International Journal of Heat and Mass Transfer 46 (2003) 851–862

www.elsevier.com/locate/ijhmt

Pool boiling characteristics of nano-fluids

Sarit K. Das a, Nandy Putra b, Wilfried Roetzel b,*

a
Heat Transfer and Thermal Power Lab, Mechanical Engineering Department, Indian Institute of Technology, Madras 600036, India
b
Institut f€ur Thermodynamik, Universit€at der Bundeswehr Hamburg, D-22039 Hamburg, Germany
Received 19 June 2001; received in revised form 23 July 2002

Abstract
Common fluids with particles of the order of nanometers in size are termed as Ônano-fluidsÕ which have created
considerable interest in recent times for their improved heat transfer capabilities. With very small volume fraction of
such particles the thermal conductivity and convective heat transfer capability of these suspensions are significantly
enhanced without the problems encountered in common slurries such as clogging, erosion, sedimentation and increase
in pressure drop. This naturally brings out the question whether such fluids can be used for two phase applications or in
other words phase change in such suspensions will be assistant or detrimental to the process of heat transfer. The
present paper investigates into this question through experimental study of pool boiling in water–Al2 O3 nano-fluids.
The results indicate that the nano-particles have pronounced and significant influence on the boiling process deterio-
rating the boiling characteristics of the fluid. It has been observed that with increasing particle concentration, the
degradation in boiling performance takes place which increases the heating surface temperature. This indicates that the
role of transient conduction in pool boiling is overshadowed by some other effect. Since the particles under consid-
eration are one to two orders of magnitude smaller than the surface roughness it was concluded that the change of
surface characteristics during boiling due to trapped particles on the surface is the cause for the shift of the boiling
characteristics in the negative direction. The results serve as a guidance for the design of cooling systems with nano-
fluids where an overheating may occur if saturation temperature is attained. It also indicates the possibility of such
engineered fluids to be used in material processing or heat treatment applications where a higher pre-assigned surface
temperature is required to be maintained without changing the fluid temperature.
Ó 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Boiling; Experimental; Heat transfer

1. Introduction operations per second are becoming quite common.

These devices are not only to operate in their respective
Heat transfer technology stands at cross roads today applications with high precision but also to do so oc-
with ever increasing demand of cooling ultra high heat cupying minimum space. This puts a challenge not only
flux equipment on one hand and unprecedented pace of to the core device design but also to their thermal
miniaturisation on the other. In the present day tech- management. While air based cooling systems are more
nology the different ranges of LASER applications, common and reliable, they fail miserably with increasing
super conducting magnets, high power X-ray and above heat flux. Therefore in almost all the high heat flux ap-
all super fast computing chips performing trillions of plications liquid cooling is preferred. The cooling liquids
usually used are water/chilled water, common refriger-
ants and liquid nitrogen or similar cryogens depending
*
Corresponding author. Tel.: +49-40-6541-2624; fax: +49- on the specific application. While water is a convenient
40-6541-2005. and safer medium, its relatively poor heat transfer
E-mail address: [email protected] (W. characteristic is a major disadvantage. Usual refriger-
Roetzel). ants are hazardous to environment and cryogens are
0017-9310/03/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 7 - 9 3 1 0 ( 0 2 ) 0 0 3 4 8 - 4
852 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862

Nomenclature

D diameter (m) Greek symbols

g acceleration due to gravity (m/s2 ) a heat transfer coefficient (W/(m2 K))
h latent heat for phase change (J/kg) k thermal conductivity (W/(m K))
L length (m) l viscosity (kg/(ms))
Nu Nuelt number r surface tension (N/m)
Pr Prandtl number q density (kg/m3 )
q heat flux (W/m2 )
Subscripts
Ra roughness parameter (lm)
f fluid
Rq roughness parameter (lm)
g vapour
Reb boiling Reynolds number
x position (mm)
z orthogonal position (lm)

costly not only due to their energy intensive production erage size <10 nm. All the above works [3–6] indicate
process but also due to whole range of costly equipment that usual theory of suspensions and slurries such as the
to be deployed to use them. classical Maxwell [7] model or the extended Hamilton–
Under the circumstances it makes sense to look at Crosser [8] or Wasp [9] model fail miserably with nano-
alternatives such as fluids with suspended solid particles fluids. An evident theory is still missing.
in them. Though the enhancement of thermal conduc- However for heat transfer engineer, this enhancement
tivity of slurries is known for more than a century, they of thermal conductivity is only a necessary condition for
have not been considered as a candidate for heat transfer using such fluids in cooling application and not a suffi-
applications in the past due to problems associated with cient condition. The real worth of such fluids can only be
them such as sedimentation, erosion, fouling and in- tested under convective conditions. The study of Ahuja
creased pressure drop. The recent advancement of ma- [10], Liu et al. [11] and Sohn and Chen [12] conclusively
terials technology [1] has been able to produce particles indicate that performance of slurries under convective
of nanometer size which when suspended in usual fluid conditions are encouraging. Eastman et al. [13] also
can overcome most of the problems encountered by mention an increase of 40% in heat transfer capability
common slurries. Choi [2] was the first one to call such for nano-fluids with 2% particle concentration under
suspension Ônano-fluidsÕ which is now widely accepted. convective conditions even though no systematic study
The stability of such fluids against sedimentation is re- is available in this regard. The theoretical observations
markably improved when very small amount of stabil- and proposition of dispersion model by Xuan and
ising agent such as laurate salt [3] is added. The erosion Roetzel [14] goes a long way in theoretical modelling
and pressure drop problems are also greatly reduced due of nano-fluids under convective condition the experi-
to small particles and the small volume fraction (usually mental validation of which is underway by the present
<5%) required for enhancement of thermal behaviour of authors.
the base fluid. A substantial enhancement of thermal While using nano-fluids for convective cooling, one
conductivity of water and ethylene glycol based nano- must be aware of its boiling characteristics. This is be-
fluids with Al2 O3 or CuO nano-particles ranging from cause even if nano-fluids are unattractive with respect to
7% to 30% with only 1–5% particle volume fraction was two (or rather three) phase applications, during con-
reported by Lee et al. [4] at room temperature. A recent vective heat transfer with high heat flux locally boiling
study by the present authors [5] shows that the en- limit may be reached. It is important that the behaviour
hancement of thermal conductivity of nano-fluids in- of nano-fluid under such conditions is accurately known
creases dramatically with temperature making it more to avoid unwanted effects such as local hot spot which
attractive for cooling at high temperature and heat flux. can cause significant deterioration of reliability of
Further enhancement of thermal conductivity of nano- components to be cooled.
fluids with pure metallic particles was reported by Xuan The present paper is aimed at an experimental study
and Li [3] who found enhancement comparable to Lee of pool boiling characteristics of water–Al2 O3 nano-fluid
et al. [4] using much bigger (100 nm) particles size. Finally, under atmospheric conditions. The thrust of the experi-
the enhancement of thermal property received a quan- ment is to compare the pool boiling parameters with
tum jump when Eastman et al. [6] reported an increase that of pure water and thus bring out the applications
of thermal conductivity by an outstanding 40% with and limitations of nano-fluids under the condition of
only 0.3 vol.% of nano-particles of copper having av- phase change.
 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862 853

2. Production and characterisation of nano-fluid

Depending on base fluid and particle combination a

number of nano-fluids can be produced. However in the
present investigation, water based nano-fluids of Al2 O3
particles have been used. The reason for this is the fact
that the boiling characteristics of the base fluid water is
most widely known and the thermal conductivity of
water–Al2 O3 nano-fluids for different particle concen-
tration and the effect of temperature on it has already
been studied [5]. Even though CuO–water nano-fluids
have shown better thermal properties, they have not
been used because CuO nano-particles have explosive
characteristics at temperature more than 100 °C with
moist air and they produced considerable fouling. The Fig. 2. TEM of dispersed nano-aluminium oxide powder.
particles were produced (by Nanophase Technologies
Corporation) by physical vapour deposition technique.
In the powder state they form loose agglomerates of
micrometer size as shown by transmission electron mi-
croscopy (TEM) photo given in Fig. 1. However it has
been observed that the agglomerates break down to a
considerable extent to produce smaller size particles and
agglomerates when dispersed in water. The dispersion
was carried out by using ultrasonic vibration for 4 h
after making a suspension of desired concentration. Fig.
2 shows the TEM picture of dispersed particles by dry-
ing a dilute suspension on silicon wafers. The volume
weighted average particle size was found to be 38 nm.
However analysis showed a wide distribution of particles
as typically shown in Fig. 3. It was found that the re-
sulting suspension is quite stable at lower (1% and 2%)
volume concentration and at higher concentration (3% Fig. 3. Particle size distribution of the nano-powder.
and 4%) very nominal amount of sedimentation occurs
after a time (6 h) long enough compared to the duration
of the experiment. No effort was made to stabilise the present study. However it was observed that due to
solution because stabilising agents such as laurate salts turbulent mixing of the fluid by bubble movements no
are known to be surfactants and themselves have con- sedimentation takes place even for 4% volume concen-
siderable effect on the boiling process [15] by reducing tration of particles during pool boiling.
surface tension which may offset the main focus of the To characterise the produced nano-fluid, first rhe-
ological study was made. The measurement with a disc
type rotating rheometer showed that shear rate-defor-
mation characteristics are almost linear and similar to
water measured by the same instruments as given in Fig.
4. This confirms a Newtonian behaviour of the fluid
between 1% and 4% particle (volume) concentration.
Fig. 5 shows the values of viscosity of the fluids which
are fond to be constant against shear rate but higher
compared to water. Thus, it can be said that this par-
ticular nano-fluid shows an increase of viscosity with
particle concentration but remains Newtonian in nature.
The enhancement of thermal conductivity with particle
concentration and temperature as measured by temper-
ature oscillation technique [5] is given in Fig. 6. Since
surface tension is known to be an important parameter
in pool boiling, it was also measured for the present fluid
Fig. 1. TEM of agglomerated nano-aluminium oxide powder. using the conventional ring method. Fig. 7 shows the
854 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862

Fig. 4. Rheological behaviour of nano-fluids at 1% and 4% Fig. 5. Dynamic viscosity of nano-fluids and pure water at
concentrations and pure water. different temperatures.

variation of surface tension with particle concentration vent to noncondensable gases. The cooling water from
which is extremely nominal to have any surfactant effect this vertical condenser is then circulated through an oval
on the boiling process. shaped copper coil (4) which performs the task of con-
densing the bulk of the vapour produced. This coil hangs
from the roof of the vessel and is designed to cool the
3. Pool boiling experiment entire vapour at the maximum rate of evaporation. At
the same time it is kept at sufficient distance from the top
3.1. Experimental set-up surface of the boiling liquid to avoid direct cooling which
may result subcooled boiling. A pressure gauge (5)
The experimental set-up was designed keeping in mounted at the top of the vessel checks the pressure at
mind the parameters the effects for which are required to which boiling takes place. As boiling surface, a cylin-
be observed. For this reason no effort has been made to drical cartridge heater (6) of 20 mm diameter and max-
fabricate a so-called standard boiling apparatus but imum 420 V, 2.5 kW rating is used. The voltage is
watch has been kept so that the experiments for different regulated at appropriate values to reach prescribed heat
nano-fluids and water are performed under identical fluxes. It is inserted from the side wall. To observe the
conditions. The test section is shown in Fig. 8. It consists boiling characteristics during water experiments, round
of 120 mm
100 mm
200 mm rectangular stainless windows (7), with double glass (inner 8 mm and outer 6
steel vessel (1) with thick insulation (2) outside. The mm thickness) was built on both the side walls. A
vessel has two cooling arrangements cascaded together. sheathed 0.5 mm thick chromel–alumel (K-type) ther-
The first one (3) is a counter current copper condenser mocouple (8) was inserted to observe the bulk liquid
which on one hand connects the vessel directly to at- temperature during boiling. To measure temperature on
mosphere maintaining an atmospheric pressure in it, on the heating cartridge 10 K type thermocouples of 0.1 mm
the other hand it also serves the purpose of after cooling thickness were welded at different radial and axial loca-
of any vapour which may try to escape as well as act as a tions as indicated in Fig. 9. The radial locations are a, b, c
 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862 855

Fig. 8. The experimental cell for pool boiling.

Fig. 9. The cartridge heater with thermocouple locations.

Fig. 6. Thermal conductivity enhancement of nano-fluids as a
function of temperature.
and the power was recorded with a Wattmeter. The
thermocouples, pressure gauge and Wattmeter were
connected to a data logger which was in turn connected
to a PC for recording and storage of data.
For characterising pool boiling phenomenon it is
important to know the heater geometry and surface
accurately. The heater surface is machine drawn. The
surface characteristics of the heater was measured using
a profilometer having a diameter tip of 2 lm and a
sensitivity of 0.02 lm. The major parameter for char-
acterisation of surface roughness are Ra and Rq (DIN
4762) which are defined as
Z
1 L
Ra ¼ jZðxÞj dx ð1Þ
L 0
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z
1 L
Rq ¼ ZðxÞ2 dx ð2Þ
Fig. 7. Surface tension of nano-fluids. L 0

A typical surface profile for the original heater is shown

and d and axial locations are 0, 1, 2, 3 and 4. The ther- in Fig. 10. This along with other parameters characte-
mocouples were planted at locations 0d, 1a, 1b, 2a, 2b, rising surface roughness were measured over a number
2c, 2d, 3c, 3d and 4a. The leads of the thermocouples of sample lengths. The heater surface for another set of
were taken out from tip of the heater in a bunch (9). The heaters were roughened with emery paper and used for a
power supply to the heater was varied by a transformer second set of experiments. The typical measured values
856 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862

Fig. 11. Surface roughness of the roughened heater.

Fig. 10. Surface roughness of the smoother heater.

available in literature for pool boiling on horizontal

Table 1 tubes. It was also done to check the repeatability of the
Roughness of the smooth heater experiment and to visually observe the boiling charac-
No. Ra Rq teristics and ensure that the surface provides a fairly
1 0.42 0.52 uniform density of nucleation sites and no local surface
2 0.36 0.48 defect dominates the boiling process.
3 0.46 0.60 During these experiments check on the pressure was
4 0.46 0.60 kept with a pressure gauge of 0.002 bar accuracy.
5 0.42 0.52 During this pre-experiment with water and the actual
6 0.42 0.54 experiment, temperatures were recorded at a number of
7 0.44 0.56 thermocouple locations and an average was taken to
8 0.42 0.54
designate the wall temperature. The thermocouple
measuring the fluid temperature was moved from the
fluid surface to the bottom level of the heater to ensure
Table 2 that the true bulk temperature of the liquid is measured.
Roughness of the roughened heater The projected end of the heater was insulated to
No. Ra Rq avoid boiling from the end surface. Before inserting, an
1 1.15 1.42 infrared camera was used to check the uniformity of
2 1.14 1.44 heating by switching on the heating keeping the heater in
3 0.988 1.19 air. Prior to each experiment, the liquid was boiled for
4 1.15 1.54 half an hour to drive out any dissolved gases. Experi-
5 1.12 1.47 ment was first performed with water from the lowest to
6 1.25 1.69 the highest power input and then it was carried out in
7 1.21 1.55
the reverse direction to eliminate the possibility of
8 1.15 1.52
measuring only one part in case hysteresis exists. At each
measuring condition stability of each thermocouple
reading was checked by recording the mean temperature
of roughness for the two surfaces are given in Tables 1 taken over a second at a sampling rate of 20 readings per
and 2 and typical surface profile of the roughened sur- second. During experiments with the nano-fluids, runs
face is shown in Fig. 11. with the same nano-fluid was repeated with a run of
boiling with pure water in between after cleaning the
3.2. Experimental procedure and error estimates heater surface with a water jet. This is to ensure that the
particles are not sticking to the heating surface to change
Even though no effort was made to build standard the surface characteristics which can readily be deter-
boiling apparatus, care was taken to ensure that the pool mined from the change in the boiling curve of water.
boiling experiment are performed under standard con- The thermocouples are soldered on the heater surface
ditions. A set of pre-experimental runs with doubly and have a diameter of 0.1 mm. The accuracy is 0.1 °C
distilled demineralised water were made to ensure that and that for fluid temperature measurement with 0.5
the experimental results are in line with similar results mm diameter (for better stability within boiling fluid) is
 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862 857

0.2 °C. The mean wall temperature is assumed to be the

arithmetic mean value of all 10 measuring points on the
surface. The Wattmeter used for recording the power
has got an accuracy of 10 W which is 1.25% at the lowest
heat flux. At the highest heat flux it changes to 0.4%
accuracy. Another error of the heat flux is caused by
heat conduction from the electrically heated cylinder
into the wall of the vessel. The vessel wall acts as a cir-
cular fin which transfers heat into the water in the vessel.
This heat flow is estimated to be below 1% of the elec-
trical heat input. The uncertainty in diameter measure-
ment was 0.05 mm. A systematic error analysis was
made to ascertain the measurement uncertainty which
turned out to be 4% for surface heat flux. The uncer-
tainty of mean temperature difference remains to be re-
stricted to a maximum of 4%. Thus the maximum
systematic error of the heat transfer coefficient is 8%, it
will have the same sign and order of magnitude for all
experiments. So the effect of nano-particles on heat
transfer is measured which much higher accuracy.

4. Results and discussion

Prior to the experiment the nano-fluids were sepa-

rately boiled to determine the boiling point. It was ob- Fig. 12. Preliminary experiments: comparison of boiling of
served that for nano-particle concentration from 0.1% to pure water in the present set-up with Cornwell–Houston cor-
4% a very nominal decrease of 0.4 °C (which is of the relation [18].
order of error in temperature measurement) in the
boiling point occurs at atmospheric pressure.
Through the observation window it was found that of the heat transfer surface, observed after four days and
for pure water at low heat flux, the flow transforms from two weeks, respectively.
natural convective to nucleate boiling and distinct bub- The above results clearly indicate that the present
bles start sliding as observed by a row of studies by apparatus yields established boiling characteristics on
Cornwell and co-workers [16–19]. As depicted in these horizontal tube where fluid property and boiling pa-
studies, the bubble sliding mechanism gains momentum rameters such as surface roughness and heater diameter
with increasing heat flux and at a high heat flux the are found to play the expected roles.
bubbles coalesce to form bigger bubbles as observed by Following the above validation, experiments were
Chun and Kang [20]. These observations show that carried out to evaluate pool boiling with nano-fluids of
qualitatively the boiling behaviour in the present 1%, 2% and 4 % Al2 O3 nano-particle concentration in
apparatus conforms to that reported in literature. To water. Due to opaque nature of the fluid, onset of nu-
evaluate these pre-experimental runs with water quan- cleate boiling could not be accurately determined but
titatively, the q  DT data of different runs for the from the surface agitation of the fluid it was ascertained
smoother surfaced heater (Fig. 10) is plotted in Fig. 12, to be not much different from pure water at the given
in terms of Nu and Reb . Nu ¼ aD=k is the usual Nuelt surface condition. Since no significant change of surface
number and Reb ¼ qD=ðhfg lf Þ is the boiling Reynolds tension with nano-particle concentration was observed,
number. Further the convective type correlation of no hysteresis as described in [22] was found to exist.
Cornwell and Houston [21] However it was observed that during increasing power
the boiling starts at a higher heat flux than nucleation
Nu ¼ 104Re0:67
b Pr
0:4
ð3Þ stops during decreasing power both for water and nano-
fluid. After a number of runs repeatable boiling char-
is plotted in the same figure (Fig. 12). The plot shows acteristics of nano-fluids were taken for which the
good agreement of Eq. (3) and the measurements with q  DT from the results are shown in Fig. 13. This
water and clean heat transfer surface giving a confidence clearly indicates that the boiling performance of the base
in the experimental set up and procedure. It also shows fluid (water) has been deteriorated with the addition of
that the heat transfer coefficient decreases due to fouling nano-particles since the boiling curves are shifted to the
858 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862

Fig. 13. Pool boiling characteristic of nano-fluids on the Fig. 14. Pool boiling characteristic of nano-fluids on the
smoother heater. roughened heater.

right. This means that without changing the boiling to be different in nature compared to the smoother
temperature the nano-fluid can cause harm to cooled heater. In this case a more drastic increase in wall su-
surface if boiling limit is reached because it will give a perheat was observed for nano-fluids up to 1% particle
higher wall superheat meaning a higher surface tem- concentration after which it seemed to slow down up to
perature compared to water at the same heat flux. It has 4% (measured range). This can be better understood
been observed that the shift of the curve to the right is from Fig. 15 where for pure water, 1% and 4% particle
not proportional to the particle concentration. For ex- concentration the boiling curves for two heaters are
ample in Fig. 13, for the case of smoother heater (sur- compared. Here all the curves are shifted towards left
face, Fig. 10), a considerable shift of the curve was for the roughened heater due to increase in surface
observed with only 0.1% particle concentration and roughness but the extent of shift for different particle
thereafter from 1% to 4% concentration a regular shift concentrations are different and are dependent on heat
of the curve was observed at lower heat fluxes. However flux.
at the upper part of the curves the difference between The present results are somewhat contrary to ex-
wall superheats for various particle concentration was pectation. Fig. 6 shows a substantial increase in thermal
fond to increase with increasing heat flux. This depicts a conductivity of fluid with nano-particles. This enhance-
regular but non linear tendency of deterioration of ment goes to as much as 60% at saturation temperature.
boiling character for nano-fluids with the increase in The surface tension and latent heat remains unaffected
particle concentration. In order to examine this deteri- and the only unfavourable change in fluid property due
oration under different heater surface conditions, the to presence of the particles is its increase in viscosity.
same experiments were repeated for the second (rough- Since fluid conduction in micro layer evaporation under
ened, surface in Fig. 11) heater and the q  DT charac- the bubble as well as in reformation of thermal bound-
teristics for this has been shown in Fig. 14. Here also ary layer at the nucleation site [23] plays a major role in
shift of the boiling curve to the right indicating a dete- heat transfer during pool boiling, with such a substantial
rioration of the boiling performance with particle con- increase in thermal conductivity, nano-fluids are ex-
centration was observed. However, the shift was found pected to enhance heat transfer characteristic during
 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862 859

Fig. 15. Effect of roughness on the pool boiling behaviour of

nano-fluids. Fig. 16. Nu–Re plots for nano-fluids on the smoother heater.

pool boiling. In case of pool boiling on horizontal tubes

at moderate heat flux, the series of works from Cornwell Heater 1
and co-workers [16–19] conclusively prove the impor- (smooth)
tance of sliding bubbles where again conduction plays Pure water Nu ¼ 97:9Re0:638
b Pr0:4
an important role. Thus both for stationary bubble de- 1% Al2 O3 Nu ¼ 78:84Re0:687
b Pr0:4
0:69 0:4
velopment and sliding bubble mechanism, the increase 2% Al2 O3 Nu ¼ 72:39Reb Pr
in thermal conductivity is expected to enhance heat 4% Al2 O3 Nu ¼ 67:56Re0:619
b Pr0:4
transfer during boiling which is just contrary to what has
Heater 2 (rough)
been observed in the present set of experiments. The
Pure water Nu ¼ 137Reb0:526 Pr0:4
fact, that the present increase of wall superheat in boil-
1% Al2 O3 Nu ¼ 99:48Re0:503
b Pr0:4
ing and as a consequence decrease in boiling heat
2% Al2 O3 Nu ¼ 94:63Re0:495
b Pr0:4
transfer coefficient is an additional effect, can be un- 0:490 0:4
4% Al2 O3 Nu ¼ 89:12Reb Pr
derstood from Figs. 16 and 17. Here in keeping with
Cornwell–Houston [21] type of correlation, Nu–Reb plot The lines indicating these correlations are shown in Figs.
for both the heaters have been presented. 16 and 17. For smoother heating surface the shift in the
It is evident that for each particle concentration the boiling character was found to be more or less uniform
Nu–Reb characteristics is different and shifted down- with concentration while for the roughened surface it
wards. This conclusively brings out the fact that the was found to be rapid at lower concentration and slower
change in boiling characteristics of nano-fluids cannot thereafter.
be explained in terms of property change alone because To find out which effect skirts the property variation,
the Nu–Reb correlations are altered. The following cor- the surface characteristics of the heaters were re-exam-
relations were obtained for the two heaters, by regres- ined after the runs with nano-fluids before jet cleaning of
sion analysis. According to Eq. (3) the exponent of Pr is the surfaces. It was found that a considerable reduction
taken as 0.4. The Pr number is formed with the actual in the surface roughness takes place which returns to
fluid properties of the nano-fluid. almost the original surface after cleaning. As an example
860 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862

Table 3
Roughness of the smooth heater after boiling
No. Ra Rq
1 0.281 0.371
2 0.264 0.337
3 0.288 0.377
4 0.358 0.439
5 0.358 0.444
6 0.347 0.445
7 0.291 0.372
8 0.299 0.377

tively uneven surface during boiling. These trapped

particles change the surface characteristics making it
smoother. This causes the degradation of the boiling
characteristics. For higher particle concentration, the
particles virtually form a layer on the heating surface
hindering the fluid flow and heat transfer. However due
to their extremely small size, they are almost completely
removable by water jet cleaning. Thus the small size of
the particles causes the surface skirting which over-
shadows the thermal conductivity enhancement of the
nano-fluids.
The large change in boiling character of the rough-
ened heater with smaller particle concentration can also
be explained in a similar way. In this case due to higher
surface roughness (1.15 lm) the cavities on the surface
Fig. 17. Nu–Re plots for nano-fluids on the roughened heater. are more and as a consequence the smoothening of the
surface by sitting particles is more abrupt. Thus with
smaller particle concentration (<1%) enough amount of
the surface characteristics of the smooth heater (shown in
particles are deposited on the uneven surface to con-
Fig. 10 and Table 1) was changed to even lower value
siderably affect the boiling character. Any additional
shown in Fig. 18 and Table 3 after boiling nano-fluid on
deposit of particles from higher concentration in the
it. This brings out the probable cause for the deteriora-
fluid brings only marginal deterioration of pool boiling
tion in boiling characteristics. Due to the fact that the
characteristics.
size of the nano-particles (20–50 nm) are one to two or-
ders of magnitude smaller than the roughness (0.2–1.2
lm) of the heating surface, the particles sit on the rela- 5. Conclusion

The modern electronics, computing and optical

technology has brought about a stream of equipment
dealing with extremely high heat flux needing more and
more cooling efficiency. In recent times nano-fluids have
been claimed to be a new possibility in meeting these
demands due to their enhanced thermal conductivity
and capability of further enhancing convective process
through particle dispersion. They have been found to be
much improved with respect to sedimentation, clogging
and pressure drop compared to common slurries. Even
though these indicate a possibility of enhancing such
fluids under phase change conditions, the present study
conclusively negates this possibility. Through a carefully
designed series of experiments it has been shown that
presence of nano-particles deteriorates boiling perfor-
Fig. 18. Surface roughness of the smoother heater after boiling mance systematically with increase in particles concen-
with nano-fluids (without surface cleaning). tration. This means with the increase in particles the wall
 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862 861

superheat is increased at a given heat flux. Using the of nano-fluids, in: Proceeding of the Symposium on
nano-fluid properties which were carefully obtained Nanophase and Nanocomposite Materials II, vol. 457,
during characterisation of the fluid, reduction of the Material Research Society, Boston, 1997, pp. 3–11.
experimental data to conventionally used in the form of [2] U.S. Choi, Enhancing thermal conductivity of fluids
dimensionless equation reveals that there is a definite with nanoparticles, in: D.A. Siginer, H.P. Wang, (Eds.),
Developments and applications of non-Newtonian glows,
additional effect which causes this deterioration of
FED-vol. 231/MD-vol. 66, ASME, New York, 1995, pp.
boiling characteristic which cannot merely be explained 99–105.
in terms of change of fluid properties in presence of [3] Y. Xuan, Q. Li, Heat transfer enhancement of nanofluids,
nano-particles. The additional effect which has been Int. J. Heat Fluid Flow 21 (2000) 58–64.
found to be at the root of this degradation in boiling [4] S. Lee, U.S. Choi, S. Li, J.A. Eastman, Measuring thermal
phenomenon is the change of surface characteristics conductivity of fluids containing oxide nanoparticles,
during boiling due to trapped particles on the surface ASME J. Heat Transfer 121 (1999) 280–289.
which are one to two orders of magnitude smaller than [5] S.K. Das, N. Putra, P. Thiesen, W. Roetzel, Temperature
the surface roughness. dependence of thermal conductivity enhancement for
On the practical aspects, the present study draws a nanofluids, J. Heat Transfer, Trans. ASME, accepted for
publication.
limit for application of nano-fluids and makes it clear
[6] J.A. Eastman, U.S. Choi, S. Li, W. Yu, L.J. Thompson,
that while designing convective cooling systems with Anomalously increased effective thermal conductivities of
nano-fluids, care should be taken so that local heat ethylene glycol-based nanofluids containing copper nano-
fluxes do not cause boiling which will leave the heated particles, Appl. Phys. Lett. 78 (6) (2001) 718–720.
surface temperature much above that expected for the [7] J.C. Maxwell, A Treatise on Electricity and Magne-
pure fluid. In future, research may also be directed to tism, 2nd ed., vol. 1, Clarendon Press, Oxford, UK, 1881,
find out whether such higher temperature of heated p. 435.
surface and its control by changing particle concentra- [8] R.L. Hamilton, O.K. Crosser, Thermal conductivity of
tion (and of course other parameters such as particle heterogeneous two component systems, I & EC Fundam.
1 (3) (1962) 187–191.
fluid combination and particle size) can be used to
[9] F.J. Wasp, Solid–liquid slurry pipeline transportation,
produce engineered fluids to attained desired tempera-
Trans. Tech. Berlin. (1977).
ture of a surface in applications such as heat treatment [10] A.S. Ahuja, Augmentation of heat transport in laminar
or material processing. flow of polystyrene suspension. Experiments and results,
J. Appl. Phys. 46 (8) (1975) 3408–3416.
[11] K.V. Liu, U.S. Choi, K.E. Kasza, Measurement of
pressure drop and heat transfer in turbulent pipe flows of
Acknowledgements
particulate slurries, Argonne National Laboratory Report,
ANL-88-15, 1988.
This research work has been carried out during a [12] C.W. Sohn, M.M. Chen, Microconvective thermal con-
Humboldt research stay of the first author at the Institut ductivity in disperse two phase mixture as observed in a
f€
ur Thermodynamik (Institute of Thermodynamics) of low velocity Couette flow experiment, J. Heat Transfer 103
the Universit€at der Bundeswehr Hamburg (University of (1981) 47–51.
the Federal Armed Forces, Hamburg). [13] J.A. Eastman, U.S. Choi, S. Li, Development of energy-
The grant of the Humboldt research Fellowship to efficient nanofluids for heat transfer applications, Research
Prof. Sarit Kumar Das is greatly appreciated. Thanks Briefs Argonne National Laboratory, 2000.
are due to the Deutscher Akademischer Austauschdienst [14] Y. Xuan, W. Roetzel, Conceptions for heat correlation of
nanofluids, Int. J. Heat Mass Transfer 43 (2000) 3701–
(German Academic Exchange Service) who finances the
3707.
present research stay of Mr. Nandy Putra in Hamburg. [15] V.M. Wasekar, R.M. Maglik, Pool boiling heat transfer in
The authors would also like to express their thanks to aqueous solutions of an anionic surfactant, J. Heat
Professor Dr. G. B€ ohme, and Professor Dr.-Ing. H. Transfer Trans. ASME (122) (2000) 708–715.
Rothe of the University of the Federal Armed Forces [16] K. Cornwell, The role of sliding bubbles in boiling on tube
Hamburg, for the rheological experiments and surface bundles, in: Proceedings of the International Heat Transfer
roughness measurements, respectively, carried out in their Conference, Jerusalem, 1990, pp. 455–460.
laboratories. We thank Dr. D. H€ user for lending her ex- [17] A.J. Addlesee, K. Cornwell, D.G. Peace, Fluid dynamics of
pertise in explaining the surface characteristics. sliding bubbles and heat transfer implications, Adv. Pool
Boiling Heat Transfer Eurotherm (8) (1989).
[18] K. Cornwell, S.D. Houston, Assessment of convective
boiling on a horizontal tube, in: Boiling 2000, UEF
References Conference, Alyeska, USA, 2000, pp. 1–13.
[19] K. Cornwell, R.B. Sch€ uller, A study of boiling outside a
[1] J.A. Eastman, U.S. Choi, S. Li, L.J. Thompson, S. Lee, tube bundle using high speed photography, Int. J. Heat
Enhanced thermal conductivity through the development Mass Transfer 2225 (5) (1982) 683–690.
862 S.K. Das et al. / International Journal of Heat and Mass Transfer 46 (2003) 851–862

[20] M. Chun, M. Kang, Effects of heat exchanger tube [22] A.B. Cohen, Hysteresis phenomena at the onset of nucleate
parameters on nucleate pool boiling heat transfer, J. Heat boiling, Pool and External Flow Boiling, ASME, New
Transfer Trans. ASME 120 (1998) 468–476. York, 1992, pp. 1–14.
[21] K. Cornwell, S.D. Houston, Nucleate pool boiling on [23] R.J. Benjamin, A.R. Balakrishnan, Nucleate pool boiling
horizontal tubes: a convection-based correlation, Int. J. heat transfer of pure liquids at low to moderate heat fluxes,
Heat Mass Transfer 37 (suppl. 1) (1994) 303–309. Int. J. Heat Mass Transfer. 39 (12) (1996) 2495–2504.