Ice nucleation activities of soot particles internally mixed with sulphuric acid at cirrus cloud conditions

Soot particles are important candidates for ice nucleating particles (INPs) in cirrus cloud formation which is known 10 to exert a warming effect on climate. Bare soot particles, generally hydrophobic and fractal, mainly exist near emission sources. Coated or internally mixed soot particles are more abundant in the atmosphere and have a higher probability to impact cloud formation and climate. However, the ice nucleation ability of coated soot particles is not as well understood as that of freshly produced soot particles. In this study, two samples, a propane (C3H8) flame soot and a commercial carbon black were coated with varying wt % of sulphuric acid (H2SO4). The ratio of coating material mass to the mass of bare soot particle was controlled 15 and progressively increased from less than 5 wt % to over 100 wt %. Both bare and coated soot particle ice nucleation activities were investigated with a continuous flow diffusion chamber operated at mixed-phase and cirrus cloud conditions. The mobility size and mass distribution of size selected soot particles with/without H2SO4 coating were measured by a scanning mobility particle sizer (SMPS) and a centrifugal particle mass analyser (CPMA) running in parallel. The mixing state and morphology of soot particles were characterized by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). In 20 addition, the evidence for the presence of H2SO4 on coated soot particle surface is shown by Energy Dispersive X-ray spectroscopy (EDX). Our study demonstrates that H2SO4 coatings suppress the ice nucleation activity of soot particles to varying degrees depending on the coating thickness, but in a non-linear fashion. Thin coatings causing pore filling in the sootaggregate inhibits pore condensation and freezing (PCF). Thick coatings promote particle ice activation via droplet homogeneous freezing. Overall, our findings reveal that H2SO4 coatings will suppress soot particle ice nucleation abilities in 25 the cirrus cloud regime, having implications for the fate of soot particles with respect to cloud formation in the upper troposphere. 30 https://doi.org/10.5194/acp-2021-645 Preprint. Discussion started: 6 September 2021 c © Author(s) 2021. CC BY 4.0 License.

requires optimum pore structure, size and appropriate surface wettability of the soot particle. Numerous factors involved in determining the ice nucleation activity of bare soot particles can render the impacts of the aging process on ice nucleation, where soot becomes internally mixed with other aerosols further changing the particle properties, more complicated. Therefore, 100 systematic laboratory experiments to reveal the ice nucleation activity of certain soot samples and soot particles in different coating states are needed.
A foreign coating material changes the chemical composition and hygroscopicity of soot particles significantly (Zhang et al., 2008;China et al., 2015;Zhang et al., 2020). Zhang and Zhang (2005) showed that H2SO4 coating onto soot particles, produced 105 by a soot generator burning methane, hexane or kerosene, is irreversible and enhances its hygroscopicity. Möhler et al. (2005b) reported that polydisperse graphite spark soot particles coated with H2SO4 nucleate ice at an ice saturation level higher than that of bare graphite spark soot particles but lower than the homogeneous freezing RH (RHhom) required by pure H2SO4 droplets, at a given T. However, the soot coating thickness effect on ice nucleation was not systematically compared in the study because H2SO4 coated soot particles were produced with different coating conditions. Kulkarni et al. (2016) observed 110 toluene or α-pinene coated diesel soot particles freezing homogeneously. Dalirian et al. (2018) investigated the coating effect of different organic compounds on the cloud droplet activation ability of soot particles and concluded that even a small amount of soluble coating is able to make hydrophobic soot particles become hydrophilic enough to be active could condensation nuclei (CCN). Therefore, the water interaction behaviour of coated soot particles can be changed significantly by the coating material. Considering its solubility in water and strong hygroscopic ability, H2SO4 coating very likely enhances soot particle 115 surface wettability, i.e. a lower soot-water contact angle (Mahrt et al., 2020b), which plays an important role in PCF (David et al., 2020;Marcolli et al., 2021). In addition, the coating process can alter soot particle morphology and influence the availability of potential pores for PCF at cirrus cloud conditions. It is reported that external material coating changes the initial soot particle size, particle internal microstructure and surface texture (Saathoff et al., 2003;Khalizov et al., 2009;Pei et al., 2018;Zhang et al., 2020). The size growth caused by considerable addition of coating material is a possible coating effect and the coating 120 material distribution may further modify the soot-aggregate structure. For example, oleic acid coating can significantly deform soot particle shape and increase particle mass whereas the removal of the coating material by heating seems to recover its morphology to some extent (Bambha et al., 2013). Schnitzler et al. (2017) also demonstrated that liquid material coating can change soot-aggregate structure by the tension induced by coating condensation on the particle surfaces. It is therefore possible that, if the particle structure can be changed, the pore volume and pore size distribution of the soot-aggregate are also alterable. 125 Considering that soot particle ice nucleation ability via PCF shows dependence on the aforementioned parameters, coating effects on soot particle ice nucleation need systematic studies with a focus on associated morphology and hygroscopicity changes.
In this study, both a propane (C3H8) combustion soot and a porous commercial carbon black are exposed to different H2SO4 130 supersaturation levels to generate coated soot particles. With progressively increasing H2SO4 coatings, size selected soot https://doi.org/10.5194/acp-2021-645 Preprint. Discussion started: 6 September 2021 c Author(s) 2021. CC BY 4.0 License. particle ice nucleation activities are systematically investigated in the mixed-phase and cirrus cloud regimes and compared to bare soot particles. The mixing state of soot particle with H2SO4 is characterized with an emphasis on the morphology and a hypothesis about the coating process of soot particle with H2SO4 and the influence on cirrus ice nucleation is proposed based on the results presented herein. 135 2 Experimental methods

Experimental samples
In this study, two types of soot particles were investigated. First, a C3H8 flame soot is produced by a miniature combustion aerosol standard (miniCAST, model 4200, Jing Ltd., Zollikofen, Switzerland) soot generator. The Jing miniCAST forms a diffusion flame by burning C3H8 under a fuel-lean conditions. The associated soot particle is termed as mCASTblack hereafter. 140 This soot sample is technically the same as the sample used by Mahrt et al. (2018;2020a;2020b), who studied the ice nucleation activities of fresh and aged mCASTblack soot particles in the continuous flow diffusion chamber HINC (Horizontal Ice Nucleation Chamber) (Lacher et al., 2017). At the outlet of the miniCAST, part of the exhaust, containing primary soot particles or small size aggregates, is diluted by filtered synthetic air. The resulting flow containing aerosol is then fed into a 0.125 m 3 continuous stirred tank reactor to generate larger and more homogeneous soot-aggregates. Finally, a 1 L min -1 sample flow is 145 sampled out of the tank and directed to a differential mobility analyser (DMA, classifier 3080, with a 3081 column and a polonium radiation source, TSI Inc.) after passing through a molecular sieve diffusion drier, in order to select monodisperse aerosol sample with mobility size of 200 or 400 nm.
The second type of soot is FW200, a commercial carbon black (Orion Engineered Carbons GmbH,OEC,Frankfurt,Main,150 Germany). FW200 carbon black is a product of incomplete combustion from hydrocarbon liquid fuels. There are two purposes here for choosing this commercial carbon black as one of the experimental samples. Firstly, it is used to represent porous and surface wettable soot particles in the atmosphere because of its large specific surface area value (550 m 2 g -1 provided by the manufacturer; 526 m 2 g -1 reported by Mahrt et al., 2018) and according to surface chemistry information provided by the manufacturer. Secondly, future studies can reproduce ice nucleation experiments with this commercial carbon black more 155 precisely for analysis and comparisons. Mahrt et al. (2020b) used a wet dispersion method to aerosolize soot particles aged in H2SO4 (pH = 4) by a nebulizer with a magnetic stirrer. Instead of using an aqueous suspension, a dry dispersion method is utilized in this study to reduce changes in morphology caused by droplet evaporation after wet suspension. A Venturi nozzle is deployed with a glass jar containing dry soot powder mounted on a strong magnetic stirrer. Soot powder is suspended by the stirrer and then can be entrained into a high pressure N2 flow by the nozzle. The soot aerosol flow generated is directed to 160 a flow cascade, to buffer the aerosol flow and particle number concentration fluctuations. Finally, a 1 L min -1 sample flow is directed to the DMA and 200 or 400 nm size selected soot particles can be generated. Figure 1 shows the experimental setup schematic illustrating aerosol sample generation, sample particle size selection, soot particle H2SO4 coating, ice nucleation experiments and particle characterization measurements. Firstly, the DMA selects 200 165 nm (sheath to sample flow ratio 13 : 1) or 400 nm (sheath to sample flow ratio 7 : 1) soot particles and a 1 L min -1 monodisperse sample flow is produced with a particle number concentration larger than 3,000 cm -3 for the following experiments. The aerosol sample (1 L min -1 ) either goes through a home-built H2SO4 coating apparatus or passes by a dilution system with synthetic air flow. This dilution system has equivalent volume and flow resistance to the coating apparatus, as well as the same dilution ratio (~ 5 : 1). The diluted or H2SO4 coated soot aerosol sample (> 5 L min -1 ) is split into five pathways, distributed to a high 170 efficiency particulate air capsule (HEPA capsule, PALL Corporation) for exhaust flow, a condensation particle counter (CPC;

Coating apparatus
The coating procedure follows the basic idea of heating to produce coating material vapour then cooling to condense or adsorb 185 the coating material onto soot particle surfaces, finally generating H2SO4 coated soot particles, as depicted in Fig. 2. The coating mechanism is attributed to two pathways, including the direct condensation of supersaturated H2SO4 vapor and the adsorption of small H2SO4 particles formed by homogeneous nucleation (Bambha et al., 2013;Pei et al., 2018). The aerosol sample flow used to generate H2SO4 vapor is heated in an aluminium heating block before going through a flask containing 50 ml pure H2SO4 (Sigma-Aldrich, 95.0-97.0 %) mounted on another aluminium heating block. The temperature of the heating 190 blocks is controlled by a LabView (National Instruments Corporation, Austin, Texas, US) program. While passing through the H2SO4 flask, soot particles will be mixed with H2SO4 vapor. Next, a dilution flow at the same temperature as the H2SO4/soot flow entrains the particles out of the flask. Downstream of the apparatus, there is a water-cooling system to generate H2SO4 supersaturation to nucleate or condense H2SO4 onto soot particle surfaces, thus generating H2SO4 internally mixed soot particles. The cooling water temperature is maintained at 20 ℃ by a thermostat (LAUDA E300). By increasing the H2SO4 195 saturation and dilution flow temperature from 30 to 95 ℃, the H2SO4 coating wt % (refer to Sect. 2.2.3 Eq. (1)) in terms of the ratio of H2SO4 coating mass to the bare soot particle mass can be increased monotonically.

200
In order to generate internally mixed soot/H2SO4, the following operation procedure is strictly followed. First, a particle free synthetic airflow is used to flush the apparatus for several minutes at the very beginning of the experiment until the CPC records a zero particle number concentration at the outlet of the coating apparatus. Then, the heated soot aerosol sample flow is connected to the inlet of the apparatus. In order to check the particle mixing state, a SMPS-CPMA combined particle size and mass distribution measurement is conducted at the outlet of the coating apparatus before starting the downstream 205 experiments. The results demonstrate that only internally mixed soot particles with H2SO4 are produced from the coating apparatus, indicating no pure nucleated H2SO4. The SMPS results for 200 nm mCASTblack soot particles coating process are presented in Appendix, Fig. A1 as an example. In the absence of soot, nucleation mode H2SO4 shows a high number concentration and a small size mode of ~ 40 nm fitted by a log-normal distribution function. After feeding soot samples into the coating apparatus, mixed aerosol particle size distribution mode shifts close to the size mode of bare soot particles, 210 meanwhile, the small size mode of H2SO4 particles becomes absent and the number concentration of H2SO4 particles decreases dramatically (see Fig. A1). The CPMA mass scan results also show the homogeneity of the 200 nm mCASTblack soot particles internally mixed with H2SO4, given that the log-normal mass distribution fitting for coated aerosol particles only shows a single distinct peak (see Fig. A2) which indicates the mass mode of the aerosol is for H2SO4 coated particles only. Considering that a 200 nm pure H2SO4 particle has a larger mass than a coated soot particle, we are confident that there is no pure 200 nm 215 H2SO4 particle existing in the aerosol flow.

Ice nucleation experiments
In this study, the HINC chamber (Kanji and Abbatt, 2009;Lacher et al., 2017) is utilized to investigate particle ice nucleation abilities under varying RH conditions at a fixed temperature. HINC is a horizontal continuous flow diffusion chamber in which aerosol particles can experience a known RH and T condition for a variable time (~ 16 s in this study). Two parallel copper 220 plates, sandwiching a polyvinylidene fluoride (PVDF; Angst+Pfister AG, Zurich, Switzerland) frame with a thickness of 20 mm, form the core chamber space. The temperature of each copper plate is controlled by a thermostat (LAUDA, RP890) and monitored by four thermocouples with an uncertainty of ± 0.1 K. Glass fibre paper (Pall Corporation, 66217) adhered on the inner wall of each copper plate is wetted before the experiment and serves as a water vapor reservoir. During the experiment, both walls are coated with a thin ice layer and remain at ice saturation condition (RHi = 100 %). By maintaining a temperature 225 difference (ΔT) between the top and the bottom wall (Ttop > Tbottom), a linear temperature distribution as well as a linear distribution of water vapor pressure develops inside the chamber in the vertical dimension. However, the saturation water vapor pressure has an exponential relation with the temperature. Hence, a nonlinear RH distribution will be generated and the RH value at a fixed vertical position inside the chamber can be determined as well. Upstream of the chamber, there are four ports drawing pure dry N2 into the chamber as the sheath flow confining the aerosol sample flow to the centre of the chamber 230 at a defined RH and T condition. A movable aerosol injector is used to direct the aerosol particles into the centre of the chamber and can also be used to adjust particle residence time in the chamber. At the chamber outlet, an optical particle counter (OPC, MetOne, GT-526S) maintains a total flow rate of 2.83 L min -1 . Before the HINC experiment, the N2 sheath flow rate is adjusted to ensure that 0.22 L min -1 aerosol sample flow makes up part of the 2.83 L min -1 . A 1:12 aerosol sample to sheath flow ratio is used during the whole course the study. During a HINC experiment, the CPC is used to count the total aerosol particle 235 number entering the chamber and the OPC is used to detect the total number of ice crystals or water droplets coming out of the chamber in six size channels (0.3, 1.0, 2.0, 3.0, 4.0 and 5.0 μm). Thus, the ratio of ice crystal or water droplet number to the total aerosol particle number can be derived, which is the aerosol particle activated fraction (AF) value. For bare and coated soot samples, HINC experiments are performed from 218 to 243 K with six measurements at different T 240 and with a 5 K interval. At each T, at least two RH scans are conducted for each soot sample from ice saturation condition to RH conditions above water saturation. The RHi scan rate is 2 % per minute. The RHw values of the aerosol sample upstream of HINC are less than 5 % and monitored by a RH sensor. Soot particle H2SO4 coating wt % for ice nucleation experiments are shown in Table 1.

Particle characterization measurements
SMPS-CPMA: The aerosol particle size distribution is measured by a SMPS system after size selection and/or coating process 250 (see Fig.1). For 200 nm size selected soot particles with/without coating, the CPC 3772 with a flow rate of 1 L min -1 is used and the SMPS measurement can cover a size scanning range from 12.6 to 572.5 nm, which means the double-charged particles selected by the DMA upstream can be detected. For 400 nm soot aerosol samples, the SMPS utilizes a low-flow mode CPC 3776 with a 0.3 L min -1 flow rate and can scan up to a size value of 914 nm, which also covers the 680 nm particles carrying two charges but treated as a 400 nm particle by the upstream DMA. The amount of double-charged particles is approximately 255 16 and 29 % for 200 and 400 nm mCASTblack bare soot particles, respectively (see Fig. A3), and double-charged 200 and 400 nm FW200 bare soot particles amount to 21 % and 26 %, respectively (see Fig. A4). The CPMA is operated in mass scanning mode at a constant speed. A water CPC 3787 running in high-flow mode (1.5 L min -1 ) is used to sample the aerosol through the instrument. The mass scanning range covers the mass value of double-charged particles (see Fig. A2). Single particle mobility size is derived from a lognormal fitting of the aerosol size distribution measured by SMPS, assuming soot 260 particles to be spherical. The mass value of a single particle is derived from CPMA mass distribution scans with a similar mathematical data processing method. Based on single soot particle mobility size and mass results, H2SO4 coating wt %, https://doi.org/10.5194/acp-2021-645 Preprint. Discussion started: 6 September 2021 c Author(s) 2021. CC BY 4.0 License. equivalent coating monolayers (assuming uniform coating) and particle effective density are calculated. The H2SO4 coating wt % is the ratio of soot particle coated H2SO4 mass to the mass of initial bare soot particle and defined as: where is the mass of H2SO4 coated by the soot particle and is bare soot particle mass. The number of equivalent coating monolayers (ML) is calculated according to Wyslouzil et al. (1994) and given by: where wt is the coating mass percentage value, $ is the Avogadro constant, % is the molecular weight of H2SO4, & is the number of H2SO4 molecules per cm 2 corresponding to one monolayer coverage (4.5×10 14 cm -2 ) (Wyslouzil et al., 1994), is 270 the specific area of the soot particle (mCASTblack 120 m 2 g -1 , FW200 526 m 2 g -1 ) (Mahrt et al., 2018). According to McMurry et al. (2002), the particle effective density ( '(( ) expression is given as: where is the single particle mass and is the particle mobility size. To control the H2SO4 coating wt %, a number of factors need to be considered, including initial soot aerosol particle number concentration, saturation and dilution flow rate and 275 temperatures, the temperature of pure H2SO4 in the flask and subsequent cooling condition (see Fig. 2), the setup volume, as well as the intrinsic soot particle properties. Here, temperatures, flow rates and the whole volume of the apparatus is well controlled. However, it is impossible to adjust the initial soot sample particle number concentration to be identical for every experiment. As a result, even though the H2SO4 coating thickness increases with increasing H2SO4 saturation temperatures, for per soot sample coating, the result shows some variance. In this study, the standard deviation value for each coating wt % 280 is provided in Table 1 and in each panel of Fig. 3.
The H2SO4 coating wt %, particle mobility size, particle '(( and the equivalent ML value for each soot sample are presented in Fig. 3. Note that the equivalent ML values highly depend on the soot particle specific surface area given by BET (Brunauer-Emmett-Teller) analysis (Brunauer et al., 1938), which can be influenced by different pre-treatment levels and also different 285 gas probes used in the physisorption measurement (Lowell et al., 2004). For example, Ouf et al. (2019) reported that soot BET specific surface area can be severely influenced by the sample outgassing level and can vary in a significant range. Furthermore, it is suggested that N2 BET and Argon BET results for the same soot sample show variance (Lowell et al., 2004). Nevertheless, the results in Fig. 3 still can provide relatively comparable information to evaluate soot particle mixing states with different H2SO4 coating thicknesses. Overall, 200 nm size selected soot particle mobility size can increase by ~ 5 % with increasing 290 H2SO4 coating wt % but starts to decrease dramatically when the H2SO4 coating wt % is larger than 20 %, which means first size growth and then shrinkage (collapse). Whereas 400 nm size selected soot particle does not show apparent size growth and starts to collapse when the coating mass percentage reaches ~ 20 %. With increasing coating wt %, the equivalent H2SO4 coverage ML and coated particle '(( increase monotonically as expected. Scanning Electron Microscopy (SEM) (Zeiss Leo 1530, Carl Zeiss AG, Oberkochen, Germany) is used to observe the H2SO4 coating state on soot particles. Both coated and uncoated FW200 and mCASTblack 400 nm soot-aggregates are collected on 300 400 mesh Cu girds with a formvar/carbon support film (TED, PELLA, Inc.). All grids are collected using the Zurich Electron Microscope Impactor (ZEMI), a home-built and semi-automated rotating drum impactor (Aerni et al., 2018;Mahrt et al., 2018) running in parallel to HINC ice nucleation experiments. During the grid sampling, ZEMI pulls a 1 L min -1 aerosol sample flow as shown in Fig. 1. The total sampling time is between 5 and 10 minutes depending on the particle concentration. In addition to bare particles, coated soot-aggregates with the lowest and the highest H2SO4 coating wt % (see Table 1) in the ice nucleation 305 experiments are collected for SEM analysis. For convenience of discussion in Sect. 3, these two coated conditions will be termed as thin coating and thick coating, respectively. Images of interest are obtained at magnification values 20k and 200k.
The small magnification value is used to have an overview of the particles. Soot particle morphology change and H2SO4 coating state are analysed from high resolution images.

310
Transmission Electron Microscopy (TEM) and High Resolution TEM (HR-TEM) (JEOL-1400+ TEM, JEOL Ltd., Tokyo, Japan, operated at 120 kV; Hitachi HT7700 EXALENS, Hitachi Ltc., Chiyoda, Japan, operated at 100kV; TFS Talos F200X, operated at 200 kV, and TFS F30, operated at 300 kV, both Thermo Fisher Scientific Inc., Waltham MA, USA) are utilized to detect morphology changes induced by H2SO4 coating and physical evidence on H2SO4 coating. These grids were collected separately from the SEM grids. To match with corresponding ice nucleation experiments, the coating wt % (see Table 2) of 315 soot-aggregates for TEM analysis was controlled to be comparable to those coating wt % shown in Table 1 for 200 nm soot samples. Again, the lower H2SO4 coating wt % will be termed as thin coating and the higher one will be termed as thick coating for further SEM results discussion. Both bare and coated soot-aggregates are collected on Quantifoil Cu 200 mesh R2/2 grids (Quantifoil Micro Tools GmbH, Großlöbichau, Germany). Open holes of the R2/2 pattern provide a view on sootaggregates without a carbon film background. 320 Energy Dispersive X-ray spectroscopy (EDX) (TFS Talos F200X equipped with a Super-X EDS system) is used for chemical evidence on the presence of H2SO4 on soot particle surfaces. These grids used for EDX are the same as those for the TEM measurements. Both bare and coated soot-aggregates are analysed and results are presented both by 2D element distribution maps and conventional element spectra. Carbon (C), oxygen (O) and sulphur (S) element mass content for soot particles with 325 different H2SO4 coating wt % are also calculated and normalized to compare their chemical composition change. 3 Results and discussion 330 Soot particle ice nucleation abilities are presented in terms of AF curves as a function of RHw and RHi in Figs. 4 to 7. The 1 μm OPC channel data is used to plot AF curves, referring to the ratio of the total number of particles exiting the HINC chamber larger than 1 μm to the total number of particles entering the chamber as sampled by the CPC. The particle property results obtained by characterization measurements in Sect. 2 will be used to understand the corresponding soot particle ice nucleation activities. Figs. 4 and 5 present the AF plots for coated 200 and 400 nm mCASTblack particles and uncoated counterparts as a function of RH values at T from 218 to 243 K. Ice nucleation activities of bare and coated 200 and 400 nm FW200 particles are shown in Figs. 6 and 7, respectively. Both bare mCASTblack and FW200 soot particle can only form water droplets above water saturation conditions at 243 and 238 K (T > HNT), regardless of particle size. At these two temperatures, OPC signals are only observed at small channels but signals are absent at the largest channel, i.e. 5 μm channel (see Figs. B5 to B8). This is because water droplet growth by vapor diffusion is less efficient than that of ice crystal growth at these conditions (RHw < 340 105%) in HINC (Lohmann et al., 2016). Thus, water droplets do not grow up to 5 μm and are thus not detected in the 5 μm OPC channel, however ice crystals can grow up to this size value (Lacher et al., 2017;Mahrt et al., 2018). Therefore, we conclude no ice nucleation onto bare soot particles at 243 and 238 K. This is in agreement with  who suggested that commercial black carbon particles require temperatures lower than HNT to freeze. For mCASTblack, ice nucleation results were the same for both 200 and 400 nm particles showing no significant activation at RH < RHhom at cirrus 345 relevant temperatures (T < HNT). FW200 particles exhibit significant ice nucleation at T < HNT unlike the mCASTblack, with 400 nm FW200 particles being more active INPs than 200 nm samples. Unsurprisingly, excellent agreement is achieved for the results of bare mCASTblack and FW200 soot between this study and Mahrt et al. (2018), who also studied these two types of soot INP samples using the same chamber. At T < HNT, AF curves for bare mCASTblack soot particles stay in the uncertainty range of homogeneous freezing. As shown in Figs. 4 and 5, only < 10 -3 of the soot sample can form ice crystals at 350 RH values slightly lower than the homogeneous RH values, even at 218 K. This is consistent with the results from Mahrt et al. (2018;2020a;2020b). As shown in Figs. 6 and 7, bare FW200 soot particles are active INPs and can form ice crystals at RH < RHhom when T < HNT, also in line with the results presented by Mahrt et al. (2018). The authors suggested that these soot particles with small mesopores can be effective INPs and form ice crystals via PCF at humidity conditions lower than homogeneous freezing conditions in the cirrus cloud regime (Mahrt et al., 2018). The paucity of mesopores relevant to PCF 355 and its low surface wettability make mCASTblack soot a poor INP at T < HNT, compared to FW200 soot. Here, PCF relevant mesopores refer to pores with right size and shape which are not only small enough to induce liquid water capillary condensation at RHw < 100% due to the inverse Kelvin effect but also large enough to allow ice growing out of the pore upon homogeneous freezing of supercooled pore water (Marcolli, 2014;Marcolli, 2020). The active ice nucleation ability of FW200 can be attributed to its low soot-water contact angle and abundant mesopore structures. 360 In addition to bare soot particles, more than eight different H2SO4 coating wt % are performed from less than one equivalent molecule monolayer to several equivalent molecule monolayers coverage. For ease of discussing the ice nucleation activity comparison between bare and coated particles, three typical coating states, namely thin, medium and thick coating, are selected as representative cases. The results for all coating wt % corresponding to Table 1 are given in the Appendix B, Figs. B1 to B4.

365
Thin coating stands for a H2SO4 coating wt % less than 3.5 % of its initial mass value. Medium coating represents a H2SO4 coating of wt % ~ 10 % whereas thick coating represents a H2SO4 coating of wt % larger than 25 %.

H2SO4 coated mCASTblack soot ice nucleation results
At T > HNT (243 and 238 K), both coated and uncoated particles require water saturation conditions to form water droplets.
With similar evidence to bare particles, we conclude no ice nucleation at T > HNT for coated soot of all coating masses. Soot 395 particles coated with H2SO4 are expected to be more hygroscopic and act as CCN if a uniform coating is assumed. However, the AF curves for both 200 and 400 nm coated mCASTblack soot stay within error bars with those of bare particles at T > HNT, as shown in Figs. 4 and 5. There is some indication that coated particles are better CCN at T > HNT. However, we cannot claim that the hygroscopicity is significantly different as seen the AF curves overlapping with those of bare soot particles, suggesting that the coatings may be non-uniform but form clusters on soot-aggregates. Non-uniform coating patterns 400 can be seen from SEM images in Fig. 8. Muller et al. (1996) and Persiantseva et al. (2004) suggested that soot particle https://doi.org/10.5194/acp-2021-645 Preprint. Discussion started: 6 September 2021 c Author(s) 2021. CC BY 4.0 License. micropores play an important role in the soot-water interaction activities if there is no soluble material available on its surfaces. Popovicheva et al. (2008a;2008b) demonstrated that capillary condensation induced by mesopores in soot-aggregates can occur when RHw values are larger than 80 %. Therefore, it is feasible that a small amount of H2SO4 coating on mCASTblack particle impacts its soot-water interaction activities if non-uniformly distributed coating material leads to pore filling which 405 reduces mesopore availability. However, if further H2SO4 coating provides soluble material on soot surfaces for water adsorption, it compensates the unavailability of pores caused by thin coating pore filling and enhances soot update ability because of its high hygroscopicity. Overall, there is no significant change for mCASTblack soot CCN activity after H2SO4 coating using the method presented in this work.

410
At T < HNT (233-218 K), all coated particles only form ice via homogeneous freezing and there is an apparent freezing depression particularly at 223 and 218 K for thinly coated soot, as can be seen in Figs. 4 and 5 by the delayed freezing onset of coated particles compared to that of bare particles at each T. This implies that H2SO4 coating inhibits the PCF mechanism at these low T. From Fig. 4, thinly coated 200 nm mCASTblack soot particles show the largest suppression in the onset of ice formation among all coating masses. This shows coherence with the results at warm temperatures (T > HNT) and can be 415 explained by the H2SO4 pore filling effect. Homogeneous freezing of supercooled water in pores is an important step in a PCF activation process (Marcolli et al., 2021;Marcolli, 2014). According to Koop et al. (2000), the homogeneous freezing rate depends on the liquid water activity. Due to H2SO4 filled in pores for thin coating cases, inverse Kelvin effect induced water uptake will form high concentration H2SO4 solution with a low water activity, which leads to depressed homogeneous freezing as it requires higher RHw conditions than RHhom for freezing. This inhibits the bulk freezing of the soot particle with low H2SO4 420 coating wt % until the particle takes up sufficient water to dilute enough for bulk droplet freezing. With more coating (in medium and thick cases), hygroscopic H2SO4 adsorption on soot surfaces competes with H2SO4 pore filling and promotes the water uptake ability of soot particles to form a bulk water droplet more readily. This can explain why the soot particle with thicker coating tends to freeze homogeneously like a bulk solution droplet at higher RH conditions. More details about the interaction of soot particle with H2SO4 will be discussed in Sect. 3.3. The similar ice nucleation activity of 400 nm coated 425 mCASTblack soot particles can be seen from Fig. 5.
Results for coated mCASTblack soot in this study are comparable to the literature. For instance, ice nucleation results at low T are in good agreement with previous H2SO4 coated miniCAST C3H8 soot results from Crawford et al. (2011), who treated polydisperse miniCAST soot aerosol in saturated H2SO4 vapor and studied its ice nucleation behaviour at T from 220 to 230 430 K. Similarly, the authors concluded that the addition of H2SO4 inhibits soot particle ice activation and shifts the freezing onset RH towards the homogeneous freezing conditions of pure H2SO4. Möhler et al. (2005b) also reported a suppression in the ice nucleation ability of polydisperse graphite spark generated soot samples (~25-300 nm) coated with H2SO4 (~40-350 nm) which freeze towards homogeneous freezing conditions. However, Mahrt et al. (2020b) reported a significant ice nucleation enhancement for 400 nm mCASTblack soot particles aged in a bulk H2SO4 aqueous solution (pH = 4) and coated by dry residues of H2SO4. The authors used a different aging method to treat the soot sample and the H2SO4 ageing effect and coating material distribution are different from the coating conditions in this study. The 400 nm particles in Mahrt et al. (2020b) were generated by atomizing a bulk aqueous suspension followed by drying and size selection. If a solution droplet spanning a size range from 400 nm to 2 μm is assumed before drying, the H2SO4 coating wt % is estimated to be in the range from 0.0026 % to 0.32 %, much less than the coating wt % in this study. Furthermore, the H2SO4 coating generated in this way can be 440 distributed more uniformly over the soot particle surface, compared to a nonuniform H2SO4 coating in this study (see Sect. 3.3). On the other hand, soot particles experienced a water interaction process when they were treated with H2SO4 aqueous solution, which means water interaction induced morphology changes and particle compaction could be responsible for this ice nucleation promotion as suggested by soot pure water ageing case in the same study by Mahrt et al. (2020b). In addition, DeMott et al. (1999) investigated the freezing RH conditions for multiple-layer H2SO4 coated polydisperse lamp black soot 445 particle (Degussa Corporation, Frankfurt, Germany) with a size distribution mode of 240 nm under cirrus cloud conditions, using a continuous flow diffusion chamber. The authors reported that these coated soot particles require RHi values 152 and 156 % respectively to reach 1 % AF level at 228 and 223 K (Demott et al., 1999), which is close to the RHi values 152 and 154 % at the same T for 200 nm mCASTblack soot particles with thick H2SO4 coatings presented in this study. It is reported that soot particles with aggregate compaction after experiencing a cloud process can nucleate ice via PCF at RH < RHhom 450 conditions when T < HNT (Mahrt et al., 2020a). However, even though the thick H2SO4 coating of mCASTblack soot particle in this study results in a significant size shrinkage (see Fig. 3a and b), i.e. some extent aggregate compaction (see Figs. 8c and 9c), its ice nucleation ability is not promoted compared to the uncoated particles. This suggests that the availability of pores (i.e. pore volume) without a freezing depression will be a prerequisite for PCF ice nucleation despite of lowering the sootwater contact angle by H2SO4 coating. 455

H2SO4 coated FW200 soot ice nucleation results
In general, H2SO4 coating has limited effects on FW200 soot particle droplet activation ability at T > HNT as shown in Figs. 6 and 7. There is no significant deviation between the AF curves of soot particles (both for 200 and 400 nm) with and without H2SO4 coating at 243 and 238 K. It is safe to conclude that the H2SO4 coating generated in this study does not enhance FW200 soot particle water interaction and CCN activation ability at these two T. The AF curves of bare and coated FW200 soot 460 particles in Figs. 6 and 7 are consistent with the study conducted by Koehler et al. (2009) who reported that the CCN activation of hydrophilic aviation kerosene soot particles requires higher RH conditions than the Kelvin RH limit (Henson, 2007), which is required by wettable particles to show CCN activation. We attribute our results to the absence of a uniform H2SO4 coating similar to that discussed above. With such a non-uniform coating, the hygroscopicity of coated H2SO4 does not exhibit its water interaction enhancement for its host soot particle. 465 From the AF curves in Figs. 6 and 7 at T < HNT, thinly and moderately coated FW200 soot particles form ice crystals at RH < RHhom showing that PCF is still well preserved. This means pores or voids with right size for PCF are not all blocked or are still available when soot particle is coated by H2SO4, also implying that the adsorbed H2SO4 by the soot particle does not distribute evenly or uniformly. In general, thin and moderate H2SO4 coatings cannot depress FW200 soot ice nucleation ability 470 for both 200 and 400 nm coated FW200 soot particles. At T < HNT, the 200 nm soot particles with thick H2SO4 coverage show an obvious ice activation depression and an apparent freezing mechanism shift from PCF to homogeneous freezing near the homogeneous freezing RH thresholds, as shown in Fig. 6. Thickly coated 400 nm FW200 soot particles in Fig.7 show a similar ice nucleation onset compared to bare soot particles but with a slight suppression in total AF observed, suggesting the ice nucleation mesoporous volume for PCF is still available. Approaching RHhom, there is a clear shift in the slope of the AF curves 475 at T ≤ 233 K in Fig.7 suggesting a mechanism change from PCF to bulk homogeneous freezing, which aligns with the predictions of Koop et al. (2000). Overall, 200 nm soot particle ice nucleation abilities are more affected by the thick H2SO4 coating compared to 400 nm particles. Given the approximate spherical structure of a FW200 soot-aggregate, a larger size and surface area will require a higher H2SO4 coating wt % to achieve the same equivalent coating coverage as that of a smaller aggregate. Soot particles with a small mobility size contain less pore volume and can be more easily covered by H2SO4 coatings, 480 thus their PCF activation is readily inhibited and they freeze homogeneously like a H2SO4 droplet with a soot particle core. It is safe to extrapolate that further H2SO4 coating for 200 nm FW200 soot particles will totally depress its PCF freezing tail (at 218 K) and the coated particle will freeze only homogeneously according to Koop et al. (2000). Comparably, 400 nm FW200 soot particles contain more abundant pore structures which are not inhibited to PCF even upon excessive H2SO4 coating. For the same H2SO4 coating wt % more PCF active pores are available in 400 nm FW200 soot particles than in 200 nm particles.

485
This again implies that the distribution of H2SO4 onto soot particle surfaces is not uniform or the H2SO4 coating states are different over soot particle local structures.

Summary of coated soot ice nucleation
To sum up, we propose three possible scenarios to explain the H2SO4 coating effects on mCASTblack and FW200 soot particle 490 IN activities. Firstly, H2SO4 may fill in and block some mesopores of the soot-aggregate making it PCF inactive at low T because the subsequent pore water capillary condensation into H2SO4 filled pores will form concentrated H2SO4 solutions with a low homogeneous nucleation rate (Koop et al., 2000). This can be attributed to thin or medium H2SO4 coating wt % scenario for mCASTblack. Meanwhile, if the soot particle is porous and contains sufficient pore structures, some mesopores can remain unfilled and are still able to induce PCF such as for thin and medium coatings of 200 nm FW200 and even for thick coating of 495 400 nm FW200. Secondly, most of the mesopores are affected by H2SO4. At low RH conditions, mesopores filled with H2SO4 solution require a higher saturation condition to freeze, in order to recover to high enough water activity and to compensate the homogeneous nucleation rate depression. At higher RH conditions near homogeneous freezing condition, the depressed PCF mechanism is not active and so these coated particles freeze homogeneously. The second scenario can help with the understanding of thick H2SO4 coating effects on mCASTblack particles and 200nm FW200 particles. Finally, excessive H2SO4 500 coating tends to block all pores and shuts down the PCF mechanism. In this case, H2SO4 might form a shell on the particle https://doi.org/10.5194/acp-2021-645 Preprint. Discussion started: 6 September 2021 c Author(s) 2021. CC BY 4.0 License. surface or encapsulate the particle so the coated particle behaves as a H2SO4 droplet and only freezes via homogeneous freezing. We suggest the last scenario is relevant to soot particles with high H2SO4 coating wt % (> 100 %). Detailed descriptions about these three coating states will be discussed in Sect. 3.3.2.

Typical internally mixed states of soot particles and sulphuric acid 505
Coupling single particle size and mass measurement results with microscopy of the particle mixing state characterization results, as well as with the ice nucleation activities of H2SO4 coated particles with different coating wt %, a hypothesis depicting the internal mixing states of H2SO4 and soot particles is proposed. In this section, particle microscopy characterization results will be presented first, and next, a three-step process for soot particle H2SO4 coating will be explained to further understand particle ice nucleation activities shown in Sects. 3.1 and 3.2. 510

SEM, TEM and EDX results to characterize particle morphology
SEM results both for bare and coated soot particles are shown in Fig. 8. These SEM images are taken at 200k magnification for 400 nm mCASTblack and FW200 soot particles with H2SO4 coating wt % similar to those coated soot particles in Figs. 5 and 7 and also termed as bare, thin and thick coatings. Images for the same soot samples at 20k magnification are provided in Fig. C1. As seen in Fig. 8a, bare 400 nm mCASTblack soot particle is fractal and looks like a long primary particle chain with 515 lacy voids. Thinly coated 400 nm soot-aggregate (as shown in Fig. 8b) is close to the bare. Some primary particles form a chunk with thick H2SO4 coating and the shape of the primary particles is more ambiguous than in the bare aggregate, which can be seen in Fig. 8c (thick coatings). In addition, thickly coated 400 nm mCASTblack soot-aggregates tend to fold by showing a slight curvature and appears less fractal, compared to bare and thinly coated aggregates. 520 Figure 8d shows that a FW200 bare 400 nm soot-aggregate is less fractal than the same size bare mCASTblack soot-aggregate in Fig. 8a. This is in agreement with the particle fractal dimension results presented in Mahrt et al. (2018), where the authors reported a higher fractal dimension value for FW200 soot particle (2.35) compared with mCASTblack (1.86). There are no distinguishable morphological feature differences between Fig. 8d and e, which means thin H2SO4 coating does not result in a significant soot surface topography change. We believe that a small amount of H2SO4 first fills into pores among primary 525 particles and is unable to modify the particle surface significantly. However, heavier mass coatings will have a part of sootaggregate embedded into the H2SO4 material, as clearly visible in Fig. 8f. These images are comparable to the SEM images for ambient soot particles in a field study ; see Fig. S4b). Besides, a ring of spray in Fig. 8f is demonstrated to be small H2SO4 droplets resulted from the impaction of soot-aggregates on the Cu grid, as seen EDX results (discussed below and see Fig. 11). Similar results about H2SO4 coating introduced morphology change for FW200 soot can also be seen 530 from TEM images shown in Fig. C2f.  TEM images of 200 nm bare, thinly and thickly coated both for mCASTblack and FW200 soot particles are shown in Fig. 9.
The coating wt % is presented in Table 2 and similar to those of the ice nucleation experiments, as aforementioned in Sect.

2.2.3.
There is no apparent morphology difference between bare and thinly coated aggregates for mCASTblack soot (as shown in Fig. 9a and b), which is consistent with the SEM results. However, the aggregate size in Fig. 9a and b is much larger than 540 200 nm. This could be an artifact during grid sample collection, which results in soot-aggregate piling or agglomeration when particles deposit onto the grid. Small aggregates can also coagulate while transporting in the aerosol flow (Kulkarni et al., 2011). In addition, soot-aggregates are of heterogeneous shapes and even size selected monodisperse soot aerosol spread a larger size distribution range than the selected size. Besides, a part of double-charged particles, which have a larger mobility size, also exist as already specified in Sect. 2.2.3 (see Fig. A3). Thus, these larger aggregates with higher inertia are more 545 effectively collected by the ZEMI impactor. Nevertheless, the aggregate in Fig. 9a or b is already much larger than the size of a double-charged 200 nm aggregate (~ 320 nm). In order to support that the large aggregate sizes are an artifact of grid sampling, the optical size of suspended soot-aggregates for TEM grid collection are also tested by the OPC and the results are presented in Appendix C, showing that the suspended aggregates are binned into the expected optical size channels and demonstrating their size selection is acceptable. The thickly coated mCASTblack aggregate in Fig. 9c shows a more compacted 2D projection 550 shape, which suggests the thick coating increases primary particle connectivity resulting in some extent aggregate compaction.
Similar to mCASTblack, thin H2SO4 coating for FW200 soot particle does not lead to distinguishable features, comparing Fig.   9d with Fig. 9e. Clearly, thick coating for FW200 soot particle results in a shrunken aggregate projection with the edge boundary being smoother than the bare and thinly coated ones. This shrinkage is also explained by the decrease in mobility size of soot particles with thick coatings shown in Fig. 3c. In addition, thick H2SO4 coating seems to show a darker visualization 555 effect for FW200 soot-aggregates compared to bare particles. In addition, TEM images provide evidence of H2SO4 presence on soot particle surfaces, which is consistent with SEM images. In Fig. 9f, the watermark indicated by a red square between the particle and the edge of the carbon support film is believed to be a H2SO4 footprint. During the electron microscopy it could be observed how the structure, which looked initially like a balloon, slowly evaporated and finally left this shrivelled skin. Evidence on the presence of H2SO4 on soot primary particle surfaces and H2SO4 pore filling effects from HR-TEM 560 images is presented in the Appendix, Figs. C4 and C5. EDX measurements demonstrate H2SO4 coating shows a non-uniform distribution on soot-aggregate surfaces. Grids with 200 nm mCASTblack and FW200 soot-aggregates are sampled. To indicate soot-aggregate surface H2SO4 distribution, composite 570 images in HAADF (High Angle Annular Dark Field) of 2D EDX elemental maps for S (red dots in Figs. 10a and 11a) are used. With a focus on the particle surface chemical composition, C, O and S element contents for areas of interest (AOI) within a soot-aggregate projection are quantified to obtain the spectra to avoid the influence of grid material, as indicated by the red square in Figs. 10a and 11a. The mass percentage of each element is normalized by the total result of these three elements and attached in Figs. 10b and 11b. Background signal corrected element spectrum results are provided in Appendix C. 575 Comparing S maps for mCASTblack in Fig. 10a, the presence of H2SO4 on the soot is indicated by an increasing signal level with increasing coating thickness, however the coating material distribution is not uniform on the soot-aggregate surfaces. A radar chart in Fig. 10b shows quantitative chemical composition changes for H2SO4 coated mCASTblack soot. With H2SO4 coating, the S content increases as expected whereas the O content decreases, which is unexpected because H2SO4 contains a 580 significant proportion of oxygen. This may suggest that there is a reaction between mCASTblack soot and H2SO4 depleting some O content in the process, considering H2SO4 as a strong oxidant with respect to mCASTblack soot (Mahrt et al., 2020b). Fig. 10b, even though thick coating has a H2SO4 coating mass percentage by more than 30 %, only a tiny amount of S is detected by the EDX measurement over AOI. It is feasible that the soot local surface (i.e. AOI) detected is with a lower coating wt % than the global soot particle, considering non-uniform H2SO4 coating distribution on the soot surfaces.

585
Additionally, a part of H2SO4 coating material is lost during the measurement because of the heating effect from high energy electron beams, which also partly explains the lower H2SO4 content detected by EDX mapping. In Fig. 11a for FW200 soot, intensive signals on coated particle surfaces also suggest random distribution of coating material and that there are preferred sites for the interaction of soot and H2SO4. Watermarks attached with and around thickly coated soot-aggregate are shown to be H2SO4 droplets. This finding coincides with the SEM result in Fig. 8f and TEM result in Fig. 9f. The normalized C, O and 590 S mass ratio result clearly suggests the increment of H2SO4 coating material in terms of O and S and the decreasing share of soot material in terms of C (Fig. 11b). Contrary to coated mCASTblack soot particles, the normalized proportion of O in H2SO4 coated FW200 soot particles increases with higher H2SO4 coating wt %, suggesting that H2SO4 coating induced surface reaction (or oxidation) for FW200 soot does not occur (or at least not to the same extent) as that for mCASTblack soot. This can be attributed to the fact that FW200 soot is already surface oxidized during its production according to the information 595 provided by the manufacturer.

(coating wt = 139.3 %) aggregates. Sulphur signal (red dots) is overlain on HAADF (high-angle annular dark-field) images. Scale bars are indicated in each image. Panel (b) the normalized carbon (C), oxygen (O) and S mass percentage of aera of interest (AOI, indicated as red square in panel (a)) for FW200 soot with different coating thicknesses.
Overall, SEM images show that thick H2SO4 coatings can change soot-aggregate surface topography. The coating material 610 either can be distributed over the soot surfaces unevenly and change the connectivity of adjacent primary particles or will pile up over primary particle clusters to form chunks and finally form a H2SO4 shell. TEM images show thin H2SO4 coatings do not change the morphology of soot-aggregates whereas thick H2SO4 coatings can modify their morphology significantly by inducing aggregate structure compaction. EDX 2D maps provide robust evidence on the presence of H2SO4 over sootaggregates and further demonstrate that the H2SO4 distribution is not uniform. Comparing the thick coating results for 200 nm 615 mCASTblack soot (Fig. 9c) to those of FW200 soot (Fig. 9f), coating effects on mCASTblack soot-aggregate morphology modification are different even though the coating equivalent ML values are similar (3.4 ± 0.4 ML and 3.6 ± 0.3 ML, respectively, according to Table 2). Moreover, the SEM image with a higher 3.5 equivalent ML coating for 400 nm mCASTblack aggregate (Fig. 8c) compared to that of a 400 nm FW200 aggregate with 1.7 equivalent ML coating (Fig. 8f) does not show more compaction of mCASTblack particles. This different coating effect between mCASTblack and FW200 620 soot results from the differences in intrinsic soot morphology properties. Nonetheless, we expect that mCASTblack sootaggregates can also be embedded into H2SO4 and appear to be as compacted as thickly coated FW200 if H2SO4 coating wt % is achieved, e.g. > 100 %. Therefore, the equivalent ML value is not a reliable proxy to describe the distribution of H2SO4 coating material on soot-aggregate surfaces.

A three-step process for soot particle H2SO4 coating 625
Based on the aforementioned soot particle ice nucleation activity and property characterization results, we hypothesize that the soot coating process presented here has three stages (as illustrated in Fig. 12). A soot-aggregate comprising of numerous primary particles has a complex intra-aggregate structure (Vander Wal et al., 2007;Bond et al., 2013), a sketch of which is shown in Fig. 12a. We first start with a small amount of H2SO4 coating (Fig. 12b) where H2SO4 molecules fill into pores among primary particles or voids in the soot-aggregate because of the capillary force but not all pores will be filled completely. 630 Meanwhile, H2SO4 molecules can form clusters (humps or small islands) on the surfaces. At this point, the particle size is not expected to change as supported by the single particle mass and size results shown in Fig. 3. The particle effective density increases following a smooth slope, showing the gain of H2SO4 as the density of bare soot particles is smaller than that of pure H2SO4. In general, the overall coating effect of the first stage is pore filling and if any, some aggregate shrinking.

635
Further increase in coating mass results in a more complete pore filling and the spread of H2SO4 on aggregate surfaces. As depicted in Fig. 12c, at this stage of the coating process, particle size growth starts to be detectable. At this stage, the particle effective density may increase by a different manner depending on soot types. In Fig. 3, mCASTblack soot particle effective density does not change whereas FW200 soot effective density continues to increase. This is because mCASTblack soot is more fractal or lacy (Figs. 8 and 9) and H2SO4 may form molecule clusters or branches sticking on the particle surfaces. Hence, 640 the size growth effect overwhelms the mass increase effect. On the contrary, FW200 soot is more compact and has abundant pore structures, thus H2SO4 can be accommodated inside the pore volume. Thus, the addition of H2SO4 for FW200 soot shows less size growth effect than for mCASTblack soot. Therefore, the main effect of the second stage in H2SO4 coating process is further pore filling with coating material spreading to the surfaces. Very likely, H2SO4 shows preference to the soot particle surfaces and will firstly stick to some special sites on the surface before forming molecule clusters. For example, Garland et 645 al. (2008) used SEM and atomic force microscopy (AFM) to investigate how the oleic acid deposits on soot particle surfaces and suggested that oleic acid islands with a height of several nano-meters will be accumulated on the surface rather than form a uniform oleic acid coverage layer.
Finally, once pores are filled and the growth of H2SO4 cluster on the surface continues, the growing H2SO4 clusters will interact 650 with each other or collapse onto the particle surfaces, or neighbour islands start to merge into big patches. This process induces collapse also of the particle structure causing particle compaction, as shown in Fig. 12d. Consequently, the third mixing state results in the particle shrinkage and finally forming H2SO4 encapsulated particles with a soot core if the coating is thick enough. This argument is supported by the dramatic size shrinkage of thickly coated soot particles (see Figs. 8f and 9f) in SEM and TEM images of particles embedded or partially embedded in H2SO4 droplets. Such a mechanism is also supported by some 655 work in the literature. For example, fractal soot agglomerates can be compacted by the condensation of ozonolysis products of α-pinene secondary organic aerosol (SOA) with the bare soot particle serving as a nucleation centre (Saathoff et al., 2003). Khalizov et al. (2009) also pointed out that small soot-aggregates gain size growth but larger aggregates show a dramatic size shrinkage due to structure compaction after being exposed to gaseous H2SO4 for more than 10 s. Most recently, Pei et al. (2018) investigated the morphological changes of soot particles after H2SO4 or limonene ozonolysis SOA coating, and suggested that 660 the coating material will change soot particle nanostructure by a two-step process. Firstly, external coating material will fill into the voids or pores among soot-aggregates. Next, the particle size will grow with further vapor condensation when the voids are filled completely. In brief, our three-stage hypothesis has robust support from measurements in this study and is also in good agreement with previous findings in the literature. 665 Figure 12. Illustration of the proposed H2SO4 coating processes for soot particles (not to scale). Bare soot particle (a), soot particle pore filling and surface coating induced by progressive increase in H2SO4 coating mass (b, c, d).

Atmospheric implications
Soot particles in the troposphere can affect cirrus cloud formation. H2SO4 coatings have clear suppression effect on soot particle ice nucleation based on the extent of the coating and the soot particle properties, with compact and porous particles being able 670 to resist suppression in ice nucleation more than fractal particles. The difference in this suppression effect is because of the higher availability of mesopores in compact soot particles. Given the long lifetime (from ~ 5 days global mean up to 3 weeks) of soot particles in the atmosphere (Liu, 2005;Shen et al., 2014;Lund et al., 2018;Liu et al., 2020), the processes soot particles undergo with respect to ageing and cloud processing can be complex . Due to the ubiquitous nature of H2SO4 in the troposphere (Pye et al., 2020), acid coatings are likely atmospheric ageing processes for soot particles. Hence, it 675 would be important to consider the sulphate-soot mixing state when evaluating the impact of soot on cirrus cloud formation.
As demonstrated in this study, soot particles internally mixed with H2SO4 exhibit a suppressed ice nucleation ability depending on their morphological properties. Thick H2SO4 coatings (wt > 100 %) effectively make the soot ice nucleation contribution to cirrus clouds formation comparable to homogeneous freezing of solution droplets. Therefore, the soot-sulphate mixing state impact on ice nucleation should be considered in predicting the impact of soot particles on cirrus formation, as well as when 680 evaluating the radiative forcing and climate impacts of ice clouds in upper troposphere (Lohmann et al., 2020). In other words, using soot proxies without sulphate coatings may only be relevant for regions where soot is quickly incorporated into the cloud upon emission. Field studies detected soot-aggregates with 0.1-1 μm sizes in ice crystal residuals sampled from cirrus clouds (Petzold et al., 1998;Twohy and Poellot, 2005;Cziczo and Froyd, 2014) and the presence of 100-800 nm BC particles in the interstitial aerosol samples increases with increasing altitude from 8 to 11 km in the cirrus cloud regime (Petzold et al., 1998). 685 Thus, we believe that the results presented in this study are of atmospherically relevant sizes (200 and 400 nm).
Considering that mCASTblack soot represents fractal and hydrophobic soot (Marhaba et al., 2019) and as an appropriate proxy for aviation soot particles (Ess and Vasilatou, 2018), H2SO4 coated mCASTblack soot particles in this study can be comparable to aviation soot particles coated by H2SO4 in the atmosphere. At T < HNT, H2SO4 coated aviation soot particles may not inhibit 690 or compete with droplet homogeneous freezing. This is because bare and H2SO4 coated 200 and 400 nm mCASTblack particles freeze at conditions for homogeneous freezing of solution droplets. Given the minor ability of ice formation for mCASTblack soot via PCF even at 218 K addressed in this study, uncoated aviation soot analogous to fractal and hydrophobic mCASTblack may not make significant contribution to cirrus cloud formation via PCF below homogeneous freezing conditions. Moreover, the PCF activation of mCASTblack soot particles with a low H2SO4 coating wt % will be further depressed and require RH > 695 RHhom conditions to form ice crystals. This implies that aviation soot particle proxies even with low H2SO4 coating mass (similar to our mCASTblack soot low H2SO4 coating wt %) could only form ice crystals above RHhom conditions at T < HNT.
Initially active soot particles (similar to our FW200 soot sample), if not extensively mixed with H2SO4 in the atmosphere, only demonstrate a slight suppression in ice nucleation ability at cirrus relevant temperatures (T < 233 K). Thus, these coated particles still nucleate ice via PCF forming ice crystals at RH < RHhom conditions, with the potential to perturb the cirrus background. Thickly coated soot particles, freezing homogeneously, are also atmospherically relevant. Bhandari et al. (2019) collected ambient soot particles under different weather conditions in a valley and reported that most ambient soot particles are covered with a thick coating or are embedded in external materials. If these thickly coated soot particles can be activated as droplets and transported to high altitudes at lower T conditions, it is likely that ice crystals can be induced by them via 705 homogeneous freezing. In water supersaturated conditions at T < HNT, it is reported that the homogeneous freezing of water droplets nucleated by soot particles is the dominant ice crystal formation pathway for cirrus clouds (Kärcher and Yu, 2009).
Consequently, the water interaction enhancement of soot particles with thick H2SO4 coatings in this study resembles this case.
According to Ditas et al. (2018), BC particles emitted by wildfires can reach the lowermost stratosphere and the authors suggested that a BC particle with a 120 nm average mass equivalent diameter can be coated by volatile organic compounds 710 (VOCs) and/or H2SO4 with a thickness of 150 nm. In this case, our findings imply that homogeneous freezing will be the dominant mechanism for ice nucleation as even the very active FW200 soot loses their ice nucleation ability with such a thick H2SO4 coating. As such, knowing the ageing history (trajectory) of soot particles in the upper troposphere would be crucial to accurately determine the role of soot particles in cirrus cloud formation. Mahrt et al. (2020a) suggested that cloud processing induced aggregate compaction of soot samples significantly enhances their ice nucleation ability in the cirrus cloud regime. 715 However, if the compaction arises from thick coatings of H2SO4 no enhancement or even a suppression of ice nucleation ability is to be expected in the cirrus cloud regime.

Summary
The ice nucleation ability of two samples, an organic carbon poor (mCASTblack) and a porous commercial carbon black (FW200) soot were investigated. The soot particles coated with H2SO4 to different degrees are systemically studied at T = 243-720 218 K under RH conditions from ice saturation to well above water supersaturation. Auxiliary measurements for soot sample property characterization, including particle mobility size and mass, H2SO4 coating wt %, soot-aggregate morphology and surface chemical composition, were performed. Overall, H2SO4 coating on soot particles can modify soot-aggregate morphological properties depending on the coating wt % and enhance soot-water interaction ability, thereby impacting soot particle ice nucleation abilities. The C3H8 flame fuel-lean soot (mCASTblack), viewed as an aviation soot proxy in laboratory 725 studies, is originally a poor INP at T < HNT. H2SO4 coating makes 200 and 400 nm mCASTblack soot particles form ice crystals via homogeneous freezing and low mass H2SO4 coating even depresses their ice nucleation activity at T < HNT to require RH > RHhom conditions. Porous FW200 bare soot particle and FW200 soot particles with thin or moderate H2SO4 coating can nucleate ice crystals via PCF below homogeneous freezing conditions at T < HNT. However, thick H2SO4 coatings exert a size dependent effect suppressing small size (200 nm) FW200 soot PCF activation more than that of larger size particles 730 (400 nm). Thickly coated 200 nm particles freeze homogeneously requiring a smaller H2SO4 coating wt % than that for 400 https://doi.org/10.5194/acp-2021-645 Preprint. Discussion started: 6 September 2021 c Author(s) 2021. CC BY 4.0 License. nm particles. Based on above results, we propose a three-step H2SO4 coating process on soot-aggregates with increasing H2SO4 coating masses, inclusion of pore filling, H2SO4 spreading over soot particle surfaces and soot particle structure collapse.
As demonstrated in this study, H2SO4 coating changes soot particle morphology and modifies its water interaction ability 735 simultaneously. Differentiating these two coating effects to identify the dominant effect on ice nucleation is still an open question for future studies. In addition to H2SO4 coating, atmospheric ageing altering soot particle ice nucleation abilities can include chemical oxidation, organics coating and clouding processing in the atmosphere. Only a small proportion of studies in the literature have examined these topics systemically. Experimental studies on these soot ageing processes will benefit the understanding of soot role in cirrus cloud formation. 740

Appendix A The mixing state of soot and H2SO4
In order to ensure the absence of pure nucleated H2SO4 particles in the mixing state aerosol flow, the particle size distribution evolution of the aerosol in the coating apparatus was measured. Here, we present the internal mixing process of 200 nm mCASTblack soot with a H2SO4 coating wt % = 15.6 % as an example. As shown in Fig. A1, particle size distribution for the 745 0-minute state of pure H2SO4 aerosol in the coating apparatus (red line) and 200 nm size selected bare soot particles (black dashed line) were conducted firstly and respectively. Afterwards, 200 nm mCASTblack bare soot aerosol was fed into the coating apparatus. The mixed aerosol particle size distribution was monitored until it reached a stable state (20-minutes black line, Fig. A1). In addition, a comparison between the aerosol particle mass distribution of bare soot particles and H2SO4 coated particles was made to further demonstrate that pure nucleated H2SO4 particles are absent in the aerosol flow produced from 750 the coating apparatus (Fig. A2). Figure A1 depicts the particle size distribution evolution process of the mixing state of soot particles with H2SO4. At the very beginning of the measurement, pure nucleated H2SO4 aerosol shows a high particle number concentration and a small size distribution with a mode peak at ~ 40 nm. After 200 nm bare soot particles are introduced into the coating apparatus for several 755 minutes, the mixed aerosol particle number concentration dramatically decreases. With more time, the pure H2SO4 particle size distribution mode is absent and a size distribution peak around 200 nm becomes more distinct, indicating that the mixing state of soot particles and H2SO4 is changing over time. particles. The green line presents the particle size distribution 9 minutes later when the soot particles were fed into the coating apparatus. The blue line shows the particle size distribution 13 minutes later. The black line stands for the stable mixing state of soot particle with H2SO4. Figure A2 shows the mass distribution of 200 nm bare and H2SO4 coated mCASTblack soot particles. The larger mass 770 distribution mode value of coated soot particles doubtlessly demonstrates the adsorption/condensation of H2SO4 on the sootaggregates. The mass distribution curve also provides evidence for the internal mixing state of soot particles with H2SO4, given that the curve only shows one single distinct peak with a log-normal fitting peak value at 1.93 fg. It is conceivable that there are no 200 nm pure H2SO4 particles in the aerosol flow because a 200 nm pure H2SO4 particle has a mass of 7.7 fg.

780
Because of the heterogeneity of soot particles and the poor assumption for its spherical structure, DMA cannot perfectly select monodisperse soot particles based on the electrical mobility size. Therefore, we measured the number size distribution of the size selected bare soot particles using an SMPS system with multiple charged particle correction, downstream of the DMA.
The DMA and SMPS system were operated with the same configurations as for all ice nucleation experiments. For 200 nm soot particles, the DMA was running with a sheath to aerosol flow ratio of 10 : 1 and the downstream SMPS system with a 785 flow ratio of 4 : 1. The system was pulled by a CPC 3772 with a sample flow rate of 1 L min -1 . For 400 nm soot particles, the DMA sheath to aerosol flow ratio was 7 : 1 and SMPS system operating with a CPC 3776 in low flow mode (aerosol sample flow 0.3 L min -1 ) had a sheath to aerosol flow ratio 6 : 1. As shown in Figs

Appendix B The ice nucleation activity of bare and H2SO4 coated soot particles
Soot particles coated by H2SO4 with progressively increasing coating masses were systematically investigated for their ice nucleation activities. The coating thickness was from less than a complete H2SO4 monolayer coverage up to several equivalent 800 monolayers. Figures B1 to B4 present the AF plots derived from 1 μm OPC channel data of all the soot samples investigated as a function of RH (RHi and RHw) and AF curves based on 5 μm OPC channel data are presented in Figs. B5 to B8. These figures show that at 243 and 238 K the signal in the 5 μm OPC channel at RHw < 105% is absent, thus suggesting no ice crystals form at these temperatures and only water droplets are present in the 1 μm channel. This is because if ice can be nucleated at 243 and 238 K, the ice crystals grow sufficiently large to be reliably detected in the 5 μm channel for RHw < 105 805 % but water droplets do not. As such, we can confidently conclude no ice nucleation for our soot particles at T = 243 and 238 K. For both mCASTblack and FW200 soot, at least eight H2SO4 coating wt % were tested for size selected soot particles of a size 200 or 400 nm.           Figure B8. RH scans for bare and coated 400 nm FW200 soot particles at nine different coating wt % corresponding to 5 μm OPC channel, presented as AF as a function of RH. Black solid lines represent water saturation conditions according to Murphy and Koop (2005). Black dashed lines denote the expected RH values for homogeneous freezing at T < HNT (Koop et al., 2000). The grey shading areas show the possible RH uncertainty calculated for water saturation and homogeneous freezing conditions. The percentage number represents the H2SO4 coating mass ratio to the bare particle mass.

860
Appendix C Supplementary SEM, TEM and EDX results to characterize particle morphology SEM and TEM images of low resolutions or with larger scale bars showing an overview of particle morphologies are presented in Figs. C1 (400 nm soot-aggregates) and C2 (200 nm soot-aggregates), respectively. The SEM images at 20k magnification in Fig. C1 show that thickly coated soot particles, both for mCASTblack and FW200, are more compacted whereas thinly 865 coated particles are similar to the bare ones. The same findings can be supported by TEM images at low magnification values in Fig. C2. Additionally, some small residues, surrounding single soot-aggregates in Fig. C2f, are believed to be small H2SO4 droplets generated during the impaction/drying of these soot-aggregates.
However, the visualized particles at low magnifications are larger than the particle sample selected size. As explained in Sect. 870 3.3.1, the most probable reason is the soot-aggregate agglomeration during impaction onto the grid, particularly when the particle concentration is high. This can explain why mCASTblack soot-aggregates in Fig. C1a, b and c present to be much larger than 400 nm, i.e. the size selection value. During TEM sample grids collection, the particle number concentration of 400 nm mCASTblack soot particles was as high as 3,000 cm -3 , providing a high probability for these mCASTblack sootaggregates to agglomerate into larger clusters (> 400 nm). In order to exclude other possible artifacts, the optical size of 875 suspended 200 and 400 nm mCASTblack size selected bare soot particle sample was measured by the OPC, operated in a same manner as for HINC experiments. The measurement aerosol flow, also with a particle number concentration ~ 3,000 cm -3 , was just sampled upstream of ZEMI. Each OPC channel logged the number of particles larger than the channel threshold value in a 5 s interval. After measuring for 5 minutes, the percentage of 200 and 400 nm mCASTblack bare soot particles in different OPC counting channels could be calculated. As shown in Fig. C3a, only 2.08 % of 200 nm size selected mCASTblack soot 880 particles are detected with an optical size larger than 0.3μm, suggesting larger soot-aggregates in Fig. 9a    HR-TEM images are also collected to determine how the H2SO4 coating changes the fine structures of soot-aggregates and the network among primary particles, i.e. soot spheres showing a distinct onion-like structure comprising of graphitic layers in Fig. C4. In order to avoid the background noise from the carbon film on the Cu grid, soot-aggregates hanging outside of the 905 mesh edge (i.e. in the holes) were selected to obtain these HR-TEM images. The primary particle profiles in these 2D projected https://doi.org/10.5194/acp-2021-645 Preprint. Discussion started: 6 September 2021 c Author(s) 2021. CC BY 4.0 License. soot-aggregate images are still distinguishable in Fig. C4d for bare and Fig. C4e for thinly coated soot particles, but single primary particle edges are more ambiguous in Fig. C4f for thickly coated soot particles. The similar results can be seen from Fig. C5, in which the outline of FW200 soot primary particles with thick coating (Fig. C5f) is smoother than those of bare and thinly coated particles (Fig. C5d and e). This can be attributed to the pore filling effect of H2SO4 coating, which fills in voids 910 or wedges among primary particles and reduces the roughness of the soot-aggregate surface.

960
Data availability: The data presented in this publication will be made available at DOI 10.3929/ethz-b-000498786. Note by the authors: Data and DOI link will be activated for public access upon acceptance of publication.
Author contributions: KG and ZAK designed the experiment and interpreted the data. KG wrote the manuscript and prepared the figures with contributions from ZAK. EJBM collected SEM, TEM and EDX results. All authors discussed and reviewed 965 the manuscript. ZAK supervised the project.
Competing interests: The authors declare that they have no conflict of interest.

Acknowledgements:
We are grateful to Franz Friebel who originally designed the coating apparatus and to the experimental 970 atmospheric physics group at ETHZ for their help with instrument trouble shooting. Franz Friebel is thanked for his valuable comments on the initial manuscript. The authors also would like to thank Fabian Mahrt for sharing his HINC data processing code. This work was supported by China Scholarship Council (Grant No. 201906020041) and the atmospheric physics professorship at ETH.