Content uploaded by Ronald Pindak
Author content
All content in this area was uploaded by Ronald Pindak on Jun 22, 2016
Content may be subject to copyright.
VOLUME 75, NUMBER 3 PHYSICAL REVIEW LETTERS 17JULY 1995
X-Ray Photon Correlation Spectroscopy Study of Brownian Motion of Gold Colloids in Glycerol
S.B. Dierker,1R. Pindak,2R.M. Fleming,2I.K. Robinson,3and L. Berman4
1Department of Physics, University of Michigan, Ann Arbor, Michigan 48109-1120
2AT&T Bell Laboratories, 600 Mountain Avenue, Murray Hill, New Jersey 07974-0636
3Department of Physics, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801-3080
4National Synchrotron Light Source, Brookhaven National Laboratory, Upton, New York 11973-5000
(Received 30 November 1994; revised manuscript received 5 May 1995)
We report x-ray photon correlation spectroscopy studies of the static structure factor and dynamic
correlation function of a gold colloid dispersed in the viscous liquid glycerol. We find a diffusion
coefficient for Brownian motion of the gold colloid which agrees well with that extrapolated from
measurements made with visible light, but which was determined on an optically opaque sample and in
a wave-vector range inaccessible to visible light.
PACS numbers: 61.10.Lx, 05.40.+j, 42.25.Kb, 82.70.Dd
Photon correlation spectroscopy (PCS) probes the dy-
namics of a material by analyzing the temporal correla-
tions among photons scattered by the material. Visible
light PCS has proven to be an indispensible technique
for studying the long wavelength hydrodynamics of flu-
ids, including simple liquids, liquid mixtures, liquid crys-
tals, polymers, and colloids [1]. However, visible PCS
cannot probe the short wavelength dynamics of materi-
als or opaque materials at all. The new field of x-ray PCS
(XPCS) offers an unprecedented opportunity to extend the
range of length scales over which a material’s low fre-
quency (10–3 to 106Hz) dynamics can be probed down to
interatomic spacings. It is clear that demonstration of the
ability to make dynamic XPCS measurements would be a
major step forward in this nascent field.
Many of the important problems in the low frequency
dynamics of condensed matter systems for which XPCS
should be uniquely suited arise in disordered materials, in
particular, liquids. These include studies of the dynamic
structure factor of liquids down to interatomic length
scales, density fluctuations in liquids undergoing a glass
transition, internal conformational dynamics and reptation
in polymers, and equilibrium concentration fluctuations
in polymer blends near a phase separation critical point.
Here we report results of equilibrium dynamic measure-
ments on a disordered system using XPCS. Specifically,
we measured the diffusion coefficient for Brownian mo-
tion of gold colloid particles dispersed in glycerol. Our
results unequivocally demonstrate the feasibility of the
XPCS technique, as well as illustrate the crucial impor-
tance of both matching the longitudinal coherence length
of the x rays to the experimental requirements and the
tremendous benefit of utilizing area detectors in XPCS
measurements.
While the principles of XPCS have been known for
decades, the very low flux of coherent x rays available
with previous sources has, until now, precluded its appli-
cation as a practical technique. The critical development
which has now made XPCS a feasible technique is the
use of insertion devices at second and third generation
synchrotron sources. The work reported here was done
on the wiggler beam line X25 at the Brookhaven National
Synchrotron Light Source (NSLS).
Previous work has consisted of measurements of the
static speckle patterns resulting from the structure of an-
tiphase domains in Cu3Au [2] or from grazing incidence
scattering from gold-coated polymer films of inhomoge-
neous thickness [3]. Identification of equilibrium critical
fluctuations in Fe3Al with XPCS has also been reported
very recently [4]. These studies benefited either from
the tremendous gain in scattering cross section of Bragg
scattering from essentially well-ordered materials [2,4] or
from the large reflectivity at grazing incidence [3]. In
contrast, the scattering typical of highly disordered mate-
rials, such as liquids or glasses, is dramatically weaker.
To demonstrate the difficulty of a coherent x-ray beam
by liquids, we estimate its average scattering efficiency.
For an incident x-ray beam of cross-sectional area Aiand
power P0(photonsysec) incident on a sample containing
Ntot uncorrelated atoms of charge Z, the fractional detected
count rate sPyP0dscattered into the solid angle DV is [5]
PyP0r2
0NtotZ2DVyAi. Here r0e2ymc22.8 3
10213 cm is the classical Thomson electron radius. In
terms of the density r, molecular weight M, Avogadro’s
number NA, and the scattering volume V, we have Ntot
rVNAyM. We take VAiymfor a sample of thickness
t1ym, where mrmmis the adsorption length and
mmis the mass absorption coefficient, and make use of the
approximate relationship [5] mmøCZl3NAyM, where C
is a constant. In a speckle experiment, the detector solid
angle of a single speckle is DV øl2yAi. Altogether this
gives PyP0ø8310226ZyClAi. This demonstrates the
benefit of having a high source brightness, i.e., P0yAi.
Aiis set by the transverse coherence length of the x-
ray beam. At X25, this requires a pinhole diameter of
5mm, for which Ai,20 mm2. For the case of liquid Au,
Z79, and taking l1.55 Å at 8 keV, C261 cm–1
for Au at 8 keV, and P0ø43107photonsysec at X25
with DEyE1.5%, we estimate the impracticably small
count rate of P,331024cps.
0031-9007y95y75(3)y449(4)$06.00 © 1995 The American Physical Society 449
VOLUME 75, NUMBER 3 PHYSICAL REVIEW LETTERS 17JULY 1995
However, if one had a liquid of Npgold particles, each
with naatoms (whether crystalline or amorphous, regard-
less of orientation), such that NpnaNtot, then the aver-
age scattering in the forward direction would be the sum
of the coherent scattering from Npparticles, i.e., PyP0~
Npn2
aNtotna, instead of Ntot. The qdependence of the
scattering will be that of the particle form factor [5] FsqRd
for a sphere of radius R(in the dilute gas limit at low q,
the structure factor will be unity),
FsqRd∑3hsinsqRd2qR cossqRdj
sqRd3∏2
.(1)
For example, for a Au particle with R200 Å, na,
23106, giving a sufficiently large P,600 photonsysec.
A constraint of this approach is that one must work at
very low q, although not so low as to be dominated by the
tail of the Fraunhofer diffraction of the main beam. We
estimate that measurements should be possible at least over
the range 131023,q,331022Å–1. This extends
far beyond the upper qrange of visible light scattering, for
which qmax ,431023Å–1.
Monodisperse gold sols were prepared by reducing
HAuCl4with Na3-citrate [6]. The hydrodynamic radius
RHof the resultant gold colloid particles was determined
to be 335 Å for a small diluted aliquot of the resulting
sol using visible PCS [1]. A cumulant analysis of the dy-
namic correlation functions indicated a polydispersity of
,10%. The hydrophobic gold sol was stabilized against
aggregation by absorbing cold water fish skin gelatin onto
the surface of the particles [7]. RHand the hydrodynamic
polydispersity were remeasured for the stabilized sol and
determined to be 425 Å and 20%, respectively. The in-
creased magnitude of these quantities presumably reflects
the increased drag and statistical size fluctuations of the
“fuzzy” molecular coat covering the spheres.
The as-prepared sol had a gold volume fraction of
2.7 31024%, whereas the XPCS measurements required
a sample with a gold volume fraction of ,1%.At
this concentration, the x-ray absorption of the gold and
the glycerol are approximately equal. Thus, 450 ml
of sol was concentrated by centrifuging, aspirating out
the supernatant, and resuspending in 1.5 ml H2Oby
ultrasonicating. The concentrated sol was remeasured
with visible PCS, and RHand the polydispersity were
determined to be 590 Å and 26%, respectively. The sol
was then centrifuged and the supernatant was aspirated
and resuspended by ultrasonicating in 0.10 ml glycerol
to which 0.4MNaCl had been added in order to screen
out Coulomb interactions between the spheres. This was
repeated twice more. The end result was a concentration
enhancement of 4000:1, giving an estimated gold volume
fraction of 1.3% in a 0.1 ml sol. This concentration was
confirmed by x-ray absorption measurements. We note
that at this high concentration the sample is completely
opaque to visible light, and thus could not be studied with
visible PCS. The 1% Au sample was placed in a 1.0 mm
diameter hole drilled in a 0.6 mm thick aluminum plate,
i.e., about one absorption length thick at 8 keV.
A laterally coherent beam was prepared by passing the
x rays through a 5 mm diameter pinhole. The longitudi-
nal coherence length lcoh of the x-ray beam is given by
lslyDldand was set by using a two crystal WyB4C multi-
layer monochromator, giving lcoh ,100 Å. This resulted
in an incident coherent x-ray flux of 43107photonsysec.
In a transmission geometry, the maximum path length dif-
ference (PLD) is given by 2hsinutanu, where his the
sample thickness. Thus, the requirement [1] that PLD
,lcoh is satisfied up to scattering angles 2uof 8 mrad, cor-
responding to q3.2 31022Å–1. We also attempted
measurements using a Si(220) monochromator, for which
lcoh 2mm and the incident coherent flux was only
,23105photonsysec. Data collected under those con-
ditions were of significantly poorer quality than those re-
ported here. However, small-angle scattering data col-
lected under incoherent illumination conditions using the
Si(220) monochromator were of good quality, and all such
data presented here (Figs. 1–3) were collected using this
monochromator. This underscores the crucial importance
of not having lcoh any greater than required by PLD in
order to have the maximum intensity available for the
experiment.
X-ray detectors used included a scintillation detector
and a two-dimensional charged couple device (CCD)
area detector operated in the direct detection mode. The
CCD detector was custom-built and used a Kodak KAF-
1400 CCD, which is a front illuminated device having a
512 3768 array of 9 mm square pixels.
Figure 1 shows the measured scattering intensity, both
with and without the colloid. The strong signal centered at
q0is the direct beam transmitted through the sample.
The background scattering without sample is due to small
angle scattering from the slits and pinholes in the beam
line. The overall shape and intensity of the colloid scat-
tering qualitatively matches that expected from Eq. (1).
Measurements of the static structure factor were also
made with the CCD area detector. Figure 2 shows a plot
of a CCD image taken with an 80 mm front pinhole.
The channel cut through the center of the plot is due
to the shadow of a 200 mm diameter gold wire used to
block the main beam, which would have saturated the
FIG. 1. Ssqdas measured with incoherent x rays by scanning
an 80 mm pinhole followed by a scintillation detector. The
incident beam diameter was 200 mm. One contribution to the
background, a Kapton window, is indicated.
450
VOLUME 75, NUMBER 3 PHYSICAL REVIEW LETTERS 17JULY 1995
FIG. 2. Log plot of scattering intensity from the colloid
collected with the CCD camera.
detector. The CCD image was obtained by summing ten
1 sec exposures in order to extend the CCD’s dynamic
range. In comparison, each line scan with the pinhole-
scintillation detector combination over a similar qrange
took more than an hour.
A comparison of the structure factor measured in the
scintillation detector scans with that deduced from radial
averages of the CCD image is shown in Fig. 3. The
data have been scaled to coincide at q,0.015 Å–1.
The scintillation detector and CCD scans show good
general agreement apart from differences in the amount
of forward scatter around the main beam between the two
measurements.
In order to determine the size of the protected colloid
spheres and the degree of polydispersity, the scintillation
detector data shown in Fig. 3 were fit by a model
consisting of a Gaussian distribution of sphere scattering
radii, with each sphere according to Eq. (1), along with
FIG. 3. Comparison of Ssqdderived from scintillator or
pinhole scans with that from the CCD image in Fig. 2. Also
shown are the results of a fit of a model for Ssqdbased on
Eq. (1) and a Gaussian distribution of particle radii.
a constant background. The results indicate that the
mean radius was 270 Å and the standard deviation of
the distribution was 57 Å. These numbers agree well
with the mean hydrodynamic radius and polydispersity
measured on dilute aqueous solutions of the colloid with
visible PCS before protecting them with gelatin. This
suggests that the increased RHmeasured for the protected
colloid results from the added hydrodynamic drag of
the macromolecular coating. The further increase in RH
upon centrifuging presumably results from changes in the
molecular conformation of the gelatin molecules and not
from aggregation of the spheres.
We next turn to consideration of the Brownian motion
dynamics of this system. The Au colloid particles form
a dilute gas having diffusively relaxing concentration
fluctuation with a corresponding diffusion coefficient
given by [1] DkbTy6phRHand a relaxation time at
wave vector qof t1yDq2. Glycerol is a prototypical
glass former with a glass transition at Tg286 ±
C, at
which the shear viscosity happroaches ,1013 P [8].
The dynamic correlation function gstdis defined as
gstdknstdns0dlykns0dl2,(2)
where kl represents a time average and nstdis the
detected photon count at time t.gstdshould have a
maximum value of 2 at t0, decaying to 1 at infinite
time. Spatial averaging of the speckles by the detector
can reduce the t0intercept of gstd.
A dramatic increase in collection efficiency was ob-
tained by using the CCD detector to measure the scatter-
ing for 6750 pixels simultaneously. Since the dynamics
depend only on the magnitude of q, we can average the
autocorrelation functions measured in all of the pixels in a
band of qvalues. This is equivalent to performing an en-
semble average over pixels, as well as a time average for
each pixel, and should reduce the time needed to measure
the correlation functions with good statistics by the num-
ber of pixels averaged over [9]. With 6750 pixelsyimage,
the reduction is substantial.
Specifically, we recorded 1920 images of the inten-
sity in two separate 90±arcs corresponding to q’s of
3.3 31023and 5.5 31023Å–1, with widths of 10% of
their average q’s and containing 1750 and 5000 pixels,
respectively (see Fig. 4). The larger qis well above the
maximum value of ,431023Å–1 attainable by visible
PCS. The images were taken once per second for a to-
tal duration of 32 min, with the exact time of exposure
recorded for each image. A total of ,3.4 3106pho-
tons were detected, corresponding to an average count rate
of 0.36 photonsypixel sec. This compares well with the
value of 0.68 derived from our earlier estimate when the
sample transmission (0.35), the ratio of the area of a pixel
to that of a speckle (0.13), the estimated quantum effi-
ciency of the CCD detector (0.05), and Eq. (1) (0.5) are
taken into account.
The time autocorrelation function of each pixel was then
calculated and the ensemble average of the resulting 1750
451
VOLUME 75, NUMBER 3 PHYSICAL REVIEW LETTERS 17JULY 1995
FIG. 4. Autocorrelation functions of colloid scattering inten-
sity as collected with the CCD camera for two different wave
vectors along with single exponential fits and relaxation times,
as indicated. The inset schematically depicts the scattering
“halo,” the beam stop, and the subarrays which were ensem-
ble averaged over. Curve (b) is offset by 0.15 for clarity.
or 5000 correlation functions calculated. The results are
shown in Fig. 4, which is plotted as gstd21. The large
signal-to-noise ratio of the data in Fig. 4 can leave no doubt
that they correspond to dynamic x-ray scattering from
colloid concentration fluctuations. The large amplitude of
the decays is also as expected considering the small size of
the pixels relative to the coherence area for the scattering.
Also shown in Fig. 4 are fits to the data of single
exponential relaxations. They give characteristic decay
times of 43.1 and 24.1 sec for q3.3 31023and 5.5 3
1023Å–1, respectively. The measurements were made at
a temperature of 229 ±
C. The viscosity of pure glycerol
[8] should be 5.5 3103P at this temperature. Thus,
the nominal expected relaxation times are 61 and 22 sec,
which are in good agreement with the measurements.
The deviation of the relaxation rates from a precise
q2dependence may be an indication of the onset of
nonhydrodynamic effects at large wave vector [10].
While these measurements clearly establish the feasi-
bility of XPCS studies, more experiments are called for
in order to fully understand the dynamics of this colloid.
For example, nonhydrodynamic effects at large wave vec-
tor [10] as well as nonlinear relaxation effects predicted
by mode coupling theories [11] near the glass transition
should be fruitful areas of future study with this new
technique.
In conclusion, we have made small angle x-ray scat-
tering measurements of the static structure factor of an
optically opaque gold colloid dispersed in glycerol. We
also reported a determination of dynamic correlation func-
tions and the diffusion coefficient for the Brownian mo-
tion of the colloid particles, at a wavelength inaccessible
to visible PCS, by using the new technique of XPCS. Our
results demonstrate the importance of using a multilayer
monochromator, or just the natural bandwidth of undula-
tor radiation (,1%), in situations where a large lcoh is not
required, such as surface diffraction or small-angle scat-
tering experiments, as well as the benefits of ensemble av-
eraging with CCD area detectors in XPCS measurements.
The present results clearly show that XPCS is a fea-
sible technique and can have wide application to cer-
tain classes of systems, especially complex fluids and
surfaces, even using existing 2nd generation synchrotron
x-ray sources such as the NSLS. This is particularly
true given several significant forthcoming enhancements:
(1) Optimization of the use of an area detector for making
ensemble measurements of gstd:3101–102. (2) Beam-
line optics which utilize the full coherent output of the
undulator in both the vertical and horizontal directions:
310. (3) Introduction of undulator sources such as the
Prototype Small Gap Undulator recently installed at the
NSLS: 310. These will enhance the data collection effi-
ciency by 103and 104, relative to the current experiment.
Those classes of problems demanding the utmost coherent
incident x-ray flux will require the 3rd generation undula-
tor sources at the European Synchrotron Radiation Facility
and the Advanced Photon Source: 310 2–103, for a total
gain of 104–106, relative to the current experiment. When
these developments come to fruition, XPCS will realize its
full potential as a unique and important new technique.
It is a pleasure to acknowledge the assistance and
advice of Christoph Schmidt on the preparation of the
gold colloids. Robert MacHarrie assisted in the x-
ray measurements. S.B. D. acknowledges support under
NSF Grant No. DMR 92-17956 and also the donors
of the Petroleum Research Fund, administered by the
American Chemical Society, for the partial support of this
research under Grant ACS-PRF No. 26389-AC9. I. K. R.
acknowledges support under NSF Grant No. DMR 93-
15691. The NSLS is supported by the U.S. Department
of Energy under Contract No. DE-AC02-76CH00016.
[1] See, for example, Benjamin Chu, Laser Light Scattering:
Basic Principles and Practices (Academic Press, San
Diego, 1991), 2nd ed.
[2] M. Sutton et al., Nature (London) 352, 608 (1991).
[3] Z.H. Cai et al., Phys. Rev. Lett. 73, 82 (1994).
[4] S. Brauer et al., Phys. Rev. Lett. 74, 2010 (1995).
[5] G.B. E. Warren, X-Ray Diffraction (Addison-Wesley,
New York, 1968).
[6] G. Frens, Nature (London) 241, 20 (1973).
[7] Colloidal Gold: Principles, Methods, and Applications,
edited by M.A. Hayat (Academic Press, San Diego,
1989).
[8] M.L. Williams et al., J. Phys. Chem. 77, 3701 (1955).
[9] Apollo P.Y. Wong and P. Wiltzius, Rev. Sci. Instrum. 64,
2547 (1993).
[10] W. Hess and R. Klein, Adv. Phys. 32, 214 (1983).
[11] J. Jackle, J. Phys. Condens. Matter 1, 267 (1989).
452