DOI: 10.1021/la902856x 2401Langmuir 2010, 26(4), 2401–2405 Published on Web 09/10/2009

pubs.acs.org/Langmuir

© 2009 American Chemical Society

Matching Constituent Fluxes for Convective Deposition of BinarySuspensions

Pisist Kumnorkaew, Alexander L. Weldon, and James F. Gilchrist*

Center for Advanced Materials and Nanotechnology, Department of Chemical Engineering, Lehigh University,Bethlehem, Pennsylvania 18015

Received August 3, 2009. Revised Manuscript Received August 21, 2009

Rapid convective deposition is an effective method for depositing well-ordered monolayers from monodispersesuspensions; however, much less is known about polydisperse suspension deposition. The addition of a much smallerspecies can enhance deposition by extending the range of ordered deposition and can induce instability, producingstripes and other complex morphologies. By considering relative species flux, we predict the volume fraction ratio ofsmaller to larger constituents necessary for steady well-ordered deposition. Experiments varying the 1 μmmicrosphereand 100 nm nanoparticle concentrations exhibit an optimum nanoparticle to microsphere volume fraction ratio atmoderate volume fractions that agrees well with theory. Average local bond order and surface density characterizecrystallinity and coverage, respectively. At lower microsphere volume fraction, monolayer crystallinity is optimized at aconstant nanoparticle volume fraction of 0.04. At lower-than-optimum nanoparticle concentrations for each micro-sphere concentration, instability occurs and alternating stripes of monolayer and submonolayer morphologies form. Athigher-than-optimum nanoparticle concentration, themicrospheres become disordered and/or formmultilayer regions.Additionally, the degree of microsphere burial in deposited nanoparticles depends solely on nanoparticle concentration.

Introduction

Tuned polydispersity in colloidal suspensions is increasinglyused in applications to control filmmorphology during drying,1,2

fabricate crystals of various periodicities,3 and study the effects ofdiffering species on suspension rheology.4,5 Polydispersity refersnot only to the size distribution in a given suspension but also tothe variability in particle shape, density, internal morphology,and surface chemistry.6 Suspension polydispersity commonlystems from the variability in nucleation and growth rates andfrom particles formed via aggregation. Heteroaggregation anddepletion flocculation are classic examples of tuned polydisper-sity.4,7 Novel combinations of particles of both tuned size andsurface polydispersity have been used to form aggregates withspecified properties.8 Long-studied particle-sorting processes in-cluding fractionation and other phoretic analyses focus on redu-cing polydispersity.

Convective deposition is a method of growing popularity inaltering surface morphology and chemistry.9 Convective deposi-tion is the process by which a suspension meniscus is drawnacross a substrate with well-ordered particle layers depositedas the contact line advances. The fundamental science continuesto elucidate novel aspects of convective deposition related to

suspension properties and fluidmechanics;10-14 however, the roleof polydispersity on convective deposition is still unclear.15Binarysuspensions of differently sized particles have found use in thefabrication of 2D and 3D colloidal crystals where smaller con-stituents fill the interstitial regions between larger species.14,15

Deposited binary suspension layers are very useful for alteringsurfaces’ optical properties.16,17 Binary suspensions with poly-dispersity in properties other than size have been minimallyexplored, though most combinations of materials will probablyresult in highly heterogeneous deposited layers. The exception tothis is the higher degree of uniformity in polymer andmicrospherebinary suspensions where the elastic stresses aid assembly.13

We extend previous work on nanoparticle and microspherebinary depositions15 by considering relative fluxes of individualconstituents during deposition. Our prior work showed thatnanoparticle addition of optimum volume fraction φnano to asuspension held at constant microsphere volume fraction φmicro

allows the convective deposition of a homogeneous monolayerthat is much larger than that deposited with a unary microspheresuspension. At lower-than-optimal φnano, stripes of alternatingmorphologies form perpendicular to the direction of the advan-cing meniscus, and at higher-than-optimal φnano, surface crystal-linity degrades or forms other morphologies such as multilayersand stripes. In deposited layers, microspheres are partially buriedwithin a nanoparticle multilayer, and the degree of exposedmicrospheres depends on φnano. At high φnano, the depletion

*Corresponding author. E-mail: gilchrist@lehigh.edu.(1) Harris, D. J.; Hu, H.; Conrad, J. C.; Lewis, J. A. Phys. Rev. Lett. 2007, 98,

148301–4.(2) Harris, D. J.; Lewis, J. A. Langmuir 2008, 24, 3681–3685.(3) Shevchenko, E. V.; Talapin, D. V.; Kotov, N. A.; O’Brien, S.; Murray, C. B.

Nature 2006, 439, 55–59.(4) Gilchrist, J. F.; Chan, A. T.; Weeks, E. R.; Lewis, J. A. Langmuir 2005, 21,

11040–11047.(5) Mohraz, A.; Weeks, E. R.; Lewis, J. A. Phys. Rev. E: Stat., Nonlinear, Soft

Matter Phys. 2008, 77, 060403–4.(6) Glotzer, S. C.; Solomon, M. J. Nat. Mater. 2007, 6, 557–562.(7) Russel, W. B.; Saville, D. A.; Schowalter, W. R. Colloidal Dispersion;

Cambridge University Press: Cambridge, U.K., 1989.(8) Snyder, C. E.; Yake, A. M.; Feick, J. D.; Velegol, D. Langmuir 2005, 21,

4813–4815.(9) Velev, O. D.; Gupta, S. Adv. Mater. 2009, 21, 1897–1905.

(10) Brewer, D. D.; Allen, J.; Miller, M. R.; de Santos, J. M.; Kumar, S.; Norris,D. J.; Tsapatsis, M.; Scriven, L. E. Langmuir 2008, 24, 13683–13693.

(11) Ghosh, M.; Fan, F.; Stebe, K. J. Langmuir 2007, 23, 2180–2183.(12) Kumnorkaew, P.; Ee, Y.-K.; Tansu, N.; Gilchrist, J. F. Langmuir 2008, 24,

12150–12157.(13) Kim, M. H.; Choi, H. K.; Park, O. O.; Im, S. H. Appl. Phys. Lett. 2006, 88,

143127–3.(14) Kitaev, V.; Ozin, G. A. Adv. Mater. 2003, 15, 75–78.(15) Kumnorkaew, P.; Gilchrist, J. F. Langmuir 2009, 25, 6070–6075.(16) Tessier, P.M.; Velev, O.D.; Kalambur, A. T.; Rabolt, J. F.; Lenhoff, A.M.;

Kaler, E. W. J. Am. Chem. Soc. 2000, 122, 9554–9555.(17) Prevo, B. G.; Hwang, Y.; Velev, O. D. Chem. Mater. 2005, 17, 3642–3651.

2402 DOI: 10.1021/la902856x Langmuir 2010, 26(4), 2401–2405

Article Kumnorkaew et al.

destabilizes microspheres, forming a gel that cannot be depositeduniformly. In this work, we propose a mechanism describingoptimal deposition concentration ratios and the formation ofstripes and other morphologies through unbalanced depositionfluxes. This allows for the tuning of suspension properties tomatch species fluxes and deposit very uniform monolayers usingbidisperse suspensions. We expand previous work to investigatethe nanoparticle influence across microsphere volume fractions0<φmicroe 0.24. In addition,we demonstrate howvaryingφmicro

under constant φnano influences the morphology but does notinfluence the degree of microsphere burial.

Materials and Methods

Suspension Preparation. The primary colloid suspensionused in this work is prepared by dispersing silica microspheres(Fuso Chemical Co, Japan) having a density of 2.2 g/cm3, anaverage diameter of 2amicro=1.01( 0.02μm,and a zeta potentialof-48mV( 1mV indeionized (DI) waterwith a volume fractionφmicro. The suspension is dispersed using a sonic dismembrator(model 550, Fisher Scientific, Pittsburgh, PA) for 10 min and isstirred for 30 min. (Fisher Scientific, model 550). A separatecolloidal suspension of diameter 2anano = 100 nm polystyrene(PS) having a zeta potential of-59mV( 1 mV prepared at φnano= 0.35 in DI water (supplied by the Emulsion Polymer Instituteat Lehigh University) is combined with the silica solution toachieve the desired suspension composition.

Substrate Preparation. Plain glass microslides (76 � 25 �1 mm3, Fisher PA) are used as deposition blades, and glasscoverslips (40� 24� 0.25mm3, Fisher PA) are used as substratesfor all samples. All glassware is cleaned by immersion inPiranha solution, 5:1 v/v sulfuric acid/hydrogen peroxide, for30 min. The cleaned glassware is rinsed with DI water until noresidual acid remains and is then immersed in DI water until use.The bottom edge of the glass deposition blade is made hydro-phobic with a thin coating of parafilm (Fisher PA). The contactangles on bare glass and on the hydrophobic surface aremeasuredto be 10 and 105�, repectively, by imaging a 10 μL stationarydroplet on the surface.

Deposition. The experimental setup was described pre-viously.12 All experiments were performed at roughly 50%relative humidity and 24 �C. The deposition blade angle is fixedat 45�, positioned approximately 10 μm above the substrate,and observed directly using a digital camera (Dinolite AM311S).The volume of colloid suspension for each experiment is10 μL.Microstructure Analysis. Deposited monolayers are ob-

served directly using scanning electron microscopy (SEM) andconfocal laser scanningmicroscopy.AHitachi 4300 field emissionSEM is used to observe the microstructure. Prior to SEMimaging, the sample is coated with iridium. Confocal laser scan-ning microscopy (VTeye, Visitech International) is used to ob-serve the microstructure after rewetting the layer with an aqueoussolution of 8 mM rhodamine B for imaging; this rewetting doesnot disturb the microstructure. The sample is scanned at 30 fpswhile the sample is translated across the microscope objective at100 μm/s using a motorized stage. This allows for the scanning oflarge regions of the sample (∼60000microspheres) to evaluate themicrostructure of those microspheres in contact with the sub-strate; specifically, relativemicrosphere substrate coverage, F, andlocal bond order, ψ6, were evaluated. The maximum theoreticalvalue of surface coverage for monosized microspheres depositedas a single crystal is F= π/(12)1/2 = 0.907. The long-range nano-particle substrate coverage is only qualitatively reported. ψ6

describes the relative orientation of particles in a plane around acentral particle. It is calculated by using all angles θ between eachparticle of interest i and its nearest neighbors j. Prior to thecomputation, vectors rij are determined for all nearest neighborsn. ψ6 = 1 is a perfect crystal, and ψ6 g 0.8 is considered to be

highly ordered for polycrystalline morphology and is reported asthe average over a large number of microspheres, N,

ψ6, ave ¼ 1

N

XN

k¼1

1

n

Xn

j¼1

exp½6iθðrijÞ� ð1Þ

Relative Fluxes

For a single-component suspension, Dimitrov and Nakaya-ma18 derived the relationship between volume fraction anddeposition speed for an advancing crystal on a substrate. For ahexagonally ordered monolayer

vmono ¼ Jeβ

2aðφDÞφ

ð1-φÞ ð2Þ

where vmono is the substrate velocity and is equal to the velocity ofthe advancing monolayer crystal front, Je is the solvent flux, 2a ismicrosphere diameter, φ and φ

D are the suspension volumefraction in solution and within the deposited thin film, respec-tively, andβdescribes particle-surface interactions.Assumptionsin implementing this equation include the fact that the bulksuspension volume fraction equals the particle volume fractionnear the advancing crystal front and that when particle-surfaceinteractions are strongly repulsive β≈ 1. Thismass balance resultsfrom calculating the amount of solvent required to remove fromthe bulk suspension to yield a thin liquid layer of particles ofheight equal to the particle diameter. The volume fraction of anorderedmonolayer immersed in liquid at the same height 2a is φD,which is the deposited microsphere volume fraction. A simplegeometric relationship gives φD

φD ¼ πR a

-aða2 -x2Þ dx2

ffiffiffi3

pa3

¼ π=ð3ffiffiffi3

pÞ ð3Þ

In binary suspensions where smaller constituents percolatethrough the interstitial spaces between larger particles, referredhenceforth as nanoparticles and microspheres respectively, a fewmodifications to this model are necessary. First, we assume thatdeposition dynamics of a microsphere monolayer primarily relateto microsphere properties (φmicro, amicro). However, when con-sidering the final microstructure of microsphere monolayersembedded within a nanoparticle layer, as shown in Figure 1,nanoparticle flux into the layer is as significant as microsphereflux. As the layer reaches maximum packing and jams near themicrosphere crystallization front, Je changes and the local freesurface curvature, solvent surface area, and effective suspensionviscosity become nontrivial. However, these physical changesaffect the deposition process comparatively little as comparedto the relative microsphere and nanoparticle fluxes into the layer.In binary depositions resulting in high-quality microspheremonolayers, nanoparticles almost entirely bury microspheres.Uniform deposition is possible only with balanced constituentfluxes. If nanoparticles exactly fill a height of 2a at the micro-spheres tops, then the relative volume fraction of each species,directly related to the ratio of constituent fluxes into the thinfilm, is

φDnano

φDmicro

¼ Jnano

Jmicro¼ ð1-π=ð3 ffiffiffi

3p ÞÞP

π=ð3 ffiffiffi3

p Þ ð4Þ

(18) Dimitrov, A. S.; Nagayama, K. Chem. Phys. Lett. 1995, 243, 462–468.

DOI: 10.1021/la902856x 2403Langmuir 2010, 26(4), 2401–2405

Kumnorkaew et al. Article

where φnanoD and φmicro

D are deposited nanoparticle and micro-sphere volume fractions respectively, Jnano and Jmicro are nano-particle and microsphere fluxes respectively, and P is the packingfraction of nanoparticles in the interstitial region between hex-agonally ordered microspheres. High nanoparticle confinementinhibits crystallization; for jammed nanoparticles19,20 with dimin-ishing nanoparticle size, anano/amicrof 0 andP≈ 0.64. For finite-sized nanoparticles, P may be significantly lower because ofconfinement.As seen inFigure 1, nanoparticles donot completelybury the microspheres, instead residing at a height h/a = 0.725above the microsphere equator. Hence, the ratio of depositednanoparticles to microspheres is

Jnano

Jnano¼ ð1-π

R 0:725a

-a ða2 -x2Þ dxÞPπR 0:725a

-a ða2 -x2Þ dx≈0:32 ð5Þ

The constituent flux ratio is exactly the volume fraction ratio ofspecies; consequently, when Jnano ≈ 0.32Jmicro, nanoparticle andmicrosphere fluxes into the thin film are matched appropriatelyfor binary depositions. Note that this result is largely independentof anano. The degree of burial depends weakly on φnano,

15 andacross observed degrees of burial this results in less than 5% errorin this calculation.

Ideal and nonideal matched Jnano and Jmicro combinations areexplored in Figure 2 in terms of their control over thin filmmorphology under otherwise ideal conditions.Under idealmono-layer-deposition conditions (Figure 2a), the fluxes are balancedthrough the relation proposed in eq 5, and the thin film is filledwith the appropriate number of particles for steady deposition.When nanoparticle flux is insufficient (Figure 2b), the movingfront of the nanoparticles lags behind that of the advancingmicrosphere crystal front. The result is instability where the liquidlayer pins on the deposited particles and interacts with binarymorphology or the microsphere monolayer. Thus, a stick-slipperiodic motion is possible, which results in stripes formingperpendicular to the deposition direction; these have featuressimilar to those seen with monodisperse suspensions.11,21-24

Packing of microspheres in the transition regions from single tomultilayer morphologies likely follows that seen in monodispersesuspension depositions.25,26 Any monodisperse suspensions withan additional nonvolatile species at high enough concentrationcanpotentially undergo the same transition.The distancebetween

stripes may correlate with the degree of mismatch between fluxes;this is seen qualitatively in experiments but is not quantified.Finally, when the nanoparticle flux exceeds that necessary to fillthemicrosphere interstitial region (Figure 2c), nanoparticles forcethis region to expand. This expansion comes about in two ways.Microspheres separate as nanoparticles inhibit crystallization anda submonolayer is formed or microspheres pack vertically totransition to multilayer morphology. If the nanoparticle concen-tration is insufficient to fill the multilayer voids, then layermorphology will oscillate between the monolayer and multilayerregions. Experiments within these three situations are summa-rized below.

Experimental Results

We begin by establishing baseline monolayer deposition speedsofmonodispersemicrosphere suspensions, 0.12e φmicroe 0.24, toquantify the relationship between vmono andφmicro (Figure 3). Thisquantification closely parallels the relationship given in eq 2. It isdifficult to produce large monolayer regions with φmicro e 0.12 inpart because of our experimental protocol of depositing a finitevolume of suspension; lower φmicro experiments result in deposi-tion areas too small for the determination of optimum conditions.However, previous studies successfully show the continuation ofthis trend at lower φmicro.

25 Optimum vmono values shown inFigure 3 are used for binary deposition with corresponding φmicro.This assumption has been validated through select binary deposi-tion trials with varying deposition speed.

The morphology of microsphere depositions can be furthertuned with the addition of moderate φnano. Optimized binary

Figure 1. (a) Nanostructure of nanoparticles surrounding micro-spheres in a well-ordered array. (b) Sketch of the local geometryand relative burial of microspheres. The nanoparticles do notcompletely bury the microspheres, covering only to h/a= 0.725.

Figure 2. Description of matched fluxes and resulting layer mor-phologies. (a) When the nanoparticles and microspheres havecomplementary concentrations, steady monolayer depositionoccurs. (b) For a lower-than-optimal nanoparticle concentration,instability arises and causes the advancing meniscus to jump.(c) For higher-than-optimal nanoparticle concentration, the inter-stitial region betweenmicrospheresmust increase by either spread-ing the microspheres or forming a multilayer. Instability betweenmonolayer and multilayer morphologies can also occur.

(19) Jaeger, H. M.; Nagel, S. R. Science 1992, 255, 1523–1531.(20) Song, C.; Wang, P.; Makse, H. A. Nature 2008, 453, 629–632.(21) Lee, J. A.; Meng, L.; Norris, D. J.; Scriven, L. E.; Tsapatsis, M. Langmuir

2006, 22, 5217–5219.(22) Snyder, M. A.; Lee, J. A.; Davis, T. M.; Scriven, L. E.; Tsapatsis, M.

Langmuir 2007, 23, 9924–9928.(23) Abkarian, M.; Nunes, J.; Stone, H. A. J. Am. Chem. Soc. 2004, 126, 5978–

5979.(24) Ray, M. A.; Kim, H.; Jia, L. Langmuir 2005, 21, 4786–4789.(25) Prevo, B. G.; Velev, O. D. Langmuir 2004, 20, 2099–2107.(26) Meng, L.;Wei, H.; Nagel, A.;Wiley, B. J.; Scriven, L. E.; Norris, D. J.Nano

Lett. 2006, 6, 2249–2253.

2404 DOI: 10.1021/la902856x Langmuir 2010, 26(4), 2401–2405

Article Kumnorkaew et al.

suspension depositions have larger uniform regions as comparedwith unary depositions as well as fewer defects and smaller edgemultilayer regions. As in prior studies,12,15 ψ6 and F quantify thequality of monolayer microstructure and coverage (Figure 4).Optimization ofφnano for φmicro=0.12, 0.16, 0.20, and 0.24 yieldsa maximum ψ6 and F for each φmicro. With increasing φmicro,optimum φnano increases. These data correlate with morphologi-cal observations; henceforth, the respective φnano for each φmicro

producing a maximum ψ6 or F will be referred to as optimumφnano, φnano* (φmicro). Experimentally obtained optimum concen-trations are found to be φnano* (0.12) = 0.04, φnano* (0.16) = 0.055,φnano* (0.20) = 0.065, and φnano* (0.24) = 0.07. For each φmicro,φnano < φnano* produces stripes of alternating monolayer andsubmonolayer regions whereas φnano > φnano* produces low-quality samples of either general disorder or alternating mono-layer and multilayer bands.

Optimalψ6 and F for each φnano* show a slight downward trendsuggesting that binary suspensions with lower φmicro produce

more uniform layers. This variation relates directly to the numberof crystalline defects and suggests that the crystal domain size islarger for lowerφmicro. Formonodisperse samples under optimumconditions with varying deposition speeds and blade angles, ψ6

and F were essentially constant.12 It is also clear that at higherφmicro, ψ6 and F are more sensitive to nanoparticle concentration.This sensitivity plays an important role in the choice of processingconditions. Also, the slopes of ψ6 and F verses φnano are steeperwith φnano < φnano* than with φnano > φnano* . This increasedsensitivity of ψ6 and F at φnano < φnano* is a result of insufficientnanoparticle flux (Figure 2b) as compared with that of micro-spheres. Sample patchiness with φnano < φnano* obviously yieldslower-than-optimum percent coverage and heterogeneity of sur-face morphology and, with a correspondingly higher number oflayer/void transition regions, a decreased overall layer crystal-linity as well. The smaller magnitudes of the decreases inψ6 and Fwith φnano > φnano* are explained through the same reasoningexcept that void spaces are replaced with multilayer regions. Onlythe microsphere layer in contact with the substrate is imaged foranalysis; thus the difference between amonolayer and the bottomof a multilayer is small except in the mono- to multilayermorphological transition regions. The same argument for theabsence of crystallinity at void/monolayer interfaces holds formonolayer/multilayer boundaries, which often exhibit squarepacking, and explains the ψ6 reduction. The relatively smallermagnitude of this reduction stems from the larger relative size ofmultilayer regions versus their void counterparts. The percentcoverage, however, drops only slightly from the maximumF at φnano* with a higher nanoparticle concentration. Thisstems from the fact that the only real decrease in the coveragein these φnano > φnano* samples with multilayer region samplesoccurs at the monolayer/multilayer boundaries; there is noφnano>φnano* analog to the void interior with submonolayerpatchiness.

Figure 5 summarizes the experimental trials of varying φnanoand φmicro and compares them to the theoretical prediction of

Figure 3. Plot of monolayer deposition velocity, vmono, vs φmicro.The data follows the trend given by eq 2 (---).

Figure 4. (a) Surface crystallinity,ψ6, and (b) percent coverage, F,versus φnano for φmicro = 0.12, 0.16, 0.20, and 0.24. Optimal φnanomaximizing ψ6 and F exists for each φmicro.

Figure 5. φnano vsφmicro where each datum is shaded by its relativeψ6. Darker points are more crystalline. X’s indicate monolayerdepositions at lowφmicrowheremonolayer regions are too short forrobust crystallinity analysis. The dashed line, from eq 5, predictsthe concentration ratio when constituent deposition fluxes arematched.

DOI: 10.1021/la902856x 2405Langmuir 2010, 26(4), 2401–2405

Kumnorkaew et al. Article

φnano* from eq 5. Each experimental datum is shaded with itsrespective ψ6. Note that ψ6 was calculated through the computa-tional analysis of large-deposition-length scans; φmicro e 0.04yielded only short-range (less than 0.5 cm) depositions. In shortdepositions, striping would not necessarily be apparent, and evenin high-quality monolayers, the data lacks statistically viablequantification. Those samples φmicro < 0.12 have a qualitativelyhigh-quality (HQ) monolayer microstructure though they nolonger follow the flux balance and instead form monolayers onlywith φnano = 0.04. Samples at φnano = 0.03 show stripes, andsamples at φnano = 0.05 produce poor-quality depositions. It isunclear why this φnano plateau exists from low tomoderate φmicro.Perhaps these samples never reach steady deposition conditions.At φmicro = 0.12, the data transitions follow our prediction fromeq 5 of proportionally increasing φmicro and φnano. This predictionis an upper limit based on themaximumpossible randompackingand results in a slight mismatch between experiments and predic-tions at φmicro = 0.24. Below our theoretical line, where φnano <φnano* , ψ6 drops because of striping from insufficient nanoparticleflux. For φnano>φnano* , ψ6 is suboptimum but remains high aspreviously discussed. At very high φnano, for all φmicro, (notshown) depletion results and the resulting gel cannot be convec-tively deposited uniformly.

Thus far this discussion has focused on tuning φnano for a givenφmicro. Likewise for a given φnano and deposition rate, changingφmicro alters the surfacemorphology. Figure 6 shows SEM imagesof long-range morphology and short-range microstructure forφnano= φnano* (0.20)=0.06 and v=30μm/s (vmono=30μm/s forφmicro = 0.20) and for φmicro = 0.04, 0.08, 0.12, 0.16, 0.20, and0.24. From the ideal conditions shown in Figure 6e, decreasingφmicro reduces F by forming microsphere-free patches with onlysmall streaks of assembled nanoparticles. For φmicro = 0.16 and0.12, vmono = 20 and 12.5 μm/s, respectively; the submonolayerformed at these relatively higher deposition speeds supports theassumption that deposition quality is dictated bymicrosphere sizeand concentration. At φmicro = 0.24, regions of microspheremultilayers form. Monolayer regions, albeit less prolific undernonoptimal conditions, have exactly the same microstructureacross these samples and thus demonstrate that the degree ofmicrosphere burial is independent of φmicro.

Conclusions

The morphology and microstructure of convectively depositedbinary microsphere-nanoparticle suspensions are controlledthrough constituent fluxes, the ratio of which directly relates tothe ratio of the volume fractions. The theoretical optimumnanoparticle concentration predicted is φnano* = 0.32φmicro, whichis verified through the quantitative analysis of thin film qualitywith ψ6 and F at various φnano and φmicro. Experiments resultingin the deposition of well-ordered microsphere monolayersfrom binary suspensions occur at φnano ≈ φ*nano; with φnano <φnano* , instability causes the monolayer to jump, and withφnano>φnano* insufficient interstitial space for nanoparticles causesmicrosphere spreading and disorder, multilayer formation, or acombination of these two phenomena. Tuning φnano and φmicro

toward optimal monolayer deposition conditions enhances thefield of convective deposition both from intellectual and applica-tion-driven standpoints. From a fundamental scientific stand-point, this work builds on the literature15 by demonstrating agreater understanding of the role and relation of species fluxduring deposition in the tuning of alternate morphologies andnano- and microscale particle interactions. From an application-driven standpoint, the ability to deposit longer and more uniformlayers as well as the ability to control the morphology precisely isvaluable, though the reason for this enhancement is not immedi-ately clear. In practice, the addition of nanoparticles may beconsidered to be a deposition aid. Limitations and questionsraised with this research include the further explanation of binarysuspension behavior at low φnano and φmicro as well as howdepositions of polydisperse suspensions will behave. We exploreunary and binary suspensions, with our binary constituents beingan order of magnitude different in size. This begs the question ofhow ternary and higher-order suspensions would behave underdeposition when intermediately sized particles do not fit withininterstices and similarly how a truly large distribution of particlesizes would behave.

Acknowledgment. We gratefully acknowledge the supportof the National Science Foundation under award number0828426. P.K. is grateful for the support of the Royal ThaiScholarship.

Figure 6. SEMimages ofdepositionswithφmicro=0.04 (a), 0.08 (b), 0.12 (c), 0.16 (d), 0.20 (e), and0.24 (f) atφnano=0.06 and v=30μm/s.Awell-ordered monolayer with high ψ6 and F is formed at φmicro = 0.20, submonolayer morphologies are formed at φmicro < 0.20, andmultilayermorphologies are formedatφmicro>0.20.Locally, the degree ofmicrosphere burial by nanoparticles is constant in regions ofwell-ordered microsphere monolayers.