See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/224145659
A Flat Heat Pipe Architecture Based on
Nanostructured Titania
Article in Journal of Microelectromechanical Systems · September 2010
DOI: 10.1109/JMEMS.2010.2051019 · Source: IEEE Xplore
CITATIONS
READS
28
93
6 authors, including:
Changsong Ding
Gaurav Soni
16 PUBLICATIONS 68 CITATIONS
47 PUBLICATIONS 90 CITATIONS
University of California, Santa Barbara
Amritsar College of Engineering and Technol…
SEE PROFILE
SEE PROFILE
Payam Bozorgi
Carl D Meinhart
7 PUBLICATIONS 41 CITATIONS
179 PUBLICATIONS 6,051 CITATIONS
University of California, Santa Barbara
SEE PROFILE
University of California, Santa Barbara
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
LTE LINK INVESTIGATION USING VST View project
All content following this page was uploaded by Carl D Meinhart on 22 March 2014.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
878
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 19, NO. 4, AUGUST 2010
A Flat Heat Pipe Architecture Based on
Nanostructured Titania
Changsong Ding, Gaurav Soni, Payam Bozorgi, Brian D. Piorek, Carl D. Meinhart, and Noel C. MacDonald
Abstract—A novel 3 cm × 3 cm × 600 µm-thick Ti-based flat
heat pipe is developed for Thermal Ground Plane (TGP) applications. The Ti-based heat pipe architecture is constructed by
laser welding two microfabricated titanium substrates to form
a hermetically sealed vapor chamber. The scalable heat pipes’
flat geometry facilitates contact with planar heat sources, such as
microprocessor chip surfaces, thereby reducing thermal contact
resistance and improving system packaging. Fluid transport is
driven by the wicking structure in the TGP, which consists of
an array of Ti pillars that are microfabricated from a titanium
substrate using recently developed high-aspect-ratio Ti processing
techniques [1]. The hydrophilic nature of the Ti pillars is increased
further by growing ∼200-nm hairlike nanostructured titania of
the pillar surfaces. The resulting super hydrophilic wick offers
the potential to generate high wicking velocities of ∼27.5 mm/s
over distances of 2 mm. The experimental wetting results show
a diffusive spreading behavior that is predicted by Washburn
dynamics [2]. The maximum effective thermal conductivity of a
heat pipe is directly related to the speed of capillary flow of the
working fluid through the wick and is measured experimentally
in the first-generation device to be k = 350 W/m · K. A dummy
TGP with a cavity volume of ∼170 µL was used to test the hermiticity level of the laser packaging technique. The device gave a
0.067% of water loss based on ∼60 µL of charged water at 100 ◦ C
in air for over a year.
[2009-0303]
Index Terms—Heat pipe, nanostructure, Thermal Ground
Plane (TGP), titanium, vapor chamber.
I. I NTRODUCTION
OOLING devices such as heat sinks, fans, and heat pipes
have long been utilized for cooling microprocessor chips
[3]. Heat pipes have been of particular use in cooling powerintensive chips because of their high thermal conductivity and
their capacity to carry high amounts of heat [4]. Heat pipes are
composed of the following three components: a hermetically
sealed cavity, a working fluid, and a wick structure on the
internal walls of the cavity. Different kinds of materials such as
C
Manuscript received December 2, 2009; revised April 13, 2010; accepted
May 3, 2010. Date of publication June 14, 2010; date of current version
July 30, 2010. This work was supported by the TGP program from the Microsystems Technology Office, Defense Advanced Research Projects Agency,
under Grant N66001-08-1-2035. A portion of this work was done in the UCSB
nanofabrication facility, which is part of the NSF-funded NNIN network.
Subject Editor R. T. Howe.
C. Ding is with the California NanoSystems Institute, University of
California, Santa Barbara, CA 93106-6105 USA (e-mail: changsong_
ding@yahoo.com).
G. Soni, P. Bozorgi, C. D. Meinhart, and N. C. MacDonald are with
the Department of Mechanical Engineering, University of California, Santa
Barbara, CA 93106-9560 USA.
B. D. Piorek is with SpectraFluidics, Inc., Santa Barbara, CA 93117 USA,
and also with Gas Reaction Technologies, Inc., Santa Barbara, CA 93110 USA.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JMEMS.2010.2051019
biporous copper [5], sintered copper [6], carbon nanotubes [7],
microstructures with different topologies [8] and patterns [9],
and micromachined channels have been investigated for micro
heat pipes for their wicking functions. The hermetically sealed
cavity is filled with a working fluid such as water or ammonia
etc., depending on the heat load and design operating temperatures. The high heat-carrying capacity of a heat pipe is due
to the latent heat and transport of the working fluid. During
operation, the working fluid evaporates by absorbing heat into
a hot region (heat source) of the heat pipe and flows as a
vapor toward a cold region (heat sink) where condensation then
occurs due to the cooler temperature. A very fine wick structure
on the internal wall of the heat pipe draws the condensate from
the cold region to the hot region using capillary action.
This transport cycle is continuous within a heat pipe, thereby
transporting large amounts of heat. Since the latent heat of
evaporation of the working fluid is high relative to sensible heat,
heat pipes can transport much higher amounts of heat than what
can be conducted through a solid rod of the same dimension of
solid materials.
Heat pipes are an essential component of contemporary
laptop cooling systems because of the high power dissipated by
modern microprocessor chips. Generally, heat pipes are cylindrical in shape and are made of copper, with a fine copper wire
mesh attached to the internal walls. Such cylindrical heat pipes
do not establish good thermal contact with the flat surfaces on
microprocessor chips, and they are not easily packaged in a planar geometry. Flat heat pipes are known as vapor chambers, and
they suffer from poor fluid transport efficiency and thereby have
a low thermal conductivity compared to cylindrical heat pipes.
Titanium offers many advantageous properties relative
to other micromechanical materials, including high fracture
toughness and excellent corrosion resistance. Using recently
developed micromachining techniques based on Cl2 /Ar dry
etching [1], [10], high-aspect-ratio structures with micrometerscale and nanoscale features are made out of titanium foil/
substrates. The etched substrates can then be bonded or welded
to other substrates to enable the realization of complex multilayered devices. These foil substrates can be obtained in
thicknesses ranging from 10 µm to many hundred micrometers,
with minimal intrinsic residual stress. The high modulus and
weldability of titanium allow greater design flexibility than
what is possible with other materials such as ceramics, low
modulus metals, and deposition-based thin films, where residual stresses can cause undesirable deformation of the device
structures [11].
Titanium is a unique material for Thermal Ground Plane
(TGP) in that it can be microfabricated and oxidized to create
1057-7157/$26.00 © 2010 IEEE
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on August 04,2010 at 16:42:19 UTC from IEEE Xplore. Restrictions apply.
DING et al.: FLAT HEAT PIPE ARCHITECTURE BASED ON NANOSTRUCTURED TITANIA
879
Fig. 1. Schematic of the Ti-based TGP architecture. The overall device
thickness is less than 1 mm. The bottom substrate is a 250-µm-thick Ti. The
super hydrophilic substrate is a combination of Ti and TiO2 and is 50 µm in
height. The vapor cavity is 100 µm tall. The sealing structure is 200 µm thick.
nanoscale features. In addition, Ti can be macromachined and
laser welded to produce a hermetically sealed chamber. The
high fracture toughness of Ti makes it an excellent material for
fabrication of thin, lightweight, flexible, and robust heat pipes.
The current work demonstrates the proof of concept for
fabrication and performance of a Ti-based flat heat pipe
(vapor chamber) for cooling planar geometries. This approach
incorporates microfabricated Ti pillars that are oxidized to
produce nanoscale features to create a super hydrophilic wicking structure. The wick can be optimized further to increase
fluid transport, and thereby, it achieves a high heat-transfer
performance.
II. D EVICE G EOMETRY
A schematic of the first-generation TGP is shown in Fig. 1.
The device is formed by microfabricating Ti pillars from a
Ti wafer. A second substrate is laser welded to the first substrate
to form a hermetically sealed vapor chamber. The overall thickness is approximately 600 µm, and it has a lateral size of 3 cm ×
3 cm. The lateral size can be readily extended to larger sizes in
future designs. The overall weight of the device is measured to
be 3.4 gm. A wicking structure is formed from Ti pillars that
have a diameter of 5 µm, a center-to-center pitch of 10 µm,
and a height of 50 µm. These Ti pillars are oxidized to form
nanostructured titania (NST) with characteristic dimensions of
approximately 200 nm.
Water is used as the working fluid. During operation, heat
is applied to the wicking substrate, causing evaporation. The
water vapor is transported to a cold region and is condensed
by rejecting heat with a heat sink. The condensed water is then
wicked to the evaporating region. This process is continuous
during the TGP operation.
III. FABRICATION P ROCESS
Bulk titanium processing techniques have been developed
since 2001 for harsh environment microsystems such as mirrors
[1], electrodes [12], and relays [13]. Argon and chlorine are
usually used for dry etching of Ti [1], [10]. For the current
TGP device, we use well polished commercial grade 1 (CP1)
Ti substrates. The fabrication process for making Ti pillars
begins with the SiO2 -masked deep etching of a 300-µm-thick
Fig. 2. SEM photograph of the deep-etched bulk Ti pillar array. The pillars
have a diameter of 5 µm, a center-to-center pitch of 10 µm, and a height of
50 µm.
Ti substrate using inductively coupled plasma etch in an Ar/Cl2
ambience. The SiO2 layer is approximately 3 µm thick. After
deep etching is complete, the sample is oxidized in a 30%
solution of hydrogen peroxide at 83 ◦ C to obtain the NST
hair at the pillar surfaces and the bottom floor surface. NST
fabrication, however, is a challenging task, which poses several
difficulties. For example, peeling of the NST layer from the
Ti surface is often observed when the sample is dipped into
hydrogen peroxide immediately after the deep etch step. We
believe that it is the TiClx passivation layer and the native oxide
which inhibit the robust formation of NST. To minimize this
problem, we apply an isotropic etch to remove the sidewall
passivation layer and contaminants. A high-pressure plasma
system, which etches away both the top SiO2 masking material
and the sidewall passivation layer and contaminants, is ideal
for this purpose. This etch is realized using a Technics PE
IIA plasma etcher (100 kHz RIE) with a CF4 : O2 chemistry
(85 : 15). With this system, an oxygen cleaning procedure
(100 W, 300 mtorr, and 1 min) is first applied to the samples to
remove the hydrocarbon that may originate from air; then, the
CF4 : O2 chemistry (85 : 15 sccm) is applied (300 W, 300 mtorr,
and 15 min) to etch away the top masking material and the
side wall passivation layer/contaminants and to expose the
fresh Ti surface. Following this etching step, the same oxygen
cleaning procedure is applied again to remove the possible CF4
accumulation on the pillar sidewalls. A very thin layer of TiO2
may form on the sample surfaces. However, it does not affect
NST formation in the following oxidization procedure.
An SEM image of the microscale pillars is shown in Fig. 2.
A close-up view of the NST hair on a pillar surface is shown in
Fig. 3. The NST film enhances the local wetting of the Ti posts.
The wicking structure has length scale ranges of 50-µm high
posts, 5-µm diameters, and ∼200-nm NST hairs. The range
of length scales enhances the wetting characteristics of the
wicking structure and is a key for optimal TGP performance.
As mentioned earlier, the NST film enhances the local wetting of the Ti posts. Once wetted, the micropillars create a capillary action over the centimeter length scale. Such a combination
of multiple length scales produces a super hydrophilic structure.
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on August 04,2010 at 16:42:19 UTC from IEEE Xplore. Restrictions apply.
880
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 19, NO. 4, AUGUST 2010
TABLE I
L ASER PARAMETER A PPLIED FOR W ELDING THE TGP D EVICE
Fig. 3. SEM image of the Ti microstructure. The Ti pillars have a height of
50 µm and a diameter of 5 µm, and they are etched into a Ti substrate with
a 5-µm gap. Ti is oxidized to produce ∼200-nm hairlike NST. The hairlike
structure enhances wetting of Ti pillars.
Fig. 5. Hydrophilic nature of the wick was tested by placing a drop of water
at the edge of the wicking substrate. The drop of water collapsed and traveled
among the pillar arrays. The left figure shows the wetting profile after 66 ms,
and the right figure shows the wetting profile at 200 ms.
Fig. 4. (Left) Photograph of the first-generation Ti-based TGP. The two Ti
substrates are laser welded to form a hermetically sealed vapor chamber. A
small quantity of water is sealed within the vapor chamber. (Right) Close-up of
the microscale laser weld that joins the two Ti substrates.
this paper. By laser welding a patch to seal the charging hole,
the degassing issue of epoxy will be avoided. We expect it to be
more reliable either by sealing with laser welding or by capping
necessary charging tubes with Swagelok end caps.
IV. D EVICE PACKAGING U SING L ASER W ELDING
V. W ETTING P ERFORMANCE OF THE TGP
For continuous operation, the working fluid and the wick
structure must be hermetically sealed in the vapor cavity.
The performance of the TGP depends significantly on the
quality of packaging. A major problem with the packaging techniques used conventionally for microfabrication, such as hightemperature thermocompression and flip-chip bonding, is the
degradation of reliability, which is caused by the excess stress
due to thermal mismatch. In order to eliminate the stresses
that occur at high temperature, we take a different approach
by applying a pulsed laser as a local and rapid heat source to
locally weld the Ti substrates, instead of bonding the substrates
by heating the entire device.
A pulsed Nd:YAG laser (neodymium-doped yttrium aluminum garnet; Nd : Y3 Al5 O12 ) with a wavelength of 1064 nm
was applied as a heat source to locally weld a 30 mm × 30 mm
titanium square lid to a titanium substrate (shown in Fig. 4).
The applied laser parameters are enumerated in Table I. The
hermiticity of the packaged device achieved the leaking rate of
10−11 std cm3 /sec using the helium leaking method.
A hole (1.8 mm in diameter) was drilled into the welded
package for water charging. The cavity was charged with a measured quantity of water at atmospheric pressure. The quantity of
water was judiciously chosen to fill the pores up to the top of
the pillars under atmospheric pressure at 100 ◦ C. After adding
water, the inlet hole was covered with a small 3 mm × 3 mm
piece of Ti and was sealed with an epoxy-based adhesive to
form the fully encapsulated TGP device. A laser-sealed (epoxy
free) prototype for hermiticity test is shown in the latter part of
Washburn [2] originally analyzed the dynamics of capillary
flow through a circular tube. By balancing surface tension
forces with viscous resistance, he showed that the wetting speed
can be written as
dx
rγ cos θ 1
=
dt
4η x
(1)
where x is the length of the penetration, t is the period of the
penetration, r is the radius of the capillary, γ is the surface
tension of water, η is the viscosity of water, and θ is the contact
angle.
An important feature of this equation is that the wetting velocity scales as x−1 . The penetration length can be calculated by
integrating (1), and it is shown to increase with the square root
of time x = (Dt)1/2 , where D is the dynamic coefficient. A
high dynamic coefficient is preferred for the wicking structure,
as this performance metric is directly related to heat transport
in the TGP.
The wetting characteristics of the wicking structure (i.e.,
dynamic coefficient) was measured experimentally. A drop of
DI water was placed at the edge of the wicking substrate at
room temperature. Once in contact with the surface, the droplet
collapsed into the wicking substrate and traveled through the
pillar array.
Fig. 5 shows two snapshots recorded during the wetting
experiment. The flow of water through the pillars was recorded
with a digital camera at 15 frames per second, and the front
boundary of water was tracked. The front boundary travels
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on August 04,2010 at 16:42:19 UTC from IEEE Xplore. Restrictions apply.
DING et al.: FLAT HEAT PIPE ARCHITECTURE BASED ON NANOSTRUCTURED TITANIA
Fig. 6. Length of the wetted region as a function of time. The symbols are
the experimental data, while the solid line is the least squares fit of x = 10.5 ×
10−3 t1/2 for pillars with NST and x = 6.5 × 10−3 t1/2 for Ti pillars without
NST.
approximately 10 mm in 1 s. The distance traveled by water
(i.e., the travel distance of the wetted region) as a function of
time is shown in Fig. 6.
A curve of the form x2 = Dt was fitted to the data for Ti
pillars and NST-coated Ti pillars. The results show that D =
4.225 × 10−5 m2 s−1 for Ti pillars (data not shown). When
NST is coated on the Ti pillars, the dynamic coefficient D
increased by 160.95% to
x2 = Dt = 1.1025 × 10−4 t.
(2)
The agreement between the experimental data and the curve fit
shown in Fig. 6 indicates that the flow distance follows a square
root scaling in time. Therefore, we conclude that the wetting in
the Ti and NST/Ti pillar array behaves according to Washburn
dynamics due to the diffusive wetting nature.
The experimental value of the liquid flow speed dx/dt was
determined by differentiating (2), such that
5.5125 × 10−5
dx
=
.
dt
x
(3)
The wetting velocity dx/dt = D/2x is proportional to the
inverse of the penetration distance and is shown in Fig. 7 as
a function of time for Ti pillars and NST-coated Ti pillars.
Similarly, the wetting velocity profile for the Ti pillars is
derived to be dx/dt = 2.12e − 5/x. The flow speed decreases
∼x1/2 . The velocity for the NST/Ti pillars is ∼27.5 mm/s
at a penetration distance of 2 mm, but it decreases to only
5.5 mm/s at a penetration distance of 10 mm. Literature data
profiles are needed for wetting speed comparisons to other
wicking materials such as porous copper, carbon nanotubes, and
microstructures.
The Washburn equation was originally derived for a circular
tube, with r being the radius [2]. The titanium wicking material
has a 5 µm × 50 µm feature geometry gap among the pillars.
By assuming that we can use r = 2.5 µm (half of the gap), that
the contact angle is very close to zero, and that water properties
are at room temperature, the Washburn equation (1) predicts
a wetting velocity profile to be dx/dt = 1.01e − 4/(2x). This
dynamic coefficient result is ∼8.2% larger than the measurement from NST wicking material.
881
Fig. 7. Velocity plot of Ti pillars and NST-coated Ti Pillars using (3). The
wetting velocity for the NST-coated pillars is 160.95% higher than just Ti
pillars. The wetting speed for the NST pillars is ∼25 mm/s at a distance of
2 mm away from the water drop.
VI. H EAT T RANSPORT P ERFORMANCE OF THE TGP
The rate of heat transport in the TGP is proportional to
the rate of evaporation of the working fluid under steady-state
conditions
dme
dq 1
=
dt
dt hf g
(4)
where dq/dt is the power input and hf g is the latent heat of
evaporation of the fluid.
The mass flow rate of capillary flow through the wick structure is given by
dmc
= ρνAp
dt
(5)
where ρ is the density of the fluid, ν is the wicking speed [given
by (3)], and Ap is the cross-sectional area occupied by the fluid
in the wick, which is a function of x. By assuming a triangular
profile of the fluid in the wick, Ap is equal to the cross-sectional
area by taking half of the pillar height.
The rate of heat transported by the TGP can be calculated by
combining (4) and (5)
dqmax
= ρhf g νAp .
dt
(6)
Substituting the wicking speed ν using the modified
Washburn equation yields
dqmax
D Ap
= ρhf g
dt
2 L
(7)
where L is the length scale over which the heat is transported.
The heat-carrying capacity decreases with the length of the heat
pipe. For liquid water at 100 ◦ C, we obtain hf g = 2256.9 kJ/kg
and ρ = 958.27 kg/m3 . The speed of capillary flow observed
in our experiments (at room temperature) is in the range of
2–25 mm/s (Fig. 7) for the first 10-mm range. By assuming
that the wicking material begins to dry out at a small region
near the heat source, we calculate the heat-carrying capacity
for a high representative velocity of 20 mm/s. The combined
pore cross-sectional area can be found in the following way:
Since the pore width and the diameter of the pillar are equal, the
combined width of the pores is half of the TGP width (which
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on August 04,2010 at 16:42:19 UTC from IEEE Xplore. Restrictions apply.
882
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 19, NO. 4, AUGUST 2010
Fig. 9. Hermiticity data test. Panel (a) is the front side of a dummy TGP. Panel
(b) shows the back side, which has a patch that is used to seal the hole used for
charging water.
Fig. 8. Effective thermal conductivity of the TGP as a function of the
temperature of the heat source. The maximum value was determined to be
∼350 W/m · K.
is 3 cm). The height of the pores is about 50 µm. This yields a
combined cross-sectional area of 7.5 × 10−7 m2 . Substituting
these values in (6) yields a characteristic rate of heat transfer of
dq/dt ≈ 18.25 W.
We have performed experiments which demonstrate the
current TGP’s performance as a function of the heat source
temperature. In these experiments, solid copper bars were used
to transfer heat into and out of the TGP. The interfacial area of
the bars was matched to the area of the hot and cold regions of
the TGP, which was 0.5 cm by 3 cm. A Kapton thin-film heater
(McMaster–Carr, item 7980T11) heated the copper bar used
for heat injection. The power applied to the Kapton heater was
controlled by a variac driven at 120 V and 60 Hz. The distal end
of the copper bar used for removing heat from the cold region
of the TGP was cooled with liquid nitrogen. To prevent thermal
communication with the ambient, the copper bars and TGP were
insulated with two layers of 1/8-in-thick butyl-coated PVC foam
thermal insulation tape (McMaster–Carr, item 75875A662).
The temperature profile of the TGP was determined by
monitoring thermocouples located at the hot and cold TGP
interfaces. The heat flux through the TGP was determined by
monitoring a third thermocouple that is located 1 cm away from
the TGP interface on the hot copper bar, which provided powertransfer calculations using the known thermal conductivity of
copper. As shown in Fig. 8, experiments were conducted at
the following four separate fixed input temperatures: 97.5 ◦ C,
100.7 ◦ C, 105.4 ◦ C, and 112.3 ◦ C. The resulting power transfer
through the TGP was 4.0, 4.8, 7.2, and 6.4 W, respectively.
Once a fixed temperature was applied to the hot side of the TGP,
the system was allowed to achieve a steady state over 15 min
before measurements were taken. TGP thermal conductivity
was calculated using the measured power transfer, heat source
(copper block) dimensions (0.5 cm × 2 cm), and TGP dimensions (height is 0.6 mm, width is 3.0 cm, and length is 3.0 cm).
Thermal paste (OMEGATHERM OT-201 Series) was used
between the TGP and the hot and cold bars, and its resistance
was considered in the TGP thermal conductivity calculations.
The thermal resistance of the device is 7.19 K/W, and the
contact resistance on each side of the contacts is 2.25 K/W. The
maximum effective thermal conductivity is k ∼ 350 W/m · K,
Fig. 10.
Percentage loss of water as a function of time.
which occurs at an input temperature of 105.4 ◦ C. These values
were calculated based on the aforementioned TGP dimensions
with a length of 3 cm. The conductivity values that are shown
in Fig. 8 are proportional to the length to be used. The effective
conductivity values under different tilt angles and accelerations
are important performance aspects of TGP devices for specific
applications. These topics are under investigation and beyond
the scope of this paper.
VII. H ERMITICITY T EST
A 3 cm × 3 cm dummy TGP device was made with the same
dimensions as Fig. 4 for hermiticity test. The device has a cavity
of 200 µm × 3 cm × 3 cm, with standoffs inside to support.
The net volume of the cavity is ∼170 µL. An amount of 60 µL
of water was charged to the device at room temperature and atmospheric pressure before sealing the patch using laser welding
technique, with argon as the shielding gas. The heat generated
by laser welding is local due to the short pulse of the laser
energy so that it does not reach the center area of the TGP
device. The TGP device, as shown in Fig. 9, was then weighted
with the Mettler Toledo lab scale (with 0.1-mg resolution),
confirming that 60 ± 1 µL of water was enclosed.
The weight of the device was recorded before it was loaded
into an oven where the device was kept at 100 ◦ C in air
ambient. The water loss was measured by weighting the device
with Mettler Toledo lab scale with 0.1-mg resolution. The
percentage of water loss is shown in Fig. 10. The device gives a
0.067% of water loss based on 60 µL of charged water for over
a year term long at 100 ◦ C in air.
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on August 04,2010 at 16:42:19 UTC from IEEE Xplore. Restrictions apply.
DING et al.: FLAT HEAT PIPE ARCHITECTURE BASED ON NANOSTRUCTURED TITANIA
VIII. C ONCLUSION
A novel Ti-based flat heat pipe is developed by laser welding microfabricated Ti substrates. The device size is laterally
3 cm × 3 cm, with a thickness of 600 µm. A high-performance
capillary wick was formed by etching 5-µm diameter pillars in
titanium. The pillar array has a 10-µm center-to-center pitch
and a height of 50 µm. The hydrophilic nature of the Ti pillars
is increased by growing nanoscale (∼200 nm) hairlike NST on
the pillar surfaces. The wetting experiments show that the wick
structure obeys Washburn dynamics and that it has a capillary
velocity of ∼27.5 mm/s at a distance of 2 mm, which decreases
to 5.5 mm/s at a distance of 10 mm. The wetting velocities
increase 161% when the Ti pillars are coated with NST, as
compared to uncoated Ti pillars.
The first-generation TGP device was constructed and found
to have an effective thermal conductivity of 350 W/m · K, with a
heat load of 7.2 W and with the hot side temperature of the TGP
being 105.4 ◦ C. The thermal conductivity value is proportional
to the length to be used for calculation, based on the thermal
resistance model. These data confirm the proof of concept of
the Ti-based heat pipe. A dummy TGP with a cavity volume
of ∼170 µL was used to test the hermiticity level of the laser
packaging technique. The device gave a 0.067% of water loss
based on ∼60 µL of charged water at 100 ◦ C in air for over a
year. We expect future generations of the Ti-based TGP to have
significantly higher performance through the optimization of
several design parameters and by eliminating noncondensable
gases with a better optimized charging system.
R EFERENCES
[1] M. F. Aimi, M. P. Rao, N. C. MacDonald, A. S. Zuruzi, and D. P. Bothman,
“High-aspect-ratio bulk micromachining of titanium,” Nat. Mater., vol. 3,
no. 2, pp. 103–105, Feb. 2004.
[2] E. W. Washburn, “The dynamics of capillary flow,” Phys. Rev., vol. 17,
no. 3, pp. 273–283, Mar. 1921.
[3] U. Vadakkan, G. M. Chrysler, and S. Sane, “Silicon/water vapor
chamber as heat spreaders for microelectronic packages,” in Proc.
21st Annu. IEEE Semicond. Thermal Meas. Manage. Symp., 2005,
pp. 182–186.
[4] H. Xie, H. Xie, A. Ali, and R. Bhatia, “The use of heat pipes in
personal computers,” in Proc. 6th Intersoc. Conf. ITHERM, 1998,
pp. 442–448.
[5] T. Semenic and I. Catton, “Experimental study of biporous wicks for
high heat flux applications,” Int. J. Heat Mass Transf., vol. 52, no. 21/22,
pp. 5113–5121, Oct. 2009.
[6] H. de Bock, K. Varanasi, P. Chamarthy, T. Deng, A. Kulkarni,
B. Rush, B. Russ, S. Weaver, and F. Gerner, “Experimental investigation
of micro–nano heat pipe wick structures,” in Proc. ASME IMECE, 2008,
pp. 1–6.
[7] Q. Cai, C.-L. Chen, G. Xiong, and Z. Ren, “Explorations of carbon
nanotube wick structure for high heat flux cooling,” in Proc. 2008 ASME
Heat Transfer Summer Conf., Jacksonville, FL, pp. 293–297.
[8] R. Ranjan, J. Murthy, and S. Garimella, “Analysis of the wicking and
thin-film evaporation characteristics of microstructures,” J. Heat Transf.,
vol. 131, no. 10, p. 101 001, Oct. 2009.
[9] Y. Chen, L. Melvin, S. Rodriguez, D. Bell, and M. Weislogel, “Capillary driven flow in micro scale surface structures,” Microelectron. Eng.,
vol. 86, no. 4–6, pp. 1317–1320, Apr.–Jun. 2009.
[10] E. R. Parker, M. F. Aimi, B. J. Thibeault, M. P. Rao, and
N. C. MacDonald, “High-aspect-ratio inductively coupled plasma etching
of bulk titanium for MEMS applications,” in Proc. 206th Meeting Electrochem. Soc., Honolulu, HI, 2004, p. 2580.
[11] C. O’Mahony, M. Hill, P. J. Hughes, and W. A. Lane, “Titanium as
a micromechanical material,” J. Micromech. Microeng., vol. 12, no. 4,
pp. 438–443, Jun. 2002.
883
[12] M. P. Rao, M. F. Aimi, and N. C. MacDonald, “Single-mask, threedimensional microfabrication of high-aspect-ratio structures in bulk silicon using reactive ion etching lag and sacrificial oxidation,” Appl. Phys.
Lett., vol. 85, no. 25, pp. 6281–6283, Dec. 2004.
[13] C. Ding, X. Huang, G. Gregori, E. R. Parker, M. P. Rao, D. R. Clarke,
and N. C. MacDonald, “Development of bulk-titanium-based MEMS
RF switch for harsh environment applications,” in Proc. ASME Conf.,
Orlando, FL, 2005, pp. 123–126.
Changsong Ding received the B.S. degree in naval
architecture and ocean engineering from Shanghai
Jiao Tong University, Shanghai, China, in 1996, and
the M.S. and Ph.D. degrees in mechanical engineering from the University of California, Santa Barbara
(UCSB), in 2003 and 2007, respectively, where he
investigated and developed nonsilicon material micromachining and nanomodification-based 3-D microelectromechanical systems (MEMS).
From 1996 to 2001, he worked for several industrial companies including Delphi Automotive Systems Inc. and CNOOC, among others. He is currently an Assistant Project
Scientist with the California NanoSystems Institute, UCSB. His current research focuses on bulk titanium-based MEMS including switch and flat heat
pipe applications.
Gaurav Soni received the B.Tech. degree in mechanical engineering from the Indian Institute of
Technology, New Delhi, India, and the M.S. and
Ph.D. degrees in mechanical engineering from the
University of California, Santa Barbara.
He is currently a Senior Analysis Engineer in
the solar power generation industry. His research
interests include fluid dynamics and heat transfer at
both small and large scales.
Payam Bozorgi received the B.S. degree in mechanical engineering from Tehran Polytechnic University,
Tehran, Iran, in 2000, and the M.S. degree in mechanical engineering from the University of Southern
California, Los Angeles, in 2004. Since 2006, he has
been working toward the Ph.D. degree at the University of California, Santa Barbara (UCSB), where
he developed titanium-based microelectromechanical system packaging by using a micropulsed laser
welding approach.
He is also a Research Assistant with the California
NanoSystems Institute, UCSB.
Brian D. Piorek received the B.S. degree in chemical engineering and the Ph.D. degree in mechanical
engineering from the University of California, Santa
Barbara (UCSB), in 2002 and 2008, respectively,
where he developed a microfluidic gas-phase chemical sensor technology.
He is a Visiting Research Scientist with the
California NanoSystems Institute, UCSB, where he
investigates the fundamental transport phenomena
for microfluidic and chemical sensor applications.
In 2008, he cofounded SpectraFluidics, Inc., Santa
Barbara, where he currently serves as the Vice-President of Research and
Development. Since 2004, he has also been with Gas Reaction Technologies,
Inc., Santa Barbara.
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on August 04,2010 at 16:42:19 UTC from IEEE Xplore. Restrictions apply.
884
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 19, NO. 4, AUGUST 2010
Carl D. Meinhart received the Ph.D. degree from
the University of Illinois at Urbana–Champaign,
Urbana, in 1995.
While at the University of Illinois, his research
involved the investigation of wall-bounded turbulent
flows. Since 1996, he has been with the University
of California, Santa Barbara, where he is currently
a Professor of mechanical engineering and where
his research has focused on developing and applying
microresolution particle image velocimetry (microPIV) to study transport phenomena in microfluidic
devices. His research group is involved in the development and analysis of bioMEMS for immunoassays and molecular diagnostics and in the development
of microfluidics for the detection of airborne molecules, such as explosives and
narcotics.
Noel C. MacDonald received the Ph.D. degree
in electrical engineering from the University of
California, Berkeley, in 1967.
He holds a joint professorial appointment in the
Department of Mechanical and Environmental Engineering and the Department of Materials, University of California, Santa Barbara (UCSB), where he
also holds the Fred Kavli Chair in Microelectromechanical Systems Technology. He was an Acting
Assistant Professor at the University of California,
Berkeley, from 1967 to 1968, he was a member of
the Technical Staff at Rockwell International Science Center, and he held
management positions at Physical Electronics Industries, Inc., and PerkinElmer Company from 1968 to 1982. Before joining UCSB, he held the
Acheson/Laibe Professorship in Engineering at Cornell University, Ithaca, NY,
and he served as the Director of the Cornell Nanofabrication Facility and the
Chairman of the School of Electrical Engineering, Cornell University. From
1997 to 1999, he served as the Director of the Microsystems Technology Office,
Defense Advanced Research Projects Agency. In 2000, he was elected to the
National Academy of Engineering for his contributions to the development of
the scanning auger microprobe and micromachined microinstruments. He has
written or coauthored over 100 papers and several book chapters. He holds
61 patents. His current research interests include advanced materials processing
as well as interdisciplinary research for micro/nanoelectromechanical systems.
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on August 04,2010 at 16:42:19 UTC from IEEE Xplore. Restrictions apply.
View publication stats