The WeIzmann Supercooled Droplets Observation on Microarray (WISDOM)
is a new setup for studying ice nucleation in an array of
monodisperse droplets for atmospheric implications. WISDOM combines
microfluidics techniques for droplets production and a cryo-optic
stage for observation and characterization of freezing events of
individual droplets. This setup is designed to explore
heterogeneous ice nucleation in the immersion freezing mode, down to
the homogeneous freezing of water (235 K) in various cooling
rates (typically 0.1–10 Kmin-1). It can also be used
for studying homogeneous freezing of aqueous solutions in colder
temperatures. Frozen fraction, ice nucleation active surface site
densities and freezing kinetics can be obtained from WISDOM
measurements for hundreds of individual droplets in a single
freezing experiment. Calibration experiments using eutectic
solutions and previously studied materials are described. WISDOM
also allows repeatable cycles of cooling and heating
for the same array of droplets. This paper describes the WISDOM
setup, its temperature calibration, validation experiments and
measurement uncertainties. Finally, application of WISDOM to study
the ice nucleating particle (INP) properties of size-selected ambient Saharan dust particles
is presented.
Introduction
In mixed phase clouds, water droplets remain stable in a supercooled
state below 273 K and ice nucleates spontaneously as droplets
reach the homogeneous freezing temperature, below 236 K
(Pruppacher et al., 1998). At warmer temperatures, ice particles may
coexist with supercooled droplets, due to heterogeneous nucleation
facilitated by the presence of ice nucleating particles (INPs) (Cantrell
and Heymsfield, 2005). In cases where INPs are immersed in the droplet
before supercooling, referred to as immersion freezing mechanism, the
droplets first grow to supercritical size before freezing occurs (de
Boer et al., 2011). Observations and modeling studies suggest that
immersion freezing is the prominent mechanism for heterogeneous ice
formation in mixed phase clouds (Rosenfeld and
Woodley, 2000; Ansmann et al., 2008; Field et al.,
2012; Nagare et al., 2016; Possner et al., 2017).
Ice particles affect the radiative and microphysical properties of
mixed phase clouds and Earth's hydrological cycle. Therefore, they can
influence present and possibly future climate (Hoose and Möhler,
2012; IPCC, 2013). Studying ice formation in clouds is hence
important, and yet, due to its complexity, this process is still not
fully understood and presents a great challenge to laboratory and
field researchers as well as for clouds and climate modelers (DeMott
et al., 2010; Schnaiter et al., 2016; Ullrich et al., 2017).
Offline studies of immersion freezing often use cold stage techniques
(Budke and Koop, 2015). The basic idea is to place an array of
droplets over a cold stage and cool continuously until all are frozen,
to obtain a quantitative measurement of their corresponding freezing
temperatures (Vali, 1971). The droplets may be microliter sized and
observed with a simple camera. Smaller droplets, down to the
picoliter range, are usually observed under a microscope. In both
cases, freezing events are identified by optical changes in the
droplets when they crystalize (Knopf and Lopez, 2009; Murray et al., 2011; Atkinson et al., 2013; Hiranuma et al.,
2015).
Cold stage techniques may suffer from technical issues such as
droplets evaporation and vapor transfer due to the
Wegener–Bergeron–Findeisen process, where ice grows on the expense
of supercooled droplets or from seeding of neighboring droplets by
formation and surface growth of frost halos (Budke and Koop,
2015). Some cold stages instruments place oil over the droplets or use
droplets in oil emulsions to prevent these effects (Murray et al.,
2012). Still, results from cold stage experiments may be biased by
effects of inhomogeneous temperature of the substrate and the
surroundings or by various contaminations caused during droplets'
preparation and measurement (Hiranuma et al., 2015). Furthermore,
supercooling is limited due to the presence of impurities, which
increases with the volume of the droplet. Hence, to allow
comprehensive studies down to the homogeneous region, low volumes (<1µL) are used,
and generation of these volumes is not
trivial and may cause further complications.
Microfluidics is a technology of fluids manipulation in microchannel
arrays on a small device. Microfluidics is widely used in a range of
fields, such as physics, chemistry, biology, life sciences and the
food industry (Whitesides, 2006; Neethirajan et al., 2011; Sackmann et al., 2014). Recent studies used a
microfluidic apparatus to
study ice nucleation processes. Riechers et al. (2013) used
a microfluidics device to produce and collect monodisperse droplets of
water in various sizes, which were subsequently observed under
a microscope to study their homogeneous freezing. Stan et al. (2009)
recorded nucleation in water droplets and silver-iodide-seeded
droplets, while droplets were flowing during cooling. Schmitz
et al. (2009) established “Dropspots”, a static microfluidic array
of droplets, later used by Edd et al. (2009) to measure nucleation
kinetics. However, in the atmospheric heterogeneous ice nucleation
field, microfluidics techniques are not widely adopted, despite many
potential advantages.
The WeIzmann Supercooled Droplets Observation on Microarray (WISDOM)
is a new instrument combining the cold stage technique with
microfluidics technology and is designed to study immersion freezing
of micrometer-sized droplets, while addressing most of the technical
issues listed above. The WISDOM setup introduces several advantages of
microfluidics to the atmospheric ice nucleation field. WISDOM is based
on the “Dropspots” static array (Schmitz et al., 2009), which
enables the separation and the fixation of the droplets, so that each
individual droplet is recorded and studied, and also can be used for
repetition of freezing cycles and further exploration of the
nucleation process of a specific sample.
In this paper, we present the WISDOM setup, its calibration and
validation procedures. For validation experiments, homogeneous and
heterogeneous freezing were examined by following homogeneous freezing
rates of pure aqueous or solutions or deriving the efficiencies of
three types of mineral dust surrogates and collected ambient Saharan
dust to validate heterogeneous freezing experiments.
Homogeneous nucleation rates are described stochastically using the
volume-dependent ice nucleation rate (Jv (T)) in supercooled
droplets, given by the frozen fraction (fice) of droplets
with volume V at a certain temperature (T) and time intervals
(Δt) (Murray et al., 2010; Alpert et al., 2011; Riechers
et al., 2013),
JvT=-ln1-fice(T)VΔt.
Heterogeneous freezing is described by a singular approach that
assumes that nucleation occurs at a certain temperature due to a special
nucleation site. Hence, a cumulative number of nucleation sites per
unit surface area, ns, is used to describe the heterogeneous
nucleation efficiency at a certain temperature,
ns(T)=-ln1-fice(T)A,
where fice is the fraction of frozen droplets at
temperature T and A is the specific surface area of the immersed
particles in each droplet (Vali, 1971; Vali et al., 2015; Whale
et al., 2015).
Validation of WISDOM was further extended below the homogeneous
nucleation temperature of pure water, using aqueous solutions as the
freezing temperatures of solutions decrease as a function of the
solution water activity.
Experimental setupDroplets production and trapping
The WISDOM setup, shown in Fig. 1,
is made of a microfluidic setup which include a pressure-controlled
pump with four independent flow channels (OB1 MK3 by Elveflow),
a stereoscope (SMZ-171 by Motic) that permits a full view of all
channels and inlets, and a charge-coupled
device (CCD) camera (GS3 by Point Grey) that enable
real-time monitoring of the droplets production. The flows in the
channels are continuous and controlled by the pressure pump. One
channel is connected to the continuous (oil) phase, and a second
channel contains the sample (aqueous solution that can contain
INPs). The two phases meet in a narrow junction where monodisperse
droplets are generated due to the pressure exerted by one phase over
the other. The ratio between the flows determines the size of the
emerging droplets; the volume increases with increasing flow rate of
the sample. In this setup, droplets are suspended in an oil mixture,
consisting of mineral oil (Sigma Aldrich) and a 2 weight percent
(wt %) nonionic surfactant (span80®, Sigma
Aldrich), added for droplets stabilization (Riechers et al.,
2013). Hence, an array of picoliter (micrometer-size) droplets is
generated directly on a device.
The WISDOM setup. (a) The design of the microfluidic
device is based on Schmitz et al. (2009). Aqueous solutions
(including the sample) and oil are connected through the inlets and
merge in a junction to generate monodisperse droplets. Subsequently,
droplets flow into a trap array and settle in them as the flow is
stopped. The device is transferred into a cooling stage for
subsequent freezing experiments. (b) Upper and (c)
side views of the device, which is made of PDMS (polydimethylsiloxane), plasma glued to
a microscope glass slide, placed over the cooling silver block.
The principle of the Schmitz et al. (2009) design is that the droplets
flow into round chambers that are connected by the constriction
channel. At a certain flow, the droplets are squeezed through the
constriction channel and the array fills up with droplets. When the
flow is too weak or stopped, the constriction channel stops the
droplets' movement and they are trapped in the chambers. The droplets
are isolated and stable in the chambers and it is safe to move the
device from the generation stage to the cold stage for the freezing
experiments.
Devices are fabricated following the Schmitz et al. (2009)
protocol. Briefly, the device pattern is imprinted on
a polydimethylsiloxane (PDMS) polymer, later glued to a 1 mm
thick microscope glass slide using air plasma treatment. After the
plasma treatment the PDMS surfaces are hydrophilic (Eddings et al.,
2008). Therefore, the devices were used only in the following day,
after their surfaces became hydrophobic, following their exposure to
the atmosphere, or after their annealing at 60 ∘C for 30 min.
Freezing experiments and detection
The droplets array is placed in a commercial cryostage (Linkam,
THMS600) coupled to an optical microscope (Olympus, BX-51 with 10x
magnification, transmission mode). Experiments are monitored by
a microscope-mounted CCD camera (Allied Vision Technologies, Oscar
F-510C) for automatic identification of droplets and their freezing
events. Both the device and the cooling stage are cleaned with
2-propanol. Then the device is placed over the stage together with
a thin layer of oil on its bottom to provide good thermal
conductivity. Each freezing experiment starts with dry N2
purging to replace the moist atmosphere inside the cryostage to
prevent condensation. During the experiment, N2 flow prevents
water condensation on the cryostage window. Freezing experiments are
conducted with a cooling rate of 1 Kmin-1, which is
relevant for atmospheric conditions and also allows good
thermalization of the droplets, as will be shown in the calibration
section (Sect. 3.1). Each cooling cycle is followed by a heating
cycle, where melting is observed. Analysis of the melting onset is
then used to verify that the thermal conductivity is good and thus
validate the measurement.
In-house LabVIEW software is used to record a freezing experiment
movie file and analyze it offline. The temperature readings by the
Linkam cryostage temperature sensor (<±0.25 K for the
operated temperature range) and the movie frames are synchronized and
integrated. In most cases, 1 s (or 0.017 K at
1 Kmin-1) per frame is used. Currently, the WISDOM setup
operates with two types of devices that differ in their droplets' trap
diameter: 40 and 100 µm. Approximately 550 and
120 droplets can be monitored per experiment in the smaller and larger
diameter devices, respectively. Statistically, for the same sample,
larger droplets encompass more INP surface area within each droplet,
which can be more sensitive for detecting rare active sites. The
device can be reused for the same sample, if it is not clogged or
destroyed during the experiment. However, because the channels of the
40 µm device are smaller they tend to clog faster (for
instance by large particles).
Automatic detection of phase transitions
The optical brightness of a droplet changes during a phase transition
(freezing or melting) due to the different interaction of light with
the liquid and the solids. For phase transition detection, an in-house
image processing LabVIEW program monitors automatically the optical
brightness change. The program detects the droplets using a spherical
shape criterion and sets a square surrounding the droplet that defines
an array of pixels that are attributed to that specific
droplet. A change in the optical brightness is represented by the gray
level value of the image's pixels, ranging from 0 to 255. Freezing is
calculated per movie frame and is defined as the subtraction of the
brightness mean value for each droplet in two consecutive frames
(ΔGL), thus allowing derivation of freezing rates. At the
beginning of the analysis, the first 15 frames are used to identify
the noise level of the signal by calculating its standard deviation
SD (ΔGL). The program then searches
for the maximal freezing signal that is also greater than 5 times the
noise level. The temperature associated with this freezing signal is
assigned as the freezing temperature for that droplet.
In this algorithm, the program can distinguish successfully between
a phase transition event and noise that arises from the camera signal,
droplet movement or any other interruption. Figure 2 presents
a spectral analysis for different types of phase transitions observed
in WISDOM. Since WISDOM operates in transmission microscopy mode, the
light is scattered more efficiently by ice crystals in comparison with
a liquid droplet and a freezing event involves droplet darkening and
a negative signal. The negative signal of a freezing event of a single
droplet is shown in Fig. 2a. In comparison, during melting, the
droplet becomes brighter until all the crystals melt and the signal
is positive. In Fig. 2b and c the analysis of a melting signal and
a eutectic melting signal are presented for the entire frame.
Spectra of different phase transition events as observed in
WISDOM. (a) Freezing, (b) eutectic melting, and
(c) melting onset and clear point (liquefaction) are the
mean of all sampled droplets in a single experiment. The phase
transition is defined optically by the brightness information
obtained by the gray level of the image pixels.
SD (ΔGL)
describes the standard deviation of the difference in mean GL for two
consecutive frames. At the beginning of the experiment the noise
level is studied and freezing or melting are detected only if
SD (ΔGL) is as least 5 times greater than the noise
standard deviation level. Freezing and melting examples are for pure water droplets and
the eutectic melting example is for aqueous solution droplets of
NaCl. The eutectic melting point of NaCl and pure water melting point
are marked by the yellow and red lines in (b) and
(c), respectively. In all cases the droplet diameter
was 100 µm.
Results and WISDOM validationTemperature calibration
Temperature accuracy is a most important parameter in ice nucleation
experiments. An error propagation analysis by Riechers et al. (2013)
demonstrated how the temperature uncertainty may lead to
a distribution of temperatures between different
instruments. Therefore, we performed a thorough temperature
calibration using the known eutectic melting points and the melting
points of several aqueous solutions as calibration reference
points. Although ice nucleation experiments are performed while
cooling, the calibration experiments were done while heating to
improve the calibration precision and to avoid biases associated with
supercooling of the liquids (Budke and Koop, 2015).
Droplets thermalization
The temperature of the Linkam stage was measured at the upper center
part of the cooling stage and hence may differ from the actual
temperature of the droplets in the device due to thermal effects such
as temperature gradients and temperature lag. During cooling or
heating, a vertical temperature gradient may develop between the top
of the device, in contact with the inner ambient of the cryostage, and
the bottom of the device, which is in contact with the cooling silver
block. This gradient is expected to increase in magnitude, as the
temperature of the stage decreases or increases below or above ambient
temperature. Edd et al. (2009) used a similar setup and found
a difference between the top temperature and the bottom temperature of
about 2 K around 237 K and 3 K around 227 K.
Stan et al. (2009) also reported a vertical gradient of
1–2 K, which was reduced to 0.5 K with a flow of
cooled N2 over their device. In addition, a thermal lag may
arise during cooling or heating as the rate of temperature change is
high and precludes proper temperature equilibration. Hence, a more
accurate measurement of the droplet temperature is taken as a sum of
the stage temperature with the contributions of both thermal gradient
and lag.
Figure 3 demonstrates the combined effects of the temperature change rate
and device properties on the thermalization of pure water droplets
(double deionized water, DDW; 18.2 MΩ⋅cm).
Specifically, freezing and melting experiments at different rates were performed. The
temperature difference (ΔT) is the difference between the
measured values and the extrapolated temperature at equilibrium
conditions (0 Kmin-1). As expected, at slower temperature
cooling (heating) rates, the droplets are more equilibrated with the
stage temperature and ΔT is negligible. However, ΔT
increases at higher temperature cooling (heating) rates (e.g.,
10 Kmin-1). We observed that during cooling (heating) the
droplet is warmer (colder) than the stage and will freeze (melt) at
colder (warmer) temperature at higher cooling (heating) rates. We also
found that because ΔT is higher, in absolute value, for
devices of thicker PDMS and/or in devices which hold larger droplets,
it should be considered in the final temperature calibration for these
scenarios. Furthermore, ΔT was found to be almost symmetric
for higher temperature cooling (heating) rates. However, for
1 Kmin-1, ΔT during cooling is higher than that
for heating. Our conjecture is that this can be an effect of the
higher thermal gradient that develops as the temperature decreases
well below ambient (236 K).
The temperature difference (ΔT), defined as the
temperature difference between the stage temperature and the droplet
extrapolated temperature at equilibrium conditions at different
cooling (heating) rates. Freezing and melting points of pure water
are represented by circles and squares (40 and 100 µm
droplet diameter, respectively) for different PDMS thicknesses
and are represented by different colors. C denotes cooling and H
denotes heating. Droplets are close to equilibrium with the stage
temperature at rates <0.1Kmin-1 and ΔT
increases with increasing temperature change rate and with the PDMS
height.
Melting of aqueous solutions
Figure 4 presents the measured melting points of NaCl solutions with
different water activities. Reported melting points represent the
temperature in which all ice crystals in the droplets completely
melted, in contrast with melting temperatures reported for pure
liquids such as water, where the onset of melting is defined as the
melting point. Melting temperature results were consistent with
theoretical melting temperatures reported in Koop and Zobrist
(2009). This provides support to our conclusion that droplets
thermalize with the cooling stage when using a heating rate of
0.1–1 Kmin-1. For faster heating rates
(i.e., 10 Kmin-1), the thermal lag was more pronounced,
leading to a melting point shift of about 2–3 K. For more
concentrated solutions, faster heating rates shifted the melting
points more.
Temperature calibration by melting points of eutectic
solutions and pure water droplets for different heating
rates. Calibration is presented for 100 µm droplets with
4 mm PDMS thickness. The onsets of pure water droplets are
also considered. Eutectic melting is used for the colder
temperature range (<253K) while clear point
(liquefaction) at various water activities is taken for the warmer
temperature range. The upper panel presents the temperature
difference between the reference value and the cooling stage
temperature after calibration. Most of the differences are within
the range ±0.2 K.
Melting of eutectic solutions
Some aqueous solutions, such as NaCl and MgCl2, arrange in
a superlattice at a certain weight percent to form a solid with
a well-defined melting point (eutectic) (252.05 K for NaCl and
at 239.95 K for MgCl2) (Borgognoni et al., 2009;
Farnam et al., 2016). Interestingly, this type of melting has
a smaller optical signature compared to that of melting points of pure
substances, as can be seen in Fig. 2b. We have set a specific water
activity for a solution by determining its quantitative composition
using the extended aerosol inorganic model (E-AIM) (Clegg et al.,
1998) at room temperature (298 K). For calibration purposes,
because eutectic melting had a negligible variation for different
water activities used in the range of 0.99 to 0.95, we decided to take
their average to achieve a single melting value. These eutectic
melting temperatures are colder than the melting point of pure water
and, therefore, are used for expanding the WISDOM calibration range.
The final calibration is obtained for a device with a specific PDMS
thickness and at a specific cooling (heating) rate. For example,
devices with 100 µm diameter sized droplet and of
4 mm PDMS thickness have a linear calibration curve of
Tdrop=0.97×Tstage-0.46 at
0.1 Kmin-1.
Measurement reproducibility and device variability
The device's inter-variability was determined from 20 devices by comparing
their corresponding homogeneous freezing temperatures of pure
water. Specifically, each device was recycled three times with freshly
prepared droplets. Our results showed high reproducibility in the
median freezing temperature, where 50 % of the probed droplets
froze (T50), and high reproducibility in the melting point
temperature. Variation within devices was always smaller than
±0.2 K at 1 and 0.1 Kmin-1 (variation within
the devices over the whole freezing range is presented in Appendix A).
Homogeneous freezing rates of pure water
Homogeneous nucleation in supercooled water occurs in WISDOM between
238 and 237 K for a cooling rate of 1 Kmin-1 and
droplet diameter of 100 µm. Figure 5 shows WISDOM
nucleation rates in comparison with other similar instruments. It is
seen that the slopes of the rate and temperatures are similar to the
slopes reported for other instruments. The temperature where 50 %
of droplets froze (T50) is also in the expected range according
to model results of Hoffer (1961). WISDOM rates are slightly slower,
but within the uncertainty of the instruments used by Stan et al. (2009) and Riechers
et al. (2013). Stöckel et al. (2005) show
a higher nucleation rate. This discrepancy can be explained by
a decrease in the number of surface nucleation events due to the oil
phase surrounding our droplets, whereas in Stöckel et al. (2005)
droplets are suspended in air which allows surface nucleation to
occur.
The volume-dependent homogeneous freezing of pure water,
derived for 100 µm droplets with 4 mm PDMS
height. WISDOM rates are compared to relevant literature data. The
obtained fit from WISDOM is Jv(T)=exp(-3.4T+817.6). Temperature uncertainty for WISDOM is
±0.3 K.
Homogeneous and heterogeneous ice nucleation temperatures for
40 and 100 µm aqueous solution droplets as a function of
solution water activity. Freezing and melting curves are derived
from Koop et al. (2000). Heterogeneous ice nucleation is performed
with 0.1 wt % of Arizona test
dust (ATD) particles immersed in the droplets.
Homogeneous and heterogeneous freezing of aqueous
solutions
The water-activity-based ice nucleation theory by Koop et al. (2000)
describes the dependence of the freezing temperature depression on the
water activity (aw) of the solution, regardless of the
solute nature. Figure 6 presents the theoretical freezing and melting
temperature curves from Koop et al. (2000) with homogeneous ice
nucleation results measured in WISDOM, for four solutions with
atmospheric relevance. Water activities for NaCl, ammonium sulfate
(AS), glucose and levoglucosan mixtures were derived from the AIM
model and were corrected for glucose and levoglucosan, for which water
activity is temperature dependent (Zobrist
et al., 2008; Knopf and Lopez, 2009). The experiments were conducted at 1 Kmin-1
for 40 and 100 µm droplet diameters. The results follow
the theoretical curves of the water-activity-based ice nucleation, and
the dependence of the homogeneous freezing on the droplet volume is as
expected (Hoffer, 1961; Kuan-Ting and Wood, 2016) as the curve of the
smaller diameter droplets (green curve) is slightly colder compared
with the larger volume droplets (dark green curve).
Accumulated active site density spectra (ns) of
K-feldspar and illite-NX particles as a function of temperature
from validation experiments of immersion freezing in
WISDOM. Frozen fraction values are represented by a color bar for a
few surface area values that are exposed in 40 and
100 µm droplets. The dependence of the nucleation site
density on the surface area is illustrated here. WISDOM
uncertainties, propagated from surface area estimation and
measured frozen fraction errors, are included within the size of
the markers. For validation, previous immersion freezing
measurements are also presented (Hiranuma et al., 2015 and
Atkinson et al., 2013). T-binned data (1 ∘C) normalized
by the Brunauer–Emmett–Teller (BET) surface area from Hiranuma et al. (2015) are presented
in the color squares only for wet suspension analysis Bielefeld Ice Nucleation ARraY (BINARY),
red; Colorado State University Ice Spectrometer (CSU-IS), orange; Leeds Nucleation by Immersed Particles Instrument
(Leeds-NIPI), purple; Mainz acoustic levitator (M-AL), green; Mainz vertical wind tunnel (M-WT), black;
North Carolina State cold stage (NC State-CS), brown; and University of
Colorado Raman microscope cold stage (CU-RMCS), blue. The Hiranuma et al. (2015) log fit and
ns (BET) parameterization are also presented.
Similar experiments were conducted for 0.1 wt % of Arizona test
dust particles (ATD, Powder Technology Inc.) immersed in glucose
solution droplets. The ATD particles facilitate the ice nucleation at
warmer temperatures, in agreement with similar studies (Niedermeier et al., 2010; Hartmann
et al., 2011), and the freezing depression
follows the water-activity-based ice nucleation curves. Here, the
dependence of the freezing point on the droplet volume is more
pronounced, as the surface area of the immersed particles is higher;
hence, they contain higher number of nucleation sites (Marcolli et al.,
2007) as will be shown in the next section for two more types of dust.
Below 223 K, ice nucleation occurs at slightly lower
temperatures than expected by the theoretical freezing curve. As the
WISDOM temperature calibration is not valid in this temperature range,
we cannot conclude if this is due to a change in the thermal
conductivity of the device or an effect of the high concentration of
the solute in the water.
Summary of immersion freezing experiments performed for WISDOM
validation.
Accumulated active site density spectra (ns) of ambient
super-micron mineral dust particles collected in Israel during
the Saharan dust event in 2017, for three different sampling stages of
the micro-orifice uniform deposit impactor (MOUDI); D50 of 1, 1.8 and 3.2 µm.
Heterogeneous nucleation and <inline-formula><mml:math id="M147" display="inline"><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi mathvariant="normal">s</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> spectra of INP in
pure waterStandard dust powder
Heterogeneous freezing efficiencies of suspended mineral dusts
K feldspar and illite NX in supercooled water droplets are presented
in Fig. 7 and summarized in Table 1 and are compared to recent
published data. The particles are suspended at different wt %
and the frozen fraction of each suspension is derived as a function of
temperature as represented by the color bar. To examine the freezing
efficiency and compare the different mineral dust types, the results
are normalized to the surface area within each droplet. Experiments
were performed at 1 Kmin-1 for 40 and 100 µm
droplet diameters. Suspension preparation and evaluation of the
surface area are described in Appendix B.
Accumulated active site density spectra (ns) of
ambient super-micron mineral dust particles collected in Israel
during the dust event in 2017 for three MOUDI stages that were analyzed
with D50 of 1, 1.8 and 3.2 µm. The fit lnns=-005T2+2468T-293 (R2=0.98)
is also presented. References of K-feldspar standard particles
activated in WISDOM, Leeds-NIPI (Atkinson et al., 2013) and
Leipzig Aerosol Cloud Interaction Simulator (LACIS; Niedermeier et al., 2015) instruments are presented, as well as
ambient dust particles that were analyzed in AIDA and included
Israeli dust (Niemand et al., 2012).
The results demonstrate the effect of dust surface area immersed in
the droplets on the freezing parameters. The freezing temperatures
increase with increasing surface area and are also reflected in the
warming of the median frozen fraction (T50) colored in
yellow. The spectra of the number of nucleation sites per unit surface
area (ns) also support surface area dependence because all
spectra converge to a single line. The ns results show the
increase of nucleation sites at colder temperatures. Results from
WISDOM are in good agreement with similar analyses from other
instruments. In particular, ns is in best agreement with the
Leeds-NIPI (Murray et al., 2011; Broadley et al., 2012) results both
for K-feldspar and for illite-NX particles. Results of illite-NX
particles are also in good agreement with the BINARY instrument (Budke
and Koop, 2015) and reside within the uncertainty of both
instruments. The linear trend of a few weight percent values supports the assumption
that particles in suspension are uniformly distributed and the
droplets contain approximately the same surface area.
Ambient mineral dust
WISDOM can also be used for analyzing collected ambient
particles. Mineral dust particles were collected in Rehovot, Israel
(31.9∘ N, 34.8∘ E; about 80 ma.m.s.l.), during
dust storm events on 12–13 March 2017). The dust was transported from
the Sahara and North Africa. Size-segregated ambient dust
particles were collected on cyclopore polycarbonate filters using
a micro-orifice uniform deposit impactor (MOUDI; MSP Corporation model
110-R, Marple et al., 1991), which operated at 30 Lmin-1
and for 24 h, similarly to Huffman et al. (2013) and Mason
et al. (2015). The MOUDI has eleven stages with cut points (D50) of
0.056, 0.10, 0.18, 0.32, 0.56, 1.0, 1.8, 3.2, 5.6, 10 and
18 µm. The size distribution of the particles was obtained
using an optical particle counter (OPC; GRIMM Technologies model 1.109) in
the range of 0.25–32 µm and used for estimations of
surface area immersed in the droplets (further details in Appendix C).
For heterogeneous freezing experiments, a quarter of each filter is
placed with 300 µL DDW in a 1.5 mL Eppendorf vial
and particles were extracted by intensive dry sonication (Hielcher;
model UP200St VialTweeter). In Fig. 8, the spectra of the nucleation
sites per unit surface area (ns) of three super-micron stages
(D50 of 1.0, 1.8, 3.2 µm) are presented and
summarized in Table 1. It is also seen that there are slightly more active
sites for the larger particles (3.2 µm), as their surface
area is higher and there is a higher probability to contain an active
site. In Fig. 9, ns curves of the collected dust are compared to
references of K-feldspar standard particles, analyzed in different
instruments (the Leeds-NIPI, Atkinson et al., 2013; LACIS, Niedermeier et al., 2015), and to measurements of ambient dust
samples, from different locations around the world, including Israeli
settled dust, which was analyzed in the Aerosol Interaction and
Dynamics in the Atmosphere (AIDA) chamber. Moreover, the
freezing of the size-resolved mineral dust analyzed in this study by
WISDOM (slope in the temperature range) is consistent with the (gray)
polygon that represents the estimated freezing efficiency for natural
concentrations of K feldspar in internally mixed mineral types
(Atkinson et al., 2013). The results are also in agreement with
Niemand et al. (2012), especially between 243 and 249 K. At
warmer temperatures, the ns of ambient dust in this study showed
lower efficiency than in Niemand et al. (2012). This difference can
extend to 1 order of magnitude in ns and is more pronounced at
smaller particles that were analyzed (around 1–1.8 µm
diameter). For the larger particles, more nucleating sites are
observed. Both the current study and Niemand et al. (2012) suggest
that K feldspar is involved with the warmer part of their data (>248K) as the results are consistent with the Atkinson
et al. (2013) scale for ambient samples. The slope of the ns
derived in this study is similar to the slope from Atkinson
et al. (2013) at warmer temperatures. For the colder regime, the slope
is similar to the slope presented by standard K-feldspar particles in
Niedermeier et al. (2015).
WISDOM in comparison to other cold stage instruments
The microfluidics technology used in WISDOM solves some substantial
issues inherent in other currently used instruments. (1) It allows for good control
of the size and number of monodisperse droplets. (2) Additionally, it also allows for the fast production
of hundreds of nearly monodisperse droplets, which minimizes sample
sedimentation or agglomeration that may occur in a suspension, leading
to a good estimation of the surface area of the suspended
material. Moreover, several droplet diameters can be employed in the
same device without its modification, which allows for (3) good statistics achieved by
the individual analysis of hundreds of droplets; (4) the individual analysis of
monodisperse droplets, in contrast to some emulsion techniques (such
as a differential scanning calorimeter, DSC), which allows for obtaining the
frozen fraction at each temperature and acquiring detailed
information about active sites and freezing rates. (5) The use of oil
minimizes possible artifacts from the droplets' evaporation, neighbor
seeding or vapor transfer due to the Wegener–Bergeron–Findeisen
processes. (6) The small volumes decrease freezing artifacts by
impurities, thus allowing us to reach the homogeneous freezing threshold
(-37 ∘C). (7) The static droplet array allows the recycling of the droplets for multiple
freeze–thaw cycles.
(8) The microfluidics method and the
small droplet volumes enable working with small sample volumes which
can be an advantage when working with atmospheric samples.
WISDOM has a very accurate temperature calibration that spans a wide
temperature range, using the eutectic freezing method. WISDOM most
resembles the instrument used by in Edd et al. (2009). However, it
seems that issues with temperature calibration in Edd et al. (2009)
led to a temperature offset and hence different freezing rates. Stan
et al. (2009) achieved better temperature accuracy and high
statistics. However, the freezing experiment was conducted in a flow
mode, which is more complicated than in the WISDOM setup and requires
complicated modeling. In addition, the cooling rates that were used
were very fast, which induces additional errors. Riechers
et al. (2013) had high temperature accuracy as they also used
a DSC. However, they had to collect the droplets from the device as
there was no static array option and this may add further complication
and contamination.
The microfluidics technology also has disadvantages. These may
include the following: (1) oil may interact with some of the analyzed particles,
possibly leading to biased data; (2) the microchannels are susceptible
to clogging; (3) it is not possible to perform any post analysis to
the droplets' content after the experiment; (4) the small droplets'
volumes reduce the sensitivity to rare active sites. This may be
solved by performing many experiments or by using larger droplets with
more surface area within the droplets.
Summary and conclusions
The new setup WISDOM is based on microfluidics technology and its
detailed validation is presented. Based on a set of validation
measurements and a good agreement with other instruments, we conclude
that WISDOM is a suitable tool for studying atmospheric ice
nucleation, both in homogeneous and heterogeneous immersion freezing
modes. Results of homogeneous freezing correspond to
water-activity-based nucleation theory in supercooled droplets and
represent volume nucleation rates well. Heterogeneous freezing in
supercooled droplets also agrees well with literature
data. Furthermore, freezing efficiency dependence on the particles'
surface area within the droplets is clearly observed. Using
microfluidics allows a mass production of picoliter monodisperse
droplets using low volumes of suspensions, which can be beneficial for
immersion freezing studies over a wide range of supercooling down to
the homogenous temperature region. The good reproducibility of the
devices, proved using pure water freezing cycles, enables the
recycling of the same device for a few freezing cycles. It is also shown
that the temperature uncertainty can be reduced if the temperature
calibration includes the microfluidic devices' properties in the
working temperature change rates, especially for melting
experiments. In this work we have also demonstrated how WISDOM can be
applied for studying the ice nucleation properties of ambient samples
that contain a very small quantity of sample. The particles were
collected using the MOUDI during Saharan dust storm event. Results are
in correspondence with literature data of ambient dust and further
support Atkinson et al. (2013) and the possible importance of
K feldspar for ice nucleation in clouds, but further analysis of the
mineralogy is still needed in order to verify that.
Data availability
Data presented in this article are available from the first
author upon request (naama.reicher@weizmann.ac.il).
Device inter-variability over the whole freezing range
Figure A1 presents the reproducibility of the microfluidic devices. For each
device, the temperature variation between different freezing cycles of pure
water is presented for the entire range of the freezing at various frozen
fractions (0.1 to 1). These experiments were conducted for three cooling
rates, 0.1, 1 and 10 Kmin-1. For 1 Kmin-1 the
device's variability was the smallest, and the deviation in the results
between different cycles was <0.2K for most cases. While in some
cycles the temperature was reproducible in <0.05K, in other
cycles the temperature varied in <0.2K. This is not valid for
frozen fractions <0.2, where the variability was the highest, as was the
case for the other two cooling rates. This may be due to contaminants that
exist in the water or in the devices themselves. The preparation of the
devices is mostly inside a hood, but ambient particles may be trapped during
the process. For a cooling rate of 0.1 Kmin-1, the variability
between the different cycles was also <0.2K, but the variability
was higher in comparison to the variability seen at 1 Kmin-1.
This can be explained stochastically and also may be attributed to better
resolution of temperature reading during slower cooling rates. For
10 Kmin-1 the variability was between 0.2 and 0.3 K. The
faster cooling rate may slow the equilibration of droplets' temperature with
respect to the stage (as demonstrated in this work), and also low resolution
of temperature reading due to the fast cooling rate. The variability
presented here is also probably affected by the uncertainty of the
temperature sensor of the Linkam stage (< 0.25 K).
Variability in the WISDOM devices for three cooling
rates. The markers present the average temperature variability for
all the devices and the error bars represent 1 standard deviation. A line is
placed at Δ=0.2K, the upper value of which explains
the variability in the results of different freezing cycles at 0.1
and 1 Kmin-1.
Suspension preparation and characterization
Illite-NX, ATD and K-feldspar powders were suspended in
DDW and sonicated twice for 30 s with a 20 s pause,
using a Hielcher UP200St VialTweeter, adjusted especially for Eppendorf
vials. K-feldspar suspensions were additionally stirred overnight as
sonication alone was not enough to achieve a good suspension and
intensive sedimentation was observed. For validation experiments,
suspensions of 0.1 to 1 wt % were used. Figure B1 presents
nucleation site densities for illite-NX and K-feldspar particles and
the freezing efficiencies as function of the surface area in the
droplets. Characterization of the powders can be found in Marcolli et al. (2007), in Atkinson et al. (2013) and in Hiranuma
et al. (2015), and quantification the powders' specific surface area was based on
N2 adsorption analysis of Brunauer–Emmett–Teller (BET)
(Brunauer et al., 1938) using a Quantachrome Instruments Nova 2200e and
resulted in 1.9±0.6m2g-1 for the K-feldspar
powder, 108.6±2.8m2g-1 for the illite-NX powder
and 37.1±1.4m2g-1 for the ATD powder. In order
to ensure a proper analysis of the surface area, and avoid possible
surface contaminants such as water, surface cleaning was done by degassing
the powders at 60 ∘C for 3 h ahead of the BET
analysis. Evaluation of the surface area in each droplet was then
calculated by the wt % which was used, knowing the approximate
surface area per mass and assuming that the mass is distributed
uniformly inside the droplet with the same volume.
Accumulated active site density spectra (ns) of
K-feldspar (a) and illite-NX (b) particles with
different surface areas suspended in water at a cooling rate of
1 ∘C per minute. The error bars are located at three
representative frozen fractions: 0.1, 0.5 and 0.9.
Collection of ambient particles during dust storm events in
Rehovot
The GRIMM measurement was synchronized to the MOUDI stages for the
estimation of the total surface area that was collected on the filter
for droplet surface area estimation. For that, two base assumptions
were made: (1) all the particles that were collected are extracted to
the water later used for the freezing experiments; (2) sphericity of
the particles is assumed. The GRIMM bins are synchronized to the MOUDI stages
based on collection efficiency of the MOUDI, obtained from Marple
et al. (1991). For example, on certain MOUDI stages, all the particles
whose own diameter is larger than the D50 have a high chance of
being impacted on that stage. All the rest of the sizes, which are smaller
in their diameter, will continue to the next stage and will have a high
chance of depositing there. Hence, the GRIMM's bins were synchronized to
the MOUDI D50 stages. For the ns calculations, the surface
area was based on the number of particles that were measured in
a certain bin and their total surface area. To calculate the surface
area of a particle (assuming sphericity) in a certain bin, the
midpoint of that bin was used as a radius. To calculate the total mass
of the particles in each filter, dust density of quartz was used
(2.65 gcm-3), as this is usually the dominant mineral
(Mahowald et al., 2014). The error of the ns data is propagated
from the error in the frozen fraction, the error of the droplet's
volume and the error of the MOUDI's collection efficiency in the
different stages – the latter was the dominant one.
For control, analysis of blank filters was done. The blanks were
sonicated before analyzing them and freezing was mostly colder than
the freezing temperatures that are presented here and hence no special
reduction of the final active sites was done.
Competing interests
The authors declare that they have no conflict of
interest.
Acknowledgements
We gratefully acknowledge support from the Ice Nuclei Research Unit
(INUIT) of the German DFG, to The Helen Kimmel Center for Planetary
Sciences, The de Botton Center for Marine Sciences and the
Weizmann–UK Making Connections program for funding this work. We
also thank Daniel Knopf, Carsten Budke, Thomas
Koop, Ido Braslavski, and Nir Freidman for their advices
as well as Ben Murray and Heinz Bingemer for sharing
the K-feldspar and illite-NX powder standards.
Edited by: Joachim Curtius
Reviewed by: two anonymous referees
ReferencesAlpert, P. A., Aller, J. Y., and Knopf, D. A.: Ice nucleation from aqueous NaCl droplets with and without marine diatoms, Atmos. Chem. Phys., 11, 5539–5555, 10.5194/acp-11-5539-2011, 2011.Ansmann, A., Tesche, M., Althausen, D., Müller, D., Seifert, P., Freudenthaler, V., Heese, B., Wiegner, M., Pisani, G., Knippertz, P., and Dubovik, O.: Influence of Saharan dust on cloud glaciation in southern Morocco during the Saharan Mineral Dust Experiment, J. Geophys. Res.-Atmos., 113, D04210, 10.1029/2007JD008785, 2008.
Atkinson, J. D., Murray, B. J., Woodhouse, M. T., Whale, T. F., Baustian, K. J., Carslaw, K. S., Dobbie, S., O'Sullivan, D., and Malkin, T. L.: The importance of feldspar for ice nucleation by mineral dust in mixed-phase clouds, Nature, 498, 355–358, 2013.
Borgognoni, C. F., Tattini Junior, V., Ayrosa, A. M. I. B., Polakiewicz, B., Leirner, A. A., Maizato, M. J. S., Higa, O. Z., Beppu, M. M., and Pitombo, R. N. d. M.: The influence of freezing rates on bovine pericardium tissue Freeze-drying, Braz. Arch. Biol. Techn., 52, 1493–1504, 2009.Broadley, S. L., Murray, B. J., Herbert, R. J., Atkinson, J. D., Dobbie, S., Malkin, T. L., Condliffe, E., and Neve, L.: Immersion mode heterogeneous ice nucleation by an illite rich powder representative of atmospheric mineral dust, Atmos. Chem. Phys., 12, 287–307, 10.5194/acp-12-287-2012, 2012.
Brunauer, S., Emmett, P. H., and Teller, E.: Adsorption of gases in multimolecular layers, J. Am. Chem. Soc., 60, 309–319, 1938.Budke, C. and Koop, T.: BINARY: an optical freezing array for assessing temperature and time dependence of heterogeneous ice nucleation, Atmos. Meas. Tech., 8, 689–703, 10.5194/amt-8-689-2015, 2015.
Cantrell, W. and Heymsfield, A.: Production of Ice in Tropospheric clouds: a review, B. Am. Meteorol. Soc., 86, 795–807, 2005.Clegg, S. L., Brimblecombe, P., and Wexler, A. S.: Thermodynamic Model of the System H+–NH4+–SO42-–NO3-–H2O at Tropospheric
Temperatures, J. Phys. Chem. A, 102, 2137–2154, available at: http://www.aim.env.uea.ac.uk/aim/model3/model3a.php, 1998.de Boer, G., Morrison, H., Shupe, M. D., and Hildner, R.: Evidence of liquid dependent ice nucleation in high-latitude stratiform clouds from surface remote sensors, Geophys. Res. Lett., 38, L01803, 10.1029/2010GL046016, 2011.
DeMott, P. J., Prenni, A. J., Liu, X., Kreidenweis, S. M., Petters, M. D., Twohy, C. H., Richardson, M. S., Eidhammer, T., and Rogers, D. C.: Predicting global atmospheric ice nuclei distributions and their impacts on climate, P. Natl. Acad. Sci. USA, 107, 11217–11222, 2010.
Edd, J. F., Humphry, K. J., Irimia, D., Weitz, D. A., and Toner, M.: Nucleation and solidification in static arrays of monodisperse drops, Lab Chip, 9, 1859–1865, 2009.Eddings, M. A., Johnson, M. A., and Gale, B. K.: Determining the optimal PDMS–PDMS bonding technique for microfluidic devices, J. Micromech. Microeng., 18, 067001, 10.1088/0960-1317/18/6/067001, 2008.Farnam, Y., Villani, C., Washington, T., Spence, M., Jain, J., and Jason Weiss, W.: Performance of carbonated calcium silicate based cement pastes and mortars exposed to NaCl and MgCl2 deicing salt, Constr. Build. Mater., 111, 63–71, 2016.
Field, P. R., Heymsfield, A. J., Shipway, B. J., DeMott, P. J., Pratt, K. A., Rogers, D. C., Stith, J., and Prather, K. A.: Ice in clouds experiment–layer clouds, Part II: Testing characteristics of heterogeneous ice formation in lee wave clouds, J. Atmos. Sci., 69, 1066–1079, 2012.Hartmann, S., Niedermeier, D., Voigtländer, J., Clauss, T., Shaw, R. A., Wex, H., Kiselev, A., and Stratmann, F.: Homogeneous and heterogeneous ice nucleation at LACIS: operating principle and theoretical studies, Atmos. Chem. Phys., 11, 1753–1767, 10.5194/acp-11-1753-2011, 2011.Hiranuma, N., Augustin-Bauditz, S., Bingemer, H., Budke, C., Curtius, J., Danielczok, A., Diehl, K., Dreischmeier, K., Ebert, M., Frank, F., Hoffmann, N., Kandler, K., Kiselev, A., Koop, T., Leisner, T., Möhler, O., Nillius, B., Peckhaus, A., Rose, D., Weinbruch, S., Wex, H., Boose, Y., DeMott, P. J., Hader, J. D., Hill, T. C. J., Kanji, Z. A., Kulkarni, G., Levin, E. J. T., McCluskey, C. S., Murakami, M., Murray, B. J., Niedermeier, D., Petters, M. D., O'Sullivan, D., Saito, A., Schill, G. P., Tajiri, T., Tolbert, M. A., Welti, A., Whale, T. F., Wright, T. P., and Yamashita, K.: A comprehensive laboratory study on the immersion freezing behavior of illite NX particles: a comparison of 17 ice nucleation measurement techniques, Atmos. Chem. Phys., 15, 2489–2518, 10.5194/acp-15-2489-2015, 2015.
Hoffer, T. E.: A laboratory investigation of droplet freezing, J. Meteorol., 18, 766–778, 1961.Hoose, C. and Möhler, O.: Heterogeneous ice nucleation on atmospheric aerosols: a review of results from laboratory experiments, Atmos. Chem. Phys., 12, 9817–9854, 10.5194/acp-12-9817-2012, 2012.Huffman, J. A., Prenni, A. J., DeMott, P. J., Pöhlker, C., Mason, R. H., Robinson, N. H., Fröhlich-Nowoisky, J., Tobo, Y., Després, V. R., Garcia, E., Gochis, D. J., Harris, E., Müller-Germann, I., Ruzene, C., Schmer, B., Sinha, B., Day, D. A., Andreae, M. O., Jimenez, J. L., Gallagher, M., Kreidenweis, S. M., Bertram, A. K., and Pöschl, U.: High concentrations of biological aerosol particles and ice nuclei during and after rain, Atmos. Chem. Phys., 13, 6151–6164, 10.5194/acp-13-6151-2013, 2013.
IPCC: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 2013.
Knopf, D. A. and Lopez, M. D.: Homogeneous ice freezing temperatures and ice nucleation rates of aqueous ammonium sulfate and aqueous levoglucosan particles for relevant atmospheric conditions, Phys. Chem. Chem. Phys., 11, 8056–8068, 2009.
Koop, T. and Zobrist, B.: Parameterizations for ice nucleation in biological and atmospheric systems, Phys. Chem. Chem. Phys., 11, 10839–10850, 2009.
Koop, T., Luo, B., Tsias, A., and Peter, T.: Water activity as the determinant for homogeneous ice nucleation in aqueous solutions, Nature, 406, 611–614, 2000.
Mahowald, N., Albani, S., Kok, J. F., Engelstaeder, S., Scanza, R., Ward, D. S., and Flanner, M. G.: The size distribution of desert dust aerosols and its impact on the Earth system, Aeolian Res., 15, 53–71, 2014.Marcolli, C., Gedamke, S., Peter, T., and Zobrist, B.: Efficiency of immersion mode ice nucleation on surrogates of mineral dust, Atmos. Chem. Phys., 7, 5081–5091, 10.5194/acp-7-5081-2007, 2007.
Marple, V. A., Rubow, K. L., and Behm, S. M.: A Microorifice Uniform Deposit Impactor (MOUDI): description, calibration, and use, Aerosol Sci. Tech., 14, 434–446, 1991.Mason, R. H., Chou, C., McCluskey, C. S., Levin, E. J. T., Schiller, C. L., Hill, T. C. J., Huffman, J. A., DeMott, P. J., and Bertram, A. K.: The micro-orifice uniform deposit impactor–droplet freezing technique (MOUDI-DFT) for measuring concentrations of ice nucleating particles as a function of size: improvements and initial validation, Atmos. Meas. Tech., 8, 2449–2462, 10.5194/amt-8-2449-2015, 2015.
Murray, B. J., Broadley, S. L., Wilson, T. W., Bull, S. J., Wills, R. H., Christenson, H. K., and Murray, E. J.: Kinetics of the homogeneous freezing of water, Phys. Chem. Chem. Phys., 12, 10380–10387, 2010.Murray, B. J., Broadley, S. L., Wilson, T. W., Atkinson, J. D., and Wills, R. H.: Heterogeneous freezing of water droplets containing kaolinite particles, Atmos. Chem. Phys., 11, 4191–4207, 10.5194/acp-11-4191-2011, 2011.
Murray, B. J., O'Sullivan, D., Atkinson, J. D., and Webb, M. E.: Ice nucleation by particles immersed in supercooled cloud droplets, Chem. Soc. Rev., 41, 6519–6554, 2012.Nagare, B., Marcolli, C., Welti, A., Stetzer, O., and Lohmann, U.: Comparing contact and immersion freezing from continuous flow diffusion chambers, Atmos. Chem. Phys., 16, 8899–8914, 10.5194/acp-16-8899-2016, 2016.
Neethirajan, S., Kobayashi, I., Nakajima, M., Wu, D., Nandagopal, S., and Lin, F.: Microfluidics for food, agriculture and biosystems industries, Lab Chip, 11, 1574–1586, 2011.Niedermeier, D., Hartmann, S., Shaw, R. A., Covert, D., Mentel, T. F., Schneider, J., Poulain, L., Reitz, P., Spindler, C., Clauss, T., Kiselev, A., Hallbauer, E., Wex, H., Mildenberger, K., and Stratmann, F.: Heterogeneous freezing of droplets with immersed mineral dust particles – measurements and parameterization, Atmos. Chem. Phys., 10, 3601–3614, 10.5194/acp-10-3601-2010, 2010.
Niedermeier, D., Augustin-Bauditz, S., Hartmann, S., Wex, H., Ignatius, K., and Stratmann, F.: Can we define an asymptotic value for the ice active surface site density for heterogeneous ice nucleation?, J. Geophys. Res.-Atmos., 120, 5036–5046, 2015.
Niemand, M., Möhler, O., Vogel, B., Vogel, H., Hoose, C., Connolly, P., Klein, H., Bingemer, H., DeMott, P., Skrotzki, J., and Leisner, T.: A particle-surface-area-based parameterization of immersion freezing on desert dust particles, J. Atmos. Sci., 69, 3077–3092, 2012.O, K.-T. and Wood, R.: Exploring an approximation for the homogeneous freezing temperature of water droplets, Atmos. Chem. Phys., 16, 7239–7249, 10.5194/acp-16-7239-2016, 2016.
Possner, A., Ekman, A. M. L., and Lohmann, U.: Cloud response and feedback processes in stratiform mixed-phase clouds perturbed by ship exhaust, Geophys. Res. Lett., 44, 1964–1972, 2017.
Pruppacher, H. R., Klett, J. D., and Wang, P. K.: Microphysics of clouds and precipitation, Aerosol Sci. Tech., 28, 381–382, 1998.
Riechers, B., Wittbracht, F., Hutten, A., and Koop, T.: The homogeneous ice nucleation rate of water droplets produced in a microfluidic device and the role of temperature uncertainty, Phys. Chem. Chem. Phys., 15, 5873–5887, 2013.Rosenfeld, D. and Woodley, W. L.: Deep convective clouds with sustained supercooled liquid water down to -37.5 ∘C, Nature, 405, 440–442, 2000.
Sackmann, E. K., Fulton, A. L., and Beebe, D. J.: The present and future role of microfluidics in biomedical research, Nature, 507, 181–189, 2014.
Schmitz, C. H. J., Rowat, A. C., Koster, S., and Weitz, D. A.: Dropspots: a picoliter array in a microfluidic device, Lab Chip, 9, 44–49, 2009.Schnaiter, M., Järvinen, E., Vochezer, P., Abdelmonem, A., Wagner, R., Jourdan, O., Mioche, G., Shcherbakov, V. N., Schmitt, C. G., Tricoli, U., Ulanowski, Z., and Heymsfield, A. J.: Cloud chamber experiments on the origin of ice crystal complexity in cirrus clouds, Atmos. Chem. Phys., 16, 5091–5110, 10.5194/acp-16-5091-2016, 2016.
Stan, C. A., Schneider, G. F., Shevkoplyas, S. S., Hashimoto, M., Ibanescu, M., Wiley, B. J., and Whitesides, G. M.: A microfluidic apparatus for the study of ice nucleation in supercooled water drops, Lab Chip, 9, 2293–2305, 2009.Stöckel, P., Weidinger, I. M., Baumgärtel, H., and Leisner, T.: Rates of homogeneous ice nucleation in levitated H2O and D2O droplets, J. Phys. Chem. A, 109, 2540–2546, 2005.
Ullrich, R., Hoose, C., Möhler, O., Niemand, M., Wagner, R., Höhler, K., Hiranuma, N., Saathoff, H., and Leisner, T.: A new ice nucleation active site parameterization for desert dust and soot, J. Atmos. Sci., 74, 699–717, 2017.
Vali, G.: Quantitative evaluation of experimental results an the heterogeneous freezing nucleation of supercooled liquids, J. Atmos. Sci., 28, 402–409, 1971.Vali, G., DeMott, P. J., Möhler, O., and Whale, T. F.: Technical Note: A proposal for ice nucleation terminology, Atmos. Chem. Phys., 15, 10263–10270, 10.5194/acp-15-10263-2015, 2015.Whale, T. F., Murray, B. J., O'Sullivan, D., Wilson, T. W., Umo, N. S., Baustian, K. J., Atkinson, J. D., Workneh, D. A., and Morris, G. J.: A technique for quantifying heterogeneous ice nucleation in microlitre supercooled water droplets, Atmos. Meas. Tech., 8, 2437–2447, 10.5194/amt-8-2437-2015, 2015.
Whitesides, G. M.: The origins and the future of microfluidics, Nature, 442, 368–373, 2006.
Zobrist, B., Marcolli, C., Peter, T., and Koop, T.: Heterogeneous ice
nucleation in aqueous solutions: the role of water activity,
J. Phys. Chem. A, 112, 3965–3975, 2008.