Surface Properties of Colloidal Particles Affect Colloidal Self-Assembly in Evaporating Self-Lubricating Ternary Droplets

In this work, we unravel the role of surface properties of colloidal particles on the formation of supraparticles (clusters of colloidal particles) in a colloidal Ouzo droplet. Self-lubricating colloidal Ouzo droplets are an efficient and simple approach to form supraparticles, overcoming the challenge of the coffee stain effect in situ. Supraparticles are an efficient route to high-performance materials in various fields, from catalysis to carriers for therapeutics. Yet, the role of the surface of colloidal particles in the formation of supraparticles using Ouzo droplets remains unknown. Therefore, we used silica particles as a model system and compared sterically stabilized versus electrostatically stabilized silica particles - positively and negatively charged. Additionally, we studied the effect of hydration. Hydrated negatively charged silica particles and sterically stabilized silica particles form supraparticles. Conversely, dehydrated negatively charged silica particles and positively charged amine-coated particles form flat film-like deposits. Notably, the assembly process is different for all the four types of particles. The surface modifications alter (a) the contact line motion of the Ouzo droplet and (b) the particle-oil and particle-substrate interactions. These alterations modify the particle accumulation at the various interfaces, which ultimately determines the shape of the final deposit. Thus, by modulating the surface properties of the colloidal particles, we can tune the shape of the final deposit, from a spheroidal supraparticle to a flat deposit. In the future, this approach can be used to tailor the supraparticles for applications such as optics and catalysis, where the shape affects the functionality.


Introduction
The assembly of colloidal particles to supraparticles is an emerging strategy to produce functional materials. 1,2 The supraparticles gain additional functionalities compared to the individual colloidal particles, as a result of the structural arrangement of the particles, combinations of different materials, and collective effects. 1 Hence, supraparticles find applications in a wide range of fields such as optics, catalysis, magnetic materials, and drug delivery. [3][4][5][6][7][8][9][10] Evaporation-induced self-assembly (EISA) of colloidal particles in self-lubricating ternary droplets, i.e. colloidal Ouzo droplets, is an efficient route to produce supraparticles. 11,12 Although it is known that the surface properties of colloidal particles can influence their assembly in droplets, 13,14 the role of the particles' surface on the formation of supraparticles in Ouzo droplets is still unknown.
Here we show that the surface properties of the colloidal particles affect and can even impede the formation of supraparticles in colloidal 'Ouzo' droplets.
Ouzo droplets are named after the Greek aperitif Ouzo which contains water, ethanol, and oil, typically anise oil or trans-anethole. 15,16 Colloidal Ouzo droplets are obtained by addition of colloidal particles to the Ouzo mixture. 11 During the evaporation of the Ouzo droplet, the concentration of ethanol decreases faster than that of water. As a result, the oil becomes less soluble, phase-separates, and forms a lubricating oil ring at the contact line. [15][16][17][18][19] This oil ring prevents pinning and leads to the formation of a supraparticle instead of the coffee ring. 11,12 The supraparticle formation can be influenced by parameters such as the size of the colloidal particles 12 or the oil-to-particle ratio 11 in the initial Ouzo mixture. However, what will happen when the surface properties of the particles are varied? Modifying the surface properties of the colloidal particles can alter the interactions of these particles with the various interfaces present in a sessile Ouzo droplet.
Depending on the surface properties, particles can adsorb onto an oil-water interface, stabilizing droplets of a dispersed phase in a continuous phase, forming Pickering emulsions. [20][21][22] In a Pickering emulsion, the stability of the dispersed phase depends on the contact angle of the particle, and thus on the hydrophobicity. Colloidal particles with contact angle close to 90° provide the highest stability to the Pickering emulsions. 23 This stability arises from the high energy required to remove the particles that sit at the liquid-liquid interface with contact angle close to 90° [refer to SI Section S1 for more details]. Furthermore, it is known from the studies on nonevaporating Ouzo mixtures that when the system is in the metastable Ouzo-regime, the oil droplets are generally negatively charged and stabilized by electrostatic repulsion. 24,25 Recently, it was shown that in the Ouzo regime, the addition of nanoparticles results in oil droplets with a narrow size-distribution. 26 Therefore, both hydrophobicity and the electrostatic interactions between the particles and oil can affect the strength of particle-oil interactions and hence the particle-assembly.
The surface properties of the colloidal particles can influence the arrangement of the particles during the evaporation of colloidal droplets even without oil. Particularly for colloidal droplets sitting on a substrate, the attractive particle-substrate interactions can suppress the coffee ring 14, 27, 28 and cause a disordered arrangement of particles on the substrate. 29 Moreover, modifying the hydrophobicity of particles can lead to the adsorption of particles at the liquid-air interface 28 or gelation close to the liquid-air interface, 30 leading to different deposition patterns than the classical coffee stain pattern. 28,[30][31][32] Specifically for supraparticles, the interactions between the particles affect the morphology, internal structure, dispersibility, and porosity of the final supraparticle. 13,[33][34][35] Thus, the surface properties of colloidal particles influence: (i) the interparticle interactions, 13,[33][34][35][36][37] (ii) the interactions between particles and substrate, 14,[27][28][29] and (iii) the behavior of the particles at the liquid-gas interface. 28,30 Therefore, to study the effect of surface properties on supraparticle formation in evaporating colloidal Ouzo droplets, we compared the following colloids: electrostatically stabilized negativelycharged silica particles, electrostatically stabilized positively charged (amine-coated) silica particles, and sterically-stabilized silica particles that were coated with a short poly(ethylene glycol) (PEG). Additionally, as hydration of the particle surface can influence the properties of colloidal particles, in particular the wetting properties, [38][39][40][41] we further compared hydrated and dehydrated silica particles. We observed that the surface modifications of the particles affect their interactions with the oil droplets and the substrate. The hydrated negatively charged silica particles and the PEGylated particles form spheroidal supraparticles. On the contrary, the negatively charged dehydrated particles as well as the positively charged amine-coated particles formed flat film-like deposits. Thus, the surface properties of particles play a crucial role in the formation of supraparticles in self-lubricating ternary droplets, and even suppress the supraparticle formation in some cases. This study provides a guideline to exploit the surface characteristics of the colloids for tuning the properties of supraparticles, particularly in the production of multi-component materials for applications such as drug delivery 3 and optoelectronics. 42,43 Results and discussion

Colloidal particles used to study the surface effects
To study the effect of surface modification of colloidal particles on the formation of supraparticles in self-lubricating droplets, we used different silica particles, as shown in Figure 1.
Negatively-charged silica particles with a diameter of 450 nm were synthesized using the Stöber method, and labeled with rhodamine B-precursor during the polycondensation for fluorescence microscopy ( Figure 1, Table 1). These silica particles were subsequently modified with amine groups or PEG, leading to positively charged or sterically stabilized PEGylated silica particles, respectively ( Figure 1). The particles were dried after the synthesis and purification. 44 To study the effect of hydration in subsequent experiments, the particles were redispersed in water and incubated overnight, leading to hydrated particles. The dehydrated particles were obtained by resuspending the dried particles in absolute ethanol. Thus, all the particles display similar size but differ in their surface properties ( Figure 1, Table 1).
Dynamic light scattering (DLS), transmission electron microscopy (TEM), and zeta potential measurements in aqueous dispersion showed the success of synthesis and modification ( Figure   1, Table 1). All particles had a size of around 450 nm, based on TEM, and a narrow size distribution in the solvents present in the Ouzo-mixture, namely ethanol and water, as demonstrated by low polydispersity indexes (PDI) in DLS. After the PEGylation, the charge of silica particles increases from -58 to -41mV (Table 1). PEGylated particles still display negative charge, as we used a trifunctional silica precursor with a low molecular weight PEG for the surface modification (see Figure 1). Hence, some silanol groups are still present on the surface under the PEG-layer. As a result, the negative zeta potential can still be detected at the slip plane. 45 The amine-modified particles were positively charged (Table 1). After the modification with amine groups, the silica particles remain colloidally stable in ethanol as evident from the DLS results (Table 1). Slight agglomeration is observed upon the transfer into aqueous media due to noncovalent interactions between the particles; similar behavior was reported earlier in the literature. 44 Overall, the measurements (Table 1) show that the particles are colloidally stable in all solvents present in self-lubricating droplets. Purification, Drying (60 °C, vacuum) hydration, particles were resuspended and incubated in water (hydrated particles). Alternatively, particles were resuspended in ethanol prior preparation of the Ouzo-mixtures (dehydrated particles). Transmission electron micrograph of negatively charged silica particles is shown. Scale bar 500 nm.  To explain the differences in the shape of the final deposit obtained from evaporating colloidal Ouzo droplets, we observed the evaporation using top-view imaging and side-view shadowgraphy  Conversely, for dried silica particles, the base length drops suddenly by 0.04 ± 0.02 mm, at t/tw = 0.98 ± 0.01 ( Figure 4b and Figure 3c). This peculiar behavior is more evident in an increase in the height of the droplet by 14 ± 6 % (or 11 ± 3 µm) at t/tw = 0.98 ± 0.01 ( Figure S6c). Finally, after the subsequent evaporation of the oil, a flat film-like deposit is obtained ( Figure 3c). Differently, for droplets containing amine-coated particles the base length shows significant pinning and depinning (close to t/tw ≈ 0.7 and t/tw ≈ 0.9; Figure 4b). The final base length, at t/tw=1, is ≈0.6 mm, which is slightly larger than in all other cases (≈0.4 mm, Figure 4b). Thus, the analysis of the side view images reveals differences in the motion of the contact-line of the droplet, leading to differences in the shape of the final deposit. This behavior indicates that the modifications in surface-properties of the particles affects the self-lubrication.

Observing particle-oil interactions in static Ouzo mixtures
Shadowgraphy raises the question whether the surface modifications affect the interaction of the particles with the oil in the droplet. Therefore, we first accessed whether the colloids interact with the oil in the Ouzo mixture and observed colloidal Ouzo-mixtures with fluorescence confocal microscopy under non-evaporating conditions. For these experiments, we chose a composition of the Ouzo mixture that is expected to correspond to the composition in the evaporating droplet close to the spinodal, where the two-phase region forms. This composition was estimated based on Diddens et al. 46 and the phase diagram by Sitnikova et al. 19 We confirmed that the composition was close to the spinodal by preparation of Ouzo-mixtures without colloidal particles (see methods for the exact composition). Confocal microscopy showed clear differences in the particle distribution around the oil microdroplets for the different colloidal particles ( Figure 5). The surface modification with PEG and amine groups, as well as drying, altered the interactions between the colloidal particles and oil, leading to adsorption of the particles (red) onto the surface of the oil microdroplets (yellow, confocal microscopy Figure 5b-d,f-h). On the contrary, in the mixture with hydrated negatively charged particles, only polydisperse oil droplets were observed due to the spontaneous emulsification and phase separation (Figure 5a; see Table S1 for drop sizes). The particles remained dispersed in the continuous water-rich phase (Figure 5e, Figure S9). The automated analysis of the red florescence signal revealed high signal intensities in the red fluorescence channel around the oil microdroplets, confirming the differences between hydrated non-modified and other particles (see Table S1 and section S4 for further analysis of microscopy data). The particle adsorption increases the stability of the oil droplets, as also recently shown by Goubault et al. 26 in their comprehensive study on the stability and sizedistribution of oil droplets in ouzo systems containing hydrophobic particles. In our experiments, the particle coated oil droplets remain stable for at least one week (see Supplementary Section  Figure S12-S14 for stability and size distribution of the phase separated oil-droplets immediately after preparation and after one week).

Figure 5. Effects of surface modifications on particle-oil interactions. The overlay of fluorescence signals from the particles (red) and oil (yellow) is shown. (a, e) Hydrated silica particles (NP) do not interact with oil microdroplets (MD). In contrast, PEGylation (b, f), dehydration (c, g), and surface modification with amine (d, h) lead to the adsorption of the particles (NP) onto the surface of oil microdroplets (MD). Thus, the surface modifications and dehydration lead to formation of structures similar to Pickering emulsions. The composition of the mixtures was the following (by weight): t-
The particle coated oil droplets in Figure 5 appear similar to Pickering emulsions. Hence, we call these particle-coated oil droplets "Pickering microdroplets" in the following. Note that there is a difference between the formation of conventional Pickering emulsions and the Pickering microdroplets. The droplets in Pickering emulsions are obtained under energy input, using homogenization. 21 In contrast, in Ouzo mixtures, the oil phase-separates spontaneously upon an increase in the concentration of a non-solvent (water) in the mixture, forming oil droplets that get coated with the colloidal particles.
The differences in interactions between the oil droplets and the differently modified colloidal particles are likely caused by the changes in hydrophobicity of particles and electrostatic effects.
Hydrated silica particles are highly hydrophilic and typically do not adsorb onto the oil droplets, as observed in classical Pickering emulsions. 21 PEG is soluble in water and in a broad range of organic solvents, which makes PEG suitable to stabilize particles in aqueous media. [47][48][49] Additionally, PEG can be used as a solvent for t-anethole suggesting attractive interactions between PEG and t-anethole. 47 Based on this affinity of PEG towards both water and t-anethole, it can be expected that PEGylated particles can adsorb onto the surface of the oil microdroplets.
For dehydrated silica particles, the removal of the hydrate water layer can lead to a slight increase in hydrophobicity, and consequently, to the adsorption of particles onto the oil microdroplets. 26,38 It is known from literature that by suspending the particles in short-chain alcohols, the surface of particles can be made more hydrophobic due to the physical adsorption of the alcohol molecules. 21 Furthermore, it was recently shown that certain oils, depending on their solubility in water, can increase the hydrophobicity of the hydrophilic particles in-situ. 50 As a result, the hydrophilic particles can still form Pickering emulsions. 50 Thus, similar to these studies, the adsorption of ethanol and t-anethole on the silica particles could further enhance the hydrophobicity of the particles, enabling them to adsorb onto the oil microdroplets. In the case of amine-coated particles, the amine precursor that contains an alkyl-chain ( Figure 1) can lead to an increase in hydrophobicity, as indicated by a slightly better dispersibility of amine-coated particle in ethanol compared with water (Table 1). Furthermore, the surface of oil droplet is usually negatively charged in an Ouzo mixture. 24,25 Hence, the electrostatic interactions between positively-charged particle and negatively charged interface of the oil droplets can further affect the adsorption. Overall, both the dehydration and the chemical modifications change the interactions between the oil droplets and the particles, and the stability of the oil droplets in the non-evaporating emulsions.

Particle assembly observed through confocal microscopy
As the surface modifications affect the particle-oil interactions, we carried out fluorescence confocal microscopy of the evaporating droplets to further understand the formation of different deposits. Confocal microscopy of the evaporating droplets confirms that modification of the particle surface alters the oil droplets that are nucleated on the substrate ( Figure 6). We further compared the time sequence images obtained from an overlay of the fluorescence emission of silica particles (rhodamine labeled, red) and oil (perylene labeled, yellow) in Figure 7 and Figure   8. Bottom view refers to the images at a horizontal plane close to the substrate, and side view refers to the vertical cross sections that are reconstructed from stacks of horizontal planes ( Figure   7 and Figure 8). These images show the spatial distribution of oil (yellow) and the silica particles (red) over time, as well as the formation of the final deposit. We summarize all the results from confocal microscopy of evaporating droplets as a scheme in Figure 9. In the following, we first discuss the formation of the final deposit for each type of particles separately, and later compare the mechanisms of the deposit formation.  The outer shell has a film like appearance

Side view
Water Air The hydrated silica particles accumulate close to the oil-water and the air-water interface, similar as in our earlier study 12 (Figure 7a and c; summary of the process in Figure 9a). We refer to this particle-rich region close to the interfaces as "outer shell" for simplicity, to distinguish it from the particle shell formed by adsorption of particles on the surface of Pickering microdroplets. Over the course of evaporation, this outer shell keeps on contracting, thereby adjusting its shape to the oil-water and air-water interfaces (Figure 7a

Pickering microdroplets in Ouzo droplet containing dehydrated unmodified silica (e)-(g) Pickering microdroplets observed in an evaporating Ouzo droplet containing dehydrated unmodified silica particles, at t = 0.64tw. (f) and (g) are zoomed in images of the dotted rectangular region in (e). (g) Fluorescence emission signals only from perylene to show oil microdroplets close to the outer shell. Scale bar 50 µm. The evolution of outer shell at the air-water interface and close to the substrate is different for dehydrated unmodified silica particles and amine-coated silica particles. Nevertheless, both particles form flat deposits at the end of the droplet evaporation process.
The dehydrated unmodified silica particles form a film like deposit (Figure 8a and c; summary of the process in Figure 9c).  Figure 8b). This observation suggests that the dehydrated silica particles mainly accumulate close to the oil-water and the glass-water interfaces, while comparatively less particles are present at the air-water interface (as seen in Figure 8a and Figure   8c).
As evaporation proceeds, the outer shell becomes asymmetric and non-circular around t = 0.77 t w, appearing different from hydrated and PEGylated silica particles (Figure 8c-ii). Further, around t ≈ 0.95 tw, the sides of the outer shell merge together becoming a film-like structure (see SI Figure S15). At the same time, the central region gets engulfed by oil (Figure 8a • Amine-coated particles accumulate close to the glass-water interface starting around t ≈ 0.7-0.8 t w (Figure 8d: ii). This accumulation coincides with the pinning of the oil-glass-air contact line in base length, at t/tw ≈ 0.7 (shown by shadowgraphy in Figure 4b). Subsequently, the droplet undergoes consecutive depinning and pinning events until t = tw, (Figure 4b). Such a pinning-depinning cycle is not seen in droplets with dehydrated unmodified silica particles.
• The conversion of the outer shell at the oil-water interface into a film-like structure and the filling of the central region by oil takes place earlier for amine-coated particles, between t ≈ 0.8 tw and t ≈ 0.9 tw, compared to t ≈ 0.95 tw for unmodified dried particles (Figure 8b: ii, Figure   8d: ii,iii, and Figure S15).
• The outer shell at the air-water interface also appears film-like (Figure 8b-ii, Figure S16).
Between t = 0.9 tw and t = tw, this film-like structure at the air-water interface merges with the other film-like region formed at the glass-water interface (Supplementary videos V2 and V3).
• At t = tw, the entire rim of this film is in contact with the glass and only the central region is floating in the oil (Figure 8d: v, Supplementary videos V2 and V3).
Overall, the different surface properties of the particles affect their accumulation at the confining interfaces and consequently affect the motion of these interfaces. These processes determine whether or not a supraparticle or a flat film-like deposit is formed. In the following, we discuss the various mechanisms and physical principles that govern the accumulation of particles at the various interfaces.

Particle accumulation at the air-water, oil-water, and glass water-interfaces
Particles and oil microdroplets accumulate at the interfaces of an evaporating droplet because of the fast movement of the interfaces compared to the diffusion of particles and oil microdroplets away from the interface. 51 The competition between these two process can be characterized suggesting that the flow field influences their motion (see SI section S7 for details). The flow field inside an evaporating Ouzo droplet is not yet completely understood, due to the hindrance in performing velocimetry, as a result of the nucleation and growth of oil-droplets. 46 We expect that a synergetic effect of the flow field and the sedimentation reduces the concentration of the dehydrated unmodified silica particles at the air-water interface and enhances the concentration preferentially at the oil-water interface where the deposit is formed.
In the case of amine-coated particles, the accumulation close to the substrate could occur due to the additional electrostatic interactions. In general, hydrophobic glass surfaces in contact with an aqueous medium show negative charge at neutral and alkaline pH values. 55-59 Therefore, we expect that the hydrophobized glass surfaces used in this study carry a negative surface charge.
Thus, electrostatic interactions between positively-charged particles and negatively charged substrate can lead to accumulation of amine-coated particles close to the substrate (Figure 8b: ii). 14 Overall, modifications in the particles' surface properties lead to differences in particle accumulations at the various interfaces. These differences eventually determine whether flatdeposits or spheroidal supraparticles are formed.

Synopsis of the mechanism: flat-deposits versus supraparticles
During the evaporation of a colloidal Ouzo droplet, high Peclet numbers lead to accumulation of particles at the moving air/water and oil/water interfaces, forming the outer shell. However, hydrated unmodified silica particles mainly did not show any detectable interactions with oil droplets or the substrate. Therefore, the outer shell can keep on contracting until close to the end of the evaporation (Figure 9a: i-iv). The shape of the outer shell is templated by the shape of the  (Figure 9a: v).
The PEGylation of particles leads to adsorption of the particles onto the oil droplets which lie on the substrate (Figure 7e). This adsorption leads to an asymmetric outer shell which appears like a film (Figure 7d). However, this film-like structure keeps on contracting (Figure 9b: i-iv). In droplets containing PEGylated particles, the steric stabilization alters the agglomeration process; in this case, the agglomeration occurs via the entanglement of polymer chains. 60 This difference in the agglomeration process and the absence of an attractive particle-substrate interaction (as Finally we compared the internal structure of the resulting deposits. The supraparticles made using self-lubricating droplets are known to have high porosity. 11 Confocal microscopy confirms the porous structure of the supraparticles made of hydrated unmodified silica particles and PEGylated silica particles, as well as of the flat deposits made of amine-coated silica particles ( Figure S18 a-c, Figure S19, and Figure S20). In contrast, flat deposits made of the dehydrated unmodified silica particles did not show appreciable porous structure ( Figure S18 d-f). Scanning electron microscopy (SEM) images also show differences in packing of the particles at the surface of the various deposits (see SI Section S9 for details). Thus, modifying the surface of colloidal particles affects the internal structure and the particle ordering at the surface of the final deposits.
Further studies can focus on quantifying such differences in the structure and porosity of these deposits and explore ways to tune the internal and surface structure of the deposits, for instance by varying the particle to oil ratio 11 or the pH value of the dispersion. 13

Conclusion
In this work, we have shown the effect of changing the surface characteristics of colloidal particles on the formation of supraparticles in evaporating ternary self-lubricating Ouzo droplets. We compared the behavior of silica particles that are electrostatically stabilized -negatively charged or positively charged by amine-coating -and sterically-stabilized by PEGylation. Moreover, we studied the role of hydration of the particles. We show that these surface modifications alter the contact line motion and the spatial distribution of particles in the droplet. Consequently, particle surface modifications alter the shape of the final particle assembly. Unmodified hydrated particles accumulate as an "outer shell" close to the air-water interface and the oil ring. During the evaporation, this outer shell remains flexible, keeps on contracting, and forms a supraparticle. All  For modification with AHAPS, the ethanolic dispersion of particles was diluted in the same way as for PEGylation, followed by the addition of 4.9 mL of aqueous ammonium hydroxide solution (25%, 4.9 mL). After stirring for 30 min, AHAPS (80 µL) was added, and the reaction mixtures were stirred overnight under Ar at a room temperature. Afterwards, the reaction mixtures were heated under reflux for 2 h at 80 °C. Zeta potential of a test sample was measured in-between to monitor the modification (particles isolated by centrifugation). To complete the modification, AHAPS (0.3 mL) was added to the reaction mixture. The stirring overnight and the heating step were repeated followed by the isolation of nanoparticles by centrifugation (15 min, 16000 g). The particles were purified by centrifugation with ethanol as (3 times, 15 min, 16000 g). After the final centrifugation step the pellet was dried overnight at 65 °C under vacuum.

Characterization of silica particles.
DLS was done with a Malvern Zetasizer Nano series S90 at a scattering angle of 90°. The nanoparticles were diluted with a solvent so that the attenuator was at step 9-11.
The zeta potential was measured with a Malvern Zetasizer Z nano using disposable cuvettes. Transmission electron microscopy (TEM) was done with a JEOL1400 microscope at an acceleration voltage 120 kV. The samples were deposited on carbon-coated copper grids. The analysis of particle sizes was done with measure particles plug in for Fiji. 66 Confocal microscopy of the non-evaporating Ouzo mixtures.

Preparation of Ouzo-mixtures for confocal microscopy in dispersion.
To prepare the Ouzo-mixtures with dried particles, silica particles were redispersed in ethanol using an ultrasonic bath at a concentration of 10 mg mL -1 . The stock solution was used to prepare a mixture of particles and t-anethole in ethanol. To label the oil, 3 µL of perylene-solution in ethanol were added. Subsequently, water was added quickly to induce the phase separation. The vial was closed quickly and vortexed for 1-3 s. The imaging was started 10-30 min after addition of water.
The samples with hydrated particles were prepared as following: silica particles redispersed in water at a concentration of 10 mg mL -1 and incubated in water over night. Subsequently, the dispersions were diluted with water to the respective concentration. In a different vial, a solution of t-anethole in ethanol was prepared and labeled by addition of perylene. Aqueous dispersion of particles was quickly added to the oil to induce the Ouzo-effect or the phase separation followed by vortexing for 1-3 s. The imaging was started 15-30 min after preparation of the samples.
Confocal microscopy (Leica TCS SP5, Wetzlar, Germany) was used to assess the droplet size

Preparation of Ouzo mixture for evaporation experiments
To prepare the colloidal Ouzo mixtures with "dehydrated particles" (Figure 1) for evaporation experiments, the silica particles were put in a glass vial and then ethanol, t-anethole, and water were added to the vial, in this order. The mixture was sonicated for at least 5 min after adding each liquid.
To prepare the colloidal Ouzo mixtures with hydrated particles (Figure 1) for evaporation experiments, the silica particles were put in a glass vial and then water was added. The mixture was sonicated for at least 5 min and then incubated overnight. On the next day, ethanol and tanethole were added to the mixture. The mixture was sonicated for at least 5 min before adding each liquid. The weight fractions of the colloidal Ouzo mixture were as follows: t-anethole / particles / water / ethanol -0.014/0.001/0.453/0.532

Preparation of substrates (glass slides) for evaporation experiments
The glass substrates were hydrophobized by dip-coating, by modifying the procedure mentioned in the literature. 11,12 The glass slides are initially cleaned by mechanical wiping, sonication in solvents, and plasma-treatment. Meanwhile, some water saturated solution of chloroform and toluene (in 2:1 ratio, by volume) is prepared (hereby called water saturated solvent). Thereafter, a solution of chloroform and toluene, in the ratio of 1:4, by volume, is prepared (hereby called main solvent). The water saturated solvent is added to the main solvent, such that volume of the water-saturated solvent is 2% of the volume of the main solvent. Finally, OTS is added to this mixture, such that the volume of OTS is 0.45% of the initial main solvent.
The OTS-solution is mixed properly. Immediately thereafter, the cleaned glass slides are put inside the OTS-solution for hydrophobization. After 30 min, the glass slides are put in a solution of chloroform and sonicated for 10 min, to remove all the unreacted OTS from the glass surface.
After 10 min, the glass slides can also be rubbed by a chloroform wetted cotton swab, if any whitish marks of non-bound OTS are visible. The glass slides are then rinsed with ethanol and water to remove all chloroform from the glass. Finally, the glass slides are blow-dried with nitrogen and stored until they are used. The substrates had an advancing contact angle of 113° ± 3° and a receding contact angle of 106° ± 6°.

Experimental Setup for shadowgraph
Details of experimental set-up can be found in Raju et al. 12 In short, the evaporating droplets were observed from the side using monochrome 8-bit CCD camera and from the top using a CMOS color camera. Both the cameras were connected to a Navitar 12X adjustable zoom lens.
LED light sources were used for illumination. A thermo-hygrometer (OMEGA; HHUSD-RP1) was used to measure the ambient temperature and humidity. The temperature was measured as 21 ± 1 °C and the relative humidity was measured as 60 ± 10%.

Characterization the final deposits
The SEM of the final deposits was done with a Zeiss MERLIN HR-SEM.

Confocal microscopy of evaporating droplets
For performing confocal microscopy on evaporating droplets, Nikon Confocal Microscopes A1 system 667 (with 10× and 20× dry objectives) were used. The bottom view images in Figure 7 and Figure 8 were obtained using the Galvano mode, while the vertical cross-sections were obtained by reconstructing data from successive horizontal planes, captured in resonant mode.
To collect the images from all the horizontal planes and create a single snapshot of a vertical cross-section, it takes ≈ 4 s.

Image processing of evaporating droplets
The side-view shadowgraph images were analyzed using MATLAB-R2020 and FIJI. 66 Details of this image processing can be found in Raju et al. 12 funding from the Canada Research Chairs program. OK acknowledges the funding from Alexander von Humboldt Foundation.

Supporting Information
Supporting information contains the following • Video V2: 2-D vertical cross-section of evaporating colloidal Ouzo droplets using fluorescence confocal microscopy.
• Zip file Z1: Full-size images of oil and particle distribution in non-evaporating Ouzo mixtures observed using confocal microscopy. Smaller cropped regions from the images in Z1 are used in Figure 5 and Figure S9.