Condensation of Satellite Droplets on Lubricant-Cloaked Droplets.

Condensation on lubricant-infused micro- or nanotextured superhydrophobic surfaces exhibits remarkable heat transfer performance owing to the high condensation nucleation density and efficient condensate droplet removal. When a low surface tension lubricant is used, it can spread on the condensed droplet and "cloak" it. Here, we describe a previously unobserved condensation phenomenon of satellite droplet formation on lubricant-cloaked water droplets using environmental scanning electron microscopy. The presence of satellite droplets confirms the cloaking behavior of common lubricants on water such as Krytox oils. More interestingly, we have observed satellite droplets on BMIm ionic liquid-infused surfaces, which is unexpected because BMIm was used in previous reports as a lubricant to eliminate cloaking during water condensation. Our studies reveal that the cloaking of BMIm on water droplets is theoretically favorable due to the fast timescale spreading during initial condensation when compared to the long timescale required for dissolution of the lubricant in water. We utilize a novel characterization approach based on Raman spectroscopy to confirm the existence of cloaking lubricant films on water droplets residing on lubricant-infused surfaces. The selected lubricants include Krytox oil, ionic liquid, and dodecane, which have drastically different surface tensions and polarities. In addition, spreading dynamics of cloaking and noncloaking lubricants on water droplets show that ionic liquid has the capability to mobilize water droplets spontaneously owing to cloaking and its relatively high surface tension. Our studies not only elucidate the physics governing cloaking and satellite droplet condensation phenomena at micro- and macroscales but also reveal a subset of previously unobserved lubricant-water interfacial interactions for a large variety of applications.

Introduction

Condensation is a phenomenon commonly found in many natural and industrial processes, including atmospheric dew formation,[1] power generation,[2,3] natural gas processing,[4,5] water harvesting,[6,7] and refrigeration.[8] Depending on surface tensions of the condensate and surface energy of the surface, two distinct condensation mechanisms can occur, either dropwise or filmwise condensation. Filmwise condensation is much easier to achieve for a variety of working fluids due to the relatively high surface energy of solid surfaces. However, dropwise condensation is preferable for most applications owing to its higher heat transfer performance.[9−11] Therefore, extensive studies have been conducted to encourage or “promote” dropwise condensation by surface modification, including hydrophobic coating[12−14] and micro- or nanotexturing.[14−22] For liquids with a low surface tension (<10 mN/m) such as natural gas and refrigerants, hydrophobic coatings and texturing are not effective due to the nonpolar nature of both the coating and fluid, resulting in film formation. In order to enable dropwise condensation of these low surface tension fluids, researchers have recently developed lubricant-infused surfaces (LISs)[23−27] by taking advantage of the low surface tension of the infusing lubricant. One of the main questions for the condensation on LISs is how the infusing lubricant affects condensation performance. Although the low surface tension of the lubricant helps to achieve dropwise condensation, it also causes problems of cloaking at the condensate–vapor interface.[26,27] The lubricant can spread on the condensate droplet thereby encapsulating it and forming a thin cloaking layer, which adds heat and mass transfer resistance and hinders growth of the droplet.

An LIS consists of a micro- or nanotextured surface that is first functionalized with a hydrophobic chemistry followed by infusion with a thin layer of lubricant oil. For condensation of liquids with relatively high surface tension (i.e., water), LISs reduce the overall thermal resistance during condensation when compared to superhydrophobic surfaces without the lubricant due to the ultralow (<3°) contact angle hysteresis and hemispherical droplet morphologies. The rapid droplet shedding facilitates fresh nucleation sites to be continuously replenished, thereby enhancing the rate of condensation and heat transfer performance. These advantages enable LISs to achieve control of droplet mobility either passively by surface structure modification,[28,29] or actively using electrical or thermal stimuli.[30,31] Therefore, in addition to the applications in condensation, LISs have recently been used as an effective method for droplet manipulation in microfluidic systems.[32] Other applications of the LIS include liquid residual removal,[33] anti-icing, antifouling, and anti-corrosion, which are well summarized in the literature.[34] Understanding of the water–lubricant–solid interfacial interaction helps to enhance the effectiveness and versatility of LISs in these applications. The droplet morphology of a high surface tension liquid (i.e., water) residing on a superhydrophobic surface and LISs is shown in Figure a–c. Apparent contact angles greater than 150° can exist on superhydrophobic surfaces. Due to the existence of air pockets beneath droplets, high probability of droplet pinning exists, resulting in potential for low droplet mobility. Infusing the surface with lubricant oil removes air pockets, creating an atomically smooth and chemically homogeneous liquid–liquid interface, and therefore eliminating pinning. Macroscopically, the elimination of pinning manifests itself in the form of a vanishing contact angle hysteresis. Depending on the spreading coefficient of the lubricant on the water droplet (Sod), two different droplet morphologies can be observed (Figure b,c). The spreading coefficient Sod is defined in eq where γd, γo, and γod represent the surface tension of the condensing droplet, lubricant oil, and interfacial tension between the droplet and oil, respectively. The subscripts d and o represent the droplet and lubricant, respectively. If Sod is negative (Sod < 0), the lubricant will not spread on the droplet; otherwise, the droplet will be cloaked by a thin layer of lubricant (Figure c). Meanwhile, in both cases of cloaking or noncloaking, a lubricant ridge forms at the droplet periphery resulting from the surface force balance at the three-phase contact line. Cloaking of condensate droplets leads to additional thermal resistance, inhibiting droplet growth due to the vapor coming in contact with the lubricant layer instead of the liquid condensate. In the case of condensing fluids with extremely low surface tensions (Figure d–f), superhydrophobic surfaces fail to enable dropwise condensation. Instead, LISs are used to achieve dropwise condensation for liquids having surface tension as low as 20 mN/m.[11,23,26,35] Similarly, lubricant cloaking and satellite droplet condensation may be present depending on the magnitude of Sod.

Figure 1

Schematics representing the water droplet morphology during condensation on (a) superhydrophobic and (b) LIS with no spreading of the oil (green) on the condensate (blue) (Sod < 0) and (c) LIS with spreading of the oil (green) on the condensate (blue) (Sod > 0). Schematics of a low surface tension fluid condensation on a (d) superhydrophobic, (e) LIS with no cloaking (Sod < 0), and (f) LIS with cloaking (Sod > 0).

It is interesting to note that the majority of lubricants in LIS formation cloak water droplets due to their low surface tensions. These include the most commonly used Krytox and silicon oils.[26,36] As a representative noncloaking lubricant, ionic liquid (i.e., BMIm) was proposed as a potential lubricant to eliminate cloaking.[17,27] Ionic liquid is hydrophobic and immiscible with water but considered slightly soluble in water. Water (γ= 72 mN/m at room temperature) saturated with BMIm has a reduced surface tension of 42 mN/m, which leads to a negative spreading coefficient (Sod<0) and indicates that BMIm does not cloak water droplets.[17] However, no direct evidence was provided to confirm whether condensate droplets can be considered saturated with the infusing ionic liquid upon initial condensation. Spreading of the lubricant on the condensate droplet takes place at characteristic timescales approaching 10–9–10–3 s,[24] which means that the lubricant can cloak freshly formed droplets immediately when condensation occurs. At such fast timescales, it is difficult for the condensate droplet to be fully saturated with the lubricant. Consequently, water droplets are cloaked with the ionic liquid prior to reaching saturation and reside on the surface in a metastable state even beyond the saturation timescale.

In this work, we conducted experiments of water vapor condensation on nanotextured superhydrophobic copper surfaces, which were infused with different types of lubricants, including Krytox 1506, BMIm ionic liquid, and carnation oil. For brevity, we term Krytox1506 and BMIm as Krytox and BMIm hereafter. Krytox is one of the most commonly used lubricant in the open literature when studying the condensation behavior of fluids on the LIS, resulting in cloaked water droplet formation due to its positive spreading coefficient. The BMIm lubricant, though used as a noncloaking lubricant in the prior literature,[17,27] has a positive spreading coefficient (Table ). Here, through the use of environmental scanning electron microscopy (ESEM), we demonstrated through independent experiments that BMIm in fact cloaks water droplets. To compare cloaking and noncloaking behaviors, we rationally chose carnation oil and dodecane due to their negative spreading coefficient and low vapor pressure. Raman spectroscopy was used as a characterization technique to confirm the existence and to estimate the thickness of cloaking films for different water–lubricant–solid systems. We analyzed the role of cloaking films during the droplet nucleation by comparing nucleation preference of satellite and host droplets. Spreading dynamics of different lubricants on water droplets were recorded with a microscopic goniometer and analyzed to reveal the key role played by the cloaking film on water droplet manipulation.

Table 1

Properties for Lubricants Used in this Work at Standard Temperature and Pressure

lubricant	liquid–vapor surface tension, γo (mN/m)	liquid density (kg/m3)	dynamic viscosity (mPa·s)	vapor pressure (kPa)	interfacial tension with water, γow (mN/m)	spreading coefficient on water, Sow (mN/m)
Krytox1506	17	1880	113	5 × 10–8	49	6
ionic liquid	34	1430	64	not available	13	29.84
carnation oil	28	810	9.7	1 × 10–2	10.5	–6.25
dodecane	25.35	749.5	1.34	1.8 × 10–2	52.55[41]	–5.1

Methods

Sample Preparation

Three different types of micro/nanostructured surfaces were used in the condensation experiments: nanostructured copper (Cu) and silicon (Si) surfaces and microstructured aluminum (Al) surfaces. To create the copper-based LIS, flat Cu sheets were cleaned with acetone, ethanol, and deionized (DI) water in a sonicator (97043-988, VWR International LLC) sequentially. After cleaning, the native oxide layer on the Cu surface was removed with a 1.0 M hydrochloric acid (HCl) solution. The Cu sheets were then immersed into a well-mixed chemical etching solution (NaOH/NaClO2/Na3PO4·12H2O/DI water = 5:3.75:10:100 wt %) to facilitate growth of a nanostructured copper oxide (CuO) layer.[37,38] Following similar cleaning procedures with Cu sheets, the bare Si wafers were coated with a thin layer of carbon–silica hybrid nanoparticles using chemical vapor deposition (CVD).[39] Carbon nanoparticles were deposited by placing the Si wafers on top of a candle flame for 10 s. Deposition of SiO2 was then conducted by placing the samples in a vacuum desiccator (F42010, Bel-Art) with 5 mL of tetraethyl orthosilicate (TEOS) and 5 mL of ammonia water solution for 2 h. Fabrication of microstructured aluminum surfaces were simply through chemical etching of bare aluminum sheets with a 2.0 M HCl solution.[40] The SEM images of the micro/nanostructured Cu, Si, and Al surfaces are shown in Figure a–c. After the formation of micro/nanostructures, the samples were coated with trichloro(1H,1H,2H,2H-perfluorooctyl)silane (TFTS) by chemical vapor deposition (CVD) in a vacuum desiccator to turn from superhydrophilic to superhydrophobic. Following functionalization, the dry superhydrophobic micro/nanotextured surfaces were spun-coat with a layer of selected lubricants, including Krytox1506, ionic liquid: 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)-imide (BMIm), carnation oil, and dodecane. Properties of the lubricants used in this work are listed in Table . Apparent water contact angles on the superhydrophobic CuO surface and Krytox1506-infused Cu, Si, and Al-based LISs were measured with a goniometer (Kyowa DM501) and are shown in Figure d–g, respectively.

Figure 2

Scanning electron microscopy images of (a) nanotextured Cu, (b) nanotextured Si, and (c) microtextured Al surfaces. Apparent water contact angles on (d) superhydrophobic CuO surface, (e) Cu-based LIS, (f) Si-based LIS, and (g) Al-based LIS. All LISs are created with a Krytox 1506 lubricant. Images (d–g) are not to scale.

Experimental Setup

Condensation experiments were conducted in an environmental scanning electron microscope (ESEM, FEI Quanta 250) at a fixed surface temperature of 2 ± 0.1 °C. A gaseous secondary electron detector (GSED) was used for image acquisition. The sample was attached to a metal stub with silver paste (lot no. 1210728, SPI Supplies) and then placed on the Peltier cooling stage. After pumping the ESEM to a high vacuum mode (∼2 × 10–3 Pa), the scanning mode was shifted to ESEM at a pressure of 100 Pa. The temperature for the cooling stage was reduced to 2 ± 0.1 °C during pump down. Once the temperature stabilized, pressure was increased gradually from 100 to 750 Pa, corresponding to a supersaturation of 1.06. Condensation of water vapor initiated at this pressure. All scans were carried out with an e-beam voltage of 20 kV and a low current of 0.2 nA to minimize heating effects.[42]

Raman spectroscopy was carried out on a WITEC 300s with a 532 nm laser excitation and a spot size of ∼10 μm. The nanotextured CuO, Krytox oil, and ionic liquid were scanned in the range of 200–1100 cm–1, while the Raman shifts for dodecane and water layers were collected in the range of 200–750 cm–1 and 2700–3700 cm–1, respectively.

Results and Discussion

Satellite Droplet Condensation on Different Surfaces

Water condensation experiments were conducted using ESEM to visualize host and satellite droplet morphologies. Once the saturation pressure was reached, host droplets started to condense on the LISs. Condensation of satellite droplets on the lubricant-cloaked host droplets occurred at later stages. As satellite droplets condensed on the cloaking lubricant layer, the presence of satellite droplets was observed to be independent of the structure on the substrate as long as the lubricant could be stabilized on the surface. Figure a–c show ESEM images of water condensation on the Krytox-infused micro/nanostructured surfaces shown in Figure , where satellite droplets were observed for all cases. Satellite droplet formation indicated the presence of the cloaking lubricant layer, which prevents coalescence of satellite droplets with their hosts. Although the existence of satellite droplets was independent of substrate structures, the number of satellite droplets on each host droplet could be affected. As the host droplet sizes and contact angles are different on different substrates, the temperature of each host droplet surface would be different, which thus affected the number of satellite droplets.

Figure 3

Environmental scanning electron microscopy images showing satellite droplets on (a) Cu, (b) Si, and (c) Al-based LIS. (d) Schematic showing condensation of water droplets on an LIS with the infusing lubricant–vapor interface and cloaking lubricant–vapor interface labeled. All LISs are created with a Krytox 1506 lubricant.

To characterize condensation of satellite droplets, differentiation of two lubricant–vapor interfaces are important. These two interfaces are illustrated in Figure d: infusing lubricant–vapor interface and cloaking lubricant–vapor interface, which have different lubricant thicknesses, different shapes (curvatures), and different temperatures. The distribution and number density of the satellite droplets were random and affected by local vapor accommodation. The size of the host droplets ranged from 10 to 100 μm depending on the nucleation time of the droplets. Compared to the host droplets, satellite droplets had much smaller and relatively uniform diameters up to 20 μm. The growth of satellite droplets was limited by the small subcooling at the cloaking lubricant–vapor interface. Growth of host droplets was dominated by coalescence as the cloaking films prevented direct vapor accommodation. When one host droplet coalesced with another, we observed movement of satellite droplets caused by the rearrangement of the cloaking lubricant film. Although some satellite droplets coalesced with the hosts due to thickness fluctuations in the host droplet cloaking film, majority of satellite droplets stayed uncoalesced on the hosts as the cloaking film did not drain from the interface. In order to ensure that the observed satellite droplet formation in ESEM was not governed by interaction between the electron beam and the lubricant (i.e., polymerization, scission of the lubricant molecules, etc.), we conducted optical microscopy experiments for water condensation on the Krytox-infused CuO surface. The experimental setup and procedures are briefly demonstrated in Section S1 of the Supporting Information. Figure S1 shows the top view of the condensed droplets. Presence of satellite droplets was observed on almost all the visible host droplets. In the next section, we use nanostructured copper plates as the substrate to compare different droplet morphologies for condensation on different lubricants.

Cloaking versus Noncloaking Lubricants

The spreading coefficient is the key parameter currently used to justify whether a lubricant will cloak a condensate droplet.[23,26,27] As listed in Table , Krytox 1506 oil will cloak water as Sow > 0, while carnation oil and dodecane are noncloaking due to Sow < 0. For BMIm, the Sow is dependent on the saturation level of BMIm in water. In the case when water droplets are fully saturated with BMIm, the surface tension of water is reduced to 42 mN/m, which results in Sow = −5, indicating the nonspreading of BMIm on the water droplet.[17] However, it is difficult for BMIm to reach full saturation in water during droplet condensation, which is a transient process. Due to the short timescale of lubricant spreading (10–9–10–3 s),[24] cloaking occurs much faster than dissolution. Therefore, BMIm cloaks the condensed droplets immediately after condensation. To confirm the large timescale of dissolution between BMIm and water, we measured the interfacial tension between the two liquids as a function of time. The details of the experiments are included in Section S2 of the Supporting Information. The constant interfacial tension over time (Figure S3) indicated that the timescale of dissolution is much longer than the timescale for lubricant spreading. In other words, interfacial dissolution is negligible during our condensation experiments, and the cloaking layer would not be eliminated by dissolution. Without dissolution, a positive spreading coefficient Sow was obtained by substituting the liquid–vapor surface tension of pure water, BMIm, and their interfacial tension into eq or using the vOCG method considering the polarity of the liquids. Details for the vOCG calculation are included in Section S2 of the Supporting Information. The vOCG prediction shows that BMIm is slightly polar, having a polar surface tension component of 9 mN/m and a nonpolar component of 25 mN/m, which in combination resulted in a positive spreading coefficient of 29.8 mN/m (Table ).

Images of water condensation on surfaces infused with Krytox, BMIm, and carnation oil are shown in Figure , including side (Figure a–c) and top (Figure d–f) views. Cloaking of the lubricant on the water droplet in the cases of Krytox and BMIm led to the condensation of satellite droplets, indicating that the host droplets are cloaked. Meanwhile, the presence of satellite droplets on the cloaked lubricant layer indicates additional nucleation sites for vapor to condense. Despite the reduction in heat transfer performance induced by the additional thermal resistance that the cloaking lubricant layer generates, the formation of satellite droplets suggests larger surface area for the vapor to condense. This provides an opportunity for enhancing the condensation rate and can in turn lead to significant heat transfer gains. Comparing Figure a,b, the differences between the size and density of satellite droplets could be attributed to the size and contact angle of the host droplets on the substrate due to the different infusing lubricants. Another reason is that droplet sizes were different when captured at different time. In Figure c and f, as expected, the surface infused with carnation oil showed no satellite droplets on the condensed water droplets.

Figure 4

Side-view ESEM images of water vapor condensation on surfaces infused with (a) Krytox 1506 oil, (b) BMIm ionic liquid, and (c) carnation oil. Top-view ESEM images of water vapor condensation on surfaces infused with (d) Krytox 1506 oil, (e) BMIm ionic liquid, and (f) carnation oil. The scale bar is the same for all images.

In order to gain an understanding of the chemistry of the lubricant films, we used Raman spectroscopy on the LIS samples prior to and post condensation. Although Raman spectroscopy has become a powerful analytical technique, relatively little effort has been placed on employing Raman spectroscopy for liquid analysis, particularly in separations including chromatography and capillary electrophoresis, due to its poor detectability. Recent instrumentation developments in laser technology, detectors, microscopes, fiber optics, and holographic filters[43] have enabled the use of Raman on liquid samples.[44,45] Note that even in aqueous solutions, Raman spectra can provide detailed vibrational information for identification and confirmation purposes.[46,47] We used the characteristic vibrational information of lubricants to confirm the presence and estimate the thickness of the thin cloaking films on water droplets. We performed z-scanning experiments, assuming the z-axis is normal to the sample surface, for a water droplet sitting on nanostructured surfaces, which were infused with Krytox, BMIm, and dodecane, respectively. The schematics of the sample surface used for collecting Raman spectra and the corresponding Raman shifts at various z positions are presented in Figure .

Figure 5

Raman spectrum characterization for water droplets placed on Cu-based LISs. (a) Schematic of the measurement steps: (I) bare CuO nanotextured superhydrophobic surface prior to infusing lubricant, (II) LIS infused with lubricant, and (III) one DI water droplet dispensed on the LIS. The size of the dispensed DI water droplet was 20 μL. The numbers represent positions of the Raman focal plane during measurement: (1) within the substrate, (2) within the infused lubricant layer, (3) within the water droplet, and (4) within the cloaking lubricant layer (or water–vapor interface for the case of noncloaking lubricant). Measured Raman spectra for water droplets sitting on the LIS infused with (b) Krytox, (c) BMIm, and (d) dodecane.

Figure a shows a schematic for the recording steps of Raman spectrum: (I) bare CuO nanotextured superhydrophobic surface prior to infusing lubricant, (II) lubricant added to the surface, and (III) one DI water droplet dispensed on the LIS. Here, in order to detect the infusing and cloaking lubricant layers via Raman, the infusing lubricants spread on the solid surface without spinning and remained thick. For each experiment, a Raman spectrum was first recorded for the dry substrate surface, which in our case was a hydrophobic silane-coated nanotextured Cu surface. This spectrum is marked as position 1, and the recorded signals are shown as black lines in Figure b–d. After infusing the nanotextured Cu surface with the lubricant, the Raman spectrum was recorded by z-scanning starting from the solid surface to the top of the infused lubricant film. The top is marked as position 2, and the recorded signals are denoted by red lines in Figure b–d. By utilizing z-scanning of the infused lubricant film in the oscilloscope mode, we reached the position (+z) where the characteristic signal for the infusing lubricant started to disappear. At this stage, a water droplet was dispensed onto the substrate surface. A Raman spectrum was recorded again by moving the focal plane of the laser into the water droplet. This position is marked as position 3, and the recorded signals are denoted by green lines in Figure b–d. Again, by z-scanning using the oscilloscope mode, we reached the point where the signal of the Raman shift for water began to disappear, marked as position 4. At this position, by performing z-scanning in the range of ±5 μm with a step size of 500 nm, we collected the Raman signals for the thin cloaking films, which are shown as blue lines in Figure b–d. For Krytox oil and BMIm lubricants, we recorded the characteristic Raman shifts at position 4 successfully, which confirmed the formation of a thin cloaking film over the water droplets, as shown in Figure b,c, respectively. However, for the dodecane lubricant, no Raman shift was observed at position 4, indicating the absence of a cloaking film, as shown in Figure d.

The characteristic Raman shifts for CuO substrates in all experiments were in close agreement with the prior literature.[48] The three Raman active modes Ag, Bg(1), and Bg(2) gave peaks at 286, 330, and 620 cm–1, respectively. The signal at position 3 is for the water droplet, which has a Raman shift peak at a larger range (>3000 cm–1), compared to the peaks for the solid substrate and lubricants, as shown in Figure c,d. The shoulder at 3250 cm–1 corresponds to the asymmetric −OH stretch, and the most intense feature at 3400 cm–1 corresponds to the symmetric −OH stretch.[45] Due to the much larger Raman shift range of water, the signal for the water droplet at position 3 is not shown in Figure b. This was done to highlight the consistent signal peaks for the infusing (red line) and cloaking (blue line) Krytox oil. The Raman spectrum of pure Krytox oil contained intense peaks at 843 and 903 cm–1 corresponding to the stretching of peroxides and CF3 bonds in the fluoropropylene oxide groups of the perfluoro-poly(ether) backbone.[49,50] The peaks at 292 and 412 cm–1 represent the torsional and deformational vibrations of the CF2 bond.[51,52] The intensity of the cloaking film is much weaker than that of the infusing film, indicating the small thickness of the cloaking lubricant film. Similarly, in Figure c, consistent absorption bands for BMIm infusing and cloaking films were recorded at 740 and 1021 cm–1, respectively.[53] Although the signal intensity for the cloaking film was weak due to its smaller thickness compared to the lubricating film at the base of the droplet, the existence of the cloaking film was confirmed, in agreement with the ESEM experiments (Figure b,e). In the case of dodecane (Figure d), only the water signal was detected for position 4, indicating that the water droplet was not cloaked by the lubricant. The characteristic absorption bands for dodecane at 1081, 1300, and 1488 cm–1 agree with those recorded in prior studies.[54] For detailed procedures for signal recording of Krytox oil, and individual signals for the substrate, Krytox oil, BMIm, dodecane, and water, please see Section S3 and Figures S4 and S5 of the Supporting Information, respectively.

To take full advantage of Raman spectroscopy, we also estimated the thickness of the cloaking film by comparing signal intensity of the infusing and cloaking films. The thickness of the infusing film can be estimated from Figure S4a. The Raman signal for position 2 (infusing lubricant) appears at the focal depth of 30 μm and disappears at 107.5 μm, indicating a film thickness on the order of 77.5 μm. As this thickness is large enough to cover the whole focal depth, the maximum value of signal intensity at position 2 corresponds to the thickness of the focal plane depth. Therefore, we first calibrate the system to obtain the thickness of the focal plane, which is found to be approximately 5 μm (Figure S6a, Supporting Information). For details of the focal depth calibration, please see Section S3 of the Supporting Information. During z-scanning at position 4, the focal plane is moved to the interface of the water/cloaking oil film, where the oil film is too thin to fully cover the focal depth, resulting to a weaker signal. As the ratio of the signal intensity is equivalent to the ratio of the lubricant amount detected at these two positions, we can estimate the thickness of the cloaking oil film (δc) from the calibrated focal depth. Mathematically, it can be expressed as δc ≈ (I4/I2)δf, where δf is the thickness of the focal plane, and I4 and I2 are the signal intensities at positions 4 and 2, respectively. The signal intensity at positions 4 and 2 for Krytox and BMIm lubricants are compared to Figure S6b,c. For the Krytox lubricant, the two peaks at 304 cm–1 have maximum signal intensity of 29 and 3 for infusing and cloaking films, respectively. Therefore, the calculated cloaking film thickness is approximately 510 nm. Comparison of the peaks at 804 cm–1 gives a similar result. For the BMIm lubricant, the calculated cloaking oil thickness was approximately 1.18 μm. The BMIm-cloaked film thickness was higher than the Krytox cloaking thickness due to the light polarity of the BMIm lubricant (Section S2, Supporting Information) and strong polar interaction at water–BMIm interface.

The thickness of the cloaking film is dictated to be a balance between the interfacial spreading force and the disjoining pressure gradient. Accordingly, the thickness of FC70 cloaking films on a water droplet having a radius of 1 mm was estimated to be 20 nm.[55] Cross-sectional imaging of a silicon–oil-cloaked water droplet obtained by a cryogenic focused ion beam (FIB) SEM technique showed a cloaking layer thickness of approximately 65 nm.[24] Although the thickness may depend on the droplet size, lubricant properties, and lubricant–water interfacial interaction, the cloaking film thicknesses estimated here via the Raman technique are larger than the previously reported data. One possible reason is that the infusing lubricant layer has thickness of approximately 77.5 μm, indicating a large excess of lubricant oil on top of the CuO nanotextures. The lubricant excess may result in an increased thickness of the cloaking film. Due to limitations of the Raman spectroscopy instrument, lubricant films thinner than 500 nm could not be detected.

Nucleation Density of Satellite and Host Droplets

The presence of the cloaking film results in condensation of the satellite droplets on the host water droplets. Nucleation of satellite droplets does not happen immediately after host droplets are cloaked with the lubricant. Instead, condensation occurs preferentially on the lubricant-infused surface rather than the cloaked droplet surface. Figure shows the dynamics for water condensation and evaporation on the nanotextured Cu surfaces infused with Krytox 1506. Recording starts at the moment when only a few droplets condensed on the surface without any satellite droplets formation (t = 0 s). Images were taken every 60 s. Initially, although the droplets were cloaked with the lubricant, no satellite droplets were observed (Figure a). Growth of the host droplets were mainly governed by two mechanisms: (1) direct diffusion of vapor molecules through the cloaking films due to relatively large subcooling and (2) coalescence with adjacent droplets. The lack of satellite droplets at the initial stage is because the LIS has a lower temperature and thus lower nucleation energy barrier when compared to the cloaked water droplet surface. Therefore, droplets prefer to condense on empty sites between droplets on the LIS instead of on the droplets themselves (t = 60 and 120 s). We estimated the LIS temperature using an analytical thermal resistance model,[12,56] while the nucleation energy barrier was calculated using classical nucleation theory[10,57] (see Section S4 and S5, Supporting Information). Once the solid surface was fully covered by the host water droplets (t = 180 s), nucleation of satellite droplets on the cloaked host droplets initiated. After nucleation, the satellite droplets grew by direct vapor diffusion at early stages and coalescence at the later stage as satellite droplets were cloaked by the lubricant as well. The relatively small size of the satellite droplets was due to the small subcooling temperature on the host droplet surface. During evaporation (Figure b), satellite droplets disappeared immediately when the chamber pressure was reduced to 700 Pa, while the host droplets evaporated gradually. In fact, the evaporation rate was very slow due to the existence of the cloaking lubricant layer, which limited mass diffusion of water molecules.

Figure 6

Satellite droplet nucleation on a Krytox LIS. Time-lapse side-view ESEM images of water vapor (a) condensation at Pv = 740 Pa and (b) evaporation at Pv = 700 Pa. (c) Schematic of the nucleation process on an LIS. (d) Predicted temperatures of different interfaces with varying water droplet diameter D. (e) Calculated nucleation energy barrier to condensation on the LIS as a function of the surface temperature for a surface infused with Krytox (blue line) and BMIm (red dotted line).

Figure c illustrates the condensation nucleation and droplet growth process with a simplified schematic. The preferential initial nucleation on the LIS (the infusing lubricant–vapor interface) occurs due to the lower temperature of the infusing lubricant–vapor interface when compared to the cloaking lubricant–vapor interface of existing droplets. Based on the analytical thermal resistance model, the temperature of different interfaces with varying water droplet diameter D is predicted and plotted in Figure d, where Tb, Tif, Ti,1, and Ti,2 represent the droplet base, infusing lubricant–vapor interface, cloaking lubricant–vapor interface, and cloaking lubricant–droplet interface temperatures, respectively. The temperature of the infusing lubricant–vapor interface is always lower than that on the cloaking lubricant–vapor interface. Furthermore, the normalized nucleation energy barrier (W/kBT) as a function of temperature is shown in Figure e. As the nucleation energy barrier increases with temperature, nucleation of host droplets is faster and more preferential than that of the satellite droplets. The theoretical results agree well with our experimental observations shown in Figure a.

Role of a Cloaking Film: Spontaneous Droplet Mobilization in a Cloaking Lubricant

To show the role of the cloaking film at a macroscopic scale, we recorded the spreading of cloaking and noncloaking lubricants around a water droplet with a pendant drop goniometer. A lubricant droplet (Krytox oil, BMIm, or dodecane) was first deposited on a flat Si wafer surface coated with 1H,2H,2H,2H-perfluorodecyl-triethoxysilane. Due to the low surface tension of Krytox, it almost completely spread on the hydrophobic surface (t = 0, Figure a), while BMIm and dodecane made similar contact angles of 30 ± 5° (t = 0, Figure b,c). A water droplet was dispensed adjacent to the lubricant having an advancing contact angle of 115 ± 3°. By increasing the water droplet size gradually, the water droplet contacted the lubricant and an oil ridge was formed immediately due to the spreading of lubricant oil. Figure a–c shows the initial and steady states of the water–lubricant system. Detailed spreading dynamics of the lubricants on the water droplets are shown in Figure S8 of the Supporting Information. The recorded spreading and stabilization process for BMIm is very different from that for the other two lubricants. As shown in Figure b, the water droplet was pulled into the BMIm film and eventually stabilized at the center of the BMIm film. In the cases of Krytox and dodecane, the water droplets stabilized at the peripheries of the lubricants. As dodecane is a noncloaking lubricant, a small portion of the water droplet was not covered by the lubricant, although a thick oil ridge was formed due to the large size of the lubricant oil film. Figure d,e illustrates the stabilization mechanism and processes governing cloaking in the BMIm–water and dodecane–water systems. As shown in Figure d(i), the curvature difference between the water droplet top and lubricant film top creates a positive pressure gradient, which pushes the water droplet towards center of the lubricant film until an equalized curvature is reached (Figure d(ii)). For noncloaking dodecane, a three-phase contact line is formed on the water droplet (green dashed line in Figure e). A force balance is achieved at the contact line between the surface tension of water, surface tension of lubricant, and interfacial tension between the two. This force balance results in the noncenter-symmetric stabilization of the system as shown in Figure e. The observation of center-symmetric stabilization in the BMIm–water system is due to unbalanced interfacial forces, which leads to spreading of the cloaking layer instead of formation of a distinct three-phase contact line.

Figure 7

Spreading of cloaking and noncloaking lubricants on water droplets. Side-view images of initial and stabilized status of water droplets (left) adjacent to (a) Krytox oil, (b) BMIm, and (c) dodecane lubricant films residing on a hydrophobic polished Si wafer. Schematic illustrations of the stabilization processes of a water droplet (blue) in (d) BMIm and (e) dodecane lubricant film. Schematics are not to scale.

Different from the immediate mobilization of the water droplet from periphery to the center of the BMIm film, the water droplet cloaked by Krytox oil was not mobile, even after 600 s of observation, as shown in Figure a. Therefore, cloaking is not the only dominant factor for spontaneous water droplet mobilization. As Krytox oil almost completely spread on the solid surface due to its low surface tension, the curvature difference does not exist in the cloaking lubricant layer. Therefore, the Krytox oil-cloaked water droplet was immobile due to the lack of driving force.

To sum up, mobilization of water droplets in the lubricant can be achieved with two determinant factors: (1) the lubricant is cloaking, and (2) the lubricant has sufficient surface tension to maintain a cap shape with a certain curvature on the solid surface. Among the selected lubricants, the ionic liquid BMIm is the only one satisfying both conditions. These results provide the insights for droplet manipulation in order to control the dynamic behavior of droplets in multifluid systems. The directional movement of water droplets in the immiscible lubricant has practical applications in chemical reactions, drug delivery, and lab-on-chips.[58−60] Spontaneous mobilization of water droplets in BMIm liquid shows the potential of involving such lubricant in these applications.

Conclusions

Condensation of satellite droplets on lubricant-cloaked water droplets is studied in this work. To show the presence of satellite droplets, ESEM experiments were first conducted on surfaces infused with different types of lubricants, including Krytox oil, BMIm, and carnation oil. BMIm was regarded as a noncloaking lubricant in the literature as it is slightly soluble in water, which makes a negative spreading coefficient. However, we observed satellite droplets from our condensation experiments, which implied the cloaking of ionic liquid on the water droplet. This is basically because cloaking happens immediately upon condensation, while dissolution takes much more time. Further, we conducted Raman spectroscopic characterization, which directly confirmed the existence of the cloaking lubricant layer for Krytox and BMIm and nonexistence for dodecane. Moreover, we estimated the cloaking layer thickness by comparing the intensity of the Raman shifts for the infusing film and cloaking film. These results present a unique advantage of micro-Raman spectroscopy in collecting the depth profiles of heterogeneous systems. We then studied the roles of cloaking films both in the condensation and mobilization of macrodroplets. The predicted temperature and nucleation energy barrier on different interfaces showed that condensation on the solid substrate surface was more preferential than that on the lubricant-cloaked droplets. The presence of satellite droplets on the cloaked lubricant layer indicates additional nucleation sites for vapor to condense. This provides an opportunity for enhancing the condensation rate and can in turn lead to significant heat transfer gains. In a macroscopic scale, the existence of the cloaking film played an important role in mobilizing water droplets without any external forces. Our work confirmed the presence of satellite droplets, characterized their formations, and analyzed different roles of the cloaking film with different tools. We believe that the characterization of Raman vibrational modes combined with environmental microscopy can provide us a facile way to directly observe and quantify the in situ growth and formation of condensate droplets on solid or lubricant-infused surfaces.