Experimental Thermal and Fluid Science 35 (2011) 1274–1281

Contents lists available at ScienceDirect

Experimental Thermal and Fluid Science

journal homepage: www.elsevier.com/locate/etfs

Application of nanofluids to a heat pipe liquid-block and the thermoelectric

cooling of electronic equipment
Nandy Putra ⇑, Yanuar, Ferdiansyah N. Iskandar
Heat Transfer Laboratory, Department of Mechanical Engineering, University of Indonesia, Kampus UI, Depok 16424, Indonesia

a r t i c l e i n f o a b s t r a c t

Article history: Microprocessor power dissipation is constantly increasing. An increase in microprocessor size has also
Received 6 March 2011 resulted in higher heat fluxes. The growth of information technology has rapidly increased over the past
Received in revised form 21 April 2011 few years, causing an increase in the demand for a microprocessor that has a very high computing ability.
Accepted 21 April 2011
The previous generation of central processing units (CPU) had 1.17 billion transistors planted in it, which
Available online 30 April 2011
indicates that a significant amount of heat was generated. The total heat dissipation resulting from a high
end CPU is approximately 110–140 W, which will increase if the CPU voltage and frequency increase.
Keywords:
Conventional air-cooled cooling systems are no longer adequate to remove these heat fluxes. For a num-
Nanofluid
Electronic cooling
ber of applications, direct air-cooling systems will have to be replaced or enhanced by other high perfor-
Heat pipe liquid-block mance compact cooling techniques. In this study, the application of nanofluids as the working fluid on a
Heat transfer heat pipe liquid-block combined with thermoelectric cooling is investigated. The type and effect of vol-
Thermoelectric ume concentrations of nanofluids, coolant temperature, and thermoelectricsystem as heat pumps of a PC
on the CPU’s temperature are considered. The results obtained from this technique are compared to those
from other conventional cooling techniques. The heat pipe liquid-block combined with the thermoelec-
tric system has a significant effect on heat transfer from the CPU. The higher thermal performance heat
pipe liquid-block and thermoelectric cooled system with nanofluids proved its potential as a working
fluid.
Ó 2011 Elsevier Inc. All rights reserved.

1. Introduction achieve much higher heat fluxes than through conventional forced
air-cooling. This explains why considerable research has been redi-
In recent years, a significant increase in microprocessor power rected towards these approaches for the thermal management of
dissipation coupled with CPU size has resulted in an increase in electronics. In direct liquid cooling, electronics are either im-
heat fluxes. Microprocessor heat fluxes have also increased and mersed in a pool or in contact with droplets, jets [1] or sprays of
are expected to exceed 100 W/cm2 for many commercial applica- a dielectric liquid [2].
tions. Therefore, thermal management is becoming one of most Among cooling techniques for electronic devices, the use of
challenging issues and an important subject in regard to cooling thermoelectric coolers (TECs) combined with air-cooling or liquid
system performance. cooling approaches, is getting a significant amount of attention
Conventional air-cooled cooling systems are no longer adequate [3–7]. The thermoelectric module is a unique solid state heat pump
to remove the aforementioned heat fluxes. For a number of appli- because its direction of heat-pumping is fully reversible. When a
cations, direct air-cooling systems will have to be replaced or en- DC current flows through thermoelectric couples, one of the sur-
hanced by other high performance compact cooling techniques. faces will be cooled whilst the other is heated, depending on the
Liquid–vapor phase change, impinging jets, spray, direct and indi- polarity of the current. The amount of heat that can be removed
rect liquid cooling, the use of thermoelectric modules and heat by the hot side depends on the cooling load and the electrical
pipes are attractive cooling solutions for removing high heat fluxes power input as well as the efficiency of thermoelectric module. A
because of their high heat transfer coefficients. Several studies thermoelectric cooling system normally consists of three basic
have been conducted in these areas by many researchers. components: the thermoelectric module, the heat sink at the hot
Two-phase heat transfer, involving evaporation of a working side of the module and the cold sink component at the cold side
fluid in a hot area and condensation of vapor in a cold area, can of the module. Any thermoelectric module has a maximum tem-
perature difference between the hot side and the cold side,
⇑ Corresponding author. depending on number of junctions and stages of the thermoelectric
E-mail address: [email protected] (N. Putra). module. To have an optimal temperature difference, which is the

0894-1777/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermflusci.2011.04.015
 N. Putra et al. / Experimental Thermal and Fluid Science 35 (2011) 1274–1281 1275

key factor for thermoelectric module cooling systems, waste heat nanofluids has a higher cooling performance than water, the clas-
from the hot side must be released as efficiently as possible into sical coolant without nanoparticles.
the atmosphere, either by natural or forced convection [4]. Huang Heat pipes and thermosyphons [11–14] are examples of indi-
et al. [5] investigated the thermal performance of a thermoelectric rect liquid cooling methods with two-phase heat transfer. These
water-cooling device for electronic equipment. They developed a methods can transport large heat rates with small temperature dif-
novel analytical model of a thermal analogy network to predict ferences. Kim et al. [11] proposed cooling using heat pipes, and
the thermal capability of the thermoelectric device. Naphon and Chang et al. [12] experimentally investigated the thermal perfor-
Wiriyasart [6] studied liquid cooling in a mini-rectangular fin heat mance of the heat pipe cooling system for electronic equipment.
sink on a CPU with and without a thermoelectric device. They Many researchers have also attempted to enhance cooling
found that the thermoelectric device had a significant effect on performance for electronic equipment by combining cooling tech-
the CPU’s cooling performance of the PC. Putra et al. [7] applied niques. Agwu Nnanna et al. [15] researched the thermal augmenta-
multi stage thermoelectric module for cryosurgery device. They tion of a thermoelectric module using a nanofluid-based heat
investigated experimentally under variations of parameters and exchanger. Based on their experimental evidence, they concluded
heat exchanger configurations in order to obtain the modules’ that nanofluid has the potential to improving the thermal contact
characterizations. The multi-stage TEC modules prototypes have resistance between a thermoelectric module, heat source and sink.
shown good results as a cooling source for the cryosurgery cooling Several researchers have also used nanofluid filled thermosyphons
system. and heat pipes [16–21]. Tsai et al. [20] studied the effect of struc-
Since Choi [8] in 1995 introduced nanofluid as an attractive heat tural character of nanoparticles on the thermal performance of a
transfer coolant, additional research has been performed on nano- heat pipe, and the results showed that the thermal resistance of
fluids for the thermal management of electronic equipment cooling the heat pipes with nanofluids was lower than that of base fluid.
systems. For example, Nguyen et al. [9] experimentally investi- Furthermore, Kang et al. [21,22] used silver nanofluid as the work-
gated the behavior and heat transfer enhancement of a particular ing fluid inside a heat pipe, and the results indicated that the ther-
nanofluid, Al2O3 nanoparticles–water mixture, for cooling micro- mal resistance of heat pipe decreased as the silver nanoparticle size
processors. The results of their experiment showed that the heat and concentration increased.
transfer coefficient increased by as much as 40% compared to that Other comprehensive experimental data do not appear to exist
of the base fluid. Jang and Choi [10] numerically studied the cool- for the combined application of thermoelectric and commercial
ing performance of a microchannel heat sink with nanoparticles– heat pipes using nanofluids as the coolant to condense vapor in a
fluid suspension. They showed that a microchannel heat sink with cold area of the heat pipe. Therefore, this study experimentally

Circulating
Thermostatic Bath Reservoir
(CTB) Pump

Differential
Pressure Transmiter

Flow
Heat Indicator
Exchanger
Motherboard
Gigabyte GA-M55SLI-S4
v2.0
Pressure

Outlet
Thermocouple
Processor Liquid Block
AMD Athlon 64 with
CPU
Thermocouple X2 6000 Heat Pipe

cDAQ 9174 NI 9203 Inlet CPU Heat

Thermocouple Exchanger
Type 1

Thermoelectric
2 Stage x 4
Lab View CPU Heat Exchanger
NI 9213 Type 2

Fig. 1. Experimental setup.

1276 N. Putra et al. / Experimental Thermal and Fluid Science 35 (2011) 1274–1281

investigates the thermal performance of nanofluid as the working The flow rates of the cooling water are controlled by adjusting the
fluid on heat pipes and thermoelectric systems. The results of the valve and are measured using the flow meter. The commercial sin-
thermal performances are with pure water and other conventional tered heat pipe liquid-block used was made of copper, and the de-
cooling systems. tails are shown in Fig. 2.
In this study, the working fluids were water and nanofluids. The
results obtained from pure water were used as a basis for compar-
2. Experimental setup
ison. The nanofluids used in this study were composed of a Al2O3–
H2O and TiO2–H2O mixture. The nanometer-sized particles of alu-
A schematic diagram of the experimental setup is shown in
mina (Al2O3) and titania (TiO2), with an average particle size of
Fig. 1. The experimental setup consists of a PC, the closed-loop
about 13 nm for alumina and 21 nm for titania, were purchased
coolant for pure water and nanofluid, and a data acquisition sys-
in the form of dry powders. The particles were then dispersed by
tem. The piping system is designed so that parts can be changed
mixing appropriate quantities of nanoparticles with the base fluid
or repaired easily. The closed-loop of the coolant (pure water and
in a container and dispersing them in an ultrasonic vibration bath
nanofluid) consists of a 10 3 m3 storage tank, a water pump, a flow
for at least 2 h to form the well dispersed alumina–water and tita-
meter, a heat pipe liquid-block and a heat exchanger. The coolant is
nia–water nanofluid. A dispersion agent was not added to prevent
chilled in a heat exchanger which was immersed in a thermostatic
the particles from agglomerating in this synthesis process. Two
bath. This thermostatic bath had a variation of coolant tempera-
volumetric fractions of the alumina–water and titania–water
tures. After the temperatures of the coolant achieved the desired
nanofluid, 0.5 and 1 vol.%, were prepared for the experiment.
temperature, the coolant was pumped out from the storage tank,
The measured temperatures of the coolant at various positions
and was then passed through a flow meter to the heat pipe li-
are also shown in Fig. 1. The coolant was pumped into a commer-
quid-block. In the heat pipe liquid block, working fluid absorbed
cial sintered heat pipe liquid-block that was installed on top of
heat from the heat pipe liquid-block (type 1) and returned it to
the CPU on the PC. Five type-K thermocouples were installed at
the storage tank. A different configuration between the CPU and
various locations in the cooling system to monitor the fluid tem-
the liquid-block was used in which a thermoelectric system was
peratures and the CPU temperature. T1 and T2 were mounted at
placed. The thermoelectric system serves as heat pump to absorb
the inlet and outlet of the heat pipe liquid-block respectively,
heat from the CPU and transfers it to the liquid-block. The thermo-
T3 was mounted at the thermostatic bath and T4 was precisely
electric system is shown in Fig. 1 as a CPU heat exchanger (type 2).

Liquid Block Specification

Liquid Volume Aprox 117 cm3

Heat Pipe Diameter 6 mm

Heat Pipe Height 55 mm

Heat Pipe Material Copper

Base Plate Material Copper

Dimension 82(W) x 106(L) x 70(H) mm

Weight 557 gr

Tubing Connector 6,4 mm / (1/4”) Diameter

Fig. 2. Heat pipe liquid-block.

Fig. 3. Two types of conventional cooling modules: (a) heat sink fan (HSF), (b) heat pipe heat sink fan (HP HSF).
 N. Putra et al. / Experimental Thermal and Fluid Science 35 (2011) 1274–1281 1277

located between the CPU and the heat pipe liquid-block. In order Transmitter (Omega, Mode PX273) was installed on the heat pipe
to minimize thermal contact resistance between the CPU and the liquid-block. Data collection was carried out using a data acquisi-
heat pipe liquid-block, a thin film of high thermal conductivity tion system (NI cDAQ-9174, NI 9213 and NI 9203). Experiments
grease was applied at their junction interface. Two other Type-K were conducted using various coolant temperatures, volume con-
thermocouples were also used to monitor the fluid temperature centrations of nanofluids, and running conditions of the PC. The
in the reservoir and the ambient air temperature. All of the ther- supplied load into the CPU was adjusted to achieve the desired le-
mocouples used were initially calibrated using a constant temper- vel by setting the PC at a constant running condition. To deter-
ature water bath, and their accuracy has been estimated to be mine how effective using a nanofluid in an electronic cooling
at ± 0.05 K. The inlet temperature of the coolant prior to entering system is compared to a conventional cooling system, the con-
the cooling section was kept constant using a thermostatic bath. ventional cooling system, such as the heat sink fan and the heat
The temperatures at each position were recorded over a 40 min pipe heat sink fan, are examined in this experiment. Fig. 3 shows
period. To determine the pressure drop between the inlet and the heat sink fan and the heat pipe heat sink fan from the original
the outlet of the heat pipe liquid-block, a Differential Pressure Advanced Micro Devices (AMD) CPU retail box.

CTB 25oC
CTB 30oC
CTB 35oC
Temperature [oC]

Time [min]

Fig. 4. Liquid-block characteristic using H2O for load and no load (CTB at 30 °C).
Temperature [oC]

Time [min]

Fig. 5. Effect of volume concentration of nanofluids on CPU temperature of PC for no load and full load condition (CTB temperature is set at 30 °C).
1278 N. Putra et al. / Experimental Thermal and Fluid Science 35 (2011) 1274–1281

3. Results and discussion the CPU versus time for no load and full load conditions are
depicted in Fig. 4, where water was circulated as coolant. The
Each experiment was conducted at a room temperature of 29– graph shows that the CPU’s temperatures for full load conditions
30 °C for 40 min. At first, the CPU ran without any load for 10 min. are higher than those at no load, for the whole CTB temperature
A full load was then supplied to the CPU for 20 min, after which it range. The average CPU temperature increased rapidly by 7 °C after
was returned to no load. The coolant in the heat pipe liquid-block the full load was added to the CPU for all variation temperatures of
was cooled using a thermostatic bath (CTB) at 25, 30 and 35 °C. the CTB. This temperature increase occurred because the full load
Consequently, the inlet temperature of coolant before entering condition generated a higher heat flux than the no load condition.
the heat pipe liquid-block remained nearly constant for each tem- Fig. 5 shows the effect of the volume concentration of nanofl-
perature. The heat produced by the CPU was absorbed by the evap- uids on the CPU’s temperature of the PC for no load and full load
orator side of the heat pipe and then transferred to the condenser conditions. As shown in Fig. 5, the average temperature of the
side of the heat pipe. Finally, it was dissipated to the liquid-block CPU using pure water as the coolant on the heat pipe liquid-block
by the coolant (water or nanofluids). The temperature profiles of was 39.7 °C. After adding a small amount of nano-particles into the

Al2O3, CTB 25oC

TiO2, CTB 25oC
Al2O3, CTB 30oC
TiO2, CTB 30oC
Al2O3, CTB 35oC
TiO2, CTB 35oC
Temperature [oC]

Time [min]

Fig. 6. Effect of coolant temperature (nanofluids 1%) on the CPU temperature for different load conditions of the CPU.

HSF
HSF + Heatpipe
Al2O3 1%
Al2O3 0,5%
TiO2 1%
TiO2 0,5%
Temperature [oC]

H 2O

Time [min]

Fig. 7. The CPU Temperature obtained from the heat pipe liquid-block compared to those from the AMD stock heatsink fan (with and without the heat pipe). Note: Ambient
temperature is 30 °C for HSF and CTB is 30 °C for liquid-block.
 N. Putra et al. / Experimental Thermal and Fluid Science 35 (2011) 1274–1281 1279

pure water, the average temperature of the CPU decreased below The effects of coolant temperature on the CPU’s temperatures
that of the heat pipe liquid-block with pure water. The tempera- for different load conditions of the CPU are observed for nanofluids
tures were 38.5 °C and 37.8 °C for alumina–water nanofluids, having a 1% volume concentration with a variation of temperatures
38.3 °C and 39.2 for titania–water nanofluids, with volume concen- of 25 °C, 30 °C and 35 °C. Fig. 6 illustrates the CPU’s temperatures
trations of 0.5% and 1% respectively. The heat transfer rate from the for different load conditions on the CPU. For the three different
CPU to the heat pipe liquid-block increased due to the higher heat coolant temperatures, a larger CPU temperature drop was found
transfer surface area for a higher volume concentration. Therefore, for a lower coolant temperature. A lower coolant temperature re-
the CPU temperatures obtained from the heat pipe liquid-block sults in a lower capacity resistance and consequently a lower over-
with a higher volume concentration are lower than those from all thermal resistance.
the heat pipe liquid-block with a lower volume concentration, Fig. 7 shows the CPU temperatures obtained from the heat pipe
especially under full load conditions. At all conditions, the temper- liquid-block compared to those from the other cooling techniques
ature obtained from the heat pipe liquid-block with nanofluid was such as heat sink fan (HSF) and heat pipe with heat sink fan (HP
lower than the heat pipe liquid-block with pure water. HSF). It can be seen from the figure that the heat pipe liquid-block
Temperature [oC]

Time [min]

Fig. 8. The CPU temperature obtained from the heat pipe liquid combined with the thermoelectric cooling system (CTB at 30 °C).
Temperature [oC]

Al2O3 1%

Al2O3 1% + TEC

TiO2 1% + TEC

H2O + TEC

Time [min]

Fig. 9. The CPU temperature obtained from the heat pipe liquid-block with and without thermoelectric cooling system (CTB at 30 °C).
1280 N. Putra et al. / Experimental Thermal and Fluid Science 35 (2011) 1274–1281

Δ Pressure [psi]

Time [min]

Fig. 10. Pressure difference between the inlet and outlet from heat pipe liquid-block under several variations of cooling system conditions.

gives a lower CPU temperature than the other systems for all con- higher thermal performance. The reason for thermal enhancement
ditions. The heat pipe liquid-block system can reduce to 20 °C if it of nanofluid can be explained as follows. Using the presented
is compared to the system with HSF and only 7 °C if it is compared nanofluid theories, mainly those correlated to forced convection,
to HP HSF. The heat pipe liquid-block system has a significant ef- nano-particles can compress the transverse temperature gradient
fect on the CPU cooling system. of the fluid, which is mostly due to increased movement of liquid
Thermoelectric (TEC) can be used to achieve a temperature be- atoms in the presence of nano-particles. Hence, nanofluids reduce
low ambient. Thermoelectric can help heat dissipation process both the thermal resistance and the temperature difference be-
more quickly. Fig. 8 shows the temperature profile of a CPU when tween the heated wall and the coolant.
the combination cooling system between the thermoelectric and
the heat pipe liquid-block is used with water and nanofluids as Acknowledgment
the working fluid. The lowest temperature can be achieved using
1% alumina–water nanofluids as a working fluid. The temperature The Authors thank the Directorate on Research and the Commu-
of the CPU decreased to 4–6 °C below the ambient temperature. nity Service University of Indonesia for financial support.
Moreover, Fig. 9 shows the comparison between the cooling sys-
tem with and without the thermoelectric. The CPU temperature
decreased considerably with the addition of the thermoelectric References
on the cooling system. Temperatures of CPU are approximately
[1] S.W. Chang, S.F. Chiou, S.F. Chang, Heat transfer of impinging jet array over
26 °C, 24.2 °C, and 23.9 °C, respectively for water, and TiO2, Al2O3 concave-dimpled surface with applications to cooling of electronic chipsets,
nanofluids have a 1% particle concentration. This CPU temperature Experimental Thermal and Fluid Science 31 (7) (2007) 625–640.
decrease appears to be almost constant for the range of loads con- [2] J. Kim, Spray cooling heat transfer: the state of the art, International Journal of
Heat and Fluid Flow 28 (4) (2007) 753–767.
sidered. A better CPU cooling system can provide an optimal CPU [3] R. Chein, G. Huang, Thermoelectric cooler application in electronic cooling,
performance and a longer CPU life. Applied Thermal Engineering 24 (14–15) (2004) 2207–2217.
The pressure difference between the inlet and the outlet of the [4] N. Putra, Design, manufacturing and testing of a portable vaccine carrier box
employing thermoelectric module and heat pipe, Journal of Medical
liquid-block is shown in Fig. 10. The graph shows less than a Engineering and Technology 33 (3) (2009) 232–237.
0.01 psi pressure difference between the inlet and the outlet dur- [5] H.S. Huang, Y.C. Weng, Y.W. Chang, S.L. Chen, M.T. Ke, Thermoelectric water-
ing the process. The results reveal that there is no significant pres- cooling device applied to electronic equipment, International Communications
in Heat and Mass Transfer 37 (2) (2010) 140–146.
sure difference effect when water is used, and a low concentration [6] P. Naphon, S. Wiriyasart, Liquid cooling in the mini-rectangular fin heat sink
over a range of ambient temperatures of nanofluid is present. with and without thermoelectric for CPU, International Communications in
Heat and Mass Transfer 36 (2) (2009) 166–171.
[7] N. Putra, Ardiyansyah, W. Sukyono, D. Johansen, F.N. Iskandar, The
4. Conclusion characterization of a cascade thermoelectric cooler in a cryosurgery device,
Cryogenics 50 (11–12) (2010) 759–764.
[8] U.S. Choi, Enhancing thermal conductivity of fluids with nanoparticles, in: D.A.
A heat pipe liquid-block combined with a thermoelectric sys- Siginer, H.P. Wang (Eds.), Developments and applications of non-Newtonian
tem was investigated as an option for CPU cooling. Alumina–water flows, FED-vol. 231/MD-vol. 66, ASME, New York, 1995, pp. 99–105.
and titania–water nanofluids were used as the working fluid. The [9] C.T. Nguyen, G. Roy, C. Gauthier, N. Galanis, Heat transfer enhancement using
Al2 O3 –water nanofluid for an electronic liquid cooling system, Applied
effects of the nanofluid volume concentration, coolant temperature
Thermal Engineering 27 (8–9) (2007) 1501–1506.
and the running conditions of a PC on the CPU’s temperature were [10] S.P. Jang, S.U.S. Choi, Cooling performance of a microchannel heat sink with
considered. Compared to the other conventional cooling tech- nanofluids, Applied Thermal Engineering 26 (17–18) (2006) 2457–2463.
niques, the combined cooling system with nanofluids allows for a [11] K.S. Kim, M.H. Won, J.W. Kim, B.J. Back, Heat pipe cooling technology for
desktop PC CPU, Applied Thermal Engineering 23 (9) (2003) 1137–1144.
greater decrease in CPU temperature. Using nanofluids with the [12] Y.W. Chang, C.H. Cheng, J.C. Wang, S.L. Chen, Heat pipe for cooling of electronic
heat pipe liquid-block and thermoelectric cooler allowed for a equipment, Energy Conversion and Management 49 (11) (2008) 3398–3404.
 N. Putra et al. / Experimental Thermal and Fluid Science 35 (2011) 1274–1281 1281

[13] L. Vasiliev, D. Lossouarn, C. Romestant, A. Alexandre, Y. Bertin, Y. Piatsiushyk, [18] S.H. Noie, S.Z. Heris, M. Kahani, S.M. Nowee, Heat transfer enhancement using
V. Romanenkov, Loop heat pipe for cooling of high-power electronic Al2O3/water nanofluid in a two-phase closed thermosyphon, International
components, International Journal of Heat and Mass Transfer 52 (1–2) Journal of Heat and Fluid Flow 30 (4) (2009) 700–705.
(2009) 301–308. [19] J. Qu, H. Wu, P. Cheng, Thermal performance of an oscillating heat pipe with
[14] K.H. Do, S.P. Jang, Effect of nanofluids on the thermal performance of a flat Al2O3–water nanofluids, International Communications in Heat and Mass
micro heat pipe with a rectangular grooved wick, International Journal of Heat Transfer 37 (2) (2010) 111–115.
and Mass Transfer 53 (9–10) (2010) 2183–2192. [20] C.Y. Tsai, H.T. Chien, P.P. Ding, B. Chan, T.Y. Luh, P.H. Chen, Effect of structural
[15] A.G. Agwu Nnanna, W. Rutherford, W. Elomar, B. Sankowski, Assessment of character of gold nanoparticles in nanofluid on heat pipe thermal performance,
thermoelectric module with nanofluid heat exchanger, Applied Thermal Materials Letters 58 (9) (2004) 1461–1465.
Engineering 29 (2–3) (2009) 491–500. [21] S.W. Kang, W.C. Wei, S.H. Tsai, C.C. Huang, Experimental investigation of
[16] S. Khandekar, Y.M. Joshi, B. Mehta, Thermal performance of closed two-phase nanofluids on sintered heat pipe thermal performance, Applied Thermal
thermosyphon using nanofluids, International Journal of Thermal Sciences 47 Engineering 29 (5–6) (2009) 973–979.
(6) (2008) 659–667. [22] S.W. Kang, W.C. Wei, S.H. Tsai, S.Y. Yang, Experimental investigation of silver
[17] M. Shafahi, V. Bianco, K. Vafai, O. Manca, An investigation of the thermal nano-fluid on heat pipe thermal performance, Applied Thermal Engineering 26
performance of cylindrical heat pipes using nanofluids, International Journal of (17–18) (2006) 2377–2382.
Heat and Mass Transfer 53 (1–3) (2010) 376–383.