Subscriber access provided by NORTH CAROLINA STATE UNIV

Langmuir is published by the American Chemical Society. 1155 Sixteenth StreetN.W., Washington, DC 20036

Article

Dielectrophoretic Assembly of MetallodielectricJanus Particles in AC Electric FieldsSumit Gangwal, Olivier J. Cayre, and Orlin D. Velev

Langmuir, 2008, 24 (23), 13312-13320 • Publication Date (Web): 31 October 2008

Downloaded from http://pubs.acs.org on December 6, 2008

More About This Article

Additional resources and features associated with this article are available within the HTML version:

• Supporting Information• Access to high resolution figures• Links to articles and content related to this article• Copyright permission to reproduce figures and/or text from this article

Dielectrophoretic Assembly of Metallodielectric Janus Particles in ACElectric Fields

Sumit Gangwal, Olivier J. Cayre, and Orlin D. Velev*

Department of Chemical and Biomolecular Engineering, North Carolina State UniVersity,Raleigh, North Carolina 27695

ReceiVed May 17, 2008. ReVised Manuscript ReceiVed August 30, 2008

“Janus” particles with two hemispheres of different polarizability or charge demonstrate a multitude of interestingeffects in external electric fields. We reported earlier how particles with one metallic hemisphere and one dielectrichemisphere self-propel in low-frequency alternating current (AC) electric fields. Here, we demonstrate the assemblyof such Janus particles driven by AC electric fields at frequencies above 10 kHz. We investigated the relation betweenfield-induced dielectrophoretic force, field distribution, and structure of the assemblies. The phase space for electricfield intensity and frequency was explored for particle concentrations large enough to form a monolayer on a glasssurface between two gold electrodes. A rich variety of metallodielectric particle structures and dynamics were uncovered,which are very different from those obtained from directed assembly of plain dielectric or plain conductive particlesunder the action of fields of similar frequency and intensity. The metallodielectric particles assemble into new typesof chain structures, where the metallized halves of neighboring particles align into lanes along the direction of theelectric field, while the dielectric halves face in alternating direction. The staggered chains may assemble in variousorientations to form different types of two-dimensional metallodielectric crystals. The experimental results on theformation of staggered chains are interpreted by means of numerical simulations of the electric energy of the system.The assembly of Janus metallodielectric particles may find applications in liquid-borne microcircuits and materialswith directional electric and heat transfer.

1. Introduction

Isotropic spherical colloids, typically made of silica or latex,have been the focus of particle assembly for more than 50 years.There is growing recognition that anisotropic particles can beused to engineer the assembly of targeted structures.1 “Janus”particles, named by Casagrande and de Gennes2,3 after the Romanmythology god (who possessed two faces), are one such classof anisotropic colloids. Janus particles have surface coverage or“patchiness” yielding surface properties that are physically orchemically different. Various techniques have been formulatedrecently to synthesize Janus particles,4-11 as they are an interestingobject of study and can find potential applications in novelmaterials.

Anisotropic particles with two hemispheres of differentpolarizability and/or conductance have been produced by thermalevaporation12 or gold sputtering.13,14 Molecular simulations havebeen used to investigate the assembly of functionally anisotropicbuilding blocks.15-19 However, few experimental studies reporthow such Janus particles assemble or respond to external fields.

Hong et al. have assembled spherical particles with oppositeelectric charge on the hemispheres in a system where the particlediameter exceeds the electrostatic screening length.20 Theseparticles formed clusters rather than strings, and Monte Carlosimulations have been performed to analyze the experimentalresults. Behrend et al. have coated magnetic and nonmagneticparticles with metal and tracked the orientation and rotation ofthese particles in magnetic fields to calculate the torque actingon the particles.21 Takei et al. have investigated the orientationof anisotropic particles in low frequency (0.1-1 Hz) electricfields after chemically modifying the gold hemisphere withcharged thiols. The orientation of these dipolar particles in anelectric field depends on the pH, which controls the charge ofthe functionalized hemispheres.12 Crowley et al. have determinedthe dipole response of a “gyricon” ball (∼100 µm diameter),which consists of two hemispheres of different colors and differentelectrical properties.22 These gyricon balls (with white- and black-colored halves) can be rotated when immersed in a liquid andexposed to a uniform electric field, which has application in“electric paper” electronic displays.

Externally applied electric fields allow the precise tuning offorces exerted on the particles and the fluid medium by the field.Dielectrophoresis (DEP) is a force that emerges upon application

* To whom correspondence should be addressed. E-mail: odvelev@unity.ncsu.edu. Telephone: 919-513-4318. Fax: 919-515-3465.

(1) Glotzer, S. C.; Solomon, M. J. Nat. Mater. 2007, 6, 557–562.(2) Casagrande, C.; Veyssie, M. C. R. Acad. Sci 1988, 306, 1423–1425.(3) de Gennes, P. G. ReV. Mod. Phys. 1992, 64, 645–648.(4) Velev, O. D.; Lenhoff, A. M.; Kaler, E. W. Science 2000, 287, 2240–2243.(5) Cayre, O.; Paunov, V. N.; Velev, O. D. J. Mater. Chem. 2003, 13, 2445–

2450.(6) Love, J. C.; Gates, B. D.; Wolfe, D. B.; Paul, K. P.; Whitesides, G. M. Nano

Lett. 2002, 2, 891–894.(7) Fialkowski, M.; Bitner, A.; Grzybowski, B. A. Nat. Mater. 2005, 4, 93–97.(8) Nie, Z.; Li, W.; Seo, M.; Xu, S.; Kumacheva, E. J. Am. Chem. Soc. 2006,

128, 9408–9412.(9) Roh, K.; Martin, D. C.; Lahann, J. Nat. Mater. 2005, 4, 759–763.(10) Perro, A.; Reculusa, S.; Ravaine, S.; Bourgeat-Lami, E.; Duguet, E. J.

Mater. Chem. 2005, 15, 3745–3760.(11) Walther, A.; Muller, A. H. E. Soft Matter 2008, 4, 663–668.(12) Takei, H.; Shimizu, N. Langmuir 1997, 13, 1865–1868.(13) Suzuki, D.; Kawaguchi, H. Colloid Polym. Sci. 2006, 284, 1471–1476.(14) Paunov, V. N.; Cayre, O. J. AdV. Mater. 2004, 16, 788–791.

(15) Glotzer, S. C.; Horsch, M. A.; Iacovella, C. R.; Zhang, Z. L.; Chan, E. R.;Zhang, X. Curr. Opin. Colloid Interface Sci. 2005, 10, 287–295.

(16) (a) Zhang, Z.; Keys, A. S.; Chen, T.; Glotzer, S. C. Langmuir, 2005, 21,11547-11551. (b) Zhang, Z.; Glotzer, S. C. Nano Lett. 2004, 4, 1407-1413.

(17) Goyal, A.; Hall, C. K.; Velev, O. D. Phys. ReV. E 2008, 77, 031401.(18) Chen, T.; Lamm, M. H.; Glotzer, S. C. J. Chem. Phys. 2004, 121, 3919–

3929.(19) Pawar, A. B.; Kretzschmar, I.; Aranovich, G.; Donohue, M. D. J. Phys.

Chem. B 2007, 111, 2081–2089.(20) Hong, L.; Cacciuto, A.; Luijten, E.; Granick, S. Nano Lett. 2006, 6, 2510–

2514.(21) Behrend, C. J.; Anker, J. N.; McNaughton, B. H.; Kopelman, R. J. J.

Magn. Magn. Mater 2005, 293, 663–670.(22) Crowley, J. M.; Sheridon, N. K.; Romano, L. J. Electrostat. 2002, 55,

247–259.

13312 Langmuir 2008, 24, 13312-13320

10.1021/la8015222 CCC: $40.75 2008 American Chemical SocietyPublished on Web 10/31/2008

of an electric field (direct current, DC, or alternating current,AC) to a suspension of particles. The application of an electricfield across a suspension of isotropic colloidal particles leads totheir polarization. The DEP force arises when the particles′induced dipoles interact with a nonuniform electric field leadingto particle movement.23-25 The particles are either attracted(positive DEP) to the region of maximum field intensity or repelled(negative DEP) from it, depending on the effective particlepolarizability relative to the media. The particle “chaining” forceis described as a result of the attraction of the induced dipoleswithin the particles. The chaining force acting between particlesof the same type is always positive and attractive.

AC electric fields can be used to manipulate many types ofcolloidal particles in different media by simply adjusting ACelectric field parameters (magnitude, frequency, wave shape,wave symmetry, and phase).26-29 The AC dielectrophoreticmanipulation, separation, and assembly of nonconductive particlesis the largest and best developed research area. For example,DEP has been used to form cell arrays,30,31 sort and trapbioparticles32,33 and DNA,34-37 and manipulate and separatelive cells and polymer spheres.38-42 Conducting particles thathave been assembled by DEP include gold nanoparticles, quantumdots (CdSe semiconductor nanoparticles), and carbon nanotubes(CNTs).26 Our research group has used DEP for the formationof colloidal crystals,43-45 linear aggregates of nanoparticles,26,28,46

and assembly of biocomposites including arrays of live cells.47

We have used simulations to describe the assembly and calculatethe electric field distribution around pairs of dielectric particlesand pairs of metallic particles subject to an AC electric fieldbetween two coplanar electrodes.26

We report here new assembly effects and patterns driven byAC electric fields applied to suspensions of Janus metallodielectric

particles. In our previous work, we showed that the applicationof low frequency (j10 kHz) AC electric fields to suspensionsof micrometer-sized Janus particles with one dielectric hemisphereand one conductive hemisphere resulted in unbalanced liquidflows and nonlinear, induced-charge electrophoretic (ICEP)motion of the particles.48 The particles moved perpendicular tothe uniform applied electric field, with their polystyrenehemisphere forward. Here, we report the dielectrophoreticassembly of micrometer-sized metallodielectric Janus particlessuspended in water, subject to high frequency (>10 kHz) ACelectric fields. We first describe the technique for Janus particlepreparation, the experimental setup, and the procedures fornumerical simulations. The experimental data are presented andsummarized in the parameter space for electric field intensityand field frequency. The metallodielectric particle concentrationswere large enough to form a monolayer (on the glass surface),and a rich variety of structures was uncovered. Results fromtwo-dimensional (2D) numerical simulations of the electric energyof the system are correlated with the experimental results.

2. Experimental ProceduresMaterials. Deionized water with a resistivity of 18.2 MΩ cm was

obtained from a Millipore Milli-Q Plus water purification system.Surfactant-free polystyrene latex microspheres (diameter, D ) 4.0and 5.7 µm) stabilized by sulfate groups were purchased as aqueousdispersions from Interfacial Dynamics Corp. (OR). Nonionicsurfactant polyoxyethylene(20)sorbitan (Tween-20) was purchasedfrom Acros Organics (NJ). Ethanol and Teflon tape were purchasedfrom Fisher Scientific (PA).

Janus Metallodielectric Particle Preparation. The metallodi-electric particles used in the experiments were prepared by partiallycoating the polystyrene microspheres with a conductive layer ofgold on one hemisphere of the particles. The polystyrene particleswere initially concentrated by centrifuging at ∼2000g for 10 minand were washed with ultrapure Milli-Q water. A convective assemblymethod that we had engineered previously49 was used to depositparticle submonolayers on precleaned glass microscope slides (FisherScientific, PA). The dried particle submonolayers were coated with10 nm of chromium followed by 20 nm of gold in a metal evaporator(Cooke Vacuum Products, model FPS2-41). The chromium layerwas deposited to ensure that gold adheres to the exposed particlesurface. The “Janus” particles formed (Figure 1b) were thenredispersed in Milli-Q water by spraying the glass slides coveredwith metal-coated particles with a 1:3 (v/v) ethanol/Milli-Q watersolution to flush the particles using a 30 cc syringe fitted with a18G11/2 needle. The remaining particles on the glass slides wereremoved by mild sonication (for less than 20 s). Nonionic surfactantTween-20 (∼0.1 wt %) was added to the suspensions, and the particleswere washed with Milli-Q water to remove the ethanol. Additionof the nonionic surfactant prevented the particles from aggregating.High concentration suspensions of Janus particles (∼7-12% solids)in small volumes (∼50-100 µL) were prepared by centrifugation.The concentration of particles in suspension was calculated basedon the number of particles necessary to form a close-packedmonolayer on the bottom surface of the experimental cell (Figure1a).

Experimental Setup. The experimental cell (Figure 1a) wasconstructed from a glass slide onto which two coplanar gold electrodes(3 mm interelectrode gap) were deposited by evaporating 10 nm ofchromium followed by 100 nm of gold. A liquid blocker pen wasused to create a “corral” between the electrodes surrounded by ahydrophobic ring. A 2-3 µL droplet suspension of the Janus particleswas placed within this area onto the bottom microscope glass incontact with both electrodes. A microscope glass coverslip was placeddirectly on top of the particle suspension, forming a thin experimental

(23) Jones, T. B. Electromechanics of Particles; Cambridge University Press:Cambridge, 1995.

(24) Pohl, H. A. Dielectrophoresis; Cambridge University Press: Cambridge,1978.

(25) Morgan, H.; Green, N. G. AC Electrokinetics: colloids and nanoparticles;Research Studies Press Ltd.: Hertfordshire, UK, 2003.

(26) Velev, O. D.; Bhatt, K. H. Soft Matter 2006, 2, 738–750.(27) Velev, O. D. In Colloids and Colloid Assemblies; Caruso, F., Ed.; Wiley-

VCH: Weinheim, 2004; pp 437-464.(28) Bhatt, K. H.; Velev, O. D. Langmuir 2004, 20, 467–476.(29) Chang, S. T.; Beaumont, E.; Petsev, D. N.; Velev, O. D. Lab Chip 2008,

8, 117–124.(30) Alp, B.; Stephens, G. M.; Markx, G. H. Enzyme Microbiol. Technol.

2002, 31, 35–43.(31) Chiou, P. Y.; Ohta, A. T.; Wu, M. C. Nature 2005, 436, 370–372.(32) Cheng, I.; Chang, H.; Hou, D.; Chang, H. Biomicrofluidics 2007, 1, 021503.(33) Clarke, R. W.; White, S.; Zhou, D.; Ying, L.; Klenerman, D. Angew.

Chem. 2005, 44, 3747–3750.(34) Parikesit, G. O. F.; Markesteijn, A. P.; Piciu, O. M.; Bossche, A.;

Westerweel, J.; Young, I. T.; Garini, Y. Biomicrofluidics 2008, 2, 024103.(35) Washizu, M. J. Electrost. 2005, 63, 795–802.(36) Bakewell, D.; Morgan, H. IEEE Trans. Nanobiosci. 2006, 5, 1–8.(37) Sung, K. E.; Burns, M. A. Anal. Chem. 2006, 78, 2939–2947.(38) Fuhr, G.; Arnold, W. M.; Hagedorn, R.; Muller, T.; Benecke, W.; Wagner,

B.; Zimmermann, U. Biochim. Biophys. Acta 1992, 1108, 215–223.(39) Huang, Y.; Joo, S.; Duhon, M.; Heller, M.; Wallace, B.; Xu, X. Anal.

Chem. 2002, 74, 3362–3371.(40) Minerick, A. R.; Ostafin, A. E.; Chang, H. C. Electrophoresis 2002, 23,

2165–2173.(41) Gray, D. S.; Tan, J. L.; Voldman, J.; Chen, C. S. Biosens. Bioelectron.

2004, 19, 1765–1774.(42) Auerswald, J.; Knapp, H. F. Microelectron. Eng. 2003, 67-8, 879–886.(43) Velev, O. D.; Lumsdon, S. O. In Handbook of Surfaces and Interfaces

of Materials; Nalwa, H. S., Ed.; Academic Press: San Diego, CA, 2002; pp125-163.

(44) Lumsdon, S. O.; Kaler, E. W.; Williams, J. P.; Velev, O. D. Appl. Phys.Lett. 2003, 82, 949–951.

(45) Lumsdon, S. O.; Kaler, E. W.; Williams, J. P.; Velev, O. D. Langmuir2004, 20, 2108–2116.

(46) Hermanson, K. D.; Lumsdon, S. O.; Williams, J. P.; Kaler, W.; Velev,O. D. Science 2001, 294, 1082–1086.

(47) Gupta, S.; Alargova, R. G.; Kilpatrick, P. K.; Velev, O. D. Soft Matter2008, 4, 726–730.

(48) Gangwal, S.; Cayre, O. J.; Bazant, M. Z.; Velev, O. D. Phys. ReV. Lett.2008, 100, 058302.

(49) Prevo, B. G.; Velev, O. D. Langmuir 2004, 20, 2099–2107.

Assembly of Metallodielectric Janus Particles Langmuir, Vol. 24, No. 23, 2008 13313

cell approximately 2-3 particle diameters in height. The liquid dropletspread out to an area of 60-100 mm2 once it was compressed betweenthe top and bottom glass slides.

Particle chaining and crystallization (for concentrations largeenough to form a monolayer) in the chamber were observed fromabove using an Olympus BX-61 optical microscope (with 40× or50× objective), and images were recorded using an Olympus DP-70digital CCD camera. Digital movies of the Janus particles’ behaviorat high and low frequencies were recorded using a Sony CybershotDSC-V1 camera fitted to the eyepiece of the microscope.

The alternating electric field within the experimental cell wasproduced by an Agilent 33120A 15 MHz function generator (AgilentTechnologies, CO) connected to a RG-91 ramp generator/amplifier(Burleigh, NY). An AC field (of square waveform) of voltage rangingfrom 1 to 90 V and frequencies from 1 to 200 kHz was applied tothe particle suspension. A 1 µF capacitor was included in the circuitto filter any direct current component of the signal. The voltageapplied in the chamber was measured with a digital multimeter(Instek, CA). A master switch allowed starting and stopping theprocess.

Numerical Simulation. During one-half cycle of the applied ACfield, the electric field distribution and electric energy distributionaround the Janus particles inside the experimental cell were simulatedby 2D electrostatic calculations using the FEMLAB multiphysicsmodeling package (COMSOL, Burlington, MA). The geometry ofthe system with metallodielectric particles (simulated with diameterof 5 µm) and the 3 mm gap between the electrodes were specifiedas a 2D cross-sectional top view of the experimental cell presentedin Figure 1a. In order to model the system accurately, we incorporatedthe effect of the counterionic atmosphere (electrical double layer)around the bare polystyrene half of the particles to account for thehigher surface conductivity of this layer.

The solution space was divided into four main subdomains: watermedia, dielectric polystyrene core, gold cap on one-half of the particle,and counterionic atmosphere on the other half of the particle. Thevalues of the physical properties of these subdomains are listed in

the Appendix. After the boundary conditions within the experimentalcell were specified, the solution space was triangulated into aconformal mesh and the mesh was refined at least four times. Thesolver was initialized to solve the Poisson equation for all elementsto obtain the electric field intensity and electrical energy densitywithin the cell. The subdomain integration function was used tocalculate the electric energy of the entire 2D system. This functionintegrated the electric energy density over the area (since this wasa 2D simulation) of the system after selecting all four of thesubdomains. The calculations were repeated with more refined meshsizes until the mesh was small enough for the final calculated valuesto vary by less than 0.05%.

3. Experimental Results

Effect of AC Field Strength and Frequency. A wide varietyof novel structures were formed and diverse particle dynamicswere observed when the voltage and frequency of the appliedAC field were varied at Janus particle concentrations large enoughto form a monolayer on the bottom substrate. The crystallizationand electrohydrodynamic mobility thresholds for the anisotropicmicrospheres were explored by adjusting the parameters of theapplied field: the Janus particles could be made to displayelectrohydrodynamic mobility in the direction perpendicular tothe electric field as demonstrated in previous work;48 form crystalsof unique symmetries by confining staggered chains of Janusparticles in a small area; or form more complex three-dimensional(3D) bundle structures. The staggered chains, 2D crystalstructures, and 3D bundle structures disassembled once the electricfield was turned off, proving that this process was reversible.

The different regions of the dynamic and structural behaviorof particles within the space of field intensity and frequency arepresented in Figure 2. The AC field-induced dielectrophoreticor electrohydrodynamic response of the metallodielectric mi-crospheres in Milli-Q water depends strongly on the frequencyof the applied field. At higher frequencies, the metallodielectricparticles formed staggered chains, 2D metallodielectric crystals,and 3D bundles, whereas at lower frequencies the anisotropicparticles formed regular chains and performed induced-chargeelectrophoretic (ICEP) motion in the direction perpendicular tothe applied field direction. The transition from staggered chainstructures formed at high AC field frequency to ICEP motion atlow frequency is shown in the Supporting Information. Theregions outlined in Figure 2 are reviewed in more detail and theirstructure is analyzed in the next subsections.

Disordered Particles Region. Below ∼25 V cm-1 over thewhole range of frequencies (1-200 kHz), the Janus metallodi-electric particles remained disordered and were randomlydistributed on the surface of the bottom glass slide due to Brownianmotion (Figure 2, bottom). Most of the Janus particles wereobserved to assume an orientation such that the dark, gold-coatedhemisphere was facing up and the light, polystyrene hemispherewas facing down, in contact with the bottom glass surface of theexperimental cell. This could result from differences in theinteractions between the metallic/polymer halves of the particlesand the glass bottom wall of the experimental cell; however, thiseffect is easily overridden by the much stronger field-inducedinteractions and orientation at higher field strengths.

Induced Charge Electrophoresis (ICEP) Region. Above fieldsof approximately 75 V cm-1, ICEP motion resulted at ACfrequencies of 1-40 kHz (Figure 2), whereby the metallodielectricJanus particles moved in directions perpendicular to the electricfield with their polystyrene hemispheres facing forward in thedirection of motion. Our previous report explains the physicalphenomenon and characterizes this ICEP motion as a functionof field strength and frequency, electrolyte concentration, and

Figure 1. (a) Schematic illustration of the experimental cell withanisotropic particle crystals formed upon applying an AC electric fieldand (b) SEM image of 4.0 µm metallodielectric particles. The gold-coated hemispheres appear brighter due to their higher conductance.The scale bar in (b) is 4 µm.

13314 Langmuir, Vol. 24, No. 23, 2008 Gangwal et al.

particle size.48 Briefly, the electric double layer on the gold-coated conductive hemisphere of the particle is more stronglypolarized in the applied field, which drives a stronger induced-charge electro-osmosis (ICEO) slip than that on the dielectricpolystyrene side, resulting in ICEP motion in the direction of theuncoated side. Although we previously reported that ICEP motiondecreases at about 12 kHz in cells of 60-80 µm heights, wefound that ICEP motion persisted up to ∼40 kHz at high fieldstrengths in experimental cells of smaller heights (10-15 µm).The transition to the ICEP region from the 2D crystallization and3D bundle region in Figure 2 was sensitive to the operatingparameters. By slightly increasing the field strength, the stationary2D crystals formed at low frequencies and lower field strengthscould be forced to disassemble and perform ICEP motion, leadingto “melting” of the crystals, which may be related to tangentialdouble layer conductance.50 Also, by slightly decreasing the fieldfrequency, the formed stationary 3D bundles at mediumfrequencies and medium field strengths could be made todisassemble and display ICEP motion. The particle chains tendedto disassemble from the ends of the structure. Individual particlesin 3D bundles or in staggered chains detached from the chainand began ICEP motion as the structure collapse proceeded inwardfrom the ends (as displayed in the Supporting Information).

3D Bundles Region. The particles formed 3D bundles (topoptical image of Figure 2, and Figure 4c) above ∼75 V cm-1

over the frequency range of 5-200 kHz. In the higher fieldstrength, the particles in the 2D metallodielectric crystals packedmore tightly. The particle chains stacked on top of each otherup to the height of the experimental cell forming the 3D bundles(which might be nuclei for 3D crystals). Large void areas areobserved between the 3D bundles. The higher DEP and particlechaining forces within the cell resulted in particle chains merginginto the 3D bundles, increasing the bundlesߣ width andheight. These bundles also tended to stretch toward each of theelectrodes. Some the particles were observed to rotate abouttheir axes within the 3D bundles when higher field strengthswere applied. Most of the particles in the 3D bundle opticalimages are observed with their dark, gold-coated hemispheresfacing up.

Regular and Staggered Chains Region. The particles formedregular, straight chains parallel to the electric field direction similar

to chains observed for plain latex particles in applied AC electricfields44,45 and staggered chains from 25 to ∼40 V cm-1 over thewhole range of frequencies studied (1-200 kHz). At frequencieslower than ∼10 kHz and low field strengths, the regular, parallelchains were oriented such that mostly the darker, gold-coatedhemisphere of the particle was facing up (away from the bottomglass surface) as seen in the bottom left optical image of Figure(50) Basuray, S.; Chang, H. Phys. ReV. E 2007, 75, 060501-1.

Figure 2. Dynamic and structural response of Janus particles to AC electric field intensity versus field frequency in a thin experimental cell. In theoptical images, the gold-coated, conductive hemispheres appear dark and the bare, dielectric hemispheres appear light. The electric field directionis between the top and bottom of the optical images. The regions were established on the basis of 28 data points.

Figure 3. Optical micrographs of (a) staggered chains formed at lowerconcentration of Janus particles (5.7 µm diameter) in an AC field of 56V cm-1 at 40 kHz and (b) concentrated staggered chains formed witha particle concentration enough to form a monolayer in an AC field of27 V cm-1 at 40 kHz. The gold-coated, conductive hemispheres appeardark and bare, and dielectric hemispheres appear light. The electric fielddirection is between the top and bottom of the images. The scale barsin (a) and (b) are 70 and 50 µm, respectively.

Assembly of Metallodielectric Janus Particles Langmuir, Vol. 24, No. 23, 2008 13315

2. At frequencies above ∼10 kHz, the Janus particles formedstaggered chains (bottom right optical image of Figure 2, andFigure 3a). When the electric field in the cell was turned on, theparticles first oriented so that the plane between their hemispheres(gold-coated, conductive hemisphere appearing dark and bare,dielectric hemisphere appearing light) aligned in the directionof the electric field and subsequently formed chains. Within thestaggered chain, the particles positioned themselves such thatthe polystyrene hemispheres of each particle were facing inalternating directions and only the gold-coated portions of the

particles were in contact near the poles of the particle. Thestaggered chains were thus seen to have a dark gold line throughoutthe length of the chain aligned with the direction of the electricfield.

When the electric field was turned off, both the straight andstaggered chains came apart and the particles redispersed. Thechain disassembly once the field was turned off could be attributedto both steric repulsion between the particles by the polyoxy-ethylene chains of the Tween 20 adsorbed on the surface of theparticle and Brownian motion. The particles usually reorientedthemselves so that their gold-coated hemispheres were facing upand the polystyrene hemisphere was in contact with the bottomglass substrate. This disassembly process contrasts with theaggregation of metallic gold nanoparticles induced by an ACelectric field (within a cell similar to the one used here), wherethe assembled structures remained stable even after the voltagewas turned off.28,46 The aggregation of these gold nanoparticlesis irreversible due to the strong van der Waals forces acting onthe nanoparticles. The disassembly process of metallodielectricJanus particles is similar to that of plain sulfate-stabilizedpolystyrene particles in AC electric fields, where the negativelycharged sulfate groups on the plain polystyrene surface promoterepulsion from overlapping electric double layers between theparticles.51

2D Crystallization Region. By further increasing the electricfield intensity within the cell, we attempted to form two-dimensional lattices of staggered chains (Figure 3b). There is anarrow window for 2D metallodielectric crystal formationbetween ∼40 and 75 V cm-1 across the range of frequenciesstudied, where lattices form due to dipolar attraction of contiguouschains. At low frequencies (1-10 kHz), 2D crystals formed withthe gold-coated hemispheres of the particles facing up and theuncoated side facing the bottom glass substrate (Figure 2, middle-left optical image). From 10-200 kHz, we observed formationof two types of crystals (Figure 2, middle optical images). Thefirst particle lattice comprised narrow parallel metallic lanesthroughout the crystal (Figure 4a). This lattice was formed whenthe applied higher voltage directed the assembly of staggeredchains (formed in the lower-field region in Figure 2) into a close-packed lattice. Additionally, a second type of lattice appearedto possess broader metallic lanes due to the different positionsof the approaching chains of particles with respect to each other.The chains forming this type of lattice were grouped togetherwith the gold-coated hemispheres of each of the chains facingone another (Figure 4b). Within these two particle chain sets, thegold portion of each particle touched the gold portion of itsneighboring four particles near the equator and the poles ratherthan just near the poles as in the first type of Janus crystal. Thedielectric half of the particle was adjacent to the dielectric halfof a particle from the next chain set. The first type of crystallattice was observed more frequently and was larger in domainareas than the second type over the range of frequencies studied.

Characterization of Metallodielectric 2D Crystals. Wefurther characterized the staggered chain and 2D crystallizationregions using the orientational order and polarization parameters,which are commonly used to characterize dipolar liquid crystals.Similarly to liquid crystals, a specific orientation of themetallodielectric particles within the 2D crystalline lattices canbe designated a specific orientation with regards to the planebetween the metallic and dielectric halves of the particles. Theorientation order parameter, S, can be expressed as the average

(51) Hunter, R. J. Foundations of Colloid Science; Oxford University Press:Oxford, 2001.

Figure 4. Optical micrographs of (a) a crystal of staggered chains, (b)a crystal of two coexisting particle chain arrangements in an AC fieldof 125 kHz frequency, and (c) 3D bundles in an AC field of 50 kHzfrequency. The gold-coated, conductive hemispheres appear dark andbare, and dielectric hemispheres appear light. The electric field is appliedin the vertical direction in the images. The scale bars in (a-c) are 50µm.

13316 Langmuir, Vol. 24, No. 23, 2008 Gangwal et al.

of (3cos2 θ- 1)/2 over all the particles.52 The angle, θ, is betweenthe orientation vector of a given particle and the director n (seeinset of Figure 5). S is equal to unity in perfectly alignedferroelectric and antiferroelectric phases and equal to zero in theisotropic phase.52 The polarization parameter P is given by

P) |1N∑i)1

N

(ei · n)| (1)

P can be expressed as the average of cos θ over all ofthe particles.53 P is equal to unity in a perfectly alignedferroelectric phase and zero in the antiferroelectric phase. Stherefore measures orientational order, and P distinguishesbetween ferroelectric and antiferroelectric phases. In a ferroelectricphase, P g S, while in an antiferroelectric phase P < S.

We measured the orientation angle for 727 particles byprocessing the optical digital image in Figure 3b. The instan-taneous director n was designated in the direction of the appliedelectric field, and the angle θ was defined as the angle betweenthe gold/polystyrene interface and the director n. The angle θ )0° was designated for the case when the gold/polystyrene interfacewas fully aligned with the direction of the electric field and thegold-coated hemisphere was situated on the left side of the particle.The angle θ ) 180° was designated for the case when the gold/polystyrene interface was fully aligned with the direction of theelectric field and the gold-coated hemisphere was situated on theright side of the particle. The distribution of the orientation anglearound θ ) 0° and θ ) 180° was determined. The number ofparticles at a specific orientation angle for the image processedis plotted in Figure 5. The two peaks in the number of particlesat 0° and 180° orientation indicate that the particles orient withhighly specific direction. The orientation order parameter S forthis image was calculated to be S ) 0.98. This is very close tounity, which indicates that the crystal is in either of the well-aligned ferroelectric or antiferroelectric phases. To furthercharacterize this 2D metallodielectric crystal, the polarizationparameter P was calculated to yield P)-0.003, which indicates

that the metallodielectric 2D lattice structures are in a well-aligned antiferroelectric lattice.

4. Modeling of Janus Particle Orientation andStaggered Chain Formation in an Electrical Field

We calculated the electric field distribution and the energy ofthe system for different particle configurations in order tounderstand why individual Janus particles orient in the directionof the electric field in such a manner and to model the formationof staggered chains. The total electric energy We of the systemcan be obtained by integrating the local energy density, wes

(defined in the Appendix), over the subdomain volume (V)

We )∫Vwes dV (2)

Particles responding to dielectrophoretic force are attracted (ifthey are more polarizable than the media) to the high field intensityarea in an electric field gradient so that the minimum potentialenergy is reached when the particles are closest to the point ofhighest electric field strength.54,55

We performed a 2D FEMLAB simulation of a complexparticle-dielectric system inserted inside a parallel plate capacitor,since our system consists of two electrodes separated by adielectric media. We kept the voltage constant and only changedthe orientation angle and/or the positions of the Janus particlesto form different types of four-particle chain configurations. Wethen calculated the total stored electric energy (effective unitsof J m-1, since this was a 2D simulation and we integrated overthe area rather than volume). This procedure was repeated untilwe found the configuration with a maximum in stored electricenergy. We refer to this as a “quasi-Monte Carlo” approach: itis broadly similar to the methodology of a Monte Carlosimulation;56 however, since our system is quite complex andthe energy calculations in each step require a very computationallyexpensive solution of a system of partial differential equations,we only model a few selected configurations and compare theirenergy.

The dielectric permittivity of the subdomains was calculatedby means of the complex permittivity (which accounts for thefrequency of the field) as given by Morgan et al.25

ε) ε0εr -i(σ)ω

(3)

where i is the imaginary unit, σ is the electrical conductivity, andω is the AC electric field frequency (for which we specified avalue of 10 kHz). The dielectric subdomains specified inFEMLAB and the details of the conductivity and permittivityvalues used are listed in the Appendix. The top and bottomelectrodes were energized in the simulations with ( 0.1 V in thecase of the particle orientation simulation and (0.25 V forthe particle configuration simulation. In both cases, the effectivefield strength simulated within the cell was 100 V cm-1. Theboundaries on the sides of the experimental cell were taken tobe electrically symmetrical.

Orientation Angle Simulation. To understand why anindividual particle is oriented with its gold/polystyrene interfacealigned with the electric field, we calculated the electric energyof the system containing a single particle at different orientationangles (Figure 6). The 0° orientation angle (Figure 6 inset) wastaken as the baseline, since this was the equilibrium orientation

(52) Pohl, L.; Finkenzeller, U. Physical Properties of Liquid Crystals. In LiquidCrystals: Applications and Uses; Bahadur, B., Ed.; World Scientific: Singapore,1990; Vol. 1, pp 140-169.

(53) Cabral, B. J. C. J. Chem. Phys. 2000, 112, 4351–4356.

(54) Wang, X. B.; Huang, Y.; Burt, J. P. H.; Markx, G. H.; Pethig, R. J. Phys.D: Appl. Phys. 1993, 26, 1278–1285.

(55) Hughes, M. P. J. Phys. D: Appl. Phys. 2004, 37, 1275–1280.(56) Frenkel, D.; Smit, B. Understanding Molecular Simulation From

Algorithms to Applications; Academic Press: San Diego, 2002.

Figure 5. Number of particles as a function of the particle orientationangle, θ, where the angle is determined by the angle between the gold/polystyrene interface and the direction of the electric field. Two highpeaks in the angle distribution are observed at orientation angles of 0°and 180°, in line with the direction of the electric field.

Assembly of Metallodielectric Janus Particles Langmuir, Vol. 24, No. 23, 2008 13317

angle that was observed experimentally. The total electric energyof the system was calculated from -90° to 90° by rotating theparticle about its gold/dielectric interface axis in 5° increments.The potential energy is maximal at -90° when the gold/polystyrene interface is perpendicular to the electric field direction.The energy difference decreases as the particle is incrementallyrotated to 0°, which results in the minimum in potential energyof the system. This corresponds well to the experimentallyobserved orientation of the particles (Figures 3a and 5), whichis between -10° and 10°. We multiplied the total integratedelectric energy of the system by the particle radius (2.5 µm) toestimate the total energy of the system (effective units of J,converted into units of thermal energy kT). The potential energydifference between -90° orientation and 0° orientation anglewas ∼100 kT, which is a reasonable value for a body of this sizethat has been arrested from thermal or hydrodynamic fluctuations.This is equivalent to the induction of a dipole of maximal strengthin the gold-coated hemisphere (aligned with the field) and isbroadly similar to the alignment of elongated and rodlike particlesalong the electric field direction.23

Particle Chain Configuration Simulation. After establishingthe origin of the orientation of an individual particle in the electricfield, we simulated the energy profiles of different chainconfigurations of Janus particles to determine which one wouldresult in the minimal potential energy of the system. In allconfigurations simulated, the gold/polystyrene interface of allJanus particles was aligned with the field direction, as we hadestablished this orientation in the previous simulation. Thebaseline configuration included four Janus particles forming astaggered chain (Figure 7c), since this was the arrangement thatwas predominantly observed experimentally. For this baselineconfiguration, the gold shells touched slightly off-center (com-pared to a regular, straight chain) near the poles of the particleand the polystyrene half of each particle was facing in alternatingdirections. This base assembly is compared with the followingalternative configurations in Figure 7: particles with gold shellsdirectly in contact with the dielectric portion of adjacent particlewithin a straight chain (Figure 7a), gold shell directly in contactwith the gold shell of an adjacent particle within a straight chain

(Figure 7b), and gold shells touching more off-center in astaggered chain (Figure 7d).

The simulation shows that the minimum in potential energydifference does indeed occur for the experimentally observedconfiguration where the gold shells are in contact with eachother slightly off-center in a staggered chain (Figure 7c). It appearsthat the gold-coated hemisphere (which is nearly infinitelypolarizable) dominates in determining particle structures anddynamics at medium and high electric field intensities. After theparticles orient to 0° or 180° to align their gold/polystyreneinterface with the electric field, their gold-coated hemispheresare attracted to each other near the poles to align the largestdipoles (created within the gold coating) with each other alongthe electric field lines. The particles alternate with the polystyrenehalf of the particle on the left and right sides so that only the goldshells are in contact with each other, minimizing the potentialenergy of the chain configuration and resulting in a narrowconductive lane throughout the length of the chain. When the

Figure 6. Effect of orientation angle of the particle on the calculatedpotential energy difference. The inset simulation illustrates the electricenergy density contour around a single Janus particle at a 0° orientationangle in one half-cycle of an AC electric field. The direction of theelectric field is indicated on the simulation image, and the yellow arcon the left side of the particle represents the gold shell. The other halfof the particle was simulated with a counterionic layer.

Figure 7. Simulations of the electric energy density contours arounddifferent particle configurations (a-d) of four Janus particles in onehalf-cycle of an AC electric field and the effect of particle configurationof four particles on the potential energy difference (e). The bar to theright indicates the intensity of the electric energy density in a-d (in Jm-3), and the direction of the electric field is indicated on each simulation.The yellow arc represents the gold shell in (a-d).

13318 Langmuir, Vol. 24, No. 23, 2008 Gangwal et al.

particles are situated so that their point of contact is further off-center within a staggered chain where the gold shells are incontact (Figure 7d), the potential energy increases dramatically.The potential energy of the system for configuration B where thegold shells are on the same side of a nonstaggered chain (Figure7b) is closest to the system that exhibits the minimum in potentialenergy. This type of chain is similar to the chains found in a typeof Janus crystal with broader metallic lanes observed experi-mentally (Figure 4b), although in Figure 4b there are two setsof chains grouped together, which may be a kinetically trappedconfiguration.

Overall, the particle orientation and particle configurationsimulations agreed well with the experimentally observed results.The same simulations were also performed in vacuum, withoutaccounting for the counterionic conductance of the bare side ofthe Janus particle and the complex permittivities of thesubdomains. We obtained a qualitatively similar result as for thesimulations performed in water, for which we take into accountthe complex permittivity of each of the subdomains. Thus, thepresence of the counterionic atmosphere on the dielectric halfof the particles appears to modulate the interactions, but the highpolarizability of the metal half is the leading effect in the assemblyprocess. This approach could in the future be used to predict theorientation and assembly of various anisotropic and patchyparticles in applied electric fields.

5. Conclusions

The phase diagram for electric field intensity and field frequencyfor a monolayer concentration of Janus particles reveals fiveregions of particle behavior: disordered particles region, regularand staggered chains region, 2D crystallization region, inducedcharge electrophoresis (ICEP) region, and 3D bundles region.The staggered chain and 2D crystallization regions were furtherinvestigated by using electrostatic simulations to calculate theelectric energy of the system to determine the most favorableparticle orientation and lattice configuration. The simulationsagreed with experimental observations and aided in the under-standing of the Janus particle orientation along the direction ofthe electric field lines, and the staggered chain formation. Thefield-directed assembly of metallodielectric particles at highfrequency could be used in the fabrication of photonic crystalsof new symmetry types, massively parallel waveguides, liquid-borne microcircuits, and materials with directional electrical andheat transfer.

The rich variety of structures formed and dynamic motion ofthe metallodielectric Janus particles demonstrated here providea glimpse of the interesting phenomena occurring when aniso-tropic particles are subjected to external fields. The metal-coatedhemispheres of the particles play a key role in the formation ofdifferent structures and in the electrohydrodynamic mobility ofthe particles once the electric field intensity within the experi-mental cell becomes strong enough to overcome Brownianmotion. The experimental technique and simulations may nowbe applied to other types of anisotropic particles, which mayform different types of novel structures and potentially lead tothe fabrication of new materials.

Acknowledgment. This research was supported by an NSF-NIRT grant (CTS-0506701), and it is part of the NSF NanoscaleInterdisciplinary Research Team for nanoscale directed self-assembly in electrical and optical fields in collaboration with theUniversity of Delaware and California Institute of Technology.We also acknowledge support from an NSF-CAREER grant(CTS-0238636,) and a Camille Dreyfus Teacher-Scholar award.

Appendix

Electric Energy Definition and Dielectric SubdomainValues Specified in FEMLAB

The electric energy of the system is directly related to theelectric field intensity in the cell. The local electric energy densitywes for particles in vacuum is given by

wes )12

DE) 12

ε0E2 (4)

where D is the electric flux density, ε0 is the dielectric permittivityof vacuum (8.854 × 10-12 C2 N-1 m-2), and E is the electricfield intensity.57,58 For our system, which is not in vacuum, theelectric flux density D is given by

D) ε0εrE (5)

where εr is the relative permittivity.The dielectric subdomains defined in FEMLAB were the water

media (ε ) 78ε0 - [i(10-4)/ω]), a thin (100 nm thickness)conductive counterionic shell (ε) 78ε0 - [i(0.2)/ω]) on the barepolystyrene half side of the particle, and the two portions of theparticle.19,59 These two parts of the particle include the dielectricpolystyrene core (ε ) 2.55) and a very thin (30 nm thickness)conductive gold shell (ε ) 109ε0 - [i(4 × 107)/ω]) on one halfside.25,59 By using a very high value for the polarizability ofgold, we effectively introduce a domain with high conductivity(i.e., ideal metal). The simulations confirm that this leads tocomplete suppression of the electric field in the metal shell, asexpected for an ideally conductive metal.

Complex permittivity was not used for the polystyrene core,since its electrical conductivity is negligible (∼10-19 S m-1) andthe constant bulk value of 2.55 was used for its permittivity.60

The gold metal coating was specified with bulk gold values forconductivity and permittivity (with εf∞ for a perfect conductor).For the Milli-Q water media, we used a value of 10-4 S m-1 forthe conductivity and 78 for the permittivity.59 Carbon dioxidedissolved in the water in contact with air yields ∼10-5 M ionconcentration (with no electrolyte added), for which we calculateda Debye length of ∼100 nm. This is the effective thickness wespecified for the conductive counterionic layer around theuncoated portion of the polystyrene half of the particle. Weestimate the conductivity of this layer to be 0.2 S m-1 using theapproximate surface conductance formulas given by Morganand Green.25 The total surface conductance is defined as

KS )KS,s +KS,d (6)

where KS,d is the conduction of the diffuse part of the doublelayer and KS,s is the conduction of the Stern layer.25 KS,s iscalculated as

KS,s )∑j

σqs,jµj (7)

where σqs,j is the surface charge density in the Stern layer ofthe double layer and µj is the ion mobility.25 For the polystyreneparticles used, the surface charge density (σqs,j) on the particlewas 0.061 C m-2 (data from the vendor), and we used H+ andOH- ion mobilities from Morgan and Green, with σqs,H+ )2.5 × 10-7 m2(V sec)-1 and σqs,OH- ) 3.63 × 10-7 m2(Vsec)-1.25 The ion mobility would be lower in the Stern layer,

(57) Griffiths, D. J. Introduction to Electrodynamics; Pearson Education:London, 1999.

(58) Halliday, D.; Resnick, R.; Walker, J. Fundamentals of Physics Extended;John Wiley & Sons, Inc.: New York, 1997.

(59) Israelachvili, J. Intermolecular and Surface Forces; Elsevier AcademicPress: San Diego, 1992.

Assembly of Metallodielectric Janus Particles Langmuir, Vol. 24, No. 23, 2008 13319

but we use the bulk ion mobility values as an approximation.Since the double layer will consist primarily of a monovalention, we calculated an average value for KS,s of 1.73 × 10-8 Sbased on averaging the value for H+ and OH- ions. We useda value of 10-9 S for KS,d as the conduction in the diffuselayer obtained from Morgan and Green.25 The total surfaceconductance KS was calculated to be 1.83 × 10-8 S, whichwas divided by the counterionic atmosphere thickness (100nm) to yield a value of 0.2 S m-1 for the conductivity of thiscounterionic layer. This surface conductivity value was

consistent with data in the literature for very low electrolyteconcentrations.25,61

Supporting Information Available: Movies (in AVI format)illustrating the phase diagram transition from staggered chains formedat high frequency AC field to low frequency ICEP motion of the Janusparticles in directions perpendicular to the applied electric field. Thestaggered chains disassemble, and the particles begin ICEP motion oncethe frequency is reduced from 25 to 1 kHz and the electric field intensityis raised from 60 to 150 V cm-1. This material is available free of chargevia the Internet at http://pubs.acs.org.

LA8015222(60) Brydson, J. A. Plastic Materials, 7th ed.; Butterworth-Heinemann: Oxford,

1999; p 435. (61) Lyklema, J.; Minor, M. Colloids Surf., A 1998, 140, 33–41.

13320 Langmuir, Vol. 24, No. 23, 2008 Gangwal et al.