Transient microscale flow boiling heat transfer ...

International Journal of Heat and Mass Transfer 90 (2015) 396–405

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt

Transient microscale flow boiling heat transfer characteristics of HFE-7000 Saptarshi Basu b,⇑, Brian Werneke a, Yoav Peles c, Michael K. Jensen a a

Department of Mechanical, Aerospace, and Nuclear Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180, United States Dielectric Systems Module, Applied Materials, Santa Clara, CA 95054, United States c Department of Mechanical and Aerospace Engineering, University of Central Florida, Orlando, FL 32816, United States b

a r t i c l e

i n f o

Article history: Received 30 March 2015 Received in revised form 14 June 2015 Accepted 14 June 2015

Keywords: Microscale Flow boiling Transient HFE-7000

a b s t r a c t A detailed experimental study was conducted to identify the important parametric trends governing the temperature response of a microdevice to transient heat loads for flow boiling of HFE-7000. The microdevice consisted of a microgap etched on a silicon wafer and placed centrally over a thin-film heater deposited on a Pyrex wafer. A step change in heat flux and a rectangular pulse were applied to the heater. The effects of mass flux, heat flux (pulse amplitude), and pulse width on the heater temperature response and boiling dynamics were investigated in detail. Conditions at which onset of boiling occurred were identified and the repeatability of the boiling process was studied. Onset of boiling and the subsequent bubble dynamics was recorded with a high-speed video camera. Boiling initiated at very high wall superheat due to the smoothness of the heater surface and low surface tension of HFE-7000. At high heat fluxes, onset of boiling resulted in the formation of a vapor film on the surface and rapid heater temperature rise was observed. Time taken to initiate boiling decreased rapidly with increasing heat flux and then reached a constant value. The wall superheat at which boiling started increased with increasing heat flux and subsequently reached a constant limit. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Thermal management is a critical issue facing the electronics packaging industry. Miniaturization of electronic devices and systems has resulted in significantly higher transistor packaging densities. This has resulted in markedly increased power generation and heat production both at the chip and device levels as well as in large-scale systems like data centers, super computers, and military ships and aircrafts. Failure to effectively dissipate the heat flux generated in the electronic devices would result in increased device temperatures. The challenge facing thermal engineers is further exacerbated by the transient nature of the power profile frequently encountered in electronic systems. High device temperatures as well as transient temperature cycling are responsible for increased device failure rates, poor performance and low reliability. Effective heat dissipation methods and cooling technologies need to be developed to efficiently cool high power electronics operating at either steady-state or transient conditions. Refrigerant based two-phase (boiling and condensation) cooling

⇑ Corresponding author. E-mail address: [email protected] (S. Basu). http://dx.doi.org/10.1016/j.ijheatmasstransfer.2015.06.038 0017-9310/Ó 2015 Elsevier Ltd. All rights reserved.

systems have showed significant promise in dissipating high heat fluxes while maintaining low surface temperatures. The heat transfer rate is significantly higher in flow boiling compared to single-phase flow due to a combination of the effects of latent heat of vaporization and enhanced mixing of the flow due to the movement of the bubbles [1]. In the literature, microscale flow boiling has been studied in detail under steady-state heating conditions [2,3] but there is lack of experimental studies investigating flow boiling heat transfer at microscale for transient heating conditions. In the majority of the steady-state studies [4–15], the boiling process was studied by increasing the heat flux in infinitesimal amounts and allowing the system to reach steady state before measuring the temperature, pressure, and identifying the corresponding bubble dynamics and boiling regimes. In transient experiments, the heat flux was increased from zero to maximum over a very short time period, and the flow regime made a rapid transition from single-phase flow to different boiling regimes [16]. Transient inputs could be classified into three categories: step response, frequency response, and impulse response [17]. In the present study, the boiling dynamics of HFE-7000 was studied for a step change in heat flux (step response) and pulsed heat inputs (frequency response).

S. Basu et al. / International Journal of Heat and Mass Transfer 90 (2015) 396–405

397

Nomenclature A G R T t q00

area of heater, m2 mass flux, kg=m2 s resistance, X temperature, K time, s heat flux, W=m2

q s

density, kg=m3 pulse width, s

Subscripts heater heater sat saturation

Greek Letters D wall Superheat, K m kinematic viscosity, m2 =s

Several conventional scale transient pool boiling experiments were conducted using thick metallic blocks [16,18–20] or thin wires [16,21–25] as the test section. The heat flux was applied in the form of a step change. The test fluids were water [25], liquid nitrogen [21,23,26], liquid helium [24], refrigerants like FC72 [19] and R113 [20] or organic fluids like pentane [16,18] and ethyl alcohol [27]. The effects of different parameters such as heat flux, preheating, saturation pressures, test section dimensions, and thermal properties of the fluid and the solid on the temperature response of the system were studied in detail. The transient input was in the form of a step change in heat flux. The effects of thermal inertia on the temperature response were studied in detail. Onset of boiling was initiated either by activation of preexisting gas or vapor nuclei trapped in the cavities on the heater surface [16,18] or by explosive vaporization similar to homogeneous nucleation [27]. A detailed review of the superheat limits for different fluids is given in Avedisian [28]. Sakurai et al. [24], on the other hand, observed heterogeneous nucleation on a thin platinum wire immersed in liquid nitrogen and helium under an exponential heat input. Explosive vaporization near the homogeneous nucleation point required a very high heating rate [27]. For steady-state experiments, nucleation was almost always due to the activation of nucleation sites on the heated surface [1]. A premature transition to film boiling at heat fluxes lower than the critical heat flux at steady-state was observed in few studies [20,22,23] while a steady nucleate boiling regime with no or delayed transition to film boiling was observed in other studies [21]. Due to the significant thermal inertia in the experiments, the heat flux boundary condition was not accurately measured in the experiments. The studies at the conventional scale were for pool boiling configurations. The force generated on the fluid by the nucleation, growth, and collapse of the bubbles is used as an actuation mechanism in various applications such as thermal ink jet printers, DNA detection, biosensors, micropumps, fuel injection system [29]. These applications motivated several microscale pool boiling studies [29–38]. Only a few flow boiling studies were conducted at the microscale [39–41]. None of the flow boiling studies [39–41] used low surface tension fluids like refrigerants or organic fluid. The heat flux was applied to a micron sized titanium or platinum heater in the form of a single ls wide pulse. The main objective of the studies was to determine the bubble nucleation mechanisms and growth dynamics for different operating conditions and heater dimensions. The heating rate was very high and generally explosive vaporization at very high superheat was observed. Under these conditions, the boiling process was highly repeatable and afforded a great deal of control on bubble shape and size. Due to the potential application of the bubbles as an actuation mechanism, a high degree of repeatability was desirable. However, explosive vaporization at high wall superheats is not particularly efficient in dissipating high heat fluxes.

Although a significant number of steady-state studies [2–15] have shown that flow boiling is an efficient cooling mechanism, the effectiveness of the boiling process in dissipating high heat fluxes under transient operating conditions has not been studied in detail. In the present work, transient flow boiling heat transfer characteristics of HFE-7000 were experimentally studied in a microchannel heat sink for a step change in heat flux and pulsed heat inputs. The effects of mass flux, heat flux (pulse amplitude), pulse width on bubble dynamics, boiling characteristics, and temperature response of the system were determined. The bubble dynamics and boiling characteristics were correlated to the temperature response of the microdevice in order to identify the best operating conditions to dissipate the heat generated in the device. Onset of boiling was identified using high-speed video imaging. The wall superheat and time taken to initiate boiling were of particular interest. The variability in the conditions at which boiling was initiated and bubble dynamics post onset of boiling was experimentally identified. For reliable and efficient operation, onset of boiling should be controllable and repeatable and result in significant temperature drop. The heating rate was considerably lower and the pulse width was longer than the previous microscale studies [29–41]. HFE-7000 is a dielectric fluid with very low surface tension and is highly wetting in nature. Previous flow boiling studies [39–41] have used water as the test fluid, which is a high surface tension non-wetting fluid. The present set of experiments were carried out on a millimeter sized heater that is in between the heater sizes used in the microscale studies [29–41] and the macroscale experiments [16,18–20].

2. Experimental apparatus and data acqusition Flow boiling experiments with HFE-7000 were conducted in a microdevice that was placed in a closed flow loop (Fig. 1). HFE-7000 was degassed prior to the experiments to remove any non-condensable gases. A deep vacuum was created in the pressure vessel containing the test fluid and the dissolved gases were allowed to escape over a period of two days. The flow loop consisted of a condenser, preheater, and an auxiliary heat exchanger that were used to condition the fluid to the correct temperature at the test section inlet. The flow loop was fitted with an array of thermocouples and pressure transducers to monitor the temperature and pressure at various points in the loop. The microdevice consisted of a silicon wafer bonded to Pyrex wafer. The microdevice was square shaped with a side length of 0.02 m. The thickness of the silicon wafer was 0.45 mm and that of the Pyrex wafer was 1.0 mm. A 10 mm long, 1.2 mm wide, and 0.2 mm high microgap was etched on the silicon wafer. The microgap was placed centrally over a 1 mm 1 mm thin film titanium heater that was deposited on the bottom surface of a Pyrex wafer (Fig. 2). The heater was in direct contact with the

398


Bypass valve

Flow Direcon Filter Preheater

Flowmeter

Tube-in-tube Condenser

Chiller

Pump

Test Secon DAQ Electrical circuit

Pressure Vessel Pressure Gauge Pressure Transducer Thermocouple Fig. 1. Schematic of the closed flow loop for HFE-7000 flow boiling experiments.

Fig. 2. Schematic of the microdevice [42].

test fluid. The heater width was less than the microchannel width to facilitate easy bonding of the Pyrex wafer with the silicon wafer. The titanium heater was only 100 nm thick and photographs of the boiling process were taken through the thin film heater and the Pyrex. The heater itself acted as a temperature sensor. The microdevice was placed in a Delrin package with fluid conduits that connected the microdevice to the flow loop. A series of O-rings were used to ensure a leak free connection. Further details about the microfabrication process and the microdevice are available in Browne [42]. A detailed description of the flow loop and experimental procedures are given in Basu [43]. The local temperature varies over the 1 mm2 heater area and also at the point of contact of the aluminum vias. Because of the overall voltage and current measurement, the heater resistance gave a measure of the heater’s area-averaged temperature. The heater temperature and resistance was correlated using a calibration equation given by Eqn. (1).

TðKÞ ¼ 2:1523 RðXÞ þ 1110:3

ð1Þ

where T is in Kelvin. The calibration curve was obtained by placing the heater in a furnace, whose temperature was varied from approximately 20 C to 100 C in steps of 5 °C. A thermocouple was placed on the heater surface, and the resistance was recorded using a digital multimeter after steady state was achieved. The calibration curve was adjusted for contact and wire resistance when the heater was placed in the test rig at room temperature under vacuum

conditions. Similar issues with contact and wire resistance were discussed in detail in Ching et al. [44]. The calibration curve showed resistance to be decreasing with increasing temperature, which is contrary to the expected trend. However, such a negative trend was also observed in Avedisian et al. [38] and Browne [42]. The heater resistance was calculated from the voltage and current measurements using the Ohm’s law. A NI LabVIEW based data acquisition system (DAQ) was used to record the voltage and current in the test heater as well as the measurements of the different thermocouples and pressure transducers present in the test rig. The DAQ system was capable of making high frequency transient measurements of the voltage and current using two oscilloscopes (Tektronix TDS 2022B and TDS 2012B) that were synchronized with a high-speed camera and a DC power source. This ensured that temperature measurement and flow visualization were synchronized with the pulsed power input. The synchronization was achieved using a power MOSFET in a switching circuit configuration to control the power source and the MOSFET was also used to trigger the oscilloscopes and the high-speed camera. The uncertainties in the derived quantities were calculated from the uncertainty in the measured quantities using the propagation of uncertainty method outlined in Kline and McClintock [45]. The uncertainty in mass flux (kg/m2-s), heater temperature (K), and heat flux (W/m2) is 5:2%; 1:5%, and 1:0%, respectively. Details of the uncertainty analysis are available in Basu [43]. 2.1. Key considerations The passage of electricity through the thin-film heater resulted in a temperature rise due to Joule heating. The heat generated in the thin film heater was dissipated by convection to the fluid (HFE-7000) and conduction to the Pyrex substrate. The heater temperature was determined by the balance between the applied power, convective heat transfer, and conduction heat loss through the Pyrex substrate. Heat loss through the Pyrex substrate could not be measured experimentally. A numerical analysis was done to estimate the heat loss in the Pyrex substrate [43]. The heat loss analysis is not discussed here as the heater temperature is directly measured in the present study. Heat loss estimate is important for calculating convective heat transfer coefficients. The electrical


399

power generated in the heater was calculated from the product of voltage and current in the heater. The heat flux was calculated based on the total electrical power generated in the heater over the total heater surface area. Due to the small size of the thin film heater (volume = 1 1013 m3), the thermal inertia of the heater was small. Therefore, there was a rapid temperature rise at the beginning of the transient before convection effects became significant as the wall-to-fluid temperature difference increased. The sampling rate of the oscilloscopes was not large enough to accurately capture the rapid temperature rise. A voltage spike was also observed at the starting edge of the square pulse when the power supply was switched on. This typically arises during switching of a power system and due to a variety of reasons like inductive pickup or high frequency cycling [38]. This resulted in erroneous measurements during the initial transients. The time scale of this artifact depended on the sampling frequency of the oscilloscope and generally lasted for less than 5% of the total time period [43]. These initial data points were not considered in the plots presented in this paper. The data were filtered using the Butterworth filter of order one to smooth the plot and remove noise. 3. Results and discussion An experimental study was conducted to determine the parametric trends governing flow boiling of HFE-7000 at microscale for transient heat loads applied in the form of a step change in heat flux and a square pulse. The control variables were heat flux or pulse amplitude (0.6–6.75 MW/m2), mass flux (128–3800 kg/m2-s), and pulse width (2 ms – 3 s) while the temperature behavior was the primary response variable. The experiments were conducted at a pressure of 120 kPa (measured at test section exit) and a test section inlet temperature of 23 C. Onset of boiling was identified using high-speed video imaging. Time taken to initiate boiling and wall superheat at that instant were determined for different transient operating conditions. Onset of boiling was followed by a transitional boiling regime that then led to either a stable nucleate boiling regime or a vigorous boiling regime. The effects of different operating parameters on the development of different boiling regimes were determined. The temperature response of the heater was correlated to the bubble dynamics. Repeatability and control of onset of boiling conditions were also investigated.

Fig. 3. Representative plot of heater temperature in response to a step change in heat flux for flow boiling of HFE-7000.

on the boiling regime. The stable boiling regime was preceded by a transitional or developing boiling regime that lasted for a short time. The stable boiling regime was determined by a combination of the transient operating parameters and fluid flow conditions. The factors controlling the temperature dynamics and different boiling regimes are discussed in detail in the subsequent sections. Onset of boiling occurred at high wall superheats under subcooled conditions due to the smoothness of the heater and highly wetting nature of HFE-7000. At standard atmospheric pressure, the saturation temperature of HFE-7000 is 34 C. Boiling started when the heater temperature was approximately 116 C. The heater surface is very smooth and has an average roughness of only 30 nm [42]. HFE-7000 is a low surface tension fluid and produces low contact angles (hc < 5 ) [46].

3.1. Heater temperature response to a step change in heat flux A step change in heat input resulted in the heat flux changing from zero to the maximum value in a short period of time (order of 0.1 s or less), and then remaining constant. For the step change experiments, the temperature response of the heater was studied for a period of three seconds. The parametric trends governing the system temperature response to a step change in heat flux are described below. 3.1.1. Representative temperature response A representative temperature plot (Fig. 3) describes the salient features of the system temperature response to a step change in heat flux for flow boiling of HFE-7000. Before onset of boiling, the flow was in single phase and the heater temperature response was characterized by a rapid conduction temperature rise (could not be captured accurately) followed by a slower convection rise. The exact instant of onset of boiling was identified with a high-speed camera synchronized with the power system and the DAQ. Onset of boiling was followed by either a temperature drop or a temperature rise that depended

Fig. 4. Effects of heat flux on transient heater temperature characteristics for flow boiling of HFE-7000.

400


3.1.2. Effects of heat flux The heater temperature characteristics for increasing heat flux are shown in Fig. 4. At very low heat fluxes (q00 < 0.66 MW/m2), the flow was single phase over the entire time period of interest as the heater did not achieve the threshold wall superheat required to initiate boiling. However, at higher applied heat fluxes, as the heater temperature increased with time, boiling was initiated. Furthermore, with increasing amplitude, boiling was initiated earlier. Prior to onset of boiling, the heater response was characterized by a rapid conduction temperature rise followed by slower convection temperature dynamics. The duration of convection dynamics depended on the time of boiling initiation. At low heat fluxes (q00 0.66–0.78 MW/m2), when boiling initiation was delayed, convection dynamics were prevalent for a longer duration, and the system reached steady state in single phase before onset of boiling. At low heat fluxes and low wall superheat, few nucleation sites were activated and, hence, the flow regime was characterized as the discrete bubble regime (Fig. 5(d) and (e)). Initiation of boiling was followed by a temperature drop at these low heating rates, and eventually the system settled to a lower temperature. The temperature dropped due to the heat transfer enhancement associated with the additional latent heat removal during nucleation and enhanced mixing due to the motion of the vapor bubbles. Further increase in the heat flux resulted in an earlier transition to boiling with a temperature drop similar to the lower heat flux case. However, on increasing the heat flux further, boiling was initiated within a very short time after switching on the power supply, and before the system could reach steady state. Due to the higher heating rate at these heat fluxes (q00 0.78 MW/m2), the surface temperature increased even after the onset of boiling although at a slower rate. The increase in surface temperature with time resulted in an increase in the number of active nucleation sites [1]. Numerous bubbles nucleated on the heater and then coalesced to form a vapor layer (Fig. 6(d) and (e)). Boiling was vigorous and the growing vapor layer resulted in further temperature rise (Fig. 4 ). Heat transfer was diminished due to the lower thermal conductivity of the vapor film and the cold fluid was prevented

Fabricaon defects

Direcon of flow

from coming close to the heater surface. This could lead to local dryout of the heater surface. As seen in Fig. 5(b) for low heat fluxes (q00 0.66 MW/m2), boiling was initiated in the form of a large vapor cloud after about 1.40 s from the time power was switched on. This was followed by the transition boiling regime (Fig. 5(c)) for about 0.15 s. Thereafter, the system settled into the discrete bubble regime (Fig. 5(d) and (e)). The heater temperature decreased after boiling initiation, and, therefore, the vapor cloud broke into discrete bubbles. The heater temperature dropped after power was switched off, and the number of nucleating bubbles decreased until the flow was single phase again (Fig. 5(f), (g), and (h)). The fluid flow is from left to right. For higher heat fluxes (q00 > 0.78 MW/m2), boiling started after 0.116 s into the pulse in an explosive manner (Fig. 6(b)) and gave rise to a vapor layer changing shape over the surface (Fig. 6(c)). Numerous bubbles nucleated on the heater and coalesced to form the vapor layer. With increasing temperature, the vapor layer increased in size to cover most of the heater area. Fig. 6(d) and (e) shows the vigorous boiling regime that remained unchanged until the power was switched off. Thereafter, the vapor layer decreased in size to form a few discrete bubbles until nucleation stopped and the flow was single phase again (Fig. 6(f), (g), and (h)). The darker patches are the vapor bubbles while the lighter patches denote liquid HFE-7000. 3.1.3. Effects of mass flux Higher heat flux was required to reach the required wall superheat to initiate boiling with an increase in mass flux. Increasing mass flux resulted in enhanced convective cooling of the heater surface. For a constant heat flux, as mass flux was decreased, the rate of increase of the heater temperature was higher. As shown in Fig. 7, at low mass fluxes (G 300 kg/m2-s), the heater reached the threshold wall superheat quickly before the system reached steady state and boiling was initiated earlier than at higher mass fluxes. Onset of boiling was not sufficient to reduce the surface temperature at these high heating rates. Therefore, as the heater temperature increased, the number of active nucleation sites increased [1]. This is seen in Fig. 6(c), (d), and (e). A vigorous

Temperature decrease

Aluminum contact pads (white region) Fig. 5. Discrete bubble regime and boiling dynamics at low heat fluxes (q00 0.66 MW/m2).


Fabricaon defects

Direcon of flow

401

Large number of nucleaon sites; Temperature rise

Fig. 6. Vigorous boiling regime and bubble dynamics at high heat fluxes (q00 0.78 MW/m2).

Fig. 7. Effects of mass flux on transient heater temperature characteristics for flow boiling of HFE-7000.

boiling regime was observed with numerous bubbles nucleating on the titanium heater and coalescing to form a vapor layer. However, at higher mass fluxes, the system reached steady state in single-phase before onset of boiling, and the heating rate was lower. Onset of boiling was driven by the activation of the nucleation sites on the heater [1]. The flow regime was characterized by discrete bubbles and the heater temperature dropped after boiling initiation as the heating rate was lower. Convection generally suppressed nucleation and this could have also resulted in less nucleation at high mass fluxes.

3.2. Heater temperature response to a pulsed heat input A rectangular pulse of width 2 ms was applied to the heater and the temperature response was studied for flow boiling of HFE-7000. Compared to a step change in heat input, the

Fig. 8. Effects of pulse amplitude on boiling heat transfer characteristics for a 2-ms pulse.

rectangular pulse was switched off before the system reached steady state. Since the heater temperature increased with increasing pulse width, a shorter heating time would require higher pulse amplitude to produce the same level of heating. Therefore, very high heat fluxes (q00 > 5.0 MW/m2) were required to produce boiling within 2 ms. Heating rate was consequently very high. The initial transients at the beginning of the pulse could not be measured due to measurement limitations and a spike in the input voltage signal. 3.2.1. Effects of pulse amplitude The effects of pulse amplitude on the heater temperature dynamics are shown in Fig. 8. Due to high heating rates, boiling was initiated in an explosive fashion due to superheating of the liquid layer near the heater

402


Fabricaon defects

Direcon of flow

Explosive expansion of vapor layer

Uniform vapor film; Temperature rise

Fig. 9. Film boiling regime and bubble dynamics at ultra-high heat fluxes (q00 > 5.0 MW/m2).

started shrinking in size until the flow returned to single phase (Fig. 9(f), (g), and (h)). The vapor film covered the entire heater surface and the wall superheat was high enough to ensure that the film did not shrink or break until power was switched off. The wavy structures were probably caused due to interactions between the main liquid flow, the vapor film, and the pressure waves created during the explosive nucleation process. 3.2.2. Effects of mass flux Mass flux did not have a significant effect on heater temperature dynamics or on onset of boiling conditions (Fig. 10). At small time scales mass flux effects were minimal. Therefore, before onset of boiling, temperature dynamics were uninfluenced by mass flux. Boiling was initiated in an explosive manner and the heater surface was covered with a vapor film. The wall superheat and the time to boiling initiation were uninfluenced by mass flux. Mass flux also did not significantly affect the boiling regime after boiling onset. Therefore, heater temperature post boiling was unaffected by variations in mass flux.

Fig. 10. Effects of mass flux on boiling heat transfer characteristics for a 2-ms pulse.

3.3. Onset of boiling conditions

surface, and a vapor film developed over the heater (Fig. 9(d) and (e)). The heater temperature increased after onset of boiling. Onset of boiling was identified with the help of a high-speed camera. Higher heat flux resulted in a higher surface temperature and marginally earlier onset of boiling. The wall superheat at onset of boiling did not significantly vary with increasing heat flux. Boiling was initiated explosively due to superheating of the liquid layer near the heater surface around 0.80 ms into the pulse (Fig. 9(b)). It was initiated near the heater edges. Explosive nucleation resulted in the vapor layer protruding outside the heater area (Fig. 9(c)), and it quickly shrunk in size to cover the heater completely. The film was stable over the heater with some wavy structures (Fig. 9(d) and (e)). After power was switched off, the film

The conditions during onset of boiling have a significant effect on the temperature dynamics of the heater. At low heat flux, boiling was mostly characterized by bubble nucleation (discrete bubble regime and vigorous boiling regime) while at higher heat fluxes boiling was explosive in nature and a vapor film was formed over the heater surface. Boiling dynamics were primarily governed by the heating rate. The heating rate was controlled by the heat flux, mass flux, and the pulse width. There was a lower and upper bound on the applied heat flux because of measurement accuracy at low power and current limit at high power. The pulse width was also limited to 3 s. All three boiling regimes were not observed for all mass fluxes due to heat flux limitations and the finite duration of the pulse. At high heat fluxes (q00 5.0 MW/m2), the mass flux effect was minimal.


From a transient perspective, it is important to determine the conditions (wall superheat and time to nucleation) during boiling initiation. The temperature dynamics of the heater and the conditions during onset of boiling were determined by the boiling regime. The observed boiling regimes were not unique for a particular mass flux or heat flux but occurred over a range of operating conditions. Representative mass flux and heat flux conditions were chosen to illustrate the relationship between boiling dynamics and the temperature response of the heater in Figs. 11–13. HFE-7000 is a highly wetting fluid with a contact angle near zero degrees [46] and the titanium heater surface is very smooth [42]. Therefore, the wall superheat during onset of boiling was very high. The variations in onset of boiling conditions for a step change in heat flux and at low heat fluxes (q00 0.66 MW/m2) are shown in Fig. 11. This corresponds to the discrete bubble regime. A total of sixteen runs were made at the same mass flux and heat flux. At low heat fluxes, heating rate was low and boiling was slow to start. Nucleation was governed by activation of nucleation sites on the heater surface. Discrete bubbles nucleated on the heater surface. Generally, at these heat fluxes, the system reached steady state in single phase before boiling started. The heater temperature before and after onset of boiling collapsed onto the same curve. The temperature at which boiling was initiated did not vary significantly, but the time of nucleation varied for different runs. There was a minimum heat flux below which nucleation was not observed as the heater surface did not reach the required superheat. For low heat fluxes (q00 0.66 MW/m2) at which boiling was observed, it is probable that the surface superheat was close to the threshold value and boiling was initiated due to local variations in the heater temperature. Activation of nucleation sites is a random process that depended on several factors such as impurities, dissolved gases etc. [1]. The heater temperature measured was area averaged over the heater surface, and, hence, spatial variations in the temperature field were not captured. Local temperature fluctuations could result in localized nucleation. Under such conditions discrete bubble regime was observed and onset of boiling was associated with temperature drop. At a slightly higher heat flux (q00 0.75 MW/m2) and step change in heat flux, boiling was initiated very quickly as the heating rate was higher. This corresponds to the vigorous boiling

403

Fig. 12. Variations in onset of boiling conditions in the vigorous boiling regime (Step change in heat flux and at high heat fluxes (q00 0.75 MW/m2)).

Fig. 13. Variations in onset of boiling conditions in the film boiling regime (Pulse input and at ultra-high heat fluxes (q00 > 5.0 MW/m2)).

Fig. 11. Variations in onset of boiling conditions in the discrete bubble regime (Step change in heat flux and at low heat fluxes (q00 0.66 MW/m2)).

regime. Under these conditions, nucleation rate was increased and a large number of bubbles nucleated on the heater surface. The bubbles coalesced to form a vapor layer. Boiling was vigorous and the film was repeatedly agitated by the formation of new bubbles. The heater temperature increased after onset of boiling. Onset of boiling was repeatable in terms of both time and surface temperature (Fig. 12). The temperature plots for different runs collapsed into a single curve. The wall superheat at nucleation was very high. As a small change in heat flux and wall superheat resulted in a huge increase in the number of nucleating bubbles, it is probable that there is a threshold wall superheat value above which majority of the nucleation sites on the heater surface were activated. For smooth surfaces, the size distribution of the nucleation sites are typically small [47]. At higher heat fluxes and heating rates, the

404


surface temperature may rise before the nucleation sites could activate at a lower temperature. At this higher temperature, nucleation occurred without significant variations in time and wall superheat as the temperature was above the threshold value. At very high heat fluxes (q00 5.0 MW/m2) and pulse input, boiling was initiated by explosive vaporization due to superheating of the liquid layer near the heater surface and a vapor film was formed over the heater surface. This corresponds to the film boiling regime. Boiling was initiated near the heater edges. Due to the extremely high heating rates and liquid superheat, vaporization was highly repeatable in terms of wall superheat and time to nucleation (Fig. 13). The heater temperatures collapsed onto a single curve both before and after onset of boiling. Heat flux had a significant effect on the wall superheat at nucleation and the time taken to start boiling. Wall superheat increased

with increasing heat flux until it reached a constant value (Fig. 14). With increasing heat flux, heating rate increased and boiling was started earlier (Fig. 15). No data were obtained at intermediate heat fluxes. Boiling occurred at different times from the beginning of the pulse for different heat fluxes. At low heat fluxes (q00 0.66–0.78 MW/m2), as boiling was initiated slowly and wall superheat was lower, nucleation seemed to be governed by the activation of nucleation sites on the heater surface [1]. It gave rise to either a discrete boiling regime or a vigorous boiling regime, where the nucleating bubbles coalesced to form a vapor layer. However, with increasing heat flux, nucleation was explosive in nature and superheating of the liquid layer near the surface was the most likely mechanism. At very high heat fluxes (q00 > 5.0 MW/m2), the heating rate was high, and the nucleation wall superheat did not change with increasing heat flux.

4. Summary An extensive experimental study was conducted to investigate heat transfer characteristics of a microgap heat sink under transient heat loads for flow boiling of HFE-7000. The transient heat load was applied in the form of a step change in heat flux and a rectangular pulse. The power supplied to the heater was dissipated by fluid convection and by conduction in the Pyrex substrate. The heater temperature was directly measured in the present study. For boiling at such small time periods, very high heat fluxes were applied and the heating rate was high. Fluids like HFE-7000 have low surface tension and are highly wetting in nature with very low contact angles. Therefore, the boiling characteristics are different compared to a high surface tension fluid like water [39–41]. The main findings of the work are summarized as follows:

Fig. 14. Effects of pulse amplitude on wall superheat at onset of boiling.

Fig. 15. Effects of pulse amplitude on time of nucleation.

Due to the smoothness of the heater surface, low contact angles of HFE-7000, and high heating rates, boiling was initiated at very high wall superheats. Heat flux determined the boiling regime which affected the onset of boiling, bubble dynamics, and temperature response of the heater. At low heat fluxes (q00 0.66–0.78 MW/m2), boiling was initiated by activation of nucleation sites on the surface that was then covered with discrete bubbles. This corresponds to the discrete bubble regime Temperature dropped after boiling initiation. At slightly higher heat fluxes, heating rate was higher, and the nucleating bubbles coalesced to form a vapor layer that covered a significant portion of the heater. Temperature increased after onset of boiling. This corresponds to the vigorous boiling regime corresponding to large number of nucleating bubbles. At very high heat fluxes (q00 5.0 MW/m2), the heating rate was extremely high, and nucleation occurred very quickly at the heater edges and was followed by the superheated liquid exploding into a vapor film over the heater. This is the film boiling regime. Onset of boiling was followed by a rise in heater temperature. In the step change study, mass flux had a significant effect on the heater temperature response. For the same heat flux, higher mass flux resulted in a discrete bubble regime and temperature drop after onset of boiling. Lower mass flux resulted in a vigorous boiling regime and temperature rise after boiling initiation. For shorter pulses, mass flux did not have a significant effect on heater temperature response. Higher heat flux resulted in higher surface temperature both before and after onset of boiling. The repeatability of the onset of boiling conditions was tested by conducting several experiments at same conditions. At low heat fluxes (q00 0.66–0.78 MW/m2), significant variations in


time taken to initiate boiling were observed. At higher heat fluxes, variations were significantly less. The wall superheat did not vary significantly in either case. Time taken to initiate boiling decreased rapidly initially with increasing heat flux but quickly reached a constant value at higher heat fluxes. The wall superheat at the onset conditions increased with increasing heat flux. A superheat limit was observed at high heat fluxes (q00 > 4.0 MW/m2). Conflict of interest None declared. Acknowledgement This work was supported by the Office of Naval Research (ONR) under the Multidisciplinary University Research Initiative (MURI) Award GG10919 entitled ‘‘System-Level Approach for Multi-Phase, Nanotechnology-Enhanced Cooling of High-Power Microelectronic Systems.’’ References [1] V.P. Carey, Liquid-Vapor Phase-Change Phenomena, second ed., Taylor and Francis Group, New York, NY, 2008. [2] S. Basu, S. Ndao, G.J. Michna, Y. Peles, M.K. Jensen, Flow boiling of R134a in circular microtubes – Part I: study of heat transfer characteristics, J. Heat Transfer 133 (2011). 051502-051501. [3] S. Basu, S. Ndao, G.J. Michna, Y. Peles, M.K. Jensen, Flow boiling of R134a in circular microtubes – Part II: study of critical heat flux condition, J. Heat Transfer 133 (2011). 051503–051501. [4] D.D. Hall, I. Mudawar, Critical heat flux (CHF) for water flow in tubes – I. Compilation and assessment of world CHF data, Int. J. Heat Mass Transfer 43 (2000) 2573–2604. [5] G.M. Lazarek, S.H. Black, Evaporative heat transfer, pressure drop and critical heat flux in a small vertical tube with R-113, Int. J. Heat Mass Transfer 25 (1982) 945–960. [6] L. Wojtan, R. Revellin, J.R. Thome, Investigation of saturated critical heat flux in a single uniformly heated microchannel, Exp. Therm. Fluid Sci. 30 (2006) 765– 774. [7] Y. Katto, H. Ohno, An improved version of the generalized correlation of critical heat flux for the forced convective boiling in uniformly heated vertical tubes, Int. J. Heat Mass Transfer 27 (1984) 1641–1648. [8] A.E. Bergles, S.G. Kandlikar, On the nature of critical heat flux in microchannels, J. Heat Transfer 127 (2005) 101–107. [9] W. Qu, I. Mudawar, Measurement and correlation of critical heat flux in twophase microchannel heat sinks, Int. J. Heat Mass Transfer 47 (2004) 2045– 2059. [10] J.R. Thome, Boiling in microchannels: a review of experiment and theory, Int. J. Heat Fluid Flow 25 (2004) 128–139. [11] J.R. Thome, State-of-the-art overview of boiling and two-phase flows in microchannels, Heat Transfer Eng. 27 (2006) 4–19. [12] S.G. Kandlikar, Fundamental issues related to flow boiling in minichannels and microchannels, Exp. Therm. Fluid Sci. 26 (2002) 389–407. [13] S.G. Kandlikar, Two-phase flow patterns, pressure drop, and heat transfer during boiling in minichannel flow passages of compact evaporators, Heat Transfer Eng. 23 (2002) 5–23. [14] W. Qu, I. Mudawar, Flow boiling heat transfer in two-phase micro-channel heat sinks—I. Experimental investigation and assessment of correlation methods, Int. J. Heat Mass Transfer 46 (2003) 2755–2771. [15] S.S. Bertsch, E.A. Groll, S.V. Garimella, Effects of heat flux, mass flux, vapor quality, and saturation temperature on flow boiling heat transfer in microchannels, Int. J. Multiphase Flow 35 (2009) 142–154. [16] M.-C. Duluc, B. Stutz, M. Lallemand, Transient nucleate boiling under stepwise heat generation for highly wetting fluids, Int. J. Heat Mass Transfer 47 (2004) 5541–5553. [17] D. Bunce, S. Kandlikar, Transient response of heat exchangers, in: Proceedings of the second ISHMT–ASME Heat and Mass Transfer Conference, Surathkal, India, 1995. [18] S. Héas, H. Robidou, M. Raynaud, M. Lallemand, Onset of transient nucleate boiling from a thick flat sample, Int. J. Heat Mass Transfer 46 (2003) 355–365. [19] H. Auracher, W. Marquardt, Experimental studies of boiling mechanisms in all boiling regimes under steady state and transient conditions, Int. J. Therm. Sci. 41 (2002) 586–598.

405

[20] K. Okuyama, Y. Iida, T. Kato, Premature transition to film boiling with stepwise heat generation 2nd report: effect of wall material and surface conditions, Heat Transfer-Jpn. Res. 25 (1996) 51–63. [21] P.J. Giarratano, Transient boiling heat transfer from two different heat sources: small diameter wire and thin film flat surface on a quartz substrate, Int. J. Heat Mass Transfer 27 (1984) 1311–1318. [22] K. Okuyama, Y. Iida, Transient boiling heat transfer characteristics of nitrogen (bubble behavior and heat transfer rate at stepwise heat generation), Int. J. Heat Mass Transfer 33 (1990) 2065–2071. [23] D.N. Sinha, L.C. Brodie, J.S. Semura, Liquid to vapour homogeneous nucleation in liquid nitrogen, Phys. Rev. B 36 (1987) 4082–4085. [24] A. Sakurai, M. Shiotsu, K. Hata, Boiling phenomenon due to quasi-steadily and rapidly increasing heat inputs in LN2 and LHe I, Cryogenics 36 (1996) 189–196. [25] A. Sakurai, M. Shiotsu, K. Hata, K. Fukuda, Photographic study on transitions from non-boiling and nucleate boiling regime to film boiling due to increasing heat inputs in liquid nitrogen and water, Nucl. Eng. Des. 200 (2000) 39–54. [26] V. Drach, J. Fricke, Transient heat transfer from smooth surfaces into liquid nitrogen, Cryogenics 36 (1996) 263–269. [27] Y. Iida, K. Okuyama, K. Sakurai, Boiling nucleation on a very small film heater subjected to extremely rapid heating, Int. J, Heat Mass Transfer 37 (1994) 2771–2780. [28] C.T. Avedisian, The homogeneous nucleation limits of liquids, J. Phys. Chem. Ref. Data 14 (1985) 695–720. [29] J. Li, G. Peterson, P. Cheng, Dynamic characteristics of transient boiling on a square platinum microheater under millisecond pulsed heating, Int. J. Heat Mass Transfer 51 (2008) 273–282. [30] K. Okuyama, S. Tsukahara, N. Morita, Y. Iida, Transient behavior of boiling bubbles generated on the small heater of a thermal ink jet printhead, Exp. Therm. Fluid Sci. 28 (2004) 825–834. [31] G. Romera-Guereca, T. Choi, D. Poulikakos, Explosive vaporization and microbubble oscillations on submicron width thin film strip heaters, Int. J. Heat Mass Transfer 51 (2008) 4427–4438. [32] Z. Zhao, S. Glod, D. Poulikakos, Pressure and power generation during explosive vaporization on a thin-film microheater, Int. J. Heat Mass Transfer 43 (2000) 281–296. [33] P. Deng, Y.-K. Lee, P. Cheng, An experimental study of heater size effect on micro bubble generation, Int. J. Heat Mass Transfer 49 (2006) 2535–2544. [34] P. Deng, Y.-K. Lee, P. Cheng, The growth and collapse of a microbubble under pulse heating, Int. J. Heat Mass Transfer 46 (2003) 4041–4050. [35] J. Xu, W. Zhang, Effect of pulse heating parameters on the microscale bubble dynamics at a microheater surface, Int. J. Heat Mass Transfer 51 (2008) 389– 396. [36] Z. Yin, A. Prosperetti, J. Kim, Bubble growth on an impulsively powered microheater, Int. J. Heat Mass Transfer 47 (2004) 1053–1067. [37] R.E. Cavicchi, C.T. Avedisian, Bubble nucleation and growth anomaly for a hydrophilic microheater attributed to metastable nanobubbles, Phys. Rev. Lett. 98 (2007) 124501. [38] C.T. Avedisian, W.S. Osborne, F.D. McLeod, C.M. Curley, Measuring bubble nucleation temperature on the surface of a rapidly heated thermal ink-jet heater immersed in a pool of water, Proc. R. Soc. London Ser. A: Math. Phys. Eng. Sci. 455 (1999) 3875–3899. [39] G. Chen, P. Cheng, Nucleate and film boiling on a microheater under pulse heating in a microchannel, Int. Commun. Heat Mass Transfer 36 (2009) 391– 396. [40] G. Chen, X. Quan, P. Cheng, Effects of surfactant additive on flow boiling over a microheater under pulse heating, Int. J. Heat Mass Transfer 53 (2010) 1586– 1590. [41] G. Chen, X. Quan, P. Cheng, Effects of pulse width and mass flux on microscale flow boiling under pulse heating, Int. Commun. Heat Mass Transfer 37 (2010) 792–795. [42] E. Browne, Experimental investigation of single-phase and flow boiling, submerged, and confined microjet array heat transfer (Ph.D. thesis), Department of Mechanical, Aerospace, and Nuclear Engineering, Rensselaer Polytechnic Institute, 2010. [43] S. Basu, Transient heat transfer characteristics of a thermal microdevice (Ph.D. thesis), Department of Mechanical, Aerospace, and Nuclear Engineering, Rensselaer Polytechnic Institute, 2013. [44] E.J. Ching, C.T. Avedisian, M.J. Carrier, R.C. Cavicchi, J.R. Young, Measurement of the bubble nucleation temperature of water on a pulse-heated thin platinum film supported by a membrane using a low-noise bridge circuit, Int. J. Heat Mass Transfer 79 (2014) 82–93. [45] S.J. Kline, F.A. McClintock, Describing uncertainties in single-sample experiments, Mech. Eng. 75 (1953) 3–8. [46] D.G. Gray, J.M. Kuhlman, S.L. Glaspell, C.A. Hunnell, P.J. Kreitzer, D. Mehra, R.M. Youssef, Electromagnetic Control of High Heat Flux Spray Impingement Boiling under Microgravity Conditions, Technical Report, West Virginia University, Morgantown, 2007. [47] C. Corty, A.S. Foust, Surface variables in nucleate boiling, in: Chemical Engineering Progress Symposium Series, vol. 51, 1955, pp. 1–12.