J Nanopart Res (2011) 13:6579–6589

DOI 10.1007/s11051-011-0565-y

RESEARCH PAPER

Adsorption kinetics of alkanethiol-capped gold

nanoparticles at the hexane–water interface
Sultana Ferdous • Marios A. Ioannidis •

Dale Henneke

Received: 22 March 2011 / Accepted: 29 August 2011 / Published online: 17 September 2011
Ó Springer Science+Business Media B.V. 2011

Abstract The pendant drop technique was used to Phys 9:6351–6358, 2007). The experiments addition-
characterize the adsorption behavior of n-dodecane-1- ally demonstrated the important role played by the
thiol and n-hexane-1-thiol-capped gold nanoparticles capping agent. At the same concentration, gold
at the hexane–water interface. The adsorption process nanoparticles stabilized by n-hexane-1-thiol exhibited
was studied by analyzing the dynamic interfacial greater surface activity than gold nanoparticles of the
tension versus nanoparticle concentration, both at same size stabilized by n-dodecane-1-thiol. These
early times and at later stages (i.e., immediately after findings contribute to the design of useful supra-
the interface between the fluids is made and once colloidal structures by the self-assembly of alkane-
equilibrium has been established). A series of gold thiol-capped gold nanoparticles at liquid–liquid
colloids were made using nanoparticles ranging in interfaces.
size from 1.60 to 2.85 nm dissolved in hexane for the
interfacial tension analysis. Following free diffusion Keywords Dynamic interfacial tension 
of nanoparticles from the bulk hexane phase, adsorp- Nanoparticle  Ligand  Diffusion 
tion leads to ordering and rearrangement of the Adsorption barrier  Surface science
nanoparticles at the interface and formation of a
dense monolayer. With increasing interfacial cover-
age, the diffusion-controlled adsorption for the nano- Introduction
particles at the interface was found to change to an
interaction-controlled assembly and the presence of The behavior of nanoparticles at fluid–fluid interfaces
an adsorption barrier was experimentally verified. has attracted significant attention in recent years, as a
At the same bulk concentration, different sizes of result of a drive to devise strategies for producing
n-dodecane-1-thiol nanoparticles showed different novel functional materials or devices via size-selective
absorption behavior at the interface, in agreement particle self-assembly (Binks and Horozov 2006).
with the findings of Kutuzov et al. (Phys Chem Chem Binding of particles at liquid–liquid interfaces, which
has been known to stabilize Pickering emulsions, has
been explained in terms of a decrease of the free
energy (Pieranski 1980). Since, placement of a single
spherical particle of radius r at the oil–water interface
S. Ferdous  M. A. Ioannidis  D. Henneke (&)
decreases the entropy by a about kB , the interfacial
Department of Chemical Engineering,
University of Waterloo, Waterloo, ON, Canada energy change, DE, must be negative for placement of
e-mail: henneke@uwaterloo.ca the particle to be thermodynamically favoured. The

123
6580 J Nanopart Res (2011) 13:6579–6589

energy change associated with placement of a single adsorption have been reported. Glaser et al. (2006)
particle at the oil–water interface depends on the used pendant drop tensiometry to demonstrate the
particle radius r and on the particle–oil (cp/o), particle– interfacial activity of Janus nanoparticles at the
water (cp/w), and oil–water (co/w) interfacial tensions hexane/water interface. Using the same method,
as follows: Kutuzov et al. (2007) studied the time evolution of
toluene–water interfacial tension as tri-n-octylphos-
pr 2 h  i2
DE ¼  co=w cp=w  cp=o : ð1Þ phine oxide (TOPO)-stabilized CdSe nanoparticles,
co=w initially suspended in toluene, diffused toward and
An extension of this result to include the influence were adsorbed at the toluene/water interface. Very
of line tension has been given by Aveyard and Clint recently, Isa et al. (2010) also used pendant drop
(1996). This extension was recently tested by molec- tensiometry in a qualitative study of self-assembly of
ular simulation (Cheung 2009). A thorough review of iron oxide poly(ethylene glycol) core–shell nanopar-
the thermodynamics of nanoparticle attachment at ticles at the n-decane/water interface.
fluid–fluid interfaces has been provided by Bresme Of the above mentioned studies, only the study of
and Oettel (2007). Kutuzov et al. (2007) addressed quantitatively the
Equation 1 highlights a key difference between kinetics of ligand-stabilized nanoparticle adsorption.
micrometer- and nanometer-sized particles—the latter These authors analyzed the time evolution of the
being much less stably adsorbed at liquid–liquid interfacial tension using the theory of Ward and Tordai
interfaces. For typical interfacial tension values, the (1946) to infer the characteristics of the adsorption
magnitude of DE is of the order of kBT for particles process at early and late stages. For TOPO-stabilized
with radius of a few nanometer. As a result, binding of CdSe nanoparticles in the size range 2.3–6 nm,
very small nanoparticles at fluid–fluid interfaces is Kutuzov et al. (2007) found the decay of interfacial
expected to be destabilized by thermal fluctuations tension to be consistent with a mixed diffusion-
(Binks 2002) and this expectation is confirmed by activation adsorption mechanism. Using a model
experiment (Lin 2003). On the contrary, attachment developed by Liggieri et al. (1996) to describe the
of micrometer-sized particles to fluid–fluid interfaces adsorption kinetics of non-ionic surfactants, they
may be considered irreversible, due to the fact that DE estimated a magnitude of the energy barrier to
exceeds kBT by many orders of magnitude (Binks adsorption of a few kBT, approximately equal to the
2002). Using Young’s equation, it can be also shown desorption energy for a single particle calculated from
that the bracketed term in Eq. 1 is maximum when the Eq. 1. They proposed that such an adsorption barrier
contact angle h of the particle with the fluid–fluid originates from particle–particle interactions near the
interface is 90° and decreases rapidly as h ! 0 (or oil–water interface (collisions of the nanoparticles
h ! 180 ), highlighting the influence of particle approaching the interface from the bulk with nano-
wettability on the stability of particle attachment to particles that are already adsorbed or that desorb from
the interface (Binks 2002). the interface due to thermal fluctuations), which
Although, a large body of literature exists on the use become increasingly important with increasing inter-
of liquid–liquid interfaces as the locus of nanoparticle facial coverage. It is worth noting that a mixed
self-assembly into a variety of supra-colloidal struc- diffusion-activation adsorption mechanism has been
tures with applications in sensing, encapsulation, data found to apply to many different surfactants, with a
storage, and catalysis (Lin 2003; Boeker 2007; Patra small adsorption barrier always in the range 2–5 kBT
2009; Krishnaswamy 2010; Crossley 2010), much less regardless of molecular weight, structure or critical
is known about the kinetics of nanoparticle adsorption micelle concentration of the surfactant—an observa-
at liquid–liquid interfaces. Dynamic interfacial ten- tion consistent with the idea that the adsorption barrier
sion measurements, made for example by pendant is related to the work, Wads, that a molecule must do
drop tensiometry, are well suited for this task. This against surface pressure (Eastoe 1997). The latter may
technique has been used extensively for interrogating be estimated as , Wads ¼ ðco  c1 ÞA  1023 , where
the adsorption kinetics of surfactants on liquid inter- ðco  c1 Þ is the equilibrium surface pressure (in mN/
faces, but only a few applications to nanoparticle m) and A is the area (in Å2) at the fluid–fluid interface

123
J Nanopart Res (2011) 13:6579–6589 6581

occupied by one adsorbed molecule. For adsorbed Table 1 Synthesis conditions for the nanoparticle suspensions
particles, A ¼ pðr sin hÞ2 from geometrical consider- Condition Ratio of gold atoms Cap
ations. Insofar as nanoparticles are considered, direct to capping molecules
evidence of an energy barrier to adsorption on liquid–
(a) 1:8 Dodecanethiol
liquid interfaces is at present too limited to establish
(b) 6:1 Dodecanethiol
mixed diffusion-activation as the operative adsorption
(c) 6:1 Hexanethiol
mechanism, let alone elucidate the physical origin of
an adsorption barrier. Gold nanoparticles capped by
alkanethiol self-assembled monolayers are easily
synthesized and are known to exhibit remarkable were free of excess capping ligand. The size distribu-
stability and ease of functionalization. Not surpris- tions of the synthesized nanoparticles were deter-
ingly, they are some of the most studied nanomaterials mined by analyzing TEM images of the material. UV–
(Zhou 2009). Yet, an investigation of the adsorption Vis and proton NMR measurements were then
kinetics of thiol-capped gold nanoparticles at liquid– performed to confirm that free ligands were com-
liquid interfaces is lacking, despite the fact that the pletely absent in the colloidal suspensions used to
adsorption process is a prerequisite for self-assembly measure nanoparticle adsorption at the hexane–water
of these nanoparticles into useful supra-colloidal interface.
structures. Such an investigation is reported in this
study. Similar to Kutuzov et al. (2007), we also High-resolution transmission electron microscopy
employ pendant drop tensiometry to interrogate the (HR-TEM) analysis
kinetics of alkanethiol-stabilized gold nanoparticles at
the hexane–water interface. The effects of nanoparti- HR-TEM was performed using a FEI Titan micro-
cle size, concentration in the bulk organic phase and scope at 300 kV. A drop of the synthesized colloid was
ligand chain length (n-dodecane-1-thiol vs. n-hexane- placed onto a 400 mesh copper TEM grid coated with
1-thiol) on adsorption behavior are investigated. The formvar. The samples were dried in air for at least 1 h
results corroborate the finding of Kutuzov et al. before analysis.
(2007) that a mixed diffusion-activation mechanism Figure 1 shows representative TEM images and
governs nanoparticle adsorption at a liquid–liquid size distribution data for nanoparticles synthesized
interface. under conditions (a), (b), and (c), respectively, and
capped with n-dodecane-1-thiol and n-hexane-1-thiol.
Nanoparticle diameters were determined by image
Experimental analysis using the ImageJ software and the resulting
data were fitted by log-normal distributions. The
Nanoparticle synthesis, preparation, average size of the n-dodecane-1-thiol smaller nano-
and characterization particles synthesized under condition (a) was found
to be l = 1.60 nm with a standard deviation of
A series of gold nanoparticle colloids were synthe- r = 0.278 nm, while the average size of the n-
sized using a technique established by Brust et al. dodecane-1-thiol-capped larger nanoparticles synthe-
(1994) and subsequently modified by Hostetler et al. sized under condition (b) was larger than those of
(1998). The nanoparticle colloids were synthesized at condition (a); they were found to be l = 2.78 nm
30 °C using the experimental conditions shown in with a standard deviation of r = 0.273 nm. The
Table 1. During the synthesis, either n-hexane-1-thiol average size of the n-hexane-1-thiol nanoparticles
or n-dodecane-1-thiol was added to the colloidal was found to be l = 2.85 nm with a standard devi-
solution to act as a capping agent, resulting in a total of ation of r = 0.188 nm.
three colloidal suspensions.
To avoid interference from free surfactant mole- UV–Vis absorbance spectra
cules, the synthesized nanoparticles were rinsed with
ethanol and re-dispersed in hexane. This process was An Ocean Optics USB2000? UV–VIS Absorbance
repeated 25–30 times to ensure that the nanoparticles Spectrophotometer was used to obtain the spectra of

123
6582 J Nanopart Res (2011) 13:6579–6589

5
35
=1.60 nm

Absorbance [arbitrary]
30
=0.278 nm 4
25
20
15 3
10
2
c
5
0
0 1 2 3 4
1
b
20 nm Core diameter [nm] a
(i) 0
300 400 500 600 700 800 900

20
Wavelength[nm]
=2.78 nm
15 =0.273 nm Fig. 2 UV–Vis spectra of nanoparticles prepared under condi-
tions (a), (b), and (c); a and b are capped with n-dodecane-1-
10 thiol, while c is capped with n-hexane-1-thiol
5

0
0 1 2 3 4
image analysis. Hostetler et al. also found a broad SP
20 nm Core diameter [nm] band superimposed on the background *500 nm.
(ii) Alvarez et al asserted that SP bands are undetectable
for the gold nanoparticles having diameters less than
20 2.0 nm.
=2.85 nm
15
=0.188 nm
NMR spectra
10

5 Proton NMR (1H-NMR) spectra were taken using a

0 BRUKER 500 Shield TM spectrometer. NMR spectra
0 1 2 3 4
20 nm of stock n-dodecane-1-thiol solution and suspensions
Core diameter [nm]
of nanoparticles capped with n-dodecane-1-thiol for
(iii)
conditions (a) and (b) are shown in Fig. 3. The stock n-
Fig. 1 TEM image and associated size distribution for nanopar- dodecane-1-thiol was prepared by dissolving 1 mL of
ticles synthesized using conditions (a), (b) capped with pure dodecanethiol into 1.5 mL deuterated chloroform
n-dodecane-1-thiol and, under condition (c) capped with (CDCl3). NMR samples of the synthesized nanopar-
n-hexane-1-thiol are shown respectively in (i), (ii), and (iii). The
curves in the distribution data represent a log-normal fit to the data
ticles were made by suspending *14 mg of the
nanoparticles into 1.5 mL of CDCl3. The resulting
concentrations of suspensions (a) and (b) for this
the nanoparticle colloids. Pure hexane, as supplied by analysis were 1.31 9 1020 and 2.54 9 1019 particles/
Sigma-Aldrich, was used for the reference spectrum in L, respectively.
these measurements. The spectra, which were taken in For nanoparticles synthesized under condition (a),
the range of 250–850 nm, can be seen in Fig. 2. Fig. 3(i), shows three multiplets located at 1.546, 1.26,
From this figure, it is evident that a broad surface and 0.892 PPM. The NMR spectra obtained for
plasmon (SP) band is superimposed over the back- nanoparticles synthesized under experimental condi-
ground at *500 nm. The SP band decreases as the tion (b) yielded three broad multiplets at 1.547, 1.266,
nanoparticle size is reduced. The expected plasmon and 0.893 PPM. These results are also consistent
resonance peak for the gold nanoparticles is not with Hostetler et al. (1998). The resulting spectra for
detectable for condition (a) or for the smallest particle the capped nanoparticles differ from the spectrum
size. A single plasmon resonance peak is just visible obtained for stock n-dodecane-1-thiol solution. For the
for conditions (b) and (c) at *510 and *518 nm, capped nanoparticles, all of the observed peaks are
respectively. This result is consistent with the study of broader than those of stock n-dodecane-1-thiol sam-
Hostetler et al. (1998) and Alvarez et al. (1997), and ple. It is known that an NMR peak at 2.58 PPM is
matches with the estimates of nanoparticle size by associated with a proton attached to the –SH group.

123
J Nanopart Res (2011) 13:6579–6589 6583

The measurement for the stock n-hexane-1-thiol was

prepared by dissolving 1 mL of the hexanethiol into
Count [arbitrary units]

a 1.5 mL CDCl3. NMR samples of the synthesized

nanoparticles were made by dissolving *10 mg of the
nanoparticles into 1.5 mL of CDCl3. The resulting
b particle concentration was 3.66 9 1019 particles/L. In
Fig. 3(ii), the peaks for the stock n-hexane-1-thiol are
found at 0.933, 1.33, 1.648, and 2.57 PPM. Three
Stock dodecane thiol multiplets at 1.55, 1.27, and 0.856 PPM are seen for
3 2 1 0 -1
the n-hexane-1-thiol-capped gold nanoparticles. The
Frequency [PPM]
NMR peak at 2.57 PPM is associated with a proton
attached to the –SH group. This peak is observed in the
(i)
spectra of the stock n-hexane-1-thiol, but it is not seen
in the spectra for the capped nanoparticles. These
results confirm that the thiol group is completely
Count [arbitrary units]

attached to the gold nanoparticle surface. Here, greater

proximity of the –CH3 group to the nanoparticle
c
surface results in more significant shift from 0.933 to
0.856 PPM.

Dynamic interfacial tension measurements

Stock hexane thiol
The effect of nanoparticles on the dynamic interfacial
3 2 1 0 -1
tension was investigated using a series of colloids of
Frequency [PPM]
varying nanoparticle concentration, achieved by dilut-
(ii)
ing each of the three kinds of synthesized nanoparti-
Fig. 3 (i) NMR spectra of stock n-dodecane-1-thiol and cles in pure hexane. A drop of deionized ultra-pure
nanoparticles produced under conditions (a) and (b) (i.e., water (  18:2 MX cm) was formed at the end of a steel
nanoparticles capped with n-dodecane-1-thiol). (ii) NMR needle placed into the colloidal suspension and images
spectra of stock n-hexane-1-thiol and nanoparticles produced
under condition (c) (i.e., nanoparticles capped with n-hexane-1- of this drop were recorded over time. Image aquisition
thiol) and determination of dynamic interfacial tension by
axisymmetric drop shape analysis (ADSA) were
performed using the VCA 2500 XE equipment and
This peak can be seen in the spectra of the stock n- software (AST Products, Billerica, MA).
dodecane-1-thiol, but is not observed in the spectra for
the capped nanoparticles. The lack of this peak Calibration and verification of ADSA system
indicates that the thiol group is attached to the gold
nanoparticle surface, thus quenching the 2.58 PPM All glassware, syringes, and equipment were well
peak associated with the proton of the thiol. The NMR cleaned using both acetone and tetrahydrofuran.
peak at 0.88 PPM in the stock n-dodecane-1-thiol is Before use in the ADSA measurements, the hexane
associated with the terminal –CH3 group. In the (Sigma-Aldrich, C99% purity) was shaken with a
capped nanoparticle samples, this peak is not signif- nearly equal volume of ultrapure water for 30 min to
icantly shifted (i.e., only a small shift from 0.88 to achieve equilibrium between the two phases. The
*0.89 PPM is observed). Such a small shift indicates water was then removed and the process repeated a
that the methyl group of the n-dodecance-1-thiol cap is total of three times to remove trace impurities from
far from the nanoparticle surface. hexane. The interfacial tension of pure water droplets
The NMR spectra of both stock n-hexane-1-thiol in hexane treated in this manner was then measured. A
and nanoparticles capped with n-hexane-1-thiol pro- constant interfacial tension value of 51.2 ± 0.4 mN/m
duced under condition (c) are shown in Fig. 3(ii). was found at a temperature of 295.5 K for these

123
6584 J Nanopart Res (2011) 13:6579–6589

droplets, which is consistent with the literature value next to the hexane–water interface) and D is the
of 51.4 mN/m reported by Goebel and Lunkenheimer diffusion coefficient of nanoparticles in the bulk
(1997). Before any interfacial measurement, the phase. The asymptotic forms of the time dependence
equipment was checked by measuring the interfacial of dynamic interfacial tension for t ! 0 and t ! 1
tension of a pure water droplet in equilibrated are given by Fainerman (1994):
hexane to verify that the measured interfacial rffiffiffiffiffi
tension had not deviated from the established value Dt
cðtÞ ¼ c0  2RTC ð6Þ
of 51.2 mN/m. p
rffiffiffiffi
dc D
Theoretical considerations þ 1 ¼ 2RTC ð7Þ
dðt 2 Þt!0 p
rffiffiffiffiffiffiffiffi
Equilibrium between nanoparticles in the bulk and RT  C2 p
cðtÞ ¼ c1 þ ð8Þ
nanoparticles adsorbed at the hexane–water interface C 4Dt
is assumed to obey the Langmuir equilibrium rffiffiffiffiffiffi
dc RT  C2 p
isotherm:  1 ¼ ð9Þ
dðt Þt!1
2 C 4D
C
C ¼ Cmax  ð2Þ where c1 is the interfacial tension at equilibrium (i.e.,
aL þ C
as t ! 1). The latter quantity is obtained from the
where Cmax is the monolayer capacity of the liquid– intercept of a plot of long-time dynamic interfacial
liquid interface, reflecting the maximum amount of 1
tension data against t2 : The Langmuir isotherm
nanoparticles that can be adsorbed, aL is the Langmuir
parameters can be estimated from a fit of the
parameter, and C is the bulk nanoparticle concentra-
Langmuir–Szyszkowski equation to equilibrium sur-
tion at equilibrium. Assuming ideal behavior, an
face pressure data, c0  c1 :
assumption valid at low nanoparticle concentrations,
the adsorption density and bulk nanoparticle concen-
tration at equilibrium are related by the Gibbs
Results
adsorption equation:
1 dc Effect of particle concentration and size
C¼ : ð3Þ
RT dðln CÞ on the interfacial properties
Combination of Eq. 2 and 3 yields the well-known
Time-dependent interfacial tension measurements
Langmuir–Szyszkowski equation which relates
were performed on the hexane–water interface. For
explicitly interfacial tension to nanoparticle concen-
these measurements, n-dodecane-1-thiol and n-hex-
tration in the bulk phase (Szyskowski 1908)
  ane-1-thiol-capped nanoparticles were suspended in
C the hexane phase before the formation of the interface.
c ¼ c0  RT  Cmax  ln 1 þ ð4Þ
aL It was found that the dynamic interfacial tension
decreases with time for all cases as shown below.
where c0 is the interfacial tension of the hexane–water
interface in the absence of nanoparticles. When
Concentration effect on the dynamic interfacial
diffusion controls the adsorption kinetics of gold
tension for n-dodecane-1-thiol-capped gold
nanoparticles, the theory of Ward and Tordai applies
nanoparticles
(1946):
0 pffi 1
rffiffiffiffi Zt Figure 4 illustrates that at higher particle concentra-
DB pffi pffi C
tions, the rate at which the interfacial tension
CðtÞ ¼ 2 @C t  Cs ð0; t  sÞdð tÞA ð5Þ
p decreases is greater and a lower interfacial tension is
0
reached at equilibrium. For each observed concentra-
where Cs is the nanoparticle concentration in the tion, the equilibrium dynamic interfacial tension (i.e.,
subsurface (a region of the bulk solution immediately t ! 1) was obtained by fitting straight lines to c

123
J Nanopart Res (2011) 13:6579–6589 6585

50

Equilibrium Tension
40

36

[mN/m ]
Interfacial Tension [mN/m]

45
32

28
Critical Concentration = 8.28 x 1016 [#/L]
40
24
6
-20 -19 -18 -17 -16 -15
5
35 ln (C [mol/L])
4
3 Fig. 5 Extrapolated equilibrium interfacial tension of
2
30 1 2.8 ± 0.27 nm n-dodecane-1-thiol-capped gold nanoparticles
showing the critical concentration. The straight line represents a
fit on the Gibbs adsorption isotherm
25
0 800 1600 2400 3200 4000
Time [sec] Table 2 Adsorption behavior for n-dodecane-1-thiol-capped
gold nanoparticles synthesized under condition (a) at the hex-
Fig. 4 Interfacial tension of a water droplet in hexane with ane–water interface
2.8 ± 0.27 nm n-dodecane-1-thiol-capped gold nanoparticles at Nanoparticle Equilibrium Diffusion coefficient (m2/s)
various concentrations: 1 1.27 9 1017 particles/L, 2 8.48 9 concentration int. tension
1016 particles/L, 3 4.24 9 1016 particles/L, 4 2.12 9 1016 parti- C (particles/L) c1 ðmN=mÞ D0 D1
cles/L, 5 8.48 9 1015 particles/L, and 6 4.24 9 1015 particles/L
2.12 9 1015 40.70 4.42 9 10-10 1.98 9 10-10
8.51 9 1015 35.58 8.14 9 10-11 5.70 9 10-12
1
versus t2 data at late times and reading c1 as the 2.12 9 10 16
32.85 9.90 9 10 -12
3.54 9 10-13
intercept (cf. Eq. 8). These results can be seen in 4.26 9 10 16
30.90 3.63 9 10 -12
1.56 9 10-13
Tables 2 and 3. Semi-log plot of the equilibrium 2.12 9 10 17
26.30 2.57 9 10 -13
7.31 9 10-15
interfacial tension versus nanoparticle concentration is 2.55 9 10 17
26.07 1.39 9 10 -13
2.73 9 10-15
shown in Fig. 5. It is evident that as the nanoparticle
These particles were sized using a log-normal fit, resulting in a
concentration increases, the equilibrium interfacial mean particles size of l = 1.60 nm with a standard deviation
tension decreases until a critical concentration, c0 , is of r = 0.278 nm
reached for which the interface reaches maximum
coverage, C1 , by absorbed nanoparticles (i.e., the Table 3 Adsorption behavior for n-dodecane-1-thiol-capped
surface is saturated). The values of c0 for smaller and gold nanoparticles synthesized under condition (b) at the
hexane–water interface
larger dodecanethiol particles are 2.12 9 1017 and
8.28 9 1016 particles/L, respectively. Nanoparticle Equilibrium Diffusion coefficient (m2/s)
concentration int. tension
C (particles/L) c1 ðmN=mÞ D0 D1
Concentration effect on the dynamic interfacial
tension for n-hexane-1-thiol-capped gold 4.24 9 1015 36.47 3.54 9 10-10 5.16 9 10-12
nanoparticles 8.48 9 1015 34.60 1.29 9 10-10 1.41 9 10-12
16 -11
2.12 9 10 32.30 1.37 9 10 3.72 9 10-13
As with Fig. 4, Fig. 6 also shows that there is an 4.24 9 10 16
31.0 3.52 9 10 -12
9.16 9 10-14
effect on the interfacial tension associated with 8.48 9 10 16
29.69 7.56 9 10 -13
1.07 9 10-14
nanoparticle concentration. The rate at which the 1.27 9 10 17
29.62 2.24 9 10 -13
6.55 9 10-15
interfacial tension decreased was also found to be
These particles were sized using a log-normal fit, resulting in a
higher for n-hexane-1-thiol-capped nanoparticles as mean particles size of l = 2.78 nm with a standard deviation
concentration was increased. Again, the equilibrium of r = 0.273 nm
interfacial tension was found to be lower at higher
nanoparticle concentrations. With an increase in the Size effect on the dynamic interfacial tension
nanoparticle concentration, the equilibrium interfa-
cial tension also decreased until the critical concen- For larger dodecanethiol particles, the interfacial
tration, c0 = 8.61 9 1016 particles/L, was reached. tension falls sharply when the droplet is initially

123
6586 J Nanopart Res (2011) 13:6579–6589

formed. The interfacial tension also falls for smaller 50

nanoparticles, but not as sharply as for the larger ones.

As soon as, the water droplet is formed in the hexane–

Interfacial Tension [mN/m]

45
Au suspension, nanoparticles go to the interface. The
total free energy of the system decreases as a result.
The change in free energy associated with adsorption 40

of a single particle is given in Eq. 1, which shows that

for constant temperature, co/w, cp/w, and cp/o, larger 5
35
particles are more stably attached at the liquid–liquid
4
interface. On the contrary, smaller particles are more
30 3
easily displaced as a result of thermal fluctuations. For
a given bulk concentration, greater surface coverage is 2
1
initially achieved by adsorption of larger particles, 25
0 800 1600 2400 3200 4000
hence a greater initial reduction of the interfacial
Time [sec]
tension is observed. As t ! 1; equilibrium coverage
of the interface is achieved faster for larger nanopar- Fig. 6 Plot of the interfacial tension of a water droplet in
ticles. For the same bulk concentration, the equilib- hexane with 2.9 ± 0.19 nm n-hexane-1-thiol-capped gold
rium interfacial tension is slightly smaller for the nanoparticles at various concentrations: 1 1.30 9 1017 parti-
cles/L, 2 8.65 9 1016 particles/L, 3 2.16 9 1016 particles/L, 4
larger nanoparticles. Similar observations have been 8.61 9 1015 particles/L, and 5 4.32 9 1015 particles/L
made by Kutuzov et al. (2007).
adsorption density, C1 and Langmuir parameter, aL.
Effect of nature of the capping agent The estimated parameters for all the nanoparticles can
on the interfacial properties be seen in Table 4. It is instructive to compare the
experimental value of C1 to what would be expected
Figure 6 demonstrates a significant effect of the chain for some simple closely packed arrangements of
length of alkanethiol ligands on adsorption behavior. particles at the liquid–liquid interface. For the smaller
Specifically, at the same bulk concentration of dodecanethiol particles,the experimental value, C1 is
(8.55 ± 0.06) 9 1015 particles/L, the longer n-dode- thus consistent with a square arrangement value, Csq :
cane-1-thiol is apparently more effective at shielding While, for the larger dodecanethiol and hexanethiol
the polar gold core than n-hexane-1-thiol, thereby particles, the triangular arrangement, Ctri value is
decreasing the hydrophilic character and reducing the consistent with experiment. Albeit rough, these esti-
surface activity of n-dodecane-1-thiol-stabilized gold mates indicate a closely packed arrangement of the
nanoparticles relative to nanoparticles of the same size adsorbed nanoparticles at the hexane–water interface,
that are stabilized with n-hexane-1-thiol. Figures 4 as also found by Kutuzov et al. (2007) for a different
and 6 show that a significantly lower value of system. Figure 7 shows fits of the Langmuir isotherm
interfacial tension is reached following the adsorption to experimental data for larger dodecanethiol nano-
of n-hexane-1-thiol-stabilized gold nanoparticles particles studied. Maximum or nearly maximum
(27.63 vs. 32.07 mN/m) for nanoparticles of the same coverage of the interface is evidently achieved at
size stabilized with n-dodecane-1-thiol). equilibrium for all bulk concentrations considered,
i.e., C ffi Cmax :
Using Eqs. 7 and 9, the diffusivity of gold
Discussion nanoparticles may be estimated from early- and late-
time interfacial tension data, respectively. These
Adsorption kinetics of n-dodecane-1-thiol- estimates provide insight into the kinetics of the
and n-hexane-1-thiol-stabilized gold nanoparticles adsorption process. Should the adsorption kinetics be
controlled by diffusion of nanoparticles in the bulk
In this section, the gold nanoparticle adsorption data phase at all times, one should find D0 ¼ D1 : For
obtained are quantitatively analyzed. The Langmuir– dilute colloidal solutions such as the ones considered
Szyszkowski isotherm is characterized by a maximum here, the Stokes–Einstein equation should be expected

123
J Nanopart Res (2011) 13:6579–6589 6587

Table 4 Adsorption parameter for 1.6 ± 0.28 nm, 2.8 ± 0.0.27 nm n-dodecane-1-thiol and 2.9 ± 0.19 nm n-hexane-1-thiol-capped
gold nanoparticles at the hexane–water interface
Size (nm) aL (mol/m3) C1  106 ðmol/m2 Þ Ctri  106 ðmol/m2 Þ Csq  106 ðmol/m2 Þ

1.6 1.15 9 10-7 1.27 3.28 ± 1.10 0.070 ± 0.024

2.8 4.8 9 10-9 0.843 1.02 ± 0.2 0.22 ± 0.43
2.9 2.38 9 10-10 0.782 0.96 ± 0.12 0.207 ± 0.027

Table 5 Stokes–Einstein diffusion coefficient for 1.6 ± 0.28

7
Adsorption Density x10

8.4304 nm, 2.8 ± 0.0.27 nm n-dodecane-1-thiol and 2.9 ± 0.19 nm n-

hexane-1-thiol-capped gold nanoparticles at the hexane–water
8.4282
interface
[mol/m ]
2

8.4260
Size (nm) Stokes–Einstein diffusion
8.4238 coefficient (m2/s)

8.4216 1.6 (8.92 ± 1.56) 9 10-10

0
-7
1x10 2x10
-7 -7
3x10 2.8 (5.00 ± 0.48) 9 10-10
Bulk Concentration [mol/L] 2.9 (4.79 ± 0.3) 9 10-10

Fig. 7 Equilibrium adsorption density of 2.8 ± 0.27 nm

n-dodecane-1-thiol-capped gold nanoparticles as a function of
bulk nanoparticle concentration
Table 6 Adsorption behavior for n-hexane-1-thiol-capped
gold nanoparticles synthesized under condition (c) at the
to provide a reasonably accurate prediction of the hexane–water interface
diffusivity of nanoparticles in hexane:
Nanoparticle Equilibrium Diffusion coefficient (m2/s)
kB T concentration int. tension
DSE ¼ ð10Þ C (particles/L) c1 ðmN=mÞ D0 D1
6pgr
4.32 9 1015 31.70 3.70 9 10-10 9.12 9 10-13
where kB is the Boltzmann constant, g is the viscosity 15 -10
8.61 9 10 29.70 2.38 9 10 5.39 9 10-13
of the solvent, and r is the nanoparticle radius. The
estimated DS–E value for the studied nanoparticles can 2.16 9 1016 28.24 2.83 9 10-11 3.54 9 10-15
16 -12
be seen from Table 5. 8.65 9 10 25.25 1.31 9 10 6.62 9 10-15
17 -13
Calculations of D0 and D1 from dynamic interfa- 1.30 9 10 25.22 5.79 9 10 2.91 9 10-15
cial tension data for each kind of gold nanoparticles These particles were sized using a log-normal fit, resulting in a
synthesized are summarized in Tables 2, 3, and 6. We mean particles size of l = 2.85 nm with a standard deviation
of r = 0.188 nm
find D0 [ D1 typically by at least an order of
magnitude, for all but the smallest nanoparticle
concentration. Both D0 and D1 decrease with increas- adsorption times ðt ! 1Þ and for adsorption from
ing concentration of the bulk solution. Early-time higher bulk concentrations, where high coverage of
ðt ! 0Þ diffusivity estimates, D0, are plotted as a the interface by nanoparticles is realized at earlier
function of bulk nanoparticle concentration in Fig. 8, times, the estimates of diffusion coefficient afforded
where the predictions of nanoparticle diffusivity by Eq. 7 are significantly lower than the DS–E values.
provided by the Stokes–Einstein equation are also Similar observations have been made by Kutuzov
shown for comparison. et al. (2007) for a different nanoparticle system. These
Only for the lowest bulk concentrations, D0 is seen authors have suggested that an adsorption barrier
to agree well with DS–E, suggesting that diffusion in rapidly sets in with increasing surface coverage, as a
the bulk hexane phase controls the adsorption kinetics result of increasing collisions in the sub-layer between
only under conditions of low coverage of the hexane– nanoparticles desorbed from the interface with nano-
water interface by nanoparticles. Over increasing particles approaching the interface from the bulk

123
6588 J Nanopart Res (2011) 13:6579–6589

10
-8
that a particle must do against surface pressure (Eastoe
-9 Stokes-Einstein Diffusivity 1997), and using experimentally measured surface
Diffusivity [m /s]

10
2

-10
10 pressure data, we find equilibrium contact angles of
-11
10 114° and 145° for the smaller and larger dodecanethiol
-12
10 particles, respectively, and 148° for the hexanethiol
-13
10
Diffusivity (t 0) particles. Hence, the particles of different size have
-14
10
different contact angles, yet the same size particles
-7 -7 -7
0 1x10 2x10 3x10 have same almost same contact angle, which is
Bulk Concentration [mol/L] consistent with the theory (Wang 2005). The change
in free energy, DE; for a single spherical particle
Fig. 8 Diffusivity of 2.8 ± 0.27 nm n-dodecane-1-thiol-
capped gold nanoparticles as a function of bulk nanoparticle
moving from oil phase to the interface is given by:
concentration
DE ¼ pr 2 co=w ½1 þ cosðHÞ2 ; ð12Þ

where DE is the change in the free energy, r is the

solution. In the context of the theory put forth by particle radius, co/w is the interfacial tension, and H is
Liggieri et al. (1996) for mixed diffusion-activation the particle–liquid contact angle. DE; r and co/w are
controlled adsorption, the preceding analysis of same for larger dodecanethiol and hexanethiol parti-
experimental data provides an apparent diffusion cles. Thus, although having the different capping
coefficient which depends on the magnitude of the ligand, they have the same contact angle. It has also
potential barrier and free diffusion coefficient as found that as the dodecanethiol concentration on the
follows: surface of nanoparticle increases, the contact angle
  also increases (Park 2007).
DE
D ¼ Df exp  ð11Þ Such interpretations, however, must be viewed with
kB T
great caution. When deformable soft organic ligands
where DE is the activation energy of the barrier and Df cover the particle surface, assessment of the energetics
is free diffusion coefficient. In the absence of an of nanoparticle localization at a fluid–fluid interface
energy barrier to adsorption (i.e., when adsorption is via determination of an equilibrium contact angle in
limited by free diffusion of nanoparticles from the the framework of Young’s theory is challenged by
bulk solution to the subsurface), Df is equal to DS–E. molecular dynamics simulations (Udayana Ranatunga
Taking the effective diffusivity equal to D1 values 2010). Further study is clearly needed before predic-
obtained from experiments with the highest nanopar- tions of the magnitude of the adsorption barrier can be
ticle concentrations, using Eq. 11, the estimated made.
values of the potential barrier to adsorption, DE; are
4.0 9 10-20 and 4.6 9 10-20 J, respectively, for the
smaller and larger dodecanethiol particles, and Conclusions
4.8 9 10-20 J for the hexanethiol particles. These
values are approximately equal to 10 kBT, as also With an aim to improve the understanding of self-
found by Kutuzov et al. (2007) for TOPO-stabilized assembly of ligand-stabilized nanoparticles at liquid–
CdSe nanoparticles of different sizes at the toluene– liquid interfaces, we studied, by pendant drop tensi-
water interface. The latter authors suggested that, ometry, the kinetics of alkanethiol-stabilized gold
under conditions of high coverage of the interface with nanoparticles at the hexane–water interface. From the
nanoparticles, the adsorption process is controlled by time evolution of the interfacial tension at the early
the rate of desorption of already adsorbed nanoparti- (t ! 0) and late (t ! 1) stages of adsorption we
cles, such that the activation energy, DE; is roughly could infer a switch, with increasing interfacial
given by Eq. 1. For all nanoparticles tested in this coverage, from diffusion-controlled kinetics to inter-
study, we find this suggestion consistent with a contact action-controlled kinetics. The potential barrier for
angle of about 157° (measured from the aqueous nanoparticle adsorption to the hexane–water interface
phase). Alternatively, equating DE to Wads, the work was estimated from the diffusion constants of the late

123
J Nanopart Res (2011) 13:6579–6589 6589

stages of adsorption and found equal to approximately solutions, based on asymptotic equations of adsorption
10 kBT, for gold nanoparticles of different size kinetic theory. Colloids Surf A Physicochem Eng Aspects
87:61–75
stabilized with different ligands (n-dodecane-1-thiol Glaser N, Adams DJ, Boeker A, Krausch G (2006) Janus par-
and n-hexane-1-thiol). These results corroborate the ticles at liquid–liquid interfaces. Langmuir 22:5227–5229
findings of a first report by Kutuzov et al. (2007) for a Goebel A, Lunkenheimer K (1997) Interfacial tension of the
different nanoparticle system. For constant bulk water/n-alkane interface. Langmuir 13(2):369–372
Hostetler MJ, Wingate JE, Zhong CJ, Harris JE, Vachet RW,
concentration and nanoparticle size, it was addition- Clark MR, Londono JD, Green SJ, Stokes JJ, Wignall GD,
ally found that adsorption of nanoparticles stabilized Glish GL, Porter MD, Evans ND, Murray RW (1998)
with shorter alkanethiol ligands leads to a lower Alkanethiolate gold cluster molecules with core diameters
equilibrium interfacial tension. from 1.5 to 5.2 nm: core and monolayer properties as a
function of core size. Langmuir 14:17–30
Isa L, Amstad E, Textor M, Reimhult E (2010) Self-assembly of
Acknowledgments The authors would like to thank Dr. iron oxide-poly(ethyleneglycol) core–shell nanoparticles
Claude Lemaire for assistance with the NMR analysis. We at liquid–liquid interfaces. Chimia 64(3):145–149
also want to thank Dr. Robert Donkers for his helpful Krishnaswamy R, Sood Ak (2010) Growth, self-assembly and
discussions. This study was supported by the Natural Sciences dynamics of nano-scale films at fluid interfaces. J Mater
and Engineering Research Council of Canada (NSERC) and the Chem 20:3539–3552
Department of Chemical Engineering at the University of Kutuzov S, He J, Tangirala R, Emrick T, Russel TP, Boeker A
Waterloo. (2007) On the kinetics of nanoparticle self-assembly at liquid/
liquid interfaces. Phys Chem Chem Phys 9:6351–6358
Liggieri L, Ravera F, Passerone A (1996) A diffusion-based
References approach to mixed adsorption kinetics. Colloids Surf A
Physicochem Eng Aspects 114:351–359
Lin Y, Skaff H, Emrick T, Dinsmore AD, Russell TP (2003)
Alvarez MM, Khoury JT, Schaaff TG, Shafigullin MN, Vezmar
Nanoparticle assembly and transport at liquid–liquid
I, Whetten RL (1997) Optical adsorption spectra of nano-
interfaces. Sci Agric 299:226–229
crystal gold molecules. J Phys Chem B 101:3706–3712
Park Y-K, Yoo S-H, Park S (2007) Assembly of highly ordered
Aveyard R, Clint JH (1996) Particle wettability and line tension.
nanoparticle monolayers at a water/hexane interface.
J Chem Soc Faraday Trans 92(1):85–89
Langmuir 23:10505–10510
Binks BP (2002) Particles as surfactants–similarities and dif-
Patra D, Ozdemir F, Miranda OR, Samanta B, Sanyal A, Rotello
ferences. Curr Opin Colloid Interface Sci 7:21–41
VM (2009) Formation and size tuning of colloidal micro-
Binks BP, Horozov TS (2006) Colloidal particles at liquid
capsules via host–guest molecular recognition at the
interfaces. Cambridge University Press, Cambridge,
liquid–liquid interface.Langmuir 25(24):13852–13854
pp 1–518
Pieranski P (1980) Two-dimensional interfacial colloidal crys-
Boeker A, He J, Emrick T, Russell TP (2007) Self-assembly of
tals. Phys Rev Lett 45(7):569–572
nanoparticles at interfaces. Soft Matter 3:1231–1248
Szyskowski BV (1908) Experimentelle studien über kapillare
Bresme F, Oettel M (2007) Nanoparticles at fluid interfaces.
eigenschaften der asserigen losungen von fettsauren.
J Phys Condens Matter 19:413101
Z Phys Chem 64:385–414
Brust M, Walker M, Bethell D, Schiffrin DJ, Whyman R (1994)
Udayana Ranatunga RJK, Kalescky Robert JB, Chiu Chi-cheng,
Synthesis of thiol-derivatized gold nanoparticles in a two-
Nielsen Steven O (2010) Molecular dynamics simulations
phase liquid–liquid system. J Chem Soc Chem Commun
of surfactant functionalized nanoparticles in the vicinity of
801–802
an oil/water interface. J Phys Chem C 114:12151–12157
Cheung DL, Bon SAF (2009) Interaction of nanoparticles
Wang D, Duan H, Moehwald H (2005) The water/oil interface:
with ideal liquid-liquid interfaces. Phys Rev Lett 102(6):
the emerging horizon for self-assembly of nanoparticles.
066103/1–4
Soft Matter 1:412–416
Crossley S, Faria J, Shen M, Resasco DE (2010) Solid nano-
Ward AFH, Tordai L (1946) Time-dependence of boundary
particles that catalyze biofuel upgrade reactions at the
tensions of solutions I. The role of diffusion in time-effects.
water/oil interface. Sci Agric 327:68–72
J Chem Phys 14:453–461
Eastoe J, Dalton JS, Rogueda PGA, Crooks ER, Pitt AR, Sim-
Zhou J, Ralston J, Sedev R, Beattie DA (2009) Functionalized
ister EA (1997) Dynamic surface tensions of nonionic
gold nanoparticles: synthesis, structure and colloid stabil-
surfactant solutions. J Colloid Interface Sci 188:423–430
ity. J Colloid Interface Sci 331(2):251–262
Fainerman VB, Makievski AV, Miller R (1994) The analysis
of dynamic surface tension of sodium alkyl sulphate

123