of 8
Nanoscale
PAPER
Cite this:
Nanoscale
, 2015,
7
, 7833
Received 21st October 2014,
Accepted 23rd March 2015
DOI: 10.1039/c4nr06208f
www.rsc.org/nanoscale
Motion-based threat detection using microrods:
experiments and numerical simulations
Barath Ezhilan,
a
Wei Gao,
b
Allen Pei,
b
Isaac Rozen,
b
Renfeng Dong,
b
Beatriz Jurado-Sanchez,
b
Joseph Wang*
b
and David Saintillan*
a
Motion-based chemical sensing using microscale particles has attracted considerable recent attention. In
this paper, we report on new experiments and Brownian dynamics simulations that cast light on the
dynamics of both passive and active microrods (gold wires and gold
platinum micromotors) in a silver
ion gradient. We demonstrate that such microrods can be used for threat detection in the form of a silver
ion source, allowing for the determination of both the location of the source and concentration of silver.
This threat detection strategy relies on the di
ff
usiophoretic motion of both passive and active microrods
in the ionic gradient and on the speed acceleration of the Au
Pt micromotors in the presence of silver
ions. A Langevin model describing the microrod dynamics and accounting for all of these e
ff
ects is pre-
sented, and key model parameters are extracted from the experimental data, thereby providing a reliable
estimate for the full spatiotemporal distribution of the silver ions in the vicinity of the source.
1 Introduction
Controlled migration and self-propulsion of microscale par-
ticles is a challenging task with a broad range of technological
and biomedical applications. A classical way to manipulate col-
loidal suspensions relies on phoretic transport, by which exter-
nally imposed gradients of various fields (such as temperature,
electric field, solute concentration
etc
.) can lead to the
migration of colloidal particles. Recent advances in micro-
fluidic technologies have spurred a renewed interest in the
theoretical understanding and practical application of such
transport mechanisms.
1,2
In recent years, phoretic e
ff
ects have also been used for the
autonomous self-propulsion of microparticles. Catalytically
active microparticles such as bimetallic microrods can indeed
self-generate and sustain surface asymmetries leading to their
propulsion through a variety of self-phoretic mechanisms.
3
These artificial micromotors o
ff
er a broad range of potential
applications ranging from drug delivery, microscale assembly
and transport, robotics, to motion-based chemical/biochemi-
cal sensing.
4
18
Potential environmental applications of micro-
motors have grown rapidly in recent years, indicating a myriad
of innovative remediation and monitoring applications based
on new motion-based phenomena.
19
24
Specifically, micro-
motors that exhibit a concentration dependence of their self-
propulsion velocities could mimic chemotactic search strat-
egies employed by microorganisms in biological settings and
hence be used to trace plumes of hazardous chemicals to their
source. Recently, Gao
et al.
25
reported that Ir/silica Janus
micromotors, powered by extremely low levels of hydrazine
fuel, display a well-defined concentration/speed dependence,
thus holding considerable promise for tracing concentration
gradients of this pollutant. Experiments have shown that
Au
Pt micromotors exhibit directed movement toward high
hydrogen peroxide concentrations through
active di
ff
usion.
26
In addition, an unusual speed acceleration of Au
Pt micro-
motors in the presence of silver ions has also been reported.
27
This specific unexpected e
ff
ect of silver upon the speed of cata-
lytic micromotors has been exploited for designing a new
motion-based trace silver sensing protocol. The specific accel-
eration has been attributed to the underpotential deposition
of silver onto a platinum segment, which increases the electro-
catalytic activity. The highly selective motion-based response is
characterized by a defined concentration dependence, with the
speed of the micromotors providing quantitative information
down to the nanomolar level. This silver-based acceleration
has also formed the basis for a motion-based detection of DNA
hybridization in connection to silver nanoparticle tags.
28
While previous studies have separately focused on phoretic
migration in an external gradient and on self-phoretic propul-
sion in the absence of external gradients, the combined use of
both transport strategies has been widely unexplored and
could lead to novel avenues for the control of colloidal
Electronic supplementary information (ESI) available. See DOI: 10.1039/
c4nr06208f
a
Department of Mechanical and Aerospace Engineering, University of California
San Diego, 9500 Gilman Drive, La Jolla, CA 92093, USA. E-mail: dstn@ucsd.edu
b
Department of Nanoengineering, University of California San Diego, 9500
Gilman Drive, La Jolla, CA 92093, USA. E-mail: josephwang@ucsd.edu
This journal is © The Royal Society of Chemistry 2015
Nanoscale
,2015,
7
,7833
7840 |
7833
Published on 30 March 2015. Downloaded by California Institute of Technology on 11/10/2017 22:49:43.
View Article Online
View Journal
| View Issue
motions or the sensing of external fields. Because of the ubi-
quity of Brownian motion at the length scales relevant to these
suspensions, the non-trivial interplay of phoretic migration,
self-phoretic propulsion and Brownian fluctuations is of sig-
nificant theoretical and practical interest. In the context of bio-
logical transport, this is analogous to the coupling between
swimming, chemotactic migration and Brownian motion,
which is known to lead to a rich variety of non-trivial
phenomena.
In this paper, we perform experiments using both passive
tracers (Au microwires) and active swimmers (Au
Pt micro-
wires) near a silver ion source using methods described in
section 2. Passive tracers in an electrolyte gradient move
towards lower concentration of electrolyte as explained by clas-
sical theories of di
ff
usiophoresis.
29,30
This e
ff
ect causes a
marked depletion of gold microwires near the silver ion
source, thus enabling spatial detection of the source. Active
swimmers (Au
Pt microwires) also exhibit a di
ff
usiophoretic
response to the externally imposed silver ion gradient (similar
to passive tracers) while additionally accelerating in regions of
higher silver ion concentration.
27
This silver-based accelera-
tion characteristic of Au
Pt micromotors is utilized to develop
a strategy for accurate spatiotemporal sensing of the silver ion
concentration in the domain by fitting experimental data to a
numerical model. In section 3, we introduce a Langevin model
and Brownian Dynamics (BD) simulation methodology to
analyze the non-trivial coupling of di
ff
usiophoresis, silver-ion
induced speed acceleration and Brownian motion. Key model
parameters are extracted from the experimental velocity data of
both passive tracers and active swimmers, from which the
spatiotemporal evolution of the silver ion concentration in the
domain is deduced. Results from the experiments and BD
simulations are reported in section 4 where the spatiotemporal
dynamics, density and velocity profiles of both passive tracers
and active swimmers are analyzed in detail. We conclude in
section 5.
2 Experimental methods
2.1 Micromotor preparation
The Au
Pt microwires were prepared by sequential electro-
deposition of gold and platinum into 200 nm diameter cylindri-
cal pores of an alumina membrane template (catalog no. 6809-
6022; Whatman, Maidstone, U.K.). A thin gold film was first
sputtered on the branched side of the membrane to serve as a
working electrode. The membrane was assembled in a Teflon
plating cell with aluminum foil serving as an electrical contact
for the subsequent electrodeposition. A sacrificial copper layer
was first electrodeposited into the branched area of the mem-
brane using a 1 M cupric sulfate pentahydrate solution
(CuSO
4
·5H
2
O; Sigma-Aldrich, St. Louis, MO), at total charge of
10 °C and a potential of
1.0 V (
vs.
Ag/AgCl reference elec-
trode) along with platinum wire as a counter electrode. Sub-
sequently, a Au segment was plated from a gold plating
solution (Orotemp 24 RTU RACK; Technic Inc., Anaheim, CA)
and electrodeposited at a total charge of 1.5 °C and a potential
of
0.9 V. Platinum was then deposited galvanostatically using
a current of
2 mA for 50 min from a platinum plating solu-
tion (Platinum RTP; Technic Inc.). The resulting Au
Pt micro-
wires had a length of around 2
μ
m. The sputtered gold layer
and the copper sacrificial layer were simultaneously removed
by mechanical polishing with 3
4
μ
m alumina slurry. The
bisegment Au/Pt microwires were then released by immersing
the membrane in 3 M NaOH for 30 min. The synthesized
micromotors were separated from solution by centrifugation at
10 000 rpm for 5 min and washed repeatedly with ultrapure
water (18.2 M
Ω
cm) until a neutral pH was achieved. Between
the washing steps, the micromotor solution was mixed with
ultrapure water and briefly sonicated (25 s) to ensure the com-
plete dispersion of micromotors. All micromotors were stored
in ultrapure water at room temperature and their speed was
tested before each experiment.
2.2 Silver ion source preparation
Planar sources of elemental silver were fabricated through a
photolithographic process. After patterning 50
μ
m cylindrical
wells in a photoresist film on a glass coverslip, 10 nm of Ti
and 100 nm of Ag were deposited onto the glass by electron
beam evaporation (Temescal BJD 1800 E-beam Evaporator).
The remaining metal film and underlying photoresist were
stripped o
ff
in acetone, resulting in well-defined Ag films to
serve as sources of Ag
+
ions. For the duration of the experi-
ments, these Ag films can be considered as infinite sources,
since they continuously release Ag
+
ions in the presence of
hydrogen peroxide.
2.3 Setup
A schematic of the experimental setup is shown in Fig. 1.
A known volume of a solution containing the Au
Pt microwires
Fig. 1
(color online) Schematic of the experimental setup. Silver source
(black dot in the center) releases Ag
+
into a fuel enhanced (0.1% hydro-
gen peroxide) solution containing Au
Pt active microwires or Au passive
microwires. The movement of the micromotors was followed using an
inverted microscope.
Paper
Nanoscale
7834
|
Nanoscale
,2015,
7
,7833
7840
This journal is © The Royal Society of Chemistry 2015
Published on 30 March 2015. Downloaded by California Institute of Technology on 11/10/2017 22:49:43.
View Article Online
was dropped onto the planar silver source. The wires were
allowed to settle in the plane of focus before the same volume
of hydrogen peroxide fuel (0.1%), essential for microwire pro-
pulsion and silver dissolution, was added using a pipette.
Control experiments performed with an inert source made of
glass (instead of the active silver ion source) showed no convec-
tive perturbations or phoretic drift due to the transient
inhomogeneities in the mixing process. This suggests that the
peroxide and microrod solutions mix rapidly on timescales sig-
nificantly higher than the phoretic timescales relevant to our
problem. Based on their self-electrophoretic mechanism, the
micromotors self-propel at a speed of 10
μ
ms
1
in a 0.1%
hydrogen peroxide solution. This micromotor speed increases
dramatically in the presence of Ag
+
ions. Micromotor speeds
were determined with movies taken at 45 frames per second
using NIS-Elements AR software (Nikon, USA). Individual
micromotor trajectories were tracked and Origin software was
used to analyze the data.
2.4 Data extraction
2.4.1 Velocity tracking.
The time of fuel addition to the
microwire sample was taken to be the start of the experiment
(
t
= 0). microwire velocities were tracked at 0 s, 2.5 s, 5 s, 10 s,
15 s, and 20 s after adding hydrogen peroxide and were
averaged over 10 frames for each time interval. The distance of
each micromotor from the center of the Ag source was
recorded at the beginning of the 10-frame interval. Over
150 micromotors were tracked per time interval per video for
3 movies.
2.4.2 Density measurements.
Density measurements were
made at 0 s, 2.5 s, 5 s, 10 s and 20 s after fuel addition.
A known volume of solution containing the Au
Pt microwires
(for experiments with active tracers) or Au microwires (for
experiments with passive tracers) was dropped onto the silver
source. The swimmers/tracers were allowed to fully settle to
the plane of focus before addition of the hydrogen peroxide
fuel (0.1%). Rings with radii ranging from 25
μ
m to 125
μ
m
and with spacing
Δ
r
=5
μ
m between them were defined on the
captured images at the specified times. The number of micro-
wires within each consecutive pair of rings was counted,
divided by the area, normalized by the initial number of micro-
wires (at
t
= 0) and ensemble-averaged over 3 movies to obtain
the wire density at discrete radial positions
r
j
= [25 + (
j
0.5)
Δ
r
]
μ
m for
j
=1
...
20 using the following expression:
c
ð
r
j
;
t
Þ¼
P
N
i
¼
1
Ð
r
j
þ
Δ
r
2
r
j

Δ
r
2
δ
r

r
i
ðÞ
d
r
hi
Nt
¼
0
ðÞ
π
r
j
þ
Δ
r
2

2

r
j

Δ
r
2

2

*+
;
ð
1
Þ
where
r
i
is the radial position of each particle,
N
(
t
= 0) is the
total number of particles at the time of fuel addition (
t
=0)
and the symbol
·
denotes an ensemble average over 3 distinct
realizations of the experiment.
3 Langevin model and Brownian
dynamics simulation
3.1 Particle fluxes
We describe the two-dimensional dynamics of
M
= 1000 par-
ticles (either passive tracers or active swimmers) using a Lange-
vin model. A particle with index
i
(1
i
M
) has center-of-
mass position
x
i
and orientation
p
i
parametrized in the plane
of motion as
p
i
= [cos
θ
i
, sin
θ
i
]. During a short time interval
δ
t
,
the center-of-mass displacement is calculated using the follow-
ing Langevin equation as the sum of self-propulsive, di
ff
usio-
phoretic and Brownian contributions:
δ
x
i
¼
V
sp
s
x
i
;
t
½
ðÞ
p
i
þ
V
dp
ð
s
x
i
;
t
½Þ
ˆ
e
r

δ
t
þ
ffiffiffiffiffiffiffiffiffiffiffiffi
2
D
t
δ
t
p

n
1
;
ð
2
Þ
where
ˆ
e
r
¼
x
i
=
j
x
i
j
is a unit vector pointing radially outward
from the position of the source at the center of particle
i
.Brow-
nian fluctuations also lead to changes in the particle orien-
tations, which are also captured by a Langevin equation:
δ
p
i
¼
ffiffiffiffiffiffiffiffiffiffiffi
2
d
r
δ
t
p
I

pp
ðÞ
n
2
:
ð
3
Þ
In eqn (2) and (3),
n
1
and
n
2
denote two-dimensional
random vectors whose components follow a random Gaussian
distribution with zero mean and unit variance. The trans-
lational and rotational di
ff
usivities
D
t
and
d
r
can be estimated
for rod-like particles
31
as:
D
t
¼
D
t
I
þ
pp
ðÞ
where
D
t
¼
k
B
T
log
1
ε

4
π
μ
ð
4
Þ
d
r
¼
3
k
B
T
log
1
ε

π
μ
3
ð
5
Þ
where
and
ε
denote the length and inverse aspect ratio of a
particle, respectively, and
μ
is the viscosity of the suspending
fluid. The micromotors have a length of 2
μ
m and aspect ratio
of
10. Based on these values, we estimate the Brownian trans-
lational di
ff
usivity coe
ffi
cient to be
D
t
4×10
13
m
2
s
1
, and
the rotational di
ff
usivity to be
d
r
1s
1
.
V
sp
and
V
dp
are the self-propulsive and di
ff
usiophoretic vel-
ocities of the catalytic micromotors, which both depend on the
evolving silver ion concentration
s
[
x
i
,
t
] in the domain. The
functional dependence of the self-propulsion velocity on the
silver concentration is obtained by fitting a logarithmic curve
to the micromotor speed
vs.
silver ion concentration data of
Kagan
et al.
:
27
V
sp
s
ðÞ¼
V
0
¼
10
μ
ms

1
;
if
s
,
3
:
4

10

7
7
:
1169 log
s
ðÞþ
116
:
05
;
otherwise
ð
6
Þ
where
s
should be expressed in Molar units. To simulate the
dynamics of passive tracers (Au microwires), we simply set
V
sp
=0.
The di
ff
usiophoretic slip velocity of a spherical object in an
ionic gradient was previously derived
29,30
and is expressed as:
V
dp
ð
s
Þ¼
D
dp
@
log
s
@
r
;
ð
7
Þ
Nanoscale
Paper
This journal is © The Royal Society of Chemistry 2015
Nanoscale
,2015,
7
,7833
7840 |
7835
Published on 30 March 2015. Downloaded by California Institute of Technology on 11/10/2017 22:49:43.
View Article Online
where the coe
ffi
cient
D
dp
is the sum of
chemiphoretic
and
electrophoretic
contributions:
D
dp
¼
k
B
T
μ
κ

2
ν
c
ð
ζ
Þþ
̃
βζν
e

:
ð
8
Þ
here,
ν
e
and
ν
c
are positive coe
ffi
cients,
κ
is the adsorption
length,
β
̃
is the di
ff
erence between anion and cation mobilities
normalized by their sum, and
ζ
is the surface zeta potential.
This peculiar
log-sensing
property was also verified by recent
experiments on di
ff
usiophoresis in controlled electrolyte gradi-
ents.
1,2
We assume in our simulations that
D
dp
is a constant
independent of the silver concentration, and the weak depen-
dence of
D
dp
on salt concentration was indeed shown by
Palacci
et al.
1
to play only a minimal role. The simpler scaling
for
D
dp
k
B
T
/
μ
l
B
has also been suggested,
32
where
l
B
is the
Bjerrum length (
l
B
= 0.7 nm in water). This estimate gives
D
dp
10
9
m
2
s
1
and
V
dp
D
dp
/
R
0
40
μ
ms
1
. As we shall
see in section 3, a more reliable estimate for
D
dp
/
R
0
can in fact
be extracted from the experimental data.
3.2 Silver ion evolution
In order to integrate eqn (2) and (3) in time, we require knowl-
edge of the silver concentration field
s
(
x
,
t
), which varies both
in space and time. Specifically, silver ions di
ff
use radially into
the domain from the boundary of the infinite silver source
located at
r
=
R
0
=25
μ
m. We consider a finite-size compu-
tational domain of exterior radius
R
domain
= 150
μ
m. Silver ions
are injected into the domain by the following surface reaction:
Ag
ð
s
Þþ
H
2
O
2
!
2Ag
þ
ð
aq
:
Þþ
2OH

:
ð
9
Þ
At the surface of the source, we expect the rate
j
0
of silver
ion produced by the surface reaction to balance the di
ff
usive
flux away from the source:

D
s
@
s
@
r
r
¼
R
0
¼
j
0
;
ð
10
Þ
where
D
s
is the silver ion di
ff
usivity and is set to 1.6 × 10
9
m
2
s
1
(di
ff
usivity of silver ions in water). The silver ion flux
j
0
at
the source surface is expected to depend on the rate constant
of the reaction and on the concentration of hydrogen peroxide
near the surface: its value is not known
a priori
but will be
extracted from the experimental data as explained in more
detail in section 3.
A second boundary condition for the silver concentration is
provided far away from the source:
s
!
0at
r
!
1
;
ð
11
Þ
and in practice we enforce this condition at
r
=
R
domain
.
Assuming an axisymmetric distribution, the evolution of
the concentration of silver ions
s
(
r
,
t
) inside this domain is
governed by the di
ff
usion equation, in spherical coordinates:
@
s
@
t
¼
D
s
r
2
@
@
r
r
2
@
s
@
r

;
ð
12
Þ
subject to boundary conditions (10) and (11). Solving this
equation analytically using Laplace transforms, we derive the
solution for the normalized silver ion concentration as a func-
tion of radial position and time:
sr
;
t
ðÞ
j
0
R
0
¼
R
0
D
s
r
erfc
r

R
0
2
ffiffiffiffiffiffiffi
D
s
t
p



exp
D
s
t
R
2
0
þ
r
R
0

1

erfc
ffiffiffiffiffiffiffi
D
s
t
p
R
0
þ
r

R
0
2
ffiffiffiffiffiffiffi
D
s
t
p

:
ð
13
Þ
At long times
t
R
2
0
/
D
s
, the normalized silver ion concen-
tration profile reaches a steady state given by:
s
s
ð
r
Þ
j
0
R
0
¼
lim
t
!
1
s
ð
r
;
t
Þ
j
0
R
0
¼
R
0
r
:
ð
14
Þ
Fig. 2(a) shows the normalized silver ion concentration pro-
files according to eqn (13) at di
ff
erent times
t
. It can be seen
that, as time increases, the normalized silver ion concentration
on the source surface (
r
=
R
0
) increases, asymptotically reach-
ing a value of 1 at infinite time:
s
(
R
0
,
)=
j
0
R
0
. The concen-
tration of silver ion in the domain also gradually increases
everywhere as a result of di
ff
usion.
Using the solution of eqn (13) for the silver ion concen-
tration, we calculate the di
ff
usiophoretic velocity
V
dp
(
s
)as
follows:
V
dp
¼
D
dp
R
0
R
0
r

1

þ
1

exp
D
s
t
R
2
0
þ
r
R
0

1

erfc
ffiffiffiffiffiffiffi
D
s
t
p
R
0
þ
r

R
0
2
ffiffiffiffiffiffiffi
D
s
t
p

erfc
r

R
0
2
ffiffiffiffiffiffiffi
D
s
t
p

0
B
B
@
1
C
C
A

1
3
7
7
7
5
:
ð
15
Þ
At long times
t
R
2
0
/
D
s
, the di
ff
usiophoretic velocity also
reaches a steady state given by:
V
s
dp
ð
r
Þ¼
lim
t
!
1
V
dp
ð
r
;
t
Þ¼
D
dp
r
:
ð
16
Þ
The spatial dependence of the di
ff
usiophoretic velocity at
di
ff
erent times is illustrated in Fig. 2(b), showing the logar-
ithm of the silver ion concentration profile. At short times,
Fig. 2
(color online) (a) Normalized silver ion concentration pro
fi
le and
(b) its logarithm as a function of radial position and time.
Paper
Nanoscale
7836
|
Nanoscale
,2015,
7
,7833
7840
This journal is © The Royal Society of Chemistry 2015
Published on 30 March 2015. Downloaded by California Institute of Technology on 11/10/2017 22:49:43.
View Article Online
when the silver ions have not yet di
ff
used significantly into the
domain (
s
0), the model of eqn (15) predicts a nonphysical
increase in the logarithmic silver ion profile. To remedy this,
we set
V
dp
to zero when the silver concentration
s
fall below
the critical value of
s
min
= 0.2
j
0
R
0
, which suppresses the non-
physical increase of
log
s
/
r
with
r
in limit of
s
0. A similar
method was previously used by Hong
et al.
33
3.3. Model parameter extraction
The model described above has two unknown parameters: the
di
ff
usiophoretic coe
ffi
cient
D
dp
, and the surface silver ion flux
j
0
on the surface of the source. As we explain here, both of
these parameters can be extracted from the experimental data,
using two sets of experiments involving passive particles and
active micromotors. First, it can be noticed that while the self-
propulsion velocity
V
sp
in eqn (6) depends on the surface ion
flux
j
0
, the di
ff
usiophoretic velocity
V
dp
in eqn (15) does not
and only involves
D
dp
. This implies that
D
dp
can first be
extracted from data for passive particles (Au microwires). Such
data for the velocity
vs.
distance from the source is shown in
Fig. 3(a), and can be fit to the model of eqn (15) to obtain the
estimate
D
dp
/
R
0
=10
μ
ms
1
. Once
D
dp
is known, the active
micromotor (Au
Pt microwire) data, shown in Fig. 3(b), can
then be used to obtain the best fit value of
j
0
R
0
= 2.52 × 10
6
M
for the silver ion flux. This estimate for
j
0
R
0
, coupled with the
analytical solution of eqn (13), enables prediction of the full
spatiotemporal silver ion concentration profile throughout the
domain with good resolution.
3.4 Simulation method
At
t
= 0, the particle positions and orientations are initialized
with a uniform isotropic distribution. The governing equations
are non-dimensionalized using the following scales for
lengths, time, and silver concentration:
L
c
¼
R
0
¼
25
μ
m
;
ð
17
Þ
t
c
¼
R
2
0
D
s

0
:
4s
;
ð
18
Þ
s
c
¼
j
0
R
0
:
ð
19
Þ
The self-propulsion and di
ff
usiophoretic contributions to
the translational flux can be calculated analytically using
eqn (6) and (15). Time integration of the Langevin equations
for the particle motions are performed using a second-order
Adams
Bashforth time-marching scheme. The swimmer/tracer
density profiles were calculated in the same fashion as the
experimental profile measurement described in section 2, but
ensemble-averaged over 12 simulation runs.
3.5 Model assumptions and limitations
The model described above is the simplest model that
accounts for all relevant e
ff
ects including self-propulsion,
di
ff
usiophoresis, and di
ff
usion of silver ions. However, it is
also based on a number of simplifications and approximations
that may have a quantitative impact on the results. First, we
note that the classical expression for the di
ff
usiophoretic vel-
ocity used here was originally derived for spherical par-
ticles.
29,30
While we do not expect the dependence on
s
to
change significantly for rod-like particles such as the micro-
motors considered here, di
ff
usiophoresis may also lead to a
rotational flux contribution, though this flux is negligible
unless strong gradients exist on the scale of the particles,
which is not the case here. Second, we have neglected the
e
ff
ect of the microrods on the distribution of silver ions. This
e
ff
ect can be twofold: on the one hand, the reaction of the
silver ions with the microrods is expected to deplete the silver
concentration in their vicinity; on the other hand, motion of
the particles may also lead to convection and stirring of the
silver ions beyond the simple e
ff
ect of molecular di
ff
usion.
Finally we note that we have ignored both steric and hydro-
dynamic interactions between micromotors on the basis that
the suspensions used in the experiments were dilute; these
interactions may play a minor role in the fronts that form near
the silver source, where the density is higher, yet we do not
expect them to significantly modify the results.
4 Results and discussion
As the fuel is added at
t
= 0, the silver ions are produced on
the surface of the source at
r
=
R
0
and rapidly di
ff
use into the
domain (as shown in Fig. 2). This drives the formation of silver
ion gradients in the domain, to which both passive tracers and
active micromotors respond in unique ways as explained
below.
4.1 Passive tracer dynamics
The spatiotemporal dynamics of passive tracers are illustrated
in Fig. 4(a) and (b) (and accompanying video). Over time, the
tracers progressively move away from the source due to di
ff
usio-
phoresis. This results in the complete depletion of tracers
near the source. As shown in the passive tracer density profiles
in Fig. 5, a sharply defined depletion front progresses radially
outward over time. The interface between the depletion region
Fig. 3
(color online) Experimental data showing velocity as a function
of radial position and time for (a) active swimmers (Au
Pt micromotors)
and (b) passive tracers (Au microwires).
Nanoscale
Paper
This journal is © The Royal Society of Chemistry 2015
Nanoscale
,2015,
7
,7833
7840 |
7837
Published on 30 March 2015. Downloaded by California Institute of Technology on 11/10/2017 22:49:43.
View Article Online
near the source and the rest of the domain also grows sharper
with time. Fig. 3(a) shows the velocity profile of passive tracers
for di
ff
erent times. It can be seen that the wires reach a vel-
ocity of up to 14
μ
ms
1
. The peak in the velocity is reached
very close to the depletion front. The scatter in the passive
tracer velocity at any given radial location is a consequence of
Brownian fluctuations. Further from the source (
100
μ
m), a
velocity of 2
4
μ
m is reported, which corresponds to contri-
butions solely from Brownian motion.
4.2 Active swimmer dynamics
The spatiotemporal dynamics of active swimmers are illus-
trated in Fig. 4(c) and (d) (and accompanying video). A
depletion front advancing radially outward as a result of di
ff
u-
siophoresis can be clearly seen. However, the depletion e
ff
ect
is reduced by the motility of the swimmers and in particular,
the speed acceleration of the micromotors near the source
Fig. 4
(color online) Spatiotemporal dynamics of passive tracers (Au microwires): (a) experiment and (b) simulation. Spatiotemporal dynamics of
active swimmers (Au
Pt micromotors): (c) experiment and (d) simulation. Also see the accompanying video.
Fig. 5
Passive tracer (Au microwires) density as a function of radial
position and time: (a) experiment and (b) simulation.
Paper
Nanoscale
7838
|
Nanoscale
,2015,
7
,7833
7840
This journal is © The Royal Society of Chemistry 2015
Published on 30 March 2015. Downloaded by California Institute of Technology on 11/10/2017 22:49:43.
View Article Online
where the silver concentration is high. This near-source accel-
eration enables the micromotors to overcome the di
ff
usio-
phoretic flux away from the source and still swim near the
source, resulting in a more di
ff
use interface between the
depletion region and rest of the domain. It can be seen from
Fig. 3(b) that the micromotors reach velocities up to 30
μ
m
because of the combined e
ff
ect of silver induced speed accel-
eration and di
ff
usiophoresis. The peak velocities are always
attained closer to the source where the acceleration due to the
presence of silver ions is the strongest. The scatter in the vel-
ocity data of active swimmers [Fig. 3(b)] at any given radial
location is much higher compared to the passive tracers [Fig.
3(a)]. This is because, even while the self-propulsion and di
ff
u-
siophoretic velocities are only functions of the radial position,
the net translational flux is given by their vector sum (as
expressed in eqn (2)) and is dependent on the swimmer orien-
tation: swimmers pointing towards the source (
p

ˆ
e
r
,
0) have
a reduced velocity compared to swimmers pointing away from
the source (
p

ˆ
e
r
>
0). Because of this e
ff
ect, the depletion
front is found to become more di
ff
use with time, and the
speed at which it advances away from the source is also
reduced compared to the case of passive tracer rods as can be
seen from Fig. 6.
4.3 Simulation
vs.
experiment
Fig. 4 (and accompanying video) illustrate that the experiments
(Fig. 4(a) and (c)) and the Brownian dynamics simulation
(Fig. 4(b) and (d)) compare well and the simulations capture
all the key qualitative features reported in experiments. The
location of the depletion front for both passive tracers and
active swimmers [see Fig. 5 and 6(a)
versus
(b)] matches well
between experiments and simulations. The only mismatch
observed is the presence of the stronger peak in the simulation
density profiles, while such clear peaks are absent in the
corresponding experimental profiles. This mismatch may be a
consequence of several e
ff
ects, including: the particular way in
which the di
ff
usiophoretic velocity is calculated (see section 2);
inaccuracies in the di
ff
usion coe
ffi
cients, which were obtained
using slender-body formulas for high-aspect-ratio rods in an
unbounded fluid whereas the particles in the experiments
reside near a substrate; and the neglect of steric repulsion
e
ff
ects (excluded volume), which likely become important
inside the high-density front at the edge of the depletion
region. The overall agreement between experiments and simu-
lations, however, validates our modeling approach, assump-
tions, and silver ion concentration sensing method.
5 Summary
We have applied a combination of experiments and Brownian
dynamics simulations to demonstrate the use of both passive
and active microrods for the spatial detection of a silver ion
source and the accurate spatiotemporal sensing of the result-
ing silver ion concentration.
First, experiments were performed, where microrods were
dropped onto a planar silver ion source. Our experiments
revealed that the dynamics can be very di
ff
erent based on
whether the microrods are passive or active. Gold microwires,
which are passive, progressively move away from the source
due to di
ff
usiophoresis in the developing gradient of silver
ions. On the other hand, Au
Pt microrods, which are active,
also undergo speed acceleration proportional to the silver ion
concentration (as has been shown in recent experiments by
Kagan
et al.
27
), which weakens the di
ff
usiophoresis-induced
depletion and enables them to swim closer to the source. We
note that this is the first experimental study to look at the com-
bined migration and self-propulsion of microparticles when
subjected to both externally-imposed and self-generated
chemical gradients.
A theoretical model was also proposed based on a simple
Langevin equation, accounting for self-propulsion, di
ff
usio-
phoresis, translational and rotational Brownian motion,
coupled with a di
ff
usion equation for silver ion concentration.
Our model was shown to have two unknown parameters,
viz
.,
the di
ff
usiophoretic coe
ffi
cient and the surface silver ion flux.
Both of these parameters were extracted from two distinct sets
of experimental data involving passive and active microrods,
respectively. Once these parameters were known, the di
ff
usion
equation was analytically solved to obtain the full spatio-
temporal evolution of the silver ion concentration in the entire
domain, and numerical simulations were performed to inte-
grate the Langevin equation for comparison with the experi-
mental data. Good overall agreement between experiments
and simulations was reported. This study, which combined
experiments and theory together with passive and active micro-
rods, suggests novel uses for such particles in technological
applications requiring biochemical sensing, from chemical
threat detection to environmental remediation.
Acknowledgements
This project received support from the Defense Threat
Reduction Agency-Joint Science and Technology O
ffi
ce
for Chemical and Biological Defense (Grant no. HDTRA1-
13-1-0002). D. S. acknowledges support from NSF grant
Fig. 6
Active swimmer (Au
Pt micromotor) density as a function of
radial position and time: (a) experiment and (b) simulation.
Nanoscale
Paper
This journal is © The Royal Society of Chemistry 2015
Nanoscale
,2015,
7
,7833
7840 |
7839
Published on 30 March 2015. Downloaded by California Institute of Technology on 11/10/2017 22:49:43.
View Article Online