## Abstract

In order to meet the increasing performance demand of high-performance computing and edge computing, thermal design power (TDP) of central processing unit (CPU) and graphics processing unit (GPU) needs to increase. This creates thermal challenge to corresponding electronic packages with respect to heat dissipation. In order to address this challenge, two-phase immersion cooling is gaining attention as its primary mode of heat of removal is via liquid-to-vapor phase change, which can occur at relatively low and constant temperatures. In this paper, an integrated heat spreader (IHS) with boiling enhancement features is proposed. Three-dimensional metal printing and metal injection molding (MIM) are the two approaches used to manufacture the new IHS. The resultant IHS with boiling enhancement features is used to build thermal test vehicles (TTV) by following the standard electronic package assembly process. Experimental results demonstrate that boiling enhanced TVs improved two-phase immersion cooling capability by over 50% as compared to baseline TTV without boiling enhanced features.

## 1 Introduction

Driven by the increasing performance demand of high-performance computing and edge computing, thermal design power (TDP) of server electronic packages is trending toward 350 W and above. For graphics processing unit (GPU) and artificial intelligence (AI) chips, the TDP sees a similar increasing trend toward or beyond 400 W. This creates challenges for server/AI electronic package design with respect to heat dissipation. Note that a server package is typically lidded with a copper-based integrated heat spreader (IHS). A lidded package usually leads to a separate “plate-type” heat sink being placed on top of the IHS to remove heat.

Given the TDP levels mentioned above, using liquid as the primary medium to carry away the heat being dissipated by the device is more appropriate. At a high level, liquid cooling can be divided into two main approaches. Indirect liquid cooling refers to liquid cooling of ingredients that helps cool the air used for cooling the system/racks in the data centers. Direct liquid cooling refers to liquid cooling directly of the chips and the examples are cold plate liquid cooling and immersion. Among the direct liquid cooling on CPU/GPU/AI chips, the more common approach is based on flowing liquid through the internal fin structures of a “cold plate”, as shown in Fig. 1. This cold plate is placed directly over the IHS of the CPU/GPU. The other approach is called immersion cooling. As the name indicates, in this approach, the entire motherboard is soaked into a dielectric fluid. The heat is then carried away by either single-phase convective heat transfer called single-phase immersion cooling or by boiling of the fluid called two-phase immersion cooling. Two-phase immersion will be the focus of this paper.

Fig. 1
Fig. 1

Since a large amount of heat can be removed by liquid-to-vapor phase change (i.e., boiling) at relatively constant temperatures determined by the boiling point of the fluid, two-phase immersion is becoming more popular for cooling high TDP devices. Previous research work in two-phase immersion cooling has been focused on the surface boiling enhancement of cooling components apart from the electronic package. There are previous studies of boiling enhancement coating (BEC) on separate copper plates or silicon surfaces [15]. These studies have been focused on boiling enhancement on the silicon surface, or copper surface by introducing different surface treatments or adding micro/nanoscale particles onto the surface. One previous study shows free particles as BEC on a heated copper surface helped achieve the best boiling heat transfer at higher heat fluxes, with a 32% reduction in the surface superheat compared to the polished surface at 112 kW/m2 [1]. This type of cooling architecture requires an additional layer of thermal interface material (TIM) to be placed between the plate and the IHS. This will reduce the effectiveness of two-phase immersion cooling due to additional thermal resistances from the plate and the TIM.

In this paper, a new IHS with boiling enhancement features directly integrated onto it is proposed. Three-dimensional (3D) metal printing and metal injection molding (MIM) approaches are used to create the new IHS. The new IHSs with two-phase immersion cooling enhancement features are then used to build thermal test vehicles (TTVs) by following standard electronic package assembly process, including die attach and TIM/IHS attach onto the organic substrate (with copper routing and vias), which is made of organic small molecules or polymers compared to ceramic substrate which is made from the ceramic material. In order to design the electronic packages with a right set of features for two-phase immersion cooling, one must start with an understanding of the three distinct regions in Nukiyama's boiling curve: nucleate boiling, transition boiling, and film boiling [6]. Entering the transition or film boiling regime is predicated on meeting the critical heat flux (CHF) condition for the surface where operating temperatures instantaneously increase several orders of magnitude, and it would result in a catastrophic effect on the immersed packaged electronic devices. Therefore, the surface must be designed such that the CHF is always greater than the power density and/or the nucleate boiling regime is extended to the greatest length possible in order to avoid this catastrophic scenario.

There are three major vectors of attack to enlarge the domain of nucleate boiling:

1. Lower the boiling temperature of coolant (incipient point),

2. Increase the slope of boiling curve in nucleate boiling region,

3. Increase the CHF.

The three vectors for improving cooling performance in nucleate boiling region can be carried out at the system level and/or component level. Improvements at the system level refer to adjustments/changes done on a global scale (e.g., adjustments that affect the entire immersion tank containing all motherboards). Component level improvements refer to corresponding feature adjustments/changes that affect an individual device's electronic package. To lower the incipient point at the system level, Raney et al. [7] have been shown that this can be achieved by pressurizing the immersion tank. They have attributed this lowered temperature of boiling to the increased range of cavity radius due to higher pressure. According to Refs. [8,9], applying ultrasonic vibration or electric field to the immersion tank can lead to higher heat flux at a given wall superheat temperature. In other words, the slope of the boiling curve is increased as a result. Raney et al. [7] further investigated the effects of subcooling and dissolved gas (air) on the pool boiling heat transfer performance of microporous enhanced and pin-finned surfaces. It has also been shown that the CHF can be increased by cooling the whole immersion fluid below its boiling temperature and increased pressure [7]. It's also reported a dissolved gas content increase with increasing pressure and subcooling.

The same three vectors of attack can be adopted at the component level. For example, to lower the starting point of boiling, it has shown that this can be achieved by creating submicron features on metallic surfaces such as IHS [1012]. Sand blasting, loose powder sintering, and laser machining are the processes proposed by these references to create the features. Although laser machining is better than the others at creating the desired submicron features, one must also consider cost and compatibility with existing assembly flow in selecting the final manufacturing process. For standard lidded server packages, both sand blasting and loose powder sintering (3D metal printing) are more appropriate in this regard. It has shown that coating the base material (copper or nickel) with high thermal conductivity carbon nanotubes can increase the slope of the boiling curve [13]. However, also according to [13], coating the base material with carbon nanotubes requires relatively high-temperature conditions (up to 550 °C) and plasma-enhanced chemical vapor deposition. This is highly likely to drive up the cost and complexity of IHS manufacturing. At the component level, the CHF can be enhanced through an extended surface area effect by introducing appropriate BEC onto the surface, like sintered copper or pin fin structure. Several works have shown potential approaches that extend surface area can be integrated onto the surface [1416]. By comparing the different approaches by utilizing extended surface area effect, circular pin fin with microstructure appears to have the best performance and fit for IHS manufacturing via 3D printing and MIM processes. Furthermore, the resultant IHS will also be compatible with existing assembly process used to build lidded electronic packages.

The rest of the paper is divided into four sections. In Sec. 2, detailed descriptions of the processes (3D print and MIM) used to manufacture the pin fin IHS are provided. Details on the TTV and its assembly process are given in Sec. 3. Experimental setup and procedures to carry out two-phase immersion experiments are provided in Sec. 4. Finally, results and discussion are shared in Sec. 5 of the paper.

## 2 Boiling Enhancement on Lidded Electronic Packages

This work is focused on IHS with pin fins containing microstructures to improve heat transfer. Two manufacturing processes are used to construct the proposed IHS architecture. The first process is based on MIM and the second process is based on 3D metal printing. Details for each manufacturing process will be presented below.

### 2.1 Metal Injection Molding Process and Optimization to Build Integrated Heat Spreader With Boiling Enhancement Features.

MIM is a manufacturing process that can be used to build complex shapes/features in one single piece of metal. The feature size can be as small as 50 μm. It takes following key steps in manufacturing a copper IHS with pin fins containing microstructures.

1. The copper powder and binder are first mixed and then the part is molded in the injection machine to form desired fin features using corresponding hard tooled jig.

2. The part then goes through debinding process.

3. The green part is then sintered to finalize the features on its peripheral surface.

4. The intermediate sample then goes through computer numerical control machining, sandblasting, and nickel-plating to deliver the final IHS with desired boiling enhancement features.

Figure 2 shows a sample of the pin fin IHS built via the above MIM process. Note the peripheral surface of each pin fin has rectangular protrusion features with a nominal cross section of ∼300 × 400 μm. Note that the primary purpose of the pin fin and the stud-like features are meant to increase the available boiling surface, which will boost the CHF. In order to lower the incipient point, bubble nucleation sites on the IHS need to be sized and created. Hsu's criterion [17] is used to determine the appropriate size for bubble nucleation sites [18]
${Rc,min,Rc,max}=δtC22C1(Tw−Tsat)(Tw−Tbulk)×[1∓1−8C1σTsat(Tw−Tbulk)ρvhlvδt(Tw−Tsat)2]$
(1)
where $Rc,min$ and $Rc,max$ represent the range of radii at the mouth of a conical crevice, as illustrated in Fig. 3, suitable for bubble nucleation, $δt$ represents the thickness of the superheated liquid film right above the boiling surface, $Tw$ is the wall temperature, $Tsat$ is the saturation temperature of the fluid, $Tbulk$ is the remote fluid temperature, $ρv$ is the density of the gaseous phase, $hlv$ is the latent heat of vaporization, and $σ$ is the surface tension of the fluid. $C1$ and $C2$ are functions of contact angle, $θ0$, as given in Eqs. (2) and (3)
$C1=(1+cos θ0)/sinθ0$
(2)
$C2=sinθ0$
(3)
Fig. 2
Fig. 2
Fig. 3
Fig. 3

Based on preliminary test data using typical two-phase coolant 3 M FC-72, the values used in Eqs. (1)(3) for sizing the nucleation sites are given in Table 1. Based on these values, the range, as given by $Rc,min$ and $Rc,max$ was 0.5–75 μm.

Table 1

Input values for sizing nucleation sites in FC-72 [17]

 Tw (K) 338 Wall temperature Tbulk (K) 324 Remote fluid temperature Tsat (K) 329 Saturation temperature of fluid θ (Rad) 0.87 Contact angle hlv (J/kg) 8.80 × 104 Laten heat of vaporization σ (N/m) 1.00 × 10−2 Surface tension of fluid δt (m) 2.53 × 10−4 Film thickness
 Tw (K) 338 Wall temperature Tbulk (K) 324 Remote fluid temperature Tsat (K) 329 Saturation temperature of fluid θ (Rad) 0.87 Contact angle hlv (J/kg) 8.80 × 104 Laten heat of vaporization σ (N/m) 1.00 × 10−2 Surface tension of fluid δt (m) 2.53 × 10−4 Film thickness

Sandblasting was used to create the features needed for bubble nucleation. Typical surface morphology posts and blasting treatment was checked using a scanning electron microscope, shown in Fig. 4. Moreover, how the radius at the mouth of the conical crevice is measured is also shown in the same figure. A design of experiments (DOE) was used to optimize the parameters in the sandblasting process to achieve the desired feature size. These process factors include sand size, sandblast pressure, and sandblast angle. Table 2 summarizes this DOE with 9 randomized runs. DOE results, as listed in Table 2, indicate that the desired surface morphology can be achieved with angle of 90-deg, sand grit size of # 80 (equivalent to 165 μm on average), and blasting load of 3 kg (measured in equivalent weight and used in machine setup). Furthermore, the main effect plot, as shown in Fig. 5, demonstrates that sand size is the key factor affecting surface roughness on the IHS surface in the parameter ranges under investigation.

Fig. 4
Fig. 4
Fig. 5
Fig. 5
Table 2

Optimization DOE matrix

PatternAngle (deg)Sand grit sizeLoad in equivalent weight (kg)Rsm (μm)
---30# 803164
00060# 1505175
-++30# 2405113
+--90# 803145
+++90# 2407131
--+30# 807187
++-90# 2403211
+-+90# 807121
-+-30# 2403109
PatternAngle (deg)Sand grit sizeLoad in equivalent weight (kg)Rsm (μm)
---30# 803164
00060# 1505175
-++30# 2405113
+--90# 803145
+++90# 2407131
--+30# 807187
++-90# 2403211
+-+90# 807121
-+-30# 2403109

### 2.2 Three-Dimensional Printing to Build Integrated Heat Spreader With Boiling Enhancement Features.

In the 3D metal printing process, direct metal laser sintering (DMLS) technique was utilized to print out an array of studs with fine features on the peripheral of each stud. Note that the array was printed on a copper base to form a single piece of lid for package assembly. The design of the lid used in DMLS was slightly different from the one used in MIM. The slight difference was in the protrusion feature size on the peripheral of the studs. In the DMLS, the nominal cross section was targeted at ∼70 × 70 μm to further enhance both bubble nucleation and CHF. Note that the minimum layer height that a DMLS process can print is in the range of 30–150 μm. For smaller features, DMLS will have a difficult time resolving. Figure 6 shows the resultant IHS with pin-fin features manufactured via a 3D metal printing process. The nickel plating on the IHS was carried out separately post-the DMLS process. The resultant IHS sample shows a rough surface, as shown in Fig. 7(a). Specifically, surface roughness was measured at multiple points on the top surface of 3D printed IHS across the whole IHS, including studs, and the surface roughness RZ(din) was measured mainly in range of 150 to 250 μm. The corresponding surface profile of the IHS made from MIM is shown in Fig. 7(b). The surface roughness RZ(din) of the IHS made from MIM was measured mainly in range of 40–80 μm. This comparison shows that the roughness of IHS made from MIM process compares better with targets than 3D printed IHS given by Eq. (1). Note that the protrusion features on the peripheral of the studs are not as pronounced when compared to a sample produced via the MIM process. This is due to limitation in capability of DMLS to fully resolve the fine features.

Fig. 6
Fig. 6
Fig. 7
Fig. 7

## 3 Thermal Test Vehicle of Lidded Electronic Package

Thermal test vehicle units have been developed using lids with the different configurations shown in Table 3. The key steps in the assembly process for building the TTVs are listed below and illustrated in Fig. 8.

Fig. 8
Fig. 8
Table 3.

Configurations of TTV units with different boiling enhancement features

Type #Configuration of lidded TTV unitNotes
1IHS without boiling enhancement featureReference
2IHS with copper meshingReference
3IHS with pin fins made from MIM-5 mm highOptimized sandblasting
4IHS with pin fins made from MIM-10 mm highOptimized sandblasting
5IHS with 3D printed pin fins-10 mm highDMLS
Type #Configuration of lidded TTV unitNotes
1IHS without boiling enhancement featureReference
2IHS with copper meshingReference
3IHS with pin fins made from MIM-5 mm highOptimized sandblasting
4IHS with pin fins made from MIM-10 mm highOptimized sandblasting
5IHS with 3D printed pin fins-10 mm highDMLS
1. Silicon thermal test wafer manufacturing. The silicon thermal test chip is built with embedded resistive temperature detectors (RTDs). Each test cell is 2.5 × 2.5 mm2 and comes with an RTD. For each TTV unit, it has 132 RTDs in total. Since each test cell has an embedded RTD, the RTDs can be seen to evenly distribute across the silicon die.

2. Singulation of silicon thermal tester wafer.

3. Silicon thermal test die-attach onto organic substrate through solder joint reflow.

4. Apply underfill to protect first-level interconnect.

5. IHS with or without boiling enhancement feature attachment onto the silicon die and organic substrate using thermal interface material between the silicon die and IHS and sealant to bond IHS to substrate.

6. Thermal calibration on assembled TTV units, including RTD calibration using standard thermal bath, and the calibration accuracy, can reach 0.5 °C and better. Twenty RTDs across the whole silicon die was selected for calibration for each TTV unit. The power calibration was conducted by running a power sweep from 40 to 800 W. The RTD temperature reading shows a linear relationship with the input power.

Please note that the IHS-es with or without boiling enhancement features were first built separately, and then assembled onto the TTV units. Moreover, all lids were grooved and pre-attached with thermocouples type T.

The assembled lidded TTV units are shown in Fig. 9. The lidded TTV without boiling enhancement feature (part (a) in Fig. 9) serves as a baseline for assessing relative improvement in boiling heat transfer for other configurations shown in Fig. 9.

Fig. 9
Fig. 9

## 4 Experimental Setup

A two-phase immersion cooling test was conducted to quantify performance of boiling enhancement features using the TTV units. A TTV unit with copper meshing attached to the IHS was also included in the study. The purpose of including this leg is to assess the effectiveness of boiling enhancement features other than the pin fin type.

The two-phase immersion cooling test setup is shown in Fig. 10. As shown in Fig. 10, this two-phase immersion cooling tester has a compact design with a copper condenser on the top of the tank. The printed circuit board containing a socketed TTV unit is vertically placed inside the test chamber. Several thermocouples are attached to the frame close to the TTV to record coolant temperature at multiple locations. A picture of the actual test board with assembled TTV unit immersed in the tester is shown in Fig. 11. Figure 11 shows corresponding spatial arrangements of the test board and key thermocouple locations. The coolant used in this experiment is 3 M FC-72 with boiling point of 56 °C under 101 kPa.

Fig. 10
Fig. 10
Fig. 11
Fig. 11

## 5 Experimental Results and Discussions

Thermal tests were conducted on TTV units of lidded electronic packages with different boiling enhancement features. The standard lidded electronic package (no boiling enhancement) is used as a reference. Each TTV unit was powered on across multiple dies and the power of each die was recorded. The temperature limit of silicon built into the TTV is 125 °C. For the TTV unit without boiling enhancement features, it could only be tested up to a total power of 600 W before reaching its temperature limit. Several lid configurations, as shown in Fig. 12, were able to go as high as 900 W and beyond during the test. The limiter is package thermal resistance of this TTV and the maximum power of this TTV in this two-phase immersion test can be extended with reduced package thermal resistance. The case temperature of the IHS and the liquid temperature of the coolant were recorded using pre-attached thermocouples, as shown in Fig. 11. These temperature values were then used to calculate effective thermal resistances from the case of the IHS to the coolant for the various builds shown in Fig. 9.

Fig. 12
Fig. 12
The thermal resistance between the case of the lid and the coolant, $φcf$, is defined as
$φcf=(Tc−Tf)P$
(4)

where $φcf$, thermal resistance between the case of the lid and coolant,$Tc$, case temperature of lid read from thermocouple attached on top of lid,$Tf$, fluid temperature of two-phase coolant read from thermocouples attached to the frame inside chamber,$P$, total power of TTV unit.

$φcf$-es of the different configurations are shown in Fig. 12. As shown in Fig. 12, the multilayer copper meshing configuration has the lowest thermal resistance in the relatively low-power regime. This is to be expected because there exist significantly more bubble nucleation sites in the copper meshes than the other configurations. In the relatively high-power regime, pin fin structures achieved the lowest thermal resistance. This is because, in the relatively high-power regime, cooling via boiling is dominated by surface area and pin structure has significantly more surface area than the other configurations. This is particularly true for pin-fin produced using MIM process.

Interestingly, 3D printed pin fin unit did not achieve the same cooling capability as the sample made with MIM process, even when the 3D printed unit had a rougher surface. Possible root cause is that 3D printed copper lid is more porous than MIM copper lid. This would translate to a lower effective thermal conductivity and result in:

1. Lower the fin efficiency of the pin fin features; this would reduce the heat that can be carried away in the vertical direction of lid.

2. Lower the capability of the lid to spread the heat evenly in the plane of the lid. This would result in hotspots and reduced how much heat can be conducted away.

The images of boiling for lidded TTV units made with different boiling enhancement features are shown in Fig. 13. The corresponding heat flux versus ΔT for the different configurations is shown in Fig. 14. A curve for pure copper surface is added as Ref. [19]. As shown in Fig. 14, the copper mesh lowers the incipient point the most versus the other configurations. Therefore, it is not surprising that it has the best performance in the low-power regime. Although the pin structure did not achieve the lowest incipient point, it was still able to lower the start of boiling point against bare IHS. Since the test was stopped before the actual CHF was reached for each of the configurations, one cannot say conclusively which configuration had the highest CHF. However, based on the slopes of the curves shown in Fig. 14, it is expected that the 10 mm pin fin IHS made with MIM will achieve the highest CHF. This observation supports the experimental result that the MIM 10 mm pin fin was able to achieve the best cooling performance in the high-power regime.

Fig. 13
Fig. 13
Fig. 14
Fig. 14

It should be noted that even though 5 mm high pin fin on MIM IHS sample shows a slightly worse $φcf$ compared to that of 10 mm version, moving into the engineering exercise, a saving of 5 mm in height for 5 mm high pin fin MIM IHS versus 10 mm high version is a big advantage for lidded electronic package from aspects of the material cost, assembly cost and overall final platform total cost of ownership in immersion cooling adoption for the data centers. When moving forward to the design down selction for engineering implementation, this needs to be included into the consideration with a large sample size in engineering build and data collection.

## Summary

In this paper, lidded electronic packages with boiling enhancement features were built and tested. Their cooling performances were compared to a standard lidded package in two-phase immersion cooling. The best results were obtained with the lid containing 10 mm high pin fin features made via MIM. Thermal resistance data showed a 40% reduction relative to a standard lidded package. With respect to cooling capability, the same boiling enhanced lidded package showed an increase of 50% over standard lidded electronic package. Based on these results, integrating boiling enhancement features directly onto lidded packages appears to be a viable path to achieve cooling capability beyond 1 kW in two-phase immersion.

## Acknowledgment

The authors would like to thank Long Win Science and Technology Corporation for their support in two-phase immersion cooling test and providing pictures of test-up and test videos, Microloops Corporation for copper meshing layer prototyping, and Amulaire Thermal Technology in collaboration of lid prototyping using MIM technique and sandblasting study, Nanotest in collaboration of TTV sample prototyping. The authors also would like to acknowledge the colleagues from liquid cooling team within Intel for valuable discussions.

## References

1.
Sarangi
,
S.
,
Weibel
,
J. A.
, and
Garimella
,
S. V.
,
2015
, “
Effect of Particle Size on Surface-Coating Enhancement of Pool Boiling Heat Transfer
,”
Int. J. Heat Mass Transfer
,
81
, pp.
103
113
.10.1016/j.ijheatmasstransfer.2014.09.052
2.
Hsu
,
Y.-T.
,
Li
,
J.-X.
, and
Lu
,
M.-C.
,
2018
, “
Enhanced Immersion Cooling Using Two-Tier Micro- and Nano-Structures
,”
Appl. Therm. Eng.
,
131
, pp.
864
873
.10.1016/j.applthermaleng.2017.12.067
3.
El-Genk
,
M. S.
, and
Ali
,
A. F.
,
2010
, “
Enhancement of Saturation Boiling of PF-5060 on Microporous Copper Dendrite Surfaces
,”
ASME. J. Heat Trans. ASME
,
132
(
7
), p.
071501
.10.1115/1.4000975
4.
Gajghate
,
S. S.
,
Barathula
,
S.
,
Das
,
S.
,
Saha
,
B. B.
, and
Bhaumik
,
S.
,
2020
, “
Experimental Investigation and Optimization of Pool Boiling Heat Transfer Enhancement Over Graphene-Coated Copper Surface
,”
J. Therm. Anal. Calorim.
,
140
(
3
), pp.
1393
1411
.10.1007/s10973-019-08740-5
5.
Rishi
,
A. M.
,
Kandlikar
,
S. G.
, and
Gupta
,
A.
,
2020
, “
Salt Templated and Graphene Nanoplatelets Draped Copper (GNP-draped-Cu) Composites for Dramatic Improvements in Pool Boiling Heat Transfer
,”
Sci. Rep.
,
10
(
1
), p.
11941
.10.1038/s41598-020-68672-1
6.
Ghiaasiaan
,
S.
, and
Mostafa
,
2008
,
Two-Phase Flow, Boiling and Condensation in Conventional and Miniature Systems
,
Cambridge University Press
, Cambridge, UK, pp.
287
318
.
7.
Rainey
,
K. N.
,
You
,
S. M.
, and
Lee
,
S.
,
2003
, “
Effect of Pressure, Subcooling, and Dissolved Gas on Pool Boiling Heat Transfer From Microporous Square Pin-Finned Surfaces in FC-72
,”
Int. J. Heat Mass Transfer
,
46
(
1
), pp.
23
35
.10.1016/S0017-9310(02)00257-0
8.
Legay
,
M.
,
Gondrexon
,
N.
,
Person
,
S. L.
,
Boldo
,
P.
, and
Bontemps
,
A.
,
2011
, “
Enhancement of Heat Transfer by Ultrasound: Review and Recent Advances
,”
Int. J. Chem. Eng.
,
2011
, pp.
1
17
.10.1155/2011/670108
9.
Yabe
,
A.
,
Nakayama
,
W.
, and
Di Marco
,
P.
,
1999
, “
Enhancement Techniques in Pool Boiling
,”
Handbook of Phase Change: Boiling and Condensation
,
Taylor & Francis
,
, pp.
121
141
(editor-in-chief
Satish G.
Kandlikar
, co-editors
Masahiro
Shoji
and
Vijay K.
Dhir
).
10.
Mudawar
,
I.
, and
Anderson
,
T. M.
,
1993
, “
Optimization of Enhanced Surfaces for High Flux Chip Cooling by Pool Boiling
,”
ASME J. Electron. Packag.
,
115
(
1
), pp.
89
100
.10.1115/1.2909306
11.
Zhang
,
S.
,
Tang
,
Y.
,
Zeng
,
J.
,
Yuan
,
W.
,
Chen
,
J.
, and
Chen
,
C.
,
2016
, “
Pool Boiling Heat Transfer Enhancement by Porous Interconnected Microchannel Nets at Different Liquid Subcooling
,”
Appl. Therm. Eng.
,
93
, pp.
1135
1144
.10.1016/j.applthermaleng.2015.10.044
12.
Gregorcic
,
P.
,
Zupancic
,
M.
,
Iztok
,
G.
,
2018
, “
Scalable Surface Micro Structuring by Fiber Laser for Controlled Nucleate Boiling Performance of High- and Low Surface-Tension Fluids
,” accessed Oct. 25, 2021, www.nature.com/scientificreports
13.
Ujereh
,
S.
,
Fisher
,
T.
, and
Mudawar
,
I.
,
2007
, “
Effects of Carbon Nanotube Arrays on Nucleate Pool Boiling
,”
Int. J. Heat Mass Transfer
,
50
(
19–20
), pp.
4023
4038
.10.1016/j.ijheatmasstransfer.2007.01.030
14.
Mori
,
S.
, and
Okuyama
,
K.
,
2009
, “
Enhancement of the Critical Heat Flux in Saturated Pool Boiling Using Honeycomb Porous Media
,”
Int. J. Multiphase Flow
,
35
(
10
), pp.
946
951
.10.1016/j.ijmultiphaseflow.2009.05.003
15.
Zhao
,
H.
,
Dash
,
S.
,
Dhillon
,
N. S.
,
Kim
,
S.
,
Lettiere
,
B.
,
Varanasi
,
K. V.
, and
Hart
,
A. J.
,
2019
, “
Microstructured Ceramic-Coated Carbon Nanotube Surface for High Heat Flux Pool Boiling
,”
Appl. Nano Mater.
,
2
(
9
), pp.
5538
5545
.10.1021/acsanm.9b01116
16.
Ramaswamy
,
C.
,
Joshi
,
Y.
,
Nakayama
,
W.
, and
Johnson
,
W. B.
,
2003
, “
Effects of Varying Geometrical Parameters on Boiling From Microfabricated Enhanced Structures
,”
ASME J. Heat Transfer
,
125
(
1
), pp.
103
109
.10.1115/1.1513575
17.
Hsu
,
Y. Y.
,
1962
, “
On the Size Range of Active Nucleation Sites on a Heating Surface
,”
ASME J. Heat Transfer
,
84
(
3
), pp.
207
216
.10.1115/1.3684339
18.
Wu
,
Z.
, and
Sundén
,
B.
,
2014
, “
On Further Enhancement of Single-Phase and Flow Boiling Heat Transfer in Micro/Minichannels
,”
Renewable Sustainable Energy Rev.
,
40
, pp.
11
27
.10.1016/j.rser.2014.07.171
19.
Auracher
,
H.
, and
Buchholz
,
M.
,
2005
, “
Experiments on the Fundamental Mechanisms of Boiling Heat Transfer
,”
J. Braz. Soc. Mech. Sci. Eng.
, 27(
1
), pp. 1–22.10.1590/S1678-58782005000100001