Mineral dust particles from deserts are amongst the most common ice nucleating particles in the atmosphere. The mineralogy of desert dust differs depending on the source region and can further fractionate during the dust emission processes. Mineralogy to a large extent explains the ice nucleation behavior of desert aerosol, but not entirely. Apart from pure mineral dust, desert aerosol particles often exhibit a coating or are mixed with small amounts of biological material. Aging on the ground or during atmospheric transport can deactivate nucleation sites, thus strong ice nucleating minerals may not exhibit their full potential. In the partner paper of this work, it was shown that mineralogy determines most but not all of the ice nucleation behavior in the immersion mode found for desert dust. In this study, the influence of semi-volatile organic compounds and the presence of crystal water on the ice nucleation behavior of desert aerosol is investigated. This work focuses on the deposition and condensation ice nucleation modes at temperatures between 238 and 242 K of 18 dust samples sourced from nine deserts worldwide. Chemical imaging of the particles' surface is used to determine the cause of the observed differences in ice nucleation. It is found that, while the ice nucleation ability of the majority of the dust samples is dominated by their quartz and feldspar content, in one carbonaceous sample it is mostly caused by organic matter, potentially cellulose and/or proteins. In contrast, the ice nucleation ability of an airborne Saharan sample is found to be diminished, likely by semi-volatile species covering ice nucleation active sites of the minerals. This study shows that in addition to mineralogy, other factors such as organics and crystal water content can alter the ice nucleation behavior of desert aerosol during atmospheric transport in various ways.
The ice phase in clouds causes one of the largest uncertainties for
understanding the role of clouds in the present climate and for projecting
future climate
Mineral dust is thought to have an influence on cloud microphysical processes
on a global scale, with global dust emission rate estimates of up to
5 Pg yr
For several decades, clay minerals were believed to be responsible for the
ice activity of mineral dust, mainly due to their high mass fraction in
airborne dust. Recently, K-feldspars have been identified to nucleate ice at
warmer temperatures or lower relative humidity than all other minerals, both
in the immersion mode
Atmospheric aging processes are challenging to observe in situ, thus several
laboratory studies have mimicked potential aging processes. These processes
often modify the surface of dust particles and, as such, the ice nucleation
ability of mineral dust.
Residues from ice nucleating biological material such as fungal proteins or
nanoscale pollenaceous INPs have been observed to adsorb to mineral dust
while retaining their ice nucleation ability
In a partner paper to this work
The current paper focuses on the ice nucleation behavior at 238–242 K of
airborne and surface-collected dust samples. We investigate ice nucleation at
a constant temperature while RH is increased from ice saturation to above
water saturation. While the partner paper
In this part of the series we present ice nucleation measurements of 18 dust
samples. Seven airborne samples were collected after advection from the Sahara.
Four of the airborne samples were collected directly from the air in August 2013
and 2014 at the Izaña observatory in Tenerife, Spain, using a custom-made
large cyclone (Advanced Cyclone Systems, S.A.: flow rate of
200 m
To investigate if the ice nucleation activity of the samples is influenced by
biological particles internally or externally mixed with the dust or by
organic coatings on the dust particle surface, selected samples were heated
to 300
Dust particles were dry dispersed using a rotating brush generator (RBG,
Palas, model RBG 1000) with
The particle size distribution of all samples was measured with a scanning
mobility particle sizer (SMPS; TSI; DMA model 3081, CPC model 3010) for
mobility diameters (
Mineralogical composition in wt % of airborne Saharan dust
samples. Crete, Egypt, Tenerife2013 (Tenerife in Part 1), and Peloponnese as
in
The quantitative mineralogical composition of the bulk dust samples was
investigated with the X-ray diffraction (XRD) Rietveld method
Thermogravimetric analysis (TGA) of six of the dust samples was conducted by
gradually heating the dust samples from 40 to 300
The morphology of one sample was investigated using scanning electron microscopy (SEM; FEI Quanta 250 FEG, ThermoFisher Scientific).
Attenuated total reflection infrared (ATR-IR) spectroscopy was carried out on
an FTIR (Bruker Vertex 80v) equipped with an ATR cell (Pike GladiATR, diamond
ATR crystal). The beam path of the spectrometer and the optical parts of the
ATR cell are under a vacuum (1.65 mbar) to minimize the influence of water
vapor and
Ice nucleation experiments were conducted with the portable ice nucleation
chamber
In the deterministic concept
The ice nucleation activity of the Australia, Atacama milled, Etosha,
Tenerife2014_1, Peloponnese, and Morocco samples was additionally measured
after they had been exposed to 300
Ice-active surface site density at
Natural log of the ice-active surface site density as a function of the sum of quartz and feldspar content of the samples. Square symbols indicate surface-collected samples, stars indicate milled samples, and circles indicate airborne samples. For clarity, the Tenerife samples are not named individually and are instead shown as open circles. The asterisk in the legend indicates that the correlation is significant at the 0.05 level.
Ice-active surface site density was determined for 18 dust samples, of which
four are from the Sahara and were collected after atmospheric transport at the
Izaña observatory in Tenerife and three were from after atmospheric transport in the
Peloponnese, in Crete, and the Sinai Peninsula in Egypt. Figure
Overview of the Pearson correlation coefficients of the sum of
selected mineral fractions and
Ice-active surface site density at
To determine how well the dust mineralogy can overall predict the ice
nucleation activity, the correlation of
Relative mass loss
In this section, the role of heat labile material on the surface of dust
particles is investigated. A representative subset of the samples was
selected to reduce the number of experiments necessary. The Australia and
Morocco sample were selected because of their exceptional high
We investigated these possible implications further by thermogravimetric
analysis, ATR-IR, and Raman spectroscopy of the samples. The relative mass
loss under increasing temperature is shown in Fig.
The Morocco, Australia, and Atacama milled samples, which all showed no change
in
The temperature ranges where the mass loss occurs can be related to different
materials which were evaporated and potentially altered the ice nucleation
behavior. The first decrease in mass at 40 to 110
Attenuated total reflectance infrared (ATR-IR) spectra for the
We performed ATR-IR spectroscopy on the Etosha and Tenerife2014_1 to
investigate the nature of the material responsible for the respective
decrease and increase in
Raman mapping results for the Australia
Raman mapping was performed on the Etosha, Tenerife2014_1, and Australia
samples. Due to strong fluorescence, however, the Tenerife2014_1 spectra did
not yield any information and are thus not presented here. The Raman maps for
the Etosha and Australia samples are shown in Fig.
Identification of the material which was released or decomposed during the
heating was hampered by fluorescence inherent to the minerals in the samples
and also possibly due to biological material if present in the unheated
samples. The ratio of signal to noise (fluorescence) was optimized by impacting
small amounts of the samples on a pure aluminum surface and by adjusting the
laser power, but the fluorescence could not be entirely suppressed. This,
together with the complexity of the samples, inhibited an unambiguous
identification of the species which were altered by the heating and may
affect the ice nucleation ability. We suggest three possible candidates for
the cluster with a strong Raman signal at 3180 cm
For the Etosha sample an effect of organic or other heat labile material on
the ice nucleation behavior appears likely. The main minerals contained in
the Etosha sample (i.e., ankerite, calcite, dolomite, and muscovite) are not
known to be particularly ice nucleation active at the investigated
temperatures. In case of ankerite the ice nucleation ability is unknown.
Based on its similarity with dolomite, a carbonate known not to be ice
nucleation active, it is assumed that ankerite is also not active. Thus, one
of the suggested candidates with the strong Raman signal at 3180 cm
In contrast to the Etosha sample, the Tenerife2014_1 sample consists of a
number of minerals ice nucleation active at the studied temperatures, e.g., orthoclase, plagioclase, and quartz (Table
XRD diffractogram of unheated and heated Tenerife2014_1 sample. Vertical text indicates changes in peak height which were used to identify the decrease in gypsum and increase in anhydrite with heating. Horizontal text indicates peaks associated with other minerals in the sample.
Another explanation for the reduction in mass at temperatures below
300
XRD analysis of the unheated and heated Tenerife2014_1 sample show the
conversion of gypsum to anhydrite (Fig.
Scanning electron microscopy images of the
In this study we showed that the fractions of quartz and the sum of quartz
and feldspars in desert dust samples correlate better than all other mineral
fractions with the ice nucleation active surface site density of the dust in
deposition and the condensation mode at temperatures between 238 and 242 K. This
is in line with the observations for the immersion mode presented in Part 1
Ice nucleation data from this study are available from
The supplement related to this article is available online at:
YB collected the Tenerife and Israel samples, conceived and lead the measurement campaign, performed the ice nucleation measurements and analysis, performed and analyzed the XRD measurements, analyzed the TGA measurements, and wrote the paper. PB performed and analyzed the SEM, ATR-IR, and Raman measurements, and contributed to the paper. JO contributed to the Raman measurements. HG supervised the ATR-IR and Raman measurements and analysis and contributed to the paper. MP performed and analyzed the XRD measurements. ZAK performed the TGA measurements. BS, ZAK, and UL supervised the project and contributed to the paper. All authors contributed to the interpretation of data.
The authors declare no competing interests.
We thank the two anonymous reviewers for their helpful comments. The various dust samples in this paper have been collected by a number of people who the authors are very thankful to Maria Kanakidou and her team (Peloponnese, Crete); Felix Lüönd (Atacama); Paolo D'Odorico and Christopher Hoyle (Etosha); Lukas Kaufmann, Konrad Kandler, and Lother Schütz (Taklamakan); André Welti (Australia, Mojave); Monika Kohn (Dubai); Joel Corbin (Morocco); Sergio Rodríguez (Tenerife); and Hamza Mohamed Hamza (Egypt). The authors would like to thank Joanna Wong for her assistance with the TGA measurements and the Laboratory of Composite Materials and Adaptive Structures at ETH Zurich for the use of their thermal analysis equipment. We thank Hannes Wydler for his technical support with the PINC. Philipp Baloh would like to thank Karin Wieland and Rita Wiesinger for helpful discussions concerning the interpretation of the Raman spectra. Hinrich Grothe and Philipp Baloh would like to thank the Analytical Instrumentation Center and the X-Ray Center at TU Wien for the use of the Raman and XRD equipment and the University Service Centre for Transmission Electron Microscopy at TU Wien for recording the SEM images in this work. Yvonne Boose and Zamin A. Kanji gratefully acknowledge support by the Swiss National Science Foundation (grant 200020 150169/1). The research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007-797 2013) under grant agreement no. 603445 (BACCHUS). Hinrich Grothe and Philipp Baloh gratefully acknowledge support by the FFG (Austrian Research Promotion Agency) for funding under project no. 850689. Edited by: Eliza Harris Reviewed by: two anonymous referees