Chemical Characterization of Submicron Aerosol Particles in Santiago de Chile

High time resolution chemical characterization of submicron particles was carried out in the South American city of Santiago de Chile using the Aerosol Chemical Speciation Monitor (ACSM). The instrumentation operated for 100 days from August 17 to November 23 2011 in an urban station located inside the University of Santiago de Chile (USACH) campus. In addition, a semi-continuous OC/EC analyzer was also run in parallel with the ACSM for some of this time. Meteorological conditions varied along the studied period due to the transition from winter to spring time. Atmospheric temperature inversions were responsible for hourly average sub-micron particulate matter levels of up to 80 μg/m, especially during the night time. The average submicron particle mass concentration (± standard deviation) for the whole period was 29.8 ± 25 μg/m. Aerosol particles were composed mainly of organics 59%, followed by nitrate, ammonium, sulfate, black carbon and chloride with contributions of 14, 12, 8, 3 and 3%, respectively. Using positive matrix factorization, the organic fraction was divided into four distinct types of organic aerosol representing fresh automobile exhaust, biomass burning, and two oxygenated organic aerosol factors with different oxidation states. The transition from winter to spring was clearly seen in the composition of OA. The emissions from primary sources, such as vehicle and biomass burning, decreased in the period leading to spring, whereas the amount of oxygenated organic aerosol increased over the same time. This study shows that high time resolution measurements of aerosol chemical composition can lead to better characterizations of the evolution and sources of pollutants in an urban atmosphere.


INTRODUCTION
Atmospheric aerosol from a broad range of natural and anthropogenic sources has a large effect on visibility reduction, climate change and impact on human health.Health effects of particles are typically most pronounced in large cities or urban areas, megacities.Air quality in megacity is usually poor due to concentrated sources of anthropogenic emissions from heavy traffic and industrial activities combined with poor geography which together produce frequent health warnings during winter and summertime.As more than half of the world's population is now living in urban areas (UNFPA, 2007) air quality and health impacts in urban regions are becoming even more important.Santiago de Chile is a large city with a population around 5.5 million (2002).It has Mediterranean climate, and it is recognized as one of the world's most polluted cities, where PM 2.5 annual average is 2.2 times the EPA standard.Santiago is under environmental concern, whereby the authorities have implemented an important number of policies in order to recover the air quality.The level of knowledge concerning emission sources in Santiago is reasonable, but the emission profile is expected to be under continuous change as rapid industrialization, urbanization, and economic growth is occurring at a fast pace.In previous studies, five factors have been identified to contribute significantly to PM 2.5 in Santiago de Chile: soil, motor vehicles, residual oil, sulfates and, to a smaller extent, in smaller contribution marine aerosols (Moreno et al., 2010).Regarding motor vehicles, about 42% of the Chilean vehicle fleet is concentrated in Santiago and it has also experienced a noticeable growth during the last five years (INE, 2011).However, due to the stricter emission laws and improvements on vehicle technologies a decrease in the contribution of traffic has been noticed (Sax et al., 2007;Moreno et al., 2010).
In this study a new monitoring technique based on aerosol mass spectrometry was deployed in order to get a detailed insight into the chemical components of PM in Santiago de Chile.Aerosol Chemical Speciation Monitor (ACSM) is able to determine all non-refractory components in submicron aerosol particles including e.g., sulfate, nitrate, ammonium, chloride and organic matter.Previously, most of the studies from Santiago de Chile have been accomplished using elemental analysis with X-ray fluorescence (XRF) technique.Therefore, only limited information on organic aerosols (OA) is available from Santiago.In general, OA can be divided into primary and secondary OA (POA and SOA, respectively).Seguel et al. (2009) estimated that the contribution of SOA to organic carbon (OC) can reach 20% of total organic aerosol matter in summer in Santiago by applying EC tracer technique (Turpin and Huntzicker, 1991) to real time measurements of organic and elemental carbon.Although this was an important result for Santiago de Chile, this methodology is rather crude and provides no information about the chemical composition of the organic mass fraction.In this study, the measured OA data were analyzed in detail by applying positive matrix factorization (PMF) for organic mass spectra.Factor analysis enabled to explore the properties, sources and atmospheric processing of organic.

ACSM
The Aerosol chemical speciation monitor (Aerodyne Research Inc.,) is able to routinely characterize non-refractory submicron aerosol species organics, nitrate, sulfate, ammonium and chloride with the time resolution from 15 minutes up (Ng et al., 2011).The instrument consists of a particle sampling inlet, three vacuum chambers and a residual gas analyzer mass spectrometer.Particles are first focused by using an aerodynamic lens system into a narrow particle beam, which is transmitted through three vacuum chambers.In the third chamber the particle beam is directed into a hot tungsten oven (600°C) where particles are flash-vaporized, ionized with a 70 eV electron impact ionizer and detected with a quadrupole mass spectrometer.
Mass calibration of the ACSM is achieved by determining the instrument response factor using ammonium nitrate calibration aerosol (RF NO3 ) as described by Ng et al. (2011).The calibration system in this study consisted of an ultrasonic nebulizer (Model NE-U17, OMRON) for primary aerosol generation, a silica gel diffusion dryer (TOPAS, DDU 570/L), a 28 cm long Hauke-type differential mobility analyzer (DMA) and a condensation particle counter (CPC, Model 3010, TSI Inc.).The system was used to generate monodisperse 300 nm ammonium nitrate aerosol particles.From the number counts of the CPC, ammonium nitrate mass concentration was calculated using the known particle density and diameter, considering spherical particles.The calculated ammonium nitrate mass concentration was compared to the ACSM values.The number concentration was varied by diluting the generated aerosol between approximately 10-1000 particles/cm 3 , which corresponds to the mass concentration of 0.15-15 μg/m 3 for nitrate.

Semi-continuous OC/EC Analyzer
Because the ACSM is able to analyze only non-refractory material, black carbon was determined by the thermal optical transmittance (TOT) method using semi-continuous carbon analyzer (Sunset Laboratory Inc., Portland, OR).The method is previously described in detail by Schauer et al. (2003).The instrument has a PM 2.5 cyclone, and allowed hourly measurements of elemental carbon from September 29 th to November 23 rd 2011.

Description of the Sites
The ACSM and the Semi-continuous OC/EC analyzer were installed at the University of Santiago de Chile (USACH) station inside a container about 10 meters above ground level, on the roof of a two-storey building.At the USACH station, in the ACSM set up, a cyclone (URG Model 1226, URG Corporation, Chapel Hill, USA) was set outside one meter above the container to cut-off particles larger than 2.5 μm in the sampling line.
O'Higgins station is part of the monitoring station network supported by the Chilean authorities (Environmental Ministry) and maintained by the National Information System of Air Quality (SINCA) in the Metropolitan Area of Santiago de Chile, http://sinca.mma.gob.cl/index.php/region/index/id/M.The station is located about two kilometers southwest of USACH station, in the central area of Santiago de Chile, Fig. 1.The USACH and O'Higgins Park stations are located in an area of Santiago that has the highest levels of PM concentration during the whole year.These levels are primarily determined by the topographic and meteorological characteristics of the area (Gramsch et al., 2006).

ACSM Data Analysis
The organic aerosol fraction was further investigated using Positive Matrix Factorization (PMF2, Paatero, 1994) version 4.2 in the robust mode and the custom software tool PMF evaluation panel (PET) version 2.4 developed by Ulbrich et al. (2009), which enables evaluation of the PMF results together with statistical analysis.

Input Matrices Pretreatment
The two input matrices consist of organic aerosol concentration in μg/m 3 and respective uncertainties or errors, where the rows represent time series and the columns represent different variables (mass-to-charge ratios, m/z's), (Ulbrich et al., 2009).In this study the matrices contain 70 m/z's and 4026 samples.However, before applying the PMF analysis a pretreatment of the input data matrices is recommended as described by Zhang et al. (2005) and Ulbrich et al. (2009).Typically, this pretreatment is applied to AMS and Q-AMS datasets which have a faster time resolution (e.g., 5 minutes) and in case of sporadic spikes of individual variables (m/z's) the whole row can be simply removed from the analysis without statistical consequences.However, because the ACSM has lower time resolution (e.g., 30 minutes), the removal of data points should be avoided.Therefore, the errors of incorrect and non-representative spikes of individual m/z's were downweighed 100 times decreasing their importance inside the fit.The result of this procedure is similar to removal, with the advantage of keeping the correct and representative m/z's in the same row.

Meteorology and Chemical Composition of Fine Particles
The meteorological conditions of Santiago de Chile have been described in detail by Muñoz (2005).Briefly, there is a predominant wind profile coming from southwest to northeast which is stronger in spring and summertime.In contrast, higher stability and poor ventilation is observed in autumn and wintertime with significant lower wind speed.During winter nights a slow mountain-valley breeze with strong increase of calms can be distinguished.As overall result, the air masses can stay 2-4 days inside the valley.During that period most of the time the only effective removal mechanisms for particles are related to precipitation or fog and consequently the highest PM levels events are observed.
Fig. 2 shows air temperature, RH, wind speed, wind direction from the O'Higgins Park station and global solar radiation from the USACH station during August 17 th -November 23 rd , 2011.The average air temperature (T avg ) and RH for the whole campaign was 14.1°C and 65%, varying from -2 to 31°C and from 6 to 97%, respectively.This period corresponds to the transition time from moderately humid winter to dry and hot summer, which is observed by the change in the average air temperature and RH for the first 30 days of campaign, T beg = 10.2°C and RH = 65.7%, in contrast to the last 30 days T end = 17°C and RH = 51.2%.In addition, the continuous increase in global solar radiation along the period of study reinforces the notion of seasonal transition.The averaged wind speed was 2 m/s and southwest is the predominant wind direction.Santiago de Chile presents a persistent valley-mountain breeze system with anticyclonic meteorological conditions throughout the year, which leads to frequent air subsidence and thermal atmospheric inversions all year long.Therefore, aerosol concentrations in Santiago de Chile are strongly influenced by the meteorological and complex topographic conditions (Jorquera et al., 2004;Muñoz, 2005;Morales, 2006;Seguel et al., 2009).For that reason aerosol concentrations are much higher during the beginning of the campaign compared to the end.Fig. 3(a) presents the time series of organics, nitrate, ammonium, sulfate and chloride measured by the ACSM.For example, during the first 30 days of the campaign, a period with T beg < T avg , fine particle mass concentration based on the ACSM and Semi-continuous OC/EC analyzer measurements was 42.9 μg/m 3 (August 17 th -September 16 th , 2011, Table 1), while during the last 30 days when the temperature was above the average (17°C), the aerosol concentration was 18.1 μg/m 3 , (October 25 th -November 23 th , 2011).The drastic reduction is a consequence of the frequent clean air masses The total aerosol mass from the ACSM (organics + sulfate + nitrate + ammonium + chloride) was compared to the PM 2.5 concentrations measured in the O'Higgins station, Fig. 4. Typically, a collection efficiency (CE) of 0.5 has been used to account for losses in the vaporizer, which are related to aerosol acidity and relative humidity of the sampled air in the sampling line (Canagaratna et al., 2007;Middlebrook et al., 2012;Sun et al., 2012).In this study both instruments present similar level of particulate mass, when collection CE = 0.5 is applied to the ACSM.The correlation between the instruments is good (R = 0.77) but the ACSM shows slightly higher concentrations than PM 2.5 even though ACSM measures only non-refractory particles with diameter < 1 μm.This fact is probably related to the uncertainty in the CE, which has been estimated by Quinn et al. (2006) and Matthew et al. (2008) to be about 30%.Moreover, it should be noticed that the O'Higgins station with PM 2.5 is located about two kilometers from the USACH station (ACSM) and, therefore, some local sources may be different.In addition, PM mass monitor typically has losses for semi volatile compounds.

Aerosol Chemical Composition
Fig. 3(b) depicts the time series of the chemical composition of submicron aerosol.On average the aerosol is composed mainly of organics (59%), followed by nitrate, ammonium, sulfate, chloride and black carbon with 14, 12, 8, 3 and 3%, respectively.This organic fraction is about three times larger than the fraction reported by Seguel et al. (2009), and at least two reasons are presented here.
First, the fraction obtained by the authors is based on particulate organic carbon (POC), which may differ from 1.6 to 2.2 times from particulate organic matter (POM).Second, the two studies report averages in different times of the year.Seguel et al. (2009) reported an average for the summer month of February 2004.In addition, artifacts such as losses due to evaporation of semi volatiles are expected to be larger during summer conditions.
When the average of the first 30 days of campaign is compared to the last 30 days no clear change is observed in terms of organic and inorganic mass fractions, Table 1.Note that BC is not included in those fractions because unfortunately, data were not available over the whole period.Sulfate fraction increases from 6 to 16%, the nitrate fraction decreases from 17 to 12%.The decrease is probably related to the raise of the air temperatures and the semi volatile character of nitrate while the increase of sulfate is likely related to long-range or regional transport of pollutants.In fact, there are three episodes during this period where sulfate represents up to 35% of aerosol mass.Most likely sulfate is in the form of ammonium sulfate because ammonium contribution also increases to 20%.Moreover, no change in SO 2 concentration in gas phase during those events, not shown, implies that the sulfate is mainly long-range or regional transported.Those events lasted from one to three days.
A clear diurnal pattern for the concentrations is shown in Fig. 5  The concentrations of nitrate and ammonium decrease with the evolution of the PBL and increase of the wind speed along the day as well as higher volatility due to higher ambient temperature.Chloride, most likely in the form of ammonium chloride, presents the lowest concentrations during the day time (12:00-18:00) due to its volatility.It evaporates rapidly with the increase of air temperature.Sulfate presents different diurnal trend from other compounds.Its concentration is rather stable throughout the day except slightly elevated concentrations from 9:00 to 18:00.When compared to the diurnal profile of sulfur dioxide SO 2 no correlation is found.However, Fig. 5(d) depicts similar behavior for both compounds only shifted about one hour, which suggests that part of the sulfate results probably from SO 2 oxidation and, therefore, it is also a product of secondary aerosol formation.We speculate that SO 2 emission is related to at least two different sources, traffic and smelters.As a consequence of the low amount of sulfur present in the diesel used by the vehicle fleet in Santiago de Chile (S10 -10 ppm of S) a small increase in sulfate concentration, from 2.5 to 3.2 μg/m 3 , is observed just one hour after the increase in SO 2 concentration from the morning rush hour.This small increase in sulfate indicates that the local emissions represent a minor source.In the afternoon the concentrations increase and stay stable until 18:00, time when the wind speed reaches its maximum.The stable concentration of both compounds during the afternoon could be related to industrial emissions, for instance smelters, one located inside Santiago and another two located about 100 km, suggesting local and regional transportation of pollutants.

Aerosol Acidity
The acidity of the particles can be estimated through a ratio between measured ammonium concentration and theoretically predicted ammonium concentration.Predicted ammonium is needed to neutralize the major inorganic anions.It is obtained using Equation 1 (Middlebrook et al., 2012) where sulfate, nitrate and chloride were measured with the ACSM.When ammonium and predicted ammonium ratio is investigated the ratio is about 1 most of the time suggesting that aerosol is neutral most of the time (not shown).Few points are observed to be lower than the unity ratio, which indicate lack of ammonium in the particle phase and that the aerosol is, therefore, acidic.Those points correspond to two nights (less than 1% of the measurement points) when chloride fraction was high.

Characterization of OA Components ACSM-PMF Factors
The number of factors was selected based on the Q-value (the total sum of the squares of the scaled residuals), the comparison of diurnal cycles of the different factors with gas phase and BC data, the comparison of mass spectra with the AMS mass spectra database as described in detail by Ulbrich et al. (2009), and using meteorological information.Six factor solution was used to identify four distinct sources of organic aerosol, low-volatility oxygenated organic aerosol (LV-OOA), semi-volatile oxygenated organic aerosol (SV-OOA), biomass burning organic aerosol (BBOA) and hydrocarbon-like organic aerosol (HOA).Two PMF factors were recombined for LV-OOA and two factors for BBOA.The recombination was based on similar mass spectra, diurnal trend and time series of the factors.The time series of the mass fraction and the mass spectra of the four factors are depicted in Figs. 6 and 7.
The LV-OOA factor is the most oxygenated fraction of OA dominated by m/z 44, which is mainly the fragment CO 2 + typical from thermal decarboxylation of organic acid groups (Alfarra, 2004).This factor is very similar to the average LV-OOA obtained by Ng et al. (2011), Pearson correlation of R = 0.99 (Table 2), and it has been previously described as highly aged, low-volatile and associated with secondary organic aerosol.In fact, the LV-OOA correlates with the sum of secondary inorganic species nitrate and sulfate (Pearson correlation, R = 0.72, not shown).Also the average diurnal profile in Fig. 6(b) presents maximum concentration around the same time as the sum of nitrate and sulfate, 10:00-15:00, which is the time when the global radiation is more intense.In addition, the LV-OOA fraction increases during the period of the campaign due to the enhancement of the photochemical activity and air temperature with the arrival of the spring season in the southern hemisphere.On average, this factor dominated the OA fraction with 44.5% varying from 34%, during the first 30 days to 63%, during last 30 days.When investigated at   f44 (ratio between the m/z 44 and the total signal in the component mass spectrum) vs. f43 space, Fig. 8, this factor presents f44 equal to 0.21, which corresponds to O:C ratio of 0.85 (Ng et al., 2010).This ratio indicates high oxygen content aerosol and is associated with low-volatile organic aerosol (Lanz et al., 2007;Aiken et al., 2008;Ulbrich et al., 2009).The low-volatility character of the LV-OOA is illustrated in Fig. 9(a) with the strong dependency of this factor with the air temperature.
The second fraction of identified OA is SV-OOA, which contributes on average to 15% of the OA decreasing quickly from 19% during the first 30 days to 12% after the end of September and staying around 10% during the last 30 days of the campaign.The decrease occured at the same time when long-ranged transported air masses start to reach the region more frequently indicating that this factor is mostly related to local emissions.The mass spectrum is dominated by the m/z's 44 and 43, which is expected to be mainly represented by the oxygenated fragment C 2 H 3 O + and not originating from acids (Ng et al., 2010).When investigated at f44 vs. f43 space, SV-OOA lays in a region where O:C ratio is lower (0.6) than for LV-OOA.The lower O:C ratio is in agreement with the other studies that characterize SV-OOA as less oxygenated and semi-volatile fraction of OOA (Jimenez et al., 2009;Ulbrich et al., 2009;Ng et al., 2010).The semi-volatile character of this factor is also depicted in Fig. 9(a) where the SV-OOA clearly decreases with the air temperature raise.Moreover, the SV-OOA position at f44 vs. f43 space indicates that this factor corresponds to an intermediate state in the aerosol evolution between the low oxidized HOA and BBOA and the highly oxidized LV-OOA.The organic aerosol evolution process will be further discussed later on in this study.The third factor identified is HOA with the average contribution of 18.8% to OA.This factor is characterized by a hydrocarbon-like structure in the mass spectrum typical of alkanes, alkenes and cycloalkanes (m/z's 27, 29,41,43,55,57,67,69,71,81,83,85) from vehicle exhaust, previously described by Canagaratna et al. (2004) and Zhang et al. (2005).Furthermore, the mass spectrum correlates well with the average HOA obtained by Ng et al. (2011), Pearson correlation R = 0.98.Also the HOA diurnal profile presents clear similarity with black carbon (BC) and the gases, such as carbon monoxide (CO) and nitrogen oxide (NO), emitted by vehicle exhaust with the maximum concentration during the local rush-hour time 7:00-9:00, illustrated in Fig. 6(b).HOA is associated with primary organic aerosol and the high contribution to OA during the first 30 days of campaign, 24%, is a consequence of the emissions from the large vehicle fleet of Santiago de Chile, meteorological conditions and the semi volatile character of this aerosol.The poor air dispersion and low temperatures favors the partitioning of semi volatile species into the particle phase.This fact changes with the arrival of the spring season when the HOA contribution decreases to 11% for the last 30 days of the campaign.Hence, the decrease of HOA should not be interpreted or associated with traffic emission decrease in Santiago de Chile in that period.It is very unlikely that the vehicle feet changed in number and type, however, it is very likely that the POA emitted by the vehicles evolved faster into more oxidized aerosol fraction, like LV-OOA, during spring time atmospheric conditions.
Fourth factor was found to contain elevated relative signal at m/z's 29, 60 and 73, suggesting the presence of wood combustion, likely from local domestic heating (Lanz et al., 2007;Aiken et al., 2008).That factor also contains elevated signal at m/z's 41, 43 and 55, which has been associated with cooking OA (COA), an important source of POA in urban environments like Zurich, Barcelona and Manchester (Lanz et al., 2007;Allan et al., 2010;Mohr et al., 2012).Mohr et al. (2012) demonstrated that food charbroiling mass spectrum differs from HOA due to the oxygen-containing ions in the m/z's 43, 55 and 57, typical from fatty acids.Moreover, in the food charbroiling mass spectra the m/z 43 presents higher relative signal than 41.However, Allan et al. (2010) demonstrated that alternative practices, for example frying instead of charbroiling will also result in the emission of substantial amount of particles from the oil vaporization and subsequent condensation and in that case the mass spectrum may contain more m/z 41 than 43.Because the fourth factor is comprised of different sources of biomass burning, it is labeled as BBOA.
The contribution of BBOA to OA is 21.2% on average, changing from 23% at the beginning to 16% in the end of campaign.The decrease in the course of the campaign is most likely related to the change in the meteorological conditions and concurrent reduction of emissions.The diurnal profile of BBOA was quite similar with HOA except the sharp peak obtained for HOA in the morning.Cooking OA has reported to cause an increase during the early afternoon in the previous studies but that was not noticed here.This fact is probably due to the strong influence of the planetary boundary layer evolution on the aerosol load, small concentration during the day and elevated concentration during the night time.

Aerosol Processing
The evolution of organic aerosol is investigated in the f44 vs. f43 space in Fig. 8.During the first half of the campaign aerosol ranged inside a narrow region of the triangle where the low f44 and f43 indicate low O:C ratio and aging of OA (colors from red to green).We speculate that the aerosol in this region is dominated by the hydrocarbon fragment C 3 H 7 + since 50% of the OA in the first 30 days of campaign is composed of factors with a high hydrocarbon content such as HOA and BBOA.This is the period when the poor air dispersion and mechanisms of aerosol removal maintained the particles inside the valley.After this period the contribution of HOA and BBOA factors decrease while the oxygenated OA (LV-OOA) increases, which in the f44 vs. f43 space indicates that the OA moved toward the top corner of the triangle.This is the region of the triangle where f44 is higher and f43 is smaller, likely due to reduction of the hydrocarbon fragment C 3 H 7 + and increase of the oxygenated fragment C 2 H 3 O + .Moreover, that is the region of the triangle where most oxygenated OA is found with high O:C ratio.Several factors contributed to the change of the OA in the f44 vs. f43 space, for example changes in the photochemical activity, air temperature and arrival of fresher air masses to the valley.The elevated concentration of O 3 and raise of air temperature may favor the formation of atmospheric radicals, which suggests a stronger oxidant character of the atmosphere during that period.The combination of all those factors contributed to further aerosol aging.
The aerosol evolution is also observed in the diurnal profiles of SV-OOA when the first and last 30 days are compared in Fig. 9(b).Differently from the first 30 days when the diurnal pattern of SV-OOA presents clear decrease during the day and evening hours 11:00-22:00, the last 30 days reveal a prominent increase from 10:00-18:00, similar to LV-OOA diurnal.This fact illustrates the intermediate character of SV-OOA changing from processed POA to fresh SOA.The change in the SV-OOA profile during the end of the campaign is in agreement with the evolution process of OA described by Jimenez et al. (2009).This type of behavior was possible to be observed probably due to the long period the campaign comprehended, which included the season transition from winter to spring time.

CONCLUSIONS
This study investigated the chemical components of submicron PM in Santiago de Chile from August 17 th to November 23 rd , 2011.The high time resolution chemical characterization of submicron particles was accomplished by using the Aerosol Chemical Speciation Monitor.Part of the time also the Semi-continuous OC/EC analyzer was run in parallel with the ACSM.Meteorological conditions varied along the studied period due to the transition from winter to spring time.The average submicron particle mass concentration for the whole period was 29.8 μg/m 3 but atmospheric temperature inversions were responsible for submicron particulate matter levels up to 80 μg/m 3 (1-hour average), especially during the night time.Aerosol particles were composed primarily of organics (59%), followed by nitrate (14%), ammonium (12%), sulfate (8%), black carbon (3%) and chloride 3%.
The measured data was further analyzed using positive matrix factorization in order to investigate the sources and time evolution of organic aerosol in Santiago de Chile.Using PMF the organic fraction was divided into four distinct types or organic aerosol representing HOA, that was likely to be related to fresh automobile exhausts, biomass burning, that mostly originated from residential wood burning, and two oxygenated organic aerosol factors with different oxidation states.Transition from winter to spring was clearly seen in the composition of OA; the contribution of primary sources, BBOA and HOA, decreased as temperature increased.For oxygenated OA-components, semi-volatile and low-volatility OOA and OOA, the difference between the beginning of the campaign and the end of campaign was even more distinct.SV-OOA constituted one fourth of OOA at the beginning of campaign whereas at the end it made only 10% of OOA.As LV-OOA was more oxidized than SV-OOA, organic matter in general was much more oxidized in springtime than in wintertime in Santiago.Besides the contributions, the diurnal trend of OOA changed coming from winter to spring.In winter SV-OOA decreased in the afternoon with higher ambient temperature and the development of boundary layer whereas in spring it had a maximum in the afternoon indicating secondary formation.Diurnal trend for SV-OOA remained rather similar during the course of the campaign.
To our knowledge this is the first time when submicron aerosol chemistry is characterized in detailed in an urban environment of South America.This study shows that air pollution is a severe issue in Santiago de Chile, but as the aerosol sources can be characterized, the actions can be implemented in order to improve the air quality.The measurements were conducted only at one site in this study but comparable observations can be expected to be achieved from other polluted environments.
The results in the current study suggest that the most effective way to reduce atmospheric submicron aerosol in Santiago de Chile is reducing emissions from the traffic and domestic heating sectors.New incentives to promote the use of public transportation instead of private should contribute significantly to reduce the vehicle fleet of Santiago de Chile.In order to improve the performance of the fleet, the use of more efficient catalysts and cleaner technologies should be promoted and encouraged by the authorities.Finally, also, cleaner sources of energy and improvement of the calefaction stoves currently in use should be employed in the domestic heating process.

Fig. 1 .
Fig. 1.Map of Santiago de Chile and surroundings.ACSM and Semicontinuous OC/EC analyzer were operated at USACH station, PM and gases were measured at O'Higgins station.The stations are indicated in the map by grey and white markers, repetitively.

Fig. 2 .
Fig. 2. Meteorological information at O'Higgins station, air temperature, relative humidity, wind speed, wind direction and global solar radiation at USACH station.

Fig. 4 .
Fig. 4. Time series of total aerosol measured by ACSM at USACH station and PM 2.5 measured at O'Higgins station.

Fig. 5 .
Fig. 5.Diurnal profile of (a) aerosol mass concentration in μg/m 3 of organics, nitrate, sulfate, ammonium and chloride (b) meteorological parameters, wind speed, air temperature, relative humidity and solar radiation (c) nitrate, NO and NO 2 and (d) sulfate and SO 2 .

Fig. 8 .
Fig. 8. Components f44 vs. f43 of aerosol data colored by time series with respectively PMF factors.

Fig. 9 .
Fig. 9. Mass concentration inμg/m 3 of LV-OOA and SV-OOA vs. air temperature (a) and diurnal profiles of SV-OOA and LV-OOA during the first and last 30 days of campaign (b). ,