pubs.acs.org/Langmuir © 2009 American Chemical Society
An Innovative Method to Control the Incipient Flow Boiling through Grafted Surfaces with Chemical Patterns R. Rioboo,*,† M. Marengo,‡ S. Dall’Olio,‡ M. Voue,† and J. De Coninck† †
Laboratory of Physics of Surfaces and Interfaces, University of Mons, Parc Initialis, Av. Copernic, 1, B-7000 Mons, Belgium, and ‡Faculty of Engineering, University of Bergamo, Viale Marconi 5, 24044 Dalmine, Italy Received February 6, 2009. Revised Manuscript Received April 17, 2009
The onset of flow boiling of a liquid is linked to the superheat condition that is necessary to activate the nucleation sites on contacting surfaces. The nucleation sites are usually represented by cavities in the rough surface of the heat exchanger. On smooth surfaces, the region where bubble detachment does not occur due to the lack of superheating may constitute a serious limitation for microfluidic devices. This paper shows the first experimental evidence that the position of the active nucleation sites can be controlled through chemical patterning of smooth surfaces: in this study, the heated surfaces are chemically grafted with alkylsilane self-assembled monolayers by microcontact printing. The analysis of the propagation of the bubble zone area quantitatively shows that the bubbles remain localized on top of the grafted zone and that, in the initial phase of the experiment, the center of mass of the bubble zone only moves along the vertical axis, without lateral drift.
The increase in functionality and power consumption of electronic components and microdevices creates a problem: heating. This heat energy has to be removed in order to ensure the reliable working of these devices, without negatively affecting their performance. One of the techniques used in electronics and microfluidic devices utilizes so-called “phase-change cooling systems” (PCCS). Among possible PCCS, the heat pipe, for example, has the advantage of being “passive”, yielding a very high degree of long-term reliability.1,2 In each PCCS, there is a zone where liquid, solid, and gas are all present together, allowing surface features, such as wettability and roughness, to play an important role.3,4 Much of the thermal effectiveness of PCCS is related to the characteristics of their evaporation and condensation zones, their length, the thickness of the thin evaporating film, and the dynamic contact angle.5 Fixing the liquid/solid/vapor contact line to predefined positions may, then, help to enhance the thermal efficiency of PCCS. This can be performed by means of heterogeneities of the surface.6 The work presented here studies the novel possibility of controlling the onset position of incipient flow boiling of a liquid during its passage through a heated smooth channel by chemically grafting specific wettability patterns. Such a specific, highly resistant surface modification results in surfaces without cavities at micro- and submicrometer scales. The onset of flow boiling of a liquid is linked to the so-called “superheat condition” which is necessary to activate the nucleation sites on a surface. It is known that any topographical heterogeneity of the solid surface may act as a nucleation site.7 *Corresponding author.
[email protected].
(1) Kakac-, S.; Bergles, A. E.; Oliveira Fernandes, E. Two-Phase Flow Heat Exchangers: Thermal-Hydraulic Fundamentals and Design (NATO Science Series E); Springer: Berlin, 1988. (2) Kakac-, S. Boilers, Evaporators, and Condensers; Wiley: New York, 1991. (3) Wang, T. A.; Reid, R. L. ASHRAE Transactions: Research 1996, 102, 427– 433. (4) Kim, S. J.; Bang, I. C.; Buongiorno, J.; Hu, L. W. Appl. Phys. Lett. 2006, 89, 153107. (5) Wen, D. S.; Wang, B. X. Int. J. Heat Mass Transfer 2002, 45, 1739–1747. (6) Basu, N.; Warrier, G. R.; Dhir, V. K. J. Heat Transfer-Trans. ASME 2002, 124, 717–728.
Langmuir 2009, 25(11), 6005–6009
Since the smaller the cavities, the higher the superheat, phase changes upon smooth surfaces, such as the ones encountered in thermal microdevices, demand a higher superheating to occur. Thus, thermal efficiency is decreased. Control of the location of the incipient boiling is therefore crucial to bypass this drawback. When a liquid flowing in a channel is heated via a diabatic solid surface, it will dissipate the heat by convection. With high heat fluxes, the liquid will boil by forming bubbles starting from nucleation sites. The formation of the vapor bubbles is not immediately linked to the attainment of the saturation temperature of the liquid at the given pressure, but it is necessary to have a superheat condition to initiate the bubbles.2 This means that there is a separation between the position where the flowing liquid reaches the equilibrium saturation temperature and the position where the first bubbles appear. This distance, named the incipient boiling distance or the incipient boiling length (Figure 1b), may vary according to changes in various experimental parameters. Yet in 1962, Hsu8 founded that the activation of the nucleation sites on smooth surfaces was only possible for large superheats. Similarly, in 1988, Bar-Cohen9 confirmed that highly wetting liquids flooded all but the smallest cavities, hence depleting the vapor embryos needed for boiling inception. In other words, a higher incipient boiling superheat is required to activate smaller cavities and to initiate boiling (Figure 1). Once very small vapor bubbles obtain enough energy to nucleate and to subsequently separate from the cavities, boiling takes place. Therefore, the incipient boiling length is a measure of vapor bubble deactivation on the heated surface. As a consequence, an increase in incipient boiling length is accompanied by lower heat transfer rates. This means that in microdevices, for example, difficulty in activating smooth surfaces will lead to poor heat transfer efficiencies, since a large part of the channels are in the region of the nonboiling condition. In the case of pool boiling, i.e., when the liquid is at rest and heated, the wettability effects are well-known, even if not fully (7) Bergles, A. E.; Rohsenow, W. M. J. Heat Transfer 1964, 86, 365–370. (8) Hsu, Y. Y. J. Heat Transfer 1962, 34, 207–214. (9) Bar-Cohen, T.W. S. Heat Transfer Eng. 1988, 9, 19–31.
Published on Web 4/30/2009
DOI: 10.1021/la900463b
6005
Letter
Figure 1. The physical meaning of the incipient boiling length is sketched in the figure, since there is a distance between the position in the channel where the liquid has reached the saturation temperature and the position of the actual formation of bubbles. In such incipient boiling length, the heat transfer is low and no bubbles are activated from the cavities.
quantitatively described. In any case, the wettability issues and in particular the effect of the local static contact angle on boiling are always linked to the presence of cavities.10,11 The fact that chemical heterogeneities on smooth surfaces may influence the boiling onset in a liquid flow was up to now, to the best of our knowledge, not considered in the literature. The common use of high surface energy solid materials (metals) and mostly low surface tension liquids that completely wet the solids has discouraged research linking two-phase heat exchange to the control of surface hydrophobic properties, such as their pattern. Thomas et al.12 used different self-assembled monolayers (SAMs) and showed that the boiling behavior of a liquid during fast transient heating events in a pool was a function of the wettability of the solid surface. The nucleation temperatures were lower with hydrophobic SAMs on solid surfaces.13 In another paper, Takata et al.14 showed that superhydrophilic TiO2 surfaces characterized by zero static contact angle presented a higher critical heat flux than a surface with a 20° static contact angle under conditions of pool boiling15 as predicted by Kandlikar.16 In the case of flow boiling, the influence of wettability, through the use of surfactant solution, has been addressed by Jeong et al.17 They experimentally demonstrated that wettability had indeed an important role, but its influence was also a function of the mass flux. In 2003, Hibiki and Ishii18 showed that the number (10) Wang, C. H.; Dhir, V. K. J. Heat Transfer-Trans. ASME 1993, 115, 659– 669. (11) Yang, S. R.; Kim, R. H. Int. J. Heat Mass Transfer 1988, 31, 1127–1135. (12) Thomas, O. W.; Cavicchi, R. E.; Tarlov, M. J. Langmuir 2003, 19, 6168– 6177. (13) Balss, K. M.; Avedisian, C. T.; Cavicchi, R. E.; Tarlov, M. J. Langmuir 2005, 21, 10459–10467. (14) Takata, Y.; Hidaka, S.; Cao, J. M.; Nakamura, T.; Yamamoto, H.; Masuda, M.; Ito, T. Energy 2005, 30, 209–220. (15) Sefiane, K.; Benielli, D.; Steinchen, A. Colloids Surf., A 1998, 142, 361–373. (16) Kandlikar, S. G. J. Heat Transfer-Trans. ASME 2001, 123, 1071–1079. (17) Jeong, Y. H.; Sarwar, M. S.; Chang, S. H. Int. J. Heat Mass Transfer 2008, 51, 1913–1919. (18) Hibiki, T.; Ishii, M. Int. J. Heat Mass Transfer 2003, 46, 2587–2601.
6006
DOI: 10.1021/la900463b
Figure 2. (a) Scheme of the experimental setup. The channel is closed on its both sides, below and top. (b) Schematic top view of a heated channel with an example of “U” shape for the patterned zone.
of nucleation sites was a function of the static contact angle of the liquid on the solid surface. Finally, Agrawal et al.19 investigated the presence of nanobubbles on nanopatterned surfaces by atomic force microscopy. They showed that, under isothermal condition, their number is larger on hydrophobic surfaces than on hydrophilic ones. The extent to which these nanobubbles play an active role in the onset of the boiling phenomenon remains an open question. The present work is the first successful experimental demonstration that not only can chemical patterning of the exchanger surface lead to bubble formation, but it may result in controlling the position of the nucleation sites and may therefore be considered an appropriate way to change the incipient flow boiling length. Such techniques as surface patterning with SAMs20 may be of benefit for small and smooth surfaces by increasing the density of vapor embryos. A flow boiling experiment with low mass flux was considered as a first experiment. Figure 2 presents the experimental setup. Deionized Milli-Q water was used after degassing using helium bubble flow in a closed reservoir. The water was heated to 90 °C and then brought to the heated channel using a magnetic gear pump (ColeParmer). The tubes providing the water from the reservoir to the gear pump and from the pump to the heated channel were insulated with polystyrene to reduce the thermal loss and to keep the temperature of the fluid entering the test section as close as possible to that of the reservoir. The water temperature
(19) Agrawal, A.; Park, J.; Ryu, D. Y.; Hammond, P. T.; Russell, T. P.; McKinley, G. H. Nano Lett. 2005, 5, 1751–1756. (20) Rioboo, R.; Marengo, M.; Dall’Olio, S.; Voue, M.; De Coninck, J. Devices and method for enhanced heat transfer. European Patent Application EP 07113887.9-1266, August 6, 2007.
Langmuir 2009, 25(11), 6005–6009
Letter
was measured at the exit of the channel using a simple K-type microthermocouple. The channel itself was heated from below with a Chromalox WS-605 heat source providing 500 W heat at its surface (approximately 120 mm 60 mm). The heater was regulated by a power controller (Watlow DIN-A-MITE), a temperature controller (Watlow EC), and a J-type thermocouple. The temperature controller was connected to the heater and to the thermocouple. The thermocouple was used to measure the temperature of the surface of the heater and to compare it with the preset value. The test section comprised a 2-mm-thick aluminum plate provided with a cavity to fit the patterned surfaces. The size of the cavity was adjusted to the size of optical microscope glass slides used for the patterning, (i.e., 1 mm thick, 75 mm 25 mm). The plate further comprised a slit to allow the filling of the channel with the incoming water. The channel walls were created using 0.6-mm-thick Teflon spacers, and a glass plate provided the top of the channel, allowing visualization of the flow. The test section was fixed on the heater with high-conductivity paste in order to ensure good contact and uniform heat flux between the two parts. The patterned surface which comprised hydrophilic and hydrophobic areas was fixed in the cavity using high-conductivity paste. The overall channel dimensions were height 0.6 mm, width 18 mm, and length ∼100 mm. The water leaving the channel was not recirculated. Two types of patterned surfaces were used: a glass plate and a silicon plate cut to the same dimensions on which an octadecyltrichlorosilane (OTS) monolayer was grafted by microcontact printing21,22 in defined zones. The surfaces were activated as follows: each surface was rinsed twice in an ultrasonic bath of chloroform for 5 min; exposed to UV/O3 for 30 min to remove all possible organic contaminants; immersed in a piranha solution (H2O2/H2SO4 30:70 v:v), and finally rinsed with Milli-Q water and dried under nitrogen. Microcontact printing was performed immediately after activation of the surface in a low humidity chamber (RH: 6%). A polydimethylsiloxane (PDMS) stamp was used, dipped in an OTS solution (10 mM in hexane). The PDMS stamp was brought to contact with the activated solid surface for 30 s.22 In these experiments, the OTS stamped areas were either a band (width of approximately 7 mm) oriented transversally with respect to the flow direction or a U-shaped area. The wettability of the surface was measured using the sessile drop method23 after several heating and cooling runs. The parts that will be referred to as “hydrophobic” parts had advancing and receding static contact angles of θadv = 107.3° ( 9.2° and θrec = 80.3° ( 8.7°. The rest of the plate, the one that will be referred to as “hydrophilic” parts, had advancing and receding static contact angles of θadv = 94.8° ( 1.8° and θrec = 55.5° ( 8.5°. The high value of these angles can be explained by the fact that the highenergy surface received, during several heating-cooling processes and hours of experiments, all the possible contaminants of the circuit. Nevertheless, it was still possible to measure a wettability contrast of approximately 23° between the hydrophobic zones and the more hydrophilic ones. The characteristics of each experiment were completely defined by fixing three variables: the volumetric flow rate of water, the water inlet temperature, and the heater surface temperature. During the experiments, these three parameters were adjusted in order to localize the position of boiling onset to within a few (21) Kumar, A.; Whitesides, G. M. Appl. Phys. Lett. 1993, 63, 2002–2004. (22) Jeon, N. L.; Finnie, K.; Branshaw, K.; Nuzzo, R. G. Langmuir 1997, 13, 3382–3391. (23) Johnson R. E.; Dettre, R. H. In Wettability Berg, J. C. Ed.; Marcel Dekker: New York, 1993; p 1.
Langmuir 2009, 25(11), 6005–6009
centimeters of the entrance of the channel for the two volume flow rates tested (2.2 cm for the 20 mL/min mass flow rate and 2.8 cm for the 25 mL/min). In both cases, the position of the boiling onset was almost the same and coincided with the position of the hydrophobic area. The temperature of the heater was fixed at 110 ( 1 °C, and we confirmed a uniform temperature distribution along the test section. Experiments were performed at two volume flow rates (resulting in liquid mean velocities of 0.0309 and 0.0386 m/s) and with the two types of surface (glass slides and silicon wafers) to test the possible influence of these parameters. The experimental parameters are summarized in Table 1. A first test was performed on a glass surface without any hydrophobic area. No particular pattern on the appearance of the bubbles and their localization was observed. This was considered the negative control experiment. Figure 3 represents the time evolution of the appearing of boiling over the patterned surfaces. Each image is a snapshot of Table 1. Experimental Parameters for the Incipient Flow Boiling Experiments case
A
B
C
surface volume flux (mL/min)
glass 20
glass 25
silicon 25
Figure 3. Bubble evolution over time from an arbitrary zero time (top) until 75.9 s (bottom). In each image the flow is going from right to left. Three cases (A, B, and C) are represented. The hydrophobic zone is presented in the first image of each sequence: it is the zone between the two white, dashed lines. The volume flux and surface is varied in the following way. Volume fluxes are for (A), 20 mL/min; (B), 25 mL/min; (C), 25 mL/min. Grafted surfaces are for (A,B), glass; and for (C), silicon. All images are taken at the same magnification; on the (C) sequence vertical lines spaced by one millimeter are present on the images. Images represent an area of 24.9 mm 12.4 mm. The scale bar is 5 mm long. DOI: 10.1021/la900463b
6007
Letter
Figure 4. (a) Time evolution of bubble zone area (expressed in percent of the SAM-grafted area). Continuous red curve: total area of the bubble zone. Dashed green curve: area of the boiling zone in contact with the SAM-grafted surface. (b) Trajectory of the center of mass of the bubble zone. Dashed black lines: limit of the SAMgrafted zone. Points 1 to 4 are described in the text.
the channel, taken from above. On each image, the liquid is going from right to left. For each case, the zone where bubbles appear, the “bubble zone”, is clearly expanding over the hydrophobic parts independently of the direction of the flow. The three cases presented on this figure clearly show the effect of the wettability contrast on the evolution of boiling. The cycles were performed at least 5 times to ensure the reproducibility of the localized boiling effect and the thermal stability of the chemical patterns. It should be stressed that the liquid present behind the expanding bubble zone (left side of the images) received more heating than the liquid in the grafted zone. Despite this, the shape of the bubble zone still coincides with that of the grafted zone. It can be seen that at 40.2 s after the first visual appearance of bubbles, for the straight zones (A and B in Figure 3), the bubble is confined to the grafted zone. The silicon surface gave similar results in the straight zone. In order to check that indeed the bubble appearing is following the hydrophobic zone we decided to use another shape (a U-shape), which is unsymmetrical to the flow direction. For the U-shaped zone (C in Figure 3), the expanding bubble follows at least half of the grafted circuit until 75.9 s after the first bubble formation. Let us now consider some quantitative aspects of the propagation of the bubbles. The series of images corresponding to the experimental case A (Table 1) was analyzed using the Compix SimplePCI image analysis software (v 5.1, Hamamatsu Corp.), and the area of the bubble zone was recorded as a function of the time. These results are presented in Figure 4a. The area covered by the bubbles is made dimensionless by dividing its value by the value of the SAM-grafted area. Two analyses were considered. In the first one, the whole images were analyzed, showing that the bubble zone grows beyond the SAM surface at least for t > 55 s (continuous red curve). At point 3, the area remains constant over 6008
DOI: 10.1021/la900463b
15 s and rapidly vanishes as soon as the liquid flow is rapidly increased to flush out the experimental cell (point 4). In the second case, the series of images has been reanalyzed in the same way but considering only the part of the bubble zone in contact with the SAM (dashed green curve). From point 1 to point 2, both curves (almost) coincide, confirming the visual result according to which the bubbles remain confined above the SAM-grafted surface. At point 2, 90% of the SAM-grafted zone is covered by the bubbles, and the bubble zone starts to grow outside the SAM zone. Two other parameters have also been analyzed: the position along the X and Y axes of the center of mass of the bubble zone (Figure 4b). The trajectory has been represented in two parts: the first one (from point 1 to point 2) corresponds to the displacement of the center of mass inside the SAM area (continuous blue curve). This part of the trajectory confirms that the bubbles remain confined above the SAM zone and that the X position of the center of mass only deviates by less than 4%. In the meanwhile, the Y position increases from the bottom of the SAM zone to approximately reach the mid-position of it. During that part of the experiment, the edges of the bubble zone are trapped by the SAM borders. The second part of the trajectory (dashed blue curve) between points 2 and 4 is characterized by a bubble zone which spreads to the left of the SAM zone. After point 4, the whole channel was flushed by an increase of volume flow rate in order to pass to another test. Subsequently, the bubble is removed from the grafted zone. Clearly, the SAM can affect the distribution of the heat on the surface by enhancing liquid-vapor phase change on the grafted zone. Such behavior is unexpected, since on the surfaces we have considered there were no microcavities and the roughness was kept very low (typically, the mean roughness amplitude of silicon wafers and microscope glass slides is below 1 nm24). A possible explanation for such behavior is the presence of nanobubbles, already proven in isothermal conditions by different authors.19,25 It has been shown that the density of nanobubbles on hydrophobic patterns is higher than on hydrophilic zones.19 It can therefore be supposed that the triggering of boiling is enhanced by the higher nanobubble density. The U-shaped experiment indeed shows that the phase change is strongly affected by the wettability contrast, which reinforces the hypothesis that the phase change is due to nanobubble density differences. We believe this study should be the first of many experiments designed to quantify the wettability effect in terms of the local heat flux and heat distribution on the solid surface as a function of the wettability contrast. Another important benefit is the simplicity of the method compared with the usual control of the surface roughness and topographical elements. In conclusion, the incipient boiling length, which results from the superheating necessary for the activation of the nucleation sites, is an important issue for the overall efficiency of small thermal devices, such as microheating exchangers and microheating pipes. In such cases and generally for smooth surfaces, the incipient boiling length may have the same length scale as the microchannels, resulting in poor heat transfer rates compared with devices where the heat transfer channels have much higher length scales. The present work provides a first qualitative and quantitative confirmation that the use of chemical grafting to produce patterns through microcontact printing on solid smooth surfaces can control the position of the incipient boiling. The silicon and glass surfaces used have no cavities bigger than a few nanometers and the coating is of molecular thickness (at most, a (24) Rioboo, R.; Marengo, M.; Tropea, C. Atomization Sprays 2001, 11, 155– 165. (25) Cavicchi, R. E.; Avedisian, C. T. Phys. Rev. Lett. 2007, 98, 124501.
Langmuir 2009, 25(11), 6005–6009
Letter
few nm): this means that our results are not due to the standard nucleation sites usually linked to roughness and/or micrometric surface defects. We suggest that the phase change is due to the local density of nanobubbles, which is dependent on the hydrophobicity of the surface. The global heat flux on the solid surface can then be strongly affected by the chemical heterogeneities and wettability contrast. Similar results coming from experiments on materials with different heat conductivity and using different volume flow rates invite the possibility to generalize this effect to other cases. The proposed methodology may be applied to
Langmuir 2009, 25(11), 6005–6009
an enormous number of applications, from boiling to cooling, where multiphase thermo-fluid-mechanical phenomena are the controlling processes. Acknowledgment. Marco Marengo and Stefano Dall’Olio were supported through Italian National funding projects, PRIN2005 and PRIN2007, “Two-phase flows in micro- and mini-channels” led by Prof. Marco Spiga, University of Parma. This work is partially supported by the Ministere de la Region Wallonne and the Belgian Funds for Scientific Research (FNRS).
DOI: 10.1021/la900463b
6009