Professional Documents
Culture Documents
Ruiz 2016
Ruiz 2016
Boiling heat transfer performance in a spiraling radial inflow microchannel cold plate
Maritza Ruiz received her Ph.D. from UC Berkeley in 2015 where she studied
two-phase high heat flux cooling systems, and was the recipient of a National
design engineer at SunPower Corporation where she develops and studies tools
for producing the next generation of solar photovoltaic modules. She is also a part time adjunct
Santa Clara University in Santa Clara, California, USA. She received her M.S.
Graduate Research Fellow and her research interests include phase change for electronics
Jorge Padilla received the B.S. degree in mechanical engineering from the
1
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Berkeley in 2007 and 2014, respectively. His research interests include liquid-vapor phase
exchange, nucleate boiling processes and renewable energy applications. He previously worked
at the solar thermal company, eSolar. He is currently working for Google on thermal
research has spanned a variety of applications areas, including fuel cells, solar power systems,
building and vehicle air conditioning, forging and casting of aluminum, phase change thermal
energy storage, Rankine cycle power for manned space missions, heat pipes for aerospace
applications, high heat flux cooling of electronics, heat transfer in porous burners, data center
energy efficiency, energy sustainability of information processing, and advanced solar absorber
ABSTRACT
This study presents an experimental exploration of flow boiling heat transfer in a spiraling radial
inflow microchannel heat sink. The effect of surface wettability, fluid subcooling, and mass fluxes
are considered. The design of the heat sink provides an inward radial swirl flow between parallel,
coaxial disks that form a microchannel of 300 microns. The channel is heated on one side, while the
opposite side is essentially adiabatic to simulate a heat sink scenario for electronics cooling. To
explore the effects of varying surface wetting, experiments were conducted with two different
2
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
heated surfaces. One was a clean, machined copper surface and the other was a surface coated with
zinc oxide nanostructures that are superhydrophilic. During boiling, increased wettability resulted
in quicker rewetting and smaller bubble departure diameter, as indicated by reduced temperature
oscillations during boiling, and achieving higher maximum heat flux without dryout. The highest
heat transfer coefficients were seen in fully developed boiling with low subcooling levels as a result
of heat transfer being dominated by nucleate boiling. The highest heat fluxes achieved were during
partial subcooled flow boiling at 300 W/cm2 with an average surface temperature of 134 degrees
3
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Exploring innovative technologies for improved cooling capability for the thermal management
of high powered electronic chips or processors, as well as other high heat flux applications, is
crucial for the safety, performance and lifetime of devices. These technologies have the
opportunity to increase performance of devices with heat fluxes on the order of 100 to 1000 of
W/cm2. High heat flux cooling systems must be able to provide high heat dissipation rates at
high conductance levels while maintaining low and uniform wall temperatures. For active
systems, minimal pumping power, as well as simple, reliable and robust designs are desired. For
electronics cooling, Agostini et al. [1] listed a cooling benchmark for modern microelectronics
and power electronics of heat fluxes beyond 300 W/cm2 at chip temperatures of 85°C. The
potential for integrating cooling systems into a 3D chip architecture also provides an attractive
solution to meet processor cooling needs and is an active area of research [2, 3].
Two-phase microchannels are researched for their ability to utilize the large latent
heat of a working fluid during boiling to achieve high heat dissipation rates. These systems have
the ability to achieve desired cooling rates at lower pumping costs and more uniform surface
temperatures when compared to single phase microchannels. However, heat dissipation through
two-phase microchannel heat sinks are limited by the critical heat flux which occurs when the
channel wall dries out due to vapor crowding. Compressibility, instabilities and oscillations in
the flow may lead to early onset of critical heat flux (CHF) and is a major drawback as discussed
by Kim and Mudawar [4]. Additionally, despite the enhanced heat transfer performance, surface
temperatures are dependent on the saturation temperature of the working fluid. Innovative
4
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
still necessary to realize the full benefits of microchannel boiling. Kandlikar et al. [2] suggest
future work should include enhancing fundamentals, providing accurate mapping of flow
regimes, using nanoscale surface features for heat transfer enhancement, and developing robust,
inexpensive suppression techniques for mitigating flow boiling instabilities. Ebadian and Lin
also stress the need for increasing reliability and reducing cost by using innovative designs and
Two design aspects that have potential for further research in microchannels, is the
use of enhancements in flow boiling with streamwise curvature and the effects of surface
wettability on two-phase flow in microchannels. Studies with flow boiling in curved channels
subject to concave heating have shown enhanced critical heat flux when compared to straight
channels [6-9]. This enhancement is due to increased buoyant forces on vapor due to centripetal
acceleration of the flow as well as increased subcooling at the heated surface due to radial
pressure gradients. This effect was most significant for low subcooling levels where vapor
proportions were higher, with enhancements up to 50% in flow using FC-72 and subcooling of
5°C in Leland and Chow [6]. Increased surface wettability is also expected to have significant
effects on boiling heat transfer mechanisms and prevent dryout of the channel wall. Liu et al.
[10] found a superhydrophilic Si-nanowire coated surface could reduce temperature oscillations
of the channel wall, likely as a result of smaller bubbles. However, they also found it was more
difficult for nucleation to occur at low superheat and high mass flux rates. This follows trends
predicted by Qu and Mudawar [11], which suggest bubble departure diameter decreases with
equilibrium contact angles. There are many additional studies on further advancements to
5
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
max heat flux over a plain microchannel [12], small-scale (piranha) pin-fins,
These most recent studies fall under a similar surface enhancement model of
desirable characteristics in heat transfer applications and the choice of substrate material, nano-
particle deposition technique, and nano-material is diverse. Liu et al. [10] used silicon nano-
wires on an aluminum substrate while Kim and Mudawar [4] deposited SiO2 nanoparticles
(600nm thick) on a silicon substrate. The hydrothermal method of nanoparticle growth, first
demonstrated by Yang et al. [13], showed how nano-particles could be grown in a liquid growth
solution. This technique of hydrothermal synthesis on copper was subsequently used by Padilla
and Carey [14]. They demonstrated that ZnO nano-particles on a copper substrate could increase
the Leidenfrost point. Increased vaporization heat transfer coefficients were also demonstrated,
which is beneficial for flow boiling in microchannels. A copper substrate was used due to its
Motivated by the need to solve high flux cooling issues, a novel, spiraling radial
inflow microchannel heat sink was developed. This system was designed to study the
enhancements possible in both single phase flow and two-phase flow boiling. Single phase flow
enhancements in this type of heat sink due to unsteady and secondary motions, resulting in high
forced convective heat transfer coefficients, have been experimentally demonstrated in Ruiz and
Carey [15]. For both single phase and two phase flow, the microchannel design was
hypothesized to aid in providing uniform surface temperatures while achieving a high heat
dissipation rate and preventing the maldistribution of flow which can be an issue in rectangular
6
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
microchannel arrays. For two-phase flow, enhancements to boiling similar to those seen in
curved channels with concave heating were expected. A radial inflow heat sink for two-phase
flow can exploit the buoyant force on vapor bubbles due to centripetal acceleration, helping to
draw vapor off the surface towards the outlet and preventing dryout. This type of microchannel
heat sink, first described in Ruiz and Carey [15], consists of a spiraling inflow of fluid sweeping
the surface of two coaxial disks with a constant gap height and exiting through a port in the
center. A diagram of the device is shown in Fig. 1. A single entrance port on the edge of the
device leads into a larger ``feeder" channel region to allow for the development of a more
axisymmetric flow distribution in the microchannel region. The tangential flow helps to provide
a uniform flow distribution over the entire heated surface which aids in temperature uniformity
and prevents the development of hot spots. The simple design with a single inlet and outlet
allows for ease in manufacturing since it does not involve microstructured surfaces or complex
manifolding. The use of microchannels on the order of 100μm allows for high heat transfer
coefficients and a compact footprint. Increased surface wettability is expected to prevent dryout
and increase the performance of boiling heat transfer. This study entails a description of this
type of heat sink and experimental characterization of two-phase boiling flow in a fabricated
device with a constant gap height of 300μm subject to high heat dissipation rates up to 300
METHODS
The fabrication of the device for this study was done using standard machining operations. A
copper alloy 145 heat sink was paired with an insulating cover made from Ultem TM
thermoplastic. The top view of the device is shown in Fig. 2 where the thermoplastic is placed on
7
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
top of the heat sink surface. The body of the heat sink has bored holes on the bottom for three
Watlow cartridge heaters which output 400W at 120 VAC. These were inserted into the bottom
of the heat sink to provide the desired heat levels. Located along the edge of the device are
twelve holes for thermocouple probes, which measure temperature within the copper during
testing.
The design parameters for this copper device are shown in Table 1. A channel height, b,
of 300 µm and a radius of 1 cm was chosen based on typical size of systems desired for
electronics cooling applications. The inlet feeder region is located in the plastic cover and
consists of a uniform channel of cross section, Ac, 2.5mm deep x 1.25 mm wide, which was
Two different heat transfer surfaces were used for this study; one was a clean, machined,
copper surface, while the other was coated with ZnO nanostructures. The growth of
nanostructures on the surface of this coated test device was done using a technique outlined in
Padilla and Carey [14]. The technique, known as hydrothermal synthesis, involves deposition of
a nanoparticle solution onto a polished copper surface. The surface is then submerged in a
growth solution and baked for ten hours. The resulting nanocoated surface has extremely high
wetting characteristics and contact angles of less than 10 degrees. The coated and uncoated test
pieces were used separately to run identical experiments, allowing for direct comparison of the
coated and non-coated surface in this high heat flux removal study.
Figure 3 shows experimental setup of the test system. Flow was maintained throughout
the system at a constant pressure and flow rate using a gear pump. Flow rate was measured by a
rotameter upstream of the device. Two preheaters were set up downstream of the rotameter to
8
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
provide elevated inlet temperatures. The inlet fluid temperature was measured just upstream of
heat sink. After the fluid passed over the heated surface of the device, the water was condensed
and cooled in a heat exchanger submerged in a cold water or ice water bath. The water then
flowed into a reservoir for pumping back into the system. To prevent losses, insulation was
wrapped around the bottom of the heat sink and around the top exposed copper regions. The
pressure drop in the system was measured through ports in the plastic cover (Fig. 2) using two
Cole-Parmer pressure transducers. The inlet and outlet fluid temperatures as well as the
temperature within the copper device was measured using T-type thermocouple probes from
Omega. The mass flow rate was measured using a 65mm variable area rotameter from Omega
Engineering.
The heat sink was tested over a range of flow rates and heat inputs as shown in Table 2,
using distilled water as a working fluid. For each flow rate, the input power was increased over a
range of heat rates and tested over this range until either (a) the maximum heat rate, based on
experimental design limits, was reached or (b) the temperature of the system began to unstably
increase, indicating dryout conditions on the surface. Once steady state was reached for each
heat rate, temperature and pressure data was recorded for approximately 1.5 minutes and
For each flow rate and heat rate pair tested, heat transfer characteristics were found by
averaging over 1.5 minutes of steady state temperature data. For boiling tests, the total heat flux
into the water from the copper was calculated from a discretized integrated total of the heat flux
from two thermocouples at each radial location divided by the total heated copper
area.
9
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
̅ ∫ (1)
Here, the value for thermal conductivity of Copper Alloy 145, = 355 W/mK and the
heated copper surface area, = 4.18 cm2 was used. This average was taken because the
asymmetry in the heaters did not provide uniform heat flux through the thermocouple locations.
However, this method agreed well with the total heat rate into the water for single phase heat
transfer measured from the bulk fluid temperature difference. This data revealed a total average
system loss of 12% of the power input, ranging from 4 to 22% over all tests.
The surface temperature, was calculated by using the average heat flux and the
̅
(2)
The distance from the surface to the top thermocouple is 1.1 mm. Because the
thermocouple distance to the top is small, using an average heat flux rather than a local heat flux
won't result in significant errors in the surface temperature. To calculate the heat transfer
coefficient a predicted bulk fluid temperature at each location was calculated from an
energy balance as the total sensible heat into the water or the saturation temperature.
̅
(3)
Because three radial thermocouple locations were recorded, the weighted average of the three
locations was taken using the corresponding radii surrounding each measurement location.
10
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Critical heat flux values for flow boiling heat transfer subject to high flux levels is typically a
result of vapor crowding on the surface. Bergles and Kandlikar [16] have discussed the distinct
nature of critical heat flux in microchannels, stating that most of the available data for critical
heat flux levels in the literature are likely a result of instabilities. Thus, models based on this data
do not predict the true level of critical heat flux in microchannels. The spiral radial inflow
microchannel device tested in this study is also subject to instabilities during flow boiling.
In order to understand the general trends of critical heat flux during flow boiling,
however, the peak heat flux levels measured were compared to three models for critical heat flux
Zuber [17] of critical heat flux values for boiling in an extensive pool. This model is based on
vapor coalescence over a horizontal heated surface and does not incorporate any additional
forced convective effects. Thus, this model will likely be a lower limit for the expected critical
heat flux levels in flow boiling. The critical heat flux for this model is found from:
( ) (4)
Lienhard and Dhir [18] previously demonstrated that for water at atmospheric pressure
and for heater sizes on the same order of magnitude to our swirl flow experimental setup, this
Zuber model was within 14% of experimental values of CHF. The second model comparison is
for subcooled flow boiling in macroscale tubes from Celata et al. [19] and based on empirical fits
to CHF data. The critical heat flux for this model is found from:
(5)
Where:
11
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
( ) (6)
{ (7)
Where P is in bar and xe is the level of subcooling at the flow exit. The last correlation
considered in this study, is the saturated flow boiling correlation proposed by Qu and Mudawar
[20] for flow boiling of water and R113 in microchannels, originally derived from correlations of
( ) ( ) (8)
This correlation is also for water at atmospheric pressure and was tested over a range of
subcooling levels. The length to diameter ratio used in Eq. (8) was tested for ratios greater than
150 and our system. Comparing channel diameter to microchannel height, our system had a flow
Boiling heat transfer experiments enabled two-phase flow characterization of the spiraling radial
inflow microchannel heat sink for heat rates up to 300 W/cm2 using distilled water as a working
fluid. Pressure drop and heat transfer data were acquired for a range of mass flow rates, heat
rates and subcooling levels. Values of the mean mass flux rates, inlet temperatures and
maximum wall heat flux levels tested on the plain copper surface are shown in Table 2. The
same values are shown for the ZnO coated surfaces in Table 3. The mean mass flux rate value
used is found from a flow length averaged mass flux rate through the channel:
∫ (9)
12
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Where G(r) is derived from a mass balance through the device, the inlet mass flow rate is
Flow images were acquired for a variety of flow conditions in order to help characterize
the modes of heat transfer and identify distinct regimes of boiling heat transfer. Boiling
experiments in the device revealed that, beyond the single phase flow regime, the flow
throughout the entirety of the device could be characterized into one of three regimes: partial
subcooled flow boiling, oscillatory flow boiling, and fully developed flow boiling. Pressure drop,
heat transfer and flow characteristics in each of these regimes are distinct. Images of the three
flow regimes on the plain copper surface are shown in Fig. 4, while bubble growth on the ZnO
During partial subcooled flow boiling, both forced convective and nucleate boiling
effects are significant. The onset of nucleate boiling in this regime leads to increased heat
transfer coefficients relative to the single phase. Sparse, bubbly flow can be seen with some
bubbles growing to the height of the channel, but major constriction of the flow due to these
bubbles does not occur. At the outlet, the bulk flow is subcooled, although vapor occasionally
gets drawn through the outlet. Pressure drop in this regime does not increase rapidly relative to
the single phase flow. For the highest flow rates tested, the flow throughout the device never
transitioned beyond the partial subcooled flow boiling regime due to high levels of subcooling
and low levels of vaporization on the surface. For the hydrophilic, coated surfaces, less bubbles
are seen to occur during subcooled boiling, indicating either smaller bubble departure size below
visible ranges which then condense back into the liquid or less active nucleation sites. This
13
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
results in the less frequent growth of larger bubbles which tend to sweep through the channel
resulting in effective evaporative heat transfer in the region below the bubble.
During oscillatory flow boiling, the system fluctuates between partial subcooled flow
boiling and fully developed nucleate boiling. Temperature and pressure oscillations were
significant in this regime due to bubble constriction of flow in the microchannel. The percentage
increase in pumping power for this regime was significantly larger than in the subcooled flow
boiling regime. This flow regime is generally undesirable due to the large oscillations in
temperatures and pressure, and thus, design constraints should be established to avoid the
oscillations from occurring. The oscillations are a result of pressure build up behind bubbles
constricting the channel, followed by subcooled liquid flooding the surface and suppressing
nucleation. The surface will then heat back up and continue this sequence in an oscillatory way,
alternating between forced convective dominated heat transfer and nucleate boiling dominated
heat transfer. A decrease in inlet subcooling levels tends to decrease the level of oscillation
because of less aggressive nucleation suppression. The coated surfaces also tended to decrease
the level of oscillations, likely as a result of sparser nucleation or smaller bubble entrainment.
The magnitude of these temperature oscillations, as well as the clear decrease in magnitude for
oscillations on the coated surface is shown in Fig. 6. As shown in Fig. 6, the onset of
temperature oscillations occurred at increasingly higher heat fluxes as the mass flux rate
increased.
During fully developed flow boiling, water vapor rapidly leaves the surface and the flow
does not sustain large oscillations, but small and rapid fluctuations of temperature and pressure
occur as vapor is continuously driven through the exit. Bubbly flow is seen throughout the
14
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
channel with coalescing of bubbles into larger slugs of vapor. This regime was always
characterized by a nonzero average exit quality and was tested up to an exit quality of 0.15.
In general, regardless of the boiling regime, the exit quality was non-equilibrium, in that
portions of the liquid were subcooled, but there was visible vapor departing from the outlet. This
was a result of the generation of large vapor slugs during all of the boiling regimes, and may
result in annular flow through the channel. In this case, one large bubble fills the majority of the
cross-section of the channel and a thin film of liquid evaporates on the surface. This mechanism
results in high heat transfer coefficients due to evaporative rather than nucleate boiling heat
transfer driving the cooling of the surface. Evidence for this is seen in the fact that large regions
of vapor are present in images of boiling yet the surface temperatures indicate dryout cannot be
occurring over these large regions. However, these regions of annular flow may be the
mechanism for initial dryout in the channel, where the wetted surface evaporates into a bubble
and dries out the surface. This mechanism of heat transfer is expected especially during boiling
on the hydrophilic coated surfaces because smaller and sparser nucleation sites are expected as
well as increased wicking of the fluid on the surface underneath large vapor slugs, allowing the
Pressure drop data was tested for all flow rates and heat fluxes. During partial subcooled flow
boiling, pressure drop was not significantly higher than the single phase pressure drop data
predicted in Ruiz [15] Fully developed and oscillatory flow boiling tend to result in increased
pressure drop across the channel. An example of a typical pressure drop curve is shown in Fig.
7. Here we compare the pressure drop across both the coated and uncoated surfaces for a mass
15
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
flux rate of 320 kg/m2s and inlet subcooling level of 80ºC. There is very little difference
between these two data sets, suggesting that the addition of nanostructures on the surface does
not negatively affect pressure drop. However, we see that roughly below 150 W/cm2, the
pressure drop remains low, but increases as two-phase oscillatory flow develops. More detailed
Heat transfer coefficients for the lowest mass flux rate tested and four different inlet subcooling
levels is shown in Fig. 8 indicating the dominant boiling regimes for each data point. This
Figure reveals that lower levels of subcooling correspond to higher heat transfer coefficients as
nucleate boiling or evaporation become the dominant method of heat transport over forced
convective effects. Additionally, the heat transfer coefficients level out as nucleate boiling
contributions reach a maximum. This same trend of increasing heat transfer coefficient values for
decreased subcooling is seen for all mass flux rates tested at varying subcooling levels. For
higher mass flux rates, the main difference in the trends is that the heat transfer coefficients did
not increase as rapidly, although higher values of heat transfer coefficients and heat fluxes were
seen, with averaged heat transfer coefficient values up to 6.5 W/cm2K at Gm = 505 kg/m2s and To
= 45°C. Heat transfer coefficients were highest when boiling or vaporization contributions were
Figure 9 shows a comparison of the heat transfer coefficients for the lowest mass flux
rate tests on the two ZnO coated surfaces and the plain Cu surface. Both coated surfaces
demonstrated enhancements in heat transfer coefficients relative to the plain copper surface. The
first coated surface tested had the most significant enhancements in heat transfer coefficients, up
16
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
to 25% higher. This enhancement occurs mostly in the oscillatory boiling regime. Studies
indicate that the temperature oscillations in the coated surfaces are decreased likely due to
reduced bubble generating regions and increased evaporative effects as vapor slugs grow in the
channel. For higher mass flux rates, the enhancements in heat transfer coefficients are less
significant as forced convective heat transfer, which is not affected by the surface chemistry,
dominates.
The maximum heat flux levels recorded during experimental tests for all three surfaces discussed
in this study are listed in Tables 2-3. The maximum value recorded is the peak heat flux levels
reached for at least two of four tests on the system for each corresponding flow rate and inlet
temperature combination. The peak heat were not recorded if either the surfaces temperature
steadily and rapidly increased, indicating thermal runaway or dryout on the surface, or if the
surface temperatures reached higher than about 175°C and were considered unstable for the test
set up. Otherwise the heat was increased at increments of about 21 W/cm2 until the maximum
experimental limit was reached, around 300 W/cm2. At the three highest mass flux rates, the
peak heat flux values were stable and had relatively low, stable surface temperatures, so it is
likely that the critical heat flux at these rates are higher than the experimental maximum tested.
From Table 2 it is evident that lower levels of inlet subcooling were associated with lower levels
of maximum heat flux values. This is likely due to the increased quality of the flow near the
outlet region resulting in higher vapor density which can result in earlier dryout of the passage.
Figure 10 shows a comparison of the peak flux values for the plain copper surface to
three models discussed in the preceding theory section. The experimental limit of about 300
17
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
W/cm2 is also plotted, and the maximum heat flux is likely higher at the highest mass flux rates.
The critical heat flux is represented in dimensionless form as the maximum boiling number of
(10)
For the Zuber model critical heat flux values, the general trend holds but under predicts
the experimental two-phase flow data by as much as 55% (with a mean absolute error of 36%).
For the Celata et al. [19] subcooled flow boiling model, the general trend holds for the
experimental data even though the diameter and subcooling levels experimentally tested are
outside the ``range of applicability" for this model. However, the model always predicts much
lower values than the experimental data suggests. The mean absolute error for this model is
42%, under predicting the critical heat flux by as much as 48%. The saturated flow
microchannel boiling model by Qu and Mudawar [20] generally under predicts the critical heat
flux by up to 28%, yet it provides the most accurate prediction with a mean absolute error of
11%. This model is the most promising predictor of critical heat flux, thus, because it
incorporates microchannel boiling data. For the last two models, the exit mass flux rate of the
experimental device is used as the G value in order to compare to the experimental data. This
would result in the highest possible prediction, but is indicative of the mass flux rate where the
initial dryout may occur because it is the location of the highest quality or lowest subcooling
levels.
Enhancements in critical heat flux due to increased wettability effects were also assessed.
Figure 11 shows the comparison between the peak heat flux values for the plain and coated
surfaces. The hydrophilic coated surfaces were able to reach higher values more consistently,
18
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
indicating slight increases in critical heat flux values for the more wetting surfaces at the lower
mass flux rates. At the highest flux rate tested for hydrophilic surfaces, Gm = 505 kg/m2s, the
peak value reached was about the same for all three surfaces. Two key contributing factors for
the observed increase in critical heat flux rates are postulated. First, studies indicated that,
especially at lower mass flux rates, the level of temperature oscillations during nucleate flow
boiling in the system are decreased. This will tend to reduce the possibility of a temperature
oscillation causing part of the surface to dryout and triggering thermal runaway. Second, the
more highly wetting surface tends to have decreased tendency to dryout due to both wicking
effects and lower contact angles, as can be seen in Fig. 12. Thus, as large vapor regions grow,
wicking of liquid underneath the vapor slugs will tend to keep the surface wetted and prevent
dryout. This is especially important at higher exit qualities which are seen at the lower mass flux
rates, where swashes of vapor may occupy a significant portion of the channel. Surface
wettability can thus allow for higher critical heat flux levels, however, these increases are only
It is of interest to explore what other design aspects could be adjusted to increase the
CHF levels further. From the data, it is clear that increasing the flow rate would result in higher
possible flux levels due to larger forced convection heat transfer contributions and higher exit
subcooling levels. This may be undesirable because it could lead to higher pumping penalties
and reduced heat transfer coefficients. Increasing the pressure in the channel is also expected to
increase critical heat flux levels. The increased density of vapor at higher pressures can lead to
more stable vaporization. However, one key disadvantage is that the higher pressure fluid will
have a higher saturation temperature. Both critical heat flux models for flow boiling predict that
19
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
increased channel pressure will result in higher critical heat flux values. Figure 13 shows the
model variation for critical heat flux with pressure plotted relative to the modeled value for
critical heat flux levels at 1 bar. For the correlation by Qu and Mudawar [20] for saturated
microchannel boiling, critical heat flux level enhancements are fairly large at increased pressure,
with values at P = 10 bar at 11 times the value for P = 1 bar. This would indicate the highest
values tested experimentally would be greater than 2500 W/cm2 at an exit pressure of 10 bar.
The Celata et al. model [19] for subcooled flow boiling has less drastic critical heat flux
enhancements at increased pressure. The critical heat flux value is modeled to increase up to 440
W/cm2 at P = 5.5 bar which is the recommended upper limit of validity for the model.
CONCLUDING REMARKS
Flow boiling heat transfer in a spiraling radial inflow microchannel heat sink was tested
and characterized. The device is simple to manufacture and can easily achieve high heat fluxes
with relatively low, uniform surface temperatures. High heat transfer performance was shown
resulting from distinct features of this type of device. Enhanced forced convective coefficients
have been shown in single phase flow due to unsteady and secondary flow motions, with similar
enhancements seen in two-phase flow. Increased drag and buoyant forces on vapor also will tend
to draw bubbles into the flow from all regions of the microchannel. Finally, evaporation on the
surface of vapor slugs during annular flow in the channel leading to enhanced heat transfer
coefficients. The sweeping motion of these vapor regions also entrain other bubbles into the
flow. The annular flow region is especially expected to provide high heat transfer coefficients on
the superhydrophilic surface where fluid wicking underneath bubbles helps to prevent dryout and
promote effective vaporization. On the plain copper surface heat transfer coefficients were seen
20
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
up to 6.5 W/cm2K during fully developed boiling in the channel, revealing promising cooling
Maximum heat flux data is compared to critical heat flux models for flow in
microchannels. Increased surface wettability provides enhanced critical heat flux levels, up to
about 16%. Further potential for enhancements of critical heat flux are proposed including
increased inlet pressure. Heat fluxes on the order of 1000 W/cm2 can be expected with moderate
pressure increases of around 10 bar. For water, this would correspond to an increase in
saturation temperature up to 180°C which would be undesirable for electronics applications, but
could be used for other high flux cooling applications. Alternate working fluids should be
investigated for lower surface temperatures. The ZnO coated surfaces showed durability during
testing, retaining their contact angles throughout the testing process over weeks of use. Using
this type of coating could be beneficial for cooling applications where long term enhancements
NOMENCLATURE
A area, cm2
Bo Biot number
b gap size, μm
g gravity, m/s2
21
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
P pressure, bar
Re Reynolds number
T temperature, ºC
We Weber number
Greek Symbols
ρ density, kg/m3
Subscripts
l liquid property
22
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
sat saturated
v vapor property
23
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
REFERENCES
[1] Agostini, B., Fabbri, M., Park, J.E., Wojtan L., Thorne, J.R., and Michel, B., State of the
Art of High Heat Flux Cooling Technologies, Heat Transfer Engineering, vol. 28, no. 4, pp.
258-281, 2007.
[2] Kandlikar, S.G., Colin, S., Peles, Y., Garimella, S., Pease, R.F., Brandner, J.J., and
[3] Koo, J.-M., Im, S., Jiang, L., and Goodson, K.E., Integrated Microchannel Cooling for
Three-Dimensional Electronic Circuit Architectures, Journal of Heat Transfer, vol. 127, no.
[4] Kim, S.-M., and Mudawar, I., Review of Databases and Predictive Methods for Heat
[5] Ebadian, M., and Lin, C., A Review of High-Heat-Flux Heat Removal Technologies,
[6] Leland, J., and Chow, L., Channel Height and Curvature Effects on Flow Boiling from an
Electronic Chip, Journal of Thermophysics and Heat Transfer, vol. 9, no. 2, pp. 292-301,
1995.
[7] Galloway, J.E., and Mudawar, I., Critical Heat Flux Enhancement by Means of Liquid
Heat and Mass Transfer, vol. 35, no. 5, pp. 1247-1260, 1992.
24
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
[8] Sturgis, J., and Mudawar, I., Assessment of CHF Enhancement Mechanisms in a Curved,
[9] Sturgice, J.C., and Mudawar, I., Critical heat Flux in a Long, Curved Channel Subjected to
Concave Heating, International Journal of Heat and Mass Transfer, vol. 42, no. 20, pp.
3831-3848, 1999.
[10] Liu, T.Y., Li, P., Liu, C., and Gau, C., Boiling Flow Characteristics in Microchannels with
[11] Qu, W., and Mudawar, I., Prediction and Measurement of Incipient Boiling Heat Flux in
Micro-Channel heat Sinks, International Journal of Heat and Mass Transfer, vol. 45, no.
[12] Yao, Z., Lu, Y-W., Kandlikar, S.G., Micro/Nano Hierarchiccal Structure in Microchannel
Heat Sink for Boiling Enhancement, Proc. of IEEE 25th International Conference on Micro
[13] Yang, Q., Li, Y., Yin, Q., Wang, P., Cheng, Y.-B., Hydrothermal Synthesis of Bismuth
Oxide Needles, Materials Letters, vol. 55, no. 1, pp. 46-49, 2002.
[14] Padilla, J., Carey, V.P., An Experimental Study of the Leidenfrost Transition for Water on
Nanostructured Superhydrophilic Surfaces, Proc. of 15th Int. Heat Trans. Conf., IHTC 15-
9581, 2014.
25
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
[15] Ruiz, M., & Carey, V. P. Experimental study of single phase heat transfer and pressure loss
in a spiraling radial inflow microchannel heat sink. Journal of Heat Transfer, vol. 137 no.7,
[16] Bergles, A., and Kandlikar, S., On the Nature of Critical heat Flux in Microchannels,
[17] Zuber, N., Hydrodynamic Aspects of Boiling Heat Transfer (thesis), Technical Review,
California University, Los Angeles, and Ramo-Wooldridge Corp., Los Angeles, 1959.
[18] Lienhard, J.H., and Dhir, V.K., and Riherd, D.M., Peak Pool Boiling Heat Flux
Measurements on Finite Horizontal Flat Plates, ASME Journal of Heat Transfer, vol. 95,
[19] Celata, G., Cumo, M., and Mariani, A., Assessment of Correlations and Models for the
[20] Qu, W., and Mudawar, I., Measurement of Correlation of Critical Heat Flux in Two-Phase
Micro-Channel Heat Sinks, International Journal of heat and Mass Transfer, vol. 47, no.
[21] Katto, Y., and Ohno, H., An Improved Version of the Generalized Correlation of Critical
heat Flux for the Forced Convective Boiling in Uniformly Heated Vertical Tubes,
International Journal of Heat and Mass Transfer, vol. 27, no. 9, pp. 1641-1648, 1984.
[22] Ruiz, M., Characterization of Single Phase and Two Phase Heat and Momentum Transport
26
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Parameter Size
b 300 μm
ro 1 cm
ri 0.15 cm
̅ ̅ 6
27
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
TABLE 2: Range of mean mass flux rates, Gm, corresponding inlet temperatures and max heat
618 21 301
716 21 293
28
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
TABLE 3: Range of mean mass flux rates, Gm, and corresponding inlet temperatures and max
184 22 196
320 22 271
505 23 293
184 22 193
255 22 227
320 22 244
505 24 284
29
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
30
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
31
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
32
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 4: Flow images in the three flow boiling regimes for the plain copper surface. A: Partial
subcooled flow boiling, B: oscillatory flow boiling, C: fully developed flow boiling
33
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 5: Flow images of bubble growth at 184 kg/m2s and heat flux of 121 W/m2 on the
34
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 6: Root mean squared of temperature oscillations on the heat sink surface for coated and
35
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 7: Recorded pressure drop data across the microchannel for the coated and uncoated heat
36
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 8: Heat transfer coefficients for a mean mass flux Gm = 184 kg/m2s for four
37
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 9: Comparison of the heat transfer coefficients for the ZnO coated surfaces and the plain
copper surfaces for tests and Gm = 184 kg/m2s with subcooling levels of about 80ºC
38
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 10: Experimental maximum heat flux rates tested on the plain copper surface
39
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 11: Comparison of the maximum heat flux for the ZnO coated surfaces to the plain
40
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 12: Contact angle comparison of the plain copper surface to the two ZnO
nanostructure coated hydrophilic surfaces. Side view and top view shown. (White spots on
41
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Figure 13: Effect of pressure variation on critical heat flux for two models, the value of the
42
ACCEPTED MANUSCRIPT