www.nature.

com/scientificreports

OPEN Critical heat flux enhancement

in pool boiling through increased
rewetting on nanopillar array
Received: 13 September 2017
Accepted: 28 February 2018 surfaces
Published: xx xx xxxx
Thien-Binh Nguyen1, Dongdong Liu1, Md Imrul Kayes2, Baomin Wang2, Nabeel Rashin1, Paul
W. Leu2 & Tuan Tran 1
Boiling is a key heat transfer process for a variety of power generation and thermal management
technologies. We show that nanopillar arrays fabricated on a substrate enhance both the critical heat
flux (CHF) and the critical temperature at CHF of the substrate and thus, effectively increase the limit
of boiling before the boiling crisis is triggered. We reveal that the enhancement in both the CHF and the
critical temperature results from an intensified rewetting process which increases with the height of
nanopillars. We develop a predictive model based on experimental measurements of rewetting velocity
to predict the enhancement in CHF and critical temperature of the nanopillar substrates. This model is
critical for understanding how to control boiling enhancement and designing various nanostructured
surfaces into specific applications.

Boiling is a central phenomenon in technological and industrial applications as diverse as thermal management
in electronics, power generation and chemical processing1–3. In such applications, it is of immense importance to
enhance the energy efficiency by increasing the critical heat flux (CHF), the highest heat flux a boiling substrate
can achieve, as well as reducing operational risks caused by the notorious “boiling crisis”, a catastrophic failure in
boilers or heat exchanging devices4. The boiling crisis occurs on a boiling substrate when excessive vaporisation
of liquid forms a vapour layer that severely impedes heat transfer through the substrate. This leads to abrupt jump
in surface temperature and subsequently irreversible damages to the substrate of the boiling equipment. The tem-
perature Tc at which CHF occurs therefore is directly connected to the boiling crisis; a boiling system operating
at temperature higher than Tc inevitably drifts towards the boiling crisis. As a result, an enhancement in boiling
performance, without triggering the boiling crisis, requires significant increases in both the critical heat flux and
critical temperature Tc.
A large body of research has been dedicated to understanding the mechanism of nucleate boiling, a boiling
regime in which latent heat transfer is dominant compared to convective heat transfer. With increasing surface
temperature, CHF is the upper bound of heat flux in the nucleate boiling regime before the boiling crisis occurs5.
Various microscale models have been proposed to capture nucleate boiling heat transfer mechanisms6. In addi-
tion, numerous key parameters affecting the heat flux in the nucleate boiling regime have been identified, e.g.,
surface roughness ranging from nano- to micro-scales7, and wettability8. Nonetheless, a robust mechanistic pre-
diction of nucleate boiling heat transfer for a wide range of boiling conditions, including the surface properties,
is still not available9.
To increase the critical heat flux, numerous methods have been explored following two major approaches:
fluid modification and surface treatment. The former approach, which uses additives10,11 or nanoparticles12, has
not been widely applicable as it puts constraints on fluid selection and operating conditions of the boiling sys-
tem13. The latter approach includes either treatments to enhance surface wettability14,15, or surface morphological
alteration with porous coatings16,17, artificial fins18–20 and nano/microstructures13,21. Although numerous surface
modification methods have been found to increase CHF, in particular those utilising nano/microstructures, the

1
School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue,
Singapore, 639798, Singapore. 2Department of Industrial Engineering and Department of Mechanical Engineering
and Materials Science, University of Pittsburgh, Pittsburgh, 15261, USA. Correspondence and requests for materials
should be addressed to T.T. (email: [email protected])

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 1

www.nature.com/scientificreports/

Figure 1. SEM images showing nanopillar substrates of different pillar dimensions: the ratio l/db, defined as the
ratio between the pillar height l and base diameter db, and the corresponding surface roughness r are indicated
in the labels. (b) Representative snapshots showing boiling phenomenon on the smooth substrate (top panel)
and the nanopillar substrate having r = 3.2 (bottom panel) at several surface temperatures T. While the smooth
surface produces vapour bubbles which have tendency to coalesce (e.g., at T = 74 °C) and form vapour columns
(e.g., at T = 78 °C), the nanopillar substrate produce bubbles which have a narrow size distribution tend to
evolve separately without merging. The vapour column and bubbles are highlighted by red dashed lines.

roles of surface structures at different length scales in changing the heat flux and the critical temperature remain
elusive. On one hand, it was shown that the wickability effect is the major cause of enhancement in CHF13,21, i.e.,
the CHF value was shown to increase linearly with the wicked volume flux for various substrate materials and
different types of structures ranging from nanoscales to microscales21. On the other hand, the separate effect of
variation in nanoscale structures on CHF and the critical temperature is still a standing question. Although it has
been suggested that nanoscale structures may intensify the wetting velocity, which is a crucial factor leading to
CHF enhancement15,22, the relation between a systematic change in nanoscale structures, wetting velocity and
CHF has not been established. Moreover, little effort has been made to relate surface wettability or nanoscale
structures to the change in the critical temperature Tc.
In this study, we investigate the boiling phenomenon of FC-72 on nanostructured surfaces. Fluorinert liq-
uid FC-72 is an ideal coolant liquid commonly used in various heat management applications owning to its
low boiling point at 56 °C. We show that modifying the surface morphology at nanoscales leads to significant
enhancement in the boiling performance, including both the critical heat flux and the critical temperature. By
using nanopillars with systematically varying heights on silicon substrates, we demonstrate scalable boiling
enhancement with increasing height of nanopillars. We propose and experimentally verify a mechanistic and
predictive model relating the nanopillar-induced increase in rewetting to the enhancement in both CHF and
critical temperature.

Results and Discussion

Enhancement in critical heat flux and nucleate boiling limit. We systematically vary the height l of
nanopillars fabricated on silicon substrates (Fig. 1a) and show that increasing l leads to enhancement in both the
critical heat flux (CHF) and the upper limiting temperature of nucleating boiling. The nanopillars are fabricated
by an inductively coupled plasma reactive ion etching (ICPRIE) process that directly modifies the surface without
adding coating layers that may induce additional thermal resistance to the substrate23. Typically, the quantity used
for characterising the upper limiting temperature is either the CHF temperature Tc, or the Leidenfrost temper-
ature TL. The nanopillars with the base diameter db ≈ 440 nm are arranged in a hexagonal lattice with the lattice
pitch p ≈ 800 nm. The pillar height l is varied from 260 nm to 1390 nm. The differences in boiling behaviours
and performances between these substrates therefore are attributed to variation in l, or alternatively the surface
roughness r = At/Ap. Here At is the total surface area due to the presence of the nanopillars, and Ap is the projected
boiling area. In our experiment, r ranges from 1 for the smooth silicon substrate to 3.2 for the nanopillar substrate
with l = 1390 nm.
To qualitatively evaluate the effect of the nanopillars on boiling, we first compare the boiling behaviours
between the smooth substrate and the nanopillar substrate with r = 3.2 at several surface temperatures T (Fig. 1b).
Although the onset of boiling occurs at T ≈ 69 °C for both substrates, the size of bubbles generated on the smooth
substrate varies much more broadly than that on the nanopillar substrate. At higher temperature, the bubble
dynamics on the two substrates become even more contrasting: while the bubbles on the smooth substrate tend to
merge and create either large bubbles or vapour columns starting from the substrate, those generated on the nan-
opillar substrate have much lesser tendency to merge and appear more uniform in size. We note that it is not likely
for bubbles to coalesce after their detachment from the substrate due to liquid inertia and surface tension, thus the
coalescence process mainly occurs between consecutive bubbles generated from the same nucleation site before

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 2

www.nature.com/scientificreports/

Figure 2. Heat flux q versus wall superheat ΔT for smooth substrate (r = 1) and nanopillar substrates. Error
bars represent the standard deviation from multiple experiments. (b) Dependence of critical heat flux qc
(open squares), CHF temperature Tc (left-pointing triangles), and Leidenfrost temperature TL (right-pointing
triangles) on surface roughness r.

detachment. This strongly suggests that resistance to coalescence, as well as the uniform size of bubbles generated
from the nanopillar substrate result from an effective rewetting process, which facilitates bubble detachment from
the surface. We hypothesise that deviation in boiling behaviours of nanopillar substrates from the smooth one,
therefore, originates from the nanopillar-induced enhancement in the rewetting process.
We now quantify the effect of nanopillars on the boiling performance by measuring the heat flux q for each
substrate as a function of the superheat ΔT = T − Tb, where Tb = 56 °C is the boiling temperature of FC-72
at atmospheric pressure. In Fig. 2a, we show the dependence of q on ΔT for the smooth and all the nanopillar
substrates (See Supplementary Information for the uncertainty analysis of the heat flux and surface temperature
measurements). Starting from ΔT = 0, q first gradually increases due to natural convection until it reaches the
onset of nucleate boiling (ΔT ≈ 30 K for all tested substrates). In this so-called natural convection regime, since
the nanopillars are much smaller than the thermal boundary layer thickness, their presence causes negligible
effect to the heat flux, which is dictated by the convective fluid flow outside of the thermal boundary layer. Here,
the boundary layer thickness is roughly 57 µm, estimated using the ratio between the natural convection heat
transfer coefficient and the thermal conductivity of FC-7224.
As ΔT increases past the onset of nucleate boiling, the heat flux on nanopillar substrates, denoted as qn, devi-
ates substantially from that on the smooth substrate. In particular, qn after the system transitions into the nucleate
boiling regime is achieved at much lower superheat compared to that when system is still in the natural convec-
tion regime (inset of Fig. 2a). For instance, for the substrate with r = 2, q = 3.3W ⋅ cm−2 is achieved at ΔT ≈ 16
K when the system is in the nucleate boiling regime, whereas a slight increase in heat flux, q ≈ 3.5W ⋅ cm−2, is
achieved at much higher superheat ΔT ≈ 33 K in the natural convection regime. On the smooth substrate, how-
ever, the heat flux increases smoothly as the system transitions into the nucleate boiling regime. The presence of
nanopillars causes a substantial jump in heat flux as soon as the system transitions to the nucleate boiling regime
because the heat transfer mechanism is more efficient for the nanopillar substrates. The homogeneous spread of
nucleation sites, high generation frequency and narrow size distribution of bubbles on nanopillar substrates (see
Supplementary Information, Fig. S1) results in the sudden discontinuity in the boiling curve.
With the superheat increasing beyond the onset of nucleate boiling, bubbles are generated with higher fre-
quency and larger size, intensifying forced convection in the bulk liquid and subsequently enhancing the heat
flux q. For each substrate, q increases to the CHF qc at the critical temperature Tc, beyond which it drops sharply
due to excessive vapour generation and lack of replenishing liquid to the substrate. The heat flux reaches its min-
imum value at the so-called Leidenfrost temperature TL, or the upper limit of nucleate boiling. The sudden drop
in heat flux at TL often triggers drastic spike in surface temperature causing the notorious “burn-out” in boiling
applications. Thus, designing heat-dissipating substrates with high TL is as practically important as enhancing the
critical heat flux.
We observe considerable enhancement in both qc and TL on nanopillar substrates. In Fig. 2b, we show the
dependences of qc and TL on the surface roughness r. Both qc and TL increase monotonically with increasing r (or
equivalently with l). Compared to the smooth substrate (r = 1), qc and TL on the nanopillar substrate with r = 3.2
increase ≈50% and ≈25 °C, respectively. The enhancement in both qc and TL is a remarkable feature of the nano-
pillar substrates as opposed to those fabricated with larger scale structures. It was observed that micro-structures
only enhance the heat flux without increasing the Leidenfrost temperature; typically, TL decreases with increasing
height of microstructures25. Separate investigations on hierarchical substrates, i.e., microstructured surfaces cov-
ered with nano-grass, reported enhancement in either the Leidenfrost temperature26–28, or the heat flux13,27,29–34.

Mechanism of heat transfer enhancement on nanopillar substrates. To control the effects of

nanopillars on enhancing qc and TL, we now investigate the heat transfer mechanism on nanopillar substrates.
From qualitative observations of the differences in bubble generation between the nanopillar substrates and
the smooth one, we postulate that the major contributing factor leading to the observed change in qc and TL

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 3

www.nature.com/scientificreports/

Figure 3. (a) Schematics showing similarities between the rewetting process in pool boiling (left panel) and the
spreading process (right panel) from a capillary, in particular at the three-phase contact line (highlighted in the
middle panel). (b) Apparent spreading velocity vs of FC-72 on different substrates over a range of temperature
from 20 °C to 46 °C. (c) Critical heat flux qc predicted using Eqn. 2 versus the measured values for smooth
substrate (r = 1) and nanopillar substrates. The uncertainty for the predicted heat flux comes from that of the
spreading velocity measurements. The shaded area indicates a 15% deviation from the measured data.

is the nanopillar-induced rewetting process. We note that in the case of hierarchical structured surfaces, i.e.,
micro-structures covered by nano-grass, enhancement in qc was observed, but due to a very different mechanism:
the enhancement in volume of liquid wicking through the microstructures brings more liquid to the surface and
subsequently leads to an increase in heat flux by latent heat of evaporation. The wicking process, however, is not
possible for nanopillars due to the tremendous viscous stress induced in flows confined in such small length scales
between pillars13. Thus, wicking is excluded as the cause of heat flux enhancement for our nanopillar substrates,
which have interspacing between pillar ≈ 360 nm.
We focus on the advancing contact line of liquid to explore how the nanopillar-induced rewetting process
leads to heat transfer enhancement. Taking advantage of the similarity between the rewetting process of vapour
bubbles and the spreading process of liquid (illustrated in Fig. 3a), the dynamics of the advancing contact line
can be studied separately in the case of spreading liquid on nanopillar substrates15,22. We note that the rewet-
ting process has been the focus of numerous detailed studies13, which revealed contributing factors such as
surface-induced capillary force13, hydrostatic and hydrodynamic forces22, reshaping effects35 and vaporisation36.
For the nanopillar substrates, we determine the dominant effects and their contribution to the heat transfer by
investigating the spreading of liquid from a capillary tip onto heated substrates (see Fig. 3a). By varying the
surface temperature and tracking the three-phase contact-line (TCL) as the liquid spreads, we determine the
temperature dependence of the apparent spreading velocity vs on each nanopillar substrate.
The surface temperature can be varied from room temperature Tr = 20 °C up to 46 °C without causing the
liquid to boil. Although the displacement of the TCL results from various coupled effects, we simplify the analysis
by decomposing vs into several independent components (Fig. 3b):
vs = vca − vn − vh, (1)
where the component vca is caused by capillary force on the substrate and therefore is a function of the surface
roughness r only. The component vh represents the reduction in spreading velocity as liquid absorbs heat from the
surface and evaporates. Thus, if the substrate is not heated, i.e., T = Tr, it follows that vh = 0 because there is no
heat transferred from the surface to the liquid. The velocity component vn represents the reduction in spreading
velocity due to natural evaporation of liquid. In the case of liquid spreading in air, this component is nonzero and
depends on the partial pressure of vapour, whereas in the case of boiling, it vanishes as bubbles generated during
boiling of liquid are filled only with saturated vapour. Here, we have neglected contributions to TCL displacement
from hydrostatic pressure of the liquid column, and from capillary pressure at the upper liquid surface in the tube
due to the small gap between the capillary tube and the substrate.

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 4

www.nature.com/scientificreports/

We postulate that the increase in heat flux on nanopillar substrates results from nanopillar-induced enhance-
ment in wetting. This is suggested from the linear correlation between the critical heat flux of different substrates
and the spreading velocity v s measured on respective substrates at surface temperature T = 20 °C (see
Supplementary Information Fig. S2). In other words, the presence of nanopillars intensifies the rewetting process,
leading to an added amount of liquid brought to and subsequently evaporated from the surface. Among the con-
tributing components to the apparent spreading velocity, vca inherits the effect of r on capillary wetting, thus has
a direct relation to the enhanced rewetting process. In addition, as the surface temperature for spreading liquid
was set lower than the liquid’s boiling temperature, we assume that with constant surface temperature, vn remains
fixed for all the substrates, i.e., vnn = vns , where the superscripts n and s respectively indicate quantities of the nan-
opillar substrates and the smooth one. We recall that vca is independent of T and vh(Tr) = 0. As a result, at room
temperature Tr, the spreading velocity difference δvs(Tr ) = vsn(Tr ) − vss(Tr ) between a nanopillar substrate and the
smooth one only depends on the change in capillary velocity between the two substrates δvca = vcan − vcas. Thus,
δvca can be estimated as δvca ≈ δvs(Tr).
We measure the spreading velocity vs for all nanopillar substrates and the smooth substrate (see Fig. 3b). The
spreading velocity increase δvs(Tr) is calculated for each nanopillar substrate to obtain δvca. The spreading velocity
increases with increasing nanopillar height due to the increasing energetic favourability of wetted nanopillar
surfaces of increasing height. The energy of the different dry nanopillar substrates is approximately the same.
However, the energy of different wetted nanopillar substrates decreases with increasing height due to the larger
interface between the liquid and solid nanopillar array. This energy difference drives the spreading of the FC-72,
and thus, the spreading velocity increases with increasing nanopillar height. We note that for a nanopillar sub-
strate with projected boiling area Ap, the increase in capillary velocity δvca leads to a mass increase of evaporated
liquid ρlAdδvca, where Ad is the time-averaged dry area during boiling on Ap. At CHF, Ad can be estimated using
the classic boiling model by Zuber37: Ad = KAp, with K ≈ 0.1314,37,38. The enhancement in heat transfer at CHF,
i.e., of a nanopillar substrate compared to the smooth substrate, can be calculated as δQc = hfgρlAdδvca. As a result,
we can estimate the heat flux at CHF, denoted as qcn, through each nanopillar substrate using that through the
smooth one qcs and the heat flux enhancement δQc/Ap = Kρlhfgδvca as follows:
qcn = qcs + Kρlh fg δvca. (2)
We note that the heat flux at CHF on smooth substrates depends on the surface temperature and can be
predicted using the classical model for nucleate boiling by Zuber37. A comparative study between the model and
our boiling data on the smooth substrate shows an excellent agreement (see Methods for details). Thus Eqn. 2
provides a direct prediction of the heat flux through nanopillar substrates based on the enhancement in spread-
ing velocity. In Fig. 3c, we show a comparison between the heat flux predicted using Eqn. 2 and one measured
experimentally for all the substrates. The agreement between the predicted CHF and the measured values for
all nanopillar substrates strengthen our hypothesis that the enhanced heat flux is caused by nanopillar-induced
enhancement in rewetting. The results also suggest that the boiling crisis is mainly dictated by the intensity of
the rewetting process, and the increase in critical heat flux is a direct consequence of the faster rewetting velocity
using higher nanopillars.

Prediction of critical heat flux temperature on nanopillar substrates. We now explore how
nanopillars-induced enhancement in rewetting affects the surface temperature Tc at CHF. Recall that the critical
heat flux for nanopillar substrates can be predicted using that on the smooth substrate and the enhancement in
rewetting velocity. We extend this argument to estimate Tc of nanopillar substrates. First, we determine the heat
transfer coefficient hTCL associated with liquid vaporisation at the three-phase contact line (TCL) by considering
the volume Ω of liquid evaporated at TCL per unit time due to heat absorption from the nanopillar substrates. If
we denote θ the contact angle of FC-72, Ω can be calculated as Ω = vhAdtanθ (see Fig. 3a for illustration). On one
hand, the rate of latent heat required to evaporate Ω is ρlΩhfg. On the other hand, the heat rate supplied from the
substrate is hTCL(T − Tl)Ad, where Tl is the liquid temperature. Balancing the rate of latent heat of evaporation and
the heat rate supplied from the substrate gives the energy balance ρlΩhfg = hTCL(T − Tl)Ad. Thus, the heat transfer
coefficient hTCL can be obtained by taking derivative with respect to T of the energy balance, assuming that hTCL
is independent of T: hTCL = ρlhfgtanθdvh/dT. Taking into account the relation between the spreading velocity vs
and vh (Eqn. 1) and recalling that vca is independent of T, we have dvh/dT = dvs/dT, thus arrive at the equation to
determine hTCL for each substrate:
dvs
hTCL = ρlh fg tan θ .
dT (3)
The central argument of our analysis for nanopillar-induced enhancement in heat flux is that the heat transfer
increase takes place at TCL where capillary wetting and evaporation are significantly altered by the presence
of nanopillars. As shown in Fig. 4a, hTCL calculated using Eqn. 3 increases linearly with the surface roughness,
or equivalently, the height of nanopillars. This implies that the enhancement due to both conduction at the
solid-liquid interface and convection on nanopillar substrates are secondary effects. The total heat transfer coef-
ficient hn on nanopillar substrates can then be estimated using the total heat transfer coefficient hs on the smooth
substrate and the enhancement in heat transfer coefficient at TCL:
d(vsn − vss )
hn ≈ hs + K (hTCL
n s
− hTCL ) + C = hs + Kρlh fg tan θ + C,
dT (4)

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 5

www.nature.com/scientificreports/

Figure 4. (a) Heat transfer coefficient at the three-phase contact line hTCL versus surface roughness r. (b)
Critical heat flux temperature Tc predicted using Eqn. 5 versus the measured values for substrates with different
r. The shaded area indicates a 15% deviation from the measured data.

where the prefactor K is needed for the enhancement at TCL and the constant C = 425W ⋅ m−2 ⋅ K−1 accounts for
the additional heat transfer caused by the flow induced by displacement of bubbles from nanopillar substrates.
Since there is no obvious difference in bubble dynamics between nanopillar substrates, we use a fixed constant C
for all nanopillar substrates to represent the heat transfer increase due to displacement of bubbles. In deriving
Eqn. 4, we also assume insignificant variation in h between the spreading and boiling processes for a fixed sub-
strate. We note that Eqn. 4 can be applied at CHF due to the linear dependence of vs on T (see Fig. 3b). The heat
transfer coefficient at CHF on the smooth substrate can be determined using the heat transfer balance
qcs = hs (Tcs − Tb), which gives hs = qcs /(Tcs − Tb). Here Tcs is the wall temperature at CHF on the smooth sub-
strate and Tb is the boiling temperature of the liquid. Similarly, we have qcn = hn(Tcn − Tb), where Tcn is the wall
temperature at CHF on nanopillar substrates. Taking the predicted heat flux qcn and the heat transfer coefficient
hn on nanopillar substrates respectively from Eq. 2 and Eq. 4, Tcn can be predicted as:
qcn
Tcn − Tb = .
hn (5)
In Fig. 4b, we compare the predicted superheat temperatures ΔTc = Tc − Tb and the experimental ones at
CHF for all substrates. Although the predicted values of ΔTc carries the uncertainty of the spreading velocity
measurement on nanopillar substrates, they fall within 15% deviation from the experimentally measured ones,
highlighting a remarkable agreement between the predicted and experimental data.

Conclusions
In summary, we show that nanopillars fabricated on boiling substrates induce a substantial effect on the boiling
behaviours in the nucleate boiling regime. In particular, increasing the height of nanopillars effectively leads to
considerable enhancement in both heat flux and surface temperature at CHF. We attribute such enhancement to
the nanopillar-induced increase in rewetting velocity, which can be measured in a separate wetting experiment.
Based on the observations that the rewetting velocity increases for substrates with higher nanopillars, as well
as the assumption that the enhancement in heat flux dominantly takes place at the three-phase contact line, we
develop a mechanistic model to predict both the heat flux and the temperature at CHF of nanopillar substrates.
Our model takes into account the nanoscales of the pillars, and thus, excludes the wicking motion, or imbibition
of fluid, as plausible mechanisms for heat transfer enhancement. As a result, the enhanced capillary force due to
the presence of nanopillars is the major cause of the intensified rewetting process and subsequent increases in heat
flux and temperature at CHF.
The enhancement in both of these critical quantities in pool boiling is a remarkable feature of nanopillar
substrates in comparison with nano/micro-engineered substrates in which only either one can be increased. By
using only nanopillars with a systematic variation in height and well-defined geometrical dimensions, we have
established a direct link between the enhancement in capillary force and the boiling performance of a substrate.
This provides new insights about design of surface textures not only to amplify the heat flux, but also to achieve
an enhancement in the temperature at CHF, an often less discussed but equally important quantity in preventing
boiling crisis.

Methods
Fabrication of nanopillar substrates. The nanopillar silicon substrates are obtained by inductively cou-
pled plasma reactive ion etching (ICPRIE). The fabrication process consists of three steps, (1) manufacture of the
etching mask, (2) etching the underlying silicon by ICPRIE, and (3) removal of etching mask. Polystyrene (PS)
nanospheres are utilised to generate the etching mask. First, a monolayer of nanospheres with the diameter of
800 nm are self-assembled in the air-water interface39–41. Subsequently the monolayer is transferred to a P-type

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 6

www.nature.com/scientificreports/

Figure 5. Comparison between the boiling curve measured on silicon smooth surface in the present study and
the one obtained by Ujereh et al.42. The solid line represents the heat flux predicted by Rohsenow model, and the
dashed line represent the critical heat flux predicted by Zuber model37.

boron-doped (100) silicon substrate, which has been cleaned with acetone, methanol, isopropanol and then dried
with nitrogen gas. After the substrate is dried in air at room temperature, reactive ion etching (RIE) with oxygen
is applied to reduce the diameter of nanospheres to the required size (≈440 nm).
After the etching mask is manufactured, ICPRIE is used to etch the silicon substrate to fabricate silicon nano-
pillars with desired dimensions. During the process, SF6 and C4F8 with optimal ratio of 33:82 are supplied. These
gases respectively serve as the etching gas and passivation gas. Different height nanopillars are obtained by vary-
ing the duration of this step. It should be noted that when the time duration is longer than 5 minutes the PS mask
is impacted and the nanopillar base diameter is different from the etching mask. When the etching is completed,
the PS mask is removed by ultrasonication in acetone for 5 minutes.

Boiling experiment. The experimental setup used to measure the heat flux through different substrates con-
sists of three sections: heating, testing and condensation. In the heating section, a cartridge heater embedded in
a cylindrical copper rod is used to generate heat. The copper rod is insulated from the surrounding environment
by a Teflon holder. In the testing section, a boiling substrate, which is either smooth silicon wafer or a nanopillar
substrate, is placed between the copper rod and a glass cuvette of lateral size 10 mm × 10 mm. The test substrate
is cleaned by ethanol and acetone in a ultrasonic bath and rinsed with distilled water before each experiment. The
cuvette is filled with degassed FC-72 as a working liquid. The cuvette is enclosed by a stainless-steel container; the
gap between the cuvette and the container is circulated with hot water of temperature 55.5 ± 0.5 °C to keep the
temperature of the working liquid close to boiling point. The container has two glass windows on two opposite
sides allowing optical recording of the boiling phenomena. A condenser is placed on top of the container to col-
lect vapour generated from the working liquid. The cooling power of the condenser is adjusted to keep the vapour
pressure inside the cuvette at atmospheric pressure.
The vapour pressure is measured by pressure transducer (Gefran) and maintained at atmospheric pressure
by adjusting the cooling power of a Peltier module. The temperature is measured by K-type thermocouples and
the boiling process is recorded by a high-speed camera (SA-5, Photron). Three thermocouples are distributed
along the asymmetrical axis of the copper cylinder. The temperatures measured by these thermocouples are used
to calculate the heat flux q through the test substrate and the surface temperature T. The heat flux in the vertical
direction is approximated with the assumption that the rate of heat loss to the side is constant43,44. Thus the heat
flux through the substrate is estimated as:

q = 0.5kc (4T2 − T1 − 3T3)Δxc−1. (6)

where kc is the thermal conductivity of copper, T1, T2, T3 are obtained from the thermocouples in the copper rod,
and Δxc = 7 mm is the distance between the thermocouples. There are three layers between the top thermocouple
and the top surface: a copper layer (7 mm), a thermal glue layer (≈10 μm) and the silicon substrate (500 μm). The
surface temperature T of the test substrate is then estimated by the one-dimensional heat conduction equation:

T = T3 − q(Δxckc−1 + Δxg kg−1 + Δxsks−1), (7)

where Δxc/kc, Δxg/kg and Δxs/ks are the heat resistance of copper, thermal glue, and silicon substrate respectively.
For each tested substrate, the heat flux q and the corresponding superheat ΔT = T − Tb are measured when
the system is in the steady state and with temperature steps of 5 K. The experimental uncertainty is obtained by
repeating the experiment three times.

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 7

www.nature.com/scientificreports/

Experimental Validation. The measurement procedure and analysis are validated by comparing the heat
transfer data obtained on the smooth substrate to the well-known dataset obtained by Ujereh et al.42, as well as to
classical model, including the one for nucleate boiling heat transfer by Rohsenow45 and the one for critical heat
flux by Zuber37 (see Fig. 5). In particular, the critical heat flux on a smooth substrate is:

qcs = Kρv1/2h fg [gσ(ρl − ρv )]1/4 , (8)

where K ≈ 0.131 is an empirical parameter.

References
1. Dhir, V. Boiling heat transfer. Annual review of fluid mechanics 30, 365–401 (1998).
2. Karwa, N., Gambaryan-Roisman, T., Stephan, P. & Tropea, C. A hydrodynamic model for subcooled liquid jet impingement at the
leidenfrost condition. International Journal of Thermal Sciences 50, 993–1000 (2011).
3. Xu, X. & Qian, T. Single-bubble dynamics in pool boiling of one-component fluids. Physical Review E 89, 063002 (2014).
4. Carey, V. P. Liquid-vapor phase-change phenomena (CRC Press, 2007).
5. Theofanous, T., Tu, J., Dinh, A. & Dinh, T.-N. The boiling crisis phenomenon: Part i: nucleation and nucleate boiling heat transfer.
Experimental Thermal and Fluid Science 26, 775–792 (2002).
6. Kim, J. Review of nucleate pool boiling bubble heat transfer mechanisms. International Journal of Multiphase Flow 35, 1067–1076
(2009).
7. Kim, D. E., Yu, D. I., Jerng, D. W., Kim, M. H. & Ahn, H. S. Review of boiling heat transfer enhancement on micro/nanostructured
surfaces. Experimental Thermal and Fluid Science 66, 173–196 (2015).
8. Betz, A. R. et al. Boiling heat transfer on superhydrophilic, superhydrophobic, and superbiphilic surfaces. International Journal of
Heat and Mass Transfer 57, 733–741 (2013).
9. Dhir, V. K. Mechanistic prediction of nucleate boiling heat transfer–achievable or a hopeless task? Journal of Heat Transfer 128, 1–12
(2006).
10. Laurila, T., Carlson, A., Do-Quang, M., Ala-Nissila, T. & Amberg, G. Thermohydrodynamics of boiling in a van der waals fluid.
Physical Review E 85, 026320 (2012).
11. Liu, J., Do-Quang, M. & Amberg, G. Thermohydrodynamics of boiling in binary compressible fluids. Physical Review E 92, 043017
(2015).
12. Kamatchi, R. & Venkatachalapathy, S. Parametric study of pool boiling heat transfer with nanofluids for the enhancement of critical
heat flux: a review. International Journal of Thermal Sciences 87, 228–240 (2015).
13. Dhillon, N. S., Buongiorno, J. & Varanasi, K. K. Critical heat flux maxima during boiling crisis on textured surfaces. Nature
communications 6 (2015).
14. Zupančič, M., Steinbücher, M., Gregorčič, P. & Golobič, I. Enhanced pool-boiling heat transfer on laser-made hydrophobic/
superhydrophilic polydimethylsiloxane-silica patterned surfaces. Applied Thermal Engineering 91, 288–297 (2015).
15. Rahman, M. M., Pollack, J. & McCarthy, M. Increasing boiling heat transfer using low conductivity materials. Scientific reports 5
(2015).
16. Xu, P., Li, Q. & Xuan, Y. Enhanced boiling heat transfer on composite porous surface. International Journal of Heat and Mass Transfer
80, 107–114 (2015).
17. Tang, B., Zhou, R., Lu, L. & Zhou, G. Augmented boiling heat transfer on a copper nanoporous surface and the stability of nano-
porosity in a hydrothermal environment. International Journal of Heat and Mass Transfer 90, 979–985 (2015).
18. Zhong, D., Meng, J., Li, Z. & Guo, Z. Experimental study of saturated pool boiling from downward facing surfaces with artificial
cavities. Experimental Thermal and Fluid Science 68, 442–451 (2015).
19. Zhong, D., Meng, J., Li, Z. & Guo, Z. Critical heat flux for downward-facing saturated pool boiling on pin fin surfaces. International
Journal of Heat and Mass Transfer 87, 201–211 (2015).
20. Li, C. H. & Rioux, R. P. Independent and collective roles of surface structures at different length scales on pool boiling heat transfer.
Scientific reports 6 (2016).
21. Rahman, M. M., Olceroglu, E. & McCarthy, M. Role of wickability on the critical heat flux of structured superhydrophilic surfaces.
Langmuir 30, 11225–11234 (2014).
22. Kim, B. S., Choi, G., Shin, S., Gemming, T. & Cho, H. H. Nano-inspired fluidic interactivity for boiling heat transfer: impact and
criteria. Scientific reports 6, 34348 (2016).
23. Zupančič, M., Može, M., Gregorčič, P., Sitar, A. & Golobič, I. Evaluation of enhanced nucleate boiling performance through wall-
temperature distributions on pdms-silica coated and non-coated laser textured stainless steel surfaces. International Journal of Heat
and Mass Transfer 111, 419–428 (2017).
24. Bejan, A. & Kraus, A. D. Heat transfer handbook, vol. 1 (John Wiley & Sons, 2003).
25. Nair, H. et al. The leidenfrost temperature increase for impacting droplets on carbon-nanofiber surfaces. Soft matter 10, 2102–2109
(2014).
26. Kwon, H.-m, Bird, J. C. & Varanasi, K. K. Increasing leidenfrost point using micro-nano hierarchical surface structures. Applied
Physics Letters 103, 201601 (2013).
27. Ahn, H. S. et al. Pool boiling experiments on multiwalled carbon nanotube (mwcnt) forests. Journal of Heat Transfer 128, 1335–1342
(2006).
28. Sathyamurthi, V., Ahn, H., Banerjee, D. & Lau, S. Subcooled pool boiling experiments on horizontal heaters coated with carbon
nanotubes. Journal of Heat Transfer 131, 071501 (2009).
29. Vemuri, S. & Kim, K. J. Pool boiling of saturated fc-72 on nano-porous surface. International communications in heat and mass
transfer 32, 27–31 (2005).
30. Wu, W. et al. Nucleate boiling heat transfer enhancement for water and fc-72 on titanium oxide and silicon oxide surfaces.
International Journal of Heat and Mass Transfer 53, 1773–1777 (2010).
31. Chen, R. et al. Nanowires for enhanced boiling heat transfer. Nano letters 9, 548–553 (2009).
32. Launay, S., Fedorov, A., Joshi, Y., Cao, A. & Ajayan, P. Hybrid micro-nano structured thermal interfaces for pool boiling heat transfer
enhancement. Microelectronics Journal 37, 1158–1164 (2006).
33. Li, C. et al. Nanostructured copper interfaces for enhanced boiling. small 4, 1084–1088 (2008).
34. Im, Y., Joshi, Y., Dietz, C. & Lee, S. S. Enhanced boiling of a dielectric liquid on copper nanowire surfaces. Int. J. Micro-Nano Scale
Transp 1, 79–95 (2010).
35. Cooper, M. & Chandratilleke, T. Growth of diffusion-controlled vapour bubbles at a wall in a known temperature gradient.
International Journal of Heat and Mass Transfer 24, 1475–1492 (1981).
36. Zou, A., Singh, D. P. & Maroo, S. C. Early evaporation of microlayer for boiling heat transfer enhancement. Langmuir 32,
10808–10814 (2016).
37. Zuber, N. Hydrodynamic aspects of boiling heat transfer (thesis). Tech. Rep., California. Univ., Los Angeles; and Ramo-Wooldridge
Corp., Los Angeles (1959).

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 8

www.nature.com/scientificreports/

38. Linehard, J. & Dhir, V. K. Extended hydrodynamic theory of the peak and minimum pool boiling heat fluxes. Tech. Rep (1973).
39. Ye, X. & Qi, L. Two-dimensionally patterned nanostructures based on monolayer colloidal crystals: controllable fabrication,
assembly, and applications. Nano Today 6, 608–631 (2011).
40. Yu, J. et al. Preparation of high-quality colloidal mask for nanosphere lithography by a combination of air/water interface self-
assembly and solvent vapor annealing. Langmuir 28, 12681–12689 (2012).
41. Moon, G. D. et al. Assembled monolayers of hydrophilic particles on water surfaces. ACS Nano 5, 8600–8612 (2011).
42. Ujereh, S., Fisher, T. & Mudawar, I. Effects of carbon nanotube arrays on nucleate pool boiling. International Journal of Heat and
Mass Transfer 50, 4023–4038 (2007).
43. Cooke, D. & Kandlikar, S. G. Pool boiling heat transfer and bubble dynamics over plain and enhanced microchannels. Journal of
Heat Transfer 133, 052902 (2011).
44. Yao, Z., Lu, Y.-W. & Kandlikar, S. Effects of nanowire height on pool boiling performance of water on silicon chips. International
Journal of Thermal Sciences 50, 2084–2090 (2011).
45. Rohsenow, W. M. A method of correlating heat transfer data for surface boiling of liquids. Tech. Rep., Cambridge, Mass.: MIT
Division of Industrial Cooporation (1951).

Acknowledgements
This work was supported by the Nanyang Technological University (NTU), A*STAR, Singapore, University of
Pittsburgh, and NSF grant #1552712. D. Liu and N. Rashin acknowledge the research fellowship supported by
A*STAR. T.B. Nguyen acknowledge the PhD scholarship from NTU.

Author Contributions
T.T. and P.W.L. conceived the experiment, T.B.N., D.L., M.I.K., B.W., and N.R. conducted the experiment, T.B.N.,
D.L., and T.T. analysed the results. All authors reviewed the manuscript.

Additional Information
Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-22693-z.
Competing Interests: The authors declare no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© The Author(s) 2018

Scientific REPOrts | (2018) 8:4815 | DOI:10.1038/s41598-018-22693-z 9