nanoparticle printing with single-particle...

© 2007 Nature Publishing Group

Nanoparticle printing with single-particleresolution

TOBIAS KRAUS1,2, LAURENT MALAQUIN1†, HEINZ SCHMID1, WALTER RIESS1,NICHOLAS D. SPENCER2 AND HEIKO WOLF1*1IBM Research GmbH, Zurich Research Laboratory, Saumerstrasse 4, 8803 Ruschlikon, Switzerland2Laboratory for Surface Science and Technology, Department of Materials, ETH Zurich, Wolfgang-Pauli-Strasse 10, CH-8093 Zurich, Switzerland†Present address: Laboratoire de Photonique et de Nanostructures, Route de Nozay, 91460 Marcoussis, France

*e-mail: [email protected]

Published online: 2 September 2007; doi:10.1038/nnano.2007.262

Bulk syntheses of colloids efficiently produce nanoparticles with unique and useful properties. Their integration onto surfaces is aprerequisite for exploiting these properties in practice. Ideally, the integration would be compatible with a variety of surfaces andparticles, while also enabling the fabrication of large areas and arbitrarily high-accuracy patterns. Whereas printing routinelymeets these demands at larger length scales, we have developed a novel printing process that enables positioning of sub-100-nmparticles individually with high placement accuracy. A colloidal suspension is inked directly onto printing plates, whose wettingproperties and geometry ensure that the nanoparticles only fill predefined topographical features. The dry particle assembly issubsequently printed from the plate onto plain substrates through tailored adhesion. We demonstrate that the process can createa variety of particle arrangements including lines, arrays and bitmaps, while preserving the catalytic and optical activity of theindividual nanoparticles.

Today, printing is the most widespread technology to depositsmall particles onto various surfaces. In book and fine-artprinting, submicrometre pigment particles carry differentcolours, thus creating brilliant and bleach-resistant images.Small particles (often ,100 nm in diameter) are increasinglyused in other fields, too. In electronics, optics and biology,small particles are the target of intense research and are usedin a number of commercial applications1. Such applicationsexploit the particle’s confined electronic systems2, their stronginteraction with light, with their matrix and with each other3,their well-defined surface properties4, their high catalyticactivity5 and—in sufficiently small particles—their quantumconfinement properties6.

A prerequisite for future applications using particles asfunctional entities is often control of their arrangement on asurface, between electrodes or in a device. Doing so withstandard microfabrication techniques is difficult, and it is oftentime-consuming and inefficient to create sparse patterns of smallparticles using subtractive top-down processing7. Printingtechniques, in contrast, scale well to nanoscopic particle sizes.Figure 1 shows how gravure printing can be scaled to handleparticles less than 100 nm in diameter. Using self-assemblyprocesses to ink nanofabricated ‘printing plates’ and controlledadhesion for the transfer, we print nanocrystals at single-particleresolution. The process maintains the efficient parallel nature ofself-assembly without requiring specially patterned substrates orelaborate surface treatments. Printing has been demonstrated onvarious targets, including soft and flexible materials that arechallenging to handle by other microfabrication processes8.

In conventional printing, the ink is treated as a continuousmedium. Its wetting properties (as in offset printing) or the

geometry of a template (as in gravure printing) define in whichareas a printing plate is inked and from where the ink will later betransferred to the substrate. The arrangement of the final,embedded pigment particles in the print is stochastic. This principleholds even for molecular inks—many researchers have used micro-contact printing to transfer reactive molecules9, proteins10 andcatalytic colloids11 according to stamp geometries. The arrangementof the printed species itself is not imposed by a template, but iseither irregular or imposed by the adsorption process12.

In the process proposed here, discrete nanoparticles arearranged deterministically on structured surfaces using ‘directedassembly’. In contrast to conventional inking, directed assemblydoes not merely fill predefined structures with randomlydispersed pigment particles, but arranges nanoparticles atpositions that are defined by the geometry of a template. Thesesilicone templates are moulded with total areas in the order of1 cm2 from hard masters with feature dimensions between 40 nmand tens of micrometres. The template geometries adopted(Fig. 2) include lines, producing close-packed nanoparticle wiresas used in molecular electronics13, spaced regular arrays fornanowire growth and arbitrary arrangements, such as the sundesigned by the seventeenth-century alchemist Robert Fludd14,which is composed of about 20,000 single particles.

HIGH-PRECISION INKING

For inking, the meniscus of a colloidal suspension is moved overthe patterned polymer layer so that the dispersed particlesarrange inside the features15. When the particles have assumedtheir desired positions, the liquid must be removed becausebrownian motion is sufficiently strong in nanoparticle suspensions

ARTICLES

nature nanotechnology | VOL 2 | SEPTEMBER 2007 | www.nature.com/naturenanotechnology570

mailto:[email protected]

http://www.nature.com/doifinder/10.1038/nnano.2007.262

www.nature.com/naturenanotechnology


to destroy the order. The dry, filled printing plates (Fig. 2) are thenstable and can be stored until the particles are transferredonto substrates.

In capillary assembly, as the colloid moves over the surface,particles are deposited in the desired regions. Excess particles areremoved through the Stokes drag exerted by the liquid meniscus(Fig. 3a), which fulfils a similar function as the doctor blade ingravure printing (Fig. 1). High yield, precise arrangement and highcontrast require control of particle transport, which is intimatelyconnected with the wetting behaviour of the colloidal suspension.

Providing sufficient particles from the bulk colloid to theassembly region near the contact line is a prerequisite for particledeposition. This mass transport can be achieved by theconvective flux of liquid towards the meniscus caused by waterevaporating from the meniscus16. For 60-nm gold particles, thetimescale of inertial response is very short—on the order ofnanoseconds—and the brownian timescale is very long comparedwith convective timescales, so that in the bulk part of the fluid,particles can be regarded as solutes that travel alongthe streamlines (see Supplementary Information, page 2). Thestreamlines depend on experimental parameters, notably thetemperature, the humidity and the substrate velocity, and can besimulated. Figure 3b shows computational fluid dynamicssimulations, taking into account the substrate motion, whichdrags liquid towards the meniscus, and the evaporation from themeniscus, which removes liquid that has to be replenished fromthe bulk.

Two main features appear in the laminar flow of the colloidaccording to our simulations (see Supplementary Information,page 5): a flow component that is directed towards the meniscusand a recirculation flow in the upper part of the liquid. As theevaporation rate increases, the recirculation zone moves awayfrom the meniscus. Higher colloid temperature and lower outside

humidity increase particle flux to the assembly region both byincreasing flow rates from the bulk and by decreasing the particlerecirculation. For the liquid flow rate Vcolloid that occurs at acolloid temperature of 27 8C and a bulk particle concentration ofcp ¼ 2 � 1014 l21, a flux of Np ¼ cpVcolloid � 1.5 � 108 m21 s21 istransported to the assembly region. This particle flux is larger thanthe flux required to form a full particle monolayer,

_NML ¼ vsubstr

p=ð3ffiffiffiffiffi2Þ

p

r2pp

� 1:96� 107m�1 s�1

at a substrate velocity of vsubstr ¼ 0.3 mm s21. Thus, sufficientparticles to fill any feature on the printing plate are supplied.However, to ink with high yield and good contrast, additionalprerequisites need to be satisfied.

In addition to the particle flux, the particle concentration in theassembly zone has to be sufficient. We can analyse the requirementsfor high-yield assembly using a microscopic model of particletransport. Whereas gaussian statistics describe the transport ofparticles to the accumulation zone, Poisson statistics apply whensmall numbers of particles need to move into specific positions.The immobilization of a single particle in a capture site of thetemplate can be modelled by assuming that any particle from acertain volume Vc above this site will be captured (Fig. 3c). If noadditional force exists, the average number knl of particles in Vc

depends solely on the particle concentration, knl ¼ cpVc. Thus,the probability of capturing at least one particle becomes

Pðn � 1Þ �X1

n¼1

PðnÞ ¼ 1� e�knl

>10 µm typical

60 nm typical

Figure 1 From traditional gravure printing to high-resolution particle printing. a, In gravure printing, a doctor blade fills the recessed features of a printing plate

with ink. Pigment particles are randomly dispersed in the ink, which is transferred from the plate onto the substrate. b, In high-resolution particle printing, a self-

assembly process controls the arrangement of nanoparticles on the printing plate. The entire assembly is printed onto the substrate, whereby the particle positions

are precisely retained at a resolution that is three orders of magnitude higher than in conventional printing.

ARTICLES

nature nanotechnology | VOL 2 | SEPTEMBER 2007 | www.nature.com/naturenanotechnology 571



so that the minimum concentration required to achieve a 90% yieldacross many capture sites is cp � 2.3/Vc, which is large whencompared with the original colloidal concentration. We thereforehave to induce local particle accumulation without destabilizingthe colloid to provide high assembly yield.

When contact angle and particle stabilization are adjustedproperly, the excess particles that flow into the assembly regionform a stable zone of high concentration close to the contact line(Fig. 3c). This ‘accumulation zone’ is self-limiting; it does notgrow in size even if the meniscus is moved over an unstructuredsubstrate, where no particles are deposited. A steady state betweenleaving and entering particles results when the accumulation zonehas reached a certain size. The zone is reminiscent of the well-known concentration polarization in ultrafiltration17, where theStokes drag on the particles is balanced by interparticle forces(mainly hydrodynamic forces, electrostatic and entropic repulsion).This dynamic diffusion–convection equilibrium is tunablethrough temperature, which strongly affects the evaporation rateand mainly changes the Stokes drag on the particles, and throughcolloid composition, which changes the particle interaction forces.

More accurate descriptions of the assembly mechanism need totake into account the dewetting dynamics, particle–surfaceinteraction and particle–particle forces. It was observed that the

template geometry not only defines the particle arrangement, butalso governs the mechanism leading to this arrangement. Largerfeatures, for example the lines shown in Fig. 3d, create microscaleflows. Particles can flow into the features and the actualarrangement often takes place at some distance from the mainassembly front. In contrast, arrays that capture single particlespin the meniscus only very slightly, and the assembly takes placevery close to the contact line. In all cases, the wetting behaviouris critical—the transition from the mobile to the assembled statetakes place only upon dewetting.

Both the microscopic dewetting and the bulk liquid motiondepend on the contact angle; inking at high yield and highprecision requires its tuning. It is possible to modify the templatesurface to this end15. However, in a printing process, higher-energy surfaces are not an option because particles wouldstrongly adhere to them, inhibiting the transfer step. Instead, weuse surfactant systems to tune the wetting behaviour with moreflexibility, as is frequently done in conventional printing, and totune particle–surface interactions simultaneously. A surfactantsystem for particle inking has to change the contact angle of thecolloidal suspension with the hydrophobic poly(dimethylsiloxane)(PDMS) to an angle between 508 and 608, where we find thehighest assembly yield. The surfactants should not, however,

200 nm

110 nm

150 nm

173 nm

220 nm

232 nm

265 nm

140 nm

1 µm10 µm

0 700 nm0

20 nm

0 1,000 nm0

50 nm

Figure 2 Particles arranged on printing plates after inking. Gold particles were assembled in poly(dimethylsiloxane) templates by hydrodynamic and surface

forces. a, AFM topography and b, SEM micrographs of 60-nm Au monodispersed particles arranged in u-grooves of different widths, and one u-groove (bottom) filled

with polydispersed particles. c, Larger areas of multiple, densely spaced lines have also been filled, for example 150-nm lines with 60-nm Au particles. d, AFM

topography of 200-nm spaced array of 100-nm Au particles. e, 1-mm-spaced array of 60-nm Au particles. f, Bright-field optical micrograph of a sun composed of

approximately 20,000 individual 60-nm Au particles.

ARTICLES




destabilize the colloidal particles, not even in the accumulationzone with its high particle concentrations, and it should onlyleave a very thin film on the inked plate. Less obvious are theeffects of the surfactant on the particle–substrate interactions.We find that some surfactants hinder the assembly, possibly bycreating a repulsive force between substrate and particles, whereasothers (such as polyvinylpyrrolidone) form deposits that embedthe particles on the substrate. A mixture of a non-ionic, ratherhydrophobic surfactant—namely Triton X-45, an octylphenolethoxylate with a short polyethyleneglycol chain—and the anionicdodecyl sulphonate appears to avoid all these issues, leaving thecolloid stable even at high concentrations.

The surfactant system ensures a proper receding contact angle on anunstructured surface, but the local contact angle during assembly alsodepends on the pinning caused by the structures on the template’ssurface16. To achieve high yields without unspecific deposition, bothan appropriate geometry and a sufficient depth of the capture siteshave to be chosen. The results shown in Fig. 2 stem from arrays ofsingle particles assembled in 40- to 45-nm deep holes and lines ofparticles assembled in 70- to 100-nm deep lines. The deeper the linesare, the more robust the assembly process, at the cost of a lowertransfer yield. To produce sufficiently precise templates, masterspatterned through electron-beam or optical lithography and etchedinto silicon or silicon oxide were used. They can be used to castmany templates in a high-elastic-modulus polydimethylsiloxanerubber18,19. First, a prepolymer mixture is poured onto the siliconmaster, then a flexible glass backplane is placed on top, and thepolymer is cured. The backplane and polymer layer are peeled off,and the template is finally extracted with ethanol to remove residuesbefore the inking of the printing plate (see SupplementaryInformation for details of printing plate fabrication).

PARTICLE TRANSFER

Inking is followed by transfer in the printing process. Duringtransfer, the printing plate and substrate are brought intoconformal contact, facilitated by the elastomer layer that adjuststo the topography of the substrate surface. On removal of theprinting plate, adhesive forces hold the particles on the substrate,thereby creating the desired arrangement (Fig. 4). Traditionalprinting technologies use liquid inks, and ink transfer is based onwetting differences between the printing plate and the substrate(for example, paper). In this work, the ink solely consists of drynanoparticles, which are far less mobile than liquids. Particlestherefore can be positioned more precisely, but they are harder toprint. Their transfer is based on the different levels of particleadhesion on the printing plate and the substrate. Such adhesiondifferences have already been used to transfer larger particles20

and large molecules10.Wetting differences are caused by differences in interfacial energy.

Likewise, the adhesion of small particles strongly depends on theinterfacial energy of the particle–surface joint21. In contrast tosolid–liquid interfaces, however, the particle–substrate interface isoften not conformal, and its geometry can be complex22. Whensurfaces with different energies are used to transfer particles, it isalso necessary to control the interfacial area, in particular whenworking with individual nanocrystals20. Such crystals often haveirregular surfaces with sharp crystal edges. The adhesive forceacting on one particle with a planar geometry can be muchstronger than that acting on spherical ones. Moreover, solutesfrom the colloidal suspension frequently form adlayers on themetal surface while drying, thereby changing the chemical identityof the interface and increasing the particle–substrate distance. We

10 µm

T = 27 °C, RH = 50% T = 20 °C, RH = 95%

Cover plate

10 µm

Vc

Figure 3 The inking process that arranges the particles. a, The meniscus of a colloidal suspension containing 60-nm Au particles is moving over the patterned

PDMS surface of a printing plate. b, Simulation of the evaporation (colours indicate concentrations) and the streamlines (red lines) of the laminar flow that drags

along particles towards the three-phase contact line, where they form an accumulation zone. The main panel is a simulation at a relative humidity (RH) of 50% and

the inset for a RH of 95%. c,d, The accumulation zone, visible in the micrographs as a bright yellow line, moves over the printing plate. Depending on the geometry

they encounter, particles at the contact line form sparse patterns, for example, arrays (c) or dense arrangements such as lines (d).

ARTICLES




find that dry transfer of isolated spherical gold nanoparticles with60 nm diameters leads to very low yields on silicon, quartz, andeven fresh gold layers with strong van der Waals attraction to goldparticles. The transfer of cubes and plates leads to much largeryields. Likewise, lines and layers of 60-nm gold spheres could beprinted with good yield onto a hard silicon surface with a cleannative oxide layer (Fig. 4). It appears that both a large crystal facein contact with the hard substrate and multiple smaller contactpoints increase adhesion and transfer yield, whereas organicadlayers on the particles decrease adhesion and yield. This isconsistent with standard microscale models of adhesion, whichstress the importance of the actual interfacial area and of contactsplitting for adhesive strength23. Inside the lines, particles stronglyadhere to each other, and each particle adheres to the substrate.Thus, to break the adhesive bond to the substrate, multipleparticle–substrate contacts have to be broken, which requires alarger force than for each single adhesive contact24.

A reliable transfer process thus requires both surface-energydifferences and defined interfacial areas. The PDMS layer on theprinting plate that we use provides a low-energy surface, wellknown for its capacity to release biomolecules onto morehydrophilic substrates10. The geometry of the PDMS layer wasdesigned such that the particles come into contact with thesubstrate when the printing plate is placed on the target. Thissometimes conflicts with the inking process, which for someparticle arrangements (for example, for lines) requires that thetemplate be deeper than the particle’s diameter. In such cases, the

gap was kept as small as possible. A height difference of 5–10 nmcan be overcome when pressure is applied on the stamp duringthe transfer process. Figure 4a shows particle lines of differentwidths printed on oxidized silicon using this geometry.

In contrast to lines, arrays of individual spherical nanocrystalscannot be transferred onto hard substrates, even if they protrudesignificantly from the template (Fig. 2e,f). However, transferwith high yields is possible onto thin polymer layers, forexample, spin-coated resists. A polymethylmethacrylate (PMMA)layer with a thickness of 10–30 nm increases the transfer yield toabove 95%. Transfer takes place at a temperature of 110–120 8C,slightly above the glass transition temperature of PMMA.The polymer embeds the protruding segment of the crystal,thus providing a sufficiently large area of contact to createthe adhesion necessary for transfer. Figure 4 shows arrays ofindividual nanocrystals printed on PMMA. If necessary, thePMMA can then be removed in hydrogen plasma, where itcleanly decomposes.

The position of the nanocrystals is preserved during the transferwith a local accuracy better than 100 nm (see also Fig. 5b). Thelong-range accuracy of the transfer is comparable to that ofmicrocontact printing with thin-film stamps, for which relativeaverage positioning errors below 20 p.p.m. have beendemonstrated across 4-inch wafers9. Our printing plates featurethe same combination of hard backplanes with soft elastomerlayers to reduce the deformations of the stamp that cause long-range inaccuracies in soft lithography techniques18.

67 nm

120 nm

124 nm

148 nm

208 nm

255 nm

1 µm1 µm

0 650 nm0

60 nm

0 330 nm0

60 nm

1 µm

Figure 4 Particle structures printed on unpatterned Si substrates. Lines are directly printed onto the native oxide layer of Si wafers, and single-particle arrays are

printed onto additional 30-nm PMMA adhesion layers. a, AFM and b, SEM images of lines from 60-nm Au particles. The bottom row in b is made from 100-nm Ag

particles, which are mostly cubical in shape. c, Larger area of 200-nm-wide lines from 60-nm Au particles. d, AFM detail and e, SEM overview of 1-mm-spaced

array of 60-nm Au particles. f, Detail (left eye) from a printed sun pattern composed of 60-nm-diameter Au particles with 280 nm pitch.

ARTICLES




FUNCTIONAL ARRAYS

It is crucial that printed nanoparticles retain their useful propertiesduring integration. Here, a surface-sensitive application was chosento demonstrate that the individual gold nanocrystals are stillcatalytically active. Figure 5a shows silicon nanowires (SiNW) thatwere grown from an array of 60-nm particles using a vapour–liquid–solid (VLS) process25. The particles were printed onto a

30-nm-thick PMMA layer on top of a hydrogen-passivated siliconk111l substrate. The PMMA adhesion layer was removed bythermal decomposition at 300 8C in vacuum, assisted byhydrogen plasma. SiNW growth was then initiated at 460 8C in areactor using silane as the precursor gas. Nanowires nucleatedand grew from the gold particles, and their arrangement waspreserved. The epitaxial relationship of the SiNW with thesubstrate is evident from the wires growing along both thevertical and the three inclined k111l growth directions. Suchnanowires are regarded as novel building blocks for futuretransistors, memory cells and sensors, for example26.

Metal nanocrystals also interact strongly with electromagneticwaves27, a property that is exploited in applications such as glassstaining, surface-enhanced Raman spectroscopy28 (SERS) andsurface plasmon resonance-based agglomeration assays29. It hasbeen noticed that the characteristics of single particles can varyconsiderably30, and the printing technique adopted here can createsamples that allow investigating such single-particle effects in detail.When the particles are arranged in regular arrays, it is easy to findextraordinarily active particles and analyse them using a variety ofmethods. Figure 5b shows how some particles scatter visible lightmore strongly than others, both when immobilized in the PDMStemplate and when printed onto a silicon substrate. Comparison ofthe dark-field micrographs with the atomic force microscopy(AFM) and scanning electron micrograph (SEM) images revealsthat this is not simply a function of size or overall shape (althoughdoublets of particles often scatter the light strongly), and it will beinteresting to apply single-particle SERS and high-resolutionelectron microscopy to identify the features that distinguishstrongly scattering particles or SERS ‘hot spots’.

CONCLUSIONS AND OUTLOOK

In conclusion, we have developed an efficient process to fabricatelines, arrays and complex arrangements of nanoparticles with highaccuracy and single-particle resolution that retains individualparticle functionality. Our printing approach is compatible withdifferent particles, including not only metals, but also polymers,semiconductors and oxides20. It can handle bulk-synthesizedparticles directly in their original colloidal state even when theyhave been modified with functional molecules. Thanks to its highresolution, nanoparticle printing can efficiently define the criticaldimensions for nanoscale devices. The long-range accuracy issimilar to that of microcontact printing methods and reachesabout 12 p.p.m. Reaching higher accuracies of 1 p.p.m., as requiredfor large-scale integration in microelectronics, is a challenge thatremains to be addressed.

Printing processes are adaptable to continuous processing andlarge areas31. Particle handling is based on adhesion, which scalesfavourably with particle size, so that the printing step can handlemuch smaller particles, and both self-assembly and directedassembly techniques have been reported to arrange particlesdown to 2 nm in diameter15. Many such techniques, includingdry processes based on surface charging32 and processes based onbiological systems33 could be combined with nanoparticleprinting. This should enable not only the integration of muchsmaller objects but also the development of printing plates thatcan be programmed for any desired particle arrangement withoutrequiring new templates.

METHODS

PARTICLE AND TEMPLATE PREPARATION

Metal nanocrystals were either purchased (gold colloid, 60 nm nominaldiameter, British Biocell International) or synthesized through reduction from

1 µm

1 µm

Figure 5 Demonstration of the activity of printed particles. In printed arrays,

60-nm Au particles retain both their catalytic and optical activities. a, Silicon

nanowires grown through a vapour– liquid–solid process from a printed array of

Au particles (inset is tilted). b, AFM topography and the corresponding dark-field

image of particles on the printing plate (top row) and an SEM image and the

corresponding dark-field micrograph on a silicon substrate (bottom row: the

images were mirrored for convenience).

ARTICLES




their respective salts. The polydispersed gold particles shown in Fig. 2 wereobtained following a method from Brown et al.34 in three steps, starting withsmall crystals produced according to Frens’ method35, which were used to seedthe next stages. The silver nanocubes shown in Fig. 4 were obtained throughpolyol reduction of silver nitrate in a process adapted from Xia’s group36.Experimental details can be found in the Supplementary Information. In brief,silver nitrate and polyvinylpyrrolidone were injected into ethylene glycol at140 8C under acidic conditions, and 100-nm cubes were obtained after 17 hof growth. Those were then transferred into water and used for inking.

Templates for assembly were cast in PDMS using nanostructured siliconmasters. See Supplementary Information, for details of the electron-beamwriting and the processing of these masters. The masters were used to cast largenumbers of silicone templates. The PDMS used is a variation of the commerciallyavailable product. It has a higher elastic modulus and can be patterned at higherfidelity. All templates were fabricated on backplanes cut from display glass(175 mm, Schott AG).

HIGH-PRECISION INKING

Of the many surfactants we tested, the best results were obtained with thefollowing mixture, which was therefore adopted for all printing experiments. Anaqueous 0.1 wt% emulsion of Triton X-45 (Fluka) was mixed with a 10 mMaqueous solution of sodiumdodecylsulphonate (Fluka), mixed and filteredthrough a PTFE 200-nm-pore syringe filter (Rotilabo). Then 50 ml of themixture were added to 200 ml of the colloid, yielding a final particleconcentration of 2 � 1014 l21.

The inking was performed in a homemade tool (the capillarity-assistedparticle-assembly or ‘CAPA’ tool) described previously16, which enables controlof the temperature and substrate velocity while the template is being movedunder a drop of the colloidal suspension. The substrate temperature was chosensuch that a clearly visible accumulation zone formed, for example, 27 8C at theusual environmental humidity of 50%. A microscope camera recorded thecontact angle of the meniscus, which was used to ensure good quality of thesubstrate and the colloid, indicated by a contact angle of 50–608. During theinking, a microscope mounted on top of the setup was used to monitor andrecord the process.

PARTICLE TRANSFER

Particle transfer was carried out on both bare silicon surfaces (lightly doped,p-type, polished silicon wafers, Siltronic) and on PMMA-coated silicon surfaces(spin-coated with a 10–30 nm PMMA layer, having a molecular weight of950 kDa, Microchem). In both cases, the printing plates with the particles werebrought into contact with the substrates either manually or in a printing tool thatallows microscopic inspection and alignment with existing structures as well asforce and temperature control during the transfer. In the case of the uncoatedsubstrates, the printing plate was pressed firmly onto the target and removedimmediately after. In the case of the substrates with PMMA adhesion layers, theprinting plate was placed on the target, the temperature was increased to 120 8C,pressure was applied on the backplane, the temperature was decreased, and theprinting plate was removed as the temperature had dropped to 40 8C.

BULK COLLOID-FLOW SIMULATION

The laminar flows occurring in the inking process were simulated with astandard finite element modelling discretization of the Navier–Stokes equationsusing a commercial code (COMSOL, FEMLAB GmbH), in which theevaporation was calculated simultaneously to produce boundary conditions forthe flow into the meniscus. (See Supplementary Information for details andresults for a range of simulation parameters).

Received 24 May 2007; accepted 25 July 2007; published 2 September 2007.

References1. Fendler, J. H. & Meldrum, F. C. The colloid chemical approach to nanostructured materials. Adv.

Mater. 7, 607–632 (1995).2. Klein, D. L. et al. An approach to electrical studies of single nanocrystals. Appl. Phys. Lett. 68,

2574–2576 (1996).

3. Maier, S. A. et al. Local detection of electromagnetic energy transport below the diffraction limit inmetal nanoparticle plasmon waveguides. Nature Mater. 2, 229–232 (2003).

4. Englebienne, P. Use of colloidal gold surface plasmon resonance peak shift to infer affinity constantsfrom the interactions between protein antigens and antibodies specific for single or multipleepitopes. Analyst 123, 1599–1603 (1998).

5. Bell, A. T. The impact of nanoscience on heterogeneous catalysis. Science 299, 1688–1691 (2003).6. Lee, T. H. & Dickson, R. M. Discrete two-terminal single nanocluster quantum optoelectronic logic

operations at room temperature. Proc. Natl Acad. Sci. USA 100, 3043–3046 (2003).7. Parviz, B. A., Ryan, D. & Whitesides, G. M. Using self-assembly for the fabrication of nano-scale

electronic and photonic devices. IEEE Trans. Adv. Packag. 26, 233–241 (2003).8. Ahn, J. H. et al. Heterogeneous three-dimensional electronics by use of printed semiconductor

nanomaterials. Science 314, 1754–1757 (2006).9. Michel, B. et al. Printing meets lithography: Soft approaches to high-resolution printing. IBM J. Res.

Dev. 45, 697–719 (2001).10. Bernard, A. et al. Microcontact printing of proteins. Adv. Mater. 12, 1067–1070 (2000).11. Hidber, P. C. et al. Microcontact printing of palladium colloids: Micron-scale patterning by

electroless deposition of copper. Langmuir 12, 1375–1380 (1996).12. Larsen, N. B. et al. Order in microcontact printed self-assembled monolayers. J. Am. Chem. Soc. 119,

3017–3026 (1997).13. Liao, J. et al. Reversible formation of molecular junctions in 2D nanoparticle arrays. Adv. Mater. 18,

2444–2447 (2006).14. Fludd, R. & de Bry, J. T. Utriusque cosmi maioris scilicet et minoris metaphysica, physica atque

technica historia, in duo volumina secundum cosmi differentiam diuisa. (Aere J. T. de Bry, typisH. Galleri, Oppenhemii, 1617).

15. Cui, Y. et al. Integration of colloidal nanocrystals into lithographically patterned devices. Nano Lett.4, 1093–1098 (2004).

16. Malaquin, L. et al. Controlled particle placement through convective and capillary assembly.Langmuir (in the press).

17. Peppin, S. S. L. & Elliott, J. A. W. Non-equilibrium thermodynamics of concentration polarization.Adv. Colloid Interface Sci. 92, 1–72 (2001).

18. Schmid, H. & Michel, B. Siloxane polymers for high-resolution, high-accuracy soft lithography.Macromolecules 33, 3042–3049 (2000).

19. Geissler, M. et al. Fabrication of metal nanowires using microcontact printing. Langmuir 19,6301–6311 (2003).

20. Kraus, T. et al. Closing the gap between self-assembly and microsystems using self-assembly, transfer,and integration of particles. Adv. Mater. 17, 2438–2442 (2005).

21. Rimai, D. S., Ezenyilimba, M. C. & Quesnel, D. J. Effects of electrostatic and van der Waalsinteractions on the adhesion of spherical 7 mm particles. J. Adhesion 81, 245–269 (2005).

22. Farshchi-Tabrizi, M. et al. On the adhesion between fine particles and nanocontacts: An atomic forcemicroscope study. Langmuir 22, 2171–2184 (2006).

23. Kendall, K. Molecular Adhesion and its Applications: The Sticky Universe vol. xix (KluwerAcademic/Plenum Publishers, New York, 2001).

24. Arzt, E., Gorb, S. & Spolenak, R. From micro to nano contacts in biological attachment devices. Proc.Natl Acad. Soc. USA 100, 10603–10606 (2003).

25. Wagner, R. S. & Ellis, W. C. Vapor-liquid-solid mechanism of single crystal growth. Appl. Phys. Lett.4, 89–90 (1964).

26. Li, Y. et al. Nanowire electronic and optoelectronic devices. Mater. Today 9, 18–27 (2006).27. Hutter, E. & Fendler, J. H. Exploitation of localized surface plasmon resonance. Adv. Mater. 16,

1685–1706 (2004).28. Freeman, R. G. et al. Self-assembled metal colloid monolayers—an approach to sers substrates.

Science 267, 1629–1632 (1995).29. Storhoff, J. J. et al. What controls the optical properties of DNA-linked gold nanoparticle assemblies?

J. Am. Chem. Soc. 122, 4640–4650 (2000).30. Krug, J. T. et al. Efficient Raman enhancement and intermittent light emission observed in single

gold nanocrystals. J. Am. Chem. Soc. 121, 9208–9214 (1999).31. Kipphan, H. Handbook of Print Media (Springer, Berlin, 2004).32. Kim, H. et al. Parallel patterning of nanoparticles via electrodynamic focusing of charged aerosols.

Nature Nanotech. 1, 117–121 (2006).33. Sleytr, U. B. et al. Crystalline bacterial cell surface layers (S layers): From supramolecular cell structure

to biomimetics and nanotechnology. Angew. Chem. Int. Edn 38, 1035–1054 (1999).34. Brown, K. R., Walter, D. G. & Natan, M. J. Seeding of colloidal Au nanoparticle solutions. 2.

Improved control of particle size and shape. Chem. Mater. 12, 306–313 (2000).35. Frens, G. Controlled nucleation for regulation of particle-size in monodisperse gold suspensions.

Nature Phys. Sci. 241, 20–22 (1973).36. Im, S. H. et al. Large-scale synthesis of silver nanocubes: The role of HCl in promoting cube

perfection and monodispersity. Angew. Chem. Int. Edn 44, 2154–2157 (2005).

AcknowledgementsWe thank U. Drechsler for her support in microfabrication as well as R. Stutz and M. Tschudy for theirtechnical support. A part of this project was funded by the Swiss Commission for Technology andInnovation. The partial support of the State Secretariat for Education and Research (SER) in theframework of the EC-funded project NaPa (Contract No. NMP4-CT-2003-500120) is gratefullyacknowledged. The content of this work is the sole responsibility of the authors. We thank P. Seidler for hiscontinuous support.Correspondence and requests for material should be addressed to H.W.Supplementary information accompanies this paper on www.nature.com/naturenanotechnology.

Competing financial interestsThe authors declare no competing financial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

ARTICLES


http://npg.nature.com/reprintsandpermissions/


nanoparticle printing with single-particle...

Documents