DOI:10.1002/cphc.201500410 Concepts

Manipulating the Coffee-Ring Effect:Interactions at Work ManosAnyfantakis *[a, b, c] and Damien Baigl *[a, b, c] In memory of Zheng Geng

The evaporation of adrop of colloidal suspension pinned on particleinteractions drastically affect the morphology of the asubstrate usually results in aring of particles accumulated at deposit. General rules are established to control the CRE by the periphery of the initial drop. Intense research has been de- tuning these interactions, and guidelines for the rational physi- voted to understanding, suppressing and ultimately controlling cochemical formulation of colloidal suspensions capable of de- this so-called coffee-ring effect (CRE). Although the crucial role positingparticles in desirable patterns are provided.This of flow patterns in the CRE has been thoroughly investigated, opens perspectivesfor the reliable control of the CRE in real- the effect of interactions on this phenomenonhas been largely world formulations and creates new paradigmsfor flexible par- neglected. This Concept paper reviews recent works in this ticle patterning at all kinds of interfaces as well for the exploi- field and shows that the interactions of colloids with (and at) tation of the CRE as arobust and inexpensive diagnostic tool. liquid–solid and liquid–gas interfaces as well as bulk particle–

1. Introduction 1.1. Motivation for Studying the CRE Particle deposition from evaporating drops of liquidscontain- ing non-volatile solutes phenomenologically seems to be Understanding and controlling solute deposition from evapo- arather uncomplicated problem. However,our everyday expe- rating liquids is akey factor in various technologies, such as rience contrasts intuition. Characteristicring-shaped deposits inkjet printing,[2] DNA and protein microarrays[3,4] andmicro- are observed after the drying of spilled coffee or tea drops. patterning,[5] whichexplainsthe intense researcheffort in the Washed tablewaredisplays white stains of the same morpholo- last two decades for revealing the underlying phenomena in gy,since salt crystals accumulate at the edge of the aseeminglycuriosity-driven research topic.[6] drops after drying. Rain drops also leave rings of dust particles From apurely scientific point of view,the CRE offers arich after drying on windows. Surprisingly,the omnipresenteffect physicochemical platform for the investigation of numerous, responsible for the formation of the above-mentioned pat- complex and usually entangled phenomena. Microscopic ef- terns, frequently referred to as the coffee-ring effect (CRE), was fects, such as particle–particle andparticle–interface interac- elucidated only less than two decades ago. Deegan et al.[1] ex- tions, macroscopic effects, such as flow patterns, and phenom- plained that, in volatile liquid droplets partially asolid ena simultaneously involving different lengthscales, such as substrate, the evaporation rate has itsmaximum close to their wetting, are all integrated in achallenging physicalproblem.[7] pinned three-phase contact line. This inhomogeneous evapora- In addition, the robust and persistentcharacter of the CRE tion profile results in the development of aradial fluid flow makesitvery intriguing. The primary requirements for the CRE from the interior of the drop to the contact line, to replenish to take place—volatile liquid,finite , and pinned the liquid lost there. Solutes present in the liquid are transport- contact line—arecommonly met in the majority of practical ed by the flow and are deposited at the edge of the drop; systemsand applications,and this accounts for its ubiquitous their progressive accumulation leads to the formation of ring- nature.[1] Moreover,itoccurs in awide range of drop sizes (di- shaped patterns after drying is complete. ameters ranging from micrometers to centimeters)[8,9] and in aplethora of systemswith diverse chemical properties and var- [a] Dr.M.Anyfantakis,Prof. D. Baigl ious characteristic length scales. Examples include molecular Ecole Normale SupØrieure-PSL Research University (i.e. salt)[8] and macromolecular solutions,[10] colloidal suspen- Department of Chemistry sions (particle sizes from nanometers to micrometers)[11] and 24 rue Lhomond, 75005,Paris (France) E-mail:[email protected] biomaterial-based complex fluids (e.g. blood, protein solutions, [11–13] [email protected] and biofilms ). [b] Dr.M.Anyfantakis,Prof. D. Baigl Sorbonne UniversitØs UPMC Univ Paris 06, PASTEUR 1.2. Strategies for Controlling the CRE 75005, Paris (France) Asignificant part of research studies devoted to the CRE was [c] Dr.M.Anyfantakis,Prof. D. Baigl CNRS, UMR 8640 PASTEUR focused on uncovering the physicochemical parameters affect- 75005, Paris (France) ing evaporative particle depositionand particularly on devising

ChemPhysChem 2015, 16,2726 –2734 2726 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts strategies for suppressing the effect and achieving homoge- cle affinity for the LG interface is optically tuned, particle depo- nous solute deposition. Given that evaporation-driven flow sition can be dynamically controlled by using light (Sec- toward the pinned contact line is always present in avolatile tion 3.1). Moreover,wedescribe recently developed biosensing sessile drop, the vast majority of these approaches was based strategies based on dry patterns resulting from particleaggre- on affecting the hydrodynamics of drying drops. Different ad- gation induced by the attraction between probe particles and ditives, including surfactants,[10, 14] polymers,[15] cosolvents[2] and the biomarker molecule to be detected, which open the way nanoparticles,[16] were introduced to manipulate the flow pat- to low-cost and simple-to-operatepoint-of-care diagnostics terns in evaporating drops. Externalmethods that requireno (Section3.2). Conclusions and perspectivesare given in additives, such as electrowetting[17] and exposure of the drying Section4. drops to vapours of miscible, low-surface-tension liquids,[18] have also been successfully utilised. Other approaches relied on the addition or removal of deposited matter by mechanical 2. Interactions of Particles with/at Interfaces means.[19] Affecting the CRE Anumber of groups have investigated the effect of the various 1.3. Scope and Organisationofthis Paper types of interactions in adrying drop of colloidal suspension on the resulting particle deposition. In the following, we cate- The interconnection between flow patterns and final deposit gorise some of the most relevant publications based on the morphology has been recently described in detail in an in- type of interaction:particle–LS interface, particle–LG interface structive review by Larson.[7] The scope of this Concept paper and bulk particle–particle interactions. However,this classifica- is primarily to underline the generally overlooked but highly tion is not mutuallyexclusive, since it is common that in important influence of interactions on deposition patterns asingle publication, more than one of the above-mentioned emerging from evaporating drops. In particular,wefocus on types of interaction are investigated. the interactions of the suspended particles with/atthe liquid– gas (LG) and liquid–solid (LS) interfaces, as well as bulk parti- cle–particle interactions. In drying drops of most complexliq- 2.1. Interactions with the LS Interface (Substrate) uids, the morphologyofthe final deposits is determined by both the hydrodynamics (i.e.flow patterns) and the interac- Bhardwajetal.[20] studied the role of the Derjaguin–Landau– tions present in the system (Figure 1). This is expectable con- Verwey–Overbeek (DLVO) interactions on the pattern morphol- ogy from evaporating nanoliter droplets of asuspension of ti- tania nanoparticles deposited on glass. Variation of the suspen- sion pH led to modification of the DLVO particle–substrate and particle–particle interactions, which in turn alterated the shapes of the deposits. Evaporating drops at low pH values yielded arelativelyhomogeneous pattern consisting of athin uniform nanoparticle layer with athicker ring at the edge. Cal- culations showed that the particle–substrate DLVO force at low pH values was attractive. Particles in close proximity to the Figure 1. The two key elementsdictatingdeposit morphology from evapo- substrate were thusattracted to it and formed auniform de- rating sessile dropsofcolloidal suspensions:flow patterns and interactions. The latter, which are at the focus of this paper,include interactions of the posit (Figure 2a). Deposits obtained from drying drops at inter- suspended particles with the LS and LG interfaces, as well as bulk particle– mediate pH consisted of aggregates covering the whole initial- particleinteractions. ly wetted area. Near the point of particleneutral charge, the particle–substrate DLVO force was weaker than the interparti- sidering that most of interactions involving colloidsand inter- cle van der Waals attraction, and hence particle aggregation faces (e.g. Coulomb and van der Waalsinteractions, capillary occurred. At the highest pH values, drop drying led to forces, hydrophobic interactions) can typicallyexceed the ther- amarked ring pattern with almost no particles in the area en- mal energyand therefore can strongly affect particle organisa- closed by the ring. In this pH region, strong repulsive particle– tion and behaviouratinterfaces. substrate forces prevented the nanoparticles from contacting The paper is structured as follows.Selected recent publica- the substrate, and they were carriedalong by the outward ca- tions that have explored the effect of interactions of different pillary flow to the drop contact line (Figure 2b). nature are highlighted in Section 2. After analysing the role of Aphase diagram was proposed forprediction of the deposit the interactions with/atthe LS (Section 2.1) and LG (Sec- shape from evaporating drops. It was suggested that the de- tion 2.2) interfaces, we analyse the effects of bulk particle–par- posit shape resultsfrom the competition of three transport ticle interactions (Section 2.3) prior to providingalist of gener- mechanisms in the drop:1)the outward capillaryflow favour- al guidelines to control particle patterning from drying drops ing ring-shaped deposits;2)particle transport towardthe sub- (Section2.4). Next, we describe afew paradigms in which strate,driven by attractive DLVO interactions and promoting these interactions have been exploited for applicationsofprac- uniform deposits;and 3) aMarangoni recirculatory flow,driven tical importance (Section3). In suspensionsinwhich the parti- by surface-tensiongradients and favouring acentral bump de-

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2727 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts

tant layer adsorbed on the substrate and that adsorbed on the colloid surface. Recently,Dugyalaand Basavarajinvestigated the influence of particle shape and DLVO interactions on the patterns formed by drying sessile droplets.[22] They employed aqueous suspensions of hematite ellipsoids, amodel system with ad- justableparticle aspect ratio (through synthesis) and surface charge (throughpHvariation). For afixed aspectratio of 4and at lowpH, hematite drops deposited on cleaned glass yielded rings enclosing an area covered with amono- or multilayer of particles. Ring formation was attributed to the repulsive parti- cle–particle interactions allowing particles to migrate to the drop edge, driven by the outward capillary flow.The deposi- tion of some particles inside the area enclosed by the ring was ascribed to particle–substrate attraction. At intermediate pH, reduced Coulomb particle–particle repulsion and van der Waals interactions resulted in particle aggregation. At the Figure 2. The influence of pH-dependent particle–LS interface (substrate) same time, particle–substrate attraction led to the deposition DLVO interactions on the deposit morphologyfrom drops of titania nano- particles drying on glass substrates.a)Atlow pH,particle–substrateinterac- of single particles and aggregates on the glass, and auniform tions are attractive, promote particle adhesion at the substrate and eventu- dry pattern was produced. At highpH, both particle–particle ally lead to ahomogeneous deposit. b) At high pH, particle–substrate inter- and particle–substrate interactions were repulsiveand led to actionsare repulsive, and the evaporation-induced capillary flow brings par- the formation of aring pattern (Figure 3). ticles to the contact line,where they form aring-shaped pattern.Adapted with permission from Ref. [20].Copyright 2010 American Chemical Society. Independent of the particleaspect ratio andfor drops drying on cleaned glass, rings were always observed for acidic and basic conditions, whereas homogeneous patternswere posit with adiameter much smaller than the initial drop obtained for intermediate pH values. Surface-tension measure- diameter. Yanetal.[21] investigated the impact of particle and substrate charge on the colloidal self-assembly close to the edge of an evaporating sessile drop, in the absence and presence of sur- factants, at concentrations below the criticalmicellar concen- tration (CMC). It was demonstrated that particle mobility,gov- erned by the particle–substrate interactions, was akey parame- ter affecting particle ordering and deposition. For surfactant- free suspensionsofparticles having the same chargeasthe substrate, ring-shaped deposits consisting of significantly or- dered self-assembled particles were observed. Contrarily,drops of particles on an oppositely charged substrate led to the for- mation of rings with an increased number of depositedparti- cles in the drop interior.Attraction betweenthe colloids and the LS interface resultedindecreased particle mobility and therefore disordered structures. In the presence of anonionic surfactant, the self-assembly behaviour was similar to that of the surfactant-free systems. When acharged surfactant with the same charge as the particle and the substratewas added to the formulation, particle ordering andpatterning were simi- lar to those of the like-charged, surfactant-free systems. For oppositely charged particle–substrate systems, addition of asurfactant havingthe oppositecharge to the particle pro- Figure 3. Deposition patternsfrom evaporating dropsofhematite ellipsoids moted the formation of rings consisting of ordered self-assem- (fixed aspectratio of 4) dispensed on glass substrates. At low pH, ring de- blies. This was explained by surfactant adsorption to the parti- posits enclosing mono- or multilayersofparticles are formed due to the combination of particle–substrate attraction,particle–substrate repulsion cles, whichdecreased the particlechargedensity and therefore and the outward capillary flow.AtintermediatepH, reducedelectrostaticin- weakened adhesion to the LS interface. Finally,inthe case of terparticlerepulsion and attractive van der Waals interactions cause aggre- like-charged particle–substrate systems, addition of an oppo- gation. Attractiveparticle–substrate interactions lead to the deposition of sitely charged surfactant led to strong adhesionofparticles to single particles and aggregates on the substrate in auniform pattern. Parti- cle–particle and particle–substrate repulsion in conjunctionwith the out- the LS interface and thus promoted disorder.Thiseffect was ward flow resulted in ring deposits at high pH values. Adapted with permis- attributed to the hydrophobic attraction between the surfac- sion from Ref. [22].Copyright 2014 American Chemical Society.

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2728 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts ments in pendant drops of suspension in decane indicated drivingparticles from the edge to the centre of the drop. This that hematite particles were adsorbed at the water–decanein- inwardmotion was attributed to capillary forces exerted on terface for intermediate pH, whereas no adsorption occurred the suspended particles when the latter were in contact with under acidic conditions. Surprisingly,atintermediate pH, the the LG interface duringdrying, caused by the geometrical con- hydrophobicity of the substrate was found to determine the straints,that is, the LG interface, the contact line and the con- dry-deposit morphology.Whereas drops on cleaned (i.e. hydro- tact angle, given that no particleprotrusion from the free in- philic) glass yieldedhomogeneous patterns,drops on silanised terfacewas observed. glass led to the formation of rings. The authors concluded Arraysdisplayinglong-range order (over several microme- that, at intermediate pH, particle–substrate interaction was the tres) were obtained from drying drops of dodecanethiol- dominant factor determining the pattern morphology:negligi- coatedgold nanocrystals dispersed in toluene.[26] Although ini- ble interaction (hydrophobic substrates) led to coffee rings, tially thought to be the result of particle–particleand particle– whereas attraction (hydrophilic substrates) led to ring suppres- substrate interactions as well as wettingofthe liquid on the sion, with particle adsorption to the free interface occurring in solid surface,[26] later studies involving in situ small-angle X-ray any case. scatteringdemonstrated that highly ordered 2D superlattices Morales et al.[23] studied the influence of nonionic surfactants were formedatthe LG interface of the evaporating drop.[27] on the dry-pattern morphologies and the attachment strength The formationofthe 2D superlattices was explained by akinet- of particlesonthe substrate from evaporating sessile drops. At ic “crushing” model describing the accumulation of nanocrys- high initial surfacetensions(i.e. low initial surfactant concen- tals at the descending LG interface. Bigionietal.[28] further trations), amorphous stains were observed, at intermediate ini- showedthat asufficiently high flux to the LG interface (con- tial surface tensions, coffee-ring patterns were obtained, and trolled by evaporation rate and particle concentration) and low surface tensions(i.e. high surfactant amount) resulted in asufficiently strong particle–interface interaction were required patterns consisting of concentric rings. These morphologies to create 2D islands, the formation of which was aprerequisite were linked to three distinct drying regimes attributedtotran- for macroscopic monolayer formation at later drying stages. sitions of contact-line dynamics between slipping, pinned and The shape of suspended particles was shown by Yunker recurrentslip–rip–stick states,respectively.The presence of the et al. to be acrucial factor determiningthe morphologyof surfactant did not modify the van der Waals and Coulombpar- evaporative deposits.[29] The authors utilised dispersions of ticle–substrate interactions. However, with increasing surfac- polystyrene spheres stretched asymmetrically to different tant concentration,adsorption of micelles on the particleand aspect ratios a.During drop drying, spherical or slightly de- substrate surfaces led to micelle-protrusion repulsion, which af- formed particles (a=1.0–1.1)were efficientlytransported to fected the interactions betweenthe contactline and the parti- the pinned contact line by the outward capillary flow,and cles and in turn influenced the deposition and attachment atypical ring was formed (Figure 4a andb). On the contrary, strength of particles on the substrate. evaporation of drops of suspensionconsisting of more aniso- We have recently investigated the influence of the interac- tropic particles (a>1.1) led to the formation of apattern dis- tions between the suspended particles and the LS interfaceon playing uniform coverage (Figure 4c). The ellipsoidal particles the deposit morphology resulting from the dryingofdrops of were only transported toward the drop periphery until they aqueous colloidal suspension on glass substrates.[24] Surfactant- reachedthe LG interface. After being carriedthere, the ellip- free dispersions always led to the formation of ring deposits, soids experienced long-range attraction between each other, since the majority of particles accumulatedatthe drop edge. because they deformed the LG interface. These strong attrac- However,the amount of particles deposited in the area encir- tions resulted in the formation of loosely packed, arrested cled by the ring varied depending on the particle andsub- structures at the interface, which gave rise to ahigh surface strate charge. For particles and substrates having the same viscosity and enabled the colloidstoresist the outward flow charge,only alimited number of particles were deposited in (Figure 4d). The strength and range of the particle–particle ca- the interior area. Contrarily,for oppositely charged particle– pillary interactions were decreased by the addition of surfac- substrate pairs, electrostatic attraction led to the formation of tant to the drops. This enabled ellipsoids to pack closely at the rings enclosing an area covered with an increased number of contact line, since they were more flexible toward rearrange- colloidal particles. The addition of ionic surfactants (at concen- ment at the interface, and eventually resulted in the formation trationslower than the CMC) was found to modify the electro- of aring pattern after drying. Drop evaporation of mixtures of static properties of the particles and the LS interface and there- spheres with ellipsoids resulted in uniform patterns, provided fore to modulate the amount of particles adsorbed on the that the sphere diameter was larger than the minor axis of the glass. ellipsoid. This effect was ascribed to the hindered motionof large spheres under or through the looselypacked particle structures, which prevented the former from reaching the con- 2.2. Interactions with/atthe LG Interface (Free Interface) tact line. Amotion opposingthe CRE, driving solute accumulationat We have recently reported on the effect of surfactants on the centerofevaporating drops, was described by Weon and the deposition patternsfrom evaporating drops of colloidal Je,[25] who experimentally observed that the initial evaporation- dispersion on glass substrates at concentrationslower than driven particle transport toward the contact line was reversed, the CMC.[24] We demonstrated that the surfactant-mediated in-

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2729 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts

interface, presumably due to hydrophobic interactions. Particle trappingand aggregation at the free interface led to the for- mation of askin phase, the deposition of whichresulted in ho- mogeneous, disc-shaped dry patterns(Figure 5b). Withafur- ther increaseinsurfactant concentration,particles were over- chargedbythe surfactant and again became hydrophilic; coffee ring deposits were once again formed after drop evapo- ration. Notably,weshowedthat the evolution of the deposit pattern followed areproducible and general trend for avariety of particle/surfactantsystems. Below the CMC, like-charged systemsalwaysled to rings, whereas oppositely charged sys- tems displayed aring–disc–ring evolution with the discmor- phologysystematicallyoccurring at intermediate surfactant concentrations that closely corresponded to neutralisation of the particle surfacecharge (zeta potential around zero). Ex- ploitingasimilarconcept, we recently demonstrated the dy- namic control of particle deposition from evaporating drops using light,[30] which is described in more detail in Section 3.1.

Figure 4. Colloidal shape influences interactions of particles with/at the LG 2.3. Bulk Particle–Particle Interactions Affecting the CRE interface and therefore deposit morphology in drying drops.a)Drops of sus- pensions of spherical colloids result in atypical ring pattern, since most of In Section 2.1, devoted to the role of the particle–LS interac- the particles are gatheredatthe drop edgedue to the evaporative capillary tions, we already described some examples in whichthe influ- flow (b). c) In stark contrasttospheres, drops containingellipsoidal particles ence of bulk interparticle attraction/repulsion on the deposit yield homogeneous patterns. Particles reachingthe LG interface experience shape had to be taken into account.[20, 22] In this section, we long-range attractions and form loosely packednetworks, which enable par- ticles to resist the evaporation-driven flow (d). Adapted by permission from focus on other studies in which the role of particle–particle in- Macmillan Publishers Ltd:Nature(Ref. [29]), copyright 2011. teractions was considered to be the predominant factor driv- ing the deposition behaviour. Talbot et al. quantitatively investigated the effect of visco- teractions between the suspended particles and the LG inter- elastic properties of aqueous suspensions of polystyrene parti- face are the key factor determining the morphology of the dry cles and laponite, adisc-shaped nanoparticle, on the deposit pattern (Figure 5). For like- chargedparticle/surfactant mix- tures, the majority of the parti- cles accumulated at the drop edge, forming ring patterns. A markedlydifferent behaviour was shown for oppositely chargedmixtures, whereby elec- trostatically driven surfactant ad- sorptiontothe particle surface was responsible for the en- hanced particleaffinity to the LG interface. For low surfactant con- centrations,surfactant adsorp- tion to the particle surfacewas limited, and ring patterns were always observedupondrying (Figure 5a). For intermediate concentra- tions, surfactant-coated particles Figure 5. The effect of surfactant-mediated particle–LG interface interactions on the pattern morphology from evaporating dropsofanionic particle–cationic surfactant mixtures. a) At zero or low surfactant concentration, par- becamealmost neutral and at ticles remain anionic and hydrophilic due to zero or limitedsurfactant adsorption on their surface. The evapora- the same time more hydropho- tion-driven convective flow leads to the formation of rings. b) At intermediate surfactant concentrations, surfac- bic, due to exposure of the tant adsorption leads to neutralisation and hydrophobisation of the particles. Trapping and aggregationofthe apolar tails of the surfactant to particles at the LG interface createsaparticle skin, the deposition of which leads to ahomogeneous disc-shaped deposit. c) At high surfactantconcentrations, surfactant molecules form abilayer at the surfaceofthe particles, the water phase. This in turn led which become cationicand hydrophilic. The outward convectiveflow leads to the formationofrings once again. to particle adsorption atthe LG Adapted with permission from Ref. [24].Copyright 2015 American Chemical Society.

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2730 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts morphologies after the drying of inkjet-deposited droplets.[16] Crivoi and Duan studied the effect of hexadecyltrimethylam- Laponite was introduced to induce asol–gel transition in the moniumbromide on the deposit morphology obtained by drying drops, which transforms the liquid from an initially evaporating droplets of dispersed aluminium oxide nanoparti- stable colloidal dispersion to asoft gel. Gelationoccurred cles on Si wafers.[31] In the surfactant-free dispersions,particles owing to the growth of alaponite network spanning the liquid tendedtoaggregate into clusters, which upon dryingdeposit- volume, which arose from the connections between the lapon- ed into auniform layer enclosed by avisible ring. Addition of ite particles as their concentrationincreased duringevapora- surfactant reduced particle aggregation and led to the forma- tion. In particular,the authors suggestedexploiting the elastici- tion of typical ring patterns, with only alimited number of ty rather than the viscosity enhancement caused by the sol– nanoparticles deposited in the pattern interior. The particle– gel transition. By varying the initial concentration of laponite substrate attraction was comparable for the surfactant-free in asuspension with afixed concentrationofpolystyrene parti- and surfactant-containing dispersions anddid not play amajor cles, the gellingtime and therefore the amount of convective role in the pattern morphology.The authors developed radial flow could be adjusted,and thus the final deposition amodel based on diffusion-limited cluster–cluster aggregation pattern could be controlled. to qualitatively explain the experimental results. By varying the Without the addition of laponite, the convectiveradial flow particlesticking probability from 0to100%, acontinuous dominated and ring-shaped deposits wereformed(Figure 6a, transformation from acoffee ring to auniform pattern was ob- served in the simulations. There- fore, controlling the particle stickingprobability by modifying the fluid properties (in this case by the addition of surfactants) offered away to manipulate the self-assembly patterns formed from evaporating drops.[31] The concept of using the LG interfaceofadrying droplet as atool for directing the assembly of colloids into microscopic ob- jects displayinglong-range order was demonstrated by Kuncicky and Velev.[32] The shape of the Figure 6. The effect of laponite-induced gelation on the deposit formationfrom evaporating drops containing afixed amount of polystyrene nanoparticles. a) SEM images and b) corresponding schematicdeposit height pro- templating free interface and files. 1–3:Laponite-free drops lead to the formationofring-shaped deposits, becausemost of the particles accu- thus the morphology of the final mulate at the drop periphery due to the convectiveflow.4–6:Additionof2wt%laponite leads to fast gelation, assemblies were determined by which rapidly suppresses the convectiveflow and produces adome-shaped deposit. 7–9:Amoderate amountof the dynamics of the receding convective flow is achieved by slower gelation, which is realised by utilising alower laponite concentration (1 wt%). Drying drops in this case led to the formation of homogeneous disc-shaped dry patterns. Adapted with contact line. The drop contact permission from Ref. [16].Copyright 2014 American Chemical Society. angle,the initial particle concen- trationand the amount of elec- trolytewere employed as the 1–3). In suspensionscontaining small amounts of laponite, parameters controlling the contact-line dynamics. The mor- asol–geltransitionoccurred after acertain amount of time phologies of dry microstructures fell into two categories. (controllable by meansofthelaponite concentration), with the Convex structures displaying aspherical cap (convex)shape drops first gelling at their edge and the gelation front propa- and ring-shaped (concave) structures dimpled inward were ob- gating toward the centre. The moderate amountofradial flow served, depending on the substrate wettability and theinitial led to the formation of homogeneous, pancake-likedeposit volumefraction of the particles. To correlate the electrostatic morphologies (Figure 6a,7–9). When higher laponite concen- particle–particleand particle–substrate interactions to the trationswere utilised, the radial flow was suppressed very dryingdynamics and the resultant deposit morphology,the re- soon after drop deposition duetofast gelation;finally,dome- searchersexamined the effect of added in drops like deposits wereobtained (Figure 6a,4–6). Althoughthe ad- dryingonhydrophobic substrates. For low salt concentrations, dition of laponite successfully homogenised the deposition particlesadheredtothe substrate during drying, and this led patterns,italso led to aggregation of the polystyrene colloids. to the pinningofthe contact line, which was initially free to This aggregation was inhibited by adding silica nanoparticles recede. The final pattern had the form of aspherical cap. At in- to the suspensions. This study clearly illustrated how particle– termediate concentrations, the contact line was pinned during particleinteractions in the bulk could directly influence the the whole lifetime of the drop. The particles in the vicinity of local structure andthe overall morphology of the deposits re- the free interface coagulated and created ashell;the dry de- sulting from drying picoliter droplets. posit consisted of the crumpledshell. Finally,high ionic strength led to enhanced particleaggregation and sedimenta-

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2731 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts tion, which resulted in a“sand-pile” micropatch. Besides the sedimentation can lead to an overall uniform but locally inho- salt-induced screening of the electrostatic repulsions, the au- mogeneous deposit morphology. Alternatively,the viscoelastic thors noted that hydrophobic attractions could also contribute properties of nanoparticledispersions that are prone to gela- to the adhesion of particles to the substrate. tion can be exploited for the even patterning of co-suspended particles, which provides asuccessful route to homogenised deposits. 2.4. General Guidelines for Controlling Patternsfrom The existence of the above-mentioned general rules relating Evaporating Drops the interactions in an evaporating drop and the resulting de- Particle patterning from drying drops is generally acomplex posit morphology open the way to the exploitationofthis re- process, accompaniedbyadditional complications specific to lation for practical applications. Afew examples toward this each liquid used. However,from the studies describedabove, end are described in Section3. we conclude that some general guidelines for controlling evaporative patterning do exist and it would be useful to list 3. Exploiting Interaction-Driven Particle them here. These guidelines are based on the interactions be- Deposition tween the components of adrying drop, that is, the particles and the interfaces. Regarding particle–LG interface interactions, The direct relationbetween the interactions occurring in an it has been shown that both neutralisation and hydrophobisa- evaporating drop and the morphology of the dry deposit from tion of the particles promotes their affinity for that interface. evaporating drops can be exploited in two ways. On the one This commonly leads to particle trapping at the LG interface hand, on-demand tuning of the relevant interactions in and the formation of interfacial aggregates, which have apro- adrying drop can lead to tailoredparticlepatterning. Con- found effect on the morphologyofthe dry pattern. The gener- versely,characterisation of the deposited pattern can give val- al trend is that promoting particletrapping at LG interface hin- uable information aboutthe interactions taking place in ders ring formation and favoursdisc-like deposits with avaria- adrying drop of amulticomponent liquid (Figure 7a). Para- ble degree of homogeneity.Sofar,this has been achieved: digmsexploiting this bidirectional nature of the interactions– 1) by the addition of oppositely charged surfactants in aprecise depositrelation are presented below. concentration range leading to the adsorption of asurfactant monolayer that causes particleneutralisation and 2) by adjust- 3.1. Tailored Particle Deposition at Interfaces ing the pH of the liquid,which determines the degree of pro- tonation/deprotonation of the surfacegroups.Weanticipate We recently reported on the design of photoresponsive colloi- that other candidates, such as molecular (through dal suspensionsbymixinganionic polystyrene particles with the screening of the electrostatic interactions) or polyelectro- the photosensitivecationic surfactant AzoTAB, whichconsists lytes (throughelectrostatically driven particle coating)could of apolar head group and ahydrophobic tail with an azoben- lead to similarphenomena and we think they are worthy of zene moiety.Byirradiating with blue (l=440 nm) or UV (l= systematic studies. Another scientific challenge is to discrimi- 365 nm) light,the isomeric state of AzoTAB could be reversibly nate the exact contributionofeach of the two main compo- changed between trans (less polar,less hydrophilic) and cis nents, electrostaticand hydrophobic interactions,whichare (more polar,more hydrophilic), respectively.This had adirect entangled in many experimental systems. consequence for the amount of surfactant molecules bound Considering particle–LS interfaceinteractions as the platform on the particlesurface:for afixed particle concentration, for guiding evaporative particle deposition, we have shown alarger amountoftrans-AzoTAB surfactant molecules were ad- that, regardless of their origin (Coulomb, van der Waals or hy- sorbed on the particle surfacecompared to the cis isomer.This drophobic), attractive interactions with the LS interface pro- enabledustofinely and reversibly tune the particle surface mote pattern homogenisation from drying drops. There is thus properties (charge and hydrophobicity) and therefore their in- astronganalogy betweenthe role of LG and LS interfaces in teraction with the LG interface, solely by irradiation with light, directingthe deposition behavior:particle affinity for any of while keeping the liquid composition unchanged.[30] these interfaces competes with the outward capillaryflow and In away analogoustothe case of common (i.e. non-photo- promotes uniform deposition. However,the nature, range and sensitive) surfactants, particle trapping and clustering at the LG intensity of the interactions at work can be very different be- interfacecould be triggered by light by exploitingparticle neu- tween these two interfaces. For instance, long-rangecapillary tralisation and hydrophobisation induced by light-dependent forces, which are specific to the LG interface, were shown to adsorption of AzoTAB on the particles. Drying of an irradiated be instrumental in directing the deposition behaviour.Con- single drop could reversibly yield aring or ahomogeneous versely, long-range electrostatic interactions can be easily disc pattern, depending on the wavelength used. Spatialcon- tuned in the specific case of the LS interface, by simple surface trol of evaporative deposition in an array of multiple identical modifications. drops could be achieved by utilising structuredlight patterns Finally,modulating particle–particle interactions can signifi- by meansofasimple photomask (Figure 7b). Notably,byad- cantly affect evaporative particledeposition as well. Generally, justingthe irradiation time, we were able to finely control the in drops of suspensionsofpoorly stabilised colloids, aggrega- amount of particles deposited at the drop edge andits interior, tion leads to clustering. When the cluster size is large enough, and therefore to precisely adjust the degree of homogeneity

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2732 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts

atarget nucleic acidmolecule that relied on the hybridisation- inducedsuppression of the CRE. They employedsuspensions of polystyrene microparticles func- tionalised with two different probe oligonucleotides that were complementary to the target DNA. When the latter was presentinsolution, simultaneous hybridisation of the target DNA with the two probes occurred and led to agglomeration throughmicroparticle bridging. Anisotropic aggregates trapped at the LG interface caused distor- tion of the surrounding menis- cus, which resulted in long- range attraction to other aggre- gates. This in turn led to the gradualformation of anetwork at the free interface, which pre- vented particlesfrom forming aring structure at the drop con- tact line, and the CRE was even- tually suppressed. On the contra- ry,when the target DNA was presentatconcentrations below Figure 7. a) The interactions/deposit morphology relation can be harnessed for programming desired particle the detection limit of this 8 1 deposition on substrates or as astraightforward and low-cost diagnostictool. b) Photocontrol of the CRE in method( 10À mol LÀ ), atypical  drying dropscontaining nanoparticles andphotosensitivesurfactants. Spatial control of colloidal deposition in an ring pattern was observed, since array of multiple drops by meansofasimple photomask.Adapted with permission from Ref. [30].c)Temporal (non-aggregated) particles were controlofdepositedparticle distribution from an evaporating drop.The distribution of depositedparticles can be tuned on demand simply by adjusting the irradiation time. Reprinted with permission from Ref. [30].d)Detection free to be transported to the of atarget DNA molecule based on the hybridisation-induced suppression of the CRE. Simultaneous hybridisation drop edge by the radial convec- of the target DNA with two colloidal particles functionalised with probe oligonucleotidescomplementary to the tive flow (Figure 7d). By simple target DNA leads to the formation of ahomogenousdeposit. On the contrary,the absence of the target DNA re- image processing, these authors sults in the formation of atypical coffee-ring pattern. Reprinted from Ref. [33],with permission from Elsevier. were able to measure the extent of coffee-ring suppression and in the final deposit (Figure 7c).[30] This is the first example in therefore to quantify the concentration of the present target which the morphology of the pattern resulting from an evapo- DNA. Besides its effective sensitivity, this method is highly spe- rating drop could be dynamically tuned at constant drop com- cific in that it can distinguish sequences with asingle mis- position, since we demonstrated photoreversible switching be- matched nucleotide. tween marked rings and homogenouspatterns. Another interesting example that made use of the morphol- We thus believe that such particles with optically tunable af- ogy of the deposited pattern to extract information about the finity for the LG interface provide astartingpoint for the pro- presence of atarget molecule in aliquid sample was recently grammable patterning of colloids. An interesting perspective is reported by Trantum et al.[34] The detection strategy was based the possibility of intradrop patterning,that is, the controlled on the combination of Marangoni flows and target-mediated formation of topological features inside asingle drop by using particleaggregation. The utilised particles were functionalised opticalpatterns.Research in this direction is currently ongoing with aM13 monoclonal antibody able to bind to the M13K07 in our group and preliminary resultsare encouraging. bacteriophage, which was the target molecule. Colloids re- mained dispersed or were aggregated when the bacteriophage was absent or present in solution,respectively.Inthe former 3.2. Patterns from Evaporating Drops as aDiagnostic Tool case, particles were transported to the contact line by Maran- The possibility of exploiting the deposition patterns resulting goni flow from the drop apex to the drop edge. Contrarily,in from drying drops as atooltodetect target molecules in bio- the presence of the bacteriophage, large aggregates settled to logical fluids has recently attracted scientific interest. Li et al.[33] the substrate, resulting in particle accumulation in the centre presented an easy and highly efficient method to detect of the drop. The biosensor signal, based on the spatial distribu-

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2733 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Concepts tion of the deposited particlesupon drop evaporation, could Keywords: aggregation · coffee-ring effect · colloids · be easily measured by means of optical microscopy.Adeposit evaporation · interfaces with only asmall amountofparticles gathered at the centre of the dropcorresponded to anegative test (indicating the ab- [1] R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber,S.R.Nagel, T. A. Witten, sence of the target molecule), whereas adeposit with alarge Nature 1997, 389,827 –829. spot at the centre corresponded to apositive test. [2] J. Park, J. Moon, Langmuir 2006, 22,3506 –3513. [3] Y. Deng, X.-Y.Zhu, T. Kienlen, A. Guo, J. Am. Chem.Soc. 2006, 128, 2768 –2769. [4] R. Blossey,A.Bosio, Langmuir 2002, 18,2952–2954. [5] H. Ma, J. Hao, Chem.Soc. Rev. 2011, 40,5457 –5471. 4. Conclusions and Perspectives [6] R. G. Larson, Angew.Chem. Int. Ed. 2012, 51,2546 –2548; Angew.Chem. 2012, 124,2596 –2598. The take-home message of this Concept paper is that, in com- [7] R. G. Larson, AIChE J. 2014, 60,1538–1571. bination with flows, the interactions between the components [8] R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S. R. Nagel, T. A. Witten, of particle-laden drops (particles and interfaces) define the Phys. Rev.E2000, 62,756 –765. characteristics of the deposition pattern after drying. This inter- [9] X. Shen, C.-M. Ho, T.-S. Wong, J. Phys. Chem.B2010, 114,5269–5274. [10] T. Kajiya, W. Kobayashi, T. Okuzono, M. Doi, J. Phys. Chem. B 2009, 113, action–deposit morphology relationcan be exploited in two 15460 –15466. directions:1)the programmable patterning of particles from [11] T. S. Wong,T.H.Chen,X.Shen, C. M. Ho, Anal. Chem. 2011, 83,1871 – drying drops and 2) as adiagnostic tool to gatherinformation 1873. about the physics/chemistry taking place between the compo- [12] W. Sempels, R. De Dier,H.Mizuno, J. Hofkens, J. Vermant, Nat. Commun. 2013, 4,1757. nents of adrop. [13] D. Brutin, B. Sobac, B. Loquet, J. Sampol, J. Fluid Mech. 2010, 667,85– Regarding the former,the vivid scientific interestofthe last 95. two decades has led to valuable accumulated knowledge of [14] T. Still, P. J. Yunker, A. G. Yodh, Langmuir 2012, 28,4984 –4988. the interconnected effects occurring in adrying drop that di- [15] L. Cui, J. Zhang, X. Zhang, L. Huang, Z. Wang, Y. Li, H. Gao, S. Zhu, T. Wang, B. Yang, ACS Appl. Mater.Interfaces 2012, 4,2775 –2780. rectly impact the deposition pattern. However,increasing the [16] E. L. Talbot, L. Yang, A. Berson, C. D. Bain, ACS Appl. Mater.Interfaces complexity of the drop liquids, required for the development 2014, 6,9572 –9583. of new applications, calls for an improvementofour under- [17] H. B. Eral, D. M. Augustine,M.H.G.Duits, F. Mugele, Soft Matter 2011, 7, standing. Such an achievementmay provideacomplete tool- 4954 –4958. [18] M. Majumder,C.S.Rendall, J. A. Eukel, J. Y. L. Wang,N.Behabtu, C. L. box for the programmable and possibly complex patterning of Pint, T. Liu, A. W. Orbaek, F. Mirri, J. Nam, A. R. Barron,R.H.Hauge, H. K. solutes at all kinds of interfaces (fluid–solid, fluid–fluid, curved, Schmidt, M. Pasquali, J. Phys. Chem.B2012, 116,6536 –6542. flat). We believe that the stronginterplaybetween the interac- [19] N. C. Schirmer,S.Strçhle, M. K. Tiwari, D. Poulikakos, Adv.Funct.Mater. tions (particle–particleand particle–interface) taking place 2011, 21,388–21395. [20] R. Bhardwaj, X. H. Fang, P. Somasundaran,D.Attinger, Langmuir 2010, inside adrying drop and the final morphologyofthe dry pat- 26,7833 –7842. tern makes possible the development of “coffee-ring diagnos- [21] Q. Yan, L. Gao, V. Sharma, Y.-M. Chiang,C.C.Wong, Langmuir 2008, 24, tics”, afield in which lookingatdeposit morphological features 11518–11522. will provide qualitative to quantitative information about the [22] V. R. Dugyala, M. G. Basavaraj, Langmuir 2014, 30,8680 –8686. [23] V. L. Morales, J.-Y.Parlange, M. Wu, F. J. PØrez-Reche, W. Zhang, W. Sang, interactions occurring in the drop to be analysed. For physico- T. S. Steenhuis, Langmuir 2013, 29,1831 –1840. chemicalinvestigations, this will allow researchers to estimate [24] M. Anyfantakis,Z.Geng, M. Morel, S. Rudiuk, D. Baigl, Langmuir 2015, the strength of interactions in situations in which classical 31,4113–4120. methods becomeinappropriate. For societal and healthcare [25] B. M. Weon, J. H. Je, Phys. Rev.E2010, 82,015305. [26] X. M. Lin, H. M. Jaeger,C.M.Sorensen,K.J.Klabunde, J. Phys. Chem. B applications, we think that analysing the CRE of samples of 2001, 105,3353 –3357. biomedical interest will become amainplayer in the booming [27] S. Narayanan, J. Wang, X. M. Lin, Phys. Rev.Lett. 2004, 93,135503. development of point-of-care diagnostic testing. [28] T. P. Bigioni, X.-M. Lin, T. T. Nguyen, E. I. Corwin, T. A. Witten, H. M. Jaeger, Nat. Mater. 2006, 5,265 –270. [29] P. J. Yunker,T.Still,M.ALohr,A.G.Yodh, Nature 2011, 476,308 –311. [30] M. Anyfantakis,D.Baigl, Angew.Chem.Int. Ed. 2014, 53,14077–14081; Angew.Chem. 2014, 126,14301 –14305. Acknowledgements [31] A. Crivoi, F. Duan, Phys. Rev.E2013, 87,042303. [32] D. M. Kuncicky,O.D.Velev, Langmuir 2008, 24,1371 –1380. This work was supported by the European Research Council (ERC) [33] Y. Li, Z. Zhao, M. L. Lam, W. Liu, P. P. Yeung, C.-C. Chieng, T.-H. Chen, [European Community’sSeventh Framework Programme (FP7/ Sens. Actuators B 2015, 206,56–64. [34] J. R. Trantum, M. L. Baglia,Z.E.Eagleton,R.L.Mernaugh,F.R.Haselton, 2007-2013)/ERC Grant agreement n 258782)] and the Mairie de 8 Lab chip 2014, 14,315–324. Paris [Emergence(s) 2012].M.A. acknowledges fundingfrom the EuropeanCommission (FP7-PEOPLE-2013-IEF/Project 624806 “DI- Manuscript received:May 24, 2015 OPTRA”). FinalArticle published:July 31, 2015

ChemPhysChem 2015, 16,2726 –2734 www.chemphyschem.org 2734 2015 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim