Thin film evaporation on microstructured surfaces is a promising strategy for high heat flux thermal management. To enhance fundamental understanding and optimize the overall heat transfer performance across a few microns thick liquid film, however, requires detailed thermal characterizations. Existing characterization techniques using infrared thermometry or contact-mode temperature sensors such as thermocouples and resistance temperature detectors cannot accurately measure the temperature of the thin liquid film near the three-phase contact line due to the restriction of low spatial resolution or temperature sensitivity. In this work, we developed a non-contact, in situ temperature measurement approach using a custom micro-Raman spectroscopy platform which has a spatial resolution of 1.5 μm and temperature sensitivity within 0.5 °C. We utilized this method to characterize thin film evaporation from fabricated silicon micropillar arrays. We showed that we can accurately measure the local thin film temperature and map the overall temperature distribution on the structured surfaces at different heat fluxes. We investigated the effects of micropillar array geometries and showed that the temperature rise of the liquid was reduced with the decreasing micropillar pitch due to the increased fraction of the thin film area. This work offers a promising method with micro-Raman to quantify phase change heat transfer on microstructured surfaces. This characterization technique can significantly aid mechanistic understanding and wick structure optimization for various phase-change based thermal management devices.
Evaporation is a common phenomenon in nature and widely used in industrial applications. It plays an increasingly important role in thermal management,1 water purification,2 humidification,3 and vapor generation.2,4 It has been widely reported that the heat transfer of evaporation can be significantly enhanced by micro/nanostructures.5–9 For this reason, thin film evaporation on microstructured surfaces has attracted particular interest for high heat flux thermal management, especially for cooling high-performance electronic devices6–9 (e.g., micro-processors1 and high-power radio-frequency amplifiers10,11 with highly concentrated heat generation >100 W/cm2).
Several unique properties highlight the advantage of thin film evaporation. First, the large enthalpy of vaporization enables efficient heat dissipation.5,8 In addition, the microstructures create liquid-vapor interfaces that generate high capillary pressure during thin film evaporation, allowing for passive liquid transport5,6,8 [Fig. 1(a)]. Furthermore, near the three-phase contact line, there is a several micron thin liquid film [defined as the thin film region in this work; see the red dashed box of the zoomed in unit cell in Fig. 1(a)].5–8 The heat transfer across this liquid thin film is of very low thermal resistance, which has been shown to account for most of the heat dissipation during evaporation.12
Measuring the temperature of this thin film region is, therefore, key to understanding the fundamental enhancement mechanisms and optimization of the structure geometry. However, directly probing local temperature of the thin film region is difficult using contact-based resistance temperature detectors (RTDs)/thermocouples or infrared thermometry.8,13 Specifically, contact-mode sensors are usually placed on the backside of the sample [Fig. 1(a)] and thus can only characterize remotely the thermal effects averaged over several millimeters rather than locally on the front structured surface.8 On the other hand, infrared thermometry, while capable of providing some temperature information of the front side, has poor spatial resolutions (∼10–30 μm)14,15 and is subject to large temperature uncertainties (∼5 °C) when measuring the microscale thermofluidic system.16 The other weakness of infrared thermography is that it does not work well at temperatures below 50 °C or 70 °C (as there are not enough photons emitted in the 3–5 μm band at these low temperatures).
In this work, we developed a non-contact, in situ temperature measurement approach using micro-Raman thermometry to investigate thin film evaporation from microstructured surfaces. Due to the superior spatial resolution (≈1 μm) and high temperature sensitivity (≈0.5 °C), micro-Raman thermometry has been widely used to measure the localized temperature rise11,12 for various semiconductor materials17–19and microelectronic devices.20,21 Meanwhile, the capability of micro-Raman in studying phase change heat transfer has not yet been demonstrated. Here, we used silicon micropillar arrays [Figs. 1(b) and 1(c)] as a platform for probing thin film evaporation with micro-Raman. Since silicon emits strong Raman scattering, we can directly focus the laser excitation on the top of each silicon micropillar and measure the temperature of the thin film region. By using this technique for various micropillar arrays, we can obtain insights into both the local thin film region and the temperature distribution of the entire structured surface. Additionally, the general thermometry method is not restricted to understand thin film evaporation but can be also widely applied to study many thermofluidic problems at the microscale.
We fabricated test samples with a total microstructured area of 1 cm2 [red dashed box in Fig. 1(b)]. The cylindrical micropillars have diameters d of 20 μm and heights h of 50 μm [see the scanning electron microscopy (SEM) image in Fig. 1(c)]. The pitch l of the micropillar arrays was varied from 30 μm to 100 μm to investigate the geometrical effects on thin film evaporation. On the backside of the sample, we deposited a 150 nm thick platinum layer which supplies uniform heat flux to the structured area (see supplementary material S1 for details). In this work, all structured samples were named as “diameter-pitch.” For example, the micropillar array with a diameter of 20 μm and a pitch of 50 μm shown in Figs. 1(b) and 1(c) was denoted as “20–50.”
To obtain accurate local temperature measurements, we designed and built a custom micro-Raman system. Micro-Raman spectroscopy analyzes the inelastic scattering between an incident photon and an optical phonon in the specimen, which is known as the Raman effect. As the phonon frequency (also known as the Raman shift or Raman peak position) is temperature dependent, micro-Raman spectroscopy can be used to measure the temperature. Figure 2(a) shows the schematic of the optical path (see supplementary material S2 for details on the micro-Raman setup). The Raman scattering was excited by a 633 nm wavelength diode laser. This laser excitation was focused by a 50X microscope objective with a numerical aperture (NA) of 0.55. The corresponding laser spot diameter on the micropillar was ∼1.5 μm. The uncertainty of the measured Raman peak position was controlled within 0.01 cm−1, and the corresponding temperature uncertainty was below 0.5 °C (see supplementary material S2 for the uncertainty analysis). We used a low laser power at 0.5 mW to avoid the laser heating effect (see supplementary material S2 for the laser heating analysis). We showed an example Raman peak of silicon [marked as star in Fig. 2(a)], calibrated by neon emission lines [see the spectrum shown in Fig. 2(a)]. We measured the temperature of the sample by detecting the change of Raman peak position. The change of the Raman peak position and the corresponding temperature rise is given by
where is the change of the Raman peak position, and is the temperature rise with respect to the reference temperature . We used the ambient temperature as the reference temperature in this study. The proportionality constant is known as the linear temperature coefficient of the structured sample, which was −0.0225 ± 0.002 cm−1/°C (see supplementary material S3 for the temperature coefficient measurement). The measured temperature coefficient in this work shows good agreement with the results reported by Beechem et al.21 (−0.022 cm−1/°C).
We integrated the micro-Raman setup with a thin film evaporation test rig. Figure 2(b) shows the schematic of the flow loop. We used a syringe pump to supply de-ionized degassed water and a second syringe pump to control the reservoir water level to avoid flooding on the structured area [see the magnified image of Fig. 2(b) for details of the sample fixture]. The reservoir was maintained at 40 °C (see supplementary material S4 for details of experimental procedures). We characterized heat loss through the system by measuring a non-wetting structured surface (see supplementary material S5 for details of heat loss characterization). In the absence of evaporation, the heating power represents the heat loss Qloss which is proportional to the temperature rise (proportionality is 0.060 W/°C). Therefore, the heat flux q″ through evaporation of water was given by
where Qtot is the total Joule heating power and S is the structured area (1 cm2 in this study).
We experimentally demonstrated the high spatial resolution and temperature sensitivity of micro-Raman thermometry by measuring the temperature distribution on top of a single micropillar. As shown in Fig. 3, we performed temperature mapping (0.5 μm scanning step size) using micro-Raman thermometry on a micropillar located at the center of a 20–60 structured sample with an applied heat flux q″ of 1 W/cm2 and 21 W/cm2. In both cases, as expected, the temperature was very uniform at 46 °C and 100 °C due to the relatively high thermal conductivity of silicon compared to liquid water (see supplementary material S6 for detailed analysis). This uniform temperature distribution was also consistent with our heat transfer simulations (see supplementary material Fig. S7 for simulation results). As the thin film region is isothermal,12 we could obtain the temperature of the liquid thin film near the edge of the micropillar by measuring any arbitrary point on top of the micropillar. In this study, therefore, only the temperature at the center of the micropillar was measured.
To understand the effects of micropillar array geometries on heat transfer performance, we characterized the evaporation process by measuring the local temperature of the liquid thin film at various heat fluxes and on different structured samples. The micro-Raman probe was focused on a micropillar located at the center of the structured area [see the inset of Fig. 4(a)] for all samples. The heat flux q″ was increased gradually from 0 to 30–40 W/cm2 until dryout was observed. We determined the dryout when the liquid at the end of the structured area started receding. The dryout heat fluxes were relatively low because of the long wicking length L (1 cm) chosen in this study, and qdryout scales with 1/L2. Figure 4(a) shows the evaporation curves of different structured samples. There are two distinct regimes based on the slope of the curves. When the temperature of the thin film region was well below the saturation point (100 °C at 1 atm), the evaporation was limited by vapor diffusion in the air and increased slowly with temperature rise. The heat dissipation in this regime significantly relied on the sensible cooling of the liquid as well as the liquid-to-vapor phase change. When the temperature approached the saturation point, evaporation became the dominant heat transfer mechanism and the heat flux increased sharply with a slight temperature rise. On the 20–100 sample, the maximum temperature was found to be 106 °C, indicating that the liquid thin film was in a superheated state during evaporation. This maximum temperature decreased as the pitch decreased. The evaporation curves generally shifted to the left with the decreasing pitch, indicating that higher heat transfer coefficients were achieved on the smaller pitch samples. Figure 4(b) shows the liquid film temperature as a function of the micropillar pitch at various heat flux conditions. The effect of the small pitch in reducing the liquid film temperature was more significant at relatively high heat fluxes (20–30 W/cm2) than at low heat flux conditions (below 15 W/cm2). For example, when the heat flux was 27 W/cm2, the liquid film temperature was reduced by approximately 8 °C when the pitch decreased from 80 μm to 30 μm. This improvement in evaporation heat transfer can be explained by the increase in the thin film area. With the decrease in the pitch, more thin film regions around the micropillar were created per unit area, which offer more effective areas to dissipate the heat load [inset of Fig. 4(a)].
We also studied the effects of micropillar array geometries on the temperature distribution of a structured surface sample. We scanned the temperature distribution from 1 mm to 9 mm along the capillary flow direction on 20–30, 20–80, and 20–100 samples (see the XY coordinate in the inset of Fig. 5). The heat flux q″ was maintained at approximately 22 W/cm2. As this heat flux was significantly smaller than the dryout heat flux for the 20–30 sample (33 W/cm2), the meniscus in each unit cell was almost flat, resulting in similar local heat transfer coefficients (see the schematic of the inset of Fig. 5). Therefore, the temperature distribution on the 20–30 sample was uniform along the flow direction. However, nonuniform temperature distributions were observed on the 20–80 and 20–100 samples (Fig. 5). Generally, along the flow direction, the temperature of the micropillar increased first and then dropped slightly (as small as 1 °C–2 °C). The increase in temperature at the inlet of the structured surface is attributed to the decrease in sensible cooling and heat conduction effects because the water in the reservoir was kept at about 40 °C (see supplementary material S4 for more details). The slight decrease in temperature from the middle part (3 mm for 20–100 and 4 mm for 20–80 samples) to the end (9 mm) is due to the interface shape change. As this heat flux (q″ = 22 W/cm2) was close to the dryout heat fluxes of 20–80 and 20–100 samples (26 W/cm2 and 23 W/cm2, respectively), the apparent contact angle of the liquid in the last unit cell approached the receding contact angle (see the schematic of the inset of Fig. 5). Consequently, the liquid films near the three-phase contact line became thinner along the flow direction. Since the thinner liquid film had smaller thermal resistance, the temperature decreased from the middle of the sample towards the end gradually.
In summary, we showed the use of micro-Raman spectroscopy for the in situ temperature measurement of the liquid thin film in micropillar arrays during evaporation, where a spatial resolution of 1.5 μm and a temperature sensitivity of 0.5 °C were achieved. We precisely measured the liquid film temperature in the superheated state which cannot be measured by the conventional thermometry. We characterized evaporation from micropillar arrays of different geometries, showing that the liquid film temperature reduced as the pitch of the micropillar decreased due to the increase in the thin film area. We also found that the nonuniform temperature distribution occurred along the flow direction when the sample approached dryout. The high resolution local temperature measurement method presented in this work is not only beneficial to the characterization and optimization of thin film evaporation but can also be potentially utilized to reveal physical insights into a variety of other phase change heat transfer processes such as condensation and pool/flow boiling on structured surfaces.
See supplementary material for more information on sample fabrication, experimental setup, sample calibration, additional experimental results, simulation, and data analysis.
The authors would like to acknowledge Arny Leroy and Dion S. Antao for valuable discussions and the MIT Microsystems Technology Lab for fabrication staff support, discussion, and use of equipment. L. Zhang gratefully acknowledges funding support from the MIT/MTL GaN Energy Initiative and the Singapore-MIT Alliance for the Research and Technology (SMART) LEES Program. Y. Zhu acknowledges funding support provided by the Office of Naval Research (ONR) with Dr. Mark Spector as the program manager under Award No. N00014-15-1-2483. Z. Lu acknowledges funding support from the Air Force Office of Scientific Research with Dr. Ali Sayir as the program manager under Award No. FA9550-15-1-0310. The authors also acknowledge the DURIP Program through ONR for the support of the equipment.