Water uptake of subpollen aerosol particles: hygroscopic growth, cloud condensation nuclei activation, and liquid–liquid phase separation

Abstract. Pollen grains emitted from vegetation can release subpollen particles (SPPs)
that contribute to the fine fraction of atmospheric aerosols and may act as
cloud condensation nuclei (CCN), ice nuclei (IN), or aeroallergens. Here, we
investigate and characterize the hygroscopic growth and CCN activation of
birch, pine, and rapeseed SPPs. A high-humidity tandem differential mobility
analyzer (HHTDMA) was used to measure particle restructuring and water
uptake over a wide range of relative humidity (RH) from 2 % to 99.5 %,
and a continuous flow CCN counter was used for size-resolved measurements of
CCN activation at supersaturations (S) in the range of 0.2 % to 1.2 %.
For both subsaturated and supersaturated conditions, effective
hygroscopicity parameters, κ, were obtained by Köhler model
calculations. Gravimetric and chemical analyses, electron microscopy, and
dynamic light scattering measurements were performed to characterize further
properties of SPPs from aqueous pollen extracts such as chemical composition
(starch, proteins, DNA, and inorganic ions) and the hydrodynamic size
distribution of water-insoluble material. All investigated SPP samples
exhibited a sharp increase of water uptake and κ above
∼95 % RH, suggesting a liquid–liquid phase separation
(LLPS). The HHTDMA measurements at RH >95 % enable closure
between the CCN activation at water vapor supersaturation and hygroscopic
growth at subsaturated conditions, which is often not achieved when hygroscopicity tandem differential mobility analyzer (HTDMA) measurements are performed at lower RH where the water uptake and effective
hygroscopicity may be limited by the effects of LLPS. Such effects may be
important not only for closure between hygroscopic growth and CCN activation
but also for the chemical reactivity, allergenic potential, and related
health effects of SPPs.


. Filtration protocol of the pollen aqueous solution. Three series of measurements performed 2 for each type of pollen. MR is the ratio of the penetrated mass to the initial mass of pollen.  Figure S2 shows a sketch of the HHTDMA setup. Both DMAs thermally insulated and operated with a closed loop sheath air setup. The sheath and aerosol flow rates in both DMAs were 3.0 and 8 0.3 l min -1 , respectively. To control the RH we used a dew point probe (Dew Master, Edgetech Instrument, remote D-probe SC) and capacitive sensors (Almemo, FHAD 46C41A) in the range of . Experimental setup of the high humidity tandem differential mobility analyzer (HHTDMA) system: MD-700 -NAFION dryer, SDDsilica gel diffusion dryer, RHrelative humidity sensor, NL -85Kr aerosol neutralizer, DMAdifferential mobility analyzer, NCA -Nafion conditioner with air, NCW -Nafion conditioner with water, CPCcondensation particle counter, Operation mode: A-hydration&dehydration (H&D), Bhydration, Cdehydration.
2-80 % RH and the ammonium sulfate scans at RH above 80 %. Based on the Extended Aerosol Inorganics Model (E-AIM, model II) (Clegg et al., 1998;Wexler and Clegg, 2002), we converted the 2 measured ammonium sulfate growth factors into RH( , − ). The residence time between the aerosol preconditioning system and DMA2 depends on the humidification mode; its minimum value 4 is 6.5 s, which corresponds to RT in the hydration operation mode (Fig.S2). The algorithm used for calculating the uncertainty of RH and growth factors discussed in detail elsewhere (Mikhailov and 6 Vlasenko, 2020). Figure S3 illustrates these uncertainties for the case of subpollen particles.

Fig. S3.
Accuracy in RH using different methods (a) and relative growth factor uncertainty due to instrumental and RH errors. The size distribution obtained by DLS is based on the scattering intensity of the particles. For the case of Rayleigh particles, the scattering intensity is proportional to the sixth power of the diameter.
12 Table S2. Sequence of relative humidity ("RH history") experienced by the investigated aerosol particles in the key elements of the HHTDMA system (DMA1, conditioner, DMA2) during different types of HHTDMA experiments (modes of operation). For each type of experiment, X represents the independent variable, i.e., the RH value taken for plotting and further analysis of the measurement results.
HHTDMA experiment (operation mode) Hydration and dehydration (H&D) min is the relative humidity that corresponds to the Db.H&,Dmin obtained in H&D experiment.
Thus, in term of intensity of light scattering the relative contribution, , . from each particle size bin is (Finsy, 1994;Li et al., 2014) 2 where is the number of particles in size bin i having the mid-point diameter . To convert the intensity-based size distribution into number particles size distribution ( , ) we let 4 As an example Fig. S4 shows the result of converting intensity-based size distribution of birch pollen colloids to a number-weighted distribution. Combining Eq. (S1) and Eq. (S2), we obtain 6 normalized number particles size distribution: If the suspended particles all have density  then their mass 0 with respect to the total number 0 , 8 is Total mass of suspended species in the filtered solution ( 0 ) was determined as the difference 10 between the total mass of the solids in the filtered solution and the mass of dissolved species (Table   1), therefore 0 can be calculated from Eq. (S4). If 0 and , are known the number of colloids 12 in each size bin is = , 0 . A material density of 1.4 g cm -3 was used for suspended organic particles suggesting that starch (1.53 g cm -3 ), membrane proteins (1.37 g cm -3 ), cellulose (1.5 g 14 cm -3 ), carotenoids (~1 g cm -3 ) and lipids (1.0 -1.2 g cm -3 ) (Haynes, 2011) are the main species in the series of water-insoluble pollen compounds (Stanley and Linskens, 1974). It should be noted that the DLS-based size distributions in some cases has a high degree of 2 uncertainty. In the DLS setup, the light scattered by fluctuations of the concentration of molecules, particles, or aggregates suspended in a tested solution is recorded. To determine the rates of decay 4 of the intensity of the scattered light, the time correlation function, ( ) of this intensity is analyzed.
Since the correlation function is a superposition of exponential decays with distributed decay rates, 6 the distribution function is the inverse Laplace transform of the correlation function. The CONTIN algorithm (Provencher, 1982;Scotti et al., 2015) carries out this transformation numerically. One 8 of the limitations of the resolution comes from the extremely ill-conditioned nature of this Laplace inversion. Practically very small differences in ( ) within typical experimental accuracy may 10 result in quite different particle size distributions after inversion (Finsy, 1994;Anderson et al., 2013;Varenne et al., 2016). Both DLS data uncertainty and inversion algorithm together with 12 approximations used in Eq. (S4) will provide an error in size distribution.

14
Hygroscopicity parameter  can be determined from an approximate formula (Petters and Kreidenweis, 2007): particles coagulation is negligible. However, it became essential after 10 hours. In the time scale of 20-50 h, the intensity of Reyleigh scattering (symmetrical phase function) progressively increasing. ⁄ as a function of initial mobility diameter (Db.i) (Fig. 4) were fitted 12 by exponential curve (Eq.10). The best fit parameters are given in Table S3.
14 The irregular shape morphology of the rapeseed SPP (Fig. 5c) approximated by a Ferret ellipsoid with 18 maximal (a) and minimal (b) axis, respectively. The envelope shape factor,  calculated by assuming that in both DMA the particles are oriented to its flow by maximal axis (prolate ellipsoid) (Fuchs, where = / is the aspect ratio. The calculated values of  are listed in Table S4. Table S4. Envelope shape parameter ( ) of rapeseed SPP as a function of size (b) and ellipsoid aspect 2 ratio () estimated from SEM images using ImageJ processing software. Twenty images of each size range were used to calculate the average ± standard deviation aspect ratio. Mikhailov, E. F. and Vlasenko, S. S.: High-humidity tandem differential mobility analyzer for accurate determination of aerosol hygroscopic growth, microstructure, and activity coefficients 34 over a wide range of relative humidity, Atmos. Meas. Tech., 13, 2035-2056, https://doi.org/10.5194/amt-13-2035-2020, 2020.  ( 2 − 1)