Regularities of new particle formation and evolution of existing atmospheric aerosol particles in a large (3200 m) isolated volume

The paper reports on an investigation of nanometre-sized new particles formation (NPF) in aerosol-free outdoor air. This phenomenon was observed after filling of Large Aerosol Chamber (LAC) RPA “Typhoon” with the volume of 3200 m with outdoor air, passed through HEPA 13 class filter (H13). During the summer-autumn period of 2018, even in the full darkness and in presence ionizing radiation only in the shape of secondary galactic cosmic rays, new particle formation with the particle 5 size greater than 15 nm starts 0.5–1 hour after the end of LAC filling. During the 2018–2019 winter periods the NPF event was not observed once only. Approximately one day after NPF narrow ”bell-shaped” spectra with number concentration up to 10 cm−3 and mass concentration up to 0.6 μg per m are formed. During the next five or more days, these size distributions evolve due to coagulation, while their asymptotic shape remains constant with relative count standard deviation σc ≈ 0.28, and relative asymmetry 10 ras ≈ 2 (ras= skewness/σc). The value ras ≈ 2 defines the analytical description of the size distribution as the gammadistribution. During additional purification of newly formed particles with the inner H13 filter, aerosol particles concentration in LAC decreases down to a few particles per cm. This concentration remained constant for more than a week. This demonstrates that new aerosol particles are formed by homogeneous gas-to-particle conversion of gaseous precursors, which passed through 15 the external H13 filter. The mass concentration of newly formed particles depends on the concentration of precursors. It was found that after filling LAC with outdoor unfiltered air, approximately after 10 hours the left-hand side of aerosol particle size distribution below 15 nm disappears, and after several days there forms an asymptotic ”bell-shaped” size spectrum with σc ≈ 0.4–0.5 and ras= 2–3. The modal diameter becomes about 150 nm after five days, while the size distribution greater than 200 nm remains unchanged. This allows concluding that aerosol particles greater than 200 nm have a life-time of more 20 than five days, while particles smaller than 15 nm, not more than five hours. 1 https://doi.org/10.5194/amt-2020-172 Preprint. Discussion started: 29 September 2020 c © Author(s) 2020. CC BY 4.0 License.

of the measurements was limited to 2 hours. It should be noted that another important result obtained with the using of this chamber is the revealing of the amorphous nature of organic aerosol particles (Virtanen et al., 2010).

Using of a CLOUD (Cosmics Leaving Outdoor Droplets) camera built in CERN (Switzerland) with Proton Synchrotron
with a volume of 26.1 m 3 with walls of electro-polished stainless steel, a description of which along with the first results is in the work (Kirkby et al., 2011), is found to be more suitable for obtaining quantitative results. In this chamber, a temperature 70 from 183 K to 300 K can be controlled with an accuracy of 0.01 K, and 373 K for cleaning the walls from contaminants can be created. The air in the chamber is created by the evaporation of liquid nitrogen and oxygen. A calibrated amount of conversion gases SO 2 , NH 3 , and other organic and inorganic gas components are added to the chamber. The ionic compound in the chamber is regulated from zero concentrations by creating a constant electric field to a natural background from galactic cosmic rays and additional ionization of the air from the Proton Synchrotron π + -meson beam. The solar radiation is simulated 75 as well. Note that due to the processes of the influence of the walls, the duration of the studied processes in this chamber was limited to five hours. This required the constant addition of the studied gas mixture. From the results presented in this and subsequent works, the conclusion is drawn about the NPF mechanism as "ternary nucleation with ions playing a major but subdominant role" (Kirkby et al., 2016), (Dunne et al., 2016). It is also fixed that the compounds of CG are a complex system of organic and inorganic molecules without a description of their specific form. Note that this chamber does not provide the 80 injection of outdoor air.
Meanwhile, when studying the processes of NPF in various outdoor conditions, it shows the diversity of the CG composition.
Thus it was shown  that in the Amazon region with no anthropogenic pollution, the major source of the condensing material is organic compounds, emitted by plants. Nevertheless, the ratio of biogenic and anthropogenic sources of these precursors in the atmosphere is still poorly understood (Lehtipalo et al., 2018). This paper also states that this ratio will be 85 very important in the future due to the decrease in emissions from fuel fossil combustion. The complexity of the composition of CG and the interaction mechanisms of individual components is indicated in a recent work (McFiggans et al., 2019), which states "that formation mechanisms of secondary organic aerosol in the atmosphere need to be considered more realistically, accounting for mechanistic interaction between the products of oxidizing precursor molecules".
In connection with the above circumstances, it is very important to study the formation and evolution of NPF processes with 90 CG in outdoor air in much larger volumes, where the influence of walls can be minimized. As such a volume, the question arises about the using of the 3200 m 3 Large Aerosol Chamber (LAC) available at the RPA "Typhoon", which provides for its filling with external air. Till recently, aerosol research was not possible due to the results of the research of (Smirnov et al., 2000). This work stated the existence of the "chamber" aerosol with the concentration of about thousands of particles per cm 3 in LAC and similar chambers. At present, the chamber is equipped with two (external and internal) HEPA13 aerosol filters 95 (HEPA, 2019). As a result of these modifications for cleaning outdoor and indoor air from aerosols, it is possible now to reach almost zero aerosol concentration in LAC for a very long period (more than two weeks). It means that chamber walls and equipment, installed inside the chamber, do not produce aerosol particles, as was suggested earlier. The large volume of the chamber allows eliminating wall influence on particle sedimentation. Thus, in LAC it is possible to study processes of atmospheric aerosol formation and evolution in close to natural conditions with obtaining asymptotic patterns. In particular, in 100 LAC, the processes of nucleation formation of new aerosol particles from the introduced gas impurities can be experimentally studied with a trace of their evolutionary development, as well as the evolution of atmospheric aerosol in the absence of new particles source. The subject of this research is the experimental investigation of the mentioned above processes.
The obtained results can clarify some stages of aerosol dynamics in the real atmosphere. They may be also useful in the design of nano-aerosol filtration techniques in industrial and medical facilities. Nanoparticles may influence the human organisms 105 as it is indicated, in particular, in (Youssefi and Waring, 2014).
2 Design of Large Aerosol Chamber (LAC) and description of experimental conditions

LAC device description
The largest cloud investigation chamber was built in 1954 in Texas, USA (Gunn and Paul, 1954). It has a spherical shape with an internal diameter of 60 feet (volume 3200 m 3 ) with a steel wall thickness of one inch. It was first used to study the dependence 110 of aerosol concentration on atmospheric aero-ion concentration (Phillips et al., 1955). Currently, publications on its use for meteorological purposes are not known to us. In Russia, a similar cylindrical chamber with the same volume 3200 m 3 was built in the town Obninsk in 1964. It is an integral part of the complex of facilities of RPA "Typhoon", designed for cloud and meteorological processes investigation. The first detailed description of its design and thermodynamic characteristics is given in (Romanov and Zhukov, 2000). This description is cited, in particular, by authors on the use of poorly soluble hygroscopic 115 substances for modification of warm clouds and fogs (Romanov et al., 2006), (Drofa et al., 2010) and regularities of cloud microstructure formation and evolution (Romanov and Erankov, 2013). However, due to the inaccessibility of the above work (Romanov and Zhukov, 2000) for English-speaking readers the description below partially duplicates the content of this work.
It describes the possibilities of studying the properties of atmospheric aerosol, which has become possible recently due to the retrofitting of its modern equipment. 120 in the building intended for the aforementioned complex and is surrounded by adjacent working premises isolated from each other. The wall thickness of the chamber is 6 mm, which allows it to withstand excess pressures in 0.07 MPa produced in about 50 minutes by water-packed ring compressor. By reducing this pressure through a system of seven valves, a simulation of the formation a cumulus cloud with an adjustable rate of air mass rise is organised. Filling the chamber with outside air is carried out by sucking it through the windows of the premise adjacent to the chamber and the inlet using an exhaust fan 130 located in the upper part of the chamber through a valve with a nominal diameter DN = 800 mm. In this case, the complete replacement of the internal air by the external environment takes place over a time of ≈ 0.5 hours. Since 2014, it was possible to clean aerosols of air drawn in through the chamber by installing a HEPA13 (High Efficiency Particulate Air) class air filter at the entrance to the vestibule with dimensions equal to the dimensions of the entrance vestibule. At the same time, the time for replacing air in the chamber with the air cleared of aerosol increases up to about one hour. A similar air filter with its fan is 135 also available inside the chamber. During its operation for ≈ 2 hours, the aerosol concentration inside the LAC decreases by about three orders of magnitude.
Unfortunately, the appearance of the chamber cannot be represented using photographs, because, as mentioned above, it is located inside another building. Therefore, we have to confine ourselves showing its internal appearance, a photograph of which is presented in Fig. 1.

140
Note the presence in the chamber of an aero filter of class (High Efficiency Particulate Air) HEPA13 (position 4) with dimensions equal to the dimensions of the entrance vestibule. As mentioned above, the second same air filter without a fan (external H13 air filter) can be installed on the outside of the chamber. The lanterns located on the walls of the chamber are used for commissioning and are usually turned off during experiments.
A description of the cloud droplet size meter developed by RPA "Typhoon" and located on suspension platform 3 is available 145 in (Drofa et al., 2010). To measure the size of condensation nuclei, they also used a laser particle size meter developed at RPA with a lower diameter limit d l = 0.2 µm. In 2014, this device was replaced by SMPS Scanning Mobility Particle Sizer model 3936L88-N(TSI Inc.), based on the measurement of electro-mobility particles, which can be flooded with water vapour at a supersaturation of the order 0.5 %. The size region used in these experiments was from 15 to 1000 nm with 115 measurement channels. The air from LAC is sampled through a 2 m long zinc coated steel tube with an inner diameter of 18 mm. A similar 150 tube for outdoor air sampling goes through a window near SMPS. The influence of these tubes on the measured particle size was considered during the data processing according to the SMPS manual. Figure 1. Photo of LAC interior view. 1 -entrance portal with size 60 × 160 cm 2 ; 2 -rail for mounting the tested equipment; 3 -hanging platform with temperature sensors (for measuring dry and wet air), optical depth sensor with 4 m optical path and a photoelectric sensor for measurement of cloud droplets size; 4 -inner HEPA13 filter.

Features of the interaction of the air with the walls of the chamber
The integral heat capacity of the walls and structures adjacent to them exceeds the integrated heat capacity of air by 10-15 times.
When simulate the cumulus cloud formation by reducing the previously created overpressure, the temperature difference be-155 tween the wall and air usually varies from plus 5 to minus 10 K. In (Romanov and Zhukov, 2000), analytical relationships for the heat transfer rate in these processes are determined with the assumption of the convective mechanism of this process. Subsequent more detailed qualitative assessments of these processes show that the heat transfer consists of 3/4 with IR emission 6 https://doi.org/10.5194/amt-2020-172 Preprint. Discussion started: 29 September 2020 c Author(s) 2020. CC BY 4.0 License. exchange of chamber walls with water vapours and 1/4 convective heat exchanges. The convective heat exchange takes place in form of ascending (descending) air flows in a 5-10 cm thick layer near vertical walls, and in form of horizontal movement 160 near the floor (Romanov, 2009). As a result, a homogeneous temperature is established in the entire horizontal layer of the chamber (except for the regions near walls).
In stationary conditions without excess pressure, ideally, the air temperature in the chamber should coincide with the temperature of its walls in the absence of any air movements. However, measurements using a GLL acoustic anemometer show the presence of turbulent air flows at any time the chamber is in a closed state without any disturbing influences. An example 165 of such flows measured is presented in Fig. 2. 7 https://doi.org/10.5194/amt-2020-172 Preprint. Discussion started: 29 September 2020 c Author(s) 2020. CC BY 4.0 License.

LAC radiation and ionization state
To study the new particle formation processes, it is important to know the concentration of aero ions, which can play a noticeable role in their nucleation. However, according to the authors of the work (Wagner et al., 2017), the speed of new particle formation (NPF) under their influence cannot exceed the speed of ion pairs (i.p.) formation under the influence of ionizing radi-175 ation. For the chamber used in this work in CERN, this speed under the influence of natural radiation was 2 i.p. per cm 3 per s.
To determine this characteristic, during the second half of 2019 and January 2020, several measurements series of 7-10 days of dose rate (DR) in LAC with a gamma radiometer using gas Geiger-Muller counters were carried out. It turned out that all the results fit into the range DR = 8-10 µR per h. Hence, using the definition of X-ray R = 2.06 · 10 9 i.p. per cm 3 per s, we obtain the ionization rate in the chamber air at normal pressure ≈ 5 i.p. per cm 3 per s. DR measurements in the room near 180 LAC and outside the building usually exceeded the indication in the chamber by ≈ 20%. The explanation for a small decrease in DR in the chamber is that in the lower atmosphere the ionization of nitrogen and oxygen molecules is mainly responsible for muons, which are formed when primary galactic protons collide with nitrogen and oxygen nuclei at altitudes of 10-20 km (Dorman, 2004). Highly penetrating muons are not filtered by LAC walls, but, at the same time, they can ionize gas molecules.
In this work, a detailed study of the aeroion formation is not conducted in the chamber, as it requires long-term experiments 185 with a variety of different parameters. A preliminary conclusion follows from several conducted experiments that the resulting ion concentration in the chamber mainly depends on the aerosol concentration and is established within a few hours.
So for a single experiment with the aerosol concentration changing from 7 · 10 3 cm −3 to 5 · 10 1 cm −3 with using internal H13, it was found that the concentration of light ions with mobility ≥ 0.4 cm 2 per V per s was established for positive ions at a level from 4 · 10 2 cm −3 to 4 · 10 3 cm −3 correspondingly.

Method of LAC cleaning from aerosol particles
At the beginning of this section, we note that the process of developing technology to achieve aerosol concentrations in LAC of up to several particles per cm 3 has gone through several stages of understanding the process of formation and transformation of aerosol. On the first stage after about one hour of LAC filling through the external HEPA filter, the air in the chamber is almost completely (up to several tens in cm 3 ) substituted by filtered air mass from aerosol particles, greater than 15 nm (this is 195 the lower limit SMPS d l for the used). However, it was found that after a while, new particles with the size near SMPS lower detection limit appear. These particles then form a narrow "bell-shaped" distribution with concentration about 10 4 cm −3 and with an obvious absence of particles from the small sizes side.
To clarify the origin of these particles, we used the procedure for removing this aerosol with an internal H13 aerosol filter.
At the same time, it turned out, after its operation for about two hours, that the aerosol concentration decrease almost to 200 zero (not more than a few particles per cm 3 ), which persists for an indefinitely long time. A more detailed description of aerosol formation processes and size spectra transformation will be given below. Here, in Fig. 3 we give the pictures one of the cycles. Dots refer to the beginning of 10 minutes cycles of size spectra measurement start. The time before t 1 refers to the measurements of outdoor aerosols. It can be seen from Fig. 3 that particle number concentration decreases by more than two orders of magnitude after the 205 external aerosol filter. Further increasing after about one hour can be explained by NPs, formed by nucleation of a specific gas component, passing through the external H13 aerosol filter. Naturally, the NP occurs at sizes below d l . In the process of increasing their size due to Brownian coagulation, the particles become greater than the SMPS lower detection limit and are included in the total number of particles recorded by the device. This can explain a new particle concentration increasing.
Registration of the full spectrum with our equipment is possible only after ≈ 20 hours of evolution (time moment t 4 ) when its 210 left-hand side passes the border of d l . At time moment t 2 switching on the internal aerosol filter for ≈ 2 hours removes these new aerosols, and particle concentration in LAC decreases to 10-20 particles per cm 3 . This concentration remains at this low value for more than 300 hours (shown in Fig. 3).
Such low concentration (almost zero), reached after the second filtration, was observed in many experiments. This low concentration value remains almost constant for more than ten days. Switching on kitting, including agitating fan (not shown 215 in Fig. 1), does not influence on aerosol concentration. It follows, that chamber walls and equipment inside do not generate aerosol particles.

The discussion of the results
The data presented above leads us to conclude that the material for NPF in the LAC is gas component contained in the outdoor air, which will be called "conversion gas" (CG), passed through an external H13 filter. A specific property of this gas fraction 220 is its ability to completely transform into new aerosol particles without any external influences. We do not know the works where such a phenomenon would be observed. In what follows we will call it "gas-particle conversion" (GPC). It was not the knowledge of GPC phenomenon that led the authors of the work (Smirnov et al., 2000) to conclude that there is a chamber aerosol in LAC. The erroneous statement obtained in this work about the existence of a chamber aerosol is based on the fact that a corona discharge, used previously for LAC cleaning from aerosol particles, produces condensing material for new particle 225 formation itself, as shown in (Lapshin et al., 2012) and approved by our LAC experiments. We also note that the appearance of new particles in enclosed spaces (Hirsikko et al., 2007), which the authors attribute to their generation by an unknown internal source, is determined by GPC process.
In this paper, we will not discuss the mechanism of GPC due to the extreme complexity of this phenomenon. However, the results obtained after research and evidence of its existence open up a huge number of research areas in solving meteorological 230 problems.
This paper presents the results of the following areas: -Investigation of the possibility of an analytical description of the asymptotic spectra of NP in the process of their coagulation evolution. In this case, the dependence of the mass concentration of CG on the time of day and season is also investigated (see Section 3).

235
-Investigation of the evolution of naturally occurring aerosol spectra in the absence of replenishment with new particles (Section 4).
-Investigation of the asymptotic regularities of the spectra of other origins (Section 5).
Soon, the following works are planned: -Investigation of aerosol particles spectra, produced in LAC by corona discharge in aerosol-free air.

240
-The study of the ability of various substances to NPF by the injection of test gas impurities into the aerosol-free LAC.
-Investigation of the dependence of the atmospheric ions concentration on the concentration of aerosols, which is regulated using an internal H13 filter.
3 Investigation of nucleation aerosol formation and evolution As pointed above, after replacing the air in the chamber by passing through an external air filter, after some time, aerosol 245 particles begin to appear from the SMPS lower detection limit, which then evolves. Investigation of this phenomenon started in 2015 after the installation of the H13 filter in LAC. Since that time 15 experiments in the summer-autumn period and 8 in Because of minimizing the attended factors, the investigation results of such a process can be as an experimental basis both for understanding the NPF mechanism and for developing theoretical models of this process. For this, the first subsection sets out the analytical definitions and relationships that will be used to present experimental results.

Data presentation format
According to coagulation and sedimentation laws rate of number concentration N (t) change proportional to the second and the 255 first degree of this quantity, where K and D are coagulation and sedimentation coefficient respectively (Fuchs and McDonald, 1964), (Anand et al., 2012), (Salimi et al., 2017), (Friedlander, 2000). This can be described by the differential Eq. (1).
The solution of Eq.
(1) can be presented as: Note, that in the work (Smirnov et al., 2000), the solution of the Eq. (1) is presented with a constant member on the righthand side, which defines the particle inflow from an extraneous source. But we do not need such a solution, and therefore it is not given here.
(2) can be transformed into a simpler form: Note that expressions Eq.
(3) imply that K and D do not depend on particle size. Typically they are used for t ≥ t 1 to forecast dependence of N (t). However, they may be used for time t 0 < t < t 1 of N (t) behaviour reconstruction on the early stages of coagulation spectra formation, which cannot be measured by instruments. Value t 0 can be defined from the condition when denominator in expression Eq. (3) is zero.
Obtaining a more complex expression for t 0 based on expression Eq. (2) does not make sense because it will lead to the same numerical value using the limit transition procedures.
The above expressions will be used for comparison of experimentally measured N (t) dependencies with theoretical ones.
For a more detailed description of experimental size distributions we will use not only commonly used distribution function n(d) (Baron and Willeke, 2001), but normalized distribution function f (d) = n(d)/N moments. This m(n) moments are 275 described by integrals Eq. (5), in which absent integration limits from 0 to inf just for writing simplicity.
At that m(0) =1, and count median diameter d m = m(1). Other relative dimensionless parameters are defined by the following expressions: Relative count standard deviation : Asymmetry coefficient (kas): Relative asymmetry (ras) is another useful parameter: Equation (6) and Eq. (7) may be presented in the form of second and third combination moments respectively.
Let's consider the most simple and widely used two-parameter distribution functions for experimental data approximation.These are the gamma (g), the log-normal (ln) and the Smirnov ( It follows that the gamma distribution has the smallest relative asymmetry ras = 2, which is independent of other distribution parameters. The form of the gamma distribution is given below: Here u = d/d m ; µ = 1/σ c 2 . When ras > 2 one can use a sum of other distributions, with weights, defined by Eq. (9). At a qualitative level, it was found that new particles are characterized by a large variety of mass concentration and narrow size distributions during the summer-autumn period. Special attention has been paid for this period due to its importance to the understanding of new particle formation processes. The period of the cyclonic synoptic situation on August 30th, 2018 was studied in detail, when LAC was filled the air mass with a temperature of 20 • C and 12 g per m 3 of water content through an 300 external HEPA filter at 12 h 17 min local time. The investigation of new particle formation and evolution was carried out for six days. The obtained experimental results are presented in Fig. 4-7. Figure 4 shows the time dynamic of aerosol spectra. The size distribution of outdoor aerosols with the concentration equal to N ext ≈ 4 · 10 3 cm −3 is marked with external. Index t = 0 corresponds to the start of measurements, and this time moment 305 coincides with the end of LAC filling with the purified air. The aerosol concentration at that moment was N 0 = 27 cm −3 .

Formation and evolution of new particles size distributions
Since at this moment the particles were registered only in separate channels, the curve t = 0 is presented as separate marker points. Next indexes refer to the period in hours from the end of the chamber filling to the time moment when the measurement started. Figure 4 shows that after 1/3 hours, the formation of additional aerosol with size above the detection limit in 15 nm is

The evolution of particle number concentration
Results of the measuring of number and mass concentration related to Fig. 4 are presented in this section.

325
Lines with marks in Fig. 7a represent measured concentration in the range from 15 to 1000 nm. Blue solid line is N 2 (t) calculated from Eq. (2) with reference point N (t = 20) = 4200 cm −3 and with coefficients K = 9.05·10 −6 (cm 3 per hour) ≡ (1), is presented by the red line. The insignificant difference between N 2 (t) and N 3 (t) one can make from the qualitative comparison. Increasing N 3 (t) over N 2 (t) takes place in t = 3 hours and quantitatively has the value in 12 % as well as decreasing takes place in t = 160 hours and has the value 25 %.
The mass concentration m was calculated from experimental data, assuming that particles are spherical, with a density equal to 1 g per cm 3 . The dashed line in Fig. 7b represents these values. As seen from this figure, the mass concentration Here, m(t 0 = 20h) = 0.44 µg per m 3 .
The comparison of the data in this figure shows that under the above assumptions about the sphericity and constancy of the NP density taking into account the sedimentation although leads to some decrease in the measured and calculated difference 345 in mass NP concentration, but does not explain the very large its value (about in 2.5 times for the five days).
The considerations may be explained by the following hypotheses: aerosol ageing, resulting in its structure compaction; evaporation of aerosol products in the time scale of several days; overestimation of the small particle size with SMPS.

350
It is clear that the considerations expressed above are hypothetical in nature however in the current state of NP nature knowledge, one will have to dwell.

Dependence of NP characteristics on day time
The measuring of the left-hand side of the size distribution (15 nm) has never vanished. It means that there is a source of condensing products at least of hourly availability. One of these sources may be plant emissions. The efficiency of such a source  Table 1.
The data, presented in Table 1, shows that maximal aerosol concentration (0.6 µg per m 3 ) of conversion gases, which is determined by the NP mass, is observed in air, which was sampled into LAC at day time. When air is sampled in the evening 360 or at night, the NP concentration decreased to 0.2 µg per m 3 and 0.09 µg per m 3 respectively. Next, we can note the almost complete coincidence of the ras values in the first and last two rows of Table 1. The overestimation of this value for the second row can be attributed to insufficient time delay. So, the values σ c = 0.28 and ras = 2 may be considered as universal asymptotic values and used in modelling of formation and evolution of conversion aerosol. The size distribution of the formed particles is narrow, therefore the estimated above sedimentation (5 %) will not influence on aerosol size spectra, and the gamma-

Experimental results of aerosol in winter time
Winter aerosol should differ from summer aerosol as noted above. The difference is primarily defined by the absence of a biogenic CG source, which is plant emission. A similar effect was observed during the investigation of NP formation in the 370 Antarctic (Jokinen et al., 2018), in which new particles were detected when the air masses were coming from the open ocean, and there were no particles detected when the air masses were coming from mainland or ice fields. The second difference is low absolute air humidity value at negative temperatures. These low values remain after air warming in LAC, while relative humidity decreases significantly.
We studied four situations during winter 2019 when there was solid snow in the region, gentle breeze, and stable negative 375 temperature. The results of these studies are presented in Table 2. The first episode on 10/01/19 with low negative temperature corresponds to the edge of an anticyclone with isothermal temperature profile in 300 meters layer after small snowfall. The number and mass concentration of outdoor aerosol was low.
During this situation, there was no NP formation in LAC for 22 hours of observation. The other episodes refer to relatively low negative temperatures with sufficiently large number and mass concentrations of aerosol. They differ in initial temperature, 380 water content and time on LAC filling, while characteristics of the formed in LAC particles with mass concentration less than 0.1 µg per m 3 are almost similar.

Results discussion
Before the discussion, it should be noted that the process of LAC filling with air filtered by H13 filter, passed CG, takes about one hour. The process of new particle formation from CG, their coagulation and coagulation with pre-existing in LAC aerosol 385 particles, as well as their and CG removal from LAC takes place during the filling. Without going into complex computational concentration starts to decrease in time. It means, that in fact, we observe coagulation of already formed new particles.
From the results presented in Fig. 5b and Fig. 6, it can be seen that, in less than a day after the start of the NP formation process, their spectra assume an asymptotic form that coincides with the gamma distribution. Note that this conclusion we have for the first time. Indeed, the formation and asymptotic behaviour of coagulation spectra is the subject of numerous theoretical studies (Fuchs and McDonald, 1964), (Friedlander, 2000). One of the last publications on the subject is (Anand et al., 2012).

395
However for atmospheric aerosols, the theoretical study of such dependencies is complicated by uncertainty in the description of the coagulation nucleus for specific situations. Until now, the modelling of such processes has been complicated by the influence of small chamber walls. An illustration of this approval is the work (Seipenbusch et al., 2008), in which for a chamber of 2 m 3 the results of experimental and theoretical studies of the platinum aerosol evolution with an average diameter of about 7 nm in a clean chamber and the presence of a simulating atmospheric aerosol are given. Obtained in this work the results of 400 aerosol coagulation up to sizes of about 100 nm for a time of up to 250 minutes had to be obtained with continuous operation of the aerosol source due to diffusion of the aerosol onto the walls. An important direction in studying the properties of the atmospheric aerosol is the fitting of analytical ratio for approximating the size distribution function, which is the critical aerosol characteristic in modelling both cloudy and general climatic processes. The question of the analytical expression of aerosol spectra was studied in detail in (Tammet and Kulmala, 2014), where two-power law four-parameter distribution is proposed.

405
The results are obtained with the participation of the authors, who provided the measurements of aerosol spectra. It also points out many previously proposed models, one of which is the sum of three lognormal distributions (John, 2011).
The above results show that, at least for NP formation, the gamma distribution is most suitable.

Evolution of the outdoor aerosol in LAC
The asymptotic behaviour of aerosol size spectra in the absence of its source is a scientific interest. The understanding of the 410 atmospheric aerosol evolution in living space or working area, closed for outdoor air, is the practical interest. Therefore we conducted several experiments to study the evolution of various types of atmospheric aerosols in LAC. The outdoor aerosol is sampled into LAC with exhaust fan till the moment when the size spectra of outdoor aerosol and aerosol in LAC coincide.
LAC is closed for more than 7-10 days after filling. Special port in LAC with HEPA filter is used to compensate for internal and external pressure.

The aerosol spectra time evolution for various isolated air masses
Spectra evolution for four different air masses is presented in Fig.8a -Fig. 8d. The presented data show that at the moment of measurements start, there are particles smaller than 15 nm. After about 10 hours they coagulate with the formation of particles greater than the lower detection limit. After about 24 hours aerosol size distribution transforms into a "bell-shaped" distribution with a maximum at about 70-100 nm. During the next five days Time dependences of relative parameters σ c and ras of the above presented spectra are illustrated in Fig. 9a -Fig. 9d. From the data in Fig. 9, it follows that relative asymmetry value is constant during all the time of observation. It is close to ras = 3 for an anticyclone (Fig. 9a), and about ras = 2 for a cyclone, a dish, and an anticyclone with smoke. The value of σ c 425 reaches its asymptotic value of 0.4-0.5 after approximately 100 hours of aerosol evolution for all presented cases.

Time dependence of number and mass concentration of outdoor isolated aerosol
One can see, that for the presented in Fig. 8 and Fig. 9 data, isolated in LAC aerosol size distribution parameters change significantly during their evolution. From theoretical studies (Friedlander, 2000) follows that the coagulation coefficient (core) depends greatly on the particle size ratio. This makes almost impossible to solve coagulation equations analytically. Numerical 430 solving of the coagulation problem also possesses great uncertainty, because coagulation core depends greatly on particle shape. And finally, even the solutions found will not always be worth the effort spent on them due to the diversity of natural situations. As was demonstrated in the previous section, the assumption on the constant value for the coagulation coefficient can describe N (t) for narrow enough size distributions. Therefore we studied the possibility of using simple Eq.
(2) and Eq. (3) to describe the evolution of outdoor aerosol. It turn out that for the presented in Fig. 8 N (t) dependence may be described by 435 Eq.
(3) with two constant values for coagulation coefficient K < 24 and K > 24, for the period less than one day and more than one day respectively. These values presented in forth and fifth columns of Table 3. From the data in Table 3  experiment (c), they are about two times higher. This may be due to the difference in atmospheric aerosols for these years. We are not present a graphical comparison of experimentally observed and calculated from Eq. (3) with coefficients from Table 3 440 values because practically the parameters fitting should be performed for each meteorological situation separately. Note, that the difference between observed and approximated values does not exceed 5 % in the time scale from one to about ten days.
Equation (3) implies that particle mass is constant with time. However, the experimentally measured mass concentration m decreases significantly (about 1.5 times for five days). Equation (2) also cannot describe the time behaviour of aerosol number and mass concentration, even at t > 24 hours. The obtained data do not allow suggesting time dependence of K(t) and D(t), 445 therefore we offer the following expression to describe experimentally observed m(t): A similar dependence for the number concentration was given in (Voloshchuk, 1984) with some suggestions about coagulation nuclear. This expression has no physical ground, but it is simple. The values for α are shown in the last column of Table 3.
The discrepancy of experimental and calculated from Eq. (12) values in Table 3 does not exceed 10 % for the whole time range 450 at t 1 = 24 hours.
Taking into account the individual nature of the coefficient α values are given in Table 3, it can be noted that the coefficient α is useful for the aerosol mass evolutionary changes estimation in a specific large volume.

Experimental results on WC-50 aerosol evolution
One of the nanoparticle generation methods is spraying of salt aqueous solutions solution, upon drying of which nanosized 455 particles are formed. LAC has ideal conditions to study the evolution of such aerosol. On the other hand, the interpretation Table 3. The values of the number concentration at the initial moment N0 and after 24 hours N24; K -coagulation coefficient in the period before and after 24 hours; mass concentration m24 and coefficient α in Eq.(12) for data, presented in 8a-8d.
It is seen from Table 4, that relative count standard deviation values stabilize after the first day and reaches a value of σ c ≈ 0.4, and relative asymmetry tends to the value about 2 after three days. So, the gamma distribution may be used for the sprayed aerosol particles. Note, that the decreasing of mass concentration, which was observed in the above described experiments, is also was observed for these aerosol particles.

Conclusion
The major results of this research may be divided into two parts. The first one is that the upgrading of Large Aerosol Chamber (LAC) of RPA "Typhoon" with the volume of 3200 m 3 with HEPA 13 filters, made it possible to study gas-to-particle formation processes. New particle formation was observed after filling LAC with filtered air by an external filter.  practically zero aerosol concentration in LAC for a very long (more than a week) period. On the one hand, this fact indicates that all the amount of CG, pumped in LAC, is the source of NP. On the other hand, this fact indicates that the chamber walls and installed in the chamber equipment do not generate aerosol particles. So, after secondary removal, it makes possible to 480 study processes of aerosol formation and evolution in well controlled conditions. Large chamber volume allows for avoiding significant particle wall and floor sedimentation. The data presented above indicate the necessity to use the two-stage procedure for cleaning of working areas, with a time gap enable for gaseous precursors to form new particles, which are removable by H13 filter.
The second part of the opportunities pointed above is the realisation the aerosol formation and evolution processes investi-485 gation. The process of homogenous aerosol formation and evolution due to gas-to-particle conversion has been studied. It was found that almost full spectrum with sizes greater than SMPS lower detection limit (d = 15 nm) appears on the coagulation stage in 2-3 hours after filling LAC the purified outdoor air. At the same time, the NP concentration exceeds the value of about 10 4 particles per cm 3 and very quickly lowers due to Brownian coagulation. Naturally, in the period of new particles nucleation that is inaccessible for complete measurements, their concentration far exceeds the value indicated above, which, in turn,

490
almost an order of magnitude exceeds the background concentration of aeroions, equal to (2-4) · 10 3 per cm 3 . Therefore, we can say that in our case aeroions do not play the main role in the NPF process. A similar statement exists in the work (Kirkby et al., 2016), where it is said that the aeroions do not play the main role in the NPF process.
The NP spectrum of this process of homogenous aerosol formation and evolution due to gas-to-particle with the constant mass concentration transforms in the narrow ''bell-shaped" spectrum with the concentration ≈ 3 · 10 3 per cm 3 about a day 495 later. During the next days (five or more) these size spectra transform due to Brownian coagulation, while their shape remains constant. The relative count standard deviation σ c for the summer-autumn period is about 0.28, and relative asymmetry ras ≈ 2.
The last value determines the analytical approximation of this asymptotic form as the shape of a gamma distribution.
The NP mass concentration calculated by measured spectra stays almost constant during the first two days, which allows correlating it with the mass CG concentration. The mass concentration calculated by measured aerosol spectra (in the assump-500 tion that specific density equals 1 g per cm 3 ) varies from 0.09 to 0.6 µg per m 3 for night and day time respectively during the summer-autumn period. For winter conditions, it does not exceed 0.1 µg per m 3 and does not depend on the time of a day.
The decrease in aerosol concentration at night is probably caused by the decrease in plant emissions. Note, that new particle formation at night was observed in Australia during episodes when air masses came from the ocean (Salimi et al., 2017).
The second research was to investigate the aerosol spectra evolution of outdoor (not filtered) air, isolated in LAC. It was 505 found that particles smaller than 15 nm disappear after 7-10 hours in typical meteorological situations. Asymptotic a "bellshaped" curve with σ c ≈ 0.4 -0.5 and ras = 2 -3 formed in LAC after several days of evolution. Spectra with value ras = 2 can be approximated by the sum of the gamma and the log-normal distributions with appropriate value weight coefficients for measured ras. After about five days the modal diameter reaches the value of about 150 nm, while the part of the size distribution for particles greater than 200 nm has the same shape. It follows, that particles smaller than 15 nm are 510 less than 10 hours old, and in anticyclone conditions with the wind velocity about 1 m per s, these particles are formed at the distance, not more than 40 km. So, they are formed from local sources of gaseous precursors. Such particles smaller than