International Journal of Heat and Mass Transfer 124 (2018) 463–474
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Effect of nano-structure coating on thermal performance of thermosyphon boiling in micro-channels Shuang-Fei Li a, Yi-Ying Bao a,b, Ping-yang Wang a, Zhen-hua Liu a,⇑ a b
School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China Shanghai Aerospace System Research Institute, Shanghai 201108, China
a r t i c l e
i n f o
Article history: Received 8 October 2017 Received in revised form 15 February 2018 Accepted 21 March 2018
Keywords: 3D chip cooling Thermosyphon Nanofluids Micro-channel Boiling
a b s t r a c t A novel micro-channel thermosyphon technology for passively cooling 3D stacked chips was provided, and the thermosyphon boiling characteristics in vertical and inclined micro-channels with two open ends which simulate the specific stacked structure of actual 3D chip were experimentally carried out. In order to improve the heat transfer of micro-channel thermosyphon by surface treatment technology, four kinds of nanoparticles (CuO, Cu, Al2O3, SiO) were added to the base fluid to make nano-structure coatings on the heater surfaces by using a long time pool boiling treatment. Then, micro-channel thermosyphon boiling experiments were carried out with four kinds of working liquids: two kinds of pure fluids (deionized water and R113) and two kinds of moist fluids (deionized water+surfactant and R113+surfactant). The gaps and heights of micro-channels tested were in the range of 30–60 lm and 30–90 mm, respectively. Experimental results show that nano-structure coatings can significantly enhance both the maximum heat flux and heat transfer coefficient of thermosyphon boiling in micro-channels, and exist both the optimal nanoparticle kind and nanoparticle concentration in the base fluids. The experimental results provided some meaningful technology support for 3D stacked chip cooling. Ó 2018 Elsevier Ltd. All rights reserved.
1. Introduction With the rapid development of micro electric technology, 3D stacked chips have been considered as a new chip integration technology in which that chips are no longer connected side by side, but the upper and lower parallel connection, and 3D integration technology is also considered as the preferred option to achieve miniaturization. If the heat generated by the chips cannot be dissipated in time, it will result in temperature excursion, which not only impacts the normal operation of the chips, but also reduces its longevity due to the fact that long term reliability drops by 50% for each 10° rise in junction temperature [1]. Cooling of 3D chips cannot simply apply the cooling methods of traditional 2D chips in which the cooling channels are mounted in the substrate. However, the most 3D chips cooling technologies are still the simple extension of 2D chips cooling technologies, in which micro-channels with various shapes are installed in encapsulation, and working fluid are driven by external power flows through the micro-channels to take away heat [2–7]. These cooling methods belong to active type methods and have several great drawbacks. In addition, many researchers carried out various ⇑ Corresponding author. E-mail address:
[email protected] (Z.-h. Liu). https://doi.org/10.1016/j.ijheatmasstransfer.2018.03.071 0017-9310/Ó 2018 Elsevier Ltd. All rights reserved.
designs, experimental and simulated studies for ThreeDimensional Integrated Circuits [8–12]. A few of researchers proposed some designs and simulations including micro flat heat pipe system and forced convective system for cooling 3-D packaged chips [13–15]. Recently, the authors provided a new passive cooling conception for 3D chips cooling and carried out an initially experimental result [19]. A vertical micro-scale thermosyphon structure was designed by utilizing 3D chips’ specific micro structure to form thermosyphon boiling with the driving force composed of buoyancy and capillarity for 3D stacked chips cooling. As well known, Conventional 3D stacked chip encapsulation is placed horizontally. But if it is deliberately arranged vertically, tens of microns gaps between chips can form vertical micro-channels with two open ends. According to this geometrical characteristic, insulating liquid is filled in the lower half of the encapsulation and submerges the 3D stacked chips, and then the vertical micro-channels submerged in saturated liquid would form the evaporation section of a microscale thermosyphon heat pipe. The upper surface of the encapsulation can be developed as heat sink for condensation. Saturated liquid flows upward along micro-channels and absorbs heat generated in chips, then evaporates into saturated steam gradually, eventually, discharges from the upper outlet of the chips continuously by buoyancy and capillarity. This heat transfer mode
464
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
Nomenclature b d h L I P q q00 S t DT
width of channel (m) gap of channel (m) heat transfer coefficient (W/m2K) length of channel (m) electric current (A) power (W) heat flux (W/m2) critical heat flux (W/m2) surface area (m2) temperature (K) mean superheating (K)
is generally called thermosyphon boiling. A lot of research have revealed the thermosyphon boiling mode and the boiling heat transfer mechanism in micro channels [16–18]. The present concept of micro-channels thermosyphon is different from that of traditional thermosyphon. In traditional thermosyphon, the governing force is only buoyancy, but in the micro-channels thermosyphon, the governing forces are both buoyancy and capillarity. In general, capillarity will be far greater than buoyancy [19]. In other hand, the present concept of micro-channels thermosyphon is also different from that of traditional micro heat pipe, which were designed as small tubes and channels with various grooves or small triangle tubes for providing capillary force [20–24]. However, in the present study, the capillarity structure is formed by micro gaps and has not any grooves or sharp-angled corners in rectangle channel, the capillary comes from its own micro-scale. On this basis, the authors extended the study objective from vertical micro-channels to inclined and horizontal microchannels [25]. Experimental results show that horizontal and inclined micro-channels structure can also cause great natural convective boiling to cool 3D chips. In addition, the suitable addition of SDS into pure R113 and deionized water to form super-moist liquid will significantly improve the both critical heat flux and heat transfer coefficient in micro-channels by 70–80%. Since the gaps of micro channels are very small in the 3D chips, so simply using of the thermosyphon boiling mode cannot meet the need of heat removal, and then various heat transfer enhancement technologies are needed. In the previous studies [19,25], super-moist fluids are applied to reduce the solid–liquid contact angle and thence enhance capillary force. In this study, another important enhanced technology for phase-change heat transfer is the surface treatment technology, which is applied to increase capillarity and activate boiling nucleate cavities on the chip surfaces. Among the all surface treatment technologies, nanoparticles deposition method by pool boiling of nanofluids on the heating surface should be the easiest and cheapest for 3D stacked chips. So far, a lot of experimental studies have been carried out for investigating both pool boiling heat transfer of nanofluids on plate surfaces [26,27], and thermal performance of conventional thermosyphon heat pipes filled up with nanofluids [28–31]. A common finding is that there exists a porous coating layer formed by nanoparticles deposited on the heating surfaces after boiling process and the coating layer enhances greatly boiling heat transfer. For the present micro-channels, it is not suitable to use directly nanofluids as the working liquid in the thermosyphon due to the fact that it may block the channels during long-term running. However, according to a lot of published research which confirmed that the nanoparticles can be attached to the heating surfaces during boiling process due to the very strong van der Waals interactions [32,33], this surface treatment method by nanofluids boiling may be used to develop a stable nano-structure coating on the chip sur-
U
voltage (V)
Greek letters h inclined angle of channel Subscripts 0 pure liquid h horizontal v vertical h inclined angle of channels
faces, then, common fluids are used as the working liquids in micro thermosyphon. In this study, spaced apart nickel heating elements which form micro-channel heat pipe structure were used to simulate 3D chips as fever ends. Both vertical and horizontal micro-channel structures were tested. Four nanoparticles (CuO, Cu, Al2O3, SiO) were added to the two base fluids (deionized water and R113) to form some especial nano-structure coating on the surfaces of chips after a long time pool boiling treatment. Then, formal experiments were carried out using four kinds of working liquids: two pure fluids (deionized water and R113) and two moist fluids (deionized water+surfactant and R113+surfactant). The gaps and heights of channels used were in the range of 30–60 lm and 30–90 mm, respectively. The research particularly focused on boiling heat transfer ability, namely the maximum heat flux, since the main obstacle to 3D chips cooling is that the intermediate chip may be overheated without adequate cooling. Experimental results show that the nano-structure coating can significantly enhance both the maximum heat flux and heat transfer coefficient of thermosyphon boiling using pure liquids as that in common pool boiling. Furthermore, it is found that there exist both the optimal nanoparticles kind and nanoparticles mass concentration to obtain the most effective coating layer. Then on this basis, different concentrations of surfactant, SDS, were dispersed to deionized water and R113 to form two super moist liquids to improve further thermosyphon boiling performance. The optimum surfactant concentrations were found out for deionized water and R113, respectively. The experimental results fulfilled the expected goals and showed potential application of horizontal micro-channel heat pipe structure for 3D stacked chip cooling. 2. Experimental apparatus, nano-structure coating making, working fluids and experimental procedure 2.1. Experimental apparatus Fig. 1 (a) shows the diagram of basic 3D stacked chip encapsulation without any cooling methods, and Fig. 1 (b) and (c) show the schematic of micro-channels structures using present cooling method with vertical and horizontal states, respectively. Insulating working fluid is filled in the lower half of the encapsulation and submerges the 3D stacked chips. The upper surface of the encapsulation is processed as heat sink. For horizontal structure, when chips generate heat, saturated liquid flows into micro-channels from two ends of the channels along the wall driven by capillarity, absorbs heat and evaporates into saturated steam along the channels, and finally saturated steam flushes out of the channels from two ends. Strictly speaking, this heat transfer is a gas–liquid counter-flow boiling and cannot belong to thermosyphon boiling. However, since there is not strict theoretical analysis in this paper,
465
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
Cooling fluid Heat sink Encapsulation Chip
Solder joint Substrate
(a) Basic installation of 3D chip encapsulation
Cooling fluid g
Heat sink
Heat sink
Cooling fluid
Substrate
g Working fluid
Solder joint
Encapsulation Chip
Encapsulation
Working fluid
Chip
(b) Vertical heat pipe structure
Solder joint
Substrate
(c) Horizontal heat pipe structure
Fig. 1. Scheme of 3D chip encapsulation.
the experimental data from two flow patterns were managed together. Fig. 2 shows the whole experimental system. The whole apparatus can be divided into thermostatic tank part, micro-channels constitution part, i.e. the test pieces, condensation part and electrical system. The micro-channels constitution part which simulates 3D chips to generate heat is submerged under the working fluid in the glass thermostatic tank wrapped with insulation material. Fig. 3 shows the structure of test piece that is adopted to simulate the actual 3D chips. The experimental study is carried out in ambient pressure. A computer and data acquisition system are used for power and temperature measurements. A silicon rectifier provides steady direct current for heating the test pieces. According to Fig. 3, three micro-channel structures are arranged in parallel, middle of which is the measuring piece. The other two side channels are used for constitution of a symmetrical geometry and reducing heat dissipation of the middle channel. The test pieces share the same width of 4 mm and have three lengths of 30 mm, 60 mm and 90 mm, and different channel gaps ranging from 100 lm to 30 lm. The test pieces are named by their gaps and lengths. For example, 30 ⁄ 0.05 mm channel means that the length and gap of the channel are 30 mm and 0.05 mm, respectively. Fig. 4 shows the detail structure of single test piece consisting of PTFE gaskets and FR 4 substrate. An I-shaped nickel foil with the thickness of 50 lm is pasted on the FR4 substrate plate to form the heating surface of the channel. Two pieces of symmetrical PTFE gaskets are pasted on two sides of the nickel foils to form an empty rectangular channel between nickel foils with two heating walls. The thin copper plates are attached to the tail of the nickel foil as electrodes. The gap of channel is determined by the thickness of
copper electrode, which can be measured precisely. The dimensions of PTFE gaskets are precisely processed as well. 2.2. Nano-structure coating making Four kinds of nanoparticles (CuO, Cu, Al2O3, SiO) with the mean diameters of 20–30 nm were used for developing the nanofluids with different mass concentrations by two step method for the formation of porous sedimentary layer (nano-structure coating) on the heating surface. Fig. 5 shows the TEM (transmission electron microscope) photo of four kinds of water base nanofluids with 1 wt% mass concentration. All nanoparticles were commercial products made by the gas-condensation method. The mass concentration ranges of nanoparticles in the base fluids were 0.1 wt% to 1.5 wt%. The base fluids were R113 and ID water. The nano-structures were formed by nanofluids using fallowing method. Firstly, The nanofluid was prepared by dispersing nanoparticles with a fixed mass concentration and surfactant with a fixed volumetric concentration into the base fluid directly and then oscillated continuously for several hours in an ultrasonic water bath with a working frequency of 25–40 kHz so that the nanoparticles can be uniformly distributed. Then, a pool boiling test was carried out for about four hours using the nanofluid as the working liquid. In this test, all the test pieces were horizontally submerged in the nanofluid and the heating heat flux was remained at about 100 kW/m2. The reason selecting this heating time is that there was no obvious change in the surficial morphology of the coating after about four hours pool boiling. Finally, the test pieces were cleaned up and the nanofluid was drained. It must be noted that the nanofluid was only used for forming the nanostructure coating on the heating surface as a preparatory treatment
466
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
Inlet of cooling water
Outlet of cooling water
$
16
1.Auxiliary heating rod 2.glass tank 3.LED lamp 4.working fluids᧤deionized water or R113᧥ 5.insulations 6.steam outlet 7.rubber hose 8.condenser 9.power source 10.computer and data acquisition system 11.silicon rectifier 12.current meter 13.wire channel 14.glass cover 15.fixture and channel 16.viewing window 17.holder 18.organic glass base 19.voltage measurement line 20.thermocouple 21.earth wire Fig. 2. A schematic view of experimental system.
and the nanoparticles have not been added into the working liquids in the formal test. Fig. 6 shows the SEM (sweep electron microscope) pictures of the heating surfaces after the pool boiling experiment using four water based nanofluids. The mass concentration of all nanofluids were 1.0 wt% and no surfactant was added. As is shown, various deposition layers were formed on the heating surfaces. It is obvious that the deposition layers formed by Cu, CuO and Al2O3 nanofluids were porous and have clear nano-structures, however, the deposition layer formed by SiO nanofluid was slurry structure and has not clear porous morphology. This different surficial morphologies will result in different influences on the thermal characteristics of thermosyphon boiling in micro channels. Fig. 7 illustrates the SEM pictures of surficial morphologies of the coating layers on the heating surfaces after pool boiling test using CuO nanofluids with different mass concentrations. Four mass concentrations (0 wt% (water), 0.1 wt%, 1.0 wt% and 1.2 wt %) were selected to show the effect of the CuO concentration in the water based fluid on the surficial morphologies of the coating layers. The all boiling processes were kept for 4 h at the atmospheric pressure. As can be seen in Fig. 7, the boiling surface keeps smooth for water boiling test and no coating exists on the surface. However, for all nanofluid boiling tests, there exist coating layers
of nanoparticles on the surfaces. Moreover, for the boiling test with mass concentration of 0.1 wt%, only a loose and partially-coated porous layer exists on the heating surface. For the boiling tests with the mass concentration of 1.0 wt% and 1.2%, however, both the boiling surfaces are covered by coating layers completely. Moreover, comparing the two kinds of coating layers, it is found that the coating of 1.0% concentration seems to have better honeycomb-like porous structure than that of 1.2% concentration.
2.3. Working fluids In the present study, deionized water and R113 were used as the working fluids at first. Meantime, surfactant, Sodium dodecyl sulfate (SDS), was also added into the base fluids for developing two kinds of moist fluids. The reason selecting SDS is that it can cause very strong solid–liquid wettability for both ID water and R113. The deionized water used in the experiment had an electrical conductivity of 1.005 lS/cm which was measured before the formal experiment. The properties of saturated deionized water and R113 under atmospheric pressure are listed in Table 1.
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
467
(B)
(C)
(a)
A. micro channel installation slot, B. PTFE gaskets, C. copper electrode, D. micro channel, E. service entrance, F. glass fiber heating plate, G. glass fiberboard cushion Fig. 3. The structure of test piece: (a) fixture for test piece installation, (b) cross section diagram of the plane 1, (c) cross section diagram of the plant perpendicular to the plane 1.
1. PTFE gaskets, 2.FR4 substrate, 3. thermocouple mounting slits, 4. heating surface, 5. copper electrode
A. PTFE gaskets, B. FR4 substrate, D. nickel foil, E. copper electrode , F. glue layer, H. micro channel
(a)
(b) Fig. 4. Detail structure of single test piece (unit: mm).
2.4. Experimental process
P ¼UI
The basic experimental process was the same as the first study [19]. The heat flux was increased gradually. After all measured wall temperatures reached steady state for 2 min, the input power was increased again until the wall temperatures sharply increased and can’t attain a steady state. Then the input power was turned off immediately in case of temperature excess. In order to improve the accuracy of the measured the critical heat flux (CHF), the test was restarted from the input power corresponding to the last steady wall temperatures, and the power was increased by a step of 1% of the former. When wall temperatures surged again, the test was stopped eventually. The CHF was obtained from the maximum power corresponding to the last steady wall temperature. Thus, the cut-off error of input power corresponding to the CHF is 1%. Four main parameters are considered during the test, including input power, effective heat flux, wall superheating and heat transfer coefficient (HTC), which are calculated as follows:
qw ¼ q qloss ¼
ð1Þ P Ploss S
ð2Þ
DT ¼ T w T s
ð3Þ
qw DT
ð4Þ
hw ¼
where P is the electric power, I is the total current flowing through the heating surface, S is the total heating area in a channel. 2.5. Uncertainty analysis The equipment used to measure parameters includes the following: GG-K-30 thermocouples for measuring temperatures, digital micrometer for measuring length, graduated cylinder for measuring volume, power meter for measuring power. The electric power can be divided into effective power and heat loss. The heat
468
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
(a) Cu
(b) CuO
(d) SiO2
(c) Al2O3 Fig. 5. TEM photo of four nanofluids (1 wt%).
loss was calculated by a theoretical simulation [20]. The measurement errors of the equipment and relative errors of the actual internal gap of channels are listed in Table 2. According to The relative standard deviation of the transfer function theory, the maximum relative deviation of input power, effective heat flux, wall superheating and HTC (Eqs. (1)-(4)) can be derived as follows and the value is summarized in Table 3
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 @ ln P @ ln P ¼ r2U þ r2I P @U @I
rP
rqw qw
ð5Þ
rhw hw
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 @ ln hw @ ln hw @ ln hw ¼ r2qw þ r2T w þ r2T l @qw @T w @T l
3. Results and discussion 3.1. Thermosyphon boiling characteristics of pure fluids on the nanostructure coating layers 3.1.1. CHF enhancement effect Fig. 8 illustrates the effect of nano-coating formed by different nanofluids on the CHF for thermosyphon boiling in vertical micro
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 2 2 @ ln qw @ ln qw @ ln qw @ ln qw @ ln qw ¼ r2U þ r2I þ r2L þ r2d þ r2nloss @U @I @L @d @nloss
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 rDT @ ln DT @ ln DT r2T w þ r2T s ¼ @T w @T s DT
ð8Þ
ð7Þ
ð6Þ
channels using ID water as the working fluid at the atmospheric pressure. Twenty-four coating surfaces were tested which were coated by four water based nanofluids with different nanoparticles concentrations (the mass ratio of nanoparticles to water) changed
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
a
Cu/H2O nanofluid(1%wt)
c
Al2O3/H2O nanofluid(1%wt)
469
b CuO/H2O nanofluid (1%wt)
d
SiO2/H2O nanofluid(1%wt)
Fig. 6. Surficial morphology of the nano-structure coating on heat transfer surfaces.
from 0 (non-coating surface) to 1.2 wt%. Firstly, it is obvious that the nanoparticle kind has very important influence on the CHF and the reason is that different nanoparticles would form different coating surficial morphologies as seen in Fig. 6. The porous coatings formed by Cu and CuO nanofluids have clear nanostructures and thence have good enhanced effect compared with common smooth surface due to the increase of capillary and the solid–liquid wettability. Since Cu coating has better heat conductivity than CuO coating, Cu coating has the best enhancement effect for CHF among all of the coatings tested. On the other hand, The CHF has not any change compared with that on the smooth surface when the coatings were formed by SiO nanofluids and ID water was used as working fluid, since deposition layer formed by SiO nanofluid was slurry structure and has not clear porous morphology. The CHF data confirmed that different coating surficial morphologies have very significant impact on the CHF. Furthermore, for the same nanoparticle kind, the mass concentration also has obvious impact on the CHF and there is the optimum mass concentration corresponding to the greatest CHF enhancement effect. It is about 0.75 wt% for Cu and about 1.0 wt % for both CuO and Al2O3. The reason why there is the optimum mass concentration can be partly explained by the coating surficial morphology as seen in Fig. 7. In Fig. 7, three typical SEM pictures of surface morphologies for the CuO coatings with different mass concentrations show clearly that mass concentration has obvious effect on coating surface morphology, and thence has also obvious impact on the CHF enhancement effect. As seen in Fig. 7, for the coating formed by the nanofluid with mass concentration of 0.5 wt%, only a loose and partially-coated porous layer exists on the heating surface. For the coatings formed by the nanofluids with the mass concentrations of 1.0 wt% and 1.2%, however, both the surfaces are covered by coating layers completely. Moreover, com-
paring the two coating layers, it is found that the coating of 1.0% concentration seems to have better honeycomb-like porous structure than that of 1.2% concentration, and thence, it corresponds to better capillary structure. According to Fig. 8, Cu coating can increase the CHF by 70%; CuO coating increases the CHF by 60%, and Al2O3 coating only increase the CHF by 15%. Although Cu coating has the best enhanced effect, if considering expensive price and electric conductivity of Cu nanoparticles, CuO nanofluids should be better selection for forming nano-structure coating.
3.1.2. HTC enhancement effect Fig. 9 gives out thermosyphon boiling curves of ID water (HTC vs. the superheating curves) for different coatings in vertical channel. The length and gap of micro channel tested are 30 mm and 100 lm, respectively. Here, zero concentration represents a smooth surface. It is well confirmed that various coatings can significantly improve the HTC and the maximum enhanced effect can reach 1 time for Cu and CuO coatings and 50% for Al2O3 coating. More interestingly, using coatings not only causes significant enhancement of the HTC, but also shows similarity to conventional thermosyphon boiling, in which HTC increases first and then decreases with increasing superheating. The former corresponds to the nucleate boiling mode in wetting liquid film zone and the latter corresponds to conduction mode in part dry out zone. It is obvious that the coating can increase capillary force carrying more liquid into the micro channel and form boiling in liquid film on the heating surface just like in conventional scale channel. Moreover, it is found that only HTC is enhanced, but the maximum superheating has hardly been extended and almost holds at a narrow range of 26–28 K. This reason is still unclear in the present stage.
470
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
(a)Water
b
(c)CuO/H2O nanofluid(1%wt)
CuO/H2O nanofluid(0.5%wt)
d
CuO/H2O nanofluid(1.2%wt)
Fig. 7. Effect of nanofluid concentration on surficial morphology after pool boiling.
Table 1 Physical property of deionized water and R113. Property
Deionized water
R113
Boiling point (K) Liquid density (kg/m3) Gas density (kg/m3) Surface tension (mN/m) Latent heat (kJ/kg)
373.15 958 0.597 58.85 2257.2
320.75 1508.2 7.424 14.71 203
surface. It is confirmed that the CHF with the inclined angle of h may be in good agreement with Eq. (9) [34] when h is greater than or equal to 5° with a maximum error of 10%. Moreover, when h is less than 5° or equal to 0°, the CHF will have not change and is still equal to that of 5° as shown in Fig. 10.
q00# ¼ ðsin hÞ
1=4 00
qv
h ¼ maxð5; hÞ
ð9Þ
Similar to the case of CHF enhancement, for the same nanoparticle kind, the mass concentration has also obvious impact on the HTC and there is the optimum mass concentration corresponding to the greatest HTC enhancement effect. These optimum values are the same as that for CHF enhancement effect.
According to Eq. (9), the CHF of horizontal micro channel is about half of that of vertical micro channel for all working liquids and coating surfaces. For thermosyphon boiling in an inclined traditional channel, the happen mechanism of the CHF is the dry out of liquid at the outlet and the CHF is proportional to the liquid mass flux, which is proportional to buoyancy. Therefore, if the effect of buoyancy on the CHF is only considered, the buoyancy in an inclined channel would
3.1.3. Effect of inclined angle on thermosyphon boiling on coating layers Fig. 10 shows the influence of the inclined angles on the CHF of thermosyphon boiling using ID water as the working fluid on different coatings formed by water based CuO nanofluids with different mass concentrations. Here, water coating means a smooth
change with ðsin hÞ , so equation of q00# ¼ ðsin hÞ q00v can be obtained [25]. Since capillarity also affects the CHF strongly in micro channel, however, the CHF in an inclined micro channel will be greatly increased compared with that in an inclined conventional channel as seen in Eq. (9). According to Eq. (9), the CHF of pure fluids on any coating layers and any inclined angle, even
1=2
1=2
Table 2 System error and gap relative error. Gap
1 mm
0.5 mm
0.1 mm
0.05 mm
0.03 mm
Gap relative error
±3.35%
±4.3%
±12.5%
±19%
±25%
Parameters U, I L, d T
Measurement error ±0.5% ±0.002 mm ±0.2 K
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474 Table 3 Measurement error of calculated parameters. Calculated parameters
Relative error
q qb hw CHF
1.3% 1.3% 2.9% 7.3%
36000 34000
Water Vertical channel Size: 30mmx0.1mm
Coating making CuO SiO2
32000
Al2O3 Cu
30000
q /W/m
28000 26000 24000 22000 20000 18000 0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.
(%) Fig. 8. Effects of coating kinds on CHF in vertical channels using ID water as working fluid.
including horizontal case, can be obtained by the CHF corresponding to the vertical case.
3.2. Thermosyphon boiling characteristics of moist fluids on the nanocoating layers 3.2.1. CHF enhancement effect Fig. 11 illustrates the effect of the coating kinds formed by the various optimum nanofluid concentrations on the CHF of thermosyphon boiling in vertical micro channels using ID water (SDS concentration equals to 0) and the water based moist fluids as the working fluid. Three kinds of water based nanofluids were used to form the coating layers and the mass concentration was fixed at the optimum concentration obtained in the previous discussion. According to the cases using ID water as the working fluid, it is confirmed that the nanoparticle kind in the same base fluid has very important influence on the CHF and the Cu coating has the best CHF enhanced effect, and CuO coating follows immediately after Cu coating. The reason is the same as that using ID water as the working fluid. Furthermore, for the same nanoparticle coating, the SDS concentration in the base fluid has also obvious impact on the CHF and there is the optimum SDS concentration corresponding to the greatest CHF enhancement effect. It is about 2000 ppm for both the smooth surface and Al2O3 coating, and about 2500 ppm for both Cu and CuO coatings. Because the porous coating layers can absorb some SDS, the optimum SDS concentrations corresponding to both Cu and CuO coatings would be slightly greater than those corresponding to the smooth and slurry-like surfaces. Comparing Fig. 11 with Fig. 8, it can be found that the CHF using water based super-moist fluid as the working fluid can be increased by 70% than that using ID water as the working fluid for the Cu coating, by 60% for the CuO coating. Furthermore, as seen in Fig. 11, if using the optimum moist fluid as the working fluid, then, the CHF can be increased by 70% for the smooth surface,
471
110% for Cu coating and about 90% for CuO coating. As a simple estimation, if simply substituting the optimum moist fluid for ID water as the working fluid, the CHF on coating surfaces would be increased by 50%. According to the previous discussion, it has been well known that there are optimum nanoparticle mass concentrations to form optimum coatings and optimum SDS concentrations to form optimum moist fluid, so in the latter discussion, only the optimum moist fluids were used as the working fluids and only optimum coating surfaces were used as the heating surfaces. Fig. 12 displays the effect of the sizes and the inclined angles of micro-channels on the CHF enhancing ratio using the optimum R113 moist fluid (2000 ppm) as the working fluid and the optimum CuO coating as the heating surface. Here, the CHF enhancing ratio is defined as the CHF ratio of the moist fluid-coating surface to the base fluid-smooth surface. The CHF enhancing ratio is in the range of 1.7–2.3 for horizontal state and 1.8–2.4 for vertical state. The mean CHF enhancing ratios in vertical state seem to be slightly larger by 10% than those in horizontal state. It is also obvious that the shorter the length is, the larger the CHF enhancing ratio is. On the other hand, although the influence of the gap on the CHF enhancing ratio is slight, there is still an optimum gap and it is equal to 0.06 mm in the test range. Overall, although the impact of the sizes of micro-channels are complex, the impact is relatively weak, therefore, can be further ignored. For the micro channel with very small gap, on the one hand, deposition of nanoparticles on the smooth surface forms porous layer and thence increases capillary. But on the other hand, it will reduce effective flow section and increase the flow resistance. However, according to the experimental results in Fig. 12, there are almost no negative effects in the present test range. The reason is that the thicknesses of the coatings are still much less than the gaps of micro-channels. 3.2.2. HTC enhancement effect Fig. 13 shows thermosyphon boiling curves of the optimum water based moist fluid (HTC vs. the superheating curves) for different optimum coatings (Cu, CuO and Al2O3 coating surfaces) in vertical channel. The heat transfer curves of ID water on the smooth surfaces are also plotted for comparison. The length and gap of micro channel tested are 30 mm and 100 lm respectively. As seen in the case using ID water as the working fluid shown in Fig. 9, it is well confirmed that various coatings can significantly improve the HTC and the maximum enhanced effect that can reach 1.7 times for Cu coating and 1.43 times for CuO coating. If comparing with Fig. 9, then, it is found that substituting the optimum moist fluid for ID water will further increase the HTC about 70% for Cu coating, 43% for CuO coating, and 65% for Al2O3 coating. Meanwhile, using optimum moist fluid causes more significant boiling characteristics due to more liquid flowing into the micro channel. Moreover, it is also found that the maximum superheating has been hardly extended and holds still at a narrow range of 26–28 K. Fig. 14 shows thermosyphon boiling curves of the optimum R113 based moist fluid for different optimum coatings in vertical channel. The boiling curves of pure R113 on the smooth surfaces are also plotted for comparison. The length and gap of micro channel tested are 30 mm and 100 lm respectively. As seen in Fig. 14, the HTC is maximally increased by 2 times for Cu coating and 1.75 times for CuO coating. It is found that substituting the R113 moist fluid for pure R113 will cause better HTC enhancing effect at the same coating and geometry conditions. For the heat transfer of pure R113 on the smooth surface (block square symbol in Fig. 14), the HTC is relatively small and the HTC curve is almost a horizontal line in most superheating range, which
472
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
1500
1500
Coating making
1400 1200 1100 1000 900 800
1300 Concentration of CuO(%) 0 0.1 0.2 0.5 0.75 1 1.2
1200 1100 1000
h /W/m2/K
1300
h /W/m 2/K
Coating making
1400 Concentration of Cu(%) 0 0.1 0.2 0.5 0.75 1 1.2
700 600 500
Water Vertical channel Size:30mmx0.1mm
400 300 200
900 800 700 600 500 400
Water Vertical channel Size: 30mmx0.1mm
300 200
100
100 0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30
0
2
4
8
10 12 14 16 18 20 22 24 26 28 30
(b) CuO coating
(a) Cu coating 1200
6
Coating making
1100 Concentration of Al2O3(%)
1000
0 0.1 0.2 0.5 0.75 1 1.2
900
h /W/m2/K
800 700 600 500 400
Water Vertical channel Size: 30mmx0.1mm
300 200 100 0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30
(c) Al2O3 coating Fig. 9. Effects of various coatings on HTC of water in vertical micro-channel.
34000
30000 28000 26000
Water based super-moist fluid Vertical channel Size: 30mm x 100 m
42000 40000 38000 36000 2
24000
q'' /W/m
q'' /W/m2
44000
Coating method Water Smooth surface Size: 30mm x 100 m 0.1 wt% CuO+water 0.2 wt% CuO+water 0.5 wt% CuO+water 0.8 wt% CuO+water 1.0 wt% CuO+water 1.2 wt% CuO+water
32000
22000 20000
34000 32000 30000 28000
18000
Coating method Smooth surface CuO(1%) Al2O3(1%)
26000
Eq. (5)
16000
24000
14000
22000
12000
20000
10000 0
20
40
60
80
100
Cu(0.75%)
18000 0
500
1000
1500
2000
2500
3000
SDS concentration /PPM Fig. 10. Effect of inclined angle on the CHF of water for various coating layers.
means that the heat transfer mode should be heat conduction in the dry-out zone and there is not boiling in micro channel when the superheating was over 8 K. When substituting CuO coating surface for the smooth surface as the heating surface, porous heating surface not only increases capillary which increases liquid mass
Fig. 11. Effects of coating kinds on the CHF of water based moist fluid in a vertical channel.
flux and the CHF, but also increases active nucleate cavities which increase boiling HTC [35–37]. Then, if further substituting the optimum R113 moist fluid for pure R113 as the working fluid, solid– liquid wettability will increase and liquid–vapor surface tension
q'' / q''0
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474 3.0 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 2.0 1.9 1.8 1.7 1.6 1.5 1.4 1.3 1.2 1.1 1.0
will decrease, both of which can increase capillary. What’s more, these effects cause the increase in liquid mass flux flowing into micro-channel and promote both the HTC and CHF.
Optimum R113 moist fluid (R113+SDS(2000ppm) Optimum CuO coating surface
4. Conclusions
Vertical channel L=30mm L=60mm L=90mm Horizontal channel L=30mm L=60mm L=90mm
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
d / mm Fig. 12. Effects of sizes and inclined angles of micro-channels on CHF enhancing ratio for the optimum R113 moist fluid and optimum CuO coating surface.
2000
Optimum Water moist fluid 1800 Size:30cmmx 0.1mm Vertical channel 1600 1400
h /W/m2/K
473
1200 1000 Water on smooth surface
800 600
Optimum Cu coating surface Optimum CuO coating surface Optimum Al2O3 coating
400 200 0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30
Fig. 13. Effects of coatings on HTC using water moist fluid in vertical channel.
700
Optimum R113 moist fluid 650 Size:30mmx 0.1mm 600 Vertical channel
A passive cooling design for 3D stacked chip is provided, which directly utilizes the micro-channels structure of the 3D chip itself to form the evaporating section of a micro thermosyphon. In order to further improve the heat transfer of micro thermosyphon by surface treatment technology, four kinds of nanoparticles (CuO, Cu, Al2O3, SiO) were added into the ID water to make nanostructure coatings on the surfaces of simulated chips by a long time pool boiling treatment. Micro thermosyphon boiling experiments were carried out on the twenty-four coating surfaces, using four kinds of working liquids: two kinds of pure fluids (deionized water and R113) and two kinds of moist fluids (deionized water+surfactant and R113+surfactant) with different surfactant concentrations. The length and gap of micro-channels used were in the range of 30–90 mm and 30–100 lm, respectively. The experimental results are given as follows: (1) Horizontal micro channels can also cause thermosyphon boiling, The CHF of horizontal micro channel is about half of that of vertical micro channel for all working liquids and coating surfaces. Eq. (9) may be applied to predict the effect of the inclined angle of channel on the CHF. (2) The addition of SDS into pure R113 and deionized water to form moist liquid will significantly improve the both HTC and CHF in micro-channels and there exists an optimum SDS concentration ranging from 2000 ppm to 2200 ppm. (3) Developing nanoparticles coatings on the heating surface can significantly promote the heat transfer of micro thermosyphon. Nanoparticles kind has very important influence on the CHF and HTC, and CuO nanoparticle is the suitable selection. Meantime, there are the optimum nanoparticles mass concentrations corresponding to the greatest CHF and HTC enhancement effect. It is about 0.75 wt% for Cu and about 1.0 wt% for both CuO and Al2O3. (4) The impact of the sizes of micro-channels on the CHF enhancing ratio is complex, but this impact is relatively weak and can be ignored. The CHF enhancing ratio is in the range of 1.7–2.3 for horizontal state and 1.8–2.4 for vertical state. The mean CHF enhancing ratio in vertical state is slightly larger by 10% than that in horizontal state. (5) Based on the experimental results, the coating technology using nanofluids is a feasible passive technology for 3D stacked ship cooling. It has highly industrial application prospect for 3D stacked ship cooling.
550
Conflict of interest
h /W/m2/K
500 450
We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.
400 350 300 250
R113 on smooth surface
Acknowledgments
200 150
Smooth surface Optimum Al2O3 coating
100
Optimum Cu coating Optimum CuO coating
50 0 0
2
4
6
8
This work was supported by the Shanghai Aerospace Science and Technology Innovation Fund (No. SAST201247). References
10 12 14 16 18 20 22 24 26 28 30
Fig. 14. Effects of coatings on HTC using R113 based moist fluid in vertical channel.
[1] N.H. Khan, S.M. Alam, S. Hassoun, System-level comparison of power delivery design for 2D and 3D ICs, in: 2009 IEEE International Conference on 3D System Integration, 2009 3DIC. IEEE2009. pp. 1–7.
474
S.-F. Li et al. / International Journal of Heat and Mass Transfer 124 (2018) 463–474
[2] H.-C. Chien, J.H. Lau, Y.-L. Chao, M.-J. Dai, R.-M. Tain, L. Li, et al., Thermal evaluation and analyses of 3D IC integration SiP with TSVs for network system applications, in: 2012 IEEE 62nd Electronic Components and Technology Conference. IEEE 2012. pp. 1866–1873. [3] R.L. Webb, Next generation devices for electronic cooling with heat rejection to air, J. Heat Transf. 127 (2005) 2–10. [4] K. Cheon, Deionized water cooling type cooling system for electronic device, Google Patents, 2002. [5] D. Enescu, E.O. Virjoghe, A review on thermoelectric cooling parameters and performance, Renew. Sustain. Energy Rev. 38 (2014) 903–916. [6] I. Chowdhury, R. Prasher, K. Lofgreen, G. Chrysler, S. Narasimhan, R. Mahajan, et al., On-chip cooling by superlattice-based thin-film thermoelectrics, Nat. Nanotechnol. 4 (2009) 235–238. [7] J. Wang, X.-J. Zhao, Y.-X. Cai, C. Zhang, W.-W. Bao, Experimental study on the thermal management of high-power LED headlight cooling device integrated with thermoelectric cooler package, Energy Convers. Manage. 101 (2015) 532– 540. [8] G. Satish Kandlikar, Review and projections of integrated cooling systems for three-dimensional integrated circuits, J. Electron. Packag. 136 (2014) 024001– 24011. [9] Fabio Alfieri, Manish K. Tiwari, Igor Zinovik, Dimos Poulikakos, Thomas Brunschwiler, Bruno Michel, 3D integrated water cooling of a composite multilayer stack of chips, J. Heat Transf. 132 (2010) 121402–121411. [10] Keiji Matsumoto, Hiroyuki Mori, Yasumitsu Orii, Thermal performance evaluation of dual-sidecooling for a three-dimensional (3D) chip stack, ICEP 2016 Proceedings, TC1-2, 163–168. [11] Md Shahriar Shamim, Amlan Ganguly, Chetan Munuswamy, Jayanti Venkatarman, Jose Hernandez, Satish Kandlikar, Co-design of 3D Wireless Network-on-Chip Architectures with Microchannel-Based Cooling, IEEE, 2015, 978-1-5090-0172-9/15. [12] S. Tan, K.C. Toh, N. Khan, D. Pinjala, V. Kripesh, Development of single phase liquid cooling solution for 3-D silicon modules, IEEE Trans. Compon. Packag. Manuf. Technol. 1 (2011) 536–544. [13] Namjae Kim, Shiho Kim, Self-convectional three-dimensional integrated circuit cooling system using micro flat heat pipe for portable devices, Heat Transf. Eng. 35 (2014) 924–932. [14] Namjae Kim, Shiho Kim, Self-convectional three-dimensional integrated circuit cooling system using micro flat heat pipe for portable devices, Heat Transf. Eng. 35 (2014) 924–932. [15] A. Sridhar, A. Vincenzi, D. Atienza, T. Brunschwiler, 3D-ICE: a compact thermal model for early-stage design of liquid-cooled ics, IEEE Trans. Comput. 63 (2014) 2576–2589. [16] M.M. Sarafraz, F. Hormozi, S.M. Peyghambarzadeh, Thermal performance and efficiency of a thermosyphon heat pipe working with a biologically ecofriendly nanofluid, Int. Commun. Heat Mass Transfer 57 (2014) 297–303. [17] M.M. Sarafraz, F. Hormozi, S.M. Peyghambarzadeh, Role of nanofluid fouling on thermal performance of a thermosyphon: are nanofluids reliable working fluid?, Appl Therm. Eng. 82 (2015) 212–224. [18] M.M. Sarafraz, F. Hormozi, S.M. Peyghambarzadeh, Pool boiling heat transfer to aqueous alumina nano-fluids on the plain and concentric circular microstructured (CCM) surfaces, Exp. Therm Fluid Sci. 72 (2016) 125–139. [19] Kai-lun Zhang, Zhen-Hua Liu, Bao-chen Zheng, A new 3D chip cooling technology using micro-channels thermosyphon with moist fluids and nanofluids, Energy Convers. Manage. 128 (2016) 44–56.
[20] Wael I.A. Aly, Moustafa A. Elbalshouny, H.M. Abd El-Hameed, M. Fatouh, Thermal performance evaluation of a helically-micro-grooved heat pipeworking with water and aqueous Al2O3 nanofluid at different inclinationangle and filling ratio, Appl. Therm. Eng. 110 (2017) 1294–1304. [21] Wang Xiaowu, Tang Yong, Chen Ping, Investigation into performance of a heat pipe with micro grooves fabricated by extrusion–ploughing process, Energy Convers. Manage. 50 (2009) 1384–1388. [22] Mehdi Famouri, Gerardo Carbajal, Chen Li, Transient analysis of heat transfer and fluid flow in a polymer-based Micro Flat Heat Pipe with hybrid wicks, Int. J. Heat Mass Transf. 70 (2014) 545–555. [23] M.M. Rahman, M. Saha, M.M.K. Bhuiya, A. Biswas, H. Md, C.Md. Alam, Feroz, Heat transfer characteristics of a parallel miniature heat pipe system, Int. Commun. Heat Mass Transfer 79 (2016) 1–8. [24] Y.P. Chen, F.W. Yu, C.B. Zhang, X.D. Liu, Experimental study on thermohydrodynamic behaviors in miniaturized two-phase thermosyphons, Int. J. Heat Mass Transf. 100 (2016) 550–558. [25] Ping-yang Wang, Shuang-fei Li, Zhen-hua, Liu, Natural convective boiling in horizontal and inclined micro-channels structure using super-moist fluids for cooling 3D stacked chip, Int. J. Heat Mass Transf. 107 (2017) 479–487. [26] Z.-H. Liu, R.-L. Hu, L. Lu, F. Zhao, H.-s. Xiao, Thermal performance of an open thermosyphon using nanofluid for evacuated tubular high temperature air solar collector, Energy Convers. Manage. 73 (2013) 135–143. [27] L.S. Sundar, M.H. Farooky, S.N. Sarada, M. Singh, Experimental thermal conductivity of ethylene glycol and water mixture based low volume concentration of Al2O3 and CuO nanofluids, Int. Commun. Heat Mass Transf. 41 (2013) 41–46. [28] V.L. Vinodhan, K.S. Suganthi, K.S. Rajan, Convective heat transfer performance of CuO–water nanofluids in U-shaped minitube: potential for improved energy recovery, Energy Convers. Manage. 118 (2016) 415–425. [29] R. Hajian, M. Layeghi, K.B. Sani, Experimental study of nanofluid effects on the thermal performance with response time of heat pipe, Energy Convers. Manage. 56 (2012) 63–68. [30] S. Vafaei, D. Wen, Flow boiling heat transfer of alumina nanofluids in single microchannels and the roles of nanoparticles, J. Nanopart. Res. 13 (2011) 1063–1073. [31] Y. Diao, Y. Liu, R. Wang, Y. Zhao, L. Guo, Experimental investigation of the Cu/ R141b nanofluids on the evaporation/boiling heat transfer characteristics for surface with capillary micro-channels, Heat Mass Transf. 50 (2014) 1261– 1274. [32] M.M. Sarafraz, F. Hormozi, M. Kamalgharibi, Sedimentation and convective boiling heat transfer of CuO-water/ethylene glycol nanofluids, Heat Mass Transf. 50 (9) (2014) 1237–1249. [33] M.M. Sarafraz, F. Hormozi, Convective boiling and particulate fouling of stabilized CuO-ethylene glycol nanofluids inside the annular heat exchanger, Int. Commun. Heat Mass Transf. 53 (4) (2014) 116–123. [34] A.E. Bergles, R.M. Manglik, Current progress and new developments in enhanced heat and mass transfer, J. Enhanc. Heat Transf. 20 (1) (2013) 1–15. [35] N. Chen, Solvent Handbook, Chemical Industry Press, Beijing, 1997. [36] Z.-H. Liu, Y.-M. Chen, Study of critical heat flux during natural convective boiling in vertical tubes and vertical annular tubes submerged in saturated liquids, JSME Int. J. Ser. B 42 (1999) 731–736. [37] H. Peng, G. Ding, H. Hu, Effect of surfactant additives on nucleate pool boiling heat transfer of refrigerant-based nanofluid, Exp. Therm Fluid Sci. 35 (2011) 960–970.