Nanofluids in Thermosyphons and Heat Pipes

International Journal of Thermal Sciences 72 (2013) 1 e17

Contents lists available at SciVerse ScienceDirect

International Journal of Thermal Sciences j o u r n a l h o m e p a g e : w w w . e l s e v i e r . co co m / l o c a t e / i j t s

Nano�uids in thermosyphons and heat pipes: Overview of recent experiments and modelling approaches Matthias H. Buschmann Institut für Luft-und Kältetechnik Dresden, Bertolt-Brecht Allee 22, 01309 Dresden, Germany

a r t i c l e

i n f o

Article history: Received 23 November 2012 Received in revised form 24 April 2013 Accepted 24 April 2013 Available online 24 June 2013 Keywords: Nano �uids Thermosyphon Heat pipe Thermal performance

a b s t r a c t

Confronted Confronted with limited energy and material material resources resources and undesirable undesirable manmade climate climate changes, changes, science science is searchi searching ng for new and innova innovativ tive e strate strategies gies to save, save, transf transfer er and store store therma thermall energy energy.. Currently, one of the most intensively discussed options are the so-called nano�uids. Nano�uids are suspensions suspensions consisting consisting of a liquid base�uid and solid particles of sizes ranging from 10 nm to 200 nm. The higher thermal conductivity of these nanoparticles leads to an increased effective thermal conductivity of the � uid which, the general expectation is, should enhance the heat transfer of the device. This overview aims to compile results of the application of nano�uids in thermosypho thermosyphons, ns, heat pipes, and oscillating heat pipes. The general goal is to draw conclusions with respect to the potentials for improvement of the thermal performance of these gadgets. Additionally, possible mechanisms which may generate these improvements are discussed. All together 38 experimental studies and 4 modelling approaches approaches are analyzed. analyzed. While most investigations investigations recognize nano�uids as an advantageous working �uid, some others report negative negative effects. Performance effects which are related to �lling ratio, inclination angle, and operation temperature seem to be similar to those for classical working � uids. Several authors report a decrease of the thermal resistance or an increase of the ef �ciency with increasing concentration, but also a reversing of this trend if a certain optimal concentration is exceeded. This observation mainly follows with a signi�cant increase of the evaporator evaporator heat transfer transfer coef �cient. The condenser heat transfer coef �cient seems to be notor only weakly affected. Base�uid, nanoparticle material, size and shape, and the stabilization of the suspension have an in�uence on the thermal performance. However, the limited number of experiments does not allow drawing �rm conclusions. The main mechanism responsible for the improved thermal performance seems to be a porous layer built from nanoparticles on the evaporator surface. Additional positive effects may follow from the changed thermophysi thermophysical cal properties of the working � uid. reserved.  2013 Elsevier Masson SAS. All rights reserved.

1. Introduction

Confronte Confronted d with limited energy and material material resources resources and undesir undesirable able manmade manmade climate climate changes, changes, science science is searching searching for new and innovative strategies to save, transfer, and store thermal energy. energy. Currentl Currently, y, one of the most intensiv intensively ely discusse discussed d options options are the so-ca so-calle lled d nano nano�uids. uids. Rapid Rapid advanc advances es in manufa manufactu cturin ring g methods allow the production of nanoparticles of various sizes, shapes shapes,, and materi materials als.. Nano Nano�uids uids are create created d by suspen suspendin ding g nanoparticles of size 10 nm e200 nm in varying base�uids. Fig. 1 compares the normalized numbers of publications in the � elds of nano�uids, uids, heat transfer, transfer, turbulen turbulence, ce, and turbulent turbulent boundary boundary layer layer.. Normali Normalizatio zation n is carried carried out with the 201 2011 1 values values which which are 425 for the keyword keyword nano�uids, 8729 8729 for heat transfer, transfer, 5736 for turbulence turbulence

E-mail address: Matthias.Buschmann@ [email protected] ilkdresden.de..

1290-0729/$ e see front matter  2013 Elsevier Masson SAS. All rights reserved. http://dx.doi.org/10.1016/j.ijthermalsci.2013.04.024

and 896 for turbulent boundary layer. Clearly the exponential increase of publications for nano �uids is visible. The motivation for this new step in heat and mass transfer can be found in upward trends in energy density of electronic devices, increasing packing density of heat transfer equipment in general, miniat miniatur uriza izatio tion n of heat heat excha exchange ngers, rs, and other other advanc advanced ed heat heat transfer concepts. The general expectation is that the higher thermal condu conducti ctivit vity y of the nanopa nanoparti rticle cle materi materialsleads alsleads to an incre increase ased d effective thermal conductivity of the �uid which in turn should enhance the heat transfer of the device. The minuteness of the nanoparticles provides hope that these advantages are not counteracted teracted by clogging, clogging, sedimenta sedimentation, tion, and abrasion, abrasion, issues issues known known for larger larger particle particles. s. Several Several recent recent overview overviewss analyzing analyzing the current current nano�uid research (e.g. Sergis and Hardalupas [1] [1],, Thomas and Sobhan [2] [2])) suppo support rt this this assump assumptio tion. n. The darker darker facet facet of nano nano�uids is their extreme complexity, complexity, preventing �rst-principle rst-principle solutions and customar customary y physic physical al models models from describi describing ng �uid uid mechan mechanic ic and

2

M.H. Buschmann / International Journal of Thermal Sciences 72 (2013) 1e17

Nomenclature di dnf Lc Le Lhp

inner diameter, mm diameter of nanoparticle, nm length of condenser section, mm length of evaporator section, mm geometric length of heat pipe, mm

Greek symbols a angle of inclination against horizontal axis,  concentration, vol. %, wt. %, ppm 4 Abbreviations Cd diamond CNT carbon nanotubes

thermodynamic behaviour (Feja and Buschmann [3]). Therefore, most nano�uid investigations are carried out experimentally and so far in rather simple geometries such as straight pipes (Prabhat et al. [4], Buschmann [5]). New challenges occur with the application of nano�uids in thermosyphons, heat pipes, and oscillating heat pipes (Liu and Li [6]). Thermosyphons and heat pipes are well understood and widespread applied heat transfer devices. They are highly ef �cient due to the utilized phase change. Design, operation principles and thermal performance are discussed in detail in several textbooks (e.g. Reay and Kew [7], Faghri [8]). Schematics of the operation principles of these three types of apparatus are shown in Fig. 2. Thermosyphons are devices of passive heat exchange employing natural convection. Fig. 2a shows the main components of thermosyphons, the ways of rising vapour, and down � owing condensate �lm. Thermosyphons are either open-loop ore closed. In the latter case the working � uid returns to the original evaporator via the down running condensate � lm. The main parts of a heat pipe (Fig. 2b) are the evaporator were heating takes place, the adiabatic section, and the condenser were cooling takes place. In a standard heat pipe condensate return is ensured via capillary force. Therefore, the inner wall is lined with a wick which has a capillary structure. Usually screen wicks, sintered powder wicks, axial or helical grooves are employed. Alternatively the condensate can be returned by centripetal, electrokinetic, magnetic, osmotic forces or other strategies [7]. In case that a heat pipe is inclined against the horizontal axis its operation is supported by gravity. Forlarge �lling

DI EA EG ESS FR gHP HP MWCNT OHP oTS sHP ST TS US FP-OHP

deionized electromagnetic vibration ethylene glycol electrostatic stabilization �lling ratio grooved heat pipe heat pipe multiwalled carbon nanotubes oscillating heat pipe open two-phase thermosyphon heat pipe with sintered powder wick surfactant closed two-phase thermosyphon ultrasonic vibration �at plate oscillating heat pipe

ratios and inclination angles the effect of capillary structure becomes insigni�cant. In a standard heat pipe the � lling ratio is just suf �cient to saturate the capillary structure.1 When employing nano�uids as working �uids, a puzzling manifold of additional parameters e nanoparticle size, shape and material, base�uid characteristics etc. e in�uences the thermal performance of these gadgets. The general goal is to lower the thermal resistance, de�ned as the transported heat divided by the temperature difference between evaporator and condenser. This target seems to be not only reached by the enhanced thermal conductivity of the working �uid as it is the case in e.g. laminar pipe �ow. It is rather a complex interplay of nano �uid thermal properties, nanoparticle interaction with evaporator surface and wick, and changed vapour bubble generation due to porous layers formed from nanoparticles. This overview aims to compile recent experimental results of thermosyphons, heat pipes, and oscillating heat pipes operated with nano�uids. The survey is supplemented by several modelling approaches. Possible physical mechanisms acting in gadgets operated with nano�uids are discussed. The general goal is to draw conclusions with respect to the potentials for improvement of the thermal performance. Focus is therewith on the differences between reference �uid and nano�uid. Additionally suggestions for future research are given. However, the application of nano �uids and especially in heat pipes and related devices is still associated with severe and unresolved questions. Many issues are open or even not commenced to investigate. Therefore, parts of the following text and the conclusions are formulated with reserve. 2. Organization of paper

The study focuses on recent advances in the application of nano�uids in closed two-phase and open thermosyphons, heat pipes, and oscillating heat pipes. The main focus of the study is directed towards the effects caused when nanoparticles are added to the working � uids of these gadgets. A general rule proposed by Fernholz and Finley [9] in their 1996 review paper on turbulent boundary layers: In the absence of any complete or convenient theoretical approach, our primary function is to describe what we see is applied. In our case this means that the results seen by the different investigators are compiled according to their logical connectedness. Based on that ordering, the four main parts of the survey discuss physical effects which are related to operating Fig. 1. Normalized number of publications in different � elds of �uid mechanics and heat transfer. Normalization is carried out with 2011-values. Data taken 2013-04-23 from ISI WEB of KNOWLEDGE.

1

The author is grateful to an anonymous reviewer for pointing this out to him.


cooling

(a)

heating

(b)

3

(c) condenser

wick

rising vapor

vapor

down flowing condensate film

upward flowing condensate

vapor bubble Q

Q

Q

working fluid

wick

liquid slug

heating

working fluid

selfsustained thermally driven bubble / slug oscillation

cooling evaporator

Fig. 2. Schematics of operation principles of thermosyphon (a), heat pipe (b) and oscillating heat pipe (c) adapted from Refs. [7] and [56].

parameters of gadgets, effects which are related to nano�uid characteristics, the general thermal performance of the gadgets, andmodelling approaches. Some of the �ndings cannotbe assigned to one of these rubrics alone and appear therefore in multiple contexts. Based on the compiled experimental results and physical arguments of technically closely related investigations, statements with respect to the different physical mechanisms acting in gadgets operated with nano�uids are discussed. Table 1 compiles all experimental studies discussed. The different gadgets are denoted as done by the original authors. This is also true for some of the so-called heat pipes which are strictly thermosyphons. The serpentines of the oscillating heat pipes are called turns. Their number is identical with the half of the number of straight tubes. In cases where the original authors provide some information with respect to the wick this is mentioned in Table 1. In some occasions �gures of the original publications are quoted to draw the reader s attention to results which are verbally dif �cult to describe. ’

3. Some statistics

With this study the results of 38 experimental and 4 modelling approaches are compiled (Table 1). Among the experiments 11 turn their attention to closed two-phase thermosyphons and one to open thermosyphon, 18 to wicked, or grooved heat pipes, and 8 to oscillating heat pipes. 51 nano �uids have been tested all together with these experiments. A majority, 41, are water-based nano�uids. Other basefluids are the refrigerant R11 (trichloro �uoro-methane, CCl3F), ethyleneeglycol mixtures (EG), and acetone ((CH3)2CO). Nanoparticles employed are metals, namely silver (Ag), gold (Au) and copper (Cu), oxides (Al 2O3, TiO2, SiO2, CuO, ZnO, Fe 2O3) or variations of carbon (diamond, carbon nanotubes). The number of experiments carried out with silver nanoparticles, 13, exceeds by far the other materials. Alumina (Al2O3) follows with 11, CuO with 5, pure copper and titanium (TiO 2) with 4, silica (SiO 2) and iron (II, III)-oxide (Fe2O3) with 2, and zinc oxide (ZnO) with one experiment. The size of the nanoparticles ranges from 2 nm to about 100 nm. Note that most publications provide only the size of the primary nanoparticles. Due to agglomeration, the actual size within the nano�uid might be much larger. Fig. 3 provides an overview of the frequency of the nanoparticle size employed. A maximum exists between 20 nm and 40 nm.

Most authors provide only short comments with respect to the preparation of their nano�uids. Four papers give even no information. The overwhelming majority (approx. 91%) employs twostep methods where purchased or at least separately produced nanoparticles are dispersed in the base �uid. Only three experiments (Tsai et al. [10], Hajian et al. [27], Manimaran et al. [28]) were carried out with nano�uids produced by one-step methods. Here the nanoparticles were directly created within the base �uid by chemical processes. Most two-step methods employ ultrasonication for dispersing the nanoparticles. The sonication time varies between 1 h and 20 h (Fig. 4). A maximum is found around 4 he6 h. About one third of the experimental groups state that they have not employed any stabilization or surfactant. Particle concentrations are either given in volume or in weight percentage or in parts per million (ppm). Nanoparticle concentrations ranging from 0.003 to 5.3 vol. %, from 0.1 to 0.5 wt. %, and from 1 to 10 4 ppm are utilized in the experiments ( Table 1). Due to the fact that it is rather dif �cult to value how the base�uid density is changed by adding stabilizers or surfactants (Feja and Buschmann [3]), no attempts have been made to convert these different data. While 16 experiments are conducted only in vertically orientated gadgets, 9 studies investigate only horizontally positioned apparatus. In seven experiments the inclination angle is varied. Only one publication (Riehl and dos Santos [54]) reports inclination angles between 90 (evaporator on top) and 90 (evaporator at bottom) for an oscillating heat pipe. 4. Effects with respect to gadget parameters 4.1. Filling ratio

It is known from classical working �uids that the �lling ratio has a non-negligible in�uence on the maximum heat throughput of heat pipes (Reay and Kew [7]). The � lling ratio of thermosyphons and gravity supported heat pipes operated in thermosyphon mode is de�ned as the ratio of working � uid volume to internal evaporator volume. Differently for oscillating heat pipes the �lling ratio is speci�ed as ratio of working � uid volume divided to total internal volume. Experiments by Khandekar et al. [56] employing classical working �uids in an OHP showed that the maximum of the heat throughput dependence on the � lling ratio is rather � at. For water and ethanol the optimal region stretches from 10% to 80% and for


h 5 S U

g n i x i M

% . t w 1 , 5 7 . 0 , 5 . 0 , 5 2 . 0

h 3 S U / T d S o 3 / n h g h n 4 i x 4 S i S U M U

5

h 5 . 1 S U

% . % . % l l . o . % . v % o % t v l . 5 t t 4 o w , w w  v 2 . 3 . 2 . % 0 , 6 . 1 1 1 1 0 1 t e , e e , w  1 1 3 1 1 . . . . . e 3 0 0 0 0 0 5 Fig. 3. Frequency of the nanoparticle size employed in the experiments compiled in Table 1.

0 0 1 <

0 8 , 0 6 9 6 0 2 3 2 5 5 3 5 2 f )

g A / O 2 H

H O ( O l 3 A - 3 D g 2 O O C 2 / O l 2 u / g D l O A C A 2 i S A C O 6 / / / H / / / O O O , O O O H 2 2 2 3 2 2 2 C H H H G E H H H

0 5

0 7

0 7 , 0 6 , 0 5 0 5 0 5 0 5

0 9 , 0

0 9  , 0 , 0 0 0 9 9 0 9

0 9

g

0 0 0 0 5 5 1 1 0 0 0 0 5 5 1 1 0 9 / 4 . 0 0 0 5 5 0 5 2 / 1 3 4 4 . 1 3 1 / e 9 3 . 1 e 6 7 . 2 0

4 6 0 / 0 0 7 7 / 4 / 0 / 0 5 5 5 5 / 1 / 0 / 5 2 5 2 1 6 2 / . / 2 1 2

s n r u t 0 4 P H O /

s s s n r n r n r u t u t u t 6 6 6 P P P H H H O / O / O /

B

] 8 4 [ . l a t e e h k a p a n n a W

, 0 1 , 8 s P n H r u t O / 6  1

] 9 4 [ . l a t e g n e h C

0 0 1 / 0 0 1 / 0 0 3 / 5 . 1 s n r u t 5 . 7 P H O /

B B B B

] ] 3 5 ] 2 [ 0 5 , u 5 1 [ . 5 [ W l a l . d t n e a a t e u u i Q J Q

, ] 4 5 [ s o t n a S ] s o 5 5 d [ l & h e l i h R e i R

. ] 9 4 [ . f e R e e s s l i a t e d r o F . s t n e m i r e p x e l l a r o f o i t a r g . n n i l o l i � t c t e n s a t s s s o n r o c c . r s . a a l i d T y u o u N g � l C n p a o y t n r m c a a . e e t r n i s m o e h r n s t  e p g d f d h n i t o l e n d a f . . r s o h r r t l o g a e s a o t c . n n t n h i a e i t e r r i o l o m m u d r t a t p a l n e a d u i i e l a n l n a v d a a i y m b e i n c i c e a r t i e s e l g i t l i d m u e o r e b t a m r c o a r a e d m i h c t d r e e i a y u i n a o h d O A D V H B T n I a b c d e f g

B

R-123 from 35% to 80%. This knowledge motivated several nano�uid experiments with varying � lling ratios. Lin etal. [47], investigating an OHP with 5 turns, found an optimal �lling ratio of 60% for DI-water and for Ag-nano�uids (100 ppm and 450 ppm). Similarly, Paramatthanuwat et al. [41,42] showed a maximum of heattransferat FR ¼ 50%for silvernano�uidin a circular thermosyphon.Butthe samewastruefor thereference �uid DI-water. Mousa [25] found that the thermal resistance of a vertical circular heat pipe was minimal for a � lling ratio of 40e50% for both DI-water and Al2O3-nano�uid. The � lling ratio was de �ned as the percentage of the evaporator � lled with working � uid. This � nding was con�rmed for a vertical thermosyphon by Mousa [37]. The optimal � lling ratio here was 48%. Manimaran et al. [28] (H2O/CuO in wire meshed HP) showed an increase of thermal ef �ciency when the � lling ratio was increased (similar as for water). The maximum was reached for an inclination angle of 30 and FR ¼ 75%. Even with an inclination angle of 0 maximal ef �ciency was reached with FR ¼ 75%. Investigating an OHP with H2O/Al2O3, Quet al. [50] found that a � lling ratio of 70% gives the largest decrease of the thermal resistance at a power input of 58.8 W. An exception was the experiment by Teng et al. [19]. These authors indicated that at an optimal tilt angle of 60 the thermal ef�ciency was higher the lower the FR was. Maximal values were achieved at FR ¼ 20% withH 2O/Al2O3. Conversely, DI-water showed a maximum at 60 when FR was 60%. To summarize, an optimal �lling ratio seems to exist between 45% and 70% depending on the design of the gadget. Clear trends with respect to different nano�uids cannot be derived. In the most

Fig. 4. Sonication time of the experiments compiled in Table 1.

6


cases the found optimal �lling ratios are the same or at least similar as for water. 4.2. Inclination angle

The inclination angle of a thermosyphon or heat pipe is of special importance for e.g. solar heating systems and cooling devices with changing spatial position. Similar as for the �lling ratio, it is known that the inclination angle affectsthe thermal performance signi�cantly (Patil and Yarasu [57], Liu and Li [6]). Wang et al. [24] analyzed a wicked heat pipe and found that the inclination angle has nearly no impact on the temperature distribution of the condenser when H2O/CuO nano�uids were employed. Therefore the condenser heat transfer coef �cient was only negligibly changed. At an inclination angle of 45 an increase of about 5% compared to the horizontal pipe was found for a concentration of 1 wt. %. The evaporator temperature depended slightly on the inclination angle. Here a minimum was found for an angle of 45  compared to the horizontal heat pipe where the heat transfer coef �cient was increased by about 22% (see Ref. [24] Fig. 5b). This maximum was not affected by operation pressure and heat input. However, all of these observations were comparable with the results for pure water. In general it was found that the ratio of the maximum heat �uxes of nano�uid and water was nearly independent of the inclination angle. The ef �ciency of the wire mesh wickedheat pipe investigatedby Senthilkumar et al. [20,21] improved for increasing angles of inclination. However, reaching angles larger than 30  for DI-water and aqueous solution of n -Hexanol and 45 for CuO DI-water and CuO aqueous solution of n-Butanol and n -Hexanol nano�uids the ef �ciency started to decrease again. Similarly, Teng et al. [19] found for their heat pipe operated with H2O/Al2O3 nano�uid that an increasing tilt angle ledto an increase of the thermal ef �ciency. This increase deteriorated above 60 . The same was true for pure water. A continuous decrease of thermal resistance with increasing inclination angle was found by Manimaran et al. [28] for their wire meshed heat pipe using copper oxide nano �uid. The trends were similar as for water but at much lower levels. A continuous more or less monotonic increase of the heat �ux from evaporator to condenser with inclination angle was found by Wannapakhe et al. [48] for an OHP. This was the case for all investigated concentrations of Ag-nanoparticles and pure water. Exceptionsare presented in thestudiesby Huminic et al.[38] and Huminic and Huminic [39], which investigated a thermosyphon at inclination angles between 30 and 90 with a water based Fe 2O3nano�uid. While the thermal resistance of the water- �lled thermosyphon had a minimum at an angle of 45  , the nano�uid-�lled thermosyphon showed the lowest values at 90 (see Ref. [39] Fig. 7). The maximal reduction rate of the thermal resistance occurredat an inclinationangleof 90 anda concentration of 5.3vol. %.Atanangleof45  theheatthroughputincreased by19%for 2 vol. % and by 22.2% for 5.3 vol. %. The increase was nearly doubled for an angle of 90 (39%at 2 vol.% and 42% at5.3 vol.%).It shouldbe noted that in these experiments the � lling ratio was with 12.6% comparably small and that very small particles (4 e5 nm) were employed. To summarize, while some authors �nd an optimum inclination angle, others see a continuous change of thermal performance with changing angle. Five from the six studies dealing with the in�uence of the inclination angle for nano �uid-�lled gadgets �nd similar behaviour as known for pure water.

only three studies address this parameter directly. For example, Paramatthanuwat et al. [41] found that with increasing working temperature the bene�ts of silver nano�uid compared to pure water became slightly more pronounced. Paramatthanuwat et al. [42] con�rmed this result for water-based Ag-nano�uids with added oleic acid surfactant; however, this was also true for the classical working �uid DI-water. A comparable result again for H2O/ Ag nano�uid is known from Wannapakhe et al. [48]. Their oscillating heat pipe showed a monotonically increasing maximum heat �ux from evaporator to condenser with increasing operation temperature regardless of the nanoparticle concentration. Finally, Wang et al. [24] showed that the ratio of the maximum heat � uxof H2O/CuO-nano�uid and water depended slightly on the operation temperature and reached a maximum of 1.4 and 1.28 at 40  C and 60  C. Taking all results together it seems clear that water-based nano�uids behave similarly to pure water with respect to operation temperature. 5. Effects with respect to nano�uid parameters 5.1. Base �uid

Certainly the characteristics of the base�uid ofa nano�uid playa dominant role with respect to its effective thermophysical properties. Nevertheless, only three studies, Senthilkumar et al. [20,21] and Putra et al. [29], address the question if different base �uids change the thermal performance of heat pipes. Senthilkumar et al. [20,21] investigated CuO nano�uids based on either DI-water or aqueous solution of n -Butanol or n -Hexanol both at a concentration of 100 mg/l. The thermal ef �ciency of nano�uids based on aqueous solutions of n -Butanol or n -Hexanol were in all cases signi �cantly better than for nano�uids based on DI-water. The authors argue that this was caused by the positive surface tension gradient with temperature possessed by n-Butanol and n-Hexanol. However, the experiments carried out with nHexanol by Senthilkumar et al. [21] seem to indicate that this effect was neither enforced nor attenuated by adding nanoparticles. Weak differences of the thermal resistance of DI-water and pure ethylene glycol were found by Putra et al. [29] for a screen mesh wick heat pipe. Adding nanoparticles to these base �uids led to the differences becoming much more pronounced. However, in every case the water based liquids showed a lower thermal resistance. While for H2O/Al2O3 (5 vol. %) the thermal resistance was decreased byabout 78% (10 W), onlya decreaseof 11% (10 W)was achieved for EG/Al2O3 (5 vol. %). A similar trend was observed for TiO 2 nanoparticles (5 vol. %, 10 W). Here a reduction of 69% was achieved for the water-based and 2.4% for the ethylene glycol-based nano �uid. The effect was slightly more pronounced for higher heating loads. Putra et al. [29] argued that this signi �cant difference was basically caused by the different enhancements of thermal conductivity of the two basefluids. While by adding 5 vol. % of TiO2 nanoparticles to water a thermal conductivity enhancement of about 14% was achieved, forethyleneglycol only an increase of about 9% was obtained. This in turn should have caused the very different increases of the evaporator heat transfer coef �cient (see Ref. [29] Fig. 12). These � ndings clearly indicate the in �uence of the base�uid on the thermal resistance. However, the number of investigations with successively varied base�uids is still too small to draw �rm conclusions.

4.3. Operation temperature

5.2. Particle size, shape, material, and suspension stabilization

The third gadget parameter discussed is the operation temperature. Despite its known relevance from classical working �uids,

This section compiles effects which are related to the characteristics of the nanoparticles directly. Numerous studies (for an


overview see Sergis and Hardalupas [1]) indicate that the material of the dispersed nanoparticles as well as their size and shape affect the effective thermophysical properties of nano �uids signi �cantly. For the effective values of density, heat capacity, and thermal expansion coef �cient, straightforward models exist which are physically correct. The situation is more complicated for thermal conductivity and dynamic viscosity. Very promising semi-empirical approaches either based on classical models like the Maxwellequation for thermal conductivity (Buongiorno et al. [58], Ehle et al. [59]) or based on physical assumptions for dynamic viscosity (Chevalier et al. [60], Corcione [61], Prasher et al. [62]) have been published in recent years. However, these models may lack generality for several reasons (Feja and Buschmann[3]). Stabilization is considered in this section because it is undertaken to prevent particles from sedimentation or forming agglomerates. Tsai et al. [10] found that enlarging the size of their gold nanoparticles from 8 nm to 24 nm led to a signi �cant reduction of the thermal resistance of the evaporator section. Kang et al. [13] and Kang et al. [15] investigated Ag/H2O suspensions with particle sizes of 10 nm and 35 nm. The general � nding was that while with the smaller particles the thermal resistance could be lowered at a maximum of 50%, with the larger particle a lowering of about 80% was achieved. The numerical model by Do and Jang [63] con�rmed this �nding. Ji et al. [51] investigated an OHP and found that the thermal resistance was reduced signi�cantly by adding Al2O3nanoparticles of sizes 50 nm and 80 nm. A slightly lower thermal resistance was achieved for the larger particles. Given unchanged primary surface roughness of the evaporators in all of these experiments,2 the surface particle interaction parameter (ratio of average surface roughness to average particle diameter) would be lowered by a factor of 1.6 [51], 3 [10] and 3.5 [13,15] when employing the larger particles. According to Harish et al. [64], the heat transfer enhancement ratio should then be lowered too. The opposite was found by the experiments of Kang et al. [13] and Kang et al. [15] which may indicate that the nanoparticles have acted in a way which cannot be explained alone by particle surface interaction. Therefore, one may hypothesize that under the in�uence of bubble formation and departure from the evaporator surface, nanoparticles might be forced to enhanced agglomeration. This would de�nitively change the particle size distribution of the working �uid. Particle size distributions measured under calm conditions would then be irrelevant for the real situation in the vicinity of the evaporator surface. Ji et al. [52] was the only work found so far which addressed the in�uence of the particle shape on the thermal performance of a heat pipe. Employing boehmite alumina ( g-AlO (OH)) particle shapes of platelets (⌀ 9 nm), blades ( ⌀ 60 nm, thickness 10 nm), cylinders ( ⌀ 10 nm, length 80 nm) and bricks (edge length 40 nm) were investigated. The same nanoparticles were analyzed with respect to their effective thermophysical properties of nano�uids by Timofeeva et al. [65]. Ji et al. [52] found that at an operation temperature of 20  C and a volume fraction of 0.3 vol. % for heat loads less than 100 W, heat transfer enhancement of their OHP was best for cylindrical nanoparticles followed by blade, plate, and brick shaped particles, respectively. However, for heat loads larger than 125 W the situation changed and brick shaped particles performed best followed by cylindrical, plate, and blade shaped particles, respectively. The experiments by Putra et al. [29] indicated strong differences between H2O/Al2O3 and H2O/TiO2 nano�uids for heat transfer

2 Note that the different gadgets are not compared among themselves. It is rather the change of effective roughness which follows from settling down of nanoparticle having different sizes within one device employing different nano �uids which is compared.

7

coef �cient at evaporator (see Ref. [29] Fig. 12). In both cases the concentration was 5 vol. %. A similar but much weaker trend was shown for EG/Al2O3 and EG/TiO2 (5 vol. %). Unfortunately, no information with respect to possible different sizes of Al2O3 and TiO2 nanoparticles were given. As mentioned above, this is of importance because particle size affects, together with primary roughness, the bubble formation process. Paramatthanuwat et al. [42], by adding oleic acid surfactant (0.5 vol. %, 1 vol. %, 1.5 vol.%) to water-based Ag-nano �uids, found that at a concentration of 1 vol. % of this dispersant the largest heat �ux and the highest effectiveness were achieved. Depending on the operation temperature an increase of the heat �ux between 150% and 50% (see Ref. [42] e.g. Fig. 7) was achieved. Rising the surfactant concentration further to 1.5 vol. % damped this increase remarkably. However, in any case the nano �uids stabilized with oleic acid surfactant showed a better thermal performance than the pure base�uid (DI-water) or the pure nano�uid. Similarly, Hajian et al. [27] found that non-stabilized nano�uids have a negative effect on thermal resistance. Qu and Wu [53] investigated an oscillating heat pipe employing water-based nano�uids with Al2O3 (30 nm) and SiO2 (56 nm) at concentrations 0.1, 0.3, and 0.6 wt. %. They found that while the silica particles increased the overall thermal resistance, the alumina particles lowered it. The reason seemed to be that the alumina particles deposited on the evaporator increased surface nucleation and thus enhanced heat transfer. The silica particles also formed a precipitate on the evaporator surface but decreased nucleation and contact angle. The straightforward conclusion that the particle material was the relevant parameter should be taken with caution. The TEM images of SiO2 and Al2O3 nanoparticles shown by Qu and Wu [53] indicate signi �cant differences of effective particle sizes due to clustering (see Ref. [53] Fig. 1). From that observation one rather could conclude that the stabilization and, therewith, the dispersion of the particles is important for the particle layer formed on the evaporator. Yang and Liu [43] investigated two types of water-based silica nano�uids. The �rst was a conventional nano�uid and the second was a specially stabilized nano �uid grafting silane (CAS number 2530-83-8) on the particle surface. The later was named functionalized nano�uid. Both �uids were manufactured employing the same technology (two-step method, 12 h ultrasonic bath). While the conventional nano�uid built a porous layer on the evaporator surfaces, this was not the case with the functionalized one. The authors reasoned that the latter enhanced the evaporator heat transfer coef �cient due to its changed thermophysical properties, mainly its increased thermal conductivity. The conventional nano�uid decreased the evaporating heat transfer coef �cient by up to 11%, depending on the operating temperature. Yang and Liu [43] argue that this deterioration was largely caused by the deposition layer on the evaporator surface. The analytical model by Alizad et al. [66] indicated a clear effect of the nanoparticle material on the decrease of the thermal resistance. Keeping base�uid, nanoparticle size, and concentration �xed showed that CuO-particles lower thermal resistance the most, followed by Al2O3-particles and TiO 2-particles. This is surprising from the point of view of a nano �uid as heat carrier because the thermal conductivity of the particle materials order as Al2O3 (35 W/ m-K), CuO (20 W/m-K) and TiO 2 (11.7 W/m-K). Therefore, one may assume that the heat transfer process depends additionally on other thermophysical properties such as heat capacity, and density, which is indeed known from classical working � uids. Additionally, the model indicated that the in �uence of the particle size diminished for particles larger than 20 nm. It is highly likelythat nanoparticle size, shape, and material have an in�uence on the thermal performance of the different gadgets. The same is true for any stabilization or surfactant employed.

8


However, most of these in�uences seem to be different than for heat transfer applications without phase changes where thermophysical properties are in focus. That one third of all experiments was carried out with non-stabilized nano �uids points in the same direction. There is some likelihood that stabilization as done for nano�uids in other thermodynamic applications is counterproductive for the gadgets investigated here. Especially the interaction between evaporator surface and wick surface on the one side and nanoparticles on the other side has to be taken in consideration. Some studies indicate that the larger the nanoparticles, the stronger the enhancement of the evaporator heat transfer coef �cient. Nevertheless, far too few experiments have been carried out to answer this and the much more complex questions with respect to optimal nanoparticle shapes and proper stabilizations. 5.3. Concentration of nanoparticles

Nanoparticle concentration is paramount with respect to a nano�uid s effective thermophysical properties. Therefore, it is not surprising that many studies aim to investigate the dependence of thermal performance on concentration. Numerous groups report a decrease of the thermal resistance or an increase of the thermal ef �ciency with increasing concentration, but also a reversing of this trend if a certain optimal concentration is exceeded. Already the early work by Wei et al. [12] reported a declining dependence of the thermal resistance on particle concentration of H2O/Ag nano�uids. At around 100 ppm this dependence seemed to have reached saturation. This �nding was basically con �rmed by Kanget al. [13] who founda loweringof the evaporator temperature by adding 1, 10 or 50 ppmAg-nanoparticles to pure water. However, this trendwas stoppedfor a concentration of100 ppm andno further improvement was observed. Another con �rmation came from Lin et al. [47], who showed that a concentration of 100 ppm of Agnanoparticles (20 nm) in DI-water led to a better ef �ciency of their OHP than a concentration of 450 ppm. An OHP with H 2O/Ag nano�uid was also investigated by Wannapakhe et al. [48], who found that a concentrationof 0.5wt. % delivered thehighestheat �uxforall operation temperatures, inclination angles, and aspect ratios L e/di. Hajian et al. [27] presented a more detailed study. Investigating a comparably large wicked heat pipe (Le ¼ 1000 mm) they found that adding 50 ppm Ag-nanoparticles to pure water led to a maximum lowering of the thermal resistance by about 30%. Higher concentrations(200 ppm and600 ppm)showed a signi�cant increaseof the thermal resistance. Analyzing the response times of their heat pipe in dependence of the employed working � uid, surprisingly no regularity in dependence of the concentration was found. Naphon et al. [14] investigated TiO2-nano�uids based on alcohol and Naphon et al. [16] based on the refrigerant R11 (trichloro�uoromethane). The �rst case showed an increase of heat pipe thermal ef �ciency up to a concentration of 0.1 vol. % (FR 50%). For 0.5 vol. % the effect disappeared completely and the ef �ciency was even lower as for the pure basefluid alcohol. With R11 the highest heat pipe ef �ciency was again found for an optimum concentration of 0.1 vol. %. Nearly no increase of the ef �ciency occurred for a concentration of 0.01 vol. % of TiO 2 in R11. At an operation temperature of 100  C and a concentration of 1.2 wt. % (H2O/CuO), an increase of about 15% of the evaporator heat transfer coef �cient was found by Lu et al. [40] for an open thermosyphon. Lower and higher concentrations showed less improvement. This result was basically con �rmed by Liu and Zhu [23] for a horizontal mesh heat pipe. Besides the signi�cantly increased heat transfer coef �cient at the evaporator with increasing concentration, these authors also reported a higher heat transfer coef �cient at the condenser. At the evaporator a distinct maximum ’

was reached at about 1 vol. %. At the condenser the maximum was �atter and ranged from concentrations of about 0.8 vol. % to about 1.25 vol. %. For higher concentrations this trend reversed. At a concentration of 2 vol. % the evaporator heat transfer coef �cient amounts to only about two thirds of the maximal value. However, in any case the heat transfer coef �cients at evaporator and at condenser were higher for the investigated nano �uid (H2O/CuO) than for pure water. These general trends were con�rmed by Wang et al. [24] for the same wicked heat pipe but with inclination angles ranging between 0 and 90 . Four independent research groups found for H 2O/Al2O3 nano�uids no optimal concentration or only a limited dependence of the thermal performance on the concentration. The two concentrations (1, 3 vol. %) investigated by Do et al. [18] clearly showed a decrease of thermal resistance. However, while for the very low heat input of 3 W the decrease signi �cantly depended on the concentration nearly identical results were found for heat inputs ranging between 5 W and 15 W (see Ref. [18] Fig. 6). No optimal point for thermal resistance of a thermosyphon in dependence of the concentration was found by Mousa [37]. However, thermal resistance was nearly constant from 1 vol. % H 2O/Al2O3 nano�uid onward. Nanoparticle size was 40 nm. In a more recent study Putra et al. [29] found that the temperature was lowered throughout their heat pipe when the Al2O3 concentration of the working � uid was increased from 1 vol. % to 5 vol. % without running through a maximum or reaching saturation. Similarly Noie et al. [34] did not �nd an ef �ciency optimum depending on concentration in the range from 1 vol. % to 3 vol. % at a � xed input power (see Ref. [34] Fig. 3). The � fth group dealing with H2O/Al2O3 nano�uids in an oscillating heat pipe (Qu et al. [50]) reported that the maximal decrease of thermal resistance occurred at a concentration of 0.9 wt. %. For higher concentrations the thermal resistance increased again. This optimum was related with a comparable high � lling ratio of 70%. To summarize, several results show that for water-based Agnano�uids an optimum exists for concentrations of about 100 ppm. Optima were also found for TiO 2 and CuO nanoparticles dispersed in different basefluids. Surprisingly, four independent studies dealing with H2O/Al2O3 nano�uids in thermosyphons and heat pipes couldnot �ndsuch an optimum inthe range of 1 vol.%e5 vol. %. The latter might be caused by different reasons; among them shape effects as they are reported by Ji et al. [52] or that the optimum simply occurs outside the investigated concentration ranges. However, that an optimum was found foran H2O/Al2O3 nano�uid in an OHP [50] indicate that nano�uids mayact differently in different types of gadgets. 5.4. Latent heat and surface tension of nano �uids

Khandekaret al.[56] compiled characteristics of thermophysical properties which an optimal working �uid should have. Among them are low dynamic viscosity, low latent heat, and low surface tension. These demands should also be valid for heat pipes and thermosyphons employing nano�uids. However, it is known from several experimental works (e.g. Chevalier et al. [60]) that the dynamic viscosity of the pure liquid is signi �cantly increased by the addition of nanoparticles. Therefore it seems to be a priori impossible that the demand for low viscosity can be satis �ed by any nano�uid. The two other thermophysical properties, especially the latent heat, are rarely addressed by nano �uid studies. The only investigation discovered which deals with the latent heat of nano�uids is Worek et al. [67]. These results show that the experimentally determined latent heat of H2O/Al2O3 nano�uid (46 nm) and H2O/BiO2 are nearly the same as the one of pure water. Worek et al. [67], in analyzing the same nano�uids as for the latent heat, found basically no changes of the surface tension


compared with pure water. Similarly, Das et al. [68] indicated that the surface tension of their H 2O/Al2O3 nano�uid had basically not changed for concentrations up to 4 vol. %. No changes up to about 5 vol. % were also found by Buschmann and Franzke [69] for a conventional H2O/TiO2 nano�uid. A steep linear decrease of surface tension with increasing concentration was found by Liu and Zhu [23] for water-based CuO-nano�uids. At a concentration of about 2 vol. % the surface tension was lowered by 15% compared to the value of pure water. For higher concentrations the surface tension stayed at this value up to the highest investigated concentration of 4 vol. %. An increase of the surface tension was found by Xue et al. [30]. The analyzed suspension with 1 vol. % CNTshowed an increase by about 113.5% compared to pure water. Some experimental results are compiled in Fig. 5. The results for the surface tension are very contradictory and do not allow �rm conclusions with respect to thermosyphons and heat pipes despite the fact that more research is urgently needed with respect to these thermophysical properties. Other thermophysical properties relevant for heat pipes and thermosyphons, especially density, heat capacity, and the thermal expansion coef �cient, can be calculated in a straightforward manner. Thermal conductivity and viscosity of nano�uids are still dif �cult to model and may have to be measured in each individual case. 5.5. Nano �uids with magnetic nanoparticles

Ferro�uids are suspensions of magnetic particles dispersed in a base�uid. The particles may be Fe 3O4, cobalt or any magnetically ordered material. The response of ferro �uids to magnetic �elds depends on the saturation magnetisation (Popplewell et al. [70]). This ability offers additional options to in �uence the working � uid and may enhance therewith the thermal performance of thermosyphons and heat pipes. Two publications present results for magnetic nano�uids or magnetically enhanced heat pipes. Chiang et al. [26] operated a speci �c heat pipe with a magnetic nano�uid consisting of water and Fe 3O4-nanoparticles (0.8 vol. %). The essential difference compared with classical heat pipes were the porous iron nozzles and the magnets at evaporator and condenser (for design details see Fig. 1 in Ref. [26]). Both design components in�uenced obviously the formation of slugs and vapour bubbles and therewith the complete �ow pattern within the heat pipe. Applying a magnetic �eld strength with a magnitude of 200 Oe the thermal resistance of the heat pipe operated with the magnetic nano�uid was lowered by about 40%. The vector of the magnetic � eld strength de�nes strength and direction of magnetic �eld at any spatial point.

Fig. 5. Experimentally determined surface tension of water-based nano �uids. Broken line denotes pure water and dotted lines show a band of  1% of pure water.

9

The second investigation employing magnetic effects was published by Heris et al. [44]. This group investigated a closed thermosyphon employing a H2O/Ag-nano�uid. A variable magnetic �ux density (0.12, 0.35, and 1.20 T) was applied to the evaporator section. It was found that with increasing magnitude of the magnetic �ux density the thermal resistance of the thermosyphon was decreased. At 0.35 T (1.20 T) a decrease of about 3.22% (5.37%) compared to pure water was observed for a particle concentration of 20 ppm. The effect became the more pronounced the higher the concentration. However, within the experimentally investigated ranges of concentration and magnetic �ux density, no optimal combination with respect to the thermal resistance was found. These are two promising results with respect to magnetic nano�uids. However, much more research is needed to establish this very special type of liquid as working �uid in heat pipes and thermosyphons. 5.6. Nano �uids with carbon nanotubes

Due to their large length to diameter ratio, carbon nanotubes (CNTs) do not really belong to nanoparticles. However, an increasing number of investigations pay attention to nano �uids with CNTs for their extraordinary mechanical, electrical, and thermophysical properties (Xie and Chen [71]). Carbon nanotubes can be single, double, or multi-walled. One of the most interesting features is their astonishing high thermal conductivity along the tube axis which may reach up to 2000 W/(m-K) for a CNT with a diameter of 9.8 nm (Fujii et al. [72]). However, due to the large length of CNTs such nano �uids may behave completely different than nano�uids with rigid and more or less spherical nanoparticles. Three research groups analyzed the thermal performance of thermosyphons and heat pipes operated with CNT-nano�uids. One of them, Xue et al. [30], reported negative results and dif �culties with the preparation of suspensions with CNTs keeping their high aspect ratio. Only when the primary nanotubeswere chemically cut in pieces, whereby some kept their tube structure and others became fragments, could stability at 1 vol. % be established. Due to the signi�cantly increased surface tension of this speci�c nano�uid, the evaporation process was modi�ed which led tothe formation of larger bubbles, diminished departure frequency, and a tendency of bubble-coalescence. All of that resulted in a decreased performance of the investigated thermosyphon. For example, at a heat �ow of 555 W the thermal resistance increased by a factor of 3.3 compared with the water � lled thermosyphon. A completely contrary result was reported by Liu et al. [36] investigating multiwalled carbon nanotubes (MWCNT) in a thermosyphon. The thermal resistance of the investigated thermosyphon was clearly lowered when employing MWCNT-nano�uids. Additionally, these authors showed that the operation range was signi�cantly larger for MWCNT suspensions (from 20 W to 465 W) than for DI-water (from 20 W to 200 W). The optimal mass concentration found was about 2 wt. %. Depending on heat �ux and operation pressure, the evaporator heat transfer coef �cient was increased up to 40%. In a very recent work Shanbedi et al. [45] investigated MWCNTs in a thermosyphon. The nano�uid was stabilized by functionalizing the nanotubes with ethylenediamine. Differently to Xue et al. [30] the nanotubes kept their cylindrical structure with a diameter ranging from 10 nm to 20 nm. Thermal resistance of the thermosyphon decreased up to a concentration of 1 wt. % and increased again for higher concentration. In any case, thermal resistance decreased with increasing input power. This effect was mainlycaused by a temperature decrease at the evaporator while the temperature at the condenser was only weakly changed. Employing CNT-nano�uids in thermosyphons and heat pipes seems to be a promising option. However, one of the main

10


problems with these special nano�uids is their proper stabilization. In addition, the mechanisms responsible for the enhanced thermal performance discussed for ordinary nano �uids might be different in the case of CNT-nano �uids.

increase of the evaporator heat transfer coef �cient. This result is in agreement with the analytical model by Alizad et al. [66].

6. Effects with respect to thermal performance of gadgets

Thequestion if thecondenser heat transfer coef �cient is changed when nano�uids employed is basically equivalent with the question if nanoparticles are torn out of the working � uid and transported by the vapour phase or not. Intuitively one may argue that this is not the case due to the much larger particle size compared with the size of the base�uid molecules. For example, the smallest oxide nanoparticles of all experiments compiled in Table 1 have a size of 2 nme3 nm [4] and 4 nme5 nm [38]. This is about 20 e50 times larger than a water molecule which has a size of about 0.1 nm. Many experimental results [31,36,43,50] indicated that the condenser heat transfer was either not or only weakly affected by nano�uids. Qu et al. [50] analyzed SEM images of the condenser surface of an oscillating heat pipe and found nearly no changes compared to a reference surface boiled in pure water. Therefore they concluded that the thermal resistance of the condenser should not have been changed. Later, the same group (Qu and Wu [53]) found that for equal concentrations the condenser showed no particle deposit for well dispersed Al 2O3 particles, but had some precipitate for less dispersed SiO 2 particles of similar primary size. Nanoparticle deposits on the condenser were also found by Han and Rhi [22]. However, the structure of this nanoparticle layer was different than the one found on the evaporator. While on the evaporator surface the depositions were planer and of large size, the deposition shape on the condensersurface looked like spherical structures. In this special experiment the thermal performance of the gadget was decreased with dispersed nanoparticles. An interesting result was reported by Riehl and dos Santos [54]. These authors investigated the temperature �uctuations at evaporator, condenser, and in the adiabatic part of an open loop pulsating heat pipe employing H 2O/CuO nano�uid (5 vol. %). For the nano�uid the �uctuations in all three components were signi �cantly increased especially at high heat inputs. It was argued that due to the nanoparticles more bubbles were generated, which led to more vapour in the pipe. Therefore it seemed that the condenser as such did not contribute to the improved thermal performance of this pulsating heat pipe. Besides the majority of experiments which indicated that the condenser was more or less not affected by nano�uids, a group from Shanghai Jiaotong University (China) showed the opposite. Liu and Zhu [23] analyzed a horizontal mesh heat pipe employing H2O/ CuO nano�uid with 50 nm nanoparticle size. The condenser heat transfer coef �cient increased signi�cantly with increasing concentration but reached a maximum at about 1 vol. % and then fell again. Similarly, Wang et al. [24] found that the temperature distribution at the condenser was slightly larger for the H 2O/CuO nano�uid than for pure water. The condenser heat transfer coef �cient of the horizontal case was increased by about 35% at a concentration of 1 wt. %. This was signi�cantly lower than the enhancement of the evaporator heat transfer coef �cient. It should be noted that the two studies [23,24] indicating an increase of the condenser heat transfer coef �cient were carried out with horizontal meshed heat pipes. To conclude, most studies indicate that the condenser is not or only weakly affected by nano�uids; however, some experiments see also strong effects. No study analyzed here investigated the vapour phase with respect to the transport of nanoparticles directly. Therefore, the question remains open as to if nanoparticles are torn out from the working � uid and transported by the vapour phase during heat pipe/thermosyphon operation. This question can be answered only with adequate experiments.

6.1. Evaporator

It seems to be that one of the major physical phenomena responsible for the decrease of the thermal resistance is a porous layer built by nanoparticles on the evaporator surface. Therefore it is clear that evaporator temperature and evaporator heat transfer coef �cient take centre stage of many investigations. Indeed, with one exception (Han and Rhi [22]), all studies indicate a decrease of evaporator temperature depending on concentration and heat input. Investigating a conventional grooved (211 mm wide  217 mm deep) cylindrical heat pipe, Kang et al. [13] found that adding 1 ppm Ag-nanoparticles to pure water lowered the temperature at the evaporator by about 1%. Even stronger effects were reported for water-based Al2O3 (Putra et al. [29], Noie et al. [34]) and CuO (Liu et al. [31], Senthilkumar et al. [20], Wang et al. [24]) nano�uids. Putra et al. [29] showed a signi�cant reduction of the evaporator temperature for H2O/Al2O3 nano�uid decreasing nearly linearly with concentration (see Ref. [29] Fig. 7). The analytical model by Alizad et al. [66] con �rmed this � nding by showing a decrease of the temperature difference between evaporator and condenser up to the maximal investigated concentration of 8 vol. %. For H2O/CuO nano�uid Liu et al. [31] found that the evaporator heat transfer coef �cient is more than doubled for the optimal concentration of 1 wt. %. The surface temperature of the evaporator was lowered byabout 13% when H2O/CuO nano�uid(100 mg/l) was employed instead of pure water (Senthilkumar et al. [20]). This temperature was again decreased when 2 ml/l of n-Butanol was added to the nano�uid. For classical working �uids the evaporator temperature corresponds to the applied heat input in an essentially linear manner.3 This might not be the case for nano �uids. The investigation by Lu et al. [40] clearly showed that it was not the heat input alone but, rather the combination of heat input and the characteristics of the employed nano�uid that matters. Lu et al. [40] investigated an open thermosyphon employing CuO-nano �uid. At low input power the temperature distribution along the evaporator tube was nearly unchanged when nanoparticles were added to the base�uid. However, above a certain threshold the evaporator temperature became lower for the nano�uid. For example, at a power input of about 1600 W the temperature decreased from 96  C (DI-water) to 90  C for the nano�uid with a concentration of 1.2 wt. %. Evaporator heat transfer coef �cient depended also stronglyon operation temperature and heat input. For lowheat �ux (up to about 3200 W/m2) the increase of this parameter compared with the values for pure water was negligible. However, for higher heat input the increase was signi �cant and reached values up to 20% (see Ref. [40] Fig. 10). Similarly, Wang et al. [24], when employing a H2O/CuO nano�uid, reported an increase of the evaporator heat transfer coef �cient up to 3e4 times depending on operation temperature and heat input. No or only a weak dependence on the inclination angle was found. To summarize, nearly 90% of all experimental results indicate either a signi�cant decrease of the evaporator temperature or an

3

The author thanks one of the reviewers for pointing out that fact and therefore elucidating the differences to nano �uids.

6.2. Condenser


11

6.3. Thermal resistance of whole gadget

Thermal resistance is one of the main characteristics of thermosyphons, heat pipes, and oscillating heat pipes. Therefore, most authors try to show that the thermal resistance of their gadget is lowered when exchanging classical working �uids with nano�uids. Already in an early work, Tsai et al. [10] found that the thermal resistance of a heat pipe wassigni�cantly lowered when employing H2O/Au nano�uid. However, the main decrease occurred at the evaporator (between 33% and 56%) while the condenser section showed only a decrease between 7% and 20%. This result was con�rmed by Park et al. [11] for H2O/Ag nano�uid, who found that the thermal resistance of their heat pipe was slightly decreased at a concentration of 10,000 ppm. Much smaller concentrations (5, 50, and 100 ppm) were investigated by Chen [17]. Here the results indicated that a lowering of the thermal resistance by about 50% was already achieved with 5 ppm. When adding Al2O3 nanoparticles to pure water, a decrease of the thermal resistance of a wickedheat pipe up to 40% wasreported by Do et al. [18], and depending on the heat input up to 78% (10 W), 84% (20 W) and 88% (30 W) by Putra et al. [29]. Huminic et al. [38] employed comparably small Fe 2O3-nanoparticles (4 nme5 nm) at high concentration (5.3 vol. %) and found that at an inclination angle of 45 the thermal resistance was decreasedby about 10% and at 90 by 42% (see Ref. [38] Fig. 9). Similarly, Manimaran et al. [28] found that, when varying � lling ratio and inclination angle,the thermal resistance of their heat pipe was in all cases lower for the H 2O/CuO nano�uid (20 nm) than for the reference �uid water. Senthilkumar et al. [20,21] found for a wire mesh wicked heat pipe that the thermal resistance was lower for H2O/CuO nano�uid than for pure water. However, with increasing inclination angle and increasing heat supplied this effect disappeared. At about 60  e70 , depending on the amount of supplied heat, the thermal resistance of both �uids was equal. These effects were weaker for n-Butanol based nano�uid. Surprisingly no differences between the pure aqueous solution of n-Hexanol and the CuO/aqueous solution of n -Hexanol nano�uid were found. A dependence of the thermal resistance on the nanoparticle concentration was reported by Liu and Zhu[23]. Thermal resistance increased with increasing concentration up to 1 vol. %. However, for higher concentrations it began to decrease again. The investigated horizontal mesh heat pipe could be run with DI-water up to a heat 2 �ux of 44 kW/m , but failed to work for higher � uxes. Employing nano�uid (H2O/CuO) the operation range was enlarged signi�cantly keeping the lower thermal resistance. A similar enlargement of the operation range was reported by Liu et al. [31] for a vertical miniature thermosyphon employing the same nano �uid. Here the maximum heat input at the evaporator was increased from 200 W (DI-water) up to 400 W (H2O/CuO). Some values of thermal resistance reduction are compiled in Fig. 6. This plot shows only data which are given by the original authors as numerical values employing vol. % for the concentration. Results given in wt. % or ppm are omitted because it is not possible to convert these values into vol. %. This is due to the fact that even if the base�uid of a nano�uid is known, its density is changed in the most cases by adding stabilizers, surfactants or the like to the nano�uid [3]. Fig. 6 does not show signi �cant trends. The reason for the large scatter is basically the manifold nano�uid properties. Therefore this plot provides only an initial notion of how thermal resistance depends on nano�uid characteristics. Thermal resistance reached e.g. a nearly constant minimal value above a concentration of 1 vol. % in a circularheat pipe (Mousa [25]) and a thermosyphon (Mousa[37]). The achieved reduction depended on the heat imprinted at the evaporator and varied between 50% and 65% (circular heat pipe,

Fig. 6. Experimentally determined reduction of thermal resistance. Full grey curves Mousa [37]  12%. Broken grey curves Mousa [37] extended above 1.2 vol. %.

40 We60 W) and 50% and 68% (thermosyphon, 10 We60 W). Mousa [37] proposed for the percentage reduction the following empirical relation

RR ¼ 0:844  0:3042

(1)

Equation (1) has a relative error of  12% according to its originator. It is in reasonable agreement with results from Senthilkumar et al. [20,21] and Manimaran et al. [28]. However, it should not be applied to concentrations higher than 1.2 vol. %, as Fig. 6 clearly indicates. Similar as for all other parameters discussed so far, negative effects on the thermal resistance caused by nano�uids were reported. Mehta and Khandekar [32] and Khandekar et al. [33] investigated water-based nano�uids (Al2O3, CuO and Laponite) and found a signi�cantly increased thermal resistance compared to the reference �uid water. In a more recent work Han and Rhi [22] investigated a grooved heat pipe (grooves are 1 mm in depth, 1 mm width on top, and 1.3 mm in bottom) employing water-based Ag- and Al2O3-nano�uids. In those cases an increase of the thermal resistance was found. The highest thermal resistance is obtained for a hybridnano�uid consisting of Ag (0.1 vol. %) andAl2O3 (0.005 vol. %). The analytical model by Shafahi et al. [73] showed that the thermal resistance was lower the smaller the particles were or the higher the concentration of those particles was. For concentrations up to 4 vol. % no pronounced optimum was found for water-based Al2O3, CuO, and TiO 2 nano�uids with particle sizes of 10 nm, 20 nm and 40 nm. To summarize, the majority of the studies indicate a decrease of thermal resistance of the whole gadget. However, the larger part of thisdecrease occursat the evaporator.Some negative results blurthe picture and indicate an increase of the thermal resistance due to nano�uids. Additionally, contradictory results with respect to the dependence of the thermal resistance on the imprinted heat at the evaporator are reported (Putra et al.[29], Senthilkumar et al. [20,21]). Obviously the thermal resistance does not only depend on nano�uid characteristics alone but also on heat input, inclination angle and �lling ratio. Therefore it is dif �cult to value the in �uences of the different nano�uids in a straightforward manner. Detailed studies varying the different parameters successively are urgently needed. 6.4. Speci �cs of oscillating heat pipes

In 1990 Akachi [75] patented the � rst oscillating heat pipe. An oscillating heat pipe consists typically of a serpentine tube of capillary diameter which is evacuated and partly �lled with

12


working �uid. Due to surface tension slugs of liquid intersperse with vapour bubbles [7,8]. Heating an OHP at the evaporator leads toa bubble growth which inturnmovesthe liquidto the cooler part of the OHP. Heat is transported via the sensible heat of the liquid phase and via the latent heat of the vapourphase. The �lling ratio of OHP is differently as for thermosyphons and heat pipes de �ned as the ratio of the volume of the working � uid to the total inner volume of the device. Advantages of OHPs are, among others, the enforcement of heat transfer additionally to the phase change by the oscillatory motion and that liquid and vapour �ow do not interfere [46,51]. General information with respect to OHPs employing classical working � uids can be found e.g. in Karimi and Culham [76] and Khandekar et al. [77]. Ma et al. [46] were probably the �rst who investigated nano�uids in an oscillating heat pipe. They found that by adding 1 vol. % diamond nanoparticles to DI-water, the thermal resistance was reduced by more than 40%. However, with increasing heat input this decrease degenerated and reached a value of about 14% at 336 W, the maximal heat input investigated. An increase of the thermal resistance was also observed for a heat input below about 30 W (see Ref. [46] Fig. 3). Cheng et al.[49] investigated a �at plate oscillating heat pipe (for design details see Ref. [49] Fig. 1) employing diamond and gold nano�uids. When adding 0.1 vol. % diamond nanoparticles to acetone, no differences between the two working �uids were found except the power limit of the FP-OHP, which was extended dramatically from about 190 W to 360 W. Only a slight decrease of thermal resistance compared with pure water was found for the 0.0003 vol.% H2O/Au nano�uid. When charged with diamond water nano�uid of 1.0 vol. %, an increase of the thermal resistance of the FP-OHP was observed. According to the authors of this study, one of the reasons for this behaviour might have been that the comparably high concentration was not stable, settled, and led to clogging. An OHP was also analyzed by Ji et al. [51]. Here Al 2O3-particles sizes of 50 nm, 80 nm, 2.2 m m, 20 m m were employed. It was found that particles in general lower the startup temperature. The startup temperature was understood as the temperature when the oscillatory motion in the OHP began. Especially the nanoparticles signi�cantly reduced this characteristic temperature (pure water: 54.2  C, nano�uid 50 nm: 40.6  C). In addition, thermal resistance was reduced signi�cantly by adding nanoparticles. However, it seems that the size of the nanoparticles (50 nm or 80 nm) did not have a signi�cant in�uence on the startup temperature and the thermal resistance. Riehl and dos Santos [54] investigated an open loop oscillating heat pipe in vertical, horizontal, and top down positions (evaporator above condenser). In all three orientations the evaporation temperature of the nano�uid � lled OHP was lower than the reference case with DI-water. It was found that the plug/slug formation was more dynamic meaning more liquid was believed to be pumped and kept the evaporator temperature lower than for the working �uid DI-water. Additional the authors found that the temperature � uctuations at evaporator and condenser were much more pronounced when employing the nano �uid. 6.5. Long term tests

Despite its importance for practical application of nano �uids, their long term stability is still a rarely analyzed subject. Only a few experiments have been noted so far. Among them are the pipe �ow experiments by Walsh et al. [78]. These authors found that a fresh H2O/Al2O3 (10  5 nm) nano�uid showed a HagenePoiseuille pro�le in laminar pipe �ow indicating Newtonian behaviour. However, repeating the experiments two weeks later, that changed to a clear non-Newtonian behaviour. Obviously the nano�uid

underwent an ageing. According to the authors, the most probable explanation for that �nding was the formation of aggregates which changed the nano�uid from Newtonian to non-Newtonian. However, an interaction with the material of the test rig might also have played a role, as the nano �uid might not have improved with an anticorrosive. Liu et al. [31] repeated their thermosyphon experiments also about two weeks after the �rst runs. The general �nding was that the results had basically not changed. The authors argued that the suspension was refreshed due to the movement of the working �uid caused by the boiling process. However, keeping in mind that a practical thermosyphon or heat pipe, for example for cooling an electronic device, would have to work several months to years, these test periods are far too short to draw conclusions from. The buildup of the porous layer is a time dependent process. Therefore ageing and changing over time are natural. White [79] investigated the enhancement of boiling surfaces using nano�uid particle deposition. He showed that nano �uid boiling performance was increased by about 25% initially before the deposited nanoparticle layer was completely formed. After this process had �nished the nano�uid boiling coef �cient declined again due to suppression of bubble nucleation and transport. These results show clearly that the time dependent character of the porous layer has to be considered in more detail. 7. Modelling and prediction

Models can basically be split into two groups. The �rst group comprises engineering approaches which are more or less based on semi-empirical correlations. Hereto belong also approaches which order experimental data in a way to create such correlations. Models solving mass, momentum, and energy conservation equations, either analytically or numerically, in combination with evaporation and condensation models build the second group. 7.1. Semi-empirical models

An example for the ordering of experimental results to obtain semi-empirical correlations was given by Paramatthanuwat et al. [35]. These authors provided a correlation for the Kutateladze number (ratio of heat �ux to critical heat �ux) in dependence of ten similarity numbers, among them Prandtl, Grashof, Archimedes, Weber and Froude number. The correlation was in reasonable agreement with their data. Leong et al. [80] developed a model for the prediction of an air to air thermosyphon heat recovery exchanger. The model was based on semi-empirical correlations for the heat transfer coef �cients at evaporator and condenser. It took the thermal resistances of the airside, the thermosyphon wall, the working �uid at evaporator, and at the condenser into consideration. Thermal conductivity and viscosity of nano�uids were considered as temperature and concentration dependent. It was assumed that nanoparticles were transported by vapour and the condensate had the same characteristic as the working �uid. As discussed above, it is rather open as to if nanoparticles are transported by the vapour phase. For this reason, the assumption by Leong et al. [80] that the condensate has the same quality as the working �uid (mainly effective thermophysical properties caused by nanoparticles) is rather surprising. One of the main results of the predictions was that the change of nano�uid thermophysical properties was not of great importance for heat transfer in the analyzed thermosyphon. 7.2. Models based on conservation equations

Do and Jang [63] provide two models to study the in �uence of effective thermophysical properties and the in�uence of the porous


layer built on the evaporator separately. Both modelswere based on continuity and momentum equations for vapour and liquid phase. Additionally, the energy equation and the YoungeLaplace equation (capillary pressure difference between liquid and vapour) were solved numerically. The simulations indicated an optimum nanoparticle concentration. The authors argued that the optimum was caused by two competing effects. One was the extension of the evaporator surface due to nanoparticle deposition and the other was the reduced evaporation heat transfer rate due to a higher pressure drop in the porous layer built by the nanoparticles. Moreover, it was shown that the thermal resistance of the heat pipe decreased with increasing nanoparticle size, which was in agreement with the experimental results of Kang et al. [15]. An analytical model for the prediction of the thermal performance of �at-shape heat pipes operated by nano�uids was provided by Shafahi et al. [74]. A similar modelwas applied to cylindrical heat pipes [73]. The model was based on descriptions of pressure, velocity, and temperature distributions inside the heat pipe. Pressure distribution was derived by integrating momentum equation and temperature distribution following from energy balances at evaporator and condenser. The velocity distribution in the wicks followed from Darcy s law and the mass balance. The YueChoi model for the thermal conductivity of nano �uids (Yu and Choi [81]) was modi�ed to account forthe porous layer. Oneof themain �ndings of this model was that there exists an optimum for wick thickness and nanoparticle concentration. The �rst �nding was unique in the sense that none of the experiments address the wick characteristic as an independent parameter. An extension of the Shafahi-Biancoe VafaieManca model [73,74] for the startup characteristics of �atshaped heat pipes was proposed by Alizad et al. [66]. Modelling and simulation are a priori condemned to failure if the physical mechanisms causing the experimentally found improvement of the thermal performance of thermosyphons and heat pipes operated with nano�uids are not understood. The statement that any model fails if the physics behind an observation are either not known or only poorly understood is a general one. An extraordinary example is the debate of the last decade with respect to wall-bounded turbulence on � at plates and in channel and pipe �ows (for an overview see Buschmann and Gad-el-Hak [82]). Therefore, all models for thermosyphons and heat pipes mentioned above bear some preliminaryassumptions.This does notnegate the fact that the majority of the physical, mathematical and numerical components of these models are correct and represent the state of the art. In all cases, the results depend strongly on the assumptions made with respect to thermophysical properties [3]. Sensitivity studies carried out for forced [83] and natural convection [84] for different ceramic/water nano�uids con�rm this. However, the models available so far capture main features of nano�uids in thermosyphons and heat pipes found experimentally. ’

8. Possible mechanisms

In essence, nano�uids are two-phase liquids. Their behaviour is in many cases not comparablewith classical single-phase liquids like water or acetone. Relevant physical processes occur on micro and macro length scales and on different time scales. In the case of phase change, additionally signi�cant interactions between free surfaces of apparatus and nano�uids have to be considered. The huge number of parameters characterizing nano�uids e particle material, size, shape; base�uid characteristics etc. e makes the drawing of �rm conclusions even more complicated. Thus, � nding explanations for the aforementioned enhancements or deteriorations of heat transfer performance of thermosyphons and heat pipes is extraordinary challenging. Therefore, the explanations given below may appear partly vague and seem not to address the underlying physical

13

phenomena completely. However, as already stated, there are many open issues with respect to nano�uids and the understanding of the underlying physics is partly poor. This may also explain the phenomenon that several authors having found positive results but do not try to explain the physical mechanism behind them. Rather, there are usually the research teams which found deterioration look for clari�cation. Here mainly discussions which are given by different authors analyzing their experimental data are re�ected. Only minor valuing comments are made. The explanation attempts for improvements discussed by several authors can be categorized according to the following assumed underlying mechanisms:  nano�uids improve thermal performance due to enhanced thermal conductivity  nano�uids enlarge operation range due to much higher critical heat � ux  bubble departure frequency at evaporator is increased due to bubble bombardment by nanoparticles  nanoparticles coat the wick, which enlarges heat transfer area or capillary forces or both of them  the often found concentration optimum follows from counter acting forces  a porous nanoparticle layer forms on the evaporator, which changes wettability

Several authors ascribe improvements to more than one of these reasons. It is only logical to seek explanations for positive effects in a similar way as done for other heat transfer applications employing nano�uids, such as pipe � ow (Prabhat et al. [4]). For example, Kang et al. [13], Noie et al. [34], and Mousa [25] argued that nano�uids can �atten temperature gradients due to the increased effective thermal conductivity and therefore the global thermal resistance of thermosyphons and heat pipes should be reduced. However, due to the usually low concentration of nanoparticles utilized and the therefore weak changes of thermophysical properties, large alterations, e.g. of the thermal resistance, cannot be explained that way. You et al. [85] showed that, by immersing a polished copper plate as heater surface in a H 2O/Al2O3 nano�uid (0e0.04 g/l), the critical heat �ux was increased up to 200% compared with pure water. This effect may explain the increased operation range found by Liu et al. [31], Liu et al. [36], and Liu and Zhu [23]. Likewise, Wei et al. [12] argued that critical heat �ux and convective heat transfer coef �cient of their nano�uids were higher than for the pure base�uid. Therefore it was expected that the thermal performance of a heat pipe with nano�uid as working �uid was better than employing pure water. Already Tsai et al. [10] argued that a large component of the thermal resistance of their heat pipe was due to the formation of vapour bubbles at the surface of the evaporator. It was then expected that the particles suspended in nano �uids bombard the bubbles during formation and, therefore, the nucleation size of the bubbles became much smaller, which in turn should have reduced the global thermal resistance of the heat pipe. Huminic et al. [38] argued similarly that the reduction of the thermal resistance of their thermosyphon was caused by a bombardment of the vapour bubbles during bubble formation. Likewise, Shanbedi et al. [45] used that argument to explain the enhancement of thermal performance by MWCNT in their thermosyphon. Do et al. [18] inspected the surface of the wick after employing H2O/Al2O3 nano�uids of 1 vol. % and 3 vol. % utilizing an optical microscope. In both cases the wick was covered with a thin porous coating consisting obviously of nanoparticles. They argued that this coating provided additional heat transfer area for evaporation. A

14


coating on the screen mesh surface consisting of nanoparticle material was also found by Putra et al. [29]. However, differently to Do et al. [18], this research team concluded that the coating provided improved capillary effects. Wannapakhe et al. [48] argued that the found optimum at a concentration of 0.5 wt. % was caused by the interplay of thermophysical properties, mainly dynamic viscosity and thermal conductivity. Higher particle concentrations increased viscosity signi�cantly and therefore obstructed bubble generation and liquid movement in their oscillating heat pipe. Qu et al. [50] employed the theory by Mikic and Rohsenow [86] to explain the found optimum. According to this approach the thermal resistance due to nucleate boiling at the evaporator can be written as

Re ¼

1 q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi 2

(2)

2N b Db f pknf rnf c p;nf

Due to the low concentration the product k nf r nf c p,nf is comparably less changed compared to the base �uid. It was rather the increase of the active nucleation site density N b, the bubble release diameter db, and the bubble release frequency f , that decreased the thermal resistance at the evaporator. Qu et al. [50], utilizing twodimensional Atomic Force Microscopy (2d-AFM) images, con�rmed that Al2O3-nanoparticles had deposited on the evaporator and modi�ed its surface completely. A nanometre-roughness was formed on top of the original micrometre-roughness. This new surface quality in�uenced bubble release frequency and bubble diameter signi�cantly. The level of particle agglomeration seemed to be increased by concentration, which in turn would enlarge structures of the nanometre-roughness. From that observation it appeared to be clear to the authors that an optimal concentration with respect to the decrease of the thermal resistance of the evaporator should exist. Liu and Zhu [23] investigated a cylindrical wicked heat pipe. While a very thin porous layer consisting mainly of cooper and oxygen, the two components of the CuO-nanoparticles, was found on the inner wall of the evaporator, almost no deposits were found on the mesh surface. An extremely thin porous layer on the inner side of the evaporator after employing nano�uid was also found by Lu et al. [40] in their open thermosyphon. Liu and Zhu [23] concluded from their experiments that the improvement mainly followed from the changed thermophysical properties, including decreased surface tension and solid eliquid contact angle. Additional reasons for the enhanced heat transfer may have occurred from the increased effective thermal conductivity and the random Brownian motion of the nanoparticles. It should be noted that in this special experiment the thermal performance of the heat pipe was dominated by the convective heat transfer in the liquid � lm. Supporting explanations for the experimental �ndings came from the two models by Do and Jang [63]. For the test cases with H2O/Al2O3-nano�uid, both models indicated a reduction of the evaporator temperature. However, this effect was signi�cantly weakerwithout the porous layer (ModelI) than with (ModelII). The lowering found by Model I was mainly explained by the increased thermal conductivity, which allowed for an improvement of the evaporating heat transfer in the liquid �lm. One of the major �ndings of these simulations was that the porous layer on the � n tops provides an additional evaporation surface which led to an almost doubling of heat transfer. However, the performance optimum was due to competing effects which followed from increasing concentration. As the particle concentration increased, the pressure drop of the growing porous layer became larger and consequently thermal resistance was enforced. Minimal thermal resistance was achieved at nanoparticle concentrations of 0.3 vol. % (40 W) and 0.8 vol. % (80 W).

As already discussed in the foregoing sections, not all experiments found an improvement of the thermal performance of thermosyphons and heat pipes. Han and Rhi [22] indicated a general deterioration of thermal performance of their grooved heat pipe. They analyzed the �uctuation (amplitude between low and high peak temperature) at evaporator and at condenser. These �uctuations were more intense at the condenser than at the evaporator. The authors argued that this was due to the abrupt release of the heat transported by the nanoparticle in the condenser. The whole effect might have been intensi�ed by particle agglomeration. Experiments with pure water were carried out by Han and Rhi [22] after nano�uid experiments. The results indicated that the performance of the heat pipe was the same as with nano�uids. It seemed to be that the porous layer established during nano�uid experiments remained and acted further in following experiments. The shown SEMphotos from different positions of the inner surface of the heat pipe clearly indicated deposits of nanoparticles everywhere. However, the quality of the deposits varied along the pipe. While in the evaporator the shape of the deposition was planer in the condenser, the deposition looked more like spherical particles. Increased thermal �uctuations were also observed by Riehl and dos Santos [54] and Riehl [55] in an oscillating heat pipe. Riehl and dos Santos [54] calculated the critical bubble diameter for water 5.45  104 m and for nano�uid (H2O/ Cu, 5 wt. %) 5.33  104 m at 20  C. Differently to Han and Rhi [22], the authors argued that the smaller critical bubble diameter directlycontributed to the increase of vapour bubble formation and thus the pulsations were more intensive with nano�uid. Additionally, it has to be kept in mind that Han and Rhi [22] employed a grooved heat pipe and that therefore the physical reason for the increased thermal � uctuations might be different in the two cases. Mehta and Khandekar [32] and Khandekar et al. [33] drew basically two conclusions from the found deterioration of the thermal performance. First, the adverse effect of nanoparticles on pool boiling was stronger than any effect which may have followed from an increased thermal conductivity and, second, the interplay between nanoparticles and heater surface affected pool boiling negatively. Mehta and Khandekar [32] argued that most models describing nucleate boiling heat �ux show a power law dependence on nucleation site density n a and superheating (T w  T sat ).

q00 ¼ n xa ðT w  T sat Þ y

(3)

The nucleation site density is affected by physical adhesion of nanoparticles and therewith changing cavity mouth radius and/or cavity mouth angle and by changing the wetting characteristics. A simple model showed that the necessary superheat indeed can increase when the cavity mouth radius is decreased due to nanoparticle deposition. In Khandekar et al. [33] the investigation was extended with respect to the initial surface roughness (Ra ¼ 0.24 m m). The surface interaction parameter (ratio of surface roughness to particle size), according to Harish et al. [64], for the CuO particles (8.6 nme13.5 nm) would then be between 17.8 and 27.9 and for the Al2O3 particles (40 nme47 nm) between 5.1 and 6.0. In both cases this would indicate an enhancement of heat transfer (see Ref. [64] Fig. 11). However, as discussed by Khandekar et al. [33], the particles may have undergone agglomeration under the boiling process. This in turn would have lowered the surface interaction parameter signi�cantly and therewith the expected heat transfer enhancement. 9. Summary

Despite their seemingly simple design, thermosyphons, heat pipes, and oscillating heat pipes are complex heat transfer devices.


Numerous parameters affect their thermal performance already when classical one-phase liquids are employed as working � uids. The number of relevant parameters becomes even larger when nano�uids are brought into action. Additional issues occur due to the diminutiveness of the particles and their interaction with the surfaces of these gadgets. Not all phenomena, such as the effective thermophysical properties, especially viscosity, surface tension, and thermal conductivity and a signi�cantly increased critical heat �ux, are still explainable using the trusted continuum approaches. The buildup of porous layers forced by boiling processes reaches far beyond �uid mechanic and thermodynamic knowledge. This complexity explains why the experimental and modelling work carried out so far is partly contradictory and does not answer all physical and technical questions relevant for the proper design of thermosyphons, heat pipes, and oscillating heat pipes operated with nano�uids. However, based on the examination of 38 experimental studies and 4 modelling approaches, the following conclusions are drawn.  The majorityof the studies indicate that �lling ratio, inclination angle, and operation temperature affect thermosyphons and heat pipes operated with nano�uids in the same or at least in a similar way as classical � uids do. However, a few experiments also indicate speci �c trends for nano�uids.  Only three,but very clear, experiments indicate the in�uence of the base�uid on the thermal resistance.  It is highly likely that nanoparticle size, shape,and material and chemical stabilization have an in �uence on the thermal performance. However, these in�uences seem to be different than for heat transfer applications without phase changes. Especially the interaction between evaporator surface and wick surface on one side and nanoparticles on the other side has to be take into consideration.  Optima of the thermal performance of thermosyphons and heat pipes in dependence on the concentration were found for Ag, TiO2 and CuO nanoparticles dispersed in different basefluids. Surprisingly, no optimum was found in thermosyphons and heat pipes by several independent research groups dealing with H2O/Al2O3 nano�uids.  Magnetic nano�uids and nano�uids with carbon nanotubesare promising options because they allow the use of additional effects like the application of additional external forces or the extraordinary high thermal conductivity along the main axis of the CNT. However, they are still far from practical use because the number of studies is far too less to create reliable knowledge with respect to the behaviour of these speci �c liquids in thermosyphons and heat pipes.  Nearly 90% of all experimental results indicate either a significant decrease of the evaporator temperature or an increase of the evaporator heat transfer coef �cient.  The majority of the studies indicate a decrease of thermal resistance of the whole gadget. The larger part of this decrease occurs at the evaporator. However, because thermal resistance does not only depend on nano �uid characteristics alone, but also on heat input, inclination angle and � lling ratio it is dif �cult to value the in �uences of the different nano�uids in a straightforward manner.  Most studies indicate that the condenser is not or only weakly affected by nano�uids. However, some experiments see also strong effects. The question if nanoparticles are torn out from the working � uid and transported by the vapour phase is still open.  Several experimental studies show that working �uids with nanoparticles allow an operation at higher input powers without dryout.

15

 Despite its strong importance for practical applications, long term stability of nano�uids and long term operation behaviour of gadgets operated with nano�uids is not investigated in detail so far.  First modelling approaches available cover main features of nano�uids in thermosyphons and heat pipes found experimentally. However, as long as some physical mechanisms are not understood more detailed these models will lack generality.

The buildup of porous nanoparticle layers on the evaporator surface or on the wick surface seems to be one of the key features related with nano�uids employed in thermosyphons and heat pipes. Interestingly enough, different research teams tried to explain the found improvements, as well as the found deteriorations, utilizing these nanoparticle layers. The general effects related to such layers which signi �cantly improve surface wettability, as shown by a reduction of the static contact angle, were investigated by Kim et al. [87]. The reduction of the static contact angle in turn is due to the changed surface energy and surface morphology caused by the nanoparticle layer. However, these experiments were not carried out in the more complex thermosyphons and heat pipes.Here different effects mayact simultaneously or even interfere. There are several open questions with respect to the porous layer formed from nanoparticles. Among them is the question if these layers could be built without nano �uids and if such arti�cial layers would be sustainable without renewal by nanoparticles from a nano�uid employed as working liquid. Solomon et al. [88] coated a wick with nanoparticles by simply immersing the wick into a nano�uid and drying it afterwards. Employing pure water as working �uid they found a decrease of the thermal resistance up to 19%. Here much more detailed experiments are needed. To summarize nano�uids are a new and promising option as working � uids in thermosyphons, heat pipes, and oscillating heat pipes. The majority of the experimental works carried out so far indicates a lowering of the thermal resistance and therewith a signi�cant improvement of the thermal performance. Several general questions with respect to gadget parameters seem to be answered. However, many questions with respect to nanoparticle characteristics, optimization, and acting physical mechanisms are still open. Despite promising modelling approaches existing, the limited knowledge with respect to the dependencies of the thermal performance on nano�uid characteristics prevents the development of reliable design tools for practical application. In any case, more detailed experiments which address speci �c questions of nano�uids employed as working �uids in thermosyphons, heat pipes, and oscillating heat pipes are needed. Acknowledgement

This work was supported by BMWi under the grants MF 110030 and MF 120019. References [1] A. Sergis, Y. Hardalupas, Anomalous heat transfer modes of nano�uids: a review based on statistical analysis, Nanoscale Res. Lett. 6 (2011) 391. http:// dx.doi.org/10.1186/1556-276X-6-391. [2] S. Thomas, C.B.P. Sobhan, A review of experimental investigations on thermal phenomena in nano �uids, Nanoscale Res. Lett. 6 (2011) 377. http://dx.doi.org/ 10.1186/1556-276X-6-377. [3] S. Feja, M.H. Buschmann, Nano�uids e potentials illusions, Adv. Comput. Methods Exp. Heat Transf. XII (2012) 195e205. [4] N. Prabhat, J. Buongiorno, L.W. Hu, Convective heat transfer enhancement in nano �uids: real anomaly or analysis artifacts?, in: Proceedings ASME/JSME 2011 8th AJTEC, USA, 2011.

16


[5] M.H. Buschmann, Thermal conductivity heat transfer of ceramic nano �uids, Int. J. Therm.Sci. 62 (2012) 19e28. http://dx.doi.org/10.1016/j.ijthermalsci.2011.09.019 . [6] Z.H. Liu, Y.Y. Li, A new frontier of nano�uid research e application of nano�uids in heat pipes, Int. J. Heat Mass Transf. 55 (2012) 6786 e6797. http:// dx.doi.org/10.1016/-j.ijheatmasstransfer.2012.06.086. [7] D.A. Reay, P.A. Kew, Heat Pipes, � fth ed., Elsevier, 2006. [8] A. Faghri, Heat Pipe Science and Technology, Mechanical Engineering, Taylor & Francis Group, 1995. [9] H.H. Fernholz, P.J. Finley, The incompressible zero-pressure-gradient turbulent boundary layer: an assessment of the data, Prog. Aerospace Sci. 32 (1996) 6245e6311. [10] C.Y. Tsai, H.T. Chien, P.P. Ding, B. Chan, T.Y. Luh, P.H. Chen, Effect of structural character of gold nanoparticles in nano �uid on heat pipe thermal performance, Mater. Lett. 58 (2004) 1461e1465. http://dx.doi.org/10.1016/ j.matlet.2003.10.009. [11] K.H. Park, W.H. Lee, K.W.K.W. Lee, I.H. Baek, S.H. Rhi, D.R. Shin, Study on the operating characteristics in small size heat pipe using nano �uids, in: Proceedings 3rd IASME/WSEAS Int. Conf. on Heat Transfer, Thermal Eng. Environment (2005), pp. 106e109. [12] W.C. Wei, S.H. Tsai, S.Y. Yang, S.W. Kang, Effect of nano �uid concentration on heat pipe thermal performance, IASME Trans. (2005) 1432 e1439. [13] S.W. Kang, W.C. Wei, S.H. Tsai, S.Y. Yang, Experimental investigation of silver nano-�uid on heat pipe thermal performance, Appl. Therm. Eng. 26 (2006) 2377e2382. http://dx.doi.org/10.1016/j.applthermaleng.2006.02.020. [14] P. Naphon, P. Assadamongkol, T. Borirak, Experimental investigation of titanium nano�uidson theheatpipe thermalef �ciency, Int.Commun.Heat MassTransf.35 (2008) 1316e1319. http://dx.doi.org/10.1016/j.icheatmasstransfer.2008.07.010. [15] S.W. Kang, W.C. Wei, S.H. Tsai, C.C. Huang, Experimental investigation of nano �uids on sintered heat pipe thermal performance, Appl. Therm. Eng. 29 (2009) 973e979. http://dx.doi.org/10.1016/j.applthermaleng.2008.05.010. [16] P. Naphon, D. Thongkum, P. Assadamongkol, Heat pipe ef �ciency enhancement with refrigerant-nanoparticles mixtures, Energy Conversion Manage. 50 (2009) 772e776. http://dx.doi.org/10.1016/j.enconman.2008.09.045. [17] Y.T. Chen, Experimental study of silver nano �uid on �at heat pipe thermal performance, J. Marine Sci. Technol. 18 (2010) 731 e734. [18] K.H. Do, H.J. Ha, S.P. Jang, Thermal resistance of screen mesh wick heat pipes using the water-based Al2O3 nano�uids, Int. J. Heat Mass Transf. 53 (2010) 5888e5894. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2010.07.050. [19] T.P. Teng, H.G. Hsua, H.E. Mob, C.C. Chenc, Thermal ef �ciency of heat pipe with alumina nano�uid, J. Alloys Compd. 504S (2010) S380eS384. http:// dx.doi.org/10.1016/j.jallcom.2010.02.046. [20] R. Senthilkumar, S. Vaidyanathan, B. Sivaraman, Performance analysis of heat pipe using copper nano �uid with aqueous solution of n -Butanol, Int. J. Mech. Mater. Eng. 4 (2010) 251e256. [21] R. Senthilkumar, S. Vaidyanathan, B. Sivaraman, Experimental analysis of cylindrical heat pipe using copper nano �uid with an aqueous solution of n Hexanol, Front. Heat Pipes 2 (2011) 033004. http://dx.doi.org/10.5098/ fhp.v2.3.3004. [22] W.S. Han, S.H. Rhi, Thermal characteristics of grooved heat pipe with hybrid nano �uids, Therm. Sci. 15 (2011) 195e206. http://dx.doi.org/10.2298/ TSCI100209056H. [23] Z.H. Liu, Q.Z. Zhu, Application of aqueous nano�uids in a horizontal mesh heat pipe, Energy Conversion Manage. 52 (2011) 292e300. http://dx.doi.org/ 10.1016/j.enconman.2010.07.001. [24] P.Y. Wang, X.J. Chen, Z.H. Liu, Y.P. Liu, Application of nano �uid in an inclined mesh wicked heat pipes, Thermoch. Acta 539 (2012) 100e108. http:// dx.doi.org/10.1016/j.tca.2012.04.011. [25] M.G. Mousa, Effect of nano �uid concentration on the performance of circular heat pipe, Int. J. Sci. Eng. Res. 2 (2011) 1 e8. [26] Y.C. Chiang, J.J. Chieh, C.C. Ho, The magnetic-nano �uids heat pipe with superior thermal properties through magnetic enhancement, Nanoscale Res. Lett. 7 (2012) 322. http://dx.doi.org/10.1186/1556-276X-7-322. [27] R. Hajian, M. Layeghi, K.A. Sani, Experimental study of nano �uid effects on the thermal performance with response time of heat pipe, Energy Conversion Manage. 5 (2012) 63e68. http://dx.doi.org/10.1016/j.enconman.2011.11.010. [28] R. Manimaran, K.N. Palaniradja, N. Alagumurthi, Effect of �lling ratio on thermal characteristics of wire-mesh heat pipe using copper oxide nano �uids, Front. Heat Pipes 3 (2012) 023004. http://dx.doi.org/10.5098/fhp.v3.2.3004. [29] N. Putra, W.N. Septiadi, H. Rahman, R. Irwansyah, Thermal performance of screen mesh wick heat pipes with nano �uids, Exp. Therm. Fluid Sci. 40 (2012) 10e17. http://dx.doi.org/10.1016/j.exptherm�usci.2012.01.007. [30] H. Xue, J. Fan, Y. Hu, R. Hong, K. Cen, The interface effect of carbon nanotube suspension on the thermal performance of a two-phase closed thermosyphon, J. Appl. Phys. 100 (2006) 104909. http://dx.doi.org/10.1063/1.2357705. [31] Z.H.Liu, J.G.Xiong,R. Bao,Boilingheat transfer characteristicsof nano �uidsina �at heat pipe evaporator with micro-grooved heating surface, Int. J. Multiphase Flow 33 (2007) 1284e1295. http://dx.doi.org/10.1016/j.ijmultiphase�ow.2007.06.009. [32] B. Mehta, S. Khandekar, Two-phase closed thermosyphon with nano �uids, in: 14th International Heat Pipe Conference (2007), pp. 22 e27. [33] S. Khandekar, Y. Joshi, B. Mehta, Thermal performance of closed two-phase thermosyphon using nano�uids, Int. J. Therm. Sci. 47 (2008) 659e667. http://dx.doi.org/10.1016/j.ijthermalsci.2007.06.005. [34] S.H. Noie, S.Z. Heris, M. Kahani, S.M. Nowee, Heat transfer enhancement using Al2O3/waternano�uidin atwo-phaseclosedthermosyphon,Int. J.Heat Fluid Flow 30 (2009) 700e709. http://dx.doi.org/10.1016/j.ijheat�uid�ow.2009.03.001.

[35] T. Paramatthanuwat, S. Rittidech, A. Pattiya, A correlation to predict heattransfer rates of a two-phase closed thermosyphon (TPCT) using silver nano �uid at normal operating conditions, Int. J. Heat Mass Transf. 53 (2010) 4960e4965. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2010.05.046. [36] Z.H. Liu,X.F.Yang,G.S.Wang,G.L.Guo,In �uenceof carbonnanotube suspensionon thethermalperformanceof a miniature thermosyphon,Int.J. Heat Mass Transf. 53 (2010) 1914e1920. http://dx .doi.org/10.1016/j.ijheatmasstransfer.2009.12.065. [37] M.G. Mousa, Thermal performance of thermosyphon charged by nano �uid for cooling electronic component, Int. J. Sci. Eng. Res. 2 (2011b) 1 e9. [38] G. Huminic, A. Huminic, I. Morjan, F. Dumitrache, Experimental study of the thermal performance of thermosyphon heat pipe using iron oxide nanoparticles, Int. J. Heat Mass Transf. 54 (2011) 656e661. http://dx.doi.org/ 10.1016/j.ijheatmasstransfer.2010.09.005. [39] G. Huminic, A. Huminic, Heat transfer characteristics of a two-phase closed thermosyphons using nano �uids, Expt. Therm. Fluid Sci. 35 (2011) 550e557. http://dx.doi.org/10.1016/j.exptherm�usci.2010.12.009. [40] L. Lu, Z.H. Liu, H.S. Xiao, Thermal performance of an open thermosyphon using nano �uids for high-temperature evacuated tubular solar collectors Part 1: indoor experiment, Solar Energy 85 (2011) 379e387. http://dx.doi.org/ 10.1016/j.solener.2010.11.008. [41] T. Paramatthanuwat, S. Boothaisong, S. Rittidech, K. Booddachan, Heat transfer characteristics of a two-phase closed thermosyphon using deionized water mixed with silver nano, Heat Mass Transf. 46 (2009) 281e285. http:// dx.doi.org/10.1007/s00231-009-0565-y. [42] T. Paramatthanuwat, S. Rittidech, A. Pattiya, Y. Ding, S. Witharana, Application of silver nano �uid containing oleic acid surfactant in a thermosyphon economizer, Nanoscale Res. Lett. 6 (2011) 315. http://dx.doi.org/10.1186/1556276X-6-315. [43] X.F. Yang, Z.H. Liu, Application of functionalized nano �uid in thermosyphon, Nanoscale Res. Lett. 6 (2011) 494. http://dx.doi.org/10.1186/1556276X-6-494. [44] S.Z. Heris, H. Salehi, S.H. Noie, The effect of magnetic �eld nano�uid on thermal performance of two-phase closed thermosyphon (TPCT), Int. J. Phys. Sci. 7 (2012) 534e543. http://dx.doi.org/10.5897/IJPS11.1019. [45] M. Shanbedi, S.Z. Heris, M. Baniadam, A. Amiri, M. Maghrebi, Investigation of heat-transfer characterization of EDA-MWCNT/DI-Water nano �uid in a twophase closed thermosyphon, Ind. Eng. Chem. Res. 51 (2012) 1423e1428. http://dx.doi.org/10.1021/ie202110g. [46] H.B. Ma, C. Wilson, B. Borgmeyer, K. Park, Q. Yu, S.U.S. Choi, M. Tirumala, Effect of nano�uid on the heat transport capability in an oscillating heat pipe, Appl. Phys. Lett. 88 (2006) 143116. http://dx.doi.org/10.1063/1.2192971. [47] Y.H. Lin, S.W. Kang, H.L. Chen, Effect of silver nano- �uid on pulsating heat pipe thermal performance, Appl. Therm. Eng. 28 (2008) 1312e1317. http:// dx.doi.org/10.1016/j.applthermaleng.2007.10.019. [48] S. Wannapakhe, S. Rittidech, B. Bubphachot, O. Watanabe, Heat transfer rate of a closed-loop oscillating heat pipe with check valves using silver nano �uid as working � uid, J. Mech. Sci. Technol. 23 (2009) 1576 e1582. http://dx.doi.org/ 10.1007/s12206-009-0424-2. [49] P. Cheng, S. Thompson, J. Boswell, H.B. Ma, An investigation of �at-plate oscillating heat pipes, J. Electron. Packag. 132 (2010) 041009. http:// dx.doi.org/10.1115/1.4002726. [50] J. Qu, H.Y. Wu, P. Cheng, Thermal performance of an oscillating heat pipe with Al2O3ewater nano �uids, Int. Comm. Heat Mass Transf. 37 (2010) 111 e115. http://dx.doi.org/10.1016/j.icheatmasstransfer.2009.10.001. [51] Y. Ji, H. Ma, F. Su, G. Wang, Particle size effect on heat transfer performance in an oscillating heat pipe, Exp. Therm. Fluid Sci. 35 (2011) 724 e727. http:// dx.doi.org/10.1016/j.exptherm�usci.2011.01.007. [52] Y. Ji, C. Wilson, H. Chen, H. Ma, Particle shape effect on heat transfer performance in an oscillating heat pipe, Nanoscale Res. Lett. 6 (2011) 296. http:// dx.doi.org/10.1186/1556-276X-6-296. [53] J. Qu, H. Wu, Thermal performance comparison of oscillating heat pipes with SiO2/water and Al 2O3/water nano�uids, Int. J. Therm. Sci. 50 (2011) 1954 e 1962. http://dx.doi.org/10.1016/j.ijthermalsci.2011.04.004. [54] R.R. Riehl, N. dos Santos, Water-copper nano�uid application in an open loop pulsating heat pipe, Appl. Therm. Eng. 42 (2012) 6 e10. http://dx.doi.org/ 10.1016/j.applthermaleng.2011.01.017. [55] R.R. Riehl, private communication, 2012. [56] S. Khandekar, N. Dollinger, M. Groll, Understanding operational regimes of closed loop pulsating heat pipes: an experimental study, Appl. Therm. Eng. 23 (2003) 707e719. http://dx.doi.org/10.1016/S1359-4311(02)00237-5. [57] A.D. Patil, R.B. Yarasu, Factors affecting the thermal performance of two phase closed thermosyphon: a review, Int. J. Emerging Technol. Adv. Eng. 2 (2012) 202e206. [58] J. Buongiorno, D.C. Venerus, N. Prabhat, T. McKrell, J. Townsend, R. Christianson, et al., A benchmark study on the thermal conductivity of nano �uids, J. Appl. Phys. 106 (2009) 094312. http://dx.doi.org/10.1063/ 1.3245330. [59] A. Ehle, S. Feja, M.H. Buschmann, Temperature dependency of ceramic nano �uids shows classical behavior, J. Thermophys. Heat Transf. 25 (2011) 378e385. http://dx.doi.org/10.2514/1.T3634. [60] J. Chevalier, O. Tillement, F. Ayela, Structure rheology of SiO2 nanoparticle suspensions under very high shear rates, Phys. Rev. E 80 (2009) 051403. http://dx.doi.org/10.1103/PhysRevE.80.051403. [61] M. Corcione, Empirical correlating equations for predicting the effective thermal conductivity dynamic viscosity of nano �uids, Energy Conversion


Manage. 52 (2011) 789e793. http://dx.doi.org/10.1016/j.enconman.2010. 06.072. [62] R. Prasher, D. Song, J. Wang, Measurements of nano�uid viscosity and its implications for thermal application, Appl. Phys. Lett. 89 (2006) 133108. http://dx.doi.org/10.1063/1.2356113. [63] K.H. Do,S.P.Jang, Effectof nano �uids onthe thermal performanceof a �at micro heat pipe with a rectangular grooved wick, Int. J. Heat Mass Transf. 53 (2010) 2183e2192. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2010.07.050. [64] G. Harish, V. Emlin, V. Sajith, Effect of surface particle interactions during pool boiling of nano �uids, Int. J. Therm. Sci. 50 (2011) 2318e2327. http:// dx.doi.org/10.1016/j.ijthermalsci.-2011.06.019. [65] E.V. Timofeeva, J.L. Routbort, D. Singh, Particle shape effects on thermophysical properties of alumina nano �uids, J. Appl. Phys. 106 (2009) 014304. http:// dx.doi.org/10.1063/1.3155999. [66] K. Alizad, K. Vafai, M. Shafahi, Thermal performance operational attributes of the startup characteristics of �at-shaped heat pipes using nano �uids, Int. J. Heat Mass Transf. 55 (2012) 140e155. http://dx.doi.org/10.1016/ j.ijheatmasstransfer.2011.08.050. [67] W.M. Worek, M.N. Golubovic, H.D.M. Hettiarachchi, Nano �uids critical heat �ux, an experimental analytical study, in: 10th UK National Heat Transfer Conference (2007). [68] S.K. Das, N. Putra, W. Roetzel, Pool boiling characteristics of nano- �uids, Int. J. Heat Mass Transf. 46 (2003) 851e862. [69] M.H. Buschmann, U. Franzke, Improvement of thermosyphon performance by employing nano�uid, in: 4th IRR Conference, Paper no. TP-014. The Netherland (Delft), June 16e20 (2013). [70] J. Popplewell, A. Al-Qenaie, S.W. Charles, R. Moskowitz, K. Raj, Thermal conductivity measurements on ferro�uids, Colloid Polym. Sci. 260 (1982) 333e338. [71] H. Xie, L. Chen, Review on the preparation thermal performances of carbon nanotube contained nano�uids, J. Chem. Eng. Data 56 (2011) 1030e1041. http://dx.doi.org/10.1021/je101026j. [72] M. Fujii, X. Zhang, H. Xie, H. Ago, K. Takahashi, T. Ikuta, H. Abe, T. Suimizu, Measuring the thermal conductivity of a single carbon nanotube, Phys. Rev. Lett. 95 (2005) 065502. http://dx.doi.org/10.1103/PhysRevLett.95.065502. [73] M. Shafahi, V. Bianco, K. Vafai, O. Manca, An investigation of the thermal performance of cylindrical heat pipes using nano �uids, Int. J. Heat Mass Transf. 53 (2010) 376e383. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2009.09.019. [74] M. Shafahi, V. Bianco, K. Vafai, O. Manca, Thermal performance of � at-shaped heat pipes using nano �uids, Int. J. Heat Mass Transf. 53 (2010) 1438 e1445. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2009.12.007.

17

[75] H. Akachi, U.S. Patent No. 4, 921, 041, May 1990. [76] G. Karimi, J.R. Culham, Review assessment of pulsating heat pipe mechanism for high heat �ux electronic cooling, in: Int. Society Conference on Thermal Phenomena (2004). [77] S. Khandekar, P.K. Panigrahi, F. Lefèvre, J. Bonjour, Local hydrodynamics of �ow in a pulsating heat pipe: a review, Front. Heat Pipes 1 (2010) 023003. http://dx.doi.org/10.5098/fhp.v1.2.3003. [78] P.A. Walsh, V.M. Egan, E.J. Walsh, Novel micro-PIV study enables a greater understanding of nanoparticle suspension �ows: nano�uids, Micro �uid Nano�uid 8 (2009) 837e842. http://dx.doi.org/10.1007/s10404-009-0553-z. [79] S.B. White, Enhancement of Boiling Surfaces Using Nano �uid Particle Deposition. PhD-Thesis, The University of Michigan, 2010 . [80] K.Y. Leong, R. Saidur, T.M.I. Mahlia, Y.H. Yau, Performance investigation of nano �uids as working � uid in a thermosyphon air preheater, Int. Commun. Heat Mass Transf. 39 (2012) 523e529. http://dx.doi.org/10.1016/ j.icheatmasstransfer.2012.01.014. [81] W. Yu, S.U.S. Choi, The role of interfacial layers in the enhanced thermal conductivity of nano �uids: a renovated Maxwell model, J. Nanoparticle Res. 5 (2003) 167e171. [82] M.H. Buschmann, M. Gad-el-Hak, Recent developments in scaling of wallbounded �ows, Prog. Aerospace Sci. 42 (2007) 419e467. http://dx.doi.org/ 10.1016/j.paerosci.2007.01.001. [83] Y. He, Y. Mena, Y. Zhao, H. Lu, Y. Ding, Numerical investigation into the convective heat transfer of TiO2 nano�uids �owing through a straight tube under the laminar � ow conditions, Appl. Therm. Eng. 29 (2009) 1965e1972. http://dx.doi.org/10.1016/j.applthermaleng.2008.09.020. [84] E. Abu-Nada, Effects of variable viscosity thermal conductivity of CuO-water nano �uid on heat transfer enhancement in natural convection: mathematical model simulation, J. Heat Transf. 132 (2010) 052401. http://dx.doi.org/ 10.1115/1.4000440. [85] S.M. You, J.H. Kim, K.H. Kim, Effect of nanoparticles on critical heat �ux of water in pool boiling heat transfer, Appl. Phys. Lett. 83 (2003) 3374 e3376. http://dx.doi.org/10.1063/1.1619206. [86] B.B. Mikic, W.M. Rohsenow, A new correlation of pool-boiling data including the effect of heating surface characteristics, J. Heat Transf. 12 (1969) 245 e250. [87] S.J. Kim, I.C. Bang, J. Buongiorno, L.W. Hu, Effects of nanoparticle deposition on surface wettability in�uencing boiling heat transfer in nano �uids, Appl. Phys. Lett. 89 (2006) 153107. http://dx.doi.org/10.1063/1.2360892. [88] B. Solomon, K. Ramachandran, B.C. Pillai, Thermal performance of a heat pipe with nanoparticles coated wick, Appl. Therm. Eng. 36 (2012) 106 e112. http:// dx.doi.org/10.1016/j.applthermaleng.2011.12.004.

Nanofluids in Thermosyphons and Heat Pipes

Recommend Documents