www.advmat.de
By Randall M. Erb, Nathan J. Jenness,* Robert L. Clark, and Benjamin
B. Yellen
Janus particles generally refer to a class of colloids with two
dissimilar faces having unique material properties.[1] The
spherical asymmetry associated with Janus particles is the key
to realizing many commercial applications, including electrophoretic displays,[2,3] nanosviscometers,[4] and self-propelling
micromachines.[5] These diverse functionalities were accomplished by using an external electric or magnetic field to control
the particle orientation,[6–8] and in the process, modulate its
reflectivity, hydrodynamic mobility, or direction of motion,
respectively. However, these same asymmetries can interfere
with optical trapping techniques that are used to control the
translational degrees of freedom of a particle.[9] Optical fields
present an effective method for controlling the three translational
degrees of freedom for particles ranging from tens of
nanometers[10] to micrometers in size.[11–14] Previously, optical
fields have been used in combination with magnetic fields to
control four degrees of freedom of an asymmetric particle or
particle aggregate.[15–18] To achieve five or more degrees of
freedom, magnetic Janus particles can theoretically be used;
however, none so far have been stable in an optical trap.
Controlling all six degrees of freedom of Janus particles,
including three translational and three rotational, would open
up new applications not only in biophysical force and torsion
measurements, but also in microfluidics and material selfassembly.
Here we report on a new type of spherical Janus that can be
manipulated by a combination of optical and magnetic fields. We
demonstrate the ability to directly control five degrees of freedom
of the particle’s motion (three translational and two orientational)
while constraining the final sixth degree of freedom. Ultimately,
this demonstration represents the most control ever achieved
over freely suspended spherical colloidal particles and opens up
many exciting applications; the most obvious being the exertion
of torsional and linear forces on biomolecules. The main
achievement reported here was to develop a method of
[*] Dr. N. J. Jenness, Dr. R. M. Erb, Prof. B. B. Yellen
Mechanical Engineering and Materials Science
Center for Biologically Inspired Materials and Material Systems
Duke University
27708 Durham, North Carolina (USA)
E-mail: [email protected]
Prof. R. L. Clark
Hajim School of Engineering and Applied Science
University of Rochester
14627 Rochester, New York (USA)
DOI: 10.1002/adma.200900892
Adv. Mater. 2009, 21, 1–5
Final page numbers not assigned
COMMUNICATION
Towards Holonomic Control of Janus Particles in
Optomagnetic Traps
synthesizing magnetically anisotropic Janus particles that are
also compatible with conventional optical trapping systems. We
developed a novel lithographic technique for forming so-called
‘‘dot’’ Janus particles, which have a metallic coating covering
<20% of their surface area. The advantage of this approach is that
the dot Janus particles behave more like normal dielectric
particles in an optical trap, while also responding to magnetic
forces and torques produced by an external magnetic field.
Purely dielectric and metallic Mie and Rayleigh particles have
been optically trapped using a variety of techniques. Both dielectric
microparticles and nanoparticles can be trapped in three
dimensions with a high degree of spatial control.[11–13] Metallic
nanoparticles can also be trapped in three dimensions because
scattering from metallic and dielectric particles are similar in this
size regime.[19] However, metallic microparticles can only be
controlled in two dimensions, due to considerations previously
documented by others.[20–23] For anisotropic Janus particles, such
as dielectric particles that are partially covered by metal, the
trapping stability in a focused optical beam depends to a great
extent on the degree of metal coverage of the particle surface.
Here we propose a general explanation for why optical trapping
is more easily accomplished with dot Janus particles than with
half-coated Janus particles. In the Mie size regime, where the
particle diameter is large compared with the trapping wavelength,
l, the momentum imparted by a focused optical beam can be
described using geometric ray optics following Ashkin’s line of
reasoning.[24] In brief, each light ray refracts and reflects at the
particle/fluid interface according to Snell’s law, and the
momentum change between the incident ray and the
refracted/reflected ray is summed over all incident rays to
determine the net force on the particle. Typically, the net force is
artificially divided into a gradient force, arising from
refraction through the particle, and a scattering force, arising
from reflection at the particle surface. The gradient force tends
to pull the particle towards the beam focus, whereas the
scattering force tends to push the particle away from the
emission source.
Figure 1 illustrates the incident light rays a and b refracted
through the particle and the gradient forces ~
Fa and ~
Fb imparted
on the particle due to each light ray. The ray optics approach
reveals the importance of the symmetry of conjugate light rays in
an optical trap. As long as the gradient force balances the
scattering force, ~
Fs , the trap will remain stable. For particles
partially coated by reflective metal, the symmetry of this process
may be broken, leading to unbalanced torques and forces that
will depend on the position and orientation of the particle.
As illustrated in Figure 1b, the metal coating inhibits light
ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1
COMMUNICATION
www.advmat.de
2
focal point. Optical trapping capability should
theoretically be improved with dot Janus
particles, as illustrated in Figure 1c, because
the momentum imbalance on opposing sides
of the particle is greatly reduced.
There are many methods to fabricate Janus
particles including masking,[25] directional
evaporation,[7,26] microcontact printing,[27,28]
and interfacial fluid techniques.[29–32] Nanometer-size metallic Janus particles have been
fabricated using a convective assembly deposiFigure 1. A simplified ray-optics approach is illustrated for optical trapping of a) transparent, b)
tion technique[33] followed by metal evaporahalf Janus, and c) dot Janus particles. The smaller magnetic coating of dot Janus particles allows
for greater optical transmission through the particle and trapping of the particle near the focal tion or through self-assembled grafting techniques.[34] On the micrometer-size scale,
point. Conversely, the metallic coating of half Janus particles leads to larger momentum
imbalance on the opposing faces of the particles and hinders its stability near the focal point. metallic Janus particles are most commonly
produced by directional metal evaporation
onto colloidal monolayers.[7,26] We used variants of this technique to produce both
conventional half metallic Janus particles
and dot Janus particles. In brief, 10-mm
polystyrene particles in ethanol were pipetted
onto glass microscope slides, which formed
consistent regions of close-packed monolayers
upon solvent evaporation (Fig. 2a). A thin
50-nm cobalt layer was directionally evaporated onto the substrate using conventional
metal evaporation techniques (Fig. 2b). Next,
the substrate was sonicated in water to
resuspend the half Janus particles (Fig. 2c).
The dot Janus particles are engineered from
the same monolayer of particles (Fig. 2a),
however, an intermediary step is required to
partially mask the region of metal evaporation.
This is accomplished by first spinning a
1.5-mm-thick layer of photoresist onto another
glass slide. The two slides are pressed firmly
together into a sandwich, causing the photoresist to fill the voids between the particles and
both slides except in the region where steric
forces between the top glass slide and the
particles exclude the photoresist. Due to mass
balance considerations, it is possible that the
photoresist develops cavities in the interparticle regions. In any case, a uniform coating of
Figure 2. Techniques for fabricating half Janus and dot Janus particles are illustrated. In both the interfaces was reliably formed enabling
processes a monolayer of 10 mm fluorescent polystyrene particles are convectively assembled on
protection of the particles by the photoresist
a glass slide. Evaporation of cobalt onto the colloidal monolayers is typically used to produce half
Janus particles, which are subsequently released by sonication. Dot Janus particles are produced except at the contact points between the
by pressing the particles into photoresist in order to protect a larger region of the particle. Cobalt particles and the top slide. The sandwich is
is evaporated onto the photoresist film, followed by development of the remaining photoresist cured by heating the sample at 90 8C for 2 min,
and sonication, in order to produce Janus particles coated with a small dot. Insets are scanning after which the two slides can be pulled
electron micrographs with scale bars of b) 10 mm, e) 15 mm, and c,g) 2 mm.
apart. The particles remain anchored to the
bottom slide due to stronger adhesion forces,
leaving a small dot of 20% uncoated surface remaining on each
transmission through half of the particle leading to unbalanced
particle (Fig. 2e). A thinner 20-nm cobalt layer is evaporated
gradient and scattering forces, consequently expelling a Janus
onto the surface (Fig. 2f). Thinner metal films allow for easier
particle from the center of the trap. This effect was recently
lift-off when the sample is next developed to remove the
observed experimentally.[9] At low optical powers (7 mW), it was
photoresist and release the dot Janus particles into solution
demonstrated that Janus beads can be confined within an optical
(Fig. 2g). The particles can then be transferred to an aqueous
trap,[9] however, the trapping region in these cases is not strongly
solvent for testing.
localized and rotation of the particle can be observed about the
ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Mater. 2009, 21, 1–5
Final page numbers not assigned
www.advmat.de
it along the external field direction, according
*
*
*
to t m ¼ m0 m p H ext , where m0 is the
permeability of free space. For the
10-mm-diameter dot Janus particles fabricated
here, the maximum magnetic torque in a 50
Oe magnetic field is calculated to be 2 1015
N m. For low Reynolds number systems, such
as this one where inertial terms are negligible,
the rotation frequency of the particle can be
Figure 3. a) Coordinate axes are defined relative to particle orientation, as depicted. b) Illus- determined through the balance between
tration of bead trajectory and c) micrograph depicting an overlay of 12 images during optical magnetic torque and viscous drag, where
3
manipulation of a dot Janus particle with magnetic moment oriented along w ¼ 08. Manipulation g ROT ¼ 8pha , is the torsional friction on a
sequence starts at the arrow and moves particles along x, y, xz, and yz global axes. Inset in
particle of radius a due to a fluid with viscosity
(c) shows the orientation of the dot. Scale bar is 10 mm.
h. For sufficiently low rotation frequencies, the
particle becomes phase-locked with the exterIn order to test the optical compatibility of these Janus
nal field and follows the equations of motion of a non-linear
particles, each type of suspension was exposed to an optical trap.
harmonic oscillator, similar to systems studied by others.[41] For
The optical trapping setup, identical to Jenness et al.,[35] used a
this system, we calculate the critical frequency to be on the order
532-nm wavelength laser coupled with a spatial light modulator.
of hundreds of hertz, thus the angular velocity of the particle will
Half Janus particles were immediately expelled from the beam
nearly always be phase locked with the field. For the sake of
focus when the laser was turned on (see Supporting Information,
comparison, the magnetic torque in our experimental system is
Video 1). Upon lowering the average beam power below a certain
three orders of magnitude larger than torques previously reported
threshold (<8.0 mW), half Janus particles, though not trapped,
in similar optomagnetic systems,[17] which is a direct result of the
would remain near the focal point. Occasionally, these proximal
large volume of ferromagnetic material on the surface of these dot
particles were observed to rotate around the focal point,
Janus particles. This torque, of course, can be adjusted by
consistent with previous results.[9] In some cases, heating of
changing the thickness of the cobalt dot, however, this additional
the particle was observed as substantiated by the formation of air
control element was not tested here.
bubbles nucleating on the particle surface, a phenomenon
Although alignment in two dimensions can be achieved using
previously observed in colloidal suspensions of micropartia single uniform field, due to the axial symmetry of the field, the
cles.[36–39] However, at these lower beam intensities, the optical
particle will have one remaining orientational degree of freedom
traps could not overcome surface and gravitational forces to
that cannot be controlled. For example, if the field direction is
translate half Janus particles in a reliable fashion.
w,c ¼ 08, then the magnetic moment will be constrained to the
Alternatively, the dot Janus particles can be trapped in the focal
plane w,c ¼ 08. However, the particle is still free to rotate in u
point of the optical trap and subsequently controlled in three
(Fig. 4). It is possible to increase the control of the particle by
orthogonal spatial directions (x, y, z) (Fig. 3, see Supporting
using an optical trap in combination with external magnetic field.
Information, Video 2). In the coordinate system defined in
The optical trap can directly control three translational degrees of
Figure 3a, the trapped dot Janus particle remains free to rotate in
freedom of the particle while constraining one rotational degree
the c and w directions, however, the optical trap effectively
of freedom (u in this case). In addition, the magnetic field can
constrains the particle’s motion in the u direction in order to
directly control the remaining two orientational degrees of
minimize interference of the cobalt dot with the transmission of
freedom leading to near holonomic control of a colloidal particle.
the optical beam. While it is theoretically possible to directly
To demonstrate this system, three iron-core solenoids are placed
control u as well by using an optical beam that can be
independently rotated, this is technically very challenging and
was not demonstrated in this work. Trapping of both dot Janus
and bare 10-mm particles was consistently observed with laser
intensities in the range of 2 to 15 mW. The compatibility of the dot
Janus particles with the full range of trapping laser intensities
demonstrates that the metal coatings of dot Janus particles do not
impair the optical trapping capability.
To test the magnetic response of both types of Janus particles,
external magnetic fields were applied using iron-core solenoids.
Cobalt is ferromagnetic, retaining a remnant magnetic moment,
*
m
p , in the absence of magnetic field. The magnetic moment of
cobalt can be approximated from the magnetization of bulk
*
cobalt, M p ¼ 5000 0e,[40] and the volume of the cobalt patch, Vp,
*
*
Figure 4. Orientational control of dot Janus particles is demonstrated.
as m p ¼ M p Vp . Thin films of cobalt have large shape anisotropy,
a) Cartoon and b) fluorescent micrograph of the alignment of dot Janus
which causes the magnetic moment to be effectively pinned in the
*
particles’ magnetic moments with an applied external field (black arrows).
plane of the film. In the presence of an external field, H ext , the
Due to the axial symmetry of the applied field, particles remain free to rotate
*
moment of the particle will feel a magnetic torque, t m , that aligns
in u.
COMMUNICATION
Adv. Mater. 2009, 21, 1–5
3
Final page numbers not assigned
ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
COMMUNICATION
www.advmat.de
Experimental
Preparation of Colloidal Crystals: To deposit a
single layer of colloids onto a glass slide, 10-mm
polystyrene particles (Thermofisher, Waltham, MA)
were transferred from aqueous solvent by spinning
the particles down in a centrifuge at 2 units of relative
centrifugal force for 1 min and resuspending them in
ethanol. These particles were pipetted onto slightly
tilted glass microscope slides, which formed consistent regions of close-packed monolayers via
gravitation induced sedimentation combined with
solvent evaporation.
Development and Stabilization of Dot Janus
Particles: After metal evaporation (Fig. 2f), the glass
Figure 5. Simultaneous optical and magnetic control of a dot Janus particle is demonstrated.
slide with dot Janus particles is placed in a small glass
a) The magnetic field setup required for ‘dot’ Janus optical and magnetic manipulation. The
vial filled with M-351 (Microchem) for 5 min and
solenoids apply the magnetic field required for rotation while the optical trap positions the
sonicated for 10 s. With the dot Janus particles in
particle in three dimensions b) Overlay of seven images as dot Janus particle is spatially moved
suspension, the slide is discarded and the fluid is
around a predefined circle (dotted line) by a holographic optical trap and rotationally oriented by a
pipetted into microcentrifuge tubes and diluted
0.1 Hz rotating magnetic field. For visual aid, illustrations of the particle orientation are provided
ten-fold with 0.05% LiquiNox. The particles are spun
with the arrows indicating the direction of the magnetic field at each instance. c–e) Sequence of
down in a centrifuge at 2 units of relative centrifugal
images showing rotation of Janus particle around w axis while optically trapped. A particle with
force for 1 min and resuspended in 0.05% LiquiNox.
small debris was intentionally used to better visualize the orientation of the Janus particle in w.
This step is repeated an additional time.
Optical Trap Set-up: The laser was a pulsed
Coherent Primsa Nd:YVO4 512 nm. The microscope used was a Zeiss
along orthogonal axes that are positioned 5 cm from the
Axiovert 200 microscope fit with Zeiss 100, 40, and 10 lenses. The
sample (Fig. 5a). A dot Janus particle is first optically trapped
SLM used was a HOLOEYE LCR 2500 and was run in phase only mode. The
within the focal point allowing for control of x, y, z and a
beam path was encoded in phase holograms to create the desired intensity
distributions and played back in movie form to create preprogrammed
constraint of u ¼ 08. An external magnetic field is applied to orient
paths. To prepare a sample for the optical trap, dilute Janus particle
the particle in w and c directions (see Supporting Information,
suspensions (0.1% volume fraction) were placed between a microscope
Video 3).
slide and a glass coverslip separated by a 120-mm-thick spacer (Invitrogen,
In order to demonstrate the degree of particle control, a series
Carlsbad, CA).
of videos is provided as a visual aid. In one, a particle is moved
in a circular path via a programmed optical beam while a 30
Gauss, 0.1 Hz horizontally rotating magnetic field simultaneously orients the particle in c, (Fig. 5b, see Supporting
Information, Video 4). In another, a 30 Gauss, 1 Hz vertically
rotating magnetic field is shown to rotate the particle’s
orientation in w, (Fig. 5c–e, see Supporting Information, Video
5). Theoretically, it is possible to use holograms generated by
spatial light modulators to control multiple particles simultaneously.[11,12,35] This approach may also be combined with local
magnetic fields to achieve near holonomic control of an
ensemble of particles.[42,43]
In this work, we have developed a new method for the
controlled synthesis of dot Janus particles that have sufficiently
small surface coverage of cobalt film to allow compatibility
with an optical trap. The near holonomic control over these
spherical Janus particles opens up a new realm of advanced
manipulation strategies that will have important ramifications
in nanoengineering. The controlled manipulation of dot
Janus particles affords new degrees of freedom in the
measurement of mechanical properties of previously investigated
molecules such as DNA, ankyrin, and many other biomolecules.[17,44,45] Recent studies have also introduced the concept
of manipulating and controlling cell function through the
magnetic actuation of nanoparticles bound to the cell surface.[46,47] These potential biological applications as well as the
ability to improve directed hierarchical particle assembly establish
the broad reaching scientific value of the capabilities demonstrated here.
4
Acknowledgements
Randall M. Erb and Nathan J. Jenness contributed equally to this work. The
authors acknowledge funding from National Science Foundation grants:
IGERT Fellowship DGE-0221632 (R. M. E., N. J. J.), CMMI-0625480 and
CMMI- 0800173 (B. B. Y), a Nanoscale Interdisciplinary Research Team
grant: DMI-0609265 (R. L. C., N. J. J.), and a Medtronic Graduate
Fellowship (R. M. E.). Supporting Information is available online from
Wiley InterScience or from the author.
Received: March 14, 2009
Published online:
[1] A. Perro, S. Reculusa, S. Ravaine, E. Bourgeat-Lami, E. Duguet, J. Mater.
Chem. 2005, 15, 3745.
[2] T. Nisisako, T. Torii, T. Takahashi, Y. Takizawa, Adv. Mater. 2006, 18, 1152.
[3] S. Kim, S. Jeon, W. C. Jeong, H. S. Park, S. Yang, Adv. Mater. 2008, 20, 4129.
[4] J. N. Anker, C. J. Behrend, H. Huang, R. Kopelman, J. Magn. Magn. Mater.
2005, 293, 655.
[5] J. R. Howse, R. A. L. Jones, A. J. Ryan, T. Gough, R. Vafabakhsh,
R. Golestanian, Phys. Rev. Lett. 2007, 99, 048102.
[6] S. Gangwal, O. J. Cayre, O. D. Velev, Langmuir 2008, 24, 13312.
[7] H. Takei, N. Shimizu, Langmuir 1997, 13, 1865.
[8] K. K. Mehta, T. H. Wu, E. P. Y. Chiou, Appl. Phys. Lett. 2008, 93, 254102.
[9] F. S. Merkt, A. Erbe, P. Leiderer, New J. Phys. 2006, 8, 216.
[10] K. Svoboda, S. M. Block, Opt. Lett. 1994, 19, 930.
ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Mater. 2009, 21, 1–5
Final page numbers not assigned
www.advmat.de
Adv. Mater. 2009, 21, 1–5
Final page numbers not assigned
[30] Z. H. Nie, W. Li, M. Seo, S. Q. Xu, E. Kumacheva, J. Am. Chem. Soc. 2006,
128, 9408.
[31] K. H. Roh, D. C. Martin, J. Lahann, Nat. Mater. 2005, 4, 759.
[32] M. A. Correa-Duarte, V. Salgueiriño-Maceira, B. Rodrı́guez-González,
L. M. Liz-Marzán, A. Kosiorek, W. Kandulski, M. Giersig, Adv. Mater.
2005, 17, 2014.
[33] B. G. Prevo, O. D. Velev, Langmuir 2004, 20, 2099.
[34] M. Lattuada, T. A. Hatton, J. Am. Chem. Soc. 2007, 129, 12878.
[35] N. J. Jenness, K. D. Wulff, M. S. Johannes, M. J. Padgett, D. G. Cole,
R. L. Clark, Opt. Express 2008, 16, 15942.
[36] G. Eyring, M. D. Fayer, J. Appl. Phys. 1984, 55, 4072.
[37] C. M. Lawson, G. W. Euliss, R. R. Michael, Appl. Phys. Lett. 1991, 58, 2195.
[38] C. P. Lin, M. W. Kelly, Appl. Phys. Lett. 1998, 72, 2800.
[39] H. Lowen, P. A. Madden, J. Chem. Phys. 1992, 97, 8760.
[40] G. J. Parker, C. Cerjan, J. Appl. Phys. 2000, 87, 5514.
[41] B. H. McNaughton, K. A. Kehbein, J. N. Anker, R. Kopelman, J. Phys. Chem.
B 2006, 110, 18958.
[42] B. B. Yellen, R. M. Erb, D. S. Halverson, O. Hovorka, G. Friedman, IEEE
Trans. Magn. 2006, 42, 3548.
[43] B. B. Yellen, R. M. Erb, H. S. Son, R. Hewlin, H. Shang, G. U. Lee, Lab Chip
2007, 7, 1681.
[44] S. M. Block, L. S. B. Goldstein, B. J. Schnapp, Nature 1990, 348, 348.
[45] G. Lee, K. Abdi, Y. Jiang, P. Michaely, V. Bennett, P. E. Marszalek, Nature
2006, 440, 246.
[46] J. Dobson, Nat. Nanotechnol. 2008, 3, 139.
[47] S. Hughes, A. J. El Haj, J. Dobson, Med. Eng. Phys. 2005, 27, 754.
ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
COMMUNICATION
[11] J. E. Curtis, B. A. Koss, D. G. Grier, Opt. Commun. 2002, 207, 169.
[12] J. Leach, K. Wulff, G. Sinclair, P. Jordan, J. Courtial, L. Thomson, G. Gibson,
K. Karunwi, J. Cooper, Z. J. Laczik, M. Padgett, Appl. Opt. 2006, 45, 897.
[13] J. Liesener, M. Reicherter, T. Haist, H. J. Tiziani, Opt. Commun. 2000, 185,
77.
[14] L. Manna, E. C. Scher, A. P. Alivisatos, J. Cluster Sci. 2002, 13, 521.
[15] S. M. Anthony, L. Hong, M. Kim, S. Granick, Langmuir 2006, 22, 9812.
[16] C. Claudet, J. Bednar, Appl. Opt. 2005, 44, 3454.
[17] G. Romano, L. Sacconi, M. Capitanio, F. S. Pavone, Opt. Commun. 2003,
215, 323.
[18] L. Sacconi, G. Romano, R. Ballerini, M. Capitanio, M. De Pas, M. Giuntini,
D. Dunlap, L. Finzi, F. S. Pavone, Opt. Lett. 2001, 26, 1359.
[19] M. E. J. Friese, T. A. Nieminen, N. R. Heckenberg, H. Rubinsztein-Dunlop,
Nature 1998, 394, 348.
[20] H. Furukawa, I. Yamaguchi, Opt. Lett. 1998, 23, 216.
[21] A. T. O’Neil, M. J. Padgett, Opt. Commun. 2000, 185, 139.
[22] K. Sasaki, M. Koshioka, H. Misawa, N. Kitamura, H. Masuhara, Appl. Phys.
Lett. 1992, 60, 807.
[23] S. Sato, Y. Harada, Y. Waseda, Opt. Lett. 1994, 19, 1807.
[24] A. Ashkin, Proc. Natl. Acad. Sci. USA 1997, 94, 4853.
[25] C. Casagrande, P. Fabre, E. Raphael, M. Veyssie, Europhys. Lett. 1989, 9, 251.
[26] A. B. Pawar, I. Kretzschmar, Langmuir 2008, 24, 355.
[27] O. Cayre, V. N. Paunov, O. D. Velev, J. Mater. Chem. 2003, 13, 2445.
[28] O. Cayre, V. N. Paunov, O. D. Velev, Chem. Commun. 2003, 2296.
[29] K. Fujimoto, K. Nakahama, M. Shidara, H. Kawaguchi, Langmuir 1999, 15,
4630.
5