Page 1
TSpace Research Repository tspace.library.utoronto.ca
Condensation in One-Dimensional Dead-End Nanochannels
Junjie Zhong, Seyed Hadi Zandavi, Huawei Li, Bo Bao, Aaron H.
Persad, Farshid Mostowfi, David Sinton
Version post-print
Citation (published version)
Zhong, J., Zandavi, S. H., Li, H., Bao, B., Persad, A. H., Mostowfi, F., & Sinton, D. (2017). Condensation in one-dimensional dead-end nanochannels. ACS Nano, 11(1), 304–313. https://doi.org/10.1021/acsnano.6b05666
Copyright / License © American Chemical Society
Publisher’s Statement This document is the Accepted Manuscript version of a Published Work that appeared in final form in ACS Nano, copyright © American Chemical Society after peer review and technical editing by the publisher. To access the final edited and published work see doi.org/10.1021/acsnano.6b05666
How to cite TSpace items
Always cite the published version, so the author(s) will receive recognition through services that track
citation counts, e.g. Scopus. If you need to cite the page number of the author manuscript from TSpace because you cannot access the published version, then cite the TSpace version in addition to the published
version using the permanent URI (handle) found on the record page.
This article was made openly accessible by U of T Faculty. Please tell us how this access benefits you. Your story matters.
Page 2
1
Condensation in One-Dimensional Dead-End Nanochannels
Junjie Zhong, †, ¶ Seyed Hadi Zandavi, †, ¶ Huawei Li, † Bo Bao, † Aaron Harrinarine Persad, † Farshid Mostowfi, ‡
David Sinton†,*
† Department of Mechanical and Industrial Engineering, University of Toronto, M5S 3G8 Canada
‡ Schlumberger-Doll Research, Cambridge, Massachusetts, 02139 USA
* corresponding author: [email protected]
¶ These authors contribute equally
Abstract
Phase change at the nanoscale is at the heart of many biological and geological phenomena. The
recent emergence and global implications of unconventional oil and gas production from
nanoporous shale further necessitate a higher understanding of phase behavior at these scales.
Here, we directly observe condensation and condensate growth of a light hydrocarbon (propane)
in discrete sub-100 nm (~70 nm) channels. Two different condensation mechanisms at this
nanoscale are distinguished - continuous growth, and discontinuous growth due to liquid
bridging ahead of the meniscus - both leading to similar net growth rates. The growth rates
agree well with those predicted by a suitably defined thermo-fluid resistance model. In contrast
to phase change at larger scales (~220 and ~1000 nm cases), the rate of liquid condensate
growth in channels of sub-100 nm size is found to be limited mainly by vapor flow resistance
(~70% of the total resistance here), with interface resistance making up the difference. The
condensation-induced vapor flow is in the transitional flow regime (Knudsen flow accounting
for up to 13% of total resistance here). Collectively, these results demonstrate that with
confinement at sub-100 nm scales, such as is commonly found in porous shale and other
applications, condensation conditions deviate from the microscale and larger bulk conditions
chiefly due to vapor flow and interface resistances.
TOC Graphic
KEYWORDS: condensation, condensate growth, nanochannel, nanofluidics, hydrocarbon, flow
resistance, Knudsen flow
Phase change and fluid flow at the nanoscale play important roles in the environment,1-3 water
desalination,4 microelectronics,5, 6 targeted drug delivery7 and unconventional oil and gas
recovery.8-10 The latter has resulted in a large shift in global energy as hydrocarbons in shale and
tight oil formations are being produced from nano-sized pores.9, 11 To understand these systems,
it is crucial to understand phase change of nanoconfined reservoir fluids, particularly as reservoir
pressurization and depressurization alter the phase state.12 Despite the emergence and global
importance of unconventional oil and gas recovery, the fundamental physics governing phase
Page 3
2
change at the nanoscale remains poorly understood, and the available classical theories and
simulation techniques have proven to be inadequate.11
As system size decreases, the classical approach for studying phase change breaks down.13 For
the condensation process specifically, the condensate growth will drive vapor flow at the
nanoscale. When the characteristic length of the system (h) becomes close to the mean free path
of the vapor molecule (l), the flow behavior can deviate from the bulk behavior predicted by the
Navier-Stokes equations,14 and affect the condensate growth process. Furthermore, at very small
scales the saturation pressure can change as a result of liquid-gas curvature leading to capillary
condensation15.
Many investigations of nanoscale condensation were carried out by gravimetric measurements of
adsorption,16 in fixed bed geometries,17 by interferometry,15, 18 through surface force
apparatuses,19, 20 or using environmental scanning electron microscopy (ESEM).21-23 Recent
advances in nanotechnology allow the investigation of phase change in deterministic structures
(e.g., nanochannels24) and direct observation of the whole process. Previous phase change studies
in nanochannels have examined liquid-to-vapor phase transition through evaporation-induced
cavitation,25 initiated by localized heating26 or capillary evaporation.27 The condensation process,
on the other hand, is physically distinct as a vapor-to-liquid phase transition, and there is a
profound need for targeted experiments and theory to better understand this phenomenon and
inform analytical methods.11, 13 Optically accessible nanofabricated channels can provide a much
needed window into the relevant phase phenomena at these scales and has been used, for
example, to observe the capillary condensation of water.28
Here, we study condensation and condensate growth of a pure substance in sub-100 nm
confinement using propane as a model fluid. Two different condensation mechanisms at the
nanoscale are distinguished and compared with previous simulation results:27, 29 continuous
growth, and discontinuous growth due to liquid bridging ahead of the meniscus. The directly
observed effects of pressure and temperature on the condensate growth and growth rate are
compared with those of larger channels (220 nm and 1000 nm) and those predicted by a
developed thermo-fluid resistance model. In contrast to phase change at larger scales, where heat
and mass transfer processes at the liquid-vapor interface dominate the condensation and
condensate growth process,30, 31 the rate of liquid condensate growth is found to be limited
mainly by vapor flow resistance at this scale. The findings in this paper demonstrate that
condensation and condensate growth dynamics in sub-100 nm dead-end nanochannels deviate
from bulk conditions. The findings will be beneficial in understanding many natural processes
and in optimizing related applications.
Results and Discussion
To study condensation in dead-end nanochannels, we fabricated chips by etching them (high-
reactive-ionic-etching, DRIE) on a silicon wafer and anodically bonding with glass (see Methods
section for details on the fabrication process). A schematic of the experimental setup is shown in
Figure 1A. As shown, the test region consisted of 10 identical dead-end nanochannels in one
group (each channel is 35 mm × 5.5 μm). Channels of this width are large enough to visualize
using optical microscopy, and small enough to avoid channel collapse during bonding. The
Page 4
3
length in-plane (35 mm) is short enough for wafer fabrication, and long enough to ensure locally
uniform temperature within the observation region. SEM images of the cross-section of a 70 nm
high channel are shown in Figure 1B and AFM images of a channel profile are shown in Figure
1C.
Figure 1. Diagram of the experimental setup and SEM/AFM characterizations of 70 nm high channels. (A) Schematic of
the experimental setup. (B) SEM images of the cross-section of one nanochannel. (C) Typical AFM results of a series of
nanochannels. Each channel has a width of 5.5 μm and a nominal height of 70 nm (actually 68.0 ± 1.1 nm). The left image
shows the height profile of a single channel; the right image is a topological scan of 3 adjacent channels.
During condensation experiments, the pressure and temperature were closely controlled. The
fluid pressure at the inlet of the nanochannel was assumed to be the same as the pressure of the
fluid reservoir (piston) since the pressure drop in the mainchannel was negligible compared to
the drop in the nanochannels (detailed calculation is provided in the Supporting Information,
section 5). The reservoir fluid pressure was maintained within ±0.5% of the desired value. To
promote condensation near the dead-end (observation region), a temperature difference across
the chip was induced by pairing a chiller and heater clamped on top of the chip (Figure 1A). The
chiller was positioned over the dead-end region and was separated by a 1-mm thick insulation
layer from the heater at the inlet region. The chiller's temperature was adjustable, within ± 0.3 K,
and the heater's temperature was kept fixed at 473 K (±1 K).
In each condensation experiment, the pressure at the inlet of the nanochannel was initially set far
below the saturation pressure at the temperature in the chiller region. After the temperature of the
system reached steady state, the pressure was increased to the target pressure (response within 10
seconds) and was kept in steady-state operation for 30 minutes to detect condensation (see
Methods section for details on experimental procedures). Vapor condensation and condensate
growth were visualized using an inverted microscope in the bright-field mode. The vapor and
liquid phases were clearly distinguished by light intensity: the vapor phase appeared bright while
the liquid phase appeared darker, as shown in Figure 2A and Figure 2B. It is noteworthy that in
microscale or larger channels the opposite is generally observed, that is, the vapor phase is
darker compared to the liquid phase due to refraction.32, 33 However, in the subwavelength silicon
nanochannel, the vapor-Si interface provides a larger refractive index contrast and thus reflects
more incident light, appearing brighter than the liquid-Si case, as observed elsewhere.25, 34 The
Page 5
4
liquid-vapor interface position and the condensate length were extracted directly from image
sequences using an image processing algorithm. Both temperature and pressure were varied in
the experiments: temperature ranged from 314.0 K to 347.9 K in steps of ~5 K (the details of
temperature measurements are shown in the Supporting Information, section 3), and pressure
ranged from 1.4 MPa to 3.4 MPa in steps of 0.2 MPa (reading from the pressure transducer at the
inlet of the mainchannel). All the experimental conditions are plotted in a P-T diagram in the
Supporting Information, section 2.
Condensation in Nanochannels
Figure 2. Direct observation of two types of condensation (Type A and Type B) in a 70 nm high channel at (A) 2.8
MPa/343.7 K and (B) 2.8 MPa/314.0 K. The time interval between each frame in (A) and (B) is 0.2 s (0.02 s for the zoom-
in image in (B)). (C) The number of channels with Type A and Type B condensation in a group of 10 nanochannels and
three repeats as a function of supercooling at 2.8 MPa.
Nanochannels within the chiller region of the chip were monitored at each pressure step for
30 minutes to detect condensation. Image sequences of the vapor condensation and condensate
growth in a nanochannel versus time (with a sampling speed of 50 frames per second) are shown
in Figure 2. The condensed liquid phase generally advanced from left to right (i.e., from the
dead- end) and the corresponding vapor flowed from right to left (i.e., from the inlet). Two
different condensation and condensate growth mechanisms were observed, depending on
conditions. Figure 2A shows a typical Type A condensation where the condensation started at (or
close to) the dead-end, and the single liquid column grew continuously towards the inlet of the
nanochannel as a result of condensation, i.e., in Type A condensation only one liquid segment
was observed. In Type B condensation, the initiation of condensation was also close to the dead-
end point. However, the progression of the liquid-vapor interface was not continuous. Rather, it
consisted of a repeating pattern of growth followed by liquid bridging ahead of the liquid column
front. An example of this pattern is shown in Figure 2B. As seen in the inset of Figure 2B, a
darkening in the channel was observed prior to the formation of a liquid bridge. Once the liquid
bridge formed, the original interface stopped growing, but the liquid bridge continued to grow in
Page 6
5
both directions and, eventually, the vapor bubble between the two liquid-vapor interfaces
disappeared. The supplemental video (Supporting Information, section 8) shows examples of
Type A and Type B condensation in the nanochannels.
The number of nanochannels with each type of condensate growth depended on the degree of
supercooling, ∆𝑇 = 𝑇𝑆 − 𝑇𝐶, where 𝑇𝑆 denotes the saturation temperature under bulk conditions
at the inlet pressure, and 𝑇𝐶 denotes the temperature of the channels in the chiller region. Figure
2C shows the distribution of nanochannels in a group of 10 nanochannels with Type A and Type
B condensation as a function of supercooling. The number of nanochannels with Type B
condensation increased as the supercooling increased.
Additional tests were performed with larger channels heights (220 nm and 1000 nm, see AFM
results of channel heights in Supporting Information, section 17), with full details provided in the
Supporting Information, section 15. In these cases multiple liquid bridges ahead of the liquid
column formed, distinct from the 70 nm case where a maximum of one liquid bridge per channel
was observed.
Figure 3. Schematic showing two different types of condensation and condensate growth in the axial direction of a dead-
end nanochannel, in the height (nanoscale) dimension. (A) Initiation of the liquid column and film formation on the glass
and silicon surfaces. (B) The growth of the liquid column and growth of the adsorbed/condensed layers in front of the
liquid-vapor interface. (C) Type A condensation, where the liquid column grows continuously. (D) Type B condensation,
where the liquid bridge happens ahead of the liquid column. It is noteworthy that the cross sectional area is assumed
rectangular to illustrate the axial movement of the meniscus. Note that the actual cross sectional area of the channel is
shown in Figure 1B.
Figure 3 shows a schematic of the expected behavior in the height (nanoscale) dimension in
Type A (Figure 2A) and Type B (Figure 2B) condensation. Initially, the vapor will adsorb and
grow on the smooth walls of the channel (Figure 3A), forming liquid column when the two
adsorbed layers interact and touch. In our experiments, it was commonly observed that the liquid
column initiated at or near the dead-end, as shown in Figure 2A and 2B. These observations are
in agreement with the simulation results reported in the literature in which condensation also
started from the dead-end point in nanoslit or cylindrical pores.27, 36, 37 Subsequent to the liquid
column formation (Figure 3A), the liquid-vapor interface moves with a relatively constant
velocity of 𝑉𝑥1 toward the channel entrance due to the accumulation of propane molecules at the
liquid-vapor interface (as seen in Figure 2A and Figure 3B). Simultaneously, the growth of liquid
Page 7
6
films on the top and bottom channel walls is expected (𝑉𝑦1, Figure 3B). When the supercooling is
relatively low (i.e., temperature in the chiller region is relatively high or pressure at the inlet of
the nanochannel is relatively low), the growth rate of the adsorbed/condensed layers (𝑉𝑦1) in front
of the liquid column is slow. In this circumstance (i.e. Type A condensation), the liquid column
grows continuously towards the inlet of the nanochannel, as indicated in Figure 3C. As the
supercooling is increased (temperature is decreased or pressure is increased), there is a higher
rate of adsorption on all nanochannel surfaces. Type B condensation results if the adsorption rate
causes the films on the silica and silicon surfaces to coalesce prior to the arrival of the liquid
column (see Figure 3D). The growth of the adsorbing layer leading up to the formation of the
liquid bridge was captured in the images shown in Figure 2B. In the post-processed images,
visible darkening appears prior to the formation of the liquid bridge for ~0.1 s, attributed here to
the growth of liquid films on the order of tens of nanometers thick. The formation of the liquid
bridge entrap a bubble during the condensate growth process (Figure 3D). This bubble collapses
asymmetrically as the vapor phase is replaced by the liquid phase (Figure 2B). The asymmetry is
a product of the density difference between liquid and vapor phase which necessitates bulk
motion from right to left (𝑉𝑚 in Figure 3D).
Liquid bridging mechanism
The formation of the liquid bridge during capillary condensation has also been predicted through
simulations.27, 29 Liquid bridging has been predicted in large and long nanopores only at high
temperatures (i.e., less supercooling).27 This behavior was attributed to the periodic undulations
of the interface of the adsorbed/condensed film. In contrast, we directly observe liquid bridging
more frequently at more supercooled conditions (Figure 2C), as would be expected for higher
condensation rates. In another recent simulation study of adsorption and condensation in
nanopores,29 it was found that after the detachment of the meniscus at the capped end, the
formation of a liquid bridge occurred close to the pore inlet. The authors concluded that the
probability of bridge formation was higher near the pore opening, particularly for longer
channels. It was found that after the detachment of the meniscus at the capped end, the formation
of a liquid bridge occurred close to the pore inlet.
In contrast to previous simulations, we observe that a liquid bridge intermittently formed close to
the liquid-vapor interface of the liquid column (~100 μm from the interface) in the dead-end
region within our observation window (~1.5 mm x 1.5 mm). There are several potential factors
that could influence the location and frequency of liquid bridge formation in our experiments: (i)
a gradient in nanochannel height or roughness, (ii) gradients of temperature and pressure along
the nanochannel, (iii) the local thermal undulation of adsorbed/condensed liquid films and (iv)
corner flow. Each of these factors is discussed in detail below.
Firstly, with respect to the gradient of channel height, we measured different locations along the
nanochannel (~10 mm as the interval) by AFM. Results showed there was no unidirectional
height gradient along the nanochannel. Moreover, the surface roughness was small (~1 nm) and
equally distributed (see Figure 1C). Therefore, the channel height variation is not expected to
play a significant role in the liquid bridging location observed here.
Page 8
7
Secondly, regarding the variation in temperature and pressure along the nanochannel, several
aspects are noteworthy. The latent heat generated by condensation can locally disturb the
temperature field by up to 0.004 K (see Supporting Information, section 9), but this effect is
negligible compared to other sources of uncertainty. Another potential source of temperature
variation comes from the experimental setup: the maximum gradient in the chiller region was 0.6
K over the 1 mm field of view (at 314.0 K) (see Supporting Information, section 3), and thus
negligible (only 0.06 K) over the typical distance between the interface and the observed liquid
bridges. Regarding the pressure, the maximum pressure gradient calculated in the chiller region
was 0.025 MPa over the 1-mm field of view (at 3.4 MPa, see Supporting Information, section
10), and thus ~0.0025 MPa over the typical distance between the interface and the observed
liquid bridges. Both the temperature and pressure gradients described above are too small to
cause significant effects on the thickness of the adsorbed/condensed layer. In addition, the effects
of temperature and pressure gradients on the thickness of the adsorbed/condensed layer act in the
opposite direction. Specifically, the effect of the pressure gradient dominates that of the
temperature (~0.07% compared to ~0.02%), and the former would cause liquid bridge formation
in the opposite direction (i.e., close to the inlet) to that observed in our experiments. Therefore,
gradients of temperature and pressure are not responsible for the nature and location of the liquid
bridging observed here.
Thirdly, regarding the thermal undulations of liquid films, previous experimental results in
microconfinement have shown liquid bridge formation as a result of unstable surface waves.38-40
Similarly, in our experiments the formation of the liquid bridge close to the liquid-vapor
interface could also be attributed, in part, to the local undulation of the adsorbed/condensed film.
Lastly, corner flow may also influence the liquid bridging. Liquid tends to flow along the corners
to minimize the interfacial energy,41 reaching out from the liquid column in thin films along the
edges.42 The effect is stronger for thinner channels with effectively sharper corners (70 nm
geometry in Figure 1).43 In our experiments, we found the cross-channel liquid bridging occurs
less frequently in these smaller channels (70 nm) and more frequently in larger channels (220
and 1000 nm), for the same experimental conditions (experiments shown in Supporting
Information, section 15). We speculate that corner flow would tend to promote the continuous
condensate growth over liquid bridging, since vapor molecules would condense more readily
onto the liquid films (condensation coefficient of unity), rather than stick onto the silicon
substrate (sticking coefficient of 2×10-5)44 to form a new liquid bridge.
In summary, considering the four potential mechanisms outlined above, we expect a combination
of, or competition between, surface film undulation and corner flow contribute to the distinct
liquid bridge phenomena observed here in sub-100 nm channels. The exact mechanisms
governing liquid bridging in nanochannels, however, remains an open question.
Effects of pressure and temperature on condensate growth
The total condensate length versus time in the 70 nm high channel at 326.7 K and five different
pressures is shown in Figure 4A. For Type B growth cases, the total condensate length was
calculated as the sum of the liquid lengths in the nanochannel. Linear relations between the
condensate length and time were found for all conditions within our observation window. This
Page 9
8
result illustrates that, regardless of the type of condensation progression (Type A or B), the net
rate of condensate accumulation was roughly constant in time (again, only within our
observation window). Specifically, the maximum deviation of condensate growth rate between
Type A and Type B condensation at the same thermodynamic condition was under 13% for all
cases, and growth rates are plotted vs. pressure in Figure 4B. Figure 4C shows the total
condensate length versus time in the 70 nm high channel at 2.8 MPa for five different
temperatures. Similarly, linear relationships between the condensate lengths versus time were
found for all temperatures within the observation window. Figure 4D shows the isobaric
condensate growth rate at different temperatures. The condensate growth rates in Figure 4B and
4D are calculated by fitting a linear regression line to the condensate length of Figure 4A and 4C
(the total measured condensate length versus time at eight different isobaric and isothermal
conditions in the 70 nm high channels are provided in the Supporting Information, section 4).
Figure 4. Condensate length and condensate growth rate as a function of pressure and temperature in 70 nm high
channel. (A) Condensate length versus time at 326.7 K (𝑻𝑪) and different pressures. Each measurement point is a result of
the average of 3 repeated experiments in the 10 nanochannels. Representative error bars are shown for 8 of the
measurement points at most (error bars for remaining data points are omitted for clarity). The time interval for each of
the two data points plotted is 0.5 s. (B) Condensate growth rate as a function of pressure calculated from the condensate
length in (A). (C) Condensate length versus time at 2.8 MPa (𝑷𝑹) and different temperatures (time interval of 0.5 s
between data points). (D) Condensate growth rate as a function of temperature calculated from the condensate length in
(C).
Accumulation of vapor molecules at the liquid-vapor interface during the condensate growth
process necessitates vapor flow through the nanochannel. To analyze the condensate growth in
the nanochannel, we employed a resistance model (see Figure 5). During the condensation
process, we consider three major resistances in series between the nanochannel entrance and the
condensed liquid phase: (i) flow resistance in the heater region of the nanochannel (RH), (ii) flow
resistance in the chiller region of the nanochannel (RC), and (iii) the resistance to crossing the
liquid-vapor interface (RI).
Page 10
9
For flow resistance in the nanochannels and nanopores of this size, the mean free path of the
vapor molecules (l) becomes comparable to the dominant scale of the system (h in our cases). In
our experiments, the Knudsen number (Kn = l/h) for our 70 nm high channels was between 0.03
and 0.053 (see Supporting Information, section 11), which is in the transitional flow regime
(0.01 < 𝐾𝑛 < 1).48 In this regime, the Knudsen flow (i.e., net flow due to diffusion interactions)
must be considered in addition to the viscous flow.
Figure 5. Flow resistance model for condensate growth in the dead-end nanochannel.
The vapor flow resistance in the heater and chiller regions of the nanochannel are the
combination of two parallel resistances48: the resistance induced by the viscous flow (RV) and the
resistance induced by the Knudsen flow (RK):
𝑅𝛽 =𝑅𝑉𝛽𝑅𝐾𝛽
𝑅𝑉𝛽+𝑅𝐾𝛽=
1
ℎ2
12𝜇𝛽𝑔
𝐿𝛽+
2√2ℎ
3𝜌𝛽𝑔
𝐿𝛽√
𝑀
𝜋𝑅𝑇𝛽
(1)
where 𝜇 𝑔 and 𝜌𝑔 are the viscosity and density of the vapor phase, respectively, ℎ is the height of
the nanochannel, L is the channel length in the region, T is the temperature inside the channel, M
is the molar mass of the fluid, and the additional subscript 𝛽 specifies to the heater (H) or the
chiller (C) regions.
For the liquid-vapor interface resistance (𝑅𝐼), the Hertz-Knudsen equation is used to express the
condensation mass flux (𝐽𝐼 ) in terms of the difference between the vapor pressure and the
equilibrium pressure:49
𝐽𝐼 = 𝑁𝛼√𝑀
2𝜋𝑅(
𝑃𝑉
√𝑇𝑉−
𝑃𝐸𝑞
√𝑇𝑙𝑣) (2)
where Nα is an empirical constant discussed in more detail below, 𝑅 is the universal gas constant,
𝑇𝑉 and 𝑇𝑙𝑣 are the temperatures in the vapor phase and at the liquid-vapor interface, respectively
(𝑇𝑉 = 𝑇𝑙𝑣 = 𝑇𝐶 in our experiments), 𝑃𝑉 is the vapor pressure at the liquid-vapor interface, and
𝑃𝐸𝑞 is the equilibrium pressure which can be calculated from the Kelvin equation.50
Following the procedure outlined in the Supporting Information (section 5), the total resistance
(𝑅𝑇𝑂𝑇) is:
Page 11
10
𝑅𝑇𝑂𝑇 =𝜌𝐶
𝑙
𝜌𝐻𝑔
ℎ2
12𝜇𝐻𝑔
𝐿𝐻+
2√2ℎ
3𝐿𝐻√
𝑀
𝜋𝑅𝑇𝐻
+𝜌𝐶
𝑙
𝜌𝐶𝑔
ℎ2
12𝜇𝐶𝑔
𝐿𝐶+
2√2ℎ
3𝐿𝐶√
𝑀
𝜋𝑅𝑇𝐶
+𝜌𝐶
𝑙
𝑁𝛼√
2𝜋𝑅𝑇𝑐
𝑀 (3)
where 𝜌𝐶𝑙 is the liquid-phase density at 𝑇𝐶.
The first, second and third terms in eq 3 represent the resistances in the heater region (𝑅𝐻),
chiller region (𝑅𝐶) and at the interface (𝑅𝐼), respectively. The condensate growth rate (𝑉𝐿) can be
written in terms of the difference between the reservoir pressure (𝑃𝑅 ) and the equilibrium
pressure (𝑃𝐸𝑞):
𝑉𝐿 =𝑃𝑅−𝑃𝐸𝑞
𝑅𝑇𝑂𝑇 (4)
Equations 3 and 4 can be solved simultaneously to predict the condensate growth rate as a
function of temperature and pressure (see Supporting Information, section 5). For isothermal
conditions (constant 𝑇𝐶 and 𝑇𝐻 ), eq 3 and eq 4 show a relationship between the condensate
growth rate (𝑉𝐿) and the reservoir pressure (𝑃𝑅). For a constant reservoir pressure (𝑃𝑅), these
equations predict a constant growth rate, which is similar to the trend found in the experiments
within the observation window (Figure 4A). When pressure is increased, based on our
experimental conditions, the gas density (𝜌𝐻𝑔
and 𝜌𝐶𝑔
) will increase more significantly compared
to the increase in viscosity (𝜇𝐻𝑔
and 𝜇𝐶𝑔
), and the liquid density (𝜌𝐶𝑙 ) remains almost constant,
which result in decreases in both 𝑅𝐻 and 𝑅𝐶. Meanwhile, 𝑅𝐼 is not affected by the increase in the
reservoir pressure (𝑃𝑅), and the total resistance (𝑅𝑇𝑂𝑇) will decrease (eq 3). The equilibrium
pressure (𝑃𝐸𝑞) remains unchanged since it is a property that depends on the channel height (ℎ)
and the temperature (𝑇𝐶), thus 𝑃𝑅 − 𝑃𝐸𝑞 will increase. Therefore, eq 3 and eq 4 predict that the
condensate growth rate (𝑉𝐿 ) increases with an increase in 𝑃𝑅 . Figure 4B shows the model
prediction (solid line) along with the experimental results in an isothermal condition (constant 𝑇𝐶
and 𝑇𝐻). Close agreement between the experimental results and our predicted condensate growth
rate is achieved by fitting the value of Nα to be 7.4× 10-5 (see Supporting Information, section 5).
We note, however, that for the 70 nm case the fit is relatively insensitive to the value of Nα, as
the interfacial resistance component as a whole is not dominant (see Supporting Information,
section 16).
For isobaric conditions (constant 𝑃𝑅), eq 3 and eq 4 show a relationship between the condensate
growth rate (𝑉𝐿) and the temperature in the chiller region (𝑇𝐶). For a constant 𝑇𝐶, these equations
predict a constant growth rate, which is similar to the trend found in the experiments (Figure 4C).
When 𝑇𝐶 is increased, the gas density (𝜌𝐻𝑔
and 𝜌𝐶𝑔
) will increase more significantly compared to
the increase in viscosity (𝜇𝐻𝑔
and 𝜇𝐶𝑔
), and the liquid density (𝜌𝐶𝑙 ) decreases, which results in
decreases in 𝑅𝐻 , 𝑅𝐶 and 𝑅𝐼, and further results in a decrease in 𝑅𝑇𝑂𝑇. However, the equilibrium
pressure (𝑃𝐸𝑞) will increase due to the saturation pressure (𝑃𝑆) increase (i.e., 𝑃𝑅 − 𝑃𝐸𝑞 decrease).
As a result, eq 3 and eq 4 predict that the condensate growth rate (𝑉𝐿) will decrease with the
increase in 𝑇𝐶 in an isobaric system. The condensate growth rate (𝑉𝐿) predicted from the flow
resistance model (solid line in Figure 4D) is in agreement with the experimental results (with 𝑁𝛼
= 7.4 × 10−5 as before). The nonlinear dependence of condensate growth rates on temperature
predicted by our model agree closely with the experiments (Figure 4D). The nonlinear behaviors
Page 12
11
are the result of both flow and interface resistances changing nonlinearly with temperature (via
density and viscosity of vapor phase).
We emphasize that the value of Nα is empirically determined from considering all experimental
conditions (15 experiments with temperatures and pressures that ranged between 314.0 and 343.7
K, and 2.0 and 3.4 MPa, respectively, see Supporting Information, section 5). The value of Nα
that gives the best agreement between our model-predicted condensation rates and those
measured in the 70 nm high channels was (7.4 ± 0.9) × 10-5 (details of the uncertainty calculation
of Nα are given in the Supporting Information, section 6). As an empirical parameter in our
experiments, Nα groups the effects of different physical phenomena that occur near the liquid-
vapor interface. For example, in nanoconfinement the sticking of the propane to silicon at the
three-phase line might contribute significantly to the condensation at the interface (details
discussed in the Supporting Information, section 14). Thus, Nα in our model does not represent
the classical condensation coefficient since it amalgamates the effects of the potential
condensation coefficient at the liquid-vapor interface and the sticking coefficient at the solid-
vapor interface44 in addition to potential effects of corner flow that result from the channel
geometry.
Figure 6. Contributions of the interface and flow resistances are shown as percentages of the total resistance (eq 3) for
both the highest and lowest condensation rates observed under isothermal and isobaric conditions. The interface, viscous
flow, and Knudsen flow resistance contributions are shown as green bars, blue bars, and hatches, respectively (see
Supporting Information, section 7).
The modeling results indicate that flow resistance (i.e., first two terms of eq 3) accounts for
~70% of the total resistance (𝑅𝑇𝑂𝑇) in our experiments. The effect is more pronounced when the
pressure ( 𝑃𝑅) or temperature (𝑇𝐶) is decreased. For example, the contribution of flow resistance
to the total resistance is increased from 71% at 3.4 MPa to 74% at 2.0 MPa in isothermal
conditions, and from 67% at 343.7 K to 76% at 314.0 K in isobaric conditions (Figure 6).
Therefore, it is essential to take both flow resistance and interface resistance into consideration
when studying condensate growth in sub-100 nanoconfinement. In contrast, when the scale of
the system is increased from 70 nm to 220 nm and further to 1000 nm, the maximum ratio of the
flow resistance to the total resistance decreases from 76% to 31% further to 7% (based on our
experimental conditions, see Supporting Information, section 11). Thus at the micro- or
macroscale, flow resistance can usually be neglected, and only heat and mass transfer processes
at the interface dominate the condensation process.30, 31 In addition, in the 70 nm channels, the
Knudsen flow contribution is ~10% of the total resistance (8% to 13% depending on the
Page 13
12
experimental conditions, as shown in Figure 6). This contribution decreases dramatically as the
channel height increases: we found it to be only ~1% at 220 nm and less than 0.1% at 1000 nm
(see Supporting Information, section 11, for details).
The present results demonstrate a deviation from bulk or microscale condensation growth that is
not attributable to capillary condensation. Capillary condensation is the early onset of liquid
formation at the nanoscale due to multilayer adsorption and occurs at pressures below the
saturation pressure or at temperatures above the saturation temperature. However, the kinetics of
capillary condensation is limited by the growth rate of the adsorbed layers, which is very slow at
conditions close to saturation.28, 52, 53 Moreover, previous experimental results showed that the
capillary condensation of water in graphitic pores with openings less than 10 nm is slow and
continues to occur for almost 3 days at room temperature.54 For the size of nanochannels
considered in this study (70 nm), the equilibrium pressure calculated from the Kelvin equation is
~0.5% less than the saturation pressure (𝑃𝐸𝑞 ≅ 0.995 𝑃𝑠, see Supporting Information, section 12),
but we did not observe any capillary condensation within 30 minutes due to the limitation of our
pumping accuracy (±0.5% of the set value). In fact, supercooling was always required to observe
the initiation of condensation in all cases tested here (see the Supporting Information, Figure S2).
The 70 nm channel size studied here represents a middle-ground in the context of nanoporous
shale reservoirs which have, in general, pore sizes that vary from single nanometers to hundreds
of nanometers. For instance the Bakken shale from Williston basin55 and the Pierre shale from
Colorado basin56 (both are significant US shale reservoirs) have dominant pore size within the
range of 60-110 nm and 40-90 nm, respectively. Collectively this work demonstrates that, in
pores on the scale of tens of nanometers in shale and related applications, phase-change
conditions deviate from bulk conditions due primarily to flow and interface resistances, rather
than capillary effects.
Conclusion
In summary, the condensation and condensate growth of propane in dead-end nanochannels are
experimentally investigated. Two different types of condensation mechanisms are distinguished
based on the number of condensed liquid segments, which is shown to be a function of
supercooling. The directly observed effects of both pressure and temperature on the condensate
growth and growth rate show very good agreement with those predicted by a suitably defined
thermo-fluid resistance model. The rate of liquid condensate growth in sub-100 nm nanochannels
is limited by vapor flow resistance (~70% of the total resistance), in contrast to larger scales (220
nm and 1000 nm) where the vapor flow resistance effect is greatly reduced. The vapor flow in
our experiments with 70 nm high channels is in the transitional flow regime where Knudsen flow
contributes up to 13% of the total resistance. The findings in this paper illustrate the
condensation and condensate growth of pure substance in the dead-end nanochannels, which will
be beneficial in understanding many natural processes and optimizing related applications. The
specific experimental conditions target tens of nanometer porous shale gas/oil formations, where
we found condensation conditions that deviate significantly from bulk conditions due primarily
to vapor flow and interface resistances.
Methods
Page 14
13
Fabrication Procedure of Nanofluidic Chip
In the nanofluidic chip, there were 10 groups of parallel dead-end nanochannels (10 separate
nanochannels in one group and each channel was 5.5 μm in width, 35 mm in length) placed
perpendicularly along the source microchannel. This identical layout was used for chips with 70
nm, 220 nm and 1000 nm channel heights. The inlet and outlet holes were at the two ends of the
source microchannel symmetrically, which was used to fill the chip with liquid propane and to
control the pressure inside the nanochannels.
To fabricate the nanofluidic chip, the pattern was first generated using AutoCAD and then
transferred onto a 5-inch chrome mask through a mask writer (Heidelberg μPG 501) with a
resolution of 1 μm. Second, the pattern on the mask was transferred to the S1818 photoresist on
top of the 1 mm thick silicon wafer by standard photolithography methods, as shown in Figure
S1A. Third, a high reactive ion etching (DRIE) instrument (Oxford Instruments PlasmaPro
Estrelas100 DRIE System) was used to etch the nanochannels with a certain height in the wafer,
as shown in Figure S1B. Next, the microchannel with a typical height of 50 μm was patterned
and etched into the same silicon wafer by repeating the standard photolithography and DRIE, as
shown in Figure S1C and S1D. Afterward, inlet and outlet holes were drilled through the silicon
substrate at both ends of the microchannel, as shown in Figure S1E. After cleaning the silicon
wafer and 2.2 mm thick borosilicate glass in the piranha solution (H2SO4:H2O2 = 3:1) for 20
minutes (Figure S1F), the silicon wafer and the borosilicate glass were bonded together through
anodic bonding (AML AWB-04 Aligner Wafer Bonder) to seal the channels, as shown in Figure
S1G. The bonding process was carried out for approximately 20 minutes at 400 ̊C with a voltage
of 600 V and a current of 4 mA. Finally, the chip was diced into the desired shape using a dicing
machine (Disco DAD3220 Automatic Dicing Saw).
Experimental Procedures
The assembled nanofluidic chip was installed on a customized manifold enabling appropriate
sealing for high pressures. The chip and manifold were connected to a sample source cylinder
(research grade propane, Praxair 99.99%), pump (TELEDYNE ISCO MODEL 260D), and
piston cylinder (HIP 70C3-10-P) via tubing and valves. The tubing, pump, piston chambers and
valves were cleaned thoroughly before connecting to the nanofluidic chip. The entire system was
vacuumed at 2 x 10-7 MPa (PFPE RV8) for 3 hours (~10 m3/h as the displacement rate) before
each set of experiments to minimize the residual air in the system (the mass fraction of the air
was less than 10−4% of the mass of propane in the system during the experiment, see Supporting
Information, section 13). Research-grade liquid propane was filled with high pressure (4 MPa)
into the nanochannels. The temperature and pressure were then set below the liquid-vapor
saturation line to ensure vaporization of liquid propane (a P-T diagram of propane is provided in
the Supporting Information, section 2). The vapor pressure was provided and controlled by the
pump and monitored by a pressure transducer (PX409-3.5KGUSBH) at the inlet of the
mainchannel. An isolation piston cylinder was used between the pump and chip to avoid
contamination of the sample (Figure 1A). The temperature in the chip was controlled by a copper
block heater (33 mm along the channel direction and 50 mm perpendicular to the channel
direction) and chiller (10 mm along the channel direction and 50 mm perpendicular to the
channel direction) clamped on the top of the chip.
Page 15
14
The test region, the field of view, (~1.5 x 1.5 mm) focused on the dead-end portion of the group
of 10 nanochannels. To ensure condensation initiated within test region, we introduced a
temperature difference across the chip by pairing the heater and chiller. The entrance region of
the nanochannels was set at a high temperature (𝑇𝐻) using electric heaters (OMEGA CIR-1030,
temperature was constantly set to 473 ± 1 K) inserted in the block heater controlled by PID,
which covered the first 29 mm length of the nanochannels. The copper dead-end portion of the
nanochannel was set at a lower temperature (𝑇𝐶). The temperature in this region was controlled
by a refrigerated circulator (Cole-Parmer Polystat® Standard 6.5L Heated Bath, temperature is
controlled between 278.2 ± 0.3 K to 368.2 ± 0.3 K) circulating water in the copper block chiller,
which covered the last 5 mm length of the nanochannels (the total length of the chiller was 10
mm). Both the heater and the chiller were in close contact with the silicon side of the chip using
thermally conductive paste (OMEGATHERM@201) for minimizing contact thermal resistance.
An insulation layer (Alumina Oxide Fiber, ~1 mm thick) was placed between them. A high-
speed camera (PCO 1200S, 50 frames per second) connected to the inverted optical microscope
(LEICA DMI 6000B) with a 10 x objective was used to observe the condensation directly and to
record the condensation growth process in the test region of the nanochannels.
For the isothermal condensation experiments (constant 𝑇𝐶), the temperature of the water bath
was set constant at 318.2 K and we waited 1 hour to reach steady-state thermal conditions. The
pressure was increased from 1MPa directly to the target pressure (1.4 MPa to 3.4 MPa, in steps
of 0.2 MPa), and for each pressure, we waited 30 minutes to detect any condensation. For the
isobaric condensation experiments, the pressure was kept at 1 MPa first, and temperature of the
water bath was increased from 303.2 K to 343.2 K (5 K as an incremental step). For each
temperature, we waited 1 hour to reach steady-state thermal conditions. Then we increased the
pressure from 1 MPa to 2.8 MPa (within 10 seconds). We also waited 30 minutes to detect the
condensation. Figure S2 shows the experimental conditions in the phase diagram of propane. As
seen there, hollow symbols indicate that condensation was not detected within 30 minutes of
observation while the solid symbols indicate that condensation occurred within that period. As
can be seen, supercooling was required to detect condensation within the 30 minutes of
observation time. Note that the abscissa in Figure S2 is the temperature in the chiller region of
the nanochannels (𝑇𝐶 ). This temperature is not necessarily the circulated water temperature
because of heat losses and heat transfer in the chip.
Acknowledgement
The authors gratefully acknowledge support from Schlumberger Canada Ltd., Alberta Innovates-
Energy and Environment Solutions, and from the Natural Sciences and Engineering Council of
Canada through a Collaborative Research and Development Grant, as well as on-going research
funding through the Discovery Grants program and the Canada Research Chairs program. In
addition, infrastructure funding provided by the Canada Foundation for Innovation and Ontario
Research Fund is gratefully acknowledged.
Supporting Information Available: Nanofluidic chip fabrication procedure,
experimental conditions, temperature calibration in the nanochannels, effects of pressure and
temperature on condensate growth, derivation of flow resistances model, error analysis of the
condensation coefficient, flow resistance contributions to the total resistance, real time
Page 16
15
observation of Type A and Type B condensation, temperature rise due to latent heat generated by
condensation, pressure gradient in the chiller region, contributions of Knusden flow and flow
resistance at isobaric and isothermal conditions for different scales, PEq/PS at different
experimental conditions, air remaining in the system, variation of empirical parameter, Nα, with
channel heights, discontinuous condensate growth, unity empirical parameter for model
prediction of the condensate growth rate in 70, 220, 1000 nm channel vs. experimental results, as
well as AFM result for 220 nm and 1000 nm channels. This material is available free of charge
via the Internet at http://pubs.ACS.org.
References
1. Ehn, M.; Thornton, J. A.; Kleist, E.; Sipilä, M.; Junninen, H.; Pullinen, I.; Springer, M.;
Rubach, F.; Tillmann, R.; Lee, B., A Large Source of Low-Volatility Secondary Organic
Aerosol. Nature 2014, 506, 476-479.
2. Kulmala, M.; Kontkanen, J.; Junninen, H.; Lehtipalo, K.; Manninen, H. E.; Nieminen, T.;
Petäjä, T.; Sipilä, M.; Schobesberger, S.; Rantala, P., Direct Observations of Atmospheric
Aerosol Nucleation. Science 2013, 339, 943-946.
3. Bones, D. L.; Reid, J. P.; Lienhard, D. M.; Krieger, U. K., Comparing the Mechanism of
Water Condensation and Evaporation in Glassy Aerosol. Proc. Natl. Acad. Sci. U.S.A. 2012, 109,
11613-11618.
4. Yang, H. Y.; Han, Z. J.; Yu, S. F.; Pey, K. L.; Ostrikov, K.; Karnik, R., Carbon Nanotube
Membranes with Ultrahigh Specific Adsorption Capacity for Water Desalination and
Purification. Nat. Commun. 2013, 4, 2220.
5. Hou, Y.; Yu, M.; Chen, X.; Wang, Z.; Yao, S., Recurrent Filmwise and Dropwise
Condensation on a Beetle Mimetic Surface. ACS Nano 2014, 9, 71-81.
6. Anand, S.; Paxson, A. T.; Dhiman, R.; Smith, J. D.; Varanasi, K. K., Enhanced
Condensation on Lubricant-Impregnated Nanotextured Surfaces. ACS Nano 2012, 6, 10122-
10129.
7. Kiwilsza, A.; Milanowski, B.; Drużbicki, K.; Coy, L. E.; Grzeszkowiak, M.; Jarek, M.;
Mielcarek, J.; Lulek, J.; Pajzderska, A.; Wąsicki, J., Mesoporous Drug Carrier Systems for
Enhanced Delivery Rate of Poorly Water-Soluble Drug: Nimodipine. J. Porous Mater. 2015, 22,
817-829.
8. Patzek, T. W.; Male, F.; Marder, M., Gas Production in the Barnett Shale Obeys a Simple
Scaling Theory. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 19731-19736.
9. Monteiro, P. J.; Rycroft, C. H.; Barenblatt, G. I., A Mathematical Model of Fluid and Gas
Flow in Nanoporous Media. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 20309-20313.
10. Kar, A.; Chiang, T.-Y.; Ortiz Rivera, I.; Sen, A.; Velegol, D., Enhanced Transport into
and out of Dead-End Pores. ACS Nano 2015, 9, 746-753.
11. Cueto-Felgueroso, L.; Juanes, R., Forecasting Long-Term Gas Production from Shale.
Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 19660-19661.
12. Devegowda, D.; Sapmanee, K.; Civan, F.; Sigal, R. F. In Phase Behavior of Gas
Condensates in Shales Due to Pore Proximity Effects: Implications for Transport, Reserves and
Well Productivity, SPE Annual Technical Conference and Exhibition, Society of Petroleum
Engineers: 2012.
13. Li, D., Encyclopedia of Microfluidics and Nanofluidics. Springer Science & Business
Media: 2008.
Page 17
16
14. Ali Beskok, G. E. K., Report: A Model for Flows in Channels, Pipes, and Ducts at Micro
and Nano Scales. Microscale Thermophys. Eng. 1999, 3, 43-77.
15. Fisher, L.; Gamble, R.; Middlehurst, J., The Kelvin Equation and the Capillary
Condensation of Water. Nature 1981, 290, 575-576.
16. Nune, S. K.; Lao, D. B.; Heldebrant, D. J.; Liu, J.; Olszta, M. J.; Kukkadapu, R. K.;
Gordon, L. M.; Nandasiri, M. I.; Whyatt, G.; Clayton, C., Anomalous Water Expulsion from
Carbon-Based Rods at High Humidity. Nat. Nanotechnol. 2016, 11, 791-797.
17. Ally, J.; Molla, S.; Mostowfi, F., Condensation in Nanoporous Packed Beds. Langmuir
2016, 32, 4494-4499.
18. Fisher, L. R.; Israelachvili, J. N., Experimental Studies on the Applicability of the Kelvin
Equation to Highly Curved Concave Menisci. J. Colloid Interface Sci.1981, 80, 528-541.
19. Kohonen, M. M.; Christenson, H. K., Capillary Condensation from Vapors of n-
Hexane/Perfluoro-n-Hexane Mixtures. J. Phys. Chem. B 2002, 106, 6685-6695.
20. Kohonen, M.; Maeda, N.; Christenson, H., Kinetics of Capillary Condensation in a
Nanoscale Pore. Phys. Rev. Lett. 1999, 82, 4667-4670.
21. Rykaczewski, K.; Scott, J. H. J., Methodology for Imaging Nano-to-Microscale Water
Condensation Dynamics on Complex Nanostructures. ACS Nano 2011, 5, 5962-5968.
22. Miljkovic, N.; Enright, R.; Wang, E. N., Effect of Droplet Morphology on Growth
Dynamics and Heat Transfer During Condensation on Superhydrophobic Nanostructured
Surfaces. ACS Nano 2012, 6, 1776-1785.
23. Anderson, D. M.; Gupta, M. K.; Voevodin, A. A.; Hunter, C. N.; Putnam, S. A.; Tsukruk,
V. V.; Fedorov, A. G., Using Amphiphilic Nanostructures to Enable Long-Range Ensemble
Coalescence and Surface Rejuvenation in Dropwise Condensation. ACS Nano 2012, 6, 3262-
3268.
24. Perry, J. L.; Kandlikar, S. G., Review of Fabrication of Nanochannels for Single Phase
Liquid Flow. Microfluid. Nanofluid. 2005, 2, 185-193.
25. Duan, C., Karnik, R., Lu, M. C., & Majumdar, A., Evaporation-Induced Cavitation in
Nanofluidic Channels. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 3688-3693.
26. Witharana, S.; Phillips, B.; Strobel, S.; Kim, H.; McKrell, T.; Chang, J.-B.; Buongiorno,
J.; Berggren, K.; Chen, L.; Ding, Y., Bubble Nucleation on Nano- to Micro-Size Cavities and
Posts: An Experimental Validation of Classical Theory. J. Appl. Phys. 2012, 112, 064904.
27. Nguyen, P. T.; Do, D.; Nicholson, D., On the Irreversibility of the Adsorption Isotherm in
a Closed-End Pore. Langmuir 2013, 29, 2927-2934.
28. Eijkel, J. C.; van den Berg, A., Water in Micro- and Nanofluidics Systems Described
Using the Water Potential. Lab Chip 2005, 5, 1202-1209.
29. Schneider, D.; Valiullin, R.; Monson, P. A., Filling Dynamics of Closed End
Nanocapillaries. Langmuir 2014, 30, 1290-1294.
30. Shekriladze, I.; Gomelauri, V., Theoretical Study of Laminar Film Condensation of
Flowing Vapour. Int. J. Heat Mass Transfer 1966, 9, 581-591.
31. Chen, Y.; Shi, M.; Cheng, P.; Peterson, G., Condensation in Microchannels. Nanoscale
Microscale Thermophys. Eng. 2008, 12, 117-143.
32. Mostowfi, F.; Molla, S.; Tabeling, P., Determining Phase Diagrams of Gas-Liquid
Systems Using a Microfluidic PVT. Lab chip 2012, 12, 4381-4387.
33. Molla, S.; Eskin, D.; Mostowfi, F., Pressure Drop of Slug Flow in Microchannels with
Increasing Void Fraction: Experiment and Modeling. Lab chip 2011, 11, 1968-1978.
Page 18
17
34. Haneveld, J.; Tas, N. R.; Brunets, N.; Jansen, H. V.; Elwenspoek, M., Capillary Filling of
Sub-10 nm Nanochannels. J. Appl. Phys. 2008, 104, 014309.
35. Yaghoubian, S.; Zandavi, S. H.; Ward, C., From Adsorption to Condensation: The Role
of Adsorbed Molecular Clusters. Phys. Chem. Chem. Phys. 2016, 18, 21481-21490.
36. Bruschi, L.; Mistura, G.; Nguyen, P. T.; Do, D. D.; Nicholson, D.; Park, S.-J.; Lee, W.,
Adsorption in Alumina Pores Open at One and at Both Ends. Nanoscale 2015, 7, 2587-2596.
37. Zeng, Y.; Phadungbut, P.; Do, D.; Nicholson, D., Anatomy of Adsorption in Open-End
and Closed-End Slit Mesopores: Adsorption, Desorption, and Equilibrium Branches of
Hysteresis Loop. J. Phys. Chem. C 2014, 118, 25496-25504.
38. Goldsmith, H.; Mason, S., The Flow of Suspensions through Tubes. II. Single Large
Bubbles. J. Colloid Interface Sci. 1963, 18, 237-261.
39. Goren, S. L., The Instability of an Annular Thread of Fluid. J. Fluid Mech. 1962, 12,
309-319.
40. Teng, H.; Cheng, P.; Zhao, T., Instability of Condensate Film and Capillary Blocking in
Small-Diameter-Thermosyphon Condensers. Int. J. Heat Mass Transfer. 1999, 42, 3071-3083.
41. Shui, L.; Eijkel, J.; Vandenberg, A., Multiphase Flow in Micro- and Nanochannels. Sens.
Actuators, B 2007, 121, 263-276.
42. Kim, E.; Whitesides, G. M., Imbibition and Flow of Wetting Liquids in Noncircular
Capillaries. J. Phys. Chem. B 1997, 101, 855-863.
43. Eijkel, J.; Dan, B.; Reemeijer, H.; Hermes, D.; Bomer, J.; Van Den Berg, A., Strongly
Accelerated and Humidity-Independent Drying of Nanochannels Induced by Sharp Corners.
Phys. Rev. Lett. 2005, 95, 256107.
44. Allendorf, M. D.; Kee, R. J., A Model of Silicon Carbide Chemical Vapor Deposition. J.
Electrochem. Soc. 1991, 138, 841-852.
45. Weh, L., Instability of the Contact Line between Three Fluid Phases During Mass
Transfer Processes. J. Colloid Interface Sci.2000, 227, 42-47.
46. Ghasemi, H.; Ward, C., Energy Transport by Thermocapillary Convection During
Sessile-Water-Droplet Evaporation. Phys. Rev. Lett. 2010, 105, 136102.
47. Zhang, H.; Nikolov, A.; Feng, J.; Wasan, D., The Dynamics of the Annular Liquid Layer
inside a Cylindrical Capillary. Phys. Fluids 2016, 28, 024107.
48. Kast, W.; Hohenthanner, C.-R., Mass Transfer within the Gas-Phase of Porous Media.
Int. J. Heat Mass Transfer 2000, 43, 807-823.
49. Marek, R.; Straub, J., Analysis of the Evaporation Coefficient and the Condensation
Coefficient of Water. Int. J. Heat Mass Transfer 2001, 44, 39-53.
50. Zandavi, S. H.; Ward, C., Characterization of the Pore Structure and Surface Properties
of Shale Using the Zeta Adsorption Isotherm Approach. Energy Fuels 2015, 29, 3004-3010.
51. Persad, A. H.; Ward, C. A., Expressions for the Evaporation and Condensation
Coefficients in the Hertz-Knudsen Relation. Chem. Rev. 2016, 116, 7727-7767.
52. Morishige, K.; Kawai, T.; Kittaka, S., Capillary Condensation of Water in Mesoporous
Carbon. J. Phys. Chem. C 2014, 118, 4664-4669.
53. Restagno, F.; Bocquet, L.; Crassous, J.; Charlaix, E., Slow Kinetics of Capillary
Condensation in Confined Geometry: Experiment and Theory. Colloids Surf., A 2002, 206, 69-
77.
54. Morishige, K.; Kittaka, S., Kinetics of Capillary Condensation of Water in Mesoporous
Carbon: Nucleation and Meniscus Growth. J. Phys. Chem. C 2015, 119, 18287-18292.
Page 19
18
55. Zargari, S.; Canter, K. L.; Prasad, M., Porosity Evolution in Oil-Prone Source Rocks.
Fuel 2015, 153, 110-117.
56. Kuila, U.; Prasad, M., Specific Surface Area and Pore-Size Distribution in Clays and
Shales. Geophys. Prospect. 2013, 61, 341-362.