Available online at www.sciencedirect.

com

International Journal of Multiphase Flow 34 (2008) 1008–1022

www.elsevier.com/locate/ijmulflow

Pool boiling heat transfer of ultra-light copper foam with open cells
Jinliang Xu a,*, Xianbing Ji a,b, Wei Zhang a, Guohua Liu a
a
Micro-Energy System Laboratory, Key Laboratory of Renewable and Gas Hydrate, Guangzhou Institute of Energy Conversion,
Chinese Academy of Science, Guangzhou 510640, PR China
b
Graduate School of Chinese Academy of Science, Beijing 100080, PR China

Received 12 December 2007; received in revised form 11 May 2008

Available online 6 June 2008

Abstract

High speed visualizations and thermal performance studies of pool boiling heat transfer on copper foam covers were performed at
atmospheric pressure, with the heating surface area of 12.0 mm by 12.0 mm, using acetone as the working fluid. The foam covers have
ppi (pores per inch) from 30 to 90, cover thickness from 2.0 to 5.0 mm, and porosity of 0.88 and 0.95. The surface superheats are from
20 to 190 K, and the heat fluxes reach 140 W/cm2. The 30 and 60 ppi foam covers show the periodic single bubble generation and
departure pattern at low surface superheats. With continuous increases in surface superheats, they show the periodic bubble coalescence
and/or re-coalescence pattern. Cage bubbles were observed to be those with liquid filled inside and vented to the pool liquid. For the
90 ppi foam covers, the bubble coalescence takes place at low surface superheats. At moderate or large surface superheats, vapor frag-
ments continuously escape to the pool liquid.
Boiling curves of copper foams show three distinct regions. Region I and II are those of natural convection heat transfer, and nucleate
boiling heat transfer for all the foam covers. Region III is that of either a resistance to vapor release for the 30 and 60 ppi foam covers, or
a capillary-assist liquid flow towards foam cells for the 90 ppi foam covers. The value of ppi has an important effect on the thermal per-
formance. Boiling curves are crossed between the high and low ppi foam covers. Low ppi foams have better thermal performance at low
surface superheats, but high ppi foams have better one at moderate or large surface superheats and extend the operation range of surface
superheats. The effects of other factors such as pool liquid temperature, foam cover thickness on the thermal performance are also
discussed.
Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Copper foam; Pool boiling; Boiling pattern; Cage bubble; Boiling curve

1. Introduction The primary issues are the mitigation of incipience temper-

ature overshoot, enhancement of nucleate boiling heat
Compared with forced convection heat transfer, pool transfer and increasing the critical heat flux (CHF).
boiling heat transfer is a more attractive alternative scheme Rainey and You (2000) examined the pool boiling of a
because it is generally less complex and easier to seal ‘‘double enhancement” technique which required a surface
(Rainey and You, 2000). Enhancements in pool boiling enhancement (micro-porous coating) and an extended area
heat transfer can be fulfilled by increasing the active nucle- enhancement (square pin fins) using 1 cm2 flush-mounted
ation sites and/or the bubble detaching frequency, which copper surfaces. An increase in surface roughness strongly
could decrease the surface superheat and increase the boiling increases the nucleate boiling heat transfer coefficient and
heat flux (Parker and El-Genk, 2005). Honda and Wei (2004) critical heat flux, due to the increased active nucleation site
gave a review on the enhanced boiling heat transfer from density. It is believed that the presence of the pin fins pro-
electronic components by use of surface micro-structures. duces a resistance to vapor bubble departure, which
increases the bubble residence time and causes a change
*
Corresponding author. Fax: +86 20 87057656. in the boiling curve slope of both plain and micro-porous
E-mail address: xujl@ms.giec.ac.cn (J. Xu). finned surfaces.

0301-9322/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmultiphaseflow.2008.05.003
 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022 1009

Modulated (periodically non-uniform thickness) por- granular media, for which the porosity varies from 0.3 to
ous-layer coatings, as shown by Liter and Kaviany 0.6. In the past several years, some studies were performed
(2001), can enhance the pool boiling critical heat flux on the pool boiling heat transfer in metallic foams. Arbe-
nearly three times over that of a plain surface. The modu- laez et al. (2000) studied pool boiling of FC-72 in highly
lation separates the liquid and vapor phases, thus reducing porous metal foam heat sinks. Experiments were con-
the liquid–vapor counterflow resistance adjacent to the ducted with samples of porosities and pore sizes in the
heated surface. Two independent mechanisms being capa- range of 90–98% and 5–40 ppi, respectively. The results
ble of causing the liquid choking that leads to the critical show that the temperature excursion usually observed for
heat flux were suggested. fluorinert fluids at the onset of nucleate boiling is not pres-
Kim et al. (2002) investigated the mechanism of nucleate ent. For the same pore size, the low porosity samples exhi-
boiling heat transfer enhancement from micro-porous sur- bit a significantly higher heat transfer coefficient in the low
faces in saturated FC-72. The test section was a 390-lm heat flux regimes. For similar porosity, enhanced heat
diameter platinum wire coated with micro-porous particles. transfer is observed with an increase in the value of ppi.
The nucleate boiling heat transfer is augmented through Athreya et al. (2002) investigated the effects of orienta-
increased latent heat transfer in the low heat flux region tion and geometry on the pool boiling heat transfer of
and increased convection heat transfer in the high heat flux FC-72 in high porosity aluminum metal foam heat sinks.
region. Hysteresis and incipience excursion are almost absent when
Ghiu and Joshi (2005) visualized pool boiling from thin the foam is vertically positioned. High ppi samples deteri-
confined enhanced structure using a dielectric fluorocarbon orate heat transfer performance in the vertical orientation.
liquid (PF 5060) as the working fluid at atmospheric pres- The low ppi sample shows the first decreased and then the
sure. Visualizations show the presence of two boiling increased heat transfer characteristics with reduction in
regimes: slug predominance regime, slugs and plugs regime. foam heights. The temperature jump, marking transition
Vapor slugs exist in the top and bottom channels even for from nucleate to film boiling, and the critical heat flux both
lowest heat fluxes and exhibit an oscillatory movement, increase with decreasing height in the low ppi sample for
indicating the transient nature of internal evaporation. both orientations.
Parker and El-Genk (2005) experimentally studied Moghaddam and Ohadi (2003) studied pool boiling heat
enhancements in nucleate boiling of FC-72 dielectric liquid transfer of water and FC-72 on thin blocks bonded with
on porous graphite and compared results with those on a copper foams of 80 ppi, 90% porosity, 30 ppi, 95% porosity
smooth copper surface of the same size (10  10 mm). and graphite foam of 75% porosity. Significant enhance-
The surface temperature excursion at boiling incipience ment was achieved in boiling of water on the 30 ppi copper
and critical heat fluxes are compared with those of other foams, while no enhancement was observed on the 80 ppi
investigators on copper, silicon, micro-finned silicon sur- copper and graphite foams. A substantial enhancement
faces and micro-porous coatings. It is shown that there is was achieved on all the foams with FC-72 as the working
no temperature excursion at boiling incipience. Both the fluid.
nucleate boiling heat transfer coefficients and critical heat Coursey et al. (2005) investigated the thermal perfor-
fluxes are increased with increases in liquid subcooling. mance of a graphite foam thermosyphon evaporator and
Nimkar et al. (2006) studied pool boiling characteristics discussed the foam’s potential for use in the thermal man-
of pyramidal shaped re-entrant cavities etched in silicon. agement of electronics. Using the graphite foam as the
Experiments were performed in saturated FC-72 at atmo- evaporator in a thermosyphon enables the transfer of large
spheric pressure. High speed visualization was performed amounts of heat, with low temperature difference and with-
to record and quantify the bubble departure frequency, out the external pumping. The system performance with
the departure diameter, the active site density, and to FC-72 and FC-87 was examined, and the effects of liquid
observe the effect of interaction between neighboring nucle- fill level, condenser temperature, and foam height, width,
ation sites. and density were studied. Performance was found to be
Meléndez and Reyes (2006) performed enhancements in similar with FC-72 and FC-87. The liquid fill level, con-
pool boiling heat transfer using iron wool and stainless denser temperature, geometry and density, of the graphite
steel wool with binary mixture as the working fluid. It is foam were found to influence the thermal performance sig-
found that the combination of porous covers and positive nificantly. The boiling was found to be surface tension
binary mixture reduces the bubble size and coalescence, dominated, and a simple model based on heat transfer
thus increases the liquid supply towards the heating sur- from the outer surface is proposed. The maximum heat flux
face, causing an increase in the convective heat transfer of 149 W/cm2 was reached.
coefficient for pool boiling. For limited vapor-release con- The above review on pool boiling heat transfer with
ditions, the bubbles grow with no increase in heat transfer enhanced micro-structures shows that the general agree-
coefficient, while for resistance-free vapor escape the bub- ment is not reached at this stage. The objective of this
bles radii increase almost linearly. paper is to pursue the possibility that metal foams can be
The available studies of flow and heat transfer in porous used for boiling heat transfer enhancement. We perform
media concern low porosity media such as packed beds and experiments over wide parameter ranges, not only for the
1010 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022

lower surface superheat applications, but also for the high  Identify the effects of ppi, thickness of foam cover, and
surface superheats and heat fluxes. The objective of the pool liquid temperature on the thermal performance.
paper is summarized as follows:

 Perform high speed visualizations of pool boiling heat 2. Copper foam and the characteristic parameters
transfer on copper foam covers, to identify boiling
patterns. Fig. 1 shows pictures of various copper foams, for the
 Perform thermal performance measurements of ppi of 30, 60 and 90, porosity of 0.88 and 0.95. The foam
enhanced boiling heat transfer using copper foams, over has an open-celled structure composed of dodecahedron-
very wide parameter ranges. like cells, having 12–14 pentagonal or hexagonal faces.
 Explain thermal performance of pool boiling heat Two parameters characterizing foam cells are the ppi and
transfer on copper foams by the observed boiling porosity. The crosssection of ligaments depends on poros-
patterns. ity, and changes from a circle at e = 0.85 to an inner

Fig. 1. Copper foam pictures.

 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022 1011

concave at e = 0.97 (Calmidi, 1998), where e is the porosity.

a
A unit cell of the foam is shown in Fig. 2, with the assumed
tetrakaidecahedron shape. The ligament length is l with pffiffiffi its
diameter of df. Thus the volume of a unit cell ispVffiffiffiffiffi¼ 8 2l3 .
The circumcircle diameter of the foam cell is 10l, which
pffiffiffiffiffiof dp, not considering
can be regarded as the pore diameter
the ligament thickness, i.e., d p ¼ 10l.
Table 1 shows the foam cell parameters measured by a
Leica M-type microscope (Germany). It is seen that larger
ppi yields smaller pore diameter of dp and ligament diame-
ter of df. When ppi is fixed, larger porosity such as 0.95
leads to slight larger dp and smaller df, compared with
the lower porosity of 0.88.

3. Experimental setup and procedures

3.1. The copper block welded with copper foam cover

The copper block was used to fabricate the test section.

Fig. 3a shows geometry and size of the test section. In the
bottom part of the copper block there are five 6.0-mm
diameter holes, in which five cartridge heaters were
inserted, providing heating power to the copper block.
Each heater provides a maximum power of 100 W at the b
8
applied AC voltage of 220 V. In the middle part of the cop-
per block, there are four 1.0-mm diameter holes, inside 7
which four K-type thermocouples are inserted. The top 9 air
of the copper block is a rectangular plate, having a thick- 10
ness of 3.0 mm. A plain smooth, sand polished copper sur- tap water
face is regarded as the reference for boiling heat transfer 6
experiments, with the size of 12.0 mm by 12.0 mm. Alterna- 5
tively, the top copper surface was welded with copper foam
11

12

4
13
3

1
insulation material

Fig. 3. Copper block test section and experimental setup. Note: all
dimensions are in mm in (a).
Fig. 2. A unit foam cell represented by a tetrakaidecahedron unit.

Table 1 covers, with the thickness of 2.0, 3.0, 4.0 and 5.0 mm,
Parameters of foam cells used in the present paper respectively. Once the copper block was ready, it was
Ppi e dp (mm) df (mm) l (mm) df/dp cleaned by methanol and baked in an oven. Then it was
taken out of the oven and heated by the cartridge heaters
30 0.88 2.762 0.314 1.074 0.114
30 0.95 3.285 0.286 1.181 0.087 until its temperature reached the melting temperature of
60 0.88 1.192 0.141 0.486 0.118 the tin at the top copper surface, leaving a thin tin film.
60 0.95 1.491 0.124 0.541 0.083 We put a clean foam cover on the copper surface and
90 0.88 0.696 0.081 0.275 0.116 turned off the cartridge heaters, thus the copper foam
90 0.95 0.772 0.064 0.299 0.083
was welded with the copper block tightly, which is ready
1012 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022

for experiments. The welding procedure ensures small ther-

mal resistance between the copper block and the foam
cover.
After the experiments, we removed the foam cover from
the copper block when the copper block temperature
approached the melting temperature of tin by heating the
cartridge heaters. Then we sliced the foam cover in its
thickness direction and visualized the separated piece of
the foam cover by a microscope, showing that the tin thick-
ness wicked into the foam cells is about 0.1 mm.

3.2. Experimental setup

Fig. 3b shows the experimental setup. A transparent

glass chamber with the size of 125  127  145 mm was
used to contain liquid. The bottom of the glass chamber
was a stainless steel plate (3), at the center of which a rect-
angular hole was drilled. The copper block (2) was fitted
with the plate (3) by filling Teflon and epoxy glue between
them for seal, ensuring the copper foam exposed in the
pool liquid. The part of the copper block under the plate Fig. 4. Experimental setup and measurement system.
(3) was surrounded by a glass sheath (1). As the thermal
insulation material, glass fiber was filled in the gap between
the copper block (2) and the glass sheath (1), as shown in could obtain an adjustable voltage from 0 to 220 V, yield-
Fig. 3b. ing a specific power supplied to the copper foam from 0 to
The top cover of the glass chamber was a stainless steel 500 W. A precise power meter gave the power reading.
plate (6) with a 2.0 mm thickness. An inclined 6.0-mm A Hewlett-Packard data acquisition system was used to
diameter coiled copper tube (11) was arranged along the collect the pool liquid temperature and four thermocouple
internal wall surface of the glass chamber. The two ports signals (see Fig. 3a). A high speed camera system (HG-
of the coiled tube penetrated the two holes of the plate 100K, Redlake Inc., USA) was used to capture the boiling
(6). The copper tube outside of the glass chamber was con- patterns. The camera used the advanced 1.7 Megapixels
nected with a tap water system. The pool liquid tempera- CMOS sensor, which has a maximum recording rate of
ture could reach a desired value by adjusting the flow 20,000 images per second, with the sensitivity of
rate of tap water in the coiled tube (6) using the valve 1504  1128 pixels. In order to clearly observe the boiling
(10). An auxiliary heater (12) was installed in a corner of pattern in and out of foam cells, a micro-lens was adapted
the glass chamber. During the boiling experiment, when in front of the camera. In the present study, the recording
the pool liquid temperature was below the desired value, rate of 5000 images per second was used.
the auxiliary heater (12) was turned on while the valve The experimental procedure involved an initial charge of
(10) was turned off. Such cases only occurred at very small liquid in the glass tank and removing non-condensable gas.
heating power applied on the test section. However, most The copper foam cover was horizontally positioned. The
cases needed to maintain an adjustable flow rate of the top liquid level was higher than the top foam cover by
tap water in the tube (11), with the auxiliary heater (12) 100 mm. Before the experiment, the cartridge heaters were
turned off. The produced vapor during the boiling entered turned on to vigorously boil the liquid for one hour to
a reflux condenser (7). The condensed liquid returned to remove non-condensable gas in foam cells and liquid.
the glass chamber by gravity. Outside of the condenser The heaters were then turned off and the liquid was cooled
was a fin heat sink cooled by forced convective air. The to reach the environment temperature. The working fluid
condenser was vented to atmosphere by the valve (8) was acetone, with the thermo-physical properties given in
through a side branch tube. Thus atmospheric pressure Table 2 (Yaws, 1999). Most of previous studies used FC-
was always kept in the glass chamber. The pool liquid tem- 72 as the working fluid. Acetone has the surface tension
perature was measured by a K-type thermocouple (9). of 0.0192 N/m, comparable with FC-72 for the value of
0.00835 N/m, at room temperature. At atmospheric pres-
3.3. Measurement system and experimental procedures sure, the latent heat of evaporation is 512.94 kJ/kg for ace-
tone, significantly larger than the value of 94.80 kJ/kg for
Fig. 4 shows the measurement system. Five cartridge FC-72.
heaters were driven by a power supply system, consisting For each experiment, we started from a small heat flux
of a 220 V voltage stabilizer, a voltage transformer and a 1–2 W/cm2 on the copper foam and fixed the pool
power meter. By adjusting the voltage transformer, one liquid temperature. If the variation of the copper block
 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022 1013

Table 2
Physical properties of acetone at atmosphere pressure
Tsat (°C) qf (kg/m3) Cpf (J/kg K) Cpg (J/kg K) hfg (kJ/kg) r (N/m) lf (Pa s) kf (W/mK)
56.29 748.01 2302.5 1380.6 512.94 0.0192 0.000237 0.518
Where Tsat is the saturated temperature, qf is the liquid density, Cpf and Cpg are the specific heat of liquid and vapor, hfg is the latent heat of evaporation,
r is the surface tension, lf is the liquid viscosity, kf is the thermal conductivity of liquid.

temperature was smaller than 1 °C in 10 min, the heat We estimate the non-condensable gas concentration in
transfer process was considered to reach a steady state. acetone. In terms of Kretschmer et al. (1946), the mole
After that, we recorded the pool liquid temperature, the fraction of dissolved air in acetone is about 6  104 at
four temperatures on the copper block and the power room temperature of 25 °C and atmospheric pressure of
meter reading. We recorded transient boiling patterns by 101.3 kPa. The saturated vapor pressure of acetone is
the high speed camera, noting that not all the cases were expressed as (Yaws, 1999)
recorded by the camera. p  2469
We then increased the heat flux by a small step of 2–5 log10 v  1:013  105 ¼ 28:588 
760 T bulk
W/cm2, and repeated the above procedure. Attention was
paid at boiling incipience, around which a very small incre-  7:351  log10 T bulk þ 2:736  106  T 2bulk ð4Þ
ment of heat flux 1 W/cm2 was used. Same procedure was
where pv and Tbulk have the units of Pascal and Kelvin.
performed for all the experiments in this paper.
The values of pv are 30.6, 52.1 and 75.9 kPa at the three
The heat flux was computed as q ¼ k dT j
dz base surface
temperatures of 25, 38 and 48 °C, respectively. It is difficult
based on the one-dimensional heat conduction equation,
to measure the non-condensable gas concentration, but we
where k is the copper thermal conductivity, dT j
dz base surface
is
calculate the gas concentration in acetone based on the
the temperature gradient at the base surface, z is the coor-
Henry’s law: Sgas = kHpgas, where Sgas is the gas solubility,
dinate perpendicular to the base surface. A least square
kH is the Henry’s constant, pgas is the gas pressure which is
correlation of temperatures versus z was written as
pgas = p  pv, p is the pressure in the glass chamber of this
T = a0 + a1z, where a0 and a1 are constants correlated
study. Our estimation gives the mole fraction of air in
based on T1, T2, T3 and T4 (see Fig. 3a). The heat flux
acetone is 4.2  104 and 2.2  104 at the liquid tempera-
uncertainty was estimated to be smaller than 6.0%. The
tures of 38 and 48 °C, respectively, showing the decreased non-
surface superheat DTsat is defined as the surface tempera-
condensable gas concentration with increasing temperatures.
ture of Tw minus Tsat, where Tw is the temperature at the
This study covers the following data ranges: ppi of 30,
base surface, Tsat is the saturated temperature of acetone
60, 90; porosity of 0.88 and 0.95; foam cover thickness of
at atmospheric pressure. The heat transfer coefficient is cal-
2.0, 3.0, 4.0 and 5.0 mm; surface superheat from 20 to
culated as
190 K; surface heat flux up to 140 W/cm2. It is noted that
h ¼ q=ðT w  T bulk Þ ð1Þ the heat flux is based on the top copper surface area of
where Tbulk is the pool liquid temperature. The surface 12.0 mm by 12.0 mm. The foam cell area is not involved
temperature, surface superheat and pool liquid tempera- in the computation of heat flux.
ture have the maximum uncertainties of 0.3 °C.
Now we estimate uncertainty of the heat transfer coeffi- 4. Results and discussion
cient, which is a function of three independent variables of
q, Tw and Tbulk . The uncertainty of h is computed as 4.1. Visualization of boiling patterns
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2  2
oh oh oh 4.1.1. Visualization for the 30 ppi foam covers
Dh ¼ Dq2 þ DT 2w þ DT 2bulk ð2Þ
oq oT w oT bulk High speed visualizations were performed. Fig. 5a shows
the boiling pattern near the boiling incipience at q = 8.3
Thus the relative uncertainty can be calculated by
W/cm2 and DTsat  4 K. The copper foam has the ppi of
substituting Eq. (1) into Eq. (2) as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 30 and porosity of 0.88 with the thickness of 3.0 mm. Min-
2  2  2ffi iature bubbles nucleate at the foam ligaments and fiber
Dh Dq DT w DT bulk
¼ þ þ ð3Þ junctions in foam cells. Initial nucleation sites mostly take
h q T w  T bulk T w  T bulk
place at specific locations. The nucleated bubble evolves a
In Eqs. (2) and (3), Dq, DTw and DTbulk are the uncer- growth and departure process, finally is released to the pool
tainties of heat flux, wall surface temperature and pool liquid. Fig. 5b shows the miniature bubble nucleation,
liquid temperature. The maximum uncertainty of h is growth and release to the pool liquid at q = 10.4 W/cm2.
obtained by a smaller surface temperature and a larger Active nucleation sites are increased compared with those
pool liquid temperature. Thus we acquire the maximum shown in Fig. 5a. The steady state boiling heat transfer
relative uncertainty of h as 8.52%. is well maintained with foam cells immersed in liquid.
1014 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022

Fig. 5. Boiling pattern at boiling incipience and low heat flux (ppi = 30, e = 0.88, Tbulk = 38 °C, d = 3.0 mm, DTsat equals to 3.7 K for (a), 4.1 K for
(b) and 8.0 K for (c–h)).

A small increase in surface superheat greatly raises the heat at the specific area, such as the foam cover center. In con-
flux. Photographs in Fig. 5a and b show the periodic single trast to Fig. 5a and b, Fig. 5c–h shows the periodic bubble
bubble generation and departure pattern without coalescence pattern, noting that cage bubble appears in
coalescence. Fig. 5c–h and it will also be shown in the following figures,
Increases in heat fluxes lead to bubble coalescence the cage bubble formation will be given in Section 4.1.2.
before release to the pool liquid, as shown in Fig. 5c–h at Further increase in heat fluxes yields spreading of
q = 20.8 W/cm2. A new cycle begins at the time when a merged bubbles over the whole foam cover. Fig. 6 shows
cage bubble is fully separated from the foam cells and the periodic bubble coalescence and re-coalescence pattern
enters the pool liquid (Fig. 5c). Cage bubble comes from at q = 83.3 W/cm2. A similar cycle definition is used. A
the coalescence of miniature bubbles in foam cells. Then mushroom-shaped cage bubble is fully separated from the
the boiling heat transfer is sustained with nucleation and foam cover, defined as the beginning of a new cycle shown
growth of miniature bubbles in foam cells for a longer time in Fig. 6a. Miniature bubbles generated in foam cells are
such as 55.5 ms, which can be considered as the waiting quickly merged to form a set of cage bubbles, covering
time for bubble coalescence. A new cage bubble appears the whole foam cover (Fig. 6b and c). The waiting time
and gradually penetrates the foam cover for t > 55.5 ms for bubble coalescence is greatly shortened and not appar-
(Fig. 5d–g), until it is out of the foam cover and the present ent. This is the first stage for bubble coalescence. Bubble re-
cycle ends at t = 70.0 ms (Fig. 5h). Thus the cycle period is coalescence begins at t = 19.0 ms (Fig. 6c). There are six
about 70 ms and the formation/release frequency of the merged bubbles at that time. Then these bubbles are re-
merged bubble is 14.3 Hz. Because the heat flux is not suf- merged to have two cage bubbles (Fig. 6d and e). Finally
ficiently high, coalescence of miniature bubbles takes place a single big mushroom-shaped cage bubble is released to
 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022 1015

Fig. 6. Boiling pattern at q = 83.3 W/cm2 (ppi = 30, e = 0.88, Tbulk = 38 °C, d = 3.0 mm).

the pool liquid at t = 59.0 ms (Fig. 6f), ending the present a typical encircled cage bubble structure at the heat flux
cycle. Again cage bubbles have ‘‘open” centers filled with same as that of Fig. 7. Two cage bubbles are shown with
liquid. Fig. 6 shows the periodic bubble coalescence and the smaller one inside the bigger one. The bigger bubble
re-coalescence pattern. is ‘‘open” so that the smaller one can be seen. Fig. 8 pro-
A similar boiling pattern is shown in Fig. 7 at vides the direct experimental evidence of the cage bubbles.
q = 112.0 W/cm2 just before CHF. Boiling heat transfer A rough explanation of the cage bubble is given here.
is so violent that bubble coalescence evolves very quickly When miniature bubbles merge to a big bubble, the big bub-
to form a very big mushroom-shaped cage bubble (Fig. ble has the size larger than a unit foam cell. A foam ligament
7d and e). The cage bubble can surround the whole foam changes its direction when it crosses the ligament junction.
cover, including the four sides of the foam cover. As the The arranged foam ligaments stab the big bubble, or we
time during which the whole foam cover is occupied by can say that the big bubble is plunged by the continuous lig-
the vapor phase consists of a larger percentage of a full aments. When the big bubble is departing from the station-
cycle period, liquid suction towards the foam cells becomes ary foam cells, the space that was previously occupied by
difficult. A small increase in heat flux such as 1 W/cm2 the foam ligaments and junctions inside the bubble should
leads to an uncontrollable temperature rise of the copper be further replaced by liquid. The newly formed liquid vol-
block, reaching the CHF condition. ume is from the integrated flow rate of the liquid film around
the foam ligament surfaces. The liquid flow rate is sucked
4.1.2. Cage bubble formation from the pool liquid to the inner bubble. When the big bubble
Enhanced boiling heat transfer has been extensively is fully detached from foam cells, the space that was previ-
studied for several decades. However, cage bubbles are ously occupied by the foam ligaments and junctions before
not reported previously. Again, Fig. 8a and b shows the bubble departure becomes liquid filled and vented to the pool
enlarged cage bubble structure for the same run case of liquid. Surface tension force smoothes vapor–liquid inter-
Fig. 6. There are two cage bubbles shown, marked as the face inside the bubble, forming a single liquid filled center.
front one and the behind one. Both of them have a vapor Cage bubble is formed due to the connected web structure
shell and a liquid center vented to the pool liquid. The of foam ligaments through junctions. The cage bubble for-
direction of the liquid center is perpendicular to the paper mation needs further investigation.
plane. The front bubble is ‘‘open” so that the vapor shell of Wei and Honda (2003) visualized boiling patterns of
the behind one can be completely visualized. Fig. 8c shows FC-72 from silicon chips with micro-pin-fins immersed in
1016 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022

Fig. 7. Boiling pattern at q = 112.0 W/cm2 (ppi = 30, e = 0.88, Tbulk = 38 °C, d = 3.0 mm).

FC-72. Single miniature bubble at low heat fluxes and big- Fig. 10 shows boiling patterns at the heat fluxes of 41.0,
ger merged bubble at high heat fluxes were reported. 62.5 and 118.1 W/cm2. Boiling process is violent without
Parker and El-Genk (2005) presented photographs of cycle behavior demonstrated. Vapor fragments are contin-
nucleate boiling of FC-72 on the porous graphite. They uously departing to the pool liquid. Liquid is sucked
showed the higher density of detaching bubbles from the towards the foam cover through miniature gaps between
porous graphite surface, which is consistent with the mea- neighboring fragments (counter-current vapor–liquid
sured enhancements in nucleate boiling heat transfer coef- flow). Small pore size of large ppi foam cells ensures strong
ficients. Other visualizations of bubble dynamics during pumping effect for liquid suction towards the foam cells,
pool boiling on enhanced structures were given by Chen which is helpful to maintain liquid film around the foam
et al. (2005), Ghiu and Joshi (2005), Nimkar et al. (2006), ligaments and junctions at moderate or high heat fluxes.
Meléndez and Reyes (2006). Cage bubble is not reported
in these references. 4.2. Boiling heat transfer of acetone on the foam covers

4.1.3. Visualization for the 90 ppi foam covers Boiling curves are shown in Fig. 11. The value of ppi is
The observed boiling patterns on the 60 ppi foam covers 30, 60 and 90 at e = 0.88 and d = 3.0 mm, where d is the
are similar to those on the 30 ppi ones. Now we show boil- foam cover thickness. Three pool liquid temperatures of
ing patterns on the 90 ppi foam covers. The larger the value 38, 48 and 55 °C were used. The pool liquid temperature
of ppi, the smaller the pore diameter of dp is. As shown in of 55 °C approaches the saturated temperature of acetone
Table 1, dp is 2.76 mm for the 30 ppi foam at e = 0.88. But at atmospheric pressure. The surface superheats vary over
dp is decreased to 0.70 mm for the 90 ppi foam at the same a wide range from 20 to 190 K, which is helpful not only
porosity. The small pore size for large ppi foams leads to for electronic cooling applications, but also for other appli-
easy coalescence of miniature bubbles at low heat fluxes, cations with much high heat flux that require large surface
see Fig. 9a and b at q = 5.8 and 6.9 W/cm2, respectively. temperature. Boiling curves on the plain smooth surface
A single merged cage bubble is generated and detached are shown for comparisons.
periodically. Fig. 9c–f shows the boiling pattern for a full Fig. 11 demonstrates that copper foams not only
cycle at q = 12.5 W/cm2. Two merged cage bubbles instead enhance boiling heat transfer, but also significantly extend
of one are being generated and departing from the foam the operation range of surface superheats. The measured
cover synchronously. CHFs are 56.6, 40.4 and 37.2 W/cm2 on the plain surface
The periodic bubble coalescence pattern does not exist at the three pool liquid temperatures. Correspondingly,
at moderate or high heat fluxes for the high ppi foams. they are 114.3, 107.2 and 98.7 W/cm2 on the 30 ppi foam
 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022 1017

Rainey and You (2000), Liter and Kaviany (2001), Kim

et al. (2002), Honda and Wei (2004). This is mainly due
to that acetone has larger latent heat of evaporation. CHFs
are not previously reported for pool boiling heat transfer
on the metallic foam covers. The range of surface super-
heats in this paper is usually quite larger than that in
references.
Boiling curves of the foam cells can be divided into three
distinct regions, which are marked as region I for liquid
natural convection heat transfer, region II for nucleate
boiling heat transfer, for all the foam cells. Region III is
that of either the resistance to vapor release, for the 30
and 60 ppi foam covers, or the capillary-assist liquid flow
towards foam cells for the 90 ppi foam covers. The different
terms used for the third region for low and high ppi foam
covers are due to the different boiling patterns observed,
i.e., the periodic bubble coalescence and/or re-coalescence
pattern for the 30 and 60 ppi foam covers, and the violent
vapor fragments release pattern for the 90 ppi foam covers.
Slopes of boiling curves are significantly changed from
region I to II, and from region II to III, corresponding to
two transition points. Boiling incipience is the first transi-
tion point, at which the surface superheats are about 4 K
on the three ppi foam covers at the pool liquid temperature
of 38 °C (see Fig. 11a). Temperature excursions are not
observed for all the foam covers in this study, which is
qualitatively consistent with other studies such as Arbelaez
et al. (2000) on the aluminum foams, Moghaddam and
Ohadi (2003) on the copper and graphite foams. For com-
parison, the surface superheat is about 21 K on the plain
smooth surface at boiling incipience. The second transition
point from region II to III occurs at the surface superheats
of 20 and 24 K on the 30 and 60 ppi foam covers, respec-
tively. Heat fluxes at the second transition point strongly
depend on the value of ppi. Higher ppi foams greatly
decrease heat fluxes at the second transition point, indicat-
ing the increased resistance to vapor release for the high ppi
foam covers. It is known from Fig. 11 that higher pool
liquid temperatures slightly decrease slopes of boiling
curves in region II. Generally the three subfigures in Fig.
11 show similar trend at different pool liquid temperatures.
In a general sense, enhancement of pool boiling heat
transfer on foam covers is attributed to the combined effect
of an extended surface area, an increased nucleation site
density, the resistance to vapor release from foam cells,
and a capillary-assist liquid flow towards foam cells. The
liquid supply and vapor release occur as a liquid–vapor
counterflow resisting each others motion. Meléndez and
Reyes (2006) gave a correlation to compute the vapor flow
Fig. 8. Cage bubble observed in the present study ((b) is elapsed with rate escaping from the porous coverings:
3.0 ms following (a)).   !
p qv r ed 3p
m¼ ð5Þ
128 lv d
covers. CHF is not reached at high surface superheats of
180–190 K on the 60 and 90 ppi foam covers. The mea- where qv and lv are the vapor density and viscosity,
sured CHFs are several times larger than those reported respectively. Eq. (5) indicates the influence of the thermo-
in references using FC-72 as the working fluid, such as physical properties (qv, r, lv) and the porous parameters
1018 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022

Fig. 9. Boiling pattern at low heat fluxes (ppi = 90, e = 0.88, Tbulk = 38 °C, d = 3.0 mm).

Fig. 10. Boiling pattern at moderate and higher heat fluxes (ppi = 90, e = 0.88, Tbulk = 38 °C, d = 3.0 mm).

(e, dp, d). A larger vapor mass flow rate represents a smaller In region III, the 30 and 60 ppi foam covers show the
resistance to vapor release. At the same porosity e, low ppi periodic bubble coalescence and re-coalescence pattern
foam covers have large pore size of dp, leading to a large such as given in Figs. 6 and 7, contrast to the continuous
value of m. On the other hand, the capillary pressure vapor fragment release pattern given in Fig. 10 for the
pumping liquid flow towards foam cells is related to 2r/ 90 ppi foam covers. This difference directly leads to differ-
dp. Thus it is seen that small pore size of high ppi foam cov- ent heat transfer performances on low and high ppi foam
ers retards vapor escaping from foam cells, but has large covers. The resistance to vapor release is important, while
capability for pumping liquid to the phase change surface. the liquid pumping effect is weak, on the 30 and 60 ppi
In the nucleate boiling region (region II), the surface foam covers. The 30 ppi foam covers (largest pore diameter
superheats are not large, showing the periodic single bub- of dp) provides less resistance to vapor release, leading to
ble generation and departure pattern (Fig. 5a and b), or better heat transfer performance than the 60 ppi foam cov-
the periodic bubble coalescence pattern (Fig. 5c–h). The ers. As shown in Fig. 11, the 30 ppi foam covers increase
extended surface area of foam cells and increased nucle- heat fluxes by 20–25 W/cm2 than the 60 ppi foam covers
ation sites significantly enhance heat transfer. The gener- at the same surface superheat, in region III.
ated vapor is freely released while the liquid suction The capillary-assist liquid pumping effect is important
towards the phase change surfaces is sufficient. A large for the 90 ppi foam covers in region III. At sufficiently high
increase in heat flux only yields a small increase in surface heat fluxes 100 W/cm2, the 90 ppi foam covers with
superheat, maintaining good thermal performance. Fig. 11 smallest pose size has strong capability to suck liquid
shows that low ppi foams have slightly better heat transfer towards the phase change surfaces, preventing fully dry-
performance in region II. out in foam cells and thus delaying appearance of CHF.
 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022 1019

140 140
(a) Tbulk=38 oC
120 I II III
114.3W/cm2
120 (a) Tbulk=48oC
100
100
q (W/cm )
2

80
80

q (W/cm2)
60
2
56.6W/cm plain surface 60
40
ppi=30, mm
20 ppi=60, mm 40
ppi=90, mm
plain surface
0 ppi=30, mm
boiling incipience 20
ppi=60, mm
-50 0 50 100 150 200 0 ppi=90, mm
sat (K)
-50 0 50 100 150 200
140 sat (K)
o
(b) Tbulk=48 C
120 107.2W/cm2
140
100
q (W/cm )
2

80 120 (b) Tbulk=55oC

60 100
plain surface
40 ppi=30, mm
2 80

q (W/cm2)
40.4W/cm ppi=60, mm
20
ppi=90, mm
60
0
40 plain surface
-50 0 50 100 150 200
ppi=30, mm
sat (K)
20
ppi=60, mm
ppi=90, mm
0
140
(c) Tbulk=55 oC
120 -50 0 50 100 150 200
98.7W/cm2
100 sat (K)
q (W/cm )
2

80
Fig. 12. Boiling curves for the foam porosity of 0.95.
60
plain surface
40
ppi=30, mm
37.2W/cm2
20 ppi=90, =0.88, mm
ppi=60, mm
identified. Compared with Fig. 11 at e = 0.88, the range
0
of surface superheats in natural convection heat transfer
-50 0 50 100 150 200
region becomes narrow for all the foam covers. But the
entire range of surface superheats for the 30 ppi foam cov-
sat (K)
ers is extended. Boiling curves are crossed not only between
Fig. 11. Boiling curves for three ppi foams at three pool liquid the 90 and 60 ppi foam covers, but also between the 90 and
temperatures. 30 ppi foam covers.
Figs. 13 and 14 shows the effects of foam ppi, thickness
The largest heat flux is always reached on the 90 ppi foam of foam covers, and pool liquid temperature on heat trans-
covers in this study. Meanwhile, the strong liquid pumping fer coefficients. Most curves show the increased, maximum
effect for the 90 ppi foam covers leads to steeper slope of and then decreased heat transfer coefficient with continu-
boiling curves than the 30 and 60 ppi foam covers, causing ous increases in heat fluxes, i.e., the quasi-parabola distri-
the crossover of boiling curves between the 90 and 60 ppi bution. The maximum heat transfer coefficient exactly
foam covers (see Fig. 11). The crossover of boiling curves takes place at the transition point from region II to III
between the 90 and 30 ppi foam covers are not apparent (see Fig. 11).
because the 30 ppi foam covers have smaller range of sur- Fig. 13a shows the effect of ppi on heat transfer perfor-
face superheats. In region III, the difference of slopes of mance. Heat fluxes at the maximum heat transfer coeffi-
boiling curves between the 30 and 60 ppi foam covers is cients are significantly decreased with increasing ppi. For
not large. instance, the heat flux is 78 W/cm2 for the 30 ppi foam
cover, but it is 24 W/cm2 for the 90 ppi foam cover,
instead. Low ppi value delays the transition from region
4.3. Parameter effects on boiling heat transfer of foam cells II to III. Heat transfer shows nucleate boiling mechanism
before the maximum heat transfer coefficient, for all the
Larger porosity such as e = 0.95 provides less space ppi foam covers. Heat transfer coefficients are decreased
occupied by the solid material and decreases the surface beyond the maximum point for the 30 and 60 ppi foams
area of ligaments and junctions. Fig. 12 shows boiling due to the increased resistance to vapor release. They are
curves at e = 0.95. Again, three heat transfer regions are almost independent of heat fluxes beyond the maximum
1020 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022

20000
from both the four sides and the top surface of the foam
(a) Tbulk=48oC cover. At moderate or high heat fluxes in region III, the
16000
generated vapor along the four internal margins of the
h (W/m K)

12000 foam cover prevents the success liquid suction from there.
2

Thus liquid suction from the top surface of the foam cover
8000
is necessary to maintain the heat transfer process.
4000 ppi=30, ε=0.88, δ=3.0mm Fig. 14 shows the effect of pool liquid temperatures on
ppi=60, ε=0.88, δ=3.0mm
0 ppi=90, ε=0.88, δ=3.0mm heat transfer coefficients. Higher pool liquid temperature
plain surface enhances nucleate boiling heat transfer in region II, before
0 20 40 60 80 100 120 140 the maximum heat transfer coefficient. This is due to that
q (W/cm2) higher pool liquid temperature activates more nucleation
sites. However, in region III, pool liquid temperatures have
20000
ppi=60, ε=0.88, δ=3.0mm small effect on the heat transfer coefficients. In such a
(b) Tbulk=48oC ppi=60, ε=0.88, δ=4.0mm
16000 ppi=60, ε=0.88, δ=5.0mm
region, heat transfer is mainly governed by the resistance
plain surface to vapor release. The pool liquid temperature only has
h (W/m K)

12000 slight influence on the thermo-physical properties of the

2

liquid phase.
8000 Correlations for the heat transfer coefficients are difficult
at this stage, due to the different heat transfer mechanisms
4000
in the three regions. Besides, the heat transfer mechanisms
are different between the 90 ppi and low ppi foam covers in
0
0 20 40 60 80 100 120 region III. Theoretical/numerical works are recommended
q (W/cm )
2 to study the complex pool boiling heat transfer in metal
foams.
Fig. 13. Effect of ppi and foam cover thickness on heat transfer
coefficients.
4.4. Comparison with other studies

20000 We performed high speed visualizations of pool boiling

on the foam covers. The periodic single bubble generation
16000 and departure pattern at low surface superheats, and the
periodic bubble coalescence and/or re-coalescence pattern
12000 at moderate or high surface superheats were observed on
the 30 and 60 ppi foam covers. The 90 ppi foam cover
h (W/m2K)

8000
makes the bubble coalescence at sufficiently low surface
superheats, while it behaves the continuous vapor fragment
ppi=60, mm,T bulk =38oC release pattern at moderate and high surface superheats.
4000 ppi=60, mm,T bulk =48oC
For the above observations, cage bubble is a new phenom-
ppi=60, mm,T bulk =55oC
plain surface, T bulk =38oC
enon, as described in Section 4.1.2. Cage bubbles are
0
plain surface, T bulk =48oC formed due to the distinct connected web structure of foam
plain surface, T bulk =55oC cells.
0 20 40 60 80 100 120 We found that copper foams with open cells could
q (W/cm2) decrease the surface superheat and eliminate the tempera-
ture excursion at boiling incipience, which is consistent
Fig. 14. Effect of pool liquid temperatures on heat transfer coefficients.
with that drawn by other studies such as Wei and Honda
(2003) for micro-pin-fin structure, Parker and El-Genk
point for the 90 ppi foam covers in region III, which is (2005) for graphite foams. Alternatively, Rainey and You
thoroughly different from those for the 30 and 60 ppi foam (2000) noted that pool boiling heat transfer from micro-
covers (Fig. 13a). porous, square pin-finned surfaces decreases the surface
Fig. 13b shows the effect of foam cover thickness on heat superheat and temperature excursion at boiling incipience.
transfer coefficients. In region II (nucleate boiling region), But the temperature excursion is not fully eliminated for
an increase in foam cover thickness increases the bubble some run cases.
nucleation site density, enhancing heat transfer, under In this paper, we used acetone as the working fluid, hav-
which the resistance to vapor release is not large and the ing larger latent heat of evaporation than FC-72. Thus
liquid supply towards foam cells is sufficient. This is not both the heat transfer coefficients and critical heat fluxes
true in region III. Large foam cover thickness increases are much higher than those reported in references such as
the resistance to vapor release, lowering the heat transfer Rainey and You (2000), Liter and Kaviany (2001), Kim
coefficients. It is noted that liquid suction can take place et al. (2002), Honda and Wei (2004). The degree that the
 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022 1021

present data deviate from other studies depends on the the ppi foams. Region III is that of either the resistance
heating surface area, the parameters of enhanced micro- to vapor release for the 30 and 60 ppi foam covers, or
structure used. the capillary-assist liquid flow towards foam cells for
We identified three heat transfer regions of pool boiling the 90 ppi foam covers.
on copper foam covers. Region I and II refer to the liquid 5. Boiling curves have different slopes in the three regions.
natural convection heat transfer and nucleate boiling heat Crossover of boiling curves is found between the 90 ppi
transfer for all the foam covers. Region III is that of either and low ppi foam covers.
the resistance to vapor release for the 30 and 60 ppi foam 6. The value of ppi has an important effect on heat transfer
covers, or the capillary-assist liquid flow towards foam cells performance. Larger value of ppi with small pore size
for the 90 ppi foam covers. Slopes of boiling curves are dif- has larger resistance to vapor release, lowering the heat
ferent in the three regions. Boiling curves are crossed transfer performance at small or moderate surface
between the 90 ppi and low ppi foam covers. superheats. However, it has larger capability for liquid
Parker and El-Genk (2005) studied pool boiling heat suction towards foam cells at high surface superheats,
transfer of FC-72 on the porous graphite. Boiling curves leading to the improved heat transfer performance and
are divided into the natural convection region and nucleate extended range of surface superheats. This is the major
boiling region. Three sub-regions are divided for the nucle- reason for the crossover of boiling curves between high
ate boiling. Region I is that of low-superheat nucleate boil- and low ppi foam covers.
ing. In region II, the slope of boiling curves is higher than 7. Heat transfer coefficients behave the quasi-parabola dis-
that in region I. Bubble coalescence in region III reduces tribution, with the increased, maximum and then
the slope of boiling curves with increased surface super- decreased values with continuous increases in heat
heats. Thus the three heat transfer regions defined in this fluxes, except for the 90 ppi foam covers. The 90 ppi
paper are different from those defined by Parker and El- foam covers almost do not change heat transfer coeffi-
Genk (2005). Arbelaez et al. (2000) investigated pool boil- cients after they attain the maximum value.
ing heat transfer of saturated FC-72 in highly porous alu- 8. Larger foam cover thickness enhances nucleate boiling
minum foams. The experiments covered porosities and heat transfer due to the increased bubble nucleation site
ppi in the range of 90–98% and 5–40 ppi. The value of density in region II, but decreases heat transfer perfor-
ppi that they used is significantly lower than that tested mance due to the increased resistance to vapor release
in the present paper. They found the increased heat transfer in region III.
coefficients and critical heat fluxes compared with the plain 9. Higher pool liquid temperature enhances nucleate boil-
surface. It is also noted that there is a transition from ing heat transfer in region II. But pool liquid tempera-
nucleate boiling to film boiling, accompanying a tempera- tures have small effect on the heat transfer
ture jump at the transition point. This behavior is not performance in region III.
observed in the present paper. However, the crossover of
boiling curves identified in the present paper is not reported
previously. Acknowledgements

5. Conclusions This paper is supported by the National Basic Research

Program (2006CB601203) and the National Natural Sci-
The conclusions drawn in this paper are summarized as ence Foundation of China (50776089).
follows:

1. The 30 and 60 ppi foam covers display the periodic sin- References
gle bubble generation and departure pattern at low sur-
Arbelaez, F., Sett, S., Mahajan, R.L., 2000. An experimental study on
face superheats. They behave the periodic bubble
pool boiling of saturated FC-72 in highly porous aluminum metal
coalescence and/or re-coalescence pattern with continu- foams. In: 34th National Heat Transfer Conference, Pittsburgh,
ous increases in surface superheats. Cage bubble is a Pennsylvania, August 20–22.
new phenomenon identified in the present paper. Athreya, B.P., Mahajan, R.L., Sett S., 2002. Pool boiling of FC-72 over
2. The 90 ppi foam covers induce easy coalescence of min- metal foams: effect of foam orientation and geometry. In: 8th AIAA/
ASME Joint Thermophysics and Heat Transfer Conference, St. Louis,
iature bubbles at small surface superheats, with the peri-
Missouri, June 24–26.
odic bubble coalescence pattern demonstrated. At Calmidi, V.V., 1998. Transport Phenomenon in High Porosity Metal
moderate or large surface superheats, vapor fragments Foams, Ph.D. Thesis. University of Colorado, CO, USA.
are found to depart from the foam cover continuously. Chen, Y., Groll, M., Mertz, R., Kulenovic, R., 2005. Visualization and
3. Copper foams could decrease the surface superheat and mechanisms of pool boiling of propane, isobutane and their mixtures
on enhanced tubes with reentrant channels. Int. J. Heat Mass Transfer
eliminate the temperature excursion at boiling incipience.
48, 2516–2528.
4. Three distinct heat transfer regions are identified. Coursey, J.S., Kim, J., Boudreaux, P.J., 2005. Performance of graphite
Region I and II are those of liquid natural convection foam evaporator for use in thermal management. J. Electron. Packag.
heat transfer, and nucleate boiling heat transfer for all 127, 127–134.
1022 J. Xu et al. / International Journal of Multiphase Flow 34 (2008) 1008–1022

Ghiu, C.D., Joshi, Y.K., 2005. Visualization study of pool boiling from Moghaddam S., Ohadi M., 2003. Pool boiling of water and FC-72 on
thin confined enhanced structures. Int. J. Heat Mass Transfer 48, copper and graphite foams. In: Proceedings of ASME InterPACK’3,
4287–4299. International Electronic Packaging Technical Conference and Exhibi-
Honda, H., Wei, J.J., 2004. Enhancement boiling heat transfer from tion, Maui, Hawaii, July 6–11.
electronic components by use of surface microstructures. Exp. Thermal Nimkar, N.M., Bhavnani, S.H., Jaeger, R.C., 2006. Effect of nucleation
Fluid Sci. 28, 159–169. site spacing on the pool boiling characteristics of a structure surface.
Kim, J.H., Rainey, K.N., You, S.M., Pak, J.Y., 2002. Mechanism of Int. J. Heat Mass Transfer 49, 2829–2839.
nucleate boiling heat transfer enhancement from microporous surfaces Parker, J.L., El-Genk, M.S., 2005. Enhanced saturation and subcooled
in saturated FC-72. J. Heat Transfer ASME 124, 500–506. boiling of FC-72 dielectric liquid. Int. J. Heat Mass Transfer 48, 3736–
Kretschmer, C.B., Nowakowska, J., Wiebe, R., 1946. Solubility of oxygen 3752.
and nitrogen in organic solvents from 25 °C to 50 °C. Ind. Eng. Rainey, K.N., You, S.M., 2000. Pool boiling heat transfer from plain and
Chem. 38, 506–509. microporous, square pin-finned surfaces in saturated FC-72. J. Heat
Liter, S.G., Kaviany, M., 2001. Pool boiling CHF enhancement by Transfer ASME 122, 509–516.
modulated porous-layer coating: theory and experiment. Int. J. Heat Wei, J.J., Honda, H., 2003. Effects of fin geometry on boiling heat transfer
Mass Transfer 44, 4287–4311. from silicon chips with micro-pin-fins immersed in FC-72. Int. J. Heat
Meléndez, E., Reyes, R., 2006. The pool boiling heat transfer enhance- Mass Transfer 46, 4059–4070.
ment from experiments with binary mixtures and porous heating Yaws, C.L., 1999. Chemical Properties Handbook. Mcgraw-Hill, New
covers. Exp. Thermal Fluid Sci. 30, 185–192. York.