Ice nucleation on surrogates of boreal forest SOA particles: effect of water content and oxidative age

. α -pinene is an abundant volatile organic compound (VOC) emitted by boreal forests and a source of atmospheric Secondary Organic Aerosol (SOA). This precursor is commonly used as a model compound for SOA studies representing boreal forest emissions. α -pinene SOA particles can have a highly viscous solid or semi-solid phase state depending on water content, temperature and oxidative age. The phase state (or viscosity) of SOA particles has multiple effects on the chemical and physical processes in which SOA particles are involved; one of the affected processes is ice formation. 5 We investigate the effect of water content and oxidative age on ice nucleation using 100 nm quasi-monodisperse particles of boreal forest SOA surrogates . Ice nucleation experiments are conducted in the temperature range between 210 and 240 K and from ice to water saturation using the Spectrometer for Ice Nuclei (SPIN). The effect of the particle water content on the ice nucleation process is tested by preconditioning α -pinene SOA at different humidity (40%, 10% and <1% RH W ). The inﬂuence of the particle oxidative age is tested by varying their O:C ratio (oxygen-to-carbon ratios, O:C ∼ 0.45, 0.8, 1.1). To assess 10 the suitability of α -pinene as a model compound to study the ice nucleation properties of boreal forest SOA and to conﬁrm the atmospheric relevance of our ﬁndings, we compare them to measurements of SOA using pine-needle oil or Scots pine tree emissions as precursors. ice nucleation measurements show that surrogates of boreal forest SOA particles promote only homogeneous ice formation. An effect of preconditioning humidity on homogeneous ice nucleation could be observed. Contrary to the expected behavior, 15 homogeneous freezing is suppressed for SOA particles with high water content (preconditioned at 40% RH W ) and was only observed for SOA preconditioned at low RH W ( ≤ 10%). No dependence of homogeneous freezing on the SOA oxidative age was observed. The results can be explained by a signiﬁcant change of particulate water diffusivity as a function of humidity (from 10% to 40% RH W ) at 293 K , where the aerosol is preconditioned. On dry SOA particles, water diffusion into the particle is slow enough to form a core-shell morphology with an outer layer that can equilibrate within the timescale of the 20 experiment and freeze homogeneously. On SOA particles with higher water content, water diffuses faster into the particle, delaying equilibration at the particle surface and preventing the formation of a diluted shell, which can delay homogeneous freezing. To predict if a core-shell develops, we propose that the partial water vapor pressure particles are exposed to prior to an experiment can serve as an indicator.


Introduction
Heterogeneous ice formation by ice nucleating particles (INP) allows the formation of cirrus clouds at lower humidity than required for ice formation by homogeneous freezing of solution droplets, which is determined by the water activity criterion (Koop et al., 2000). Typical cirrus cloud INPs are water insoluble particles such as mineral dusts, fly ash, metallic particles (DeMott et al., 2003) or soots (Bond et al., 2013). Prompted by the realization that secondary organic aerosol (SOA) particles 5 can exist in a highly viscous, (semi-) solid state (Zobrist et al., 2008;Virtanen et al., 2010), the possibility that SOA particles could act as INPs has been investigated in the recent years (Wang et al., 2012;Schill et al., 2014;Ignatius et al., 2016;Ladino et al., 2014;Möhler et al., 2008;Prenni et al., 2009;Charnawskas et al., 2017;Wagner et al., 2017).
SOA particles are composed of oxidation products of volatile organic compounds (VOC), some of which are water soluble.
In contrast to low viscosity (more liquid-like) particles, highly viscous aerosol are slow to take up or lose water or other vapors 10 in response to variations in gas-phase composition (Mikhailov et al., 2009;Koop et al., 2011;Shiraiwa et al., 2013;Yli-Juuti et al., 2017). This resistance is particularly pronounced under dry conditions or low temperatures where SOA particles can exist in a highly viscous, or glassy state (Koop et al., 2011;Virtanen et al., 2010;Zobrist et al., 2008).
Upon updraft-driven humidification in the atmosphere, (semi-) solid amorphous organic aerosol could act as INP in several ice nucleation mechanisms. SOA particles that remain glassy during humidification could trigger ice formation via deposition 15 nucleation (Murray et al., 2010). Upon humidification beyond the amorphous deliquescence relative humidity, water diffusion might be slow enough to form core-shell morphologies in particles with a highly viscous glassy matrix, allowing ice formation via immersion freezing (Berkemeier et al., 2014;Lienhard et al., 2015). This situation could take place during fast updrafts (> 3 m s −1 ) in which the equilibration time with the surrounding humidity is limited (Lienhard et al., 2015;Price et al., 2015).
Typical equilibration times for 100 nm α-pinene SOA particles range from hours at 220 K to minutes at 230 K (Price et al., 20 2015). Continued humidification of the organic particles leads to complete liquefaction and homogeneous freezing as the only possible ice formation mechanism (Koop et al., 2011).
Previous studies found that the ice nucleation (IN) ability of SOA particles varies between species: naphthalene-derived (Wang et al., 2012), and methylglyoxal with methylamine (Schill et al., 2014) particles were classified as effective, heterogeneous INP, exhibiting ice formation onsets at humidities clearly below homogeneous freezing conditions. α-pinene SOA 25 was indicated to be ineffective at nucleating ice heterogeneously (Möhler et al., 2008;Ladino et al., 2014;Prenni et al., 2009;Charnawskas et al., 2017;Wagner et al., 2017) while Ignatius et al. (2016) found α-pinene SOA particles to be effective INPs in the deposition mode. Although the rapid cooling of the aerosol during most experiments could result in the development of a core-shell morphology, none of the above studies observed ice formation by immersion freezing. Using a cloud particle model capable of simulating water diffusion in individual aerosol particles, Fowler et al. (2020) have recently shown that at 30 temperatures between 200 and 220 K a water layer condenses on α-pinene SOA particles, which can freeze homogeneously.
In accordance with laboratory studies showing only homogeneous ice formation of α-pinene SOA particles, the same model indicates that at higher temperatures, water diffuses into particles, and freezing only takes place at humidities high enough to enable the freezing of solution droplets. Mie resonance measurements in levitating, highly viscous aerosol particles show 2 https://doi.org/10.5194/acp-2021-10 Preprint. Discussion started: 18 February 2021 c Author(s) 2021. CC BY 4.0 License. steep gradients in water content between the core and outer layer (Bastelberger et al., 2018) supporting the interpretation that viscous SOA can adopt a core-shell structure, where the outer layer can freeze homogeneously. We present experimental results of ice formation of atmospherically relevant surrogates of SOA particles, generated from biogenic emissions. To test the effect of water content and core-shell formation, α-pinene SOA particles with different diffusivity are generated by varying the oxidative ageing and exposure of particles to different levels of humidity prior to cooling. The atmospheric relevance of the ice 5 nucleation results from the α-pinene SOA is confirmed by comparing to experiments with SOA particles formed from natural, boreal forest VOC.

Methods
The SOA Ice Nucleation Experiment (SINE) campaign was carried out at the University of Eastern Finland Aerosol Physics Laboratory in June -July 2019. The experiments focused on studying the IN efficiency of SOA formed from boreal forest 10 emissions and the influence of particle water content and oxidation level on IN activity.

General experimental setup
A schematic of the experimental setup is depicted in Figure 1. SOA was generated either in an atmospheric simulation chamber  For the ice nucleation measurements with a spectrometer for ice nuclei (SPIN; Droplet Measurement Technologies) the aerosol was size-selected using a differential mobility analyzer (DMA; TSI, model 3082) and preconditioned at 40%, 10% or <1% RH W .

Aerosol generation
2.2.1 α-pinene SOA with different O:C generated with the PAM reactor 25 We used a PAM reactor to form α-pinene SOA particles of different oxidative ages, characterized by their O:C ratio. A syringe pump (Nexus 3000, Chemyx Inc.) was used to create a constant injection of α-pinene into a nitrogen carrier gas flow heated to 60°C, which was introduced into the PAM reactor. In the PAM reactor, particles are formed by a) dry ozonolysis, b) wet ozonolysis, or c) wet photooxidation with OH radicals. The O 3 concentration, and in the photooxidation experiments the irradiation level, were adjusted to create SOA particles with a specific O:C ratio (see Tab. 1 for settings). Low-O:C SOA

Boreal forest surrogate SOA from ASC
A second set of experiments was conducted utilizing the 12 m 3 collapsible PTFE ASC to form SOA under atmospheric conditions (see Fig. 1). Aqueous ammonium sulfate particles were generated with an atomizer (Aerosol Generator Model 3076, TSI), subsequently dried below their efflorescence point in a silica gel dryer and injected into the ASC to reach seed concentrations of 5.0 -7.5e4 cm −3 at the start of VOC ozonolysis. For experiments #13 and #15, α-pinene or pine needle oil were 15 added to the chamber from diffusion sources. The injection times were adjusted to reach the desired 5 -50 ppb in the ASC. The emissions from six 10 year-old pine saplings had been collected onto stainless steel multibed adsorbent cartridges containing

Ice nucleation measurements
The ice nucleation ability of 100 nm SOA particles was measured with the SPIN instrument (Garimella et al., 2016). SPIN is a parallel plate, continuous flow diffusion chamber (CFDC). The specific instrument (SPIN5) used in this study has been modified to perform low temperature experiments (Welti et al., 2020). Inside the SPIN chamber, particles are exposed to RHand T-conditions (temperature range 210 -240 K and from ice to water saturation) relevant for ice formation in cirrus clouds. Ice 5 formation during a residence time of 10 s is detected with an optical particle counter (OPC). The ratio of ice forming particles, measured with the OPC and the number of particles introduced into SPIN, measured with a condensation particle counter (CPC; Airmodus A20), is reported as activated fraction (AF). The evaporation section at the end of the chamber guarantees that droplets are not counted as ice with the OPC. In this study, the size-selected SOA particles were exposed to different humidity conditions before entering the SPIN, to probe the effects of particle water content on IN activity. The samples were 10 introduced a) directly into the SPIN, b) through a silica gel diffusion dryer, or c) sequentially through a silica gel diffusion dryer and a liquid nitrogen cold trap (see Fig. 1). The preconditioning led to sample RH W at the SPIN inlet (Vaisala HMP110 humidity sensor) of 40% , 10%, or <1%, respectively. Later, we refer to these precondition settings as wet, dry and super dry, respectively.

Effect of O:C ratio and water content on ice formation conditions
The effects of oxidative age (approximated by the O:C ratio) and water content of α-pinene SOA particles on ice formation were investigated. Note, that the water content of the particles was preconditioned after generation and before measuring their ice nucleation ability with the SPIN. The AF-spectra of α-pinene SOA particles generated with different O:C and preconditioned at different humidity are shown in Fig. 2.

20
The O:C ratio appears to have a minor impact on the conditions of ice formation of the α-pinene SOA particles. Schill and Tolbert (2012) reported that the O:C ratio of organic acids affects heterogeneous ice nucleation and suggested that this could also be the case for α-pinene SOA. Aged particles (higher O:C ratio) would nucleate ice at lower humidity due to higher surface hydrophilicity, allowing the adsorption of an ice-like layer from which the ice phase can develop. The decrease in ice formation onset humidity was strongest for 0.3 < O:C < 0.5. In the range of O:C ratios covered here (0.45-1.1) no pronounced change in 25 ice formation humidity was observed. Fig. 3   column in Fig. 2) almost completely suppressed the formation of ice up to water saturation. SOA particles preconditioned at higher RH W contain more water, which acts as a plasticizer, and they are therefore less viscous. It was expected that the higher water content of the particles would facilitate the diffusion of condensed water molecules, enabling the homogeneous freezing of SOA solution droplets. To our knowledge, a retardation of homogeneous freezing of SOA particles with higher water content has not been observed experimentally and only recently been suggested by Fowler et al. (2020), based on simulations. Fig. 4.

5
illustrates the mechanism how water content could impact homogeneous freezing of SOA particles during SPIN measurements.
When dry SOA particles (Fig. 4a) are exposed to the elevated humidity inside the SPIN, water diffusion towards the particle core is slow enough to create a core-shell structure with a steep gradient in water content. The liquid outer layer equilibrates   For SOA particles with higher water content (Fig. 4b), water diffuses faster into the particle, leading to a core-shell morphology with a more concentrated outer layer. The outer shell of the particle does not reach equilibrium and inhibits homogeneous freezing. The mechanism depicted in Fig. 4 suggests that slower diffusion enables the outer shell to equilibrate faster with the 5 gas phase. For a constant amount of water uptake (within 10 s residence time in SPIN), wet SOA particles incorporate water faster into the particle, resulting in a more concentrated droplet or shell. With longer residence times equilibrium could be reached and homogeneous freezing would theoretically be observed again.  Delayed onsets of homogeneous freezing above the humidity defined by the water activity criterion (Koop et al., 2000) at low temperatures could indicate that particles are not in equilibrium before freezing in SPIN. A similar offset for homogeneous 10 freezing of sulfuric acid particles at low temperature has been observed in the AIDA chamber (Möhler et al., 2003), which was also attributed to diffusion-limited uptake of water vapor during cooling. The systematic delay of homogeneous freezing towards higher humidity, suggests a decreasing, low water diffusion rate of SOA towards low temperature. Fowler et al. (2020) pointed out the same aspect, and proposed even an inhibition of homogeneous ice nucleation below 200 K. Experimental evidence of homogeneous freezing inhibition at such low temperatures had been previously reported by Murray (2008) for We have investigated the influence of particle water content and oxidative age on the ice nucleation ability of boreal forest SOA surrogates, using α-pinene SOA generated in a PAM reactor as a model compound. The suitability of α-pinene SOA as a proxy for the IN ability of more complex boreal forest surrogates was addressed by comparing IN measurements of SOA from two other precursors (pine-needle oil and Scots pine tree VOC) generated in an ASC under more atmospherically relevant 5 conditions. For the first time, the IN ability of pure lab-generated SOA from real pine emission has been measured.
Boreal forest SOA surrogates from all precursors were found to be inefficient ice nucleating particles, in agreement with previous studies (Möhler et al., 2008;Ladino et al., 2014;Wagner et al., 2017;Charnawskas et al., 2017). Our observations indicate that the IN ability of α-pinene SOA can be considered representative of more complex SOA produced in monoterpenedominated precursor mixes from boreal environments.

10
Homogeneous ice nucleation was observed for α-pinene SOA preconditioned at low RH W (≤ 10%) and contrary to the expected behavior, homogeneous freezing was suppressed for SOA with higher water content (40% RH W ). The onset humidity for homogeneous freezing did not depend on the oxidative age. The experiments indicate that SOA water content was the main variable controlling the onset humidity for homogeneous freezing at a certain temperature and point to a dependence on the water diffusion rate.

15
α-pinene SOA with low water content, presumably in a highly viscous state, did not act as INP, but a decreased water diffusion rate into the particle allowed the formation of a core-shell morphology, enabling the homogeneous freezing of the diluted outer layer. In contrast, SOA with higher water content, into which water diffused more efficiently during the residence time in SPIN, completely liquefied or developed a core-shell morphology in which the liquid phase was highly concentrated, inhibiting homogeneous freezing. The coincidence of SOA phase transition conditions with partial water vapor pressures 20 between 3-5 hPa also point at the potential dominant role of water content in particle properties like viscosity or diffusivity that affected the SOA ice nucleation behavior as measured by SPIN.
Further investigations should include SOA generation under a wider range of RH and T conditions and measurements of initial viscosity for a clearer connection between the ice nucleation measurements and the freezing pathways leading to those observations.

25
Data availability. Data sets are available from the authors upon request.
Author contributions. AV and AL proposed the initial research question. AAP and AW conducted the ice nucleation experiments with contributions from KK. AB, IP and IS were in charge of particle formation, characterization, and the corresponding data analysis. AAP and AW prepared the manuscript with contributions from AB, AL, and AV. All authors commented the manuscript. AV and AL acquired funding and supervised the project.