Characterization of a Smog Chamber for Studying Formation and Physicochemical Properties of Secondary Organic Aerosol

An indoor smog chamber facility has been built for carrying out secondary organic aerosol (SOA) formation and for studying physicochemical properties of SOA. This facility comprises of ~7 m FEP Teflon reactor placed in temperature controlled room coupled with instruments for gas and particle phase data. Detailed characterization experiments have been presented describing control of reactor temperature, relative humidity (RH), effective mixing time, wall loss rates of gases and particles, light source, and air purification. This chamber showed a wide range of temperature control with acceptable precision (i.e., 18–33 ± 0.5°C). The gas wall loss rates for NO, NO2, and O3 were found to be 3.78 × 10 min, 4.48 × 10 min, and 6.47 × 10 min, respectively. NO2 photolysis rate constant was 0.17 min. Particle wall loss constant was found to be 3.96 × 10 min at Dp = 100 nm. SOA yields of dark α-pinene ozonolysis ranged from 0.025 to 0.378 for α-pinene concentrations from 10 ppb to 100 ppb.


INTRODUCTION
It is well recognized that atmospheric aerosols play a very important role in environmental processes of climate change, visibility degradation, and toxicology (IPCC, 2013;US EPA, 2013;WHO, 2013).Secondary organic aerosols (SOA) are formed as a result of atmospheric oxidation of VOCs.Atmospheric organic aerosols are composed of significant amount of SOA equivalent to 45-60% (Griffin et al., 1999;Kleindienst et al., 1999;Hallquist et al., 2009).Furthermore, they have a vital part in atmospheric radiative balance because of their capability of absorbing and scattering sunlight (Andreae and Crutzen, 1997;Boucher and Haywood, 2000).These also affect the cloud formation process by acting as cloud condensation nuclei (Lohmann and Feichter, 2004).These pose a significant influence on human health as their exposure leads to detrimental and harmful effects on lungs and cardiovascular systems (Harrison and Yin, 2000;Davidson et al., 2005).However, because of the complications and lack of knowledge of their sources, properties, composition, and formation mechanisms, our understanding of their impact on human health and atmosphere is still finite and limited (Pöschl, 2005).
In SOA researches, smog chambers have been proved as a potent tool (Odum et al., 1996;Kleindienst et al., 1999;Cocker et al., 2001b;Takekawa et al., 2003;Paulsen et al., 2005;Carter et al., 2005;Hu et al., 2014;Wang et al., 2014).Availability of natural sunlight is one of the main advantages in outdoor chambers.The disadvantages include the variation of sunlight intensity and ambient temperature conditions causing difficulty in replicate experiments (Cocker et al., 2001b).For indoor chambers, temperature and relative humidity control are more accurate.Also, rates of some photolysis reactions are different due to usage of artificial lights rather than natural sunlight.Despite of this difference, indoor chambers have a good potential to reproduce experiments (Takekawa et al., 2003;Carter et al., 2005;Paulsen et al., 2005).
Initially, in late 1970s and early 1980s smog chambers have been developed to study formation of aerosols from anthropogenic and biogenic precursors (McMurry and Grosjean, 1985;Stern et al., 1987).Later on in next couple of decades many indoor and outdoor chambers were developed to study SOA formation.Description of smog chambers is given in Table 1.
In this study a new chamber facility has been described to study SOA formation and chemistry.This facility has been built at Kyungpook National University (KNU), Daegu, South Korea.Detailed characterization experiments have been presented, which authenticate the applicability of this facility for SOA formation in both absence (reaction initiated by O 3 ) and presence (reaction initiated by OH

FACILITY DESCRIPTION
The schematic of KNU smog chamber facility is shown in Fig. 1.This facility includes Teflon reactor, radiation source, air purification system, and variety of analytical instruments.

Teflon Reactor
The facility has a ~7 m 3 2 mil FEP (Saint-Gobain, France) reactor with a dimension of 1.95 m (L) × 1.9 m (W) × 1.9 m (H), placed on a net of elastic strings to support its weight.This supporting net is itself hanged to the ceiling with pulleys.This net of elastic strings aids the reactor to maintain its shape during filling and flushing of air.FEP film is flexible, thus air can be taken out of reactor without any change in pressure within the reactor.Also, it is chemically inert and highly transparent specifically to UV radiation.

Reactor Enclosure
The reactor is placed in an air-conditioned (temperature controlled) room with two 9.0 KW air conditioners and a 3.3 KW heater (Shinil Electric, Korea).The room is lined with highly reflective aluminum sheets and foils (A1050, Sumitomo, Japan) to maximize and to make homogeneous interior light intensity.For monitoring temperature, each thermocouple (Omega Engineering, CT, USA) has been installed at ~15 cm inside enclosure wall and ~10 cm inside FEP reactor surface (Fig. 1).

Light Source
Light source used in this smog chamber is 60 UVA (UVA-340, Q-Lab, FL, USA) and 12 UVB 40 W lamps (UVB-313EL, Q-Lab).To prevent overheating of reactor surface, the lamps have been placed at least 70 cm away from the reactor.These lamps are divided into four separate groups.Each group has 15 UVA and 3 UVB lamps.Hence, light intensity can be adjusted from no light (level 0) to all groups turned on (level 4).UV light source spectrum has been measured by NIST certified UV-Visible spectrometer (BLK-C, StellarNet, USA).

Air Purification and Supply System
As shown in Fig. 1 ambient air is compressed by using Anestiwata oil free scroll compressor and then dried by condensing air dryer (NSE-10B, Orion Korea) and adsorption dryer (PD015SFDR07, Walker, WA, USA) for supplying compressed air up to 420 L min -1 and 0.8 MPa.Then, this compressed and dry air is allowed to pass through 2 consecutive beds containing mixture of activated carbon and activated carbon impregnated with phosphoric acid and KMnO 4 impregnated alumina (Incul-aire, Quebec, Canada).Lastly, air passes through 2 parallel HEPA filters (12144, Pall, WA) before it finally enters FEP reactor.The purified air in the reactor has negligible amount of VOCs (C 5 -C 10 ) < 1 ppb, NO x < 1 ppb, ozone < 1 ppb, and RH < 3%.The background particle number concentration is measured using Grimm SMPS+C particle sizer (10-1083.3nm) and is always less than ~10 cm −3 after standard cleaning procedure prior to running each experiment.First two flushings are done with the purified air.Then, residual organics are removed by reacting with ~200 ppb O 3 for 1 hr followed by flushing.Further photo-chemical cleaning is made using ~200 ppb each of H 2 O 2 and NO for 1 hr.Last 4 flushings are performed with the purified air.
For varying relative humidity in Teflon bag, water evaporator using half inch U-shaped stainless steel tubing wrapped with a heating tape is connected to main inlet line of purified air.Milli-Q water is injected to the tubing using a peristaltic pump (BT 100-2J, Longer pump, China).Temperature of the evaporator is controlled using a temperature controller (TZ4ST, Autonics, Korea) coupled with a K-type thermocouple (TJ36-CAXL-116U-18, Omega, USA).

Reactant Injection
In each experiment, a known amount of reactive organic gases (ROGs) was injected using micro-liter syringes into U shaped stainless steel tubing (called evaporation tubing) through septum fixed in one of the bores of stainless steel union tee wrapped with a heating tape.It was introduced into the reactor by N 2 stream, gradually heated from room temperature to set temperature (e.g., 80°C) using a temperature controller (TZ4ST, Autonics, Korea).
Ozone is generated by ozone generator (Green Engineering, Korea) with stainless steel casing fixed with a UV lamp.Ultra- high pure oxygen was fed to ozone generator to prevent NO x formation.Ozone was introduced to FEP reactor under controlled flow rate.
For the photochemical decay of ROG, hydrogen peroxide was used as a source of OH in photochemical experiments.H 2 O 2 was introduced into the reactor as vapors by bubbling 50% H 2 O 2 (516813, Aldrich, MO, USA) with a known amount of N 2 .H 2 O 2 concentrations were estimated using saturation vapor pressure of H 2 O 2 and the volume of H 2 O 2 vapor introduced into reactor.The reactants were well mixed with a mixing fan (A17238V2HBT-C, Active-White, Taiwan) with a dimension of 17.2 cm × 15 cm × 3.8 cm (L × W × H) and a flow capacity of 4.5 m 3 min -1 .It remained turned on during the introduction of reactants.It was then switched off ~30 min before turning on UV lamps to make reactor atmosphere calm and steady.Fans have been used for internal mixing of reactants in smog chambers (Bloss et al., 2005;Carter et al., 2005;Saathoff et al., 2008;Wang et al., 2014).

Seed Aerosol Generation
Non-volatile submicron particles were generated by atomizing aqueous solution of seed material (i.e., 0.3 M ammonium sulfate) using home-made collision atomizer similar to TSI Model 3076.A diffusion dryer containing silica gel with color indicator was used for drying droplets from the atomizer.Because of the electrostatic charge, the generated aerosol can deposit on different surfaces and tube walls.So, aerosol charge neutralizer (4530 HCT, Korea) is used to neutralize the aerosol.

Analytical Instrumentation
The list of instruments used for gas phase and aerosol phase measurements is presented in Table 2.For monitoring aerosol size distribution, number concentration, and volume concentration, sequential mobility particle sizer (SMPS+C, Grimm, Germany) consisting of differential mobility analyzer (DMA) and condensation particle counter (CPC) was used.Aerosol and sheath flow rates are 0.3 and 3.0 L min -1 .NO and NO 2 were continuously monitored by using a NO x analyzer (CM2041, Casella, UK).The calibration of NO x analyzer was carried once in a week with standard NO cylinder approved for calibration.Ozone concentration is measured by O 3 analyzer (400E, Teledyne, CA).RH and temperature within Teflon chamber are also monitored by a hygro-thermometer (HMT333, Vaisala, Finland).Online measurement of VOCs inside the chamber was carried out by using Syntech Spectras GC-PID GC955 series 600.This VOC analyzer has been effectively used for measurement of ambient VOCs that contribute to ozone formation in troposphere (Xie et al., 2008;Belalcazar et al., 2009;Shao et al., 2009).In our system, samples were pre-concentrated on Tenax GR at normal temperature.A sample flow rate of 1.5 mL min -1 was used for VOC analysis.Then, samples were thermally desorbed and separated on a DB-1 column at 1 mL of N 2 as carrier gas and finally detected and quantified by a photo-ionization detector (PID) (Syntech, 2006).The VOC analyzer is suitable for measurement of hydrocarbons from C 5 -C 10 .Calibration of VOC analyzer was performed for 8 VOC precursors (α-pinene, d-limonene, isoprene, toluene, benzene, ethyl benzene, styrene and 1,3,5-trimethylbenzene).
The VOC analyzer showed a retention time of 2.1 for isoprene (C 5 H 8 ) and 18 min for d-limonene (C 10 H 16 ), respectively.The particulars of the calibration are provided in supplementary material.

RESULTS AND DISCUSSION
KNU smog chamber was characterized for basic control parameters such as temperature, RH, mixing time, gas and particle wall loss rates, NO 2 photolysis rate, and background air reactivity.

Chamber Temperature Control
The evolution and control of temperature in KNU smog chamber is demonstrated in Fig. 2. Initially, temperature was allowed to increase using a heater to a target temperature of 35°C.Then, at 0 min all UV lamps and 2 air conditioners were brought online simultaneously.Set temperatures of air conditioners were changed in four steps to lower temperature of 33°C, 30°C, 27°C, and 24°C.Controlled temperatures of 33 ± 0.2°C, 30 ± 0.6°C, 27 ± 0.4°C, and 24 ± 0.3°C were observed.It was observed that after every decrease, temperature inside the smog chamber reached at stable temperature within 10 minutes and then remained stable.The heater remained online throughout the duration of experiment.In summer, the minimum controlled temperature without the heater was 18°C.But in winter lowest temperature can be as low as ambient temperature.All later experiments were performed at 297 K and RH < 3%.

RH Control
Water vapor introduction was achieved during the filling of Teflon reactor with purified air.A peristaltic pump was used to inject Milli-Q water corresponding to RH of 25%, 50%, 75%, and 100% into water evaporator at 180°C.Actual RH readings were 24.3 ± 0.06%, 48.2 ± 0.4%, 73.8 ± 0.3%, and 90.3 ± 0.6% as shown in Fig. 3. RH remained stable for ~40 min.Particle number concentrations in FEP bag remained < 30 cm -3 during the humidification process.

Mixing Time
NO was selected as a tracer for the evaluation of a mixing time in smog chamber.Mixing fan in the reactor remained turned on during this evaluation experiment.Three stepwise NO injections were made using 50 ppm NO gas at 0.8 L min -1 for 5 min.Fig. 4 shows that NO was efficiently mixed within 3 min after each injection.It is comparable to the mixing time of 2 min for GIG-CAS chamber (Wang et al., 2014).

Wall Loss Rates of Gases
The primary reason for decrease in concentration of gases in Teflon reactor is adsorption of gases on the chamber walls.In this chamber, the wall loss rates of NO, NO 2 , O 3 , and two organics (i.e., α-pinene and toluene) were examined.A specific amount of NO, NO 2 , O 3 , and organics was individually injected in dark chamber.
Their respective wall loss rate constants were determined by considering as a first order process for their concentrations variation with time as shown in Fig. 5.The wall loss rates were found to be 3.78 × 10 -4 min -1 , 4.48 × 10 -5 min -1 , and 6.47 × 10 -5 min -1 for NO, NO 2 , and O 3 , respectively.Negligible wall loss for organics was observed in KNU chamber.The loss rate of organics was < 1.71 × 10 -4 min -1 .The concentration of α-pinene and toluene remained at 75.55 ± 0.83 ppb, and 91.76 ± 0.82 ppb, respectively.Also negligible wall loss for organics has been reported in Wang et al. (2014) and Hu et al. (2014).The wall loss constants are in agreement with those of other chambers in Table 3.

Light Intensity
UV spectra were measured at 4 points, 70 cm horizontally from each bottom corner of FEP reactor.Averaged spectrum showed a broad peak in the UV range of 280 nm to 400 nm with a peak intensity of 0.23 W m -2 at 340 nm and several line peaks in visible region.These showed homogenous UV intensity within ± 10%.Fig. 6 shows a comparison with solar spectrum, which is a yearly mean of 48 contiguous states of USA at zenith angle of 48.18° (ASTM, 2012).The lamp and solar spectra have almost same pattern in UV range.Photolysis rate of NO 2 is being used for the representation of light intensity in smog chamber.Light intensity has been evaluated by steady state actinometry, which includes the injection of NO 2 in the chamber followed by irradiation with UV lights.The concentrations of NO 2 , NO, and O 3 are subsequently measured for calculating photolysis rate of NO 2 which is represented by following equation: where, [O 3 ], [NO], and [NO 2 ], are concentrations of O 3 , NO, and NO 2 , respectively and k NO+O 3 is the reaction rate constant for NO and O 3 with the value of 1.9 × 10 -14 cm 3 mol -1 sec -1 (Jet Propulsion Laboratory, 2000).The photolysis of NO 2 is shown in Fig. 7.At full light intensity, k NO 2 equivalent to 0.17 ± 0.017 min -1 (corrected for O 3 and NO reaction in dark sampling lines) was calculated.Thus NO 2 photolysis rate constant can be varied from 0 to 0.17 min -1 by using four separate groups of UV lights.The k NO 2 values in other chambers are presented in Table 3.

Particle Wall Loss
In SOA experiments, generally concentration of aerosol decreases because of deposition as a result of Brownian motion, gravitational settling, and electrostatic forces.The particle wall loss can be expressed as first order process by the following equation: where, N(D p ) is particle number concentration with diameter D p , k dep (D p ) is wall loss rate constant of particles with diameter D p (Cocker et al., 2001a).The variation in the particle number concentration for each particle size was used to measure the particle wall loss rate constant.Two experiments were performed to measure the wall loss coefficients.The mean along with standard deviation of wall loss coefficients as function of D p is shown in Fig. 8.For SOA formation, measured particle concentrations were  corrected using the size-dependent loss constant as below.
The values of k dep for our chamber were in range of 1.7 × 10 -3 min -1 to 1.8 × 10 -2 min -1 .In Cocker et al. (2001), values of k dep are in range from 1.5 × 10 -3 to 3 × 10 -3 min -1 .Comparison with other studies at D p = 100 nm is shown in Table 3.

Air Purity
Three types of background reactivity of purified air were examined for the assessment of air purity (see Fig. 9).One dark experiment with addition of O 3 was performed.After 5 hr of O 3 injection at 218 ppb, total particle concentration was ~10 particles cm -3 in number and ~0.20 µg m -3 in mass, respectively.These are comparable with those of Hu et al. (2014).In second experiment, ~200 ppb each of NO, NO 2 and H 2 O 2 were injected and UV light was used to irradiate the reaction mixture.After irradiation of 5 hr, total particle concentration was ~800 particles cm -3 in number, ~0.35 µg m -3 in mass, and ozone generation was only 1 ppb, respectively.Lastly, 5 hr photooxidation of only purified air showed minimal particle formation of ~800 particles cm -3 in number and ~0.8 µg m -3 in mass, respectively.The magnitude of particle formation is comparable with those in Shan et al. (2007).O 3 formation rate of ~9 ppb hr -1 was ~5 times higher than previous reports (Carter, 2002;Shan et al., 2007;Lee et al., 2010;Hu et al., 2011).Considering negligible C 5 -C 10 VOCs in the purified air, residual organics deposited on surface of FEP reactor and low molecular weight hydrocarbons (e.g., methane, ethane, and propane) in purified air might be mainly responsible for the higher O 3 formation.Further improvement in bag cleaning procedure might be needed for the application of this chamber to researches on O 3 formation.

SOA Formation and Yield of α-pinene
A sequence of SOA formations by dark α-pinene ozonolysis has been performed to validate the utility of KNU chamber for further studying SOA formation and atmospheric chemistry.The yield data for dark α-pinene ozonolysis is readily available in literature (Griffin et al., 1999;Cocker et al., 2001a;Saathoff et al., 2008;Wang et al., 2014;Hu et al., 2014).The initial conditions and yield for each ozonolysis experiment are presented in Table 4.
In general, the SOA yield, Y is given by following equation: where ∆M o is the aerosol mass concentration in µg m -3 and ∆ROG is the mass concentration of reactive organic gas (ROG) in µg m -3 (Cocker et al., 2001a).All the SOA formations were carried out in dry conditions (RH < 3.0%) and temperature 297 ± 1°C.Initial conditions of α-pinene were varied from 10 ppb to 100 ppb in the series of experiments.All experiments were performed without seed and OH scavenger.Aerosol and gas phase profiles along with experimental conditions (RH and temperature) for Exp 1 are shown in Fig. 10.Density of aerosol equivalent to 1 g cm -3 was assumed for the evaluation of SOA mass concentrations.A comparison of yield data with other studies (Griffin et al., 1999;Cocker et al., 2001a;Saathoff et al., 2008;Wang et al., 2014;Hu et al., 2014) is shown in Fig. 11.Only yield data was considered that was obtained under dark conditions without OH scavenger and seed particles.The yield data of KNU chamber are generally in agreement with other data in literature.However, SOA yields of this study were relatively higher than those of SOA formed at higher temperature of 302 K-309 K (Griffin et al. (1999); Cocker et al. (2001a); Saathoff et al. (2008)).SOA yield varies inversely with temperature due to enhanced partitioning of semi volatile species to particle phase at lower temperature (Stanier et al., 2007;Johnson and Marston, 2008).Our SOA yields measured at dry conditions were higher than those of Hu et al. (2014) at humid conditions.For α-pinene ozonolysis under humid conditions, water vapors suppress the formation of organic acids resulting in decreased SOA formation (Cocker et al., 2007).The SOA yields in our study were comparable to those of Wang et al. (2014) measured at similar temperature and relative humidity conditions.

CONCLUSIONS
It is evident from detailed characterization experiments that KNU smog chamber meets all the necessary requirements for investigating SOA formation and chemistry.The chamber demonstrated good control of temperature, RH, and efficient mixing time.Gas and particle wall loss rate constants are in agreement with other chambers.The wall loss rates for NO, NO 2 , and O 3 were found to be 3.78 × 10 -4 min -1 , 4.48 × 10 -5 min -1 , and 6.47 × 10 -5 min -1 , respectively.Particle wall loss constant is found to be 3.96 × 10 -3 min -1 at D p = 100 nm.NO 2 photolysis rate constant is 0.17 min -1 .Furthermore, SOA yields of dark α-pinene ozonolysis are in range of 0.025 to 0.378 for α-pinene concentrations from 10 ppb to 100 ppb.These are also comparable with other chamber studies.

Fig. 4 .
Fig. 4. Variation of NO concentration with time in three consecutive injections for mixing time measurement in KNU smog chamber.

Fig. 7 .
Fig. 7. Profiles of NO, NO 2 , and O 3 concentrations for a series of turn on and off of UV lamps to estimate NO 2 photolysis rate.

Fig. 8 .
Fig. 8. Evolution of particle size distribution during particle wall loss experiments.(a) Without correction, (b) With correction, (c) Particle wall loss rate constant k dep (D p ) as a function of particle diameter.

Fig. 9 .
Fig. 9. Profiles of particle size distribution and relevant gases (NO, NO 2 , and O 3 ) in the experiments for purified air reactivity.(a) Particle for ozonolysis, (b) Gases for ozonolysis, (c) Particle for high NO x photochemical reaction, (d) Gases for high NO x photochemical reaction, (e) Particle for air only photochemical reaction, (f) Gases for air only photochemical reaction.

Fig. 10 .
Fig. 10.Phase profiles and experimental conditions for Exp 1.(a) Aerosol phase profile, (b) Gas phase profile, temperature, and RH conditions.

Fig. 11 .
Fig. 11.SOA yield data comparison of dark α-pinene ozonolysis for this study and other studies.

Table 1 .
Description of smog chamber facilities.

Table 2 .
Description of analytical instruments coupled to KNU smog chamber.

Table 4 .
Initial conditions and SOA yield for dark α-pinene ozonolysis SOA formation experiments.