International Journal of Heat and Mass Transfer 79 (2014) 989–1001
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer
journal homepage: www.elsevier.com/locate/ijhmt
Development of a two-step electrodeposition process for enhancing pool
boiling
Chinmay M. Patil a, K.S.V. Santhanam b, Satish G. Kandlikar a,⇑,1
a
b
Department of Mechanical Engineering, Rochester Institute of Technology, United States
School of Chemistry and Material Science, Rochester Institute of Technology, United States
a r t i c l e
i n f o
Article history:
Received 17 August 2014
Accepted 25 August 2014
Available online 19 September 2014
Keywords:
Pool boiling
Heat transfer enhancement
Porous
Microporous
Multiscale enhancement surfaces
Electrodeposition
a b s t r a c t
The continuous development of high performance chips and the growing miniaturization trend in the
electronics and microelectronics industry require efficient systems to remove large amounts of heat over
a small footprint. Pool boiling has the ability to remove large heat fluxes at small values of wall superheat
and this can be further augmented by using porous or microporous surfaces. In this work, a two-step
electrodeposition technique involving application of high current density for a short time, followed by
a lower current density for a longer time was investigated. This technique allowed a close control of
the pore size and porous layer thickness. The electrodeposition process was carefully studied and parameters for creating different morphologies of enhanced surfaces were obtained. Thickness of the coating
was in range of 50–100 lm. After testing a variety of morphologies for pool boiling heat transfer with distilled water, a maximum heat flux of 1400 kW/m2 and a significant enhancement in heat transfer coefficient (HTC) of 179 kW/m2 °C was obtained from a copper chip with cauliflower-like morphology using an
initial current density of 400 mA/cm2.
Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction
There is an increased demand for dissipating high heat fluxes
from electronic chips as faster and high power devices are being
developed in various fields including industrial equipment, military and space applications, and high-end computing. The conventional air cooling systems are not able to meet these demands due
to their low heat transfer coefficients (HTC), low specific heat and
high specific volume. Improvement in HTC will result in reducing
the size of cooling equipment. Compared to most other cooling
techniques, pool boiling is being investigated by many researchers
due to its ability to carry large amounts of heat at low wall superheat without any moving parts [1]. Pool boiling heat transfer can
be further enhanced with active devices like ultrasonic vibrations,
electrostatic fields etc., or passive techniques such as porous/
microporous surfaces, structured surfaces like finely cut grooves,
finned or knurled surfaces, integrated micro-nanostructures, etc.
Microporous coatings are attractive due to their ability to improve
⇑ Corresponding author at: Mechanical Engineering Department, Rochester
Institute of Technology, 76 Lomb Memorial Drive, Rochester, NY 14623, United
States. Tel.: +1 585 475 6728; fax: +1 585 475 7710.
E-mail addresses: [email protected] (C.M. Patil), [email protected] (K.S.V. Santhanam),
[email protected] (S.G. Kandlikar).
1
ASME Membership 608711, Fellow member.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2014.08.062
0017-9310/Ó 2014 Elsevier Ltd. All rights reserved.
both critical heat flux and HTC. The present work is focused on
development and control of the process for deposition of microporous surfaces for enhancing pool boiling heat transfer using the
electrodeposition technique.
2. Literature review
A recent exhaustive literature review provides details of various
manufacturing techniques to obtain porous enhanced surfaces [2].
The manufacturing techniques discussed in [2] can be broadly classified into five different categories as shown in Fig. 1.
Porous surfaces have been studied quite extensively in literature. Bergles and Chyu [3] created porous surfaces on bronze using
a proprietary technique, achieving 250% enhancement in HTC with
water and 400–800% with refrigerants. Anderson and Mudawar [4]
created a surface by vapor blasting on copper to create a microporous surface, achieving an enhancement of 170% in critical heat
flux (CHF) when tested with FC-87. You et al. [5], Memory et al.
[6], and Golobic and Ferjancic [7] created microporous surfaces
using a spray gun and fine powder of copper or other metals and
their oxides and spraying them on the surfaces. You et al. [5] tested
these surfaces with FC-72 and achieved an enhancement of 280% in
CHF. Memory et al. [6] investigated pool boiling of R-114 on finned,
structured and porous tubes and found that the enhancement with
990
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
Nomenclature
Abbreviations
CHF
critical heat flux
HTC
heat transfer coefficient
SEM
scanning electron microscope
A
area of the electrode
A1
nucleation rate
By
bias error
°C
degrees celsius
C⁄
concentration of copper ion
D
diffusion coefficient
Epa
anodic peak potential
Epc
cathodic peak potential
F
faraday constant
Ipa
anodic peak current
Ipc
cathodic peak current
Id
diffusion controlled current
porous surfaces was significant only at low heat fluxes. Golobic and
Ferjancic [7] created a porous medium of different metals and their
oxides on a ribbon heater and tested with FC-72, and achieved 29–
130% enhancement in CHF. Chang and You [8] applied a mixture of
copper powder, binder and a carrier fluid (Methylethylketone) on a
copper surface and heated it in an oven to evaporate the carrier
liquid. With this porous medium consisting of copper held together
by binder, they achieved 100% enhancement in CHF and 30%
enhancement in HTC when tested with FC-72.
Based on Albertson’s patent [9], Kim [10] devised a technique to
electrodeposit enhanced surfaces on copper substrates in multiple
stages. In the first stage, they applied high current density to evolve
hydrogen bubbles, simultaneously depositing copper and leaving
behind copper foam. Porosity of this foam is reduced by slowly
depositing additional copper at a lower current density. When
tested with water, they obtained 250–700% enhancement in boiling
performance and 50–60% improvement in CHF over a plain surface.
El-Genk and Ali [11] created similar surfaces and tested them with
PF-5060 which enhanced the CHF by 700% without temperature
overshoots. Kim [10] prepared porous nickel surfaces by soldering
the particles together using a plumbing solder, yielding 40–60%
enhancement in CHF. Im et al. [12] created thin films of flowerlike
CuO and tested with PF-5060, attaining 58% enhancement in CHF
for flat surfaces and 30% for microgrooved surfaces.
The literature review indicates that microporous surfaces
enhance the pool boiling heat transfer performance and increase
i
kcu
M
im
n
Py
q00
T
t
tm
Up
Uy
x
value of instantaneous current
thermal conductivity of copper
molar mass
maximum current in chronoamperometry
number of electrons
precision error
heat flux
temperature
time duration of electrolysis
time at maximum current during cyclic voltammetry
uncertainty in a given parameter
uncertainty
distance
Greek symbol
q
density
its CHF. The results also indicate that the surfaces enhance boiling
heat transfer not only for water but also for other fluids, suggesting
that this enhancement technique could be used for electronic cooling applications.
3. Electrodeposition technique development
From literature review, it was evident that microporous surfaces in general yield high heat transfer coefficients and enhance
the critical heat flux. Most of the techniques described in the literature utilize high temperatures and longer durations which are in
excess of 1 h, resulting in safety and energy consumption concerns
for large scale manufacturing. Binder-based techniques have limitations on wall superheat due to chemical stability of the binder.
Also, precise control over thickness and pore quality is necessary
to achieve good performance. The present day need is to develop
a technique that overcomes these limitations and also is amenable
for mass production at a low cost in order to gain commercial
acceptance.
From the manufacturing processes discussed in the literature
review, electrodeposition is a particularly attractive option for
the following reasons:
The process takes relatively less time.
It involves room temperature, reducing hazards of warping or
annealing due to higher temperatures.
Fig. 1. Manufacturing processes to generate porous surfaces for pool boiling heat transfer enhancement [2].
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
Thin porous films in range of 20–100 lm thickness can be readily fabricated.
Pore sizes can be controlled by controlling current density,
potential, time of deposition and magnitude of agitation.
No special care is necessary for the finished surface.
However, the disadvantages of this technique are:
High initial cost of equipment and consumables.
Special training to handle chemicals.
Special care to store and dispose the chemicals.
The goal of the present work is to develop an electrodeposition
technique in which the morphology can be controlled to yield
higher pool boiling performance.
Fig. 2 below illustrates the setup used for electrodeposition of
copper. In this setup, the test section is the cathode and a piece
of copper, referred to as copper chip, is used as the anode. The
anode could however be any other metal that undergoes oxidation.
A direct current power source (Princeton Applied Research VersaSTAT 3 and Gamry Instruments Series G™ 300) is used for all the
experiments. The cathode and anode are immersed vertically in
an electrolytic bath containing metal salt dissolved in distilled
water (in this case, CuSO4 in water). Upon application of direct current, positive ions (Cu2+) are formed at the anode; reducing copper
ion to copper at cathode This way, copper is coated on the test section connected to the cathode.
Fig. 2. Setup for electrodeposition.
991
The electrochemical process that takes place upon application
of direct current consists of two reactions. In a generalized scheme
of metal undergoing dissolution and deposition at the cathode, the
electrons are supplied to the metal cation. The reaction at the cathode is given by
Mnþ þ ne ¼ M
ð1Þ
where M is a metal such as Cu. At the anode, the electrons are liberated by oxidation process to produce the metal cation. This reaction is given by
M ¼ M nþ þ ne
ð2Þ
From the above reactions it can be observed that the metal cation is formed at the anode when it loses its free electrons. Thus, the
cation migrates from anode to cathode where it gains the electrons. This completes the electrical circuit. As the cation receives
electrons at the cathode, it is deposited there. By Faraday’s Laws
of electrolysis, the amount of metal deposited is proportional to
the current and the duration [13,14]. This process can be done by
either passing direct current through the electrolytic solution (galvanostatic) or by holding the substrate electrode (anode) at a
potential (potentiostatic). In the present work, galvanostatic
method is used to create microporous enhanced surfaces.
The electrolytic bath used in the current work had 0.8 M copper
sulfate and 1.5 M sulfuric acid. To prepare this solution, copper sulfate (CuSO4) and sulfuric acid (H2SO4) manufactured by EMD Millipore Chemicals and Avantor™ Performance Materials, Inc.
respectively were used.
The test sections were initially created using a method suggested by Shin and Liu [15]. This process involves wet electrochemical deposition accompanying evolution of hydrogen
bubbles, which were deliberately suppressed in conventional processes to produce dense metallic components. These hydrogen
bubbles function as dynamic templates during deposition. The surface pore size of the deposit increased with the time of deposition
due to coalescence of these hydrogen bubbles. It can be visualized
from the process steps shown in Fig. 3. A large number of test chips
were fabricated with microporous surfaces using different electroplating parameters based on [15]. These test chips however did not
bond properly with the substrate, and came off during pool boiling
tests yielding unstable results. Thus, bonding the porous matrix to
the copper substrate was an important issue to tackle.
Albertson [9] found that nodules are formed by plating at high
current densities that may be further plated at lower current densities to strengthen and enlarge them. He invented a technique in
which a heat transfer enhancing surface can be obtained by initially coating at higher current densities for a short duration of
time, followed by lower current densities for a longer duration.
In the first step, there was deposition of copper along with simultaneous evolution of hydrogen bubbles, leaving behind porous copper as suggested by Shin and Liu [15] and seen in Fig. 3. In the
second step, the current density is chosen such that there is deposition of copper without evolution of hydrogen. This improved the
bonding of the porous copper to the substrate. This approach was
Fig. 3. Deposition of porous copper using hydrogen bubbles as templates.
992
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
chosen to prepare the enhanced surfaces. The electrodeposition
process is studied in depth further in Section 3.1.
3.1. Electrodeposition process
The performance of the electrodeposited copper in pool boiling
application has been investigated earlier [9–11] with the conclusion that morphology of copper is an important factor in such applications. The morphology of electrodeposited copper is controlled by
varying the electrochemical parameters such as potential, current
density and magnitude of agitation in the solution. In the present
investigation, a new atomic layer of copper is deposited followed
by concurrent evolution of hydrogen gas on the surface.
The electrochemical deposition of copper was carried out using
a plain copper chip (area 1 cm2) as the working electrode along
with another copper chip as the counter electrode in a solution
containing 0.8 M CuSO4 and 1.5 M H2SO4 and using a current step
program with reversal of current as shown in Fig. 4. A saturated
calomel electrode (SCE) was used as the reference electrode and
all the voltages were recorded with respect to the SCE. The electrochemical cell is subjected to a high current density for a period of
15 s followed by a lower current density for durations ranging from
2500 to 4500 s. The magnitudes of the upper current densities
were in the range of 300–650 mA/cm2. The morphologies of the
deposits have been examined to the suitability of pool boiling
application. In order to understand the electrochemical processes
that occur during the current step electrolysis, chronopotentiometric, chronoamperometric and cyclic voltammetric analysis have
been carried out.
3.1.1. Chronopotentiometric analysis of copper deposition
Chronopotentiometry is the process in which the rate of change
of potential at an electrode is measured by controlling the current
[14]. The chronopotentiometric recordings were carried out at the
current densities shown in Table 1 to obtain information on working electrode processes. When a current density of 0.30 A/cm2 is
applied to the working electrode, the potential rapidly shifted
from 0 to about 0.68 V and the transition time for the cathodic
process is estimated at 0.24 s. Thereafter, the working electrode
potential reached a value of 1.10 V. Upon stepping the current
to 0.04 A/cm2, the potential reached 0 V. The product of current
density and square root of the transition time (Table 1) is nearly
constant suggesting that the copper deposition is diffusion
controlled.
The diffusion controlled nature of the deposition has been
examined through cyclic voltammetry and chronoamperometry. A
well-defined cathodic peak voltage (Epc) appeared at Epc = 0.24 V.
Upon reversal of the sweep past Epc, a well-defined sharp
anodic peak voltage (Epa) appeared at 0.06 V. If the Epc is extended
to 0.84 V, background electrolysis sets in with evolution of
hydrogen. These peak potentials are identical to the ones reported
in literature for copper ion reduction and oxidation of deposited
copper metal [14].
Cuþþ þ 2e ¼ Cu
Cathodic process ½Epc ¼ 0:24 V
2Hþ þ 2e ¼ H2
Cathodic process ½E > 0:84 V
ð4Þ
Cu ¼ Cu2þ þ 2e
Anodic process ½E > 0:06 V
ð5Þ
When the chronopotentiometric recording is carried out, the
potential of the working electrode initially stayed for about 0.26–
1.44 s at the copper deposition region. Thereafter the potential
shifted to 1.1 V, which corresponds to the evolution of hydrogen
along with copper deposition for a period of 8.56–9.74 s, depending on the current density used in the experiment. The initial
amount of copper deposited is shown in Table 2.
As observed from experimentation, the amount of copper
deposited decreases with higher current densities due to increased
evolution of hydrogen bubbles. At the highest current density of
300 mA/cm2, the number of atoms of copper deposited is
2.5 1017 atoms/cm2. Taking the area occupied by a copper atom
based on the radius as 1.3 108 cm, it would amount to about
1.9 1015 atoms of copper in one square centimeter area. This
value suggests that the atomic layer coverage occurs in the first
step of the deposition process. The subsequent growth of copper
is controlled by the hydrodynamics that sets up at a more negative
potential.
3.1.2. Chronoamperometric analysis of copper deposition
Chronoamperometry is an electrochemical technique in which
the potential of the working electrode is stepped to the copper
deposition region and the resulting current from the process occurring at the electrode is monitored as a function of time. Since the
current is integrated over relatively longer time intervals, chronoamperometry gives a better signal to noise ratio in comparison to
other amperometric techniques [14]. The potential step electrolysis of copper deposition has been examined by stepping the potential of the copper electrode from 0 V to 1.1 V. Step size was
selected automatically by the equipment. The current decay with
time is shown in Fig. 5.
For a diffusion controlled process, the decay of current is analyzed by Fick’s law of diffusion and the resulting current–time relationship as derived by Cottrell in reference [14] is given by the
following equation.
Id ¼
nFAD1=2 C ð6Þ
ðptÞ1=2
Fig. 4. Preparation of the electrodeposited copper chips.
Table 2
Amount of copper deposited in the transition region.
Table 1
Transition times for copper deposition.
2
1/2
Current density (io) (A/cm )
Transition time (s) (s)
io s
0.15
0.20
0.30
1.44
0.99
0.26
0.18
0.19
0.15
(s
1/2
2
) (A/cm )
ð3Þ
Current
density
mA/cm2
Total
coulombs
C/cm2
Transitional copper
deposition at-0.50 V
(lg/cm2)
Coulombs consumed for
mixed copper and
hydrogen gas (C/cm2)
150
200
300
0.22
0.20
0.08
72.4
65.8
26.3
1.28
1.80
2.92
993
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
Fig. 5. Decay of current with time.
where n is the number of electrons, F is Faraday constant, A is the
area of the electrode, D is the diffusion coefficient, C⁄ is the concentration of copper ions and t is the duration of electrolysis. This equation predicts that a plot of id vs t1/2 is constant with a slope given
by the bracket terms divided by P. Analysis of the current transient
shows linearity for the first 2.5 s with a slope of 57.745. This slope
yields a value of diffusion coefficient of 0.43 106 cm2/s. The
value obtained is slightly lower than reported in the literature
[16–18] due to differences in the experimental conditions. It has
been shown that the copper ion diffusion coefficient is dependent
on sulfuric acid concentration, and it decreases with increasing concentration [19]. The values give a good estimation of the process.
The current–time transient analysis shows one type of linearity
up to 2.5 s and another type of linearity from 2.5 s to 12 s. While
the first linearity is smooth as seen in Fig. 6(a), the second one
had divergence due to convection caused by hydrogen gas evolution at the electrode. These are shown Fig. 6(a) and (b).
3.1.3. Cyclic voltammetric analysis
Cyclic voltammetry is a technique in which the working electrode potential is ramped linearly against time. Once the experiment reaches its set potential, the potential ramp is inverted.
This process is used to study electrochemical properties of constituent in an electrolyte solution [14]. Fig. 7 shows the cyclic voltammetric peak for copper ion reduction and its complementary
oxidation peaks. It also shows the potential at which the background electrolysis (hydrogen ion reduction) occurs. Table 3 shows
Fig. 7. Cyclic voltammetric peak for the copper ion reduction and its complementary oxidation peaks.
Table 3
Cyclic voltammetric data for copper ion reduction.
Sweep rate (mV/s)
Epc (V)
Ipc (lA)
Epa (V)
Ipa (lA)
Ipc/v1/2
10
20
50
100
0.258
0.267
0.286
0.309
9.40
12.68
17.49
22.30
0.019
0.078
0.067
0.058
46.94
61.32
74.05
76.87
2.97
2.83
2.47
2.23
Working electrode: Pt disc electrode, Concentration of CuSO4: 1 mM.
the cyclic voltammetric peak currents and potentials at different
sweep rates. The near constancy of the current function shows that
the copper reduction is diffusion controlled. However, the analysis
of currents in the potential region beyond 0.80 V becomes difficult as the process becomes convection controlled.
All the electrochemical results indicate that there is an atomic
layer deposited initially over which further growth occurs [20].
This later growth is convection controlled growth which produces
the porous structure.
3.2. Modeling of nucleation
The morphology suitable for pool boiling application was
arrived at by using the galvanostatic condition discussed earlier
and followed by testing under pool boiling. Based on this data,
we considered the possibility of adopting instantaneous 2D
Fig. 6. (a) Linearity of current against square root of time, (b) divergence in the later stage due to convection caused by hydrogen gas evolution at electrode.
994
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
nucleation of copper or progressive nucleation of copper. The
instantaneous nucleation as an exponential growth is given by
[21] the following equation.
i
im
¼
t
1 ðt 2 t2m Þ
exp tm
2
t 2m
!
ð7Þ
where i is the current at any time t and im is the maximum current
at tm. This equation is modified from the progressive nucleation as
i
im
¼
!
2
t
2 ðt 3 t 3m Þ
exp tm
3
t 3m
ð8Þ
The experimental current–time transient shows the plots of i/im
vs t/tm and i/im vs (t/tm)2 in Fig. 8(a) and (b).
The results suggest that there is progressive nucleation occurring in copper deposition at a later stage of deposition [22]. As
there is no increase followed by the decrease observed in i/im in
the above curves in Fig. 8(a) and (b) as predicted by the models,
the analysis will have to be confined to the decreasing part of
the expected behavior proposed in [23]. Scharifker and Hills [19]
proposed a model for 3D nucleation and diffusive growth under
diffusion controlled deposition which follows the current response
as given by the following equation.
i ¼ bt
1=2
½1 expðct 2 Þ
ð9Þ
where,
b¼
c¼
5. Pool boiling test setup
nFD1=2 C 5.1. Test setup description
p1=2
An experimental setup was designed to study pool boiling heat
transfer of water over the enhanced surfaces at atmospheric
pressure. The schematic of this setup can be seen in Fig. 10. The
setup had two large windows, enabling visual access to study
A1 pk1 D
2
k1 ¼
dimensions of 10 10 2 mm deep machined on one of its sides.
This side for simplicity purposes is referred to as the heater side.
This groove facilitated 1-dimensional flow of heat. The other side
of this chip was polished with 1500 grit sand paper, over which
microporous coatings were deposited. This side of the chip for simplicity purposes is referred to as the boiling side. The chip was initially cleaned with distilled water and isopropyl alcohol. The
central 10 10 mm area on the boiling side of the chip was
marked. This area was exposed to the electrolytic bath, while
remainder area was insulated electrically using an electrical insulation tape.
Electrodeposition parameters used in preparing the test chips
investigated for pool boiling heat transfer are shown in Table 4.
An X-ray diffraction (XRD) experiment was conducted on chip
6P to detect presence of any impurities in the deposit. XRD is a
technique which records a scattering pattern produced when a
beam of X-ray interacts with a material. Every material has a
unique scattering pattern. The recorded pattern was compared to
an exhaustive database from International Center Diffraction Data
(ICDD). Fig. 9 shows scattering pattern of chip 6P. Vertical lines in
Fig. 9 indicate the peaks available in the database of ICDD, whereas
red line is the observed diffraction pattern. Superimposing the
available data on the obtained XRD, it was observed that the
deposit was pure copper, with no impurities.
1=2
4 8pC M
3
q
A1 is the nucleation rate, n is number of electrons, C⁄ is the concentration of copper ions, M is molar mass and q is the density. If
we consider a situation when the exponential term tends to 1, Eq.
(9) simplifies to Eq. (6). Hence Fig. 6(a) suggests that 3D model is a
good fit for the nucleation mechanism involving diffusional growth
of copper atoms.
4. Test matrix
All the test surfaces were prepared on 20 20 3 mm
thick copper test chips. It had a 2 mm wide groove with inner
Table 4
Test matrix of electrodeposition parameters and resulting coating thickness.
Chip number
Initial current
density
(mA/cm2)
Time
(sec)
Final current
density
(mA/cm2)
Time
(sec)
Thickness
(lm)
Chip
Chip
Chip
Chip
Chip
Chip
Chip
650
650
650
600
500
400
300
15
15
15
15
15
15
15
40
40
40
40
40
40
40
4500
3600
2500
2500
2500
2500
2500
83.3
76.9
66.2
53.1
59.9
61.7
69.7
1P
2P
3P
4P
5P
6P
7P
Fig. 8. Experimental current–time transients shows the plots of (a) i/im vs (t/tm)2, and (b) i/im vs t/tm.
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
995
Fig. 9. XRD of Chip 6P showing presence of pure copper without any oxides. Peaks of copper are shown using vertical lines and the values are acquired from the database.
Fig. 10. Pool boiling test setup.
the bubble dynamics. Overall dimensions of the setup were
215 150 76.2 mm. The setup primarily consisted of an aluminum block machined as the holding block for the pool boiling tests.
Observation windows from both the sides were held together
using 6 M8 bolts, compression plates and gaskets to making the
area leak proof. The entire assembly was held on the copper heater
by four screws fixed to the aluminum base. The copper heater
block comes in contact with the test chip from underneath on
the heater side of the chip. This heater block sits on an insulating
ceramic base, which is supported by the aluminum block. The
excess area of the chip beyond the central 100 mm2 electrodeposited section on the boiling side is sealed tightly using a rubber
silicone gasket and a stainless steel plate, exposing only the
enhanced surface to boiling water. Two 125 W auxiliary heaters
were provided from sides of the base block. The purpose of these
auxiliary heaters was to maintain water inside the test chamber
at saturation temperature.
The copper heater block was insulated using fiberglass pipe
insulator and held vertically on a ceramic block. Arctic SilverÒ 5
thermal paste is used to minimize contact resistance between heater top and the test section. Also heat loss analysis was conducted,
with results indicating that the conduction can be considered one
dimensional in the copper block underneath the chip.
The test setup had specific spots for five K-type thermocouples.
Three thermocouples were used to measure T1, T2 and T3 for calculating the input heat flux on the copper heater as shown in
Fig. 11. The holes for these thermocouples are 300 lm in diameter
and 5 mm deep. They are spaced 8 mm from each other. Thermocouple T4 measures the temperature at the center of the chip.
The hole is 300 lm in diameter. This reading helps in calculating
the surface temperature of the boiling surface. Thermocouple T5
is inserted from top of the setup to measure the bulk temperature
of water.
The heat flux is calculated using Fourier’s Law.
Q 00 ¼ kCu
dT
dx
ð10Þ
The temperature gradient dT=dx is calculated using Taylor’s backward series approximation.
dT 3T 3 4T 2 þ T 1
¼
dx
2Dx
ð11Þ
where T1–T3 are temperatures in the heater rod as shown in Fig. 11,
and Dx is the distance between two successive thermocouples. A
thermocouple was inserted inside a chip, as described in Section 4,
and the reading was used to calculate surface temperature of the
boiling surface. This is given by following equation.
T wall ¼ T c q00
x1
kCu
ð12Þ
996
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
instrument (VI), with a display as shown below. Upon achieving
steady state, data was logged for 10–15 s at a rate of 5 data points
per second. Before logging the readings water was allowed to boil
at steady state for about 300 s. The voltage across the cartridge
heater inserted inside the copper heater block was increased in
steps of 5–8 V and above procedure was followed. The increment
in voltage adjusted input power to this cartridge heater. This procedure was followed until heat fluxes of 1 MW/m2. Beyond this
value of heat flux, the voltage was increased in smaller increments
of 2–3 V. Necessary precautions were taken to check for CHF condition by closely and continuously monitoring the boiling surface,
the bubble dynamics and temperature readings on the LabVIEWÒ
VI. Upon reaching CHF, power supply to the heater was cut off
and subcooled water was injected on the boiling surface to break
the vapor film and restore boiling and maintain the temperature
of the test section.
5.3. Uncertainty analysis
Fig. 11. Cross section of the heater assembly, cartridge heaters in the base not
shown.
where x1 is the distance between the thermocouple and the boiling
surface of the chip. A national instruments cDAQ-9172 data acquisition system was used to measure the temperature and a LabVIEWÒ virtual instrument (VI) was programmed for recording
data and calculating surface temperature and heat fluxes. Fig. 12
shows a schematic of the entire setup and the data acquisition
system.
5.2. Experimental procedure
A DC power source was used to power the main heater and the
two auxiliary heaters. The voltage of the power source was
adjusted to set the heat input to the copper test section. Distilled,
degassed water at atmospheric pressure was used as the test
liquid. The temperatures were monitored by a LabVIEWÒ virtual
A careful uncertainty analysis was performed, similar to Cooke
and Kandlikar [24] and Kim et al. [25]. Errors arising from calibration or resolution are called bias errors and errors due to fluctuations or sensitivity are called precision errors. The cumulative
error can be expressed by
Uy ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B2y þ P2y
ð13Þ
where Uy is the uncertainty or error, By is the bias error and Py is
the precision error. Each thermocouple was calibrated, and its precision error was determined statistically to be ±0.1 °C, considering
calibration accuracy and precision. There were uncertainties in
thermal conductivity of copper due to variation of temperatures
during testing. Also, spacing between the thermocouples on the
copper heater and thickness of the test section carry some uncertainties, based on the resolution of the measurement equipment.
These uncertainties are illustrated in Table 5.
The uncertainties in measurement of temperature and distance
between thermocouples propagate an error in evaluation of heat
flux. Also, errors propagate in calculation of surface temperature
due to errors in the measurement of thickness of the test section.
These errors are calculated by
Fig. 12. Schematic of the experimental setup and data acquisition system.
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
997
Table 5
Uncertainties in various parameters.
Sr. No.
Parameter
Value
Unit
Up
Value
% Uncertainty
1
2
3
4
5
6
kCu
Lcu
Dx
T1
T2
T3
391
0.003
0.008
Varies
Varies
Varies
W/m K
m
m
°C
°C
°C
Uk,Cu
UL,Cu
Ux,Cu
UT1
UT2
UT3
9
0.00025
0.0001
0.11
0.15
0.11
2.24
8.33
1.25
N/A
N/A
N/A
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
Xn @p
Up ¼
u
ai
i¼1 @a
ð14Þ
where Up is the uncertainty in the parameter p, and uai is the
uncertainty of measured parameter ai . The uncertainty in the heat
flux can thus be expressed by the following equation.
Fig. 14. Heat loss study showing variation of temperature over the distance.
v"
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 2 2 2 2 2 #
3U T 1 kCu
4U T 2 kCu
U T 3 kCu
U q00 u
Uk
U Dx
t
¼
þ
þ
þ
þ
q00
k
Dx q00
Dx q00
Dx q00
Dx
ð15Þ
Using Eq. (15) and values from Table 5, uncertainties for different
values of heat fluxes are illustrated in Fig. 13.
From the graph, it can be observed that the uncertainty at the
low heat fluxes was estimated to be higher as compared to higher
heat fluxes. Thus, the range of error was estimated to be less than
5 percent in the range of interest above a heat flux of about
300 kW/m2.
To ensure that the heat generated from the cartridge heater was
transferred to the test section through 1D conduction, a heat loss
study was conducted. According to Fourier’s law of conduction,
the temperature profile over the length of the heater in the region
where temperatures are measured should be linear. This ensures
that there are no heat losses, and that the heat transfer is one
dimensional.
Fig. 14 shows a graph of temperature plotted against distance at
heat fluxes 270.6, 619.2, 1,152 and 1,714 kW/m2. It was seen that
the profile of the line shows linear progression as R2 was very close
to 1 at all heat fluxes. This indicated that the temperature profile is
linear, with minimal losses from the sides of the copper rod.
Fig. 13. Plot showing variation of uncertainty with increasing heat flux.
6. Results and discussion
Initially the performance of three chips, Chip 1P, 2P and 3P was
investigated. The test chips were prepared to study the effect of
duration of the second stage electroplating on pool boiling heat
transfer. The electrochemical parameters employed for these chips
are listed in Table 1. A VCA Optima contact angle measurement
equipment was used to measure the contact angle of the boiling
surfaces developed on the test chips. The surfaces were seen to
be hydrophilic in nature with their contact angles 18°, 29° and
12° for Chips 1P, 2P and 3P, respectively as shown in Fig. 15.
The surfaces were tested for their pool boiling heat transfer performance. The pool boiling curves are plotted in Fig. 16. It can be
observed from the pool boiling curves that all the three test chips
performed identically. Critical heat fluxes for all chips were
approximately 1,300 kW/m2 at wall superheats of close to 18 °C.
This showed that reducing the time in the second step had no significant effect on the pool boiling performance. Hence, the lowest
time step of 2500 s was chosen as the time step for the second
stage in electrodeposition in order to reduce the overall process
time.
Contact angles for test chips 3P–7P were measured and are
shown in Fig. 17. They are in the range of 7–14°. Without the
microporous surface coating, the contact angle of water on a plain
copper chip was 69°. This showed that microporous coatings made
the surface hydrophilic in nature.
Laser confocal images of these structures are shown in Fig. 18.
Though all images are not presented in the article, multiple images
of the same microporous surfaces at different locations were captured to ensure uniformity of the deposit. The dimensions of the
microstructures were measured using ImageJ software and the
range is presented as quantitative measurement at multiple locations. Cauliflower-like microstructures are clearly seen in these
images. The diameter of the cauliflower-like structure for Chip 7P
is seen to be approximately 30–50 lm and 20–30 lm for Chip 6P
as observed in Fig. 18(a) and (b). For increased current density
for Chips 4P and 3P, the electrodeposited surfaces showed a reduction in the density of the cauliflower structures. This reduction in
density is observed in Fig. 18(c). Open dish-like structures of diameters in range of 30–90 lm are observed as seen in Fig. 18(d). As
observed earlier, there is simultaneous evolution of hydrogen bubbles along with electrodeposition of copper. At lower current density, there is comparatively lower frequency of evolution of
hydrogen bubbles. As these bubbles escaped the surface, they coalesce with neighboring ones forming a larger bubble and prevent
copper deposition over those surfaces. Due to this late coalescence
998
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
Fig. 15. Contact angles of (a) Chip 1P, (b) Chip 2P and (c) Chip 3P.
Fig. 16. Pool boiling curves of Chips 1P, 2P and 3P with saturated distilled water,
and their comparison with plain surface.
after departure the microstructures looked like cauliflower-like
structure. At higher current density however, there is higher frequency of evolution and coalescence of hydrogen bubbles, resulting in an open dish-like morphology.
To understand the morphology further, SEM images of chip 3P
and 6P were taken as seen in Fig. 19(a) and (b). The SEM in
Fig. 19(a) showed that there were open dish-like structures similar
to what was seen in the laser confocal microscope images. On the
contrary, in Fig. 19(b) it could be seen that at lower current densities, there was reduced evolution of hydrogen bubbles resulting in
a cauliflower-like structure. These SEM images reinforced the initial hypothesis of concurrent deposition of copper and evolution
of hydrogen bubbles as described earlier. Diameter of the dish-like
structure was observed to be approximately 70 lm. Also, pore
sizes of 0.7–5 lm were observed. The measurements obtained
were based on a comparison of pixel size of the scale bar and pixel
size of the features seen on the SEM images. Diameters of the cauliflower-like structures were around 40 lm and the pore sizes
were in the range 0.5–2 lm.
Fig. 17. Contact angles of (a) Chip 3P (b) Chip 4P (c) Chip 5P (d) Chip 6P and (e) Chip 7P.
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
999
Fig. 18. Laser confocal images of (a) Chip 7P (B) Chip 6P (c) Chip 4P and (d) Chip 3P.
Fig. 19. SEM images of (a) Chip 3P and (b) Chip 6P.
The surfaces with porous coating were tested on pool boiling
setup using the procedure described in Section 4.2. The comparisons of pool boiling curves for a plain chip and Chips 3P–7P with
saturated water are shown in Fig. 20. From the pool boiling curves,
enhancements are observed in all chips prepared under different
current densities. The general trend observed may be summarized
as follows: as the current density is reduced initially, the pool boiling curve shifts towards the left. The optimum value of the current
density is observed for Chip 6P. Further reduction in the current
density however shifts the boiling curve to the right again. The
enhancement of CHF for these surfaces is in the approximate range
of 50–90%. It was seen that cauliflower-like microstructures performed better than the open dish-like structures.
At low current density (Chip 7P), there was not enough evolution of hydrogen bubbles, possibly creating a surface with a lower
porosity. These surfaces do not show substantial enhancement in
their pool boiling performance. There was an optimum value (Chip
6P) of initial current density which produced required pore sizes.
The cauliflower-like structures developed at this current density
enhanced the heat transfer performance the highest. This attained
a critical heat flux of approximately 1,400 kW/m2 at a wall superheat of 7 °C, giving a heat transfer coefficient of 179 kW/m2 °C.
Error bars are included in the pool boiling curve for Chip 3P. At
higher current densities (Chips 3P–5P), density of cauliflower-like
structures reduced due to a higher frequency of hydrogen bubble
nucleation. Open dish-like structures, created as a result of coalescence of bubbles, were seen. This results in reduction in copper
deposition. Error bars are indicated in the pool boiling curve for
Chip 3P. Thus from pool boiling curve it can be observed that there
is an optimal value of initial current density beyond which the
curve shifts to the right. This optimal value was used for further
investigation.
1000
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
applied for a longer time duration (2500–4500 s). XRD plot showed
that the deposition consisted of pure copper without oxides and
other impurities. The microporous coatings were hydrophilic in
nature with contact angles less than 15°. From this investigation,
it was concluded that the time duration of greater than 2500 s in
the second stage of the electrodeposition had no significant effect
on pool boiling performance. Chip 6P with cauliflower-like
structure gave the highest enhancement of heat transfer over a
plain surface, yielding an HTC of 176 kW/m2 °C. It was observed
that cauliflower-like microstructures performed better than open
dish-like microstructures, thus indicating that pool boiling
performance is a characteristic of the morphology.
Conflict of interest
None declared.
Fig. 20. Pool boiling curves for Chips 3P–7P with saturated distilled water.
Acknowledgments
The work on modeling using cyclic voltammetry and chronopotentiometry was carried out in Nanomaterial and Electrochemistry
Laboratory of School of chemistry and Materials Science. The work
on boiling was performed in the Thermal Analysis, Microfluidics
and Fuel Cell Laboratory in the Mechanical Engineering Department at Rochester Institute of Technology, Rochester, NY. The
authors gratefully acknowledge the financial support provided by
the National Science Foundation – United States, under CBET
Award No. 1335927.
References
Fig. 21. Heat transfer coefficients for Chips 3P–7P at different heat fluxes.
Another curve of heat flux against HTC was plotted to compare
the performance of the test chips as shown in Fig. 21. As seen from
this figure, the highest HTC of 179 kW/m2 °C was obtained with
Chip 6P. The maximum enhancement was approximately 275%
over a plain copper surface. Initially all the chips performed the
same, but at higher heat fluxes, Chip 6P performed better than
the other chips. The lowest enhancement of approximately 50%
was shown by Chip 3P. This variation could be attributed to the
variation of morphology as seen from the laser confocal images.
7. Conclusions
In this study, a two-step electrodeposition process was
investigated for creating porous structures on copper chips for pool
boiling enhancement. In the first step, a high current density
(300–650 mA/cm2) was applied for a short time (15 s). It was
observed that there was a concurrent evolution of hydrogen
bubbles along with deposition of copper, following an initial
atomic layer deposition of copper. A detailed study was conducted
to understand the bubble dynamics and nucleation of copper at the
cathode, and it was observed that the nucleation model was in
accordance with the 3-D model proposed by Scharifker and Hills
[19]. In the second step, a low current density (40 mA/cm2) was
[1] S.G. Kandlikar, M. Shoji, V.K. Dhir, Handbook of Phase Change, Taylor and
Francis, Philadelphia, PA, 1999.
[2] C.M. Patil, S.G. Kandlikar, Review of the manufacturing techniques for porous
surfaces used in enhanced pool boiling, Heat Transfer Eng. 35 (10) (2014) 887–
902.
[3] A.E. Bergles, M.C. Chyu, Characteristics of nucleate pool boiling from porous
metallic coatings, J. Heat Transfer 104 (1982) 279–285.
[4] T.M. Anderson, I. Mudawar, Microelectronic cooling by enhanced pool boiling
of a dielectric fluorocarbon liquid, Trans. ASME 111 (1989) 752–759.
[5] S.M. You, T.W. Simon, A. Bar-Cohen, A technique for enhancing boiling heat
transfer with application to cooling of electronic equipment, IEEE Trans.
Compon. Hybrids Manuf. Technol. 15 (5) (1992) 823–831.
[6] S.B. Memory, D.C. Sugiyama, P.J. Marto, Nucleate pool boiling of R-114 and R114-oil mixtures from smooth and enhanced surfaces—I. Single tubes, Int. J.
Heat Mass Transfer 38 (8) (1995) 1347–1361.
[7] I. Golobič, K. Ferjančič, The role of enhanced coated surface in pool boiling CHF
in FC-72, Heat Mass Transfer 36 (6) (2000) 525–531.
[8] J.Y. Chang, S.M. You, Enhanced boiling heat transfer from microporous
surfaces: effects of a coating composition and method, Int. J. Heat Mass
Transfer 40 (18) (1997) 4449–4460.
[9] C.E. Albertson, Boiling heat transfer surface and method, US Patent number
572,376, 1977.
[10] J.H. Kim, Enhancement of pool boiling heat transfer using thermally
conductive microporous coating techniques (Ph.D. thesis), University of
Texas at Arlington, TX, USA, 2006.
[11] M.S. El-Genk, A.F. Ali, Enhanced nucleate boiling on copper micro-porous
surfaces, Int. J. Multiphase Flow 36 (10) (2010) 780–792.
[12] Y. Im, C. Dietz, S.S. Lee, Y. Joshi, Flower-like CuO nanostructures for enhanced
boiling, Nanoscale Microscale Thermophys. Eng. 16 (3) (2012) 145–153.
[13] A.J. Bard, K.S.V. Santhanam, Application of controlled potential coulometry to
the study of electrode reactions, in: A.J. Bard (Ed.), Electroanalytic Chemistry,
vol. 4, Marcell Dekker, New York, NY, 1970. Chapter 3.
[14] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and
Applications, Wiley, New York, 2001.
[15] H.-C. Shin, M. Liu, Copper foam structures with highly porous nanostructured
walls, Chem. Mater. 16 (25) (2004) 5460–5464.
[16] A.I. Danilov, E.B. Molodkina, A.V. Rudnev, Y.M. Polukarov, J.M. Feliu, Kinetics of
copper deposition on Pt(111) and Au(111) electrodes in solutions of different
acidities, Electrochim. Acta 50 (25–26) (2005) 5032–5043.
[17] G. Gunawardena, G. Hills, I. Montenegro, Electrochemical nucleation: part IV.
Electrodeposition of copper onto vitreous carbon, J. Electroanal. Chem. 184 (2)
(1985) 357–369.
[18] M. Palomar-Pardavé, I. González, N. Batina, New insights into evaluation of
kinetic parameters for potentiostatic metal deposition with underpotential
C.M. Patil et al. / International Journal of Heat and Mass Transfer 79 (2014) 989–1001
and overpotential deposition processes, J. Phys. Chem. B 104 (15) (2000)
3545–3555.
[19] B. Scharifker, G. Hills, Theoretical and experimental studies of multiple
nucleation, Electrochim. Acta 28 (7) (1983) 879–889.
[20] O.R. Brown, H.R. Thirsk, The rate-determining step in the electro-deposition of
copper on copper from aqueous cupric sulphate solutions, Electrochim. Acta
10 (4) (1965) 383–393.
[21] M.S. Moats, J.B. Hiskey, D.W. Collins, The effect of copper, acid and
temperature on the diffusion coefficient of cupric ions in simulated
electrorefining electrolytes, Hydrometallurgy 56 (2000) 255–268.
1001
[22] M.E. Hyde, R.G. Compton, A review of the analysis of multiple nucleation with
diffusion controlled growth, J. Electroanal. Chem. 549 (2003) 1–12.
[23] M. Fleischmann, J.A. Harrison, H.R. Thirsk, Electrocrystallization of thin films of
nickel, Trans. Faraday Soc. 61 (1965) 2742–2753.
[24] D. Cooke, S.G. Kandlikar, Pool boiling heat transfer and bubble dynamics over
plain and enhanced microchannels, J. Heat Transfer 133 (5) (2011) 052902.
[25] J.H. Kim, T.W. Simon, R. Viskanta, Journal of heat transfer policy on reporting
uncertainties in experimental measurements and results, J. Heat Transfer 115
(5–6) (1993) 5–6.