International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 Contents lists available at SciVerse ScienceDirect International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt Microdroplet evaporation on superheated surfaces Shawn A. Putnam a,b,⇑, Alejandro M. Briones c,d, Larry W. Byrd c, Jamie S. Ervin c,d, Michael S. Hanchak c,d, Ashley White c,d, John G. Jones a a Air Force Research Laboratory, Materials and Manufacturing Directorate, Thermal Sciences and Materials Branch, Wright-Patterson AFB, OH 45433, United States Universal Technology Corporation, 1270 N. Fairfield Rd., Dayton, OH 45432, United States c Air Force Research Laboratory, Propulsion Directorate, Thermal and Electrochemical Branch, Wright-Patterson AFB, OH 45433, United States d University of Dayton Research Institute, University of Dayton, Dayton, OH 45410, United States b a r t i c l e i n f o Article history: Received 13 December 2011 Received in revised form 21 May 2012 Accepted 22 May 2012 Available online 30 June 2012 Keywords: Evaporation Water Microdroplet Numerical simulation Structured surfaces Depinning a b s t r a c t A comprehensive experimental investigation on water microdroplet evaporation is presented from the standpoint of phase-change cooling technologies. The study investigates microdroplet evaporation on a variety of different surface materials at surface temperatures ranging from 25 °C to 250 °C. The temporal evolution of the droplet profile and contact line dynamics are measured using high-speed photography and image analysis methods. The material systems tested consist of (1) thin-film surface coatings on glass substrates (e.g., Al, Ti, Cu, and SAMs) and (2) Cu substrates with different surface morphologies (e.g., mirror polished, 800-grit sanded, and micron-sized Cu pillar arrays). As expected, changes in surface energy influences the contact line dynamics. Changes in surface energy, however, showed no systematic influ_ LG Þ scales linearly with ence on the evaporation efficiency. For all systems studied, the evaporation rate ðm _ LG / R) and stick-slip contact line dynamics are observed. For the microdroplets contact radius (i.e., m example, the evaporation efficiency reduces after contact line depinning due to a reduction in the total _ LG is a constant and directly length of the solid-liquid-vapor contact line, whereas, before depinning, m proportional to the surface temperature, contact radius, and substrate thermal conductivity. For micro_ max droplet evaporation on thin-film surfaces, maximum evaporation rates ðm LG ffi 32 8 lg/s) and evaporative heat fluxes (qmax ffi 600 ± 100 W/cm2) are observed at superheats of 70 °C [ DT [ 130 °C. These maxima in the evaporative heat transfer performance signify a transition in the heat transfer process from a purely microdroplet evaporation regime to a droplet/film boiling regime (which is analogous to the critical heat flux observed in pool boiling). For microdroplet evaporation on Cu substrates, droplet/film boiling occurs at much lower superheats (e.g., 10 °C [ DT [ 25 °C); yet, comparable maximum evapora2 _ max tion rates (m LG ffi 36 8 lg/s) and evaporative heat fluxes (qmax ffi 750 ± 150 W/cm ) are observed. In _ max ffi 35 5 lg/s during short, this work suggests a reliable upper limit for the evaporation efficiency of m LG water microdroplet evaporation on superheated surfaces (which turns out to be independent of substrate thermal conductivity, surface structure, and surface hydrophobicity). Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction Many of the today’s cooling technologies have stringent requirements to accomplish one of two challenges: (1) reliably avoid critical heat flux (CHF) while efficiently dissipating enormous amounts of heat from very small areas [1] (e.g., heat fluxes beyond 2000 W/cm2) or (2) regulate system temperatures within sometimes fractions of a degree over a broad range of operation loads/ conditions (e.g., active, not passive, temperature control is imperative during operation because of irregularities in the environment and changing power loads). These requirements are especially stringent for the new generation of high power electronics, lasers, ⇑ Corresponding author. E-mail address: [email protected] (S.A. Putnam). 0017-9310/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2012.05.076 propulsion, and energy conversion systems, where the cooling efficiency is dictated by both the cooling configuration and the coolant itself. To date, spray cooling with water is arguably the most effective cooling process due to its ability to: (1) uniformly remove large heat fluxes, (2) use small fluid volumes, (3) take advantage of large heat of vaporization, (4) use low droplet impact velocity, and (5) provide optimal control and regulation of system temperatures[2]. Not only the liquid coolant, but also the surface properties and surface-liquid interactions are important. This is especially true at micro- and nano-scale dimensions, where interactions at liquid interfaces can have a dramatic influence on the macroscopic properties of the system, dictating physical properties like the fluid viscosity, thermal conductivity, and surface tension. Over the past several years, several groups have shown that micro- and nano- 5794 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 Nomenclature A cp Dh2 oair D F G h h kB LV Mw _ LG m NA p P q r R RH t T * v V We z surface or contact area (m2) specific heat capacity at constant pressure (J/kg/K) binary diffusivity of water vapor to air (m2/s) droplet diameter & contact diameter (lm) surface tension force per unit length of contact line (mN/m) interfacial Gibbs free energy (J) heat transfer coefficient (kW/m2/K) droplet height at centerline (lm) Boltzmann’s constant (1.3806503 10-23 m2 kg/s2/K) latent heat of vaporization for water (2.26 106 J/kg) molecular weight (g/mol) evaporation rate or liquid-to-gas mass transfer rate (lg/ s) Avogadro’s number (6.022137 106 J/kg) gauge pressure (Pa) pressure (Pa) heat flux (W/cm2) radial direction at centerline along the surface (lm) droplet radius & contact radius (lm) relative humidity (%) time & droplet lifetime (s) temperature (K or °C) velocity (m/s) droplet volume (pL) Weber number z-direction at centerline (lm) Acronyms CHF critical heat flux (W/cm2) FS fused silica VOF volume of fluid Superscript exp experiment max maximum sim simulation structures can enhance the overall heat transfer coefficient in pool boiling [3–5], flow boiling [6,7], spray cooling [8–10], heat pipes, and post-CHF film boiling [11]. Nanowire-like structures have also been predicted to enhance the long-range van der Waals interactions at solid interfaces [12,13], which, in turn, dictate the formation of specific liquid-wetting [14] and gas-spreading [15] states at a liquid-solid interface. However, a cogent explanation of the operative mechanisms that cause enhancements in heat transfer performance has not been established. Insight into the fundamental mechanisms responsible for wetting, fluid dynamics, and two-phase heat transfer can be obtained from single droplet studies. In this regard, single droplet studies are related to all mainstream multi-phase cooling processes (e.g., evaporation, spray and jet impingement cooling, flow and pool boiling, heat pipes, and thermosyphons). Thermal transport during single droplet impingement and evaporation on a heated surface is, however, a very complex phenomenon involving a multitude of physical-chemical events such deposition, spreading, receding, rebound, jetting, splash, contact line pinning, depinning, fluid and vapor convection, air and vapor bubble entrapment, and vapor bubble nucleation. There is a relatively vast literature on these topics, and we refer the reader to Refs. [16–30] and the citations therein. This work focuses on surface or interface cooling during evaporation with small, water microdroplets – e.g., microdroplets with Greek symbols |raG| droplet interfacial surface area density (m2) evaporation ratio bLG DT superheat (K) dF depinning force (mN/m) dG excess free energy (J) dGl excess free energy per unit length of contact line (J/m) dGA excess free energy per unit solid-liquid surface area (J/ m2) dT Tolman length (nm) e accommodation coefficient or evaporation efficiency c surface tension (mN/m) or surface energy density (J/m2) j curvature (m-1) k mean-free-path (m) K thermal conductivity (W/m/K) l dynamic viscosity (kg/m/s) q density (g/cm3 or kg/m3) p_ LG evaporation rate per unit length of contact line (kg/s/m) h contact angle (deg) sD depinning time (s) sE total evaporation time or droplet lifetime (s) / volume fraction w mass fraction Subscripts 0 initial condition & contact condition air air/gas h2o water D depinning im impingement Eq equilibrium G gaseous-phase L liquid-phase LF Leidenfrost S surface & solid-phase SC spherical cap V vapor-phase initial volumes (radii) less than 2500 pL (85 lm). To minimize the role of surface structure, microdroplets evaporating on smooth, metal thin-film surfaces are studied. Evaporation studies on microstructured and micro-roughened surfaces are also conducted for comparison. We find, for these small water droplets, that the key mechanisms/properties influencing the heat transfer performance are the droplet contact line dynamics (i.e., pinning and depinning), substrate thermal conductivity, and air/vapor bubble entrapment and nucleation. The goal of this work is to characterize how the interfacial heat transfer performance is coupled to the contact line dynamics during microdroplet evaporation on surfaces heated up to the Leidenfrost temperature. With this said, the aim is to improve single droplet impact dynamics and evaporation models for later enhancement of spray cooling models. To accomplish this, we combine numerical simulations with high-speed photography measurements. High-speed photography facilitates characterization of droplet wetting behavior, which is coupled to the evaporation efficiency. The experiments and simulations are conducted in parallel with small water microdroplets to facilitate later advancements in modeling methods in a reasonable time frame. The remainder of the manuscript is organized as follows: In Section 2, the experimental apparatus, measurement technique, and data analysis methods are described. Section 3 presents the S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 numerical model used to simulate microdroplet evaporation on an isothermal surface. Section 4.1 covers, in detail, microdroplet evaporation on Al thin-films with surface temperatures up to 250 °C. Section 4.2 overviews the heat transfer performance in terms of the heat flux and evaporation rate for a variety of different thinfilm surfaces (e.g., Al, Cu, Ti, and SAM thin-film coatings on glass substrates). Section 4.3 overviews the heat transfer performance during microdroplet evaporation on solid Cu substrates with different surface structures. In Section 4.4, the role of contact line depinning is discussed in terms of its influence on the evaporation efficiency, where calculations of the depinning force and free energy barriers for contact line depinning are made for all the material systems tested. The conclusions of this work are provided in Section 5. 2. Experimental methods 2.1. Materials and sample fabrication methods The materials systems studied in this work consists of micronsized water droplets in contact with (1) Al thin-films, (2) Ti thin-films, (3) Cu thin-films, (4) self-assembled monolayer (SAM) coatings, (5) polished Cu substrates, and (6) micro-structured Cu substrates. For all experiments, deionized (DI) water purified and filtered with a Barnstead Nanopure Water Purification System is used. All metal thin films are deposited by magnetron sputtering on flat fused silica (FS) glass windows [31]. The metal thin-film thicknesses ranged between 60 and 80 nm with surface roughness (Ra) less than Ra [ 0.01 lm. The polished Cu substrates are 2 mm thick oxygen-free Cu discs (1’’ in diameter) with surface roughness of either Ra ffi 0.05 lm (800-grit finish) or Ra [ 0.01 lm (mirror finish). The micro-structured Cu sample [32] consists of a simple-cubic array of circular-pillars separated by 20 lm with a height and diameter of 20 lm and 8.5 lm, respectively. A transmission-electron-micrograph (TEM) of the pillared Cu surface is provided in Fig. 13. The SAM surface consists of an octadecyl-trimethoxysilane monolayer chemically grafted to an Al thin-film on a glass substrate. The SAM surface is hydrophobic. For preparation, an Al/glass sample is immersed in a toluene solution containing 0.01 M octadecyl-trimethoxysilane (SAM layer) and 0.01 M triethylamine (catalyst) and then left overnight to self-assemble. The silane reacted and covalently bonded to hydroxyls on the Al surface while the 18-carbon chains drove self assembly and formed the hydrophobic SAM layer. 2.2. Microdroplet impingement and evaporation measurements Data is acquired with a standard dispensing/imaging system shown in Fig. 1. The micron-sized water droplets are generated with an ink-jet dispensing unit [33]. In short, the ink-jet consists 5795 of a cylindrical piezo element surrounding a glass capillary with a specific inner nozzle diameter. Three different ink-jets with inner nozzle diameters of either 120 lm, 80 lm, or 50 lm are used. In practice, the diameter of a spherical water droplet dispensed from the ink-jet equals the inner nozzle diameter (within a few microns). Experiments comprise of dispensing single microdroplets on a sample having a controlled surface temperature. Data for multiple droplets impinging on top of each other are omitted from analysis (unless stated otherwise). Video recordings during microdroplet impingement and evaporation are captured with a high-speed camera [34]. The high-speed camera is coupled with imaging optics resembling a standard compound microscope setup (i.e., eye-piece lens and microscope objective). The camera is equipped with an internal 8 Gbit buffer, 1280 800 megapixel CMOS sensor, and programmable electronic shutter with a minimum exposure time of 300 ns. Evaporation and impingement data is acquired with video acquisition rates ranging between 1200 fps and 120,192 fps. An exposure time (electronic shutter setting) of 10 ls is used in all experiments. The compound microscope optics consists of an achromatic eye-piece lens (120 mm focal length and 30 mm lens diameter) and a Mitutoyo long working distance microscope objective. Image resolution is then based on the magnification of the microscope objective selected. For example, both 20X and 50X objectives are used, facilitating image resolutions on the megapixel senor of 1.29 lm/ pixel and 0.667 lm/pixel, respectively. External TTL triggers from a computer controlled pulse generator are used to coordinate droplet dispensing and video recordings. The data acquisition procedure for measuring a single evaporation event follows: First, a digital TTL trigger initiates simultaneously (1) high-speed video recording and (2) microdroplet dispensing. Then, a 2nd TTL trigger initiates saving the acquired data. A final TTL trigger then initiates rotation of the Al-glass sample on the sample stage for the next droplet evaporation event. The surface temperature (TS) of the sample is regulated with an Omega temperature controller using a platinum RTD (resistancetemperature-device) and a 50 O resistive heating element/wire contained within a custom X, Y, Z, h sample-stage assembly. This custom sample-stage assembly is shown in Fig. 1b, where the RTD and resistive heating wire are part of the black sample-base located between the sample-stage (upper-arrow) and the sample. A temperature calibration was performed (and routinely checked) to correlate the surface temperature with the set-point temperature of the temperature controller. For this calibration, we used both IR thermometry and a custom procedure using another RTD in contact with the top surface of different sample substrates (e.g., Cu discs, Al discs, stainless steel discs, and Al thin-films on glass). For the IR thermometry measurements, the sample surfaces were painted with black (emissive) paint. We note that the doubleRTD method yields more systematic temperature calibration Fig. 1. (a) Schematic diagram of the microdroplet impingement and evaporation apparatus. (b) Close-up image of the temperature controlled X, Y, Z, h sample stage, Al-glass sample, microdrop dispenser (ink-jet), and microscope objective. (c) Schematic representation of droplet evaporation in a pinned contact line mode of evaporation. In a stickto-slip evaporation mode, the contact line (solid-liquid-vapor interface) recedes after a critical contact angle (hD) is reached (resulting in a sudden decrease in the contact radius (R) and increase in the droplet height (h)). 5796 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 results. Nevertheless, the error in surface temperature is denoted with errors bars throughout the rest of this manuscript. 2.3. Image and data analysis A custom image analysis program written in LabView is used to measure the microdroplet’s contact radius (R), height (h), and contact angle (h). The image analysis program (based on edge-contrast algorithms) defines the best edge/contour of the microdroplet and then fits a circle to the edge contour. The volume (V) of the microdroplet is then calculated based on the volume equation for a spherical cap VSC ¼ ph 6 pR 2 ð3R2 þ h Þ ¼ 3 sin h ð2 3 cos h þ cos3 hÞ; ð1Þ where the droplet contact radius (R), droplet height (h), and the surface location are measured in each image. For a spherical cap, the interfacial liquid/vapor surface area (A) is 2 ASC ¼ pðR2 þ h Þ ¼ 2p R 2 2 sin h * @ð/L qL Þ _ LG þ r ð/L qL v Þ ¼ m @t ð3Þ Fluid zone momentum equation * 3 3 fields for all variables and material properties are shared by the phases and represent volume-averaged values. The liquid phase is composed of only water, whereas the gaseous phase is composed of water vapor and air. The axisymmetric governing equations of continuity, momentum, energy, and species are solved using the segregated pressure-based solver [38]. The governing equations in differential notation for the fluid zones are as follows. Fluid zone VOF equation ð1 cos hÞ; ð2Þ where the contact angle (h) is measured directly or calculated from R and h (of course, the right-hand sides of Eqs. (1) and (2) are only valid for h < 180° and/or R > 0). In Section 4.1, an example of the image analysis process is provided (see, Fig. 3). For small, stationary water droplets these equations work well since gravity effects do not distort the curvature of the droplet. In some cases, the microdroplet surface reflection is used to improve/refine the edge contrast prior to edge detection. Typically, this is not necessary. However, it useful in some experiments at high temperatures where a vapor bubble would nucleate, grow and oscillate within the interior of the water microdroplet. The role of vapor bubbles inside evaporating droplets will be discussed in subsequent study [35]. For improved data analysis statistics, data for multiple droplet evaporation events are recorded. The variations in impingement time (dtim) and impact location (drim) are small (e.g., at most, dtim = ±100 ls and drim = ±4 lm). Again, the sample stage is rotated after each individual microdroplet impingement and evaporation event to ensure that footprint effects are minimized. These footprints are presumably due to the accumulation of small dust particles from the laboratory air at the three-phase contact line. The effects, however, only have a small influence on the impingement dynamics and initial contact conditions. For example, ten or more droplets usually need to impact the same surface location before dust accumulation is visible and begins to influence the droplet contact line dynamics (e.g., spreading, receding, and depinning). Nevertheless, the sample stage is rotated to minimize footprint effects. ** * * @ðq v Þ 2cjL q þ r ðq v v Þ ¼ rpþ r ½lðr v þr v T Þþ r /L qL þ qG @t ð4Þ Fluid zone energy equation * @ðqcp TÞ 1* _ LG þ r v qcp T þ v 2 þ p ¼ r ðKr TÞ LV m @t 2 ð5Þ Fluid zone species equation @ð/G qG wh2 o Þ * _ LG þ r ð/G qG wh2 o v Þ ¼ r ð/G qG Dh2 oair r wh2 o Þ þ m @t ð6Þ The volume fraction of gas (/G) and mass fraction of air (wair) are obtained from /G = 1 - /L and wair ¼ 1 wh2 o , respectively. The density (q), dynamic viscosity (l), and thermal conductivity (K) in Eqs. (4) and (5) are obtained from q = /LqL + /GqG, l = /LlL + /GlG, and K = /LKL + /GKG [16,17]. The thermodynamic and transport properties for water, water-vapor, and air are exactly those used in Refs. [16] and [17]. The calculation of the vapor-air mixture properties also used the procedure in Refs. [16] and [17]. 3.2. Computational domain In Fig. 2, a subset of the computational domain is shown for a water microdroplet evaporating after impact. Only a subset of 3. Numerical methods To simulate microdroplet evaporation, an explicit volume of fluid (VOF) model in FLUENT is used. Simulations are preformed for water microdroplets evaporating on an isothermal wall as a function of accommodation coefficient (e). A thorough description of our numerical approach is provided in Refs. [16,17], and [18]. In the following, the numerical approach is summarized in terms of (1) the physical fields and governing equations, (2) the computational domain, (3) evaporation using Schrage’s mass flux equation [36]. and 4) the initial conditions of the simulations presented. 3.1. Physical fields and governing equations The model tracks the time-dependent volume fractions of liquid (/L) and gas (/G) throughout the computational domain [37]. The Fig. 2. Subset of the computational domain for a numerical simulation of a water microdroplet evaporating on an isothermal domain wall at TS = 100 °C. The figure is color coded to show the liquid volume fraction (/L), where red indicates liquid, white indicates gas/air, and the colors between indicate the liquid/vapor interface. The velocity vectors show the convection patterns in both the liquid- and gasphases. The solid black lines indicated the temperature isotherms in units of K. The time stamp indicates the time elapsed during sessile droplet evaporation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article) S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 the computational domain is provided because the domain actually extends into both the z– and r–directions a distance 10 times the microdroplet radius. The figure presents the instantaneous droplet with velocity vectors and temperature isotherms. No solid zone is needed for modeling on an isothermal wall. The governing equations above correspond to the fluid zone which contains both the liquid and gas phases. The fluid zone is enclosed by the wall interface, centerline, surroundings, and enclosing walls for the gas phases. The enclosing walls for the gas phases have adiabatic boundary conditions. The centerlines and surroundings are modeled with axis of symmetry and pressure-outlet boundary conditions, respectively. 3.3. Simulating the liquid-vapor mass transfer rate To model evaporation or the volumetric mass transfer rate _ LG Þ, the expression ðm _ LG ¼ jraG j m " !# 1=2 2e Mw pL pG ð2 eÞ 2pNA kB T L1=2 T 1=2 G ð7Þ is used, which is the product of the droplet interfacial surface area density (|raG|) and the Schrage’s mass flux (bracketed term) [36], where e is the evaporation efficiency or accommodation coefficient, Mw is the molecular weight, NA is Avogadro’s number, kB is Boltzmann’s constant, pG (pL) is the gauge pressure of the gas (liquid), and TL (TG) is temperature of the liquid (gas). It is assumed at the liquid-vapor interface that the temperature of a simulation cell is T = TG = TL, where PL is the saturation pressure at T (i.e., Psat(T) = PL) and PG is the absolute vapor partial pressure. In this context, if PL (the saturation pressure at T) is greater than wh2 o Pcell ðTÞ, then _ LG is positive. Condensation occurs the liquid evaporates and m 5797 when the converse is true. Pcell refers to the absolute pressure at the interface cell. The value of Pcell is calculated from the equation of state (i.e., Pcell is a function of T), noting that Pcell can also be obtained by summation of the absolute pressure (i.e., operating pressure) and the gauge pressure at that cell. Likewise, the water saturation pressure is also computed to obtain the pressure of the liquid at the interface cell, where PL = Psat(T). During a simulation, all variables on the right-hand side of Eq. (7) evolve in time except for e (the accommodation coefficient). To date, the accommodation coefficient is not well established for a given fluid. In fact, in some cases, e has shown to differ between experiments and/or theory by up to three orders of magnitude [39]. We assume, as others have, that e is a constant, using either e = 0.1 or e = 1.0 in the simulations. Modeling evaporation with the Schrage mass flux (i.e., bracketed term in Eq. (7)) has been done by others. For example, Dhavaleswarapu et al. [40] recently implemented the Schrage mass flux equation to simulate steady-state meniscus evaporation in a channel, using the heat flux and meniscus profile as model inputs. In this study, unsteady evaporation is simulated, using the ‘‘initial’’ droplet shape and ‘‘initial’’ wall temperature as model inputs (thus, everything is evolving in time). How Eq. (7) is implemented is also different. Unlike Ref. [40], this study uses the Schrage equation as a _ LG source/sink term in the fluid governing equations, where m serves as a sink term in the liquid volume of fluid (VOF) transport equation (Eq. (3)), as a source term in the water vapor species equation (Eq. (6)), and as a sink term in the energy equation (Eq. (5)). 3.4. Details of simulations presented Two simulations were conducted for the evaporation dynamics. Impingement simulations were not carried out because droplet impingement on an isothermal wall was studied previously [16]. For both simulations, the temperature of isothermal wall was set to Ts = 373.15 K and the initial temperatures of the surroundings and the microdroplet were set to ambient conditions (i.e., TG = 300 K, RH = 30%). The droplet was placed with an initial contact angle of h0 = 90° and contact radius of R0=35.05 lm with no * impingement velocity ðv im ¼ 0m=sÞ. For the first simulation the accommodation coefficient was fixed at e = 0.1, whereas, for the second, e = 1.0. The numerical simulations were conducted in parallel using three processors in conjunction with dynamic meshing, Non-iterative time advancement (NITA) [41] and variable time-stepping. The model is first-order accurate in time and second-order accurate in space. Eq. (7) is implemented in FLUENT via C programming language subroutines. Previously, we showed that the flow fields (i.e., liquid and vapor convection patterns in Fig. 2) are accurately resolved with our grid size and meshing schemes [16,17]. 4. Results and discussion Fig. 3. (a) Dimensionless droplet-profile data as a function of time for a water microdroplet evaporating on an Al/glass substrate with a surface temperature of TS ffi 175 °C, where R/R0 (black-symbols), h/h0 (red-symbols), h/h0 (green-symbols), and V/V0 (blue-symbols) are the dimensionless contact radius, droplet height, contact angle and droplet volume, respectively. (b) Microdroplet image acquired with high-speed camera at t ffi 40 ms. (c) Same droplet image provided in (b) with a circle fit (yellow-line) to the edge contour (red-lines). Experimental Details: TS 175 ± 6.5 °C, RH ffi 30%, R0 ffi 52 lm, h0 ffi 96 lm, h0 ffi 117, and V0 ffi 871 pL. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article) The study focuses on water microdroplets with small impinge* ment-diameters (Dim) and small impingement-velocities ðv im Þ varying within typically (50 lm[ Dim [ 170 lm) and * ð0:4m=s K v im K 1:1 m=sÞ, respectively. Thus, the Weber numbers * 2 ðWe ¼ q v im Dim =cÞ are also small (We [ 3.0). In this regard, phenomena such as droplet recoil, jetting, splash, break-up and rebound do not take place during the short-time impingement dynamics (t [ 500 ls). Oscillations in the droplet’s contact angle and height do take place after impact; however, during these short-time impingement oscillations, the contact line is pinned at the maximum spreading diameter. In the following, the impingement dynamics are not discussed. Instead, the study focuses on the longer-time evaporation kinetics (t [ 500 ls). 5798 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 4.1. Microdroplet evaporation on Al thin-film surfaces 4.1.1. Changes in droplet size and shape during evaporation Evaporation experiments with isothermal Al/glass surfaces covered surface temperatures ranging from room temperature up to superheated temperatures of TS ffi 250 °C. Fig. 3 shows a raw data set acquired with the high-speed camera during a single evaporation experiment at Ts ffi 175 °C. Fig. 3 actually plots the dimensionless microdroplet-profiles (R/R0, h/h0, h/h0, and V/V0) as a function of time. As shown, the droplet’s contact radius (black-lines) is relatively constant until 33 ms into the evaporation process. After that, the contact radius rapidly decreases to another fixed value for 3 ms, and then, subsequent stepwise reductions in the contact radius take place until the entire droplet has evaporated. The microdroplet’s contact angle (green-lines) changes accordingly during this stick-slip contact line behavior. The blue-lines correspond to the droplet volume calculated with Eq. (1) using the measured contact radius (R) and droplet height (h). As shown in Fig. 3, the rate of change in the microdroplet volume is not constant (i.e., the evaporation rate is not constant). Fig. 4 provides droplet-profile data for another experiment at TS ffi 118.5 °C. The data is similar to that provided in Fig. 3, showing (in dimensionless form) the changes in the microdroplet’s radius, height, contact angle, and volume as a function of time. Like the data in Fig. 3, stick-slip contact line behavior is observed. For example, the droplet’s contact radius is fixed (pinned) until 100 ms into the evaporation process. After that, the contact radius rapidly decreases to another fixed value for 12 ms, and then, subsequent stepwise reductions in the contact radius take place until the entire droplet has evaporated. The contact angle changes accordingly during this pinning and depinning of the contact line. Again, the rate of change in droplet volume is not constant. For example, after the droplet has evaporated for T ffi 120 ms the slope of V(t) changes from, arguably, a constant value of dV/dt ffi -6.33 pL/ms to a smaller value of dV/dt ffi -2.89 pL/ms. Hence, at t ffi 120 ms the average _ LG ffi 6:2 lg/s to evaporation rate decreases in magnitude from m _ LG ffi 2:8 lg/s (i.e., a reduction in m _ LG greater than 50%). A nearly m identical V(t) dependence is shown in Fig. 3. 4.1.2. Comparisons between simulation and experiment Fig. 5 compares experimental data with simulation results. Fig. 5a shows V(t) data for small water microdroplets ðR0 35 lmÞ evaporating at TS ffi 100 °C. Fig. 5b provides the same data in dimensionless form (i.e., V(t/sE)/V0, where sE is droplet lifetime). In both experiments, contact line depinning occurs at t/sE ffi 0.55. The simulations do not account for contact line depinning (i.e., the contact radius is held constant, R(t) = R0, throughout the entire simulation). Simulation results are provided for two different accommodation coefficients (i.e., e = 0.1 and e = 1.0). As shown, the simulations under predict the evaporation rate when an accommodation coefficient of e = 0.1 is used. Better agreement is found for e = 1.0. Similar V(t) predictions are obtained for accommodation coefficients within the range 0.4 6 e 6 1.0. This is because the predicted value of sE is not sensitive to changes in e for 0.46 e 6 1.0. This is discussed in more detail in Refs. [17] and [18], where comparisons are made between simulation data as a function of accommodation coefficient, droplet size, and droplet curvature. Fig. 6 compares the experimental and simulated droplet-profile data. For these small water microdroplets, the measured evapora_ LG / dVðtÞ=dt) is relatively constant until contact tion rate (i.e., m line depinning occurs (indicated by the arrow symbols in Fig. 5 and the change in R/R0 and h/h0 in Fig. 6). The simulated evaporation rate also decreases in magnitude as the droplet evaporates. However, this is not due to depinning because the contact line is pinned in the simulations (i.e., R(t)/R0 = 1 for the simulation data Fig. 4. Dimensionless droplet-profiles as a function of droplet lifetime during evaporation on Al/glass at TS ffi 118.5 °C. Droplet images are provided at two different droplet lifetimes, where the 100 lm scale-bar applies to both images. Experimental Details: TS = 118.5 ± 4.6 °C, RH ffi 30%, V0 ffi 977.5 pL, R0 ffi 64.2 lm, h0 ffi 91.3 lm, and h0 ffi 111°. in Fig. 6). Instead, the simulated evaporation rate decreases in magnitude as the droplet height decreases. As discussed in Ref. [18], in a pinned contact line mode of evaporation, reductions in the droplet height coincide with reductions in the velocity of circulative fluid motion (i.e. vorticity), which, in turn, reduces the magnitude of the simulated evaporating rate. Based on these results, the simulations are over estimating the influence of fluid convection on the evaporation rate. An improved simulation model is currently under development which can accurately account for (1) the constant evaporation rate observed during the pinned mode of microdroplet evaporation and (2) stick-slip contact line motion. 4.1.3. Microdroplet evaporation at different surface temperatures Fig. 7 shows temporal volume data for microdroplets evaporating on Al/glass at three different surface temperatures. The data is plotted in dimensionless form. The initial droplet volumes are not the same in each experiment. However, this does not change the functional form of the dimensionless data presented. For example, contact line depinning occurs at t/sE ffi 0.5 in all three experiments, and, in result, V(t/sE)/V0 is nearly identical for the three experiments. Again, the evaporation rate is constant before contact line depinning and then decreases in magnitude as the microdroplet gets smaller. Typically, the average evaporation rate decreases by more than 50% after contact line depinning. Similar results were obtained by Mollaret et al. [42] with millimeter-sized water droplets evaporating on both Al and Teflon at surface temperatures within 25 °C[ DTS [ 100 °C. It is interesting that V/V0 as a function t/sE is independent of surface temperature. Non-dimensional relationships have shown to be powerful in predicting the functional form of the hydrodynamic boundary layer in fluid dynamics. Although, care should be taken in interpreting that V/V0 as a function of t/sE is truly independent of the surface temperature. This will only be true if the temporal depinning dynamics are also temperature independent (i.e., the change in the contact radius relative to the droplet lifetime, R(t/sE), is temperature independent). 5799 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 droplet-profile (dimensionless) 1.0 0.8 0.6 0.4 0.2 0 0 15 30 time (ms) 45 60 Fig. 6. Predicted (lines) and measured (symbols) dimensionless droplet-profiles for water microdroplets evaporating on Al/glass at TS ffi 100 °C. Simulation Details: e = 1.0, RH = 20%, TS = 100 °C, V0 = 89.8 pL, R0 = 35 lm, h0 = 90°, and sE ffi 46.3 ms. Experimental details: RH = 30 ± 3%, TS = 100 ± 2 °C, R0 ffi 34.6 ± 0.5 lm, 91 ± 1°, sE ffi 58.5 ms, sD ffi 32 ms. Fig. 5. (a) Predicted (lines) and measured (symbols) droplet volume as a function of time for microdroplets evaporating on Al/glass at TS ffi 100°C. (b) Volume ratio (V/ V0) as a function of dimensionless droplet lifetime (t/sE). Simulation results are provided using accommodation coefficients of e = 0.1 (black lines) and e = 1.0 (red lines). Experimental data are provided for two experiments with initial droplet volumes of V0 ffi 100 pL (black circles) and V0 ffi 90 pL (blue triangles), where the arrows signify when depinning occurs (sD). Simulation Details: RH = 20%, TS = 100 °C, V0 = 89.8 pL, R0 = 35 lm, h0 = 90°; black lines – (e = 0.1, sE ffi 155 ms); and red lines – (e = 1.0, sE ffi 46.3 ms). Experimental Details: RH = 30 ± 3%, TS = 100 ± 2 °C; black circles – (R0 ffi 35.7 ± 0.5 lm, h0 = 91 ± 1°, sE ffi 64.5 ms, sD ffi 36 ms); and blue triangles – (R0 ffi 34.6 ± 0.5 lm, h0 ffi 91 ± 1°, sE ffi , 58.5 ms, sD ffi 32 ms). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article) We point out that this ‘‘temperature-independent’’ transition in _ LG occurs when the microdroplets are still relatively large. That is, m the droplets are still several micrometers in size (R dT, h dT), where dT, the Tolman length, is on the order of nanometers. There_ LG is not attributed with size-dependent fore, the transition in m deviations in the droplet’s surface tension due to disjoining pressures, which are important for liquid thin-films and droplets with sizes comparable to dT [43,44]. Fig. 8 summarizes the results for water microdroplets evaporating on Al/glass at surface temperatures within 25 °C [ DTS [ 230 °C. Each data point at a given surface temperature is based on the average of four or more ‘‘independent’’ experiments (usually more). Data at surface temperatures beyond 230 °C is omitted because the droplets start bouncing, exploding, and flipping after Fig. 7. Dimensionless-droplet-volume ratio (V/V0) plotted as a function of dimensionless-droplet-lifetime ratio (t/sE) for microdroplet evaporation on Al/glass at three different surface temperatures. Experimental details: RH = 30 ± 3%; red-square data – (TS ffi 175 °C, V0 ffi 923 pL, h0 ffi 117°, sE ffi 85 ms, sD ffi 42 ms); black-circle data – (TS ffi 119 °C, V0 ffi 981 pL, 111, sE ffi 203 ms, sD ffi 102 ms); and blue-triangle data – (TS ffi 48 °C, V0 ffi 265 pL, h0 ffi 107°, sE ffi 770 ms, sD ffi 380 ms). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article) impingement. Three different evaporation rates are provided at each surface temperature because the evaporation rate decreases in magnitude after contact line depinning (i.e., the temporal form sD _ t< of V(t) follows that discussed previously). For example, m LG (open-circles) corresponds to the average evaporation rate measD _ t< sured before contact line depinning (i.e., m ¼ qdV=dt, where LG dV/dt is the average rate of change of the droplet volume before _ sLGE (filled-circles) corresponds to depinning). On the other hand, m the average evaporation rate measured for complete dry out (i.e., _ sLGE ¼ qV0 =sE ). The evaporation rate before depinning is always m the greatest. All evaporation rates are provided because they are all important in an evaporative cooling technology. 5800 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 The remainder of this manuscript focuses on the evaporation rate measured before contact line depinning; stressing again that sD _ t< m is a constant, whereas for t > sD the evaporation rate graduLG ally decreases in magnitude during the stepwise reductions in the contact radius. The reason why a change in the contact radius (length of the contact line) influences the measured evaporation rate is discussed in the following. 4.1.4. Influence of Al surface quality and changes in the contact line length on the evaporation rate Fig. 9 summarizes all the evaporation data measured with water microdroplets on different Al thin-films. Experimental data for three different Al/glass samples are reported. Two of the data sets are for samples that have been used in multiple experiments prior to these measurements. For example, these samples have been wiped clean of dust several times, rinsed with water, alcohol, and acetone, and went through several heating and cooling cycles. Hence, these Al thin-film samples (filled-circles and red-star-circles) are referred to as ‘‘practical’’ Al thin-films because they better represent an Al thin-film surface in a cooling system. The ‘‘cleanroom’’ data (open-circles), on the other hand, is for an Al/glass sample with an ideal Al thin-film surface (e.g., the ‘‘cleanroom’’ sample was only exposed to the laboratory air prior to testing). In principle, it is difficult to maintain and produce ideal surfaces. Thus, in many cases, it is impractical to base system-level cooling designs on ‘‘cleanroom’’ quality material systems. The data for multiple Al/glass samples is presented because system reliability is an important parameter in two-phase cooling [45]. At first glance it appears that the evaporation efficiency is improved with the cleanroom quality Al surface. However, for the data in Fig. 9, the initial droplet volume (V0) also contributes to _ LG measured. For example, the ‘‘cleanroom’’ the magnitude of m data are for the slightly larger droplets (relative to the ‘‘practical’’ data) and, in result, have the largest evaporation rates. Many experiments have been conducted with microdroplet volumes ranging from 50 to 42,000 pL. The results of these experiments have consistently shown that the evaporation rate increases with increasing droplet size. However, it is not actually the droplet Fig. 8. Temperature dependent evaporation rates for water microdroplets evaporating on an Al /glass substrate. Three evaporation rates are provided for each set of experiments at a given surface temperature. The symbols correspond to the average evaporation rate measured (1) before contact line depinning (open circles), (2) after contact line depinning (open triangles), and (3) for complete dry out (filled circles). Experimental Details: 5% [ RH [ 30%, 40 lm [ R0 [ 56 lm, 103°[ h0 [ 125°, and 260 pL [ V0 [ 1150 pL. volume or droplet contact area that influences the magnitude of the evaporation rate measured. Instead, it is the total length of the contact line that influences the evaporation rate. Fig. 10 shows the evaporation rate (before contact line depinning) per unit length of contact line: sD sD _ t< p_ t< ¼m LG LG =ð2pR0 Þ: ð8Þ As shown, all evaporation data collapses onto a single evaporation sD _ t< curve by scaling m by the length of the contact line (2pR0). This LG not surprising, our simulations show significant increases in the evaporative mass flux near the contact line (e.g., the simulation data in Fig. 2 shows the significant difference between the magnitude of the mass flux near the contact line and the mass flux at the droplet sD _ t< apex). Others have also shown that the m scales linearly with LG the droplets contact radius [46]. A systematic trend like that shown sD _ t< in Fig. 10 is not found by scaling m by other powers of R0 (e.g., LG 2 R0 ), supporting that phase-change heat transfer is dominated by the liquid-vapor phase transformations at the solid-liquid-vapor (a.k.a., three-phase) interface. 4.1.5. Maximum evaporation rates at high surface temperatures _ max A maximum (critical) evaporation rate of m LG ¼ 32 3 lg/s is found for water droplet evaporation on Al thin-films. This is signi_ LG measured at fied in Fig. 9 by the maximum values of m sD TS ffi 180 °C. Based on the p_ t< data in Fig. 10, the maximum evapLG oration rate per unit length of contact line is p_ max LG ¼ 65 5 ng/s/ lm. For the data discussed thus far, the highest temperature data points correspond to the surface temperature at which the droplet’s evaporation dynamics become non-sessile (perhaps chaotic in nature [47]). Data analysis at surface temperatures beyond the highest temperature data point yields results inconsistent with the lower temperature data. This transition in impingement and evaporation dynamics at large superheats can be attributed to the Leidenfrost effect [48]. which is prevalent in both boiling and sessile droplet evaporation (beyond the CHF) where a thermallyresistive vapor layer forms between the droplet and a heated Fig. 9. Temperature dependent evaporation rates measured before contact line depinning for water microdroplets on three different Al thin-films on glass. Experimental Details: open circles – (largest droplets, V0 = 1650 ± 580 pL, R0 = 88 ± 13 lm, h0 = 95 ± 9°, RH = 48 ± 3%); red-star circles – (smallest droplets, V0 = 340 ± 110 pL, R0 = 54 ± 5 lm, h0 = 91 ± 10°, RH = 23 ± 7%); and filled circles – (intermediate droplet volumes, V0 = 610 ± 308 pL, R0 = 50 ± 5 lm, h0 = 115 ± 8°, sD _ t< RH = 18 ± 12%, note: same m data provided in Fig. 8). (For interpretation of the LG references to colour in this figure legend, the reader is referred to the web version of this article.) S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 5801 conducted with this practical thin-film at TS ffi 200 °C. In these experiments a majority of the microdroplets either elastically bounced after impact or exploded during the evaporation process. The data at TS ffi 200 °C represents the few small microdroplets that actually wet the Al surface and did not explode. As shown, the heat flux starts to decrease at surface temperatures beyond 180 °C, sig2 nifying a critical evaporative heat flux of qCHF Al ¼ 600 100 W/cm . 4.2. Microdroplet evaporation on different thin-film surfaces Fig. 10. Evaporation rate per unit length of contact line measured before contact sD sD _ t< is calculated with Eq. (8) using R0 and m measured at a line depinning. p_ t< LG LG given surface temperature. Experimental details: provided in Fig. 9 caption. surface [45]. In general, a transition from a purely microdroplet evaporation regime to a droplet/film boiling regime is observed, where this transition is coupled to both (1) a critical surface tem_ max perature — T max and (2) a critical evaporation rate — m LG ). Thus, S this transition is analogous to that observed in pool boiling at CHF, where, at CHF, the boiling process is switching from a nucleate-boiling regime to a film-boiling regime. As shown in Fig. 10, the maximum surface temperatures measured with microdroplets on Al thin-films ranges within 170 °C [ DTmax [ 230 °C, which corresponds to maximum superheats of DTmax = 100 ± 30 °C. In fact, this measured value of Tmax for the filled-circle data corresponds very well with the Leidenfrost temperatures (TLF) measured by Mudawar et al.[45] during water droplet impingement on polished Al surfaces (e.g., see Table 4 in Ref. [45]). However, Tmax is not equated with TLF becasue Tmax is really a surface temperature somewhere between the surface temperature at critical heat flux (TCHF) and the true Leidenfrost temperature of the system (i.e., TCHF [ DTmax [ DTLF). The size of the droplet also influences the surface temperature _ max at which the maximum evaporation rate ðm LG Þ is observed. As _ max shown in Fig. 9, m is reached by increasing either 1) the surface LG _ max temperature or 2) the droplet size. Thus, m LG is coupled both DT and R0 (similar to the correlation between qCHF and the vapor-bubble release size (RVB) observed in boiling). In addition to Al thin-films, several other thin-film surfaces are studied to understand how changes in thin-film thermal conductivity and thin-film surface energy influence the heat transfer performance. Fig. 12 shows the heat transfer performance at different surface temperatures for all the thin-film surface materials tested. We note that the Cu thin-film (blue-triangles), Ti thin-film (redstars), and SAM surface (yellow-diamonds) are all ‘‘cleanroom’’ quality surfaces. Fig. 12a provides the average evaporation rate measured per unit length of contact line, whereas Fig. 12b provides the average evaporative heat flux (q) measured based on Eq. (9) (i.e., q before contact line depinning). As shown in Fig. 12, the maximum surface temperatures range within 170 °C[ DTmax [ 230 °C, corresponding to maximum superheats of DTmax = 100 ± 30 °C. The data also shows that the thin-film thermal conductivity has no significant influence on the heat transfer performance. For example, the thermal conductivity (K) of these thin-film samples ranged from that of Cu (KCu ffi 400 W/m/K) to Al (KAl ffi 200 W/m/K) to Ti (KTi ffi 20 W/ m/K) and then to presumably the smallest value for the SAM sD (KSAM [ 5 W/m/K); yet, both the heat flux (q) and p_ t< fall within LG the scatter of the data. This null result may be due to several factors, such as (1) the presence of the oxide layer on the metal surfaces, (2) the low thermal conductivity of the glass substrate, and/or (3) that largest resistance to heat transfer is at the liquidvapor interface (not the solid-liquid interface) [18]. The thermal conductivity of an oxide surface layer is estimated to range between that of aluminum oxide ðKAl2 O3 ffi 45 W=m=KÞ and that of titanium oxide ðKTiO2 ffi 8 W=m=KÞ. In this case, the variation in the thermal conductivity of an oxide surface layer is within a factor of 5; yet, a small change in both the evaporation rate and 4.1.6. Evaporative heat flux before contact line depinning Fig. 11 shows the heat flux measured as a function of surface temperature with water microdroplets on heated Al/glass substrates. The heat flux is calculated based on sD 2 _ t< q ¼ Lv m LG =ðpR0 Þ; ð9Þ sD _ t< m LG where is the average evaporation rate measured before depinning, R0 is the measured contact radius, and Lv is the latent heat of vaporization of water. As shown in Fig. 11, the smallest droplets (red-star-circles) have the largest heat flux and biggest droplets (open-circles) have the smallest heat flux. Thus, larger droplets have increased evaporation rates but reduced heat fluxes (as clearly shown by comparing the heat flux data in Fig. 11 with the evaporation data in Fig. 9). The maximum heat flux measured with water microdroplets on superheated Al thin-films is qmax ffi 700 W/cm2. This was measured in one experiment of the four conducted at TS ffi 180 °C with a practical Al thin-film (red-star-circle data). Several experiments were Fig. 11. Heat flux as a function of surface temperature for water microdroplets on three different Al thin-film samples. The heat flux is calculated with Eq. (9) using R0 sD _ t< and m measured. Experimental details: provided in Fig. 9 caption, open circles – LG (largest droplets, V0 = 1650 ± 580 pL), red-star circles – (smallest droplets, V0 = 340 ± 110 pL), and filled circles – (intermediate droplets, V0 = 610 ± 308 pL). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article) 5802 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 heat flux is observed. Others have shown that heat conduction through the substrate can be the rate-limiting mechanism for droplet evaporation [46,49,50] (i.e., the substrate thermal conductivity limits the heat flux, which, in turn, regulates the evaporation rate). In this regard, a small change in the heat transfer performance would be expected for these different thin-film surfaces because the thermal conductivity of the glass substrate is regulating the heat flux (which, in turn, dictates the heat transfer coefficient in a purely evaporative heat transfer regime). Interestingly, the change in surface energy also has a small influence on the heat transfer performance (i.e., q(DT)). The surface energy of these thin-film samples presumably ranged from the most hydrophilic Cu, to neutrally hydrophobic Al and Ti surfaces, and then, to the hydrophobic SAM surface. We note that just measuring the static contact angle after impingement is not a reliable metric for qualitatively estimating the changes in surface energy. For example, static contact angle measurements for all thin-film samples at 25 °C [ TS [ 95 °C suggested that the Ti surface was the most hydrophilic; yet the receding contact angles were the smallest with the Cu and ‘‘cleanroom’’ Al thin-films [51]. Nevertheless, changing the surface material did influence the depinning dynamics. For example, the microdroplets evaporating on the SAM surface started to depin much earlier relative to all the metal thin-film surfaces. However, the evaporation rates and heat fluxes for all samples are within error of each other (especially at high surface temperatures approaching T max where the maximum evaporation S rates are observed). The small changes in the evaporation rate and heat flux for the SAM surface relative to the metal thin-films also follows previous evaporation studies with monomolecular surfaces [52,53], where in Ref. [52] only a 30% reduction in the evaporation rate was observed by changing surface from highly hydrophilic (hEq ffi 20°) to hydrophobic (hEq ffi 110°). These results showing the small influence on the thin-film thermal conductivity and surface energy suggests that the greatest resistance to evaporative heat transfer exists at the liquid-vapor interface. Yet, not only the liquid-vapor interface but the interface in the proximity of the solid-liquid-vapor contact line. Larger _ LG Þ relative microdroplets will have increased evaporation rates ðm to smaller ones; yet, the heat flux is reduced by the corresponding _ LG is increase in the wetted surface area. Moreover, even though m greater with larger microdroplets, the evaporation rate per unit length of contact line ðp_ LG Þ is arguably constant and independent of the surface material. Traditionally, it is expected that hydrophilic surfaces will provide the best heat transfer performance. This is not true in regards to the evaporative heat flux with water microdroplets. As shown in Fig. 12b, the heat flux performance with the hydrophobic SAM surface is the best of the ‘‘cleanroom’’ quality samples. This suggests that thin-film surfaces that yield microdroplet contact angles between 95 and 115° will, in turn, remove the most heat. Moreover, regardless of the thin-film surface material to be used in a spray cooling technology, evaporation rates and evaporative heat fluxes CHF _ max _ max beyond m = 600 LG ¼ 32 5 lg/s, pLG ¼ 63 7 ng/s/lm, and q ± 100 W/cm2 should not be expected with water microdroplets of volumes within 100 pL[ V0 [ 2500 pL. Thus, maximizing the contact line length per unit surface area is the most important design parameter to consider in a spray cooling system (which can be accomplished by simply spraying small water microdroplets). 4.3. Microdroplet Evaporation on Cu Surfaces Fig. 12. Heat transfer performance for water microdroplets evaporating on a variety of different thin-film surfaces. (a) Evaporation rate per unit length of contact line measured before contact line depinning. (b) Evaporative heat flux before contact line depinning. For clarity, the error bars are removed from most of the ‘‘practical’’ Al thin-film data. Experimental details: Cu thin-film – (V0 = 1631 ± 1000 pL, R0 = 82 ± 18 lm, h0 = 97 ± 7°); Ti thin-film – (V0 = 1460 ± 723 pL, R0 = 92 ± 15 lm, h0 = 84 ± 8°); hydrophobic SAM – (V0 = 1404 ± 618 pL, R0 = 68 ± 11 lm, h0 = 113 ± 5°); ‘‘cleanroom’’ Al thin-film – (V0 = 1650 ± 580 pL, R0 = 88 ± 13 lm, h0 = 95 ± 9°); and ‘‘practical’’ Al thin-films – (red-star circles: V0 = 340 ± 110 pL, R0 = 54 ± 5 lm, h0 = 91 ± 10°; open circles: V0 = 610 ± 308 pL, R0 = 50 ± 5 lm, h0 = 115 ± 8°). A variety of Cu material systems are also studied to explore the role of surface microstructure and substrate thermal conductivity on microdroplet evaporation. The Cu systems tested are Cu thin-films on glass and solid Cu substrates with different surface structure/roughness (i.e., mirror-polished, 800-grit sanded, and micron-sized Cu pillars). Fig. 13 shows droplet profile data during evaporation on the Cu pillar surface at TS ffi 90 °C. For this data, multiple 1000 pL microdroplets were used to create/deposit a 42,000 pL microdroplet on the pillar surface (e.g., t=0 sec corresponds to the droplet after the last impingement event with a 1000 pL microdroplet). As shown, the evaporation kinetics with this big microdroplet on the pillar Cu surface is analogous to the thin-film data discussed previously (i.e., the evaporation rate decreases after contact line depinning). In fact, all surfaces tested in this study show stick-slip contact line behavior and constant evaporation rates before depinning. Fig. 14 summarizes the heat transfer performance for microdroplets evaporating on the different Cu surfaces. The data focuses on microdroplets with initial volumes within the range 1000 pL[ V0 [ 2000 pL. Evaporation data with bigger microdroplets like that shown in Fig. 13 is only provided for the Cu pillar surface. Again, the highest temperature data points correspond to surface temperatures approaching the Leidenfrost temperature, where T max is proprotional to R0 and signifies the surface temperS ature for transition from a purely evaporative heat transfer regime S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 5803 Fig. 13. Dimensionless droplet-profiles as a function of microdroplet lifetime during microdroplet evaporation on a Cu surface structured with Cu pillars at TS ffi 90 °C. Droplet images are provided at two different droplet lifetimes, where the 160 lm scale-bar applies to both images. The inset image is a TEM micrograph of the Cu pillar surface. Experimental details: TS = 90 ± 2 °C, RH ffi 25%, V0 ffi 41,700 pL, R0 ffi 123.5 lm, h0 ffi 395.5 lm, and h0 ffi 146°. to a droplet/film boiling regime. Data at TS [ 90 °C for the bigger microdroplets on the Cu pillars is not provided because boiling takes place within these bigger microdroplets at surface temperatures slightly beyond 100 °C (i.e. the saturation temperature). The microdroplets on all solid Cu substrates (Cu mirror, Cu 800-grit and Cu pillars) start to boil at much lower surface temperatures relative to that for microdroplets evaporating on a Cu thinfilm. For all the solid Cu substrate materials tested, the maximum superheats observed fall within the range 3 °C [ DTmax [ 30 °C, in comparison to 70 °C [ DTmax [ 130 °C for the thin-film surface coatings on glass substrates. Again, this is due the difference in the thermal conductivity substrate, which dictates the maximum allowable heat flux during evaporation (i.e., the Cu thin-film is only 70 nm thick so the heat flux through the thin-film on the glass substrate is limited by the thermal conductivity of glass). Comparison of the evaporation data suggests maximum achiev_ max _ max able evaporation rates of m LG ¼ 36 4 lg/s and pLG ¼ 80 4 ng/ s/lm (which is comparable to that observed with the thin-film materials discussed in Section 4.2). Comparison of the evaporative heat flux suggests a critical heat flux of qCHF = 750 ± 300 W/cm2 (also comparable to that observed with the thin-film materials). Moreover, as before, larger droplets have increased evaporation rates; yet, reduced heat fluxes (as clearly shown in Fig. 14 by comparing the Cu pillar data of different droplet volumes). In regards to the heat transfer performance, the microstructured Cu pillar sample yields the best heat transfer coefficient (h) of all substrate materials tested, assuming h = q/DT and that the average surface temperature underneath the microdroplet is within the error of the surface temperature calibration. For example, the heat transfer coefficients measured at a superheat of DT 10 °C are hpil2 2 lar = 650 ± 200 kW/m /K, h800-grit = 260 ± 50 kW/m /K, hmirror = 240 ± 90 kW/m2/K, and hthin-film = 50 ± 25 kW/m2/K. Thus, a 2.5 increase in the heat transfer performance is observed with the Fig. 14. Heat transfer performance for water microdroplets evaporating on different Cu substrates. (a) Evaporation rate per unit length of contact line before depinning (Eq. (8)). (b) hD/h0 Evaporative heat flux before depinning (Eq. (9)). Experimental details: RH = 48 ± 3%; Cu thin-film – (V0 = 1631 ± 1000 pL, R0 = 82 ± 18 lm, h0 = 97 ± 7°); Cu mirror – (V0 = 1405 ± 539 pL, R0 = 87 ± 12 lm, h0 = 89 ± 12°); Cu 800grit – (V0 = 1200 ± 610 pL, R0 = 74 ± 17 lm, h0 = 100 ± 13°); Cu pillars – (V0 = 1000 ± 410 pL, R0 = 45 ± 6 lm, h0 = 130 ± 7°); and Cu pillars with bigger droplets – (V0 = 6000 ± 1200 pL, R0 = 105 ± 10 lm, h0 = 125 ± 10°). Cu-pillars relative to the other Cu substrates with either a mirror or 800-grit surface finish. This suggests that the Cu pillar surface could be promising for enhancing performance in current and future evaporative heat exchangers. This enhanced heat transfer performance is attributed to the significant increase in the contact line length per unit surface area. The water microdroplets on the pillared Cu surface are effectively hydrophobic and are in either (1) a partial wetting or (2) nonwetting (Cassie) state [54,24,55]. In this case, vapor convection and evaporation is possible underneath the microdroplet (i.e., within the pillar region that separates the microdroplet from the Cu solid substrate). Hence, an increased heat transfer coefficient is obtained because there is an increase in contact line length per unit area (i.e., the region where the majority of evaporative heat transfer takes place). 4.4. Contact Line Depinning and Surface-Liquid Intermolecular Forces 4.4.1. Changes in the evaporation efficiency after depinning Fig. 15 shows the effects of contact line depinning on the evaporation efficiency. As shown, changes in the surface material changes the depinning angle, but does not systematically change 5804 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 sD _ t>sD Fig. 15. Measured evaporation ratio ðbLG ¼ p_ t< LG =pLG Þ as a function of the depinning contact angle (hD). Data is based on all water microdroplet experiments at surface temperatures within 25 °C[ DTS [ 125 °C (260 pL[ V0 [ 2200 pL). The symbols for each surface material are same as that used previously (see the legends in Figs. 12 and 14). the ratio of the evaporation rates (per unit length of contact line) sD _ t<sD before and after contact line depinning (i.e., bLG ¼ p_ t> LG =pLG ). A direct correlation between hydration forces and the evaporation efficiency is not observed in this study. In general, the measured evaporation ratio scatters around bLG ffi 1 for all surface materials tested. However, the data suggests that bLG may start to increase beyond unity for highly hydrophilic surfaces (i.e., hD ? 0°). This result is reasonable given that not all the evaporation takes place at the contact line. Some evaporation takes place at the droplet apex. So, for thin-film like droplets the evaporation rate at the apex may start to approach that at the contact line as the droplet height reduces. Yet, this is surprising because reduced evaporation rates are expected for small water microdroplets in contact with a hydrophilic surface in comparison to a hydrophobic surface. For example, a transition in the evaporation kinetics due to water hydration has recently been suggested by Golovko et al.[44] How_ LG Þ was ever, in Ref. [44], the reduction in the evaporation rate ðm observed with submicron-sized water droplets on hydrophilic Si surfaces (not micron-sized water droplets, as studied here). More_ LG Þ is reported (not, p_ LG , over, in Ref. [44], the evaporation rate ðm the evaporation rate per unit length of contact line, as studied here). Thus, our results are limited to micron-sized water droplets (i.e. V0 [ 5 pL) and are not expected to hold when the droplet size (or water film thickness) is reduced to a length scale within an order of magnitude of either the Tolman length (R0 10 dT) or the mean-free-path of the vapor phase ðR0 10 kV Þ. Nevertheless, even though surface/liquid interactions directly influence the depinning process, the relative change in the evaporation efficiency before and after contact line depinning is not systematically dependent on the surface material. Further studies of p_ LG with smaller water droplets (R0 [ 2 lm) on surfaces approaching the super–hydrophilic and super–hydrophobic wetting regimes are needed. 4.4.2. Surface tension force required for contact line depinning It is instructive to estimate the energy barriers and surface-liquid forces associated with contact line depinning. Fig. 16 provides the depinning force (dF) associated with contact line depinning. The depinning force is calculated based on the unbalanced Young–Laplace equation: [56] Fig. 16. Depinning force (dF) for water droplets on different surfaces p lotted as a function of (a) the dynamic contact angle for depinning (hD/h0) and (b) the depinning contact angle (hD). dF is calculated with Eq. (11) using measured and data. The symbols with error bars correspond to the surface materials tested in this work with water microdroplets at 25 °C[ DTS [ 125 °C (260 pL[ V0 [ 2200 pL), where the symbol descriptions are provided in Figs. 12 and 14. The symbols without error bars correspond to data taken from Ref. [56] for millimeter-sized water droplets on different surface materials at room temperature (3.0 lL[ V0 [ 3.5 lL): glass/Teflon (), glass/Al (asterisks), glass/Si (purple-diamond), Si (hexagon), Paryelene (bronze-star), and Teflon (six-point-star). F ¼ ðcSL cSV Þ þ cLV cos h; ð10Þ where F is the surface tension force per unit length of contact line and cSL, cSV, and cLV are the surface tensions (energy densities) at the solid-liquid, solid-vapor, and liquid-vapor interfaces, respectively. At equilibrium, the contact angle equals the equilibrium contact angle (i.e., h = hEq) and surface tension force is zero. However, as the water microdroplet evaporates in a pinned contact line mode of evaporation, the surface tension force at the contact line builds up in direction toward the bulk liquid. At the threshold of contact line depinning, the depinning force per unit length of contact line is dF ¼ cLV ðcos hD cos h0 Þ; ð11Þ assuming that the difference between h0 and the equilibrium tact angle is small (i.e., h0 ffi hEq) and that cSL, cSV, and cLV are stants during evaporation. The effects of the contact curvature [57] (i.e., j0 = 1/R0) on dF are neglected because the tact radius is constant (pinned) during evaporation. conconline con- S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 Fig. 16 shows the depinning force calculated for each surface material tested in this study. The predicted values of dF are based on experimental data at surface temperatures within 25 °C [ DTS [ 125°C. For TS > 100 °C, a few data sets were omitted from analysis because vapor bubbles nucleated and oscillated within the evaporating microdroplets. Data analysis included the temperature dependence of cLV to account for the measurements at different surface temperatures. Also included in Fig. 16 are recent data by Shanahan et al.[56] for millimeter-sized water droplets evaporating on different surfaces at room temperature. As shown, dF is the greatest for the most hydrophilic surfaces and approaches zero as the surfaces become more and more hydrophobic (i.e., dF? 0 as hD/h0? 1). Error bars are provided in a ‘‘slope format’’ because the scatter in the data for multiple experiments is not uniformly distributed around the average value reported. Instead, the data scatter follows the slope of the error bars. For example, the depinning contact angles measured with water microdroplets on Ti/glass (red-stars) ranged within 30°[ hD[ 60°; for droplets depinning around hD = 56 ± 4° the corresponding depinning force is dFTi = 26 ± 5 mN/m, whereas for hD = 34 ± 4° the depinning force is dFTi = 42 ± 5 mN/m. Error bars are not provided for the data from Ref. [56] because multiple experiments on each surface material were not reported. Nevertheless, dF increases with increasing surface energy (hydrophobicity), supporting that dF is expected to be greater than the surface tension of water for superhydrophilic surfaces. In this regard, larger deviations are also expected between h0 measured and hEq with high surface energy surfaces because droplet spreading during impingement is also inhibited by the intrinsic energy barrier for contact line motion. dF appears to be greater for the water microdroplets in comparison to millimeter-sized water droplets. The difference in the contact line tension [57] (curvature) between millimeter-sized and micron-sized water droplets may be contributing to this trend (which is not included in the calculations of dF using Eq. (11)). 4.4.3. Excess free energy barrier for contact line depinning It is also instructive to estimate the change in the free energy required for contact line depinning. Fig. 17 provides the excess free energy per unit solid-liquid contact area (dGA) associated with 5805 contact line depinning. The free energy barrier for contact line depinning is also calculated based on Young’s equation: [58] G ffi pR2 ðcSL cSV Þ þ AcLV þ ; ð12Þ where G is the total excess free energy of a solid-liquid-vapor system, R is the droplet’s contact radius, and A is the droplet’s interfacial liquid/vapor surface area. The ellipsis term in Eq. (12) represents higher order contributions to the free energy. For small droplets (not affected by gravity) the interfacial surface area follows that for a spherical cap (see, Eq. (2)). If equilibrium is reached after droplet impingement, then the free energy is a minimum. In this case, the excess free energy at the threshold of contact line depinning is dG ffi cLV ðAD A0 Þ; ð13Þ where AD and A0 are the surface areas of the liquid-vapor interface at the threshold of depinning and at equilibrium (before evaporation), respectively. Eq. (13) describes the total excess free energy, assuming that h0 = hEq, cSL, cSV, and cLV are constant during evaporation, and that the higher order contributions to G (e.g., contact line curvature and viscous dissipation due to fluid motion at the droplet interface) can be neglected. At the threshold of depinning, the corresponding excess free energies per unit length of contact line and per unit solid-liquid surface area are dGl ¼ dG=lCL ¼ cLV ðAD A0 Þ=ð2pR0 Þ; ð14Þ dGA ¼ dG=ASL ¼ cLV ðAD A0 Þ=ðpR20 Þ; ð15Þ where the contact line length (lCL = 2pR0) and surface-liquid contact area ðASL ¼ pR20 Þ are constants until depinning. Fig. 17 shows dGA calculated for (1) each material system tested in this work (i.e., 25 °C [ DTS [ 125 °C) and (2) the millimetersized water droplets studied at room temperature in Ref. [56]. In general, dGA scatters within -5 mJ/m2 [ dGA [ -30 mJ/m2 for all material systems studied. In this regard, we estimate dGA ffi 5 mJ/m2 as a minimum value for solid-liquid-vapor systems that exhibit stick-slip contact line behavior. The magnitude of dGA may be slightly over estimated with the microdroplets because mechanical equilibrium is probably not achieved by the start of the evaporation process (e.g., only advancing contact line motion takes place after impact, so the measured values of h0 are expected to be slightly greater than hEq). Nevertheless, the contact line is initially pinned during all these evaporation experiments, so dGA is expected to be within a factor of two of that reported. The greatest variation in dGA is observed with the microdroplets evaporating on the Cu-pillar surface. This is reasonable because ASL –pR20 for the pillar surface. Future studies with different pillar geometries and surfaces approaching the super-hydrophilic and super-hydrophobic wetting regimes are needed to confirm the applicability of this result. 5. Conclusions Fig. 17. Excess free energy barrier per unit solid-liquid surface area, dGA, plotted as a function of hD/h. dGA is calculated with Eq. (15) using measured droplet-profile data (i.e., R0, AD, and A0). The symbols with error bars are for the surface materials tested with microdroplets in this work (260 pL[ V0 [ 2200 pL). The symbols without error bars are for the surface materials tested with millimeter-sized water droplets in Ref. [56] (3.0 lL[ V0 [ 3.5 lL). The symbols for each surface material are the same as that used previously. (see, the caption in Fig. 16). A comprehensive experimental investigation of water microdroplet evaporation on superheated surfaces is provided. Measurements are made with microdroplets on a variety of different surface materials. The main experimental properties/parameters measured with each material system at different temperatures _ LG Þ, evaporation rate per unit length are the evaporation rate ðm of solid-liquid-vapor contact line ðp_ LG Þ, and evaporative heat flux (q). The key findings in this work include: (1) For microdroplet evaporation, a maximum (critical) surface Þ exists that represents the transition from temperature ðT max S a purely evaporative heat transfer regime to a droplet/film 5806 S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 boiling regime (analogous to the superheat at CHF in boiling). This critical surface temperature is inversely proportional to both the droplet’s contact radius (R0) and the substrate thermal conductivity (K), where increasing R0 and/or K will reduce T max toward the saturation temperaS ture of water. _ max (2) At T max , a maximum (critical) evaporation rate ðm LG Þ is S _ max found. m is independent of droplet size and substrate LG thermal conductivity. In fact, a universal value of _ max m LG ffi 35 5 lg/s is found for all surface materials tested. For example, regardless of the properties of surface material (i.e., thermal conductivity, surface hydrophobicity, and surface microstructure), the maximum evaporation rate mea_ max _ max does sured at transition is m LG ffi 35 5 lg/s. Thus, mLG not change by increasing the substrate thermal conductivity. _ max Instead, m LG ffi 35 5 lg/s is just reached at a lower surface temperature (superheat) on a higher thermal conductivity substrate. _ max (3) Coupled to T max and m LG are the corresponding evaporative S heat flux (q) and evaporative heat transfer coefficient (h). In this work, the largest evaporative heat flux measured with water microdroplets is q = 1100 ± 100 W/cm2. This was measured with small (140 pL) microdroplets on roughened Cu substrates. The largest heat transfer coefficient, on the other hand, was measured with microdroplets evaporating on Cu substrates having a Cu-pillar surface structure, where hpil2 lar = 650 ± 200 kW/m /K at a superheat of DT 10 °C (in comparison to 50 ± 25 kW/m2/K (for different thin-films on glass substrates) and 230 ± 90 kW/m2/K (for smooth and 800-grit sanded Cu substrates)). These results support that maximizing the solid-liquid-vapor contact line length per unit surface area is most important parameter for enhancing the evaporative heat transfer performance (e.g., q>1000 W/ cm2 and h>600 kW/m2/K), which can be achieved by simply spraying tiny water microdroplets on microstructured surfaces. (4) For all droplet-surface systems studied, stick-slip contact line behavior is observed during evaporation. As expected, the surface energy controls the contact line dynamics, where the magnitude of the contact line depinning force is directly proportional to surface energy. This is supported by comparison of data with free energy and contact line depinning force calculations for all material systems studied, where fair agreement is found between the microdroplets studied here and millimeter-sized water droplets studied by others. Due to stick-slip contact line motion, droplets evaporate in a pinned contact line mode until a critical contact angle (depinning force) is reached for contact line motion. Before _ LG is a constant and is contact line depinning (slippage), m directly proportional to the surface temperature, contact radius, and substrate thermal conductivity, whereas, after depinning, the evaporation rate reduces in magnitude due to a reduction in the contact line circumference (i.e., 2pR). Thus, even though changes in surface energy can dramatically change the depinning dynamics, changes in surface energy, however, has no systematic influence on the ratio of the evaporation efficiency before and after contact line depinning, suggesting that phase-change heat transfer is dominated by the evaporation process at the solid-liquidvapor contact line. Acknowledgements This material is based on research sponsored by U.S. Air Force Office of Scientific Research under Grant No. 2303BR5P. The authors are thankful to Mike Check, Chris Murtatore, John Ferguson, Andrey Voevodin, and Chad Hunter for their help and frequent technical discussions. The authors thank Arthur Safriet for technical contributions during design and construction of the experimental apparatus. The authors are also grateful to HPCMO for the computing resources at the AFRL/DSRC. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of U.S. Air Force Office of Scientific Research or the U.S. Government. References [1] I. Mudawar, Assessment of high-heat-flux thermal management schemes, IEEE Trans. Comput. Packag. Technol. 24 (2001) 122–141. [2] J. Kim, Spray cooling heat transfer: the state of the art, Int. J. Heat Fluid Flow 28 (2007) 753–767. [3] I. Mudawar, S. Ujereh, T. Fisher, Effects of carbon nanotube arrays on nucleate pool boiling, Int. J. Heat Mass Transfer 50 (2007) 4023–4038. [4] R. Chen, M.C. Lu, V. Srinivasan, Z. Wang, H.H. Cho, A. Majumdar, Nanowires for enhanced boiling heat transfer, NanoLetters 9 (2009) 548–553. [5] J.S. Ahn, N. Sinha, M. Zhang, D. Banerjee, S. Fang, R.H. Baughman, Pool boiling experiments on multiwalled carbon nanotube (MWCNT) forests, J. Heat Transfer 128 (2006) 1335–1342. [6] V. Khanikar, I. Mudawar, T. Fisher, Effects of carbon nanotube coatings on flow boiling in a micro-channel, Int. J. Heat Mass Transfer 52 (2009) 3805–3817. [7] S.V. Garimella, Advances in mesoscale thermal management technologies for microelectronics, Microelectron. J. 37 (2006) 1165–1185. [8] S. J Thiagarajan, S. Narumanchi, C. King, W. Wang, and R. Yang, ‘‘Enhancement of heat transfer with pool and spray impingement boiling on microporous and nanowire surface coatings’’. in: Proceedings 14th International Heat Transfer Conference, Washington, D.C., August 8–13, 2010. [9] J. Wendelstorf, K.-H. Spitzer, R. Wendelstorf, Spray water cooling heat transfer at high temperatures and liquid mass fluxes, Int. J. Heat Mass Transfer 51 (2008) 4902–4910. [10] J.H. Kim, S.M. You, S.U.S. Choi, Evaporative spray cooling of plain and microporous coated surfaces, Int. J. Heat Mass Transfer 47 (2004) 3307–3315. [11] V. Sathyamurthi, H.-S. Ahn, D. Banerjee, S.C. Lau, Subcooled pool boiling experiments on horizontal heaters coated with carbon nanotubes, J. Heat Transfer 131 (2009) 071501. [12] R. Spolenak, S. Gorb, H. Gao, E. Arzt, Effects of contact shape on the scaling of biological attachments, Proc. Roy. Soc. Lond., Ser. A (Math., Phys. Eng. Sci.) 461 (2005) 305–319. [13] E. Arzt, R. Spolenak, S. Gorb, Adhesion design maps for bio-inspired attachment systems, Acta Biomater. 1 (2005) 5–13. [14] F. Brochard-Wyart, J.M. di Meglio, D. Quere, G.P. de Gennes, Spreading of nonvolatile liquids in a continuum picture, Langmuir 7 (1991) 335–338. [15] Z. Li, X. Zhang, L. Zhang, X. Zeng, J. Hu, H. Fang, Analysis of the gas states at a liquid/solid interface based on interactions at the microscopic level, J. Phys. Chem. B 111 (2007) 9325–9329. [16] A.M. Briones, J.S. Ervin, S.A. Putnam, L.W. Byrd, L. Gschwender, Micrometersized water droplet impingement dynamics and evaporation on a flat dry surface, Langmuir 26 (2010) 13272–13286. [17] A.M. Briones, J.S. Ervin, L.W. Byrd, S.A. Putnam, A. White, and J.G. Jones, in: 42nd AIAA Thermophysics Conference, Honolulu, HI, June 27–30 2011. [18] A.M. Briones, J.S. Ervin, L.W. Byrd, S.A. Putnam, A. White, J.G. Jones, ‘‘Evaporation characteristics of pinned water microdroplets’’, J. Thermophys. Heat Transfer 26 (2012). [19] R. Bhardwaj, J.P. Longtin, D. Attinger, ‘‘Interfacial temperature measurement, high-speed visualization and finite-element simulations of droplet impact and evaporation on a solid surface’’, Int J. Heat Mass Transfer 53 (2010) 3733– 3744. [20] R.D. Deegan, Pattern formation in drying drops, Phys. Rev. E 61 (2000) 475– 485. [21] F. Duan, C.A. Ward, Investigation of local evaporation flux and vapor-phase pressure at an evaporative droplet interface, Langmuir 25 (2009) 7424–7431. [22] F. Girard, M. Antoni, K. Sefiane, On the effect of marangoni flow on evaporation rates of heated water drops, Langmuir 24 (2008) 9207–9210. [23] Y. Hamamoto, J.R.E. Christy, K. Sefiane, Order-of-magnitude increase in flow velocity driven by mass conservation during the evaporation of sessile drops, Phys. Rev. E 83 (2011) 051602. [24] J.B. Lee, S.H. Lee, Dynamic wetting and spreading characteristics of a liquid droplet impinging on hydrophobic textured surfaces, Langmuir 27 (2011) 6565–6573. [25] P.L. Kelly-Zion, C.J. Pursell, S. Vaidya, J. Barta, Evaporation of sessile drops under combined diffusion and natural convection, Colloids Surf. A: Physicochem. Eng. Aspects 381 (2011) 31–36. [26] H. Gelderblom, A.G. Marin, H. Nair, A. van Houselt, L. Lefferts, J.H. Snoeijer, D. Lohse, How water droplets evaporate on a superhydrophobic substrate, Phys. Rev. E 83 (2011) 026306. [27] H. Hu, R.G. Larson, Analysis of the effects of marangoni stresses on the microflow in an evaporating sessile droplet, Langmuir 21 (2005) 3972–3980. S.A. Putnam et al. / International Journal of Heat and Mass Transfer 55 (2012) 5793–5807 [28] L.Y. Barash, T.P. Bigioni, V.M. Vinokur, L.N. Shchur, Evaporatioin and fluid dynamics of a sessile drop of capillary size, Phys. Rev. E 79 (2009) 046301. [29] V. Starov, K. Sefiane, On evaporation rate and interfacial temperature of volatile sessile drops, Colloids Surf. A: Physicochem. Eng. Aspects 333 (2009) 170–174. [30] X. Xu, J. Luo, D. Guo, Criterion for reversal of thermal marangoni flow in drying drops, Langmuir 26 (2010) 1918–1922. [31] McMaster Carr, www.mcmaster.com, 600 N. Country Line Rd., Elmhurst, IL 60126, USA. (PN: 1357T11). The FS windows have dimensions of 25.4 mm in diameter and 1.7 mm in thickness. [32] Hoowaki, LLC., www.hoowaki.com, 511 Westinghouse Rd., Pendleton, SC 29670, USA. [33] MicroFab Technologies, Inc., www.microfab.com, 1104 Summit Ave., Suite 140, Plano, TX 75074, USA. [34] Vision Research, Inc., www.visionresearch.com, 100 Dey Rd., Wayne, NJ 07470, USA. Model: Phantom V12. [35] S.A. Putnam, L.W. Byrd, A.M. Briones, M.S. Hanchak, J.S. Ervin, J.G. Jones, ‘‘Role of entrapped vapor bubbles during microdroplet evaporation’’ Appl. Phys. Lett. Accepted for publication (2012). [36] R.W. Schrage, A Theoretical Study of Interphase Mass Transfer, Columbia University Press, New York, 1953. [37] C.W. Hirt, B.D.J. Nichols, Volume of fluid (VOF) method for the dynamics of free boundaries, J. Comput. Phys. 39 (1981) 201–225. [38] ANSYS FLUENT 12.0, Theory Guide (2009). [39] R. Marek, J. Straub, Analysis of the evaporation coefficient and the condensation coefficient of water, Int. J. Heat Mass Transfer 44 (2001) 39–53. [40] H.K. Dharaleswarapu, J.Y. Murthy, S.V. Garimella, Numerical investigation of an evaporating meniscus in a channel, Int. J. Heat Mass Transfer 55 (2012) 915–924. [41] R.I. Issa, Solution of the implicitly discretised fluid flow equations by operatorsplitting, J. Comput. Phys. 62 (1986) 40–65. [42] R. Mollaret, K. Sefiane, J.R. Christy, D. Veyret, Experimental and numerical investigation of the evaporation into air of a drop on a heated surface, Chem. Eng. Res. Design 82 (2004) 471–480. [43] R.C. Tolman, Effect of droplet size on surface tensión, J. Chem. Phys. 17 (1949) 333–343. [44] D.S. Golovko, H.-J. Butt, E. Bonaccurso, Transition in the evaporation kinetics of water microdroplets on hydrophilic surfaces, Langmuir 25 (2009) 75–78. 5807 [45] J.D. Bernardin, I. Mudawar, The Leidenfrost point: experimental study and assessment of existing models, J. Heat Transfer (Trans. ASME) 121 (1999) 894– 903. [46] G.J. Dunn, S.K. Wilson, B.R. Duffy, K. Sefiane, The strong influence of substrate conductivity on drople evaporation, J. Fluid Mech. 623 (2009) 329–351. [47] M. Shoji, Studies of boiling chaos: a review, Int. J. Heat Mass Transfer 47 (2004) 1105–1128. [48] J.G. Leidenfrost, On the fixation of water in diverse fire, Int. J. Heat MassTransfer 9 (1966) 1153–1166. [49] G.J. Dunn, S.K. Wilson, B.R. Duffy, K. Sefiane, Evaporation of a thin droplet on a thin substrate with a high thermal resistance, Phys. Fluids 21 (2009) 052101. [50] S. David, K. Sefiane, L. Tadrist, Experimental investigation of the effect of thermal properties of the substrate in the wetting and evaporation of sessile drops, Colloids Surf. A: Physicochem. Eng. Aspect. 298 (2007) 108–114. [51] Static contact angles measured with thin-film samples (25 °C [ TS [ 95 °C): Cu Al SAM hTi ffi 113 20 (in order from 0 ffi 79 60; h0 ffi 91 30; h0 ffi 95 50, and h0 most hydrophilic to most hydrophobic). Average receding (depinning) contact angles measured (25 °C [ TS [ 95 °C): AlðcÞ AlðpÞ hCu ffi 38 190; hTi ffi 66 160, and 0 ffi 29 70; h0 D ffi 45 130; h0 SAM h0 ffi 101 50. [52] G. Li, S.M. Flores, C. Vavilala, M. Schmittel, K. Graf, Evaporation dynamics of microdroplets on self-assembled monolayers of dialkyl disulfides, Langmuir 25 (2009) 13438–13447. [53] M.K. Bernett, L.A. Halper, N. Jarvis, T.M. Thomas, Effect of adsorbed monomolecular films on the evaporation of volatile organic liquids, I&EC Fundamentals 1 (9) (1970) 150. [54] T. Deng, K.K. Varanasi, M. Hsu, N. Bhate, C. Keimel, J. Stein, M. Blohm, Nonwetting of impinging droplets on textured surfaces, Appl. Phys. Lett. 94 (2009) 133109. [55] L. Gao, T.J. McCarthy, How Wenzel and Cassie were wrong, Langmuir 23 (2007) 3762–3765. [56] D. Oregon, K. Sefiane, M.E.R. Shanahan, Stick-slip of evaporating droplets: substrate hydrophobicity and nanoparticle concentration, Langmuir 27 (2011) 12834–12843. [57] T. Pompe, S. Herminghaus, Three-phase contact line energetics from nanoscale liquid surface topographies, Phys. Rev. Lett. 85 (2000) 1930–1933. [58] M.E.R. Shanahan, Simple theory of stick-slip wetting hysteresis, Langmuir 11 (1995) 1041–1043.
© Copyright 2026 Paperzz