the course of deriving the
procedure, it is shown that similar considerations can
be extended to spectra from nonspherical particles that
are subject to strong interactions. We demonstrate that
polydispersity and particle orientational averaging affect
scattering spectra in qualitatively Similar ways. The
applicability of the described procedure to real data
analysis is shown by discussing four examples in detail.
To derive the procedure, we begin by conSidering the
basic expression for the coherent scattering component
of the differential cross section per unit volume given
by the following ensemble average:
~ (Q) = V"1(1t. b , exp(iQ' r
,
) I). (1)
Here, V is the sample volume, N is the number of nu-
clei in the sample, and b, is the bound coherent scatter-
ing length of the lth nucleus. The position of the lth
nucleus is denoted by r, and Q is the scattering vector.
In a SANS experiment, the magnitude of Q is equal to
(411/A)sin(e/2), where A is the wavelength of incident
neutrons and e is the scattering angle. Suppose the
sample contains N" particles immersed in a homoge-
neous solvent. Then we can imagine the sample volume
to be partitioned into N" cells, each containing exactly
one particle. Without lOSing any generality, we can re-
label each nucleus with two subscripts i and j as follows:
(2)
where b
j
i and rj j are the scattering length and position
of the jth nucleus contained in the ith cell. N
j
is the
number of nuclei in the ith cell. Denoting the center of
mass of particle i by RI and the position of a nucleus
at rlJ relative to this point by Xi' we make the decom-
position rli=Rj +Xj' So,
(3)
Now, define the form factor FI(Q) of the ith particle
as
H
FI(Q)=t bliexp(iQ' Xi}
(4)
Then,
HI> HI>
~ (Q) = V-
1
(L LF1,(Q)FI(Q} exp[iQ' (R
I
-R
I
.}]). (5)
lAO. 1.1 I' =1
The form factor FI(Q} is expressed in terms of the in-
dividual scattering lengths of the nuclei in the cell con-
taining the ith particle. For many cases, it is more
convenient to express the form factor' in terms of scat-
tering-length densities, thus rewriting Eq. (4) as
FI(Q) = f drpI(r} exp(iQ' r} ,
cell I
(6)
where
J. Chern. Phys. 79(5), 1 Sept. 1983 0021-9606/83/172461-09$02.10 1983 American Institute of Physics 2461
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
2462
M. Kotlarchyk and S.-H. Chen: Neutron scattering from colloids
HI
PI(r) == L: b
lj
1i(Y -Xi)
j=t
is the scattering-length density at position r of the ith
cell. We would like to transform the form factor to an
integral over the particle volume only. Equation (6) can
be reexpressed as
+ P
s
f drexp(iQ r) ,
cell i
(7)
where P. is the scattering-length density of the solvent.
Since the second integral is simply a delta function cen-
tered at Q =:: 0, we obtain
FI(Q)=j dr[p,(r)-p.lexp(iQr) , Q*O. (8)
particle I
Up to this point, the derivation of the scattering cross
section has been completely general. When considering
particles which are polydisperse or nonspherical, the
form factor varies from particle to particle. For the
remainder of this paper, we will assume that the par-
ticle size and orientation are uncorrelated with the posi-
tions of the particles. In this case, we can write
(Q) = v-t(t (F
I
(Q)Fi (Q)
x exp[iQ . (iii - iin 1) , (9)
where the inner brackets represent an average weighted
by the distribution of particle sizes and orientations.
This average can be decomposed into
(FI(Q)Fj. (Q) = [(IF(Q) 12) -I (F(Q) 1 2]li
w
+ 1 (F(Q) 12 .
(10)
Then, the cross section reduces to
where np = Np/V is the average number density of par-
ticles in the sample and S(Q) is the interparticle struc-
ture factor defined by
(12)
Strictly speaking, when investigating a polydisperse
system, one should deal with partial structure factors
9
SI j(Q) determined by solving the matrix form of the
Ornstein-Zernicke equation
to
for the pair potential ,j
between particles with hard-core diameters (], and (]j.
By assuming that particle size and position are uncor-
related, we are forced to deal with the average struc-
ture factor S(Q) of Eq. (12). In Sec. II, examples will
demonstrate that it is reasonable to approximate S(Q)
with a structure futor calculated from the one-compo-
nent Ornstein-Zernicke equation using an effective po-
tential with a physically justifiable hard-core diameter
(].
To observe the effects of polydispersity and orienta-
tional averaging on the scattering cross section, it is
best to rewrite Eq. (ll) as
d"L, ('" (- '(-
dO QI=npP Q)S Q),
defining the terms
S'(Q) == 1 +13(Q)[S(Q) -1] ,
13(Q) == 1 (F(Q) 1
2
/( 1 F(Q) 12) ,
(13)
(14)
(15)
where p(Q) =( IF(Q)1
2
). S'(Q) acts as an apparent inter-
particle structure factor. 13 is a Q-dependent factor be-
tween zero and one that suppresses the oscillations of
the true structure factor S(Q) in the observed scattering
spectrum from a poly disperse or nonspherical system
of particles. It is the Significance of this factor that is
of primary interest in this paper.
We can immediately see the limiting forms of Eqs.
(13)-(15). For the case of scattering from monodis-
perse spheres, a(Q) == 1 and we obtain the particularly
simple form
d"L, (- -
dO Q) =npP Q)S Q) .
(16)
In the limit of very low particle density, the solution
becomes ideal and S '(Q) =S(Q) = 1, so that
(17)
Since the remainder of the paper deals with isotropic
systems containing homogeneous particles, we will re-
place the scattering vector with its magnitude Q .
II. ANALYSIS OF POL YDISPERSE SYSTEMS
For the case of a system of poly disperse homogeneous
spheres, the following expressions can be used to cal-
culate the averages necessary for computing the scat-
tering cross section:
(IF(Q)1
2
)= [" IF(Q,R)[2j(R)dR,
o
1 (F(Q)j2 = I F(Q, R)j(R) dR r '
F(Q, R) = (41T/3)R
3
.
(18)
(19)
(20)
F(Q, R), the form factor for a sphere of radius R, de-
pends on the first order spherical Bessel function it (x).
j(R) dR is the probability of a sphere having a radius be-
tween R and R + dR. is the difference between the
scattering length density of the particle and the solvent.
Equations (18) and (19) assume that the particle size
distributionj(R) is known. However, for small Q, each
integral can be replaced by an infinite series in even
powers of Q, such that the coefficients are only related
to the moments of the distribution function. So-called
cumulant expansions for (I F(Q) 12) and 1 (F(Q 12 can be
developed and are somewhat useful for predicting the
effects of polydispersity on the small-Q behavior of
scattering curves. For example, one can easily see
that
(21)
In addition, one can show that
J. Chern. Phys . Vol. 79. No.5. 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
M. Kotlarchyk and S.-H. Chen: Neutron scattering from colloids
2463
p(Q)-exp (- r (R8)/(R6)
- exp( - E:Q2 if! 15) , (22)
where R is the mean of the size distribution and E: is
generally a factor greater than unity that depends on the
spread of the distribution. Thus, for a dilute system
[S(Q)=1], a plot of lntn;ldn VS Q2 gives a slope equal
to - E:R2/5. Therefore, neglecting polydispersity (i. e.,
forcing E: = 1) can lead to a gross overestimate of the
mean particle size.
For certain distribution functions, it turns out that it
is almost as easy to calculate the integrals (18) and (19)
exactly as to calculate the various moments. Since it
is reasonable to expect that the detailed shape of the dis-
tribution is not of critical importance in analyzing sys-
tems that are of low to moderate polydispersity (i. e.,
less than about 20%), extremely useful results can be
obtained by performing the integrals exactly for a few
useful distribution functions. Shown below are explicit
expressions for p(Q) and (3(Q) for two commonly used
distributions.
A. Schultz distribution
The Schultz distribution is a two-parameter function:
fs(R) =e; 1 r+
t
R
Z
exp [- (Z; 1)RJ Ir(z + 1), z> -1 ,
(23)
where R is the mean of the distribution and Z is a width
parameter. r(X) is the Gamma function. The function
is skewed toward larger sizes, tending to a Gaussian
form at large values of Z. The distribution approaches
a delta function at R = R as Z approaches infinity. The
root mean square deviation from the mean is given by
(24)
The integrals (18) and (19) can be performed explicitly.
The result is
(25)
(26)
where
Gt(Q) = a-(z+u - (4 +a 2r( Z+1l/2 cos [(Z + + (Z + 2)
x (Z + 1){a-(Z+3) + (4 + a
2
r(Z+3>/2 cos [(Z +
(27)
G
2
(Q) = sin [(Z +
- (Z + 1)(1 + a
2
r1!2 cos [(Z + ,
a = (Z + 1)/QR .
(28)
(29)
Reference 4 has a discussion of light scattering from
dilute spheres with a Schultz polydispersity, and Eq.
(3.11) of that reference is equivalent to our Eq. (25),
except for a constant factor. In addition, Fig. 2 of Ref.
4 shows p(Q) as a function of QR for various values of Z.
For completeness we mention that E: in Eq. (22) is
simply
(Z + 8)(Z + 7)
E: = (Z + 1)2
B. Rectangular distribution
We define the rectangular distribution as
IR-RI ::<sW
IR -RI >W
(30)
(31)
with the constraint that W::<sR. Again, R is the mean of
the distribution and W is the half-width. The root mean
square deviation is a R = W 1,13. The evaluation of Eqs.
(18) and (19) leads to
,
(3(Q) = ,
where
(32)
(33)
Ht(Q) = - -NW + + tQ
3
W
3
-t cos 2QR sinQW cosQW + iQ2R2 cos 2QR sin 2QW + iQ
2
W
2
cos2QR sin2QW
+Q2 RW sin 2QR cos 2QW + 3QWcos
2
QR cos
2
QW +3QWsin
2
QR sin
2
QW - 6QR cosQR sinQR cosQWsin QW,
(34)
H
2
(Q) =2 sinQR sinQW -QR cosQR sinQW -QWsinQR cosQW . (35)
For this distribution,
1
E: - 3 9
- 1 + 5r2 + 3r
4
+ !r6
7
(36)
where r = W IR.
Figure 1 shows (3 vs QR for various values of the
polydispersity, = a RIR, both for the Schultz and rect-
angular distributions. Note that for $ O. 20, (3(QR) is
approximately the same for the two distributions. This
is in keeping with our earlier statement that the detailed
shape of the distribution is not critical in this region of
The most important part of Fig. 1 to examine is
QR $11, since this is the region where the first interfer-
ence peak of S(Q) will occur. Note the zeroes of J3(QR),
the first one being between QR equal to 4.0 and 4.5.
This apparent anomaly is also reflected in the results
of Ref. 8, where the apparent interparticle structure
factor shows a peculiar hump around QRmu. 4. 5, Rmu
being the maximum of the Schultz distribution [Rmu
J. Chern. Phys., Vol. 79, No.5, 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
2464
M. Kotlarchyk and S.-H. Chen: Neutron scattering from colloids
0.6
{3(QRI
0.2
0.6
{3(QRI
0.4
0.2
FIG. 1. i3 vs QR for spheres with various polydispersities
= !J"R/R for (a) a Schultz size distribution and (b) a rectangular
size distribution.
= (Z Iz + 1)1l]. Mathematically, it is obvious that the
zeroes in (3(QR) are simply due to the complexity in the
structure of F(Q,R), the form factor for a homogeneous
sphere.
To illustrate the effects of a Schultz polydispersity on
the apparent structure factor, Fig. 2 shows S'(Qa) cal-
culated for system of polydisperse hard spheres with
volume fraction 1/ = 0.30 and hard-sphere diameter a,
such that a
3
= 8(R 3>. The monodisperse structure factor
S(Qa) (i. e., the = 0 case) was computed using the
closed-form solution to the Percus-Yevick equation. 11-13
The apparent structure factors S'(Qa) for > 0 were
computed with Eqs. (14) and (26). This figure should
be compared with Fig. 5 of Ref. 8, which gives results
of the analogous calculation of the apparent structure
factor performed by applying the closed form solution
to the Percus-Yevick approximation for a mixture of
hard spheres. The results of Ref. 8 show smaller val-
ues for both S '(Q = 0) and the height of the first inter-
ference peale At first glance, it also appears that
there is a difference in the position of the first inter-
ference peak. However, this is not true because Ref.
8 shows S'(2QR
max
) and our figure graphs S'(Qa). By
noting that
2QR
max
/Qa = zl[(z + l)(Z + 2)(Z + 3)]113,
it is seen that the peak positions are indeed the same.
The usefulness of our formalism becomes apparent
when it is applied to speCific SANS spectra from poly-
disperse interacting colloids. Here, we present three
such examples illustrating the feasibility of applying our
procedure. All three spectra were obtained from ex-
periments conducted at the National Center for Small
Angle Scattering Research of the Oak Ridge National
Laboratory. The measurements were performed with
a neutron wavelength of 4. 75 A at a flux between 10
4
and
10
5
n/cm
2
/s. All the samples were contained in quartz
cells with 1 mm thick windows. Sample temperature
was controlled to within 0.1 C with an external water
bath. In each case, the raw scattering data was mani-
pulated in a standard way to subtract incoherent back-
ground and correct for detector sensitivity. The abso-
lute intenSity calibration of each spectrum was estimated
to be about 5% accurate.
1. Three-component water-in-oil microemulsion far
from a phase boundary (weakly interacting system)
Our microemulsion consisted of 3% (wt. IvoI.) sodium
di-2-ethylhexylsulfosuccinate, a surfactant known as
AOT (specific volume 1. 13 g/cc), and 2% (vol. Ivol.)
heavy water in n-decane. The AOT, obtained from
American Cyanimid was twice recrystallized from
methanol and the n-decane was a 99% + gold label prod-
uct from Aldrich Chemical Co. The heavy water con-
tained 99. 7% D
2
0 and 0.3% H
2
0 (atom %). At the sample
temperature of 41. 7C, the transmission was measured
to be 52.4% for the 1 mm sample path length.
The data was analyzed by assuming that polydisperse
droplets containing a water core were formed in the oil
solvent. The scattered intensity was modeled using
Eqs. (13)-(15), assuming a Schultz polydispersity and
the Percus-Yevick hard-sphere solution for S(Q). The
volume fraction of the minor component was estimated
to be 1/ = O. 0465 and the effective hard-sphere diameter
a was calculated by using
(J = (!!!L)1!
3
,
rrnp
where
,
(
R3> _ (Z + 2)(Z + 3) 113
- (Z + 1)2 ,
S'(Qa-)
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
(37)
(38)
(39)

Qa-
FIG. 2. The apparent structure factor S'(Qa) for an 7) = O. 30
volume fraction hard-sphere solution calculated with the
Percus-Yevick approximation. A Schultz polydispersity was
assumed and the hard-core diameter a was taken to be
2(R3)1/3.
J. Chem. Phys., Vol. 79, No.5, 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
M. Kotlarchyk and S.-H. Chen: Neutron scattering from colloids 2465
and cP is the volume fraction of heavy water. The fit to
the scattered intensity, therefore, depends only on the
two parameters If and Z. Visually monitoring the fit
made it apparent that there existed a unique combination
of Ii and Z for an optimum fit. The best fit to the data,
shown by curve a in Fig. 3, results in the parameters
Ii = 23 A and Z = 10, which corresponds to a polydis-
persity of = O. 30. For comparison, curve b illustrates
the scattering curve expected for If = 23 A and Z - 00,
i. e., the monodisperse case. Clearly, one effect of
polydispersity is to smooth out oscillations in the tail
of the scattered intensity. Although not obviOUS from
the figure, neglecting (3(Q) would cause the calculated
curve to fall below the data points in the small-Q re-
gion.
2. Three-component water-in-oil microemulsion near
a critical point (strongly attractive system)
This microemulsion was made from 3% AOT and 5%
D
2
0 in n-decane. The components were taken from the
same stock as the previous example. Again, the path
length was 1 mm with a transmission of 52.4%. The

-I,
dn em
0.1

\.
\
\
\
\
\ b
\
\
\
\
\
\
\ \
\\

\
\
\

0.00
FIG_ 3. SANS spectrum from the weakly interacting micro-
emulsion (2% water) described in example 1. Curve (a): Best
fit to data assuming spheres with a Schultz polydispersity and
hard-sphere interactions modelled with the Percus-Y evick
approximation <R =23.0 A. Z=10.0, 0'=66.3 A. 11=0.0465).
Curve (b): Spectrum calculated by assuming monodisperse
spheres with hard-sphere interactions <R = 23. O. 0'= 60. 9 A.
11 = 0.0465). Curve (b) is artificially normalized to the level
of curve (a).
100

.000 .004 .008 .012 .016 .020 .024 .028 .032.036
Q(,\-')
FIG. 4. SANS spectra from microemulsion described in ex-
ample 2 (5% water). A comparison is made of the scattering
spectra collected far from the critical temperature (6) and
close to the critical temperature ( ). The best-fit parameters
for the calculated curve (A) near the critical temperature are:
A, X=285, R=44.6 A. Z=12.5.
sample temperature was stabilized at 42. 6C, which is
known to be less than one degree below a lower critical
point phase separation previously observed by light
scattering. 14
In Fig. 4, the scattering spectrum from this sample
is shown along with the spectrum for the same sample
at a temperature far below the critical point. The
marked increase in the forward scattering intensity can
be modeled with a modified Ornstein-Zernicke
10
struc-
ture factor
(40)
where is the correlation length of critical fluctuations
and X is equal to n, kBTKT' KT being the osmotic isother-
mal compressibility.
The additive factor of unity in Eq. (40) accounts for
the fact that the structure factor should approach unity
at large Q for a dilute hard sphere system. As before,
p(Q) and (3(Q) were computed using a Schultz distribu-
tion of sphere sizes. It is important to realize that the
spectrum shown in Fig. 4 only shows data for Q
::; 0.036 k\ even though a complete spectrum was ob-
tained up to about Q '" O. 2 A -1. The model was applied
to the complete spectrum. 15
With the above assumptions, the scattered intenSity
depends independently on the parameters X, Ii, and
Z (along with a small residual background and normal-
ization correction factor near unity). A quasi-Newton
J. Chem. Phys., Vol. 79, No.5, 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
2466 M. Kotlarchyk and S.-H. Chen: Neutron scattering from colloids
25,----------------------------------,
20
15
10
5
-'.
\
\
\
\
\
\
5'(0)'
\
\
'\.
"'-
--------
FIG. 5. Individual components contrIbuting to the calculation
of curve (A) in Fig. 4.
algorithm was used to perform a weighted nonlinear
least squares fit with the above parameters. The solid
curve kin Fig. 4 shows the best fit to the data. Al-
though it is not shown here, the fit is reliable over ap-
proximately five decades of intensity. 15 The best fit
parameters are A, X=285, R=44. 6 A, and
Z=12.5 Figure 5 shows the various com-
ponents of the fitted curve for Q:S 0.036 A-l. The
marked difference between S(Q) and the apparent struc-
ture factor S '(Q) illustrates that the values of the critical
parameters and X are significantly affected by incor-
porating polydispersity into the data analysis.
3. Ionic micellar solution (strongly repulsive system)
The sample was a 0.294 M micellar solution of the
ionic detergent lithium dodecyl sulfate (LDS) in heavy
water. The LDS, obtained from Sigma Chemica:! Com-
pany, was 99% + pure. The heavy water was 99.8%
D
2
0 and 0.2% H
2
0. A 2 mm path length gave a trans-
mission of greater than 65%, and the sample tempera-
ture was maintained at 37. OC.
The spectrum that was analyzed was one of the data
sets used in Ref. 16. LDS micelles were modeled as
homogeneous prolate spheroids interacting through a
screened Coulombic pair potential. S(Q) was calculated
with a model developed by Hayter and Penfold. 11 Their
calculation uses a mean spherical approximation (MBA)
for spherical macroions embedded in a solvent of ionic
strength I. The details of the model are discussed else-
where, 16.11 so only information essential to following
our analysis will be included here. Although the model
is derived for spherical macroions, it will be shown
that the application to spheroids can be affected by choos-
ing a physically plausible effective hard-core diameter
a.
As in the previous examples, the scattering curve is
modeled with Eqs. (13)-(15). In this case, np is the
number denSity of micelles in the solution. If the mean
aggregation number n (number of monomers per micelle)
is determined, np can be computed by
(41)
where NA is Avogadro's number and C is the molar con-
centration of LDS monomers in the micellar state. That
is, C is equal to the total detergent concentration minus
the critical micellar concentration (c. m. c.). For no
added salt, the c. m. c. is 0.0089 M LDS.18 Previous
measurements
19
show that the hydrocarbon core of the
LDS micelle in D
2
0 can be modeled by a prolate sphe-
roid with semiminor axis b
c
fixed at 16.7 A (the length
of a fully extended dodecyl chain) and a variable semi-
major axis a
c
' computed by assuming a close-packed
micellar core. Then the mean volume of the hydrocar-
bon core is
(42)
thus allowing one to compute the mean aggregate number
n = Vc/v = , (43)
where v 364 }..3 is the elemental chain volume of an
LDS monomer. 19
To use the Hayter-Penfold model for S(Q), the fol-
lowing parameters need to be speCified: the effective
hard-core diameter a, the volume fraction 17, a screen-
ing parameter k, and the contact potential ye-
k
(ex-
pressed in units of kB T).
We postulate that a is best computed by considering
the excluded volume of a rigid prolate spheroid deter-
mined by calculating the second virial coefficient using
the method outlined by Isihara. 20 This leads to
where
3 ( sin'l(e) ( 1 - e
2
1 + e)
{j ="4 1 + e(l _ e2)1/2 . 1 .... "2e" In 1 _ e '
e
2
= (li
2
_ b
2
)/a
2

In the case of LDS micelles
a=a
e
+5. 56}..,
b=b
c
+5.56}.. ,
(44)
(45)
(46)
(47)
(48)
where 5. 56 }.. is the diameter of the LDS head group.
It has been previously estimated
19
that the axial ratio of
LDS micelles is in the neighborhood of ae/b
e
= 1. 3. Ap-
plication of the above relations show that {j 3.0, which
is the result one obtains for a hard sphere with volume
equal to that of the spheroid.
The screening parameter k depends on the inverse
screening length K:
J. Chern. Phys., Vol. 79, No.5, 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
M. Kotlarchyk and S.-H. Chen: Neutron scattering from colloids 2467
k = Ka ,
K = ( 81TN A e
2
[ )1/2
\iOOOE:kBT
[=c.m.c. +aC/2,
(49)
(50)
(51)
where a is the mean degree of ionization of a micelle
defined such that the number of charges on a micellar
surface is z =an. In Eq. (50), e is the electronic charge
and E: = 78.25 is the dielectric constant of D
2
0. The
volume fraction of the solution TJ satisfies Eq. (37) and
the contact potential is
-k 1. 6711X10
5
z
2
ye = (1 + k/2)2 Ta (A) .
(52)
According to the above description, the data fitting
depends on the two independent parameters a
c
and a.
However, to compute p(Q) and /3(Q) assuming a polydis-
perse major spheroid axis we require the introduction
of a third parameter ~ = a./a
c
Strictly speaking, the
calculation of p(Q) and /3(Q) requires both a size average
and an orientation average. However, for the ranges of
a
c
and.ac/b
c
that are expected, we can eliminate the need
for the orientation average and replace it with a sphere
having the same volume as the spheroidal hydrocarbon
core. This assumption will be confirmed in the next
section. The Appendix shows in detail how p(Q) and
/3(Q) depend on a
e
and ~ .
By visually monitoring the fit, we found that the best
fit parameters are unique: a
c
= 21.7 A, ~ = 0.13, and
a = O. 40. The fit to the data is shown in Fig. 6. The
above fit leads to the following values for the other
parameters of physical interest: ae/b
e
= 1. 3, n = 70,
a=47.6A, TJ=0.14, k=3.94, andye-
k
=12.72. The
values of the best fit parameters are the same as those
quoted in Ref. 16, obtained by assuming a monodisperse
micelle. However, the addition of the polydispersity
dn
lUI
[cm'),------------------------,
6.0
5.0
4.0
3.0
2.0
1.0
.-
.r"",
I
i \
i
/
,/
.. /
- - ~
5(0) _
2.0
1.0
FIG. 6. SANS spectrum and best fit curve for 0.294 M LDS
micellar solution described in example 3. The best fit param-
eters are a
e
= 21. 7 ..t, ~ = 0.13, O! = 0.40. The resultingparam-
eters used in the Hayter-Penfold MSA are u=47.li..t, 1')=0.14,
k = 3. 94, and 'Ye-
k
= 12.72. Both the calculated true structure
factor S(Q) (solid curve) and apparent structure factor 51 (Q)
(dashed curve) are shown.
1.0
0.8
0.6
,8(00)
0.4
(a) Pro'ate Spheroid
0.2
~
0.0
0
1.0
0.8
0.6
,8(00)
0.4
(b) Oblate Spheroid
0.2
0.0
0 10
FIG. 7. (3 vs Qa for monodisperse spheroids with various
axial ratios a /b for (a) prolate spheroids and (b) oblate spheroids.
parameter ~ = 0.13 improves the quality of the fit con-
siderably in the region of small Q.
III. EFFECTS OF AVERAGING PARTICLE
ORIENTATION
In Sec. I, it was shown that averaging the orientations
of nonspherical particles modifies the apparent struc-
ture factor S '(Q). To illustrate the effect, we calculated
/3(Q) for homogeneous prolate and oblate spheroids. The
necessary averages were computed by
(IF(QW)= I
1
IF (Q,a,b,!J.)1
2d
!J.,
o
I (F(Qj2 = I ~ 1 F(Q, a, b,!J.) d!J.\2
F(Q, a, b,!J.) = V(a, b).lp[3it(u)/u] ,
(53)
(54)
(55)
where F(Q, a, b, !J.) is the form factor for a spheroid
with semimajor axis a and semiminor axis b. The in-
tegration is over the orientation variable !J., equal to
the cosine of the angle between Q and the direction of the
spheroid major axis. For a prolate spherOid,
u=Q[a
2
IJ.2+b
2
(1_!J.2)]1I2 , (56)
V(a, b) = (41T/3) a b
2
. (57)
For an oblate spherOid,
u =Q[a
2
(1 -!J. 2) + b
2
IJ. 2]1/2 ,
V(a,b)=(41T/3)a
2
b.
(58)
(59)
Figure 2 of Ref. 3 shows the effect of varying the axial
ratio on the form of p(Q). 21 Our Fig. 7 shows the re-
sults of numerically integrating Eqs. (53) and (54) to
J. Chern. Phys., Vol. 79, No.5, 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
2468
M. Kotlarchyk and S.-H. Chen: Neutron scattering from colloids
determine the factor fj(Q) as a function of Qa for various
axial ratios. An important point to observe is that
fj(Q = 0) = 1 and fj(Q) does not decrease very far below
unity until Qa reaches values that are quite large. This
justifies our neglecting orientation effects in the pre-
vious Sec. III B. 3, where Qa
c
was always less than 3.5
and the axial ratio for the spheroidal core was about
1. 3. From Fig. 7(a), it is clear that in that case
95.
Here we present an example illustrating how orienta-
tion averaging can affect the analysis of scattering data:
4. Globular protein in aqueous solution (strongly
repulsive system with nonspherical particles)
SANS spectra of solutions containing various concen-
trations of bovine serum albumin (BSA) in heavy water
were collected at the Brookhaven National Laboratory.
The BSA (molecular weight 66 700) was obtained from
Sigma Chemical Company and the heavy water was like
that described in Sec. In B. 3. The neutron wavelength
was 5.44 A and the flux at the sample was about 7.4
x 10
5
n/cm
2
/s. Samples were contained in cells with
1 mm windows and 2 mm path lengths, giving transmis-
sion values in the vicinity of 75%. The sample tempera-
ture was stabilized at 21. 8C and the pH was measured
to be 7. 0, which is generally agreed to correspond to
7.4.
3.o..-------------------,1.5
(0)
1.0
1.0 0.5

0.00 0.05 0_1 0.10
Q(A )
FIG. 8. SANS spectra and best fit curves for BSA solutions
described In example 4. (a) 8.73% BSA lnD
2
0. (b) 16.36%
BSA in D
2
0. The best fit parameters used in the Hayter-
Penfold MSA are summarized in Table 1.
TABLE 1. Results of the data analysis of SANS intensity spec-
tra from BSA solutions.
BSA (wt./vol.) u(A) 1) k -;e- I(M) z
8.73% 65 0.064 2.1 2.7 0.0095 -10.1
16.36% 65 0.12 3.9 2:3 0.033 -13.4
The two spectra presented in Fig. 8 correspond to
BSA concentrations of 8.73% and 16.36% (wt. IvoI.).
These concentrations were determined by dividing the
weight of BSA by the volume of 20 plus the dry volume
of the BSA, where the specific volume of dry BSA is
taken as 0.734 cc/g. 22 The supplier speCified that the
BSA stock contained 3% water (wt. Iwt.), and this was
corrected for in the concentration determination.
Previous measurements
23
have shown that the par-
ticle form factor for BSA in 20 is that for a prolate
spheroid with a=70 A and b=20 A. The data were fitted
by performing the orientational averages (53) and (54)
to calculate p(Q) and /3(Q). S(Q) was determined with
the Hayter-Penfold MSA described in Sec. In B. 3. To
satisfactorily fit the data, it was found that the param-
eters CJ, 1j, k, and ye- had to be varied independently.
The best fit to each spectrum is shown in Fig. 8. It is
important to realize that these fits require that different
normalization factors be applied to each spectrum, a
fact that we have not been able to explain as of yet.
Table I summarizes the best fit parameters. The ionic
strength I is backed out from Eq. (50) and the charge z
is determined from Eq. (52). The results should be
compared with those of Ref. 23. In that study, CJ was
fixed at 2(ab
2
)1/3 = 60. 7 A and the computation of I was
based on the concentration of univalent ions dissociated
from protein molecules. Note the following points about
our results:
(i) The value of CJ= 65 A is approximately midway be-
tween 2(ab
2
)1/3 == 60.7 A and (1 determined from excluded
volume considerations (69.3 A) by Eqs. (44)-(46).
(ii) Although the values of 1j are much smaller than
predicted by Eq. (37), they scale exactly as the BSA
concentrations. In addition, the values are in approxi-
mate agreement with the trends of the Ref. 23 results.
(iii) The values of z are considerably higher than
those determined in Ref. 23. However, for pD= 7. 4,
the dependence of the fitted z values on the fitted ionic
strengths is in agreement with the predictions of Ref.
24, which give the mean charge of BSA at various solu-
tion conditions deduced from hydrogen ion titration data
and chlorine ion binding data.
IV. CONCLUSIONS
Exact models for polydispersity in the presence of
speCific interparticle interactions are rare, and the few
cases available are not in a form readily useable for
SANS spectra analysis. In this paper, we have demon-
strated, through real examples with a wide range of in-
teractions, the applicability of a simple procedure for
incorporating polydispersity and orientational disorder
into the interpretation of SANS spectra from strongly
J. Chem. Phys., Vol. 79, No.5, 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
M. Kotlarchyk and S.H. Chen: Neutron scattering from colloids 2469
interacting colloid systems. This procedure appears to
provide very reasonable results for the four examples
given in this paper. Close inspection of the S(Q) and
S'(Q) curves shown with the respective scattering data
illustrates that the factor /3(Q) improves the quality of
the fit in each instance.
In the case of polydispersity, the suggested procedure
for determining the apparent structure factor S '(Q) is
not exact since S(Q) must be approximated by solving
the Ornstein-Zernicke equation for monodisperse par-
ticles. In addition our formalism ignores coupling be-
tween particle orientation and position for systems con-
taining nonspherical pa;ticles. However, for situations
where these approximations are not too drastic, the au-
thors feel that the methods described in this paper pro-
vide an excellent way to estimate corrections to the ap-
parent structure factor.
We acknowledge both the staff of the National Center
for Small Angle Scattering Research at the Oak Ridge
National Laboratory and Dr. B. Schoenborn of the
Brookhaven National Laboratory for their generous
allocations of spectrometer time. The microemulsion
samples were prepared by J. S. Huang and M. W. Kim
of the Exxon Research and Development Company. We
would also like to thank them for collaborating in the
experiments. This work was primarily supported by
the National Science Foundation.
APPENDIX
We show here our method for approximating the aver-
ages necessary to compute p(Q) and /3(Q) in Sec. m B. 3.
The form factor, F(a
e
, beIJ.) for a prolate spheroid with
major semiaxis a
e
and minor semiaxis be [given by Eqs.
(55)-(57)] should be averaged over both the size distri-
bution of the major axis j(a
e
) and the spheroid orienta-
tion IJ.. For example, to compute (IF(Q)1
2
), we have
(IF(Q)12)= J.""daejldIJ.IF(ae,be,IJ.)12j(ae). (Al)
o 0
The orientation average is replaced by the form factor
of a sphere with the same volume as the spheroid, i. e. ,
11 dIJ. IF(a
e
, be' IJ.) 12 "" ,
o
R = (a
e
.
Then we obtain
For a rectangular distribution,
{
l/2W
j(a
e
)= 0
lac - lie I :5 W
lac -lie I > W
(A2)
(A3)
(A5)
where
Xl = - W) ]1/3 ,
.
(A6)
(A7)
(AS)
Similarly, one obtains an approximate expression for
1 (F(Q) 12:
(A9)
Equations (A6) and (A9) can be integrated analytically
to obtain expressions for p(Q) == (I F(Q) 12) and /3(Q)
= 1 (F(Q)1
2
/(IF(Q)1
2
) in terms of lie and the polydis-
persity parameter = (W /l3li
e
)
lA. Guinier and G. Fournet, Small Angle Scattering of X-rays
(Wiley, New York, 1955).
2C. G. Shull and L. C. Roess, J. Appl. Phys. 18, 295 (1947).
3L . C. Roess and C. G. Shull, J. Appl. Phys. 18, 308 (1947).
4S. R. Aragon and R. Pecora, J. Chern. Phys. 64, 2395
(1976).
5H. A. Mook, J. Appl. Phys. 45, 43 (1974).
sL. Blum and G. Stell, J. Chern. Phys. 71, 42 (1979); 72,
2212(E) (1980).
7A. Vrij, J. Chern. Phys. 71,3267 (1979).
sp. van Beurten and A. Vrij, J. Chern. Phys. 74, 2744 (1981).
9N W. Ashcroft and n. C. Langreth, Phys. Rev. 156, 685
(1967).
lOL. S. Ornstein and F. Zernicke, Proc. Akad. Sci. 17, 793
(1914).
llJ. K. Percus and G. J. Yevick, Phys. Rev. 110, 1 (1958).
12E. Thiele, J. Chern. Phys. 39, 474 (1968).
13M. S. Wertheim, Phys. Rev. Lett. 10, 321 (1963).
14J . S. Huang and M. W. Kim, Phys. Rev. Lett. 47, 1462
(1981).
15The complete spectrum can be found in a more complete
discussion of the SANS determination of the critical behavior
of this microemulsion: M. Kotlarchyk, S. H. Chen, and J.
S. Huang, Phys. Rev. A (to be published).
lSn. Bendedouch, S. H. Chen, and W. C. Koehler, J. Phys.
Chern. (to be published).
17J. B. HayterandJ. Penfold, Mol. Phys. 42,109(1981).
18
p
. Mukerjee, K. J. Mysels, and P. Kapauan, J. Phys.
Chern. 71, 4166 (1967).
190. Bendedouch, S. H. Chen, and W. C. Koehler, J. Phys.
Chern. 87, 153 (1983).
2oA. Isihara, J. Chern. Phys. 18, 1446 (1950).
21Note that the figure mentioned also incorporates the added
complexity of including a fixed polydispersity.
22C. Tanford, Physical Chemistry of Macromolecules, 1st ed.
(Wiley, New York, 1961).
23n. Bendedouch and S. H. Chen, J. Phys. Chern. 87, 1473
(1983).
24n. Tanford and J. G. Buzzell, J. Phys. Chern. 60, 225
(1956).
J. Chern. Phys., Vol. 79, No.5, 1 September 1983
Downloaded 07 Dec 2011 to 143.167.2.135. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

Vous aimerez peut-être aussi