Characterization of aerosol particles during a high pollution episode over Mexico City

More than 7 thousand wildfires were recorded over Mexico in 2019, affecting almost 640 thousand hectares. Most of these fires occurred during the spring season generating dense smoke plumes, impacting urban areas in the central part of the Mexican plateau. From May 10 to 17, 2019, biomass burning (BB) plumes affected Mexico City (MC) and diffused across the basin, producing PM2.5 levels ~ 2 times higher than the nation's air quality standards. Average PM2.5 concentrations increased sharply from 29.4 ± 7.2 µg m−3 to 65.1 ± 13.6 µg m−3 when the dense smoke plumes were detected. The higher particle concentration impacted the aerosol optical depth (AOD) as values ~ 3 times greater than the annual mean (0.32 ± 0.12) were measured, which resulted in a 17% loss of global horizontal irradiation (GHI). Under these severe pollution conditions, the visibility (Va) was reduced by ~ 80%. The high incidence of strong absorbent particles, such as soot and tarballs was revealed through electron microscopy and X-ray fluorescence (XRF) analysis. These techniques show chemical similarities between MC aerosols and those from the high-altitude (~ 4010 m. a. g. l.) Altzomoni Atmospheric Observatory, evidencing a strong influence of the BB emissions, suggesting a regional transport of these pollutants.

www.nature.com/scientificreports/ MC was affected by a severe pollution episode in May 2019, detected by the Automatic Atmospheric Monitoring Network (RAMA by its acronym in Spanish) stations. The concentrations of PM 2.5 presented a rapid and dramatic increase from May 10 to 17. This HAP episode significantly affected the V a and degraded the air quality in MC, a fact that alarmed MC's authorities, who decreeing an environmental contingency to mitigate their impacts. Although most of the airborne pollutants are assumed to be generated by forest fires and the ubiquitous anthropogenic activities within the MC basin, the detection of dense smoke plumes at the high-altitude Altzomoni Atmospheric Observatory (AAO) suggests that BB aerosols can be transported on a regional scale.
In this study, aerosol optical properties were measured during the HAP period (May 10 to 17, 2019) to evaluate the effect of high PM 2.5 concentrations on the GHI and AOD values, as well as in the V a degradation. Additionally, a morphological and chemical analysis of the aerosol samples collected in MC and the AAO was performed with transmission (TEM) and scanning (SEM) electron microscopes coupled with X-ray analyzers [19][20][21] . This analysis allows identifying the primary sources of aerosol particles that significantly reduced air quality in MC.

Methods and instrumentation
A direct comparison of the locally emitted particles vs. the long-range transported aerosol particles into the MCMA was assessed at the high-altitude AAO [22][23][24] . Given that the AAO is ~ 4010 m a.g.l. (and 1500 m above MC) 22,25 , it allows the collection of aerosol particles within the free troposphere (FT) and the mixed layer (ML) 22,26 . Due to its location and its research facilities, the AAO is part of the Network for the Detection of Atmospheric Composition Change (NDAC) since 2015 22,26,27 . Studies conducted at AAO have focused on volcanic emissions [28][29][30] , ML pollutants 22,23 , and mixed-phase cloud formation 25 . Sampling sites. Mexico city. The MCMA consists of 60 agglomerated municipalities 14 . It is considered the largest megalopolis in North America with more than 20 million inhabitants residing in ~ 1500 km 2 . MC located in a subtropical zone is affected by multiple fires during the dry-warm (DW) season, from March to May 11 . The sampling site was placed on the roof of the Institute of Atmospheric Sciences and Climate Change (ICACC) building at the UNAM campus located in southern MC (19°20′N, 99°10′W, 2440 m a.s.l.). Figure 1a shows the MC location in central Mexico. Although PM 2.5 concentrations were measured at four RAMA stations (Fig. 1b), the primary sampling site was located at the ICACC station (Fig. 1c), which has a Solar Radiation Observatory (OSR) with a suitable platform for the installation of the CIMEL sun photometer (Fig. 1d).
Altzomoni atmospheric observatory. The sampling was carried out at the AAO (19.117°N, 98.654°W), located approximately 60 km southeast of MC, 70 km northeast of Cuernavaca, and 50 km west of Puebla (Fig. S1). Its proximity to the active Popocatepetl volcano (~ 12 km), makes the AAO a strategic site for studying the impact Fine particulate matter (PM 2.5 ) measurements. Continuous PM 2.5 concentrations were obtained from the RAMA (http:// www. aire. cdmx. gob. mx) using the beta-attenuation methodology. The RAMA air quality network has 30 stations deployed throughout MC. Additionally, the RAMA has a laboratory for equipment maintenance and calibration. PM 2.5 concentrations for this study were obtained from the following 4 monitoring stations (Fig. 1b): • Instituto de Ciencias Atmosféricas y Cambio Climático (ICACC) station, located in the south of MC. Aerosol morphology and elemental analysis. TEM, SEM, and EDS analysis. The electron microscopy analyses were performed at the Central Microscopy Laboratory of the Institute of Physics of the UNAM (IFU-NAM). Electron microscope images for morphological characterization were taken in a high-resolution TEM, JEOL JEM-2010F microscope (FasTEM, JEOL, Tokyo, Japan) operated at 200 kV near the Scherrer focus, with a theoretical point-to-point resolution of 0.20 nm and a spherical aberration of 0.5 mm. To obtain the aerosol images, copper TEM-grids with particle samples were introduced into the microscope. The images were obtained after reaching the optimum vacuum. The images were recorded with a charge-coupled device (CCD) camera and processed with the GATAN digital micrography system (version 3.7.0, Roper Technologies, Inc., Sarasota, FL, USA). To obtain the micrographs of the soot particles the TEM was operated at high magnifications. SEM images were obtained with a field emission ultra-high-resolution SEM JEOL-JSM-7800F equipped with an Oxford Instruments Energy-dispersive X-ray Spectroscopy (EDS, Oxford Instruments, Abingdon, Oxfordshire, UK) detector and the AZtec 2.1 analysis software. The SEM operating voltages depend on the substrate and fluctuated between 1 and 10 kV. Micrographs obtained with SEM and TEM were analyzed using the ImageJ and Scion 4.0 software, which are public domain. EDS spectra were used to analyze the elemental compositions of particles with diameters from nanometers to micrometers. Expanded electron beams covering whole particles (~ 1 µm) were used. In this way, the average compositions of individual particles were determined. Various particles were measured in the TEM employing magnifications from ~ 2000 × to 5000 ×.
X-ray fluorescence (XRF) analysis. A custom-built XRF spectrometer for environmental applications equipped with an X-ray tube with Rh anode operated at 50 keV and 500 µA (Oxford Instruments, Mountain View, CA, USA) was used for PM 2.5 elemental analysis. The detection system consisted of an Amptek X-123SDD spectrometer with a resolution of 120 eV at 5.9 keV. The Teflon filters (Whatman) with PM 2.5 samples collected with a MiniVol (AirMetrics) from 12 h at a flow rate of 5 L min −1 were placed in the analysis chamber at a high vacuum (10 −6 torr). The XRF spectrum was collected for 900 s and subsequently integrated with the Quantitative X-ray Analysis System (QXAS). The XRF spectrometer calibration procedure was achieved using thin-film standards (MicroMatter Co., Vancouver, Canada), irradiated for 300 s under the same conditions as the sample analysis. Then, accuracy checks were performed using the NIST standard reference material 2783 38 www.nature.com/scientificreports/ Extinction coefficient and AOD measurements. Measurements of the β ex were obtained from the database of the University Network of Atmospheric Observatories (RUOA, for its acronym in Spanish) https:// www. ruoa. unam. mx 26 . The β ex is obtained from a Photoacoustic Extinctiometer (PAX) (Droplet Measurement Technologies, Boulder, CO), operated at a flow rate of ~ 1 L min −1 , which is a sensitive and high-resolution instrument that measures the optical properties of aerosol particles 40 . The PAX uses a 532 nm diode laser to simultaneously measure absorption and scattering coefficients. AOD measurements were performed with a CIMEL sun-photometer (CE-318) associated with AERONET 41 and located at the UNAM main campus in southern MC. The CE-318 device is a spectral radiometer (340, 380, 440, 500, 675, 870, and 1020 nm) programmed to automatically track the Sun following scheduled procedures to estimate the AOD. All measured optical parameters can be download from the AERONET database (https:// aeron et. gsfc. nasa. gov), as well as the fine-(FM) and coarse-modes (CM) contributions to the total AOD, estimated via the Spectral De-convolution Algorithm (SDA) 42,43 . Cloud-screened (Level 1.5) data, following the methodology described by Smirnov et al. (2000) 44 , was used in this study.
Estimation of the visibility. V a is defined as the maximum horizontal distance that the human eye can see.
In the present study V a was calculated from β ex values by using the Koschmieder equation (Eq. 1) 17 : where β ex is the extinction coefficient measured at 532 nm.
Additionally, V a was calculated using Eq. (2) from Baumer et al. (2008) 17 , assuming β ex independent of height (i.e., along the atmospheric vertical column). This estimate also assumes that all aerosol is located within the mixing layer with a height Z i .
where the mixing layer height (Z i ) was considered constant with values of 1.0, 1.5, and 3 km taken from Ceilometer measurements. AOD 500 is the AOD at 500 nm measured by the CIMEL sun-photometer. To determine the influence of atmospheric suspended particles on reduced solar irradiance, the GHI maximum values for every day were correlated with the corresponding PM 2.5 values. The average daily maximum GHI registered for May was calculated and subtracted from the maximum irradiation values of each day. This allows observing the solar radiation reduction during the HAP days.

Global solar radiation measurements. Hourly Global
Additionally, solar irradiances under cloud-free sky conditions for the ICACC station in MC were calculated with the European Solar Radiation Atlas (ESRA), which is used in the Heliosat-2 model 18 . ESRA is an empirical model based on a climatological monthly means of the Linke turbidity (T L ) factor. Values of T L for different regions of the world can be found on the Solar Radiation Data (SODA) website (http:// www. soda-pro. com/).

Ceilometer measurements. Continuous vertical profiles of the backscattered laser signal were measured
with a Vaisala CL31 Ceilometer based on the lidar technique, which uses light pulses sent to the atmosphere from a laser source. Subsequently, the Ceilometer measures the elastically scattered waves returning to the surface 23 . The Vaisala CL31 uses an indium-gallium-arsenide pulsed-diode laser emitting 910 nm pulses at a repetition rate of 10 kHz. The device also uses a single lens for transmitting and receiving light. The maximum detectable cloud-base height of this instrument is 7500 m a.g.l. Backscattered profiles were recorded every 2 to 120 s at a maximum vertical resolution of 10 m 23 .
HYSPLIT model. The 24-h air-mass back-trajectories arriving at the AAO at the 500, 1000, and 1500 m a.g.l. were calculated using the Hybrid Single-Particle Lagrangian Integrated Trajectory Model (HYSPLIT) from the National Oceanic and Atmospheric Administration (NOAA). The HYSPLIT model allows computing simple air parcel trajectories of air pollutants 45 . The back trajectories for the 4 days of the HAP episode (i.e., May 14-17) were calculated using the Real-time Environmental Application and Display System (READY).

Results and discussions
Heavy aerosol pollution episode in Mexico City. The central part of Mexico during the dry-warm season is dominated by an anticyclonic system, which leads to sunny, warm weather and low precipitation in the MCMA 46,47 . These hot and dry conditions favor the presence of dust and BB emissions from wildfires 11 . Nearly 7410 fires and almost 589,371 hectares burned were reported in Mexico during 2019 (CONAFOR, 2020) 48 . Figure 2 shows the location of active fires in Mexico detected by the Visible Infrared Imaging Radiometer Suite (VIIRS) on May 11, and 13. Fires occur throughout Mexico but are concentrated in the western, central, and southwestern regions. Satellite imagery shows the dense smoke plumes emitted by the wildfires, which reached www.nature.com/scientificreports/ the central Mexico plateau. There were also a significant number of fires that occurred inside and around the Mexico City basin.  Table 1.    10 reported similar SDA results in MC, demonstrating that fine particles highly contribute to AOD. Similarly, another study of aerosols in the MC during spring 2019 found that fine particles originate mainly from BB and local traffic emissions, while the primary source of coarse particles is dust from re-suspended soil dust 38 . A remarkable reduction in V a was also noted in MC during the HAP event. Figure 5 shows the scatter plot of AOD at 500 nm and both V a calculations (i.e., β ext and AOD 500 ), where a good correlation for Z i = 3.0 km was obtained. The impact of the atmospheric pollution on V a can be estimated for the 7-days of the HAP period, as seen in Table 1 for 24-h averages, V a suffered degradation of ~ 80% during the HAP days.

Aerosol optical depth and visibility.
TEM analysis of aerosol particles sampled at MC during the HAP episode evidences the presence of soot particles and tarballs (TB) (Fig. 6). TEM micrographs show particles of different sizes and morphologies. For example, Fig. 6a shows spherical particles with diameters (d p ) < 1 µm. These particles have an elemental composition dominated by a strong C and minor O signal (Fig. 6c), a characteristic composition of TB, a type of particles originating from incomplete combustion of fossil fuels and biomass 49,50 . Although the samples collected showed a high amount of TB particles, it is difficult to know the exact origin of these particles, due to the various sources within the MC. Figure 6b shows the TEM image of TB particles attached to soot agglomerates, whose EDS spectra consist mainly of C, O, and high-intensity K and S signals (Fig. 6d).
Aerosol effect on global solar irradiance. The influence of aerosols on daily solar irradiation is analyzed in Fig. 7. The observed increase in PM 2.5 concentrations from 10 to 17 May had an obvious effect on GHI (Fig. 7a, shaded period). The data indicate that GHI gradually decreases as PM 2.5 levels increased, mainly due to the scattering and absorption of sunlight 51 . The significant reduction in irradiance occurs mainly between 11:00 h and 18:00 h LST (Fig. S2a), when maximum irradiance values are detected, coincides with the highest presence of smoke. To quantitatively estimate the PM 2.5 impact on the solar irradiance, the maximum daily GHI measurements were subtracted from the monthly mean GHI m value. This difference (∆GHI = GHI − GHI m ) taken as an anomaly in percent (or departures from the mean monthly value) has a direct impact on visibility. Figure 7b shows the ∆GHI and the PM 2.5 trends, where it can be observed that the maximum GHI has a signifi-      Figure S2b shows the correlation between these quantities, with a correlation coefficient of 0.54. The negative tilt matches solar irradiation reduction, corresponding with the PM 2.5 values increasing. The average values of the GHI and its anomaly during the VPAQ and PAQ days of May 2019 are shown in Table S1.
Aerosol characterization at the AAO. Vertical transport of pollutants within the mixing layer. Ceilometer measurements were used to estimate the height of the ML, required to infer the vertical distribution of aerosols. The derivation of the ML height during spring is reliable due to suitable meteorological conditions (low humidity, clear skies, and the absence of precipitation) and high aerosol loading in the atmosphere. Figure S3 shows the resulting profile of the ML height estimates for the MC on May 14, 2019. The sharp decrease in aerosol backscattering between the ML and FT (contrasting colors in Fig. S3a) indicates the boundary between the ML and the FT. The ML time series starts in the early morning hours, with an average height of 900 m a.g.l, until midday when it increases rapidly, reaching heights > 3500 m a.g.l. due to turbulence and dry convective processes. The ML collapses after sunset, as seen in the height decrease around 18:00 LST. ML expansion was also evident with the increase in particle concentration at the AAO. Figure S3b shows the total particle concentration (sizes > 30 nm) measured with a condensation particle counter (CPC) at the AAO. A rapid increase in particle concentration was observed at 11:00 LST, and subsequently, the maximum value was reached at 11:30 LST. Similar results were published by Baumgardner et al. (2009) 22 , who found that particulate matter concentration and other pollutants reach their maximum concentration at mid-afternoon.
Aerosol particle types at AAO. Analysis was carried out on 120 particles and based on the morphology and elemental composition as the main criteria, allowing the particle classification into five groups (Table 2): soot, organic, mineral dust, S-rich, and complex secondary particles. Statistical analysis of the spectra showed that 90% of the particles contain C and O, 50% of the particles contain Si, and 30% present S. Fe and Al were also detected although in a low number of particles with weak signals. Copper was not considered for this classification because this element is present in the TEM grids. In this work, the morphology and the elemental analyses were the main factors used to identify soot (chain aggregates) and TB (spherical particles) present in almost all the TEM-grids analyzed.
Soot. Figure 8 shows TEM images of individual soot particles sampled at the AAO site within the FT (i.e., 12:00-05:00 h) during the HAP days. All images show chain-like agglomerated structures of nano-sized primary spherical particles with d p < 60 nm, a typical structure present in soot originated in combustion processes 52 . Those soot particles sampled at the AAO could probably have been produced by BB events or were transported by the ML convective process. The soot particle in Fig. 8a has a d p of ~ 1.7 µm, while the particle in Fig. 8b with a d p ~ 0.7 µm is attached to the bigger one and more compact aerosol. EDS spectra in Fig. 8b consist mainly of three peaks; the most intense is the amorphous carbon signal (~ 0.28 keV), Si, and oxygen (~ 0.53 keV). However, other peaks observed in the EDS (i.e., S and Cl) show the mixed state of the soot. Figure S4 shows an SEM image and the EDS elemental composition mapping of soot particles sampled at the AAO within the ML. The EDS map shows the presence of Si, Al, K, Ca, and Fe, homogeneously distributed throughout the particle which indicates that this particle is an aged aerosol. The high percentage of Si, Al, and O suggests the presence of material with a geological origin that resulted from the resuspension of soil dust.
Mineral dust particles. Mineral dust aerosol at the AAO comes mainly from the resuspension of soils and probably from rocks eroded by the wind. These particles with d p < 600 nm (Fig. 8 c and d) show compact shapes and are mixed with other inorganic materials. Figure 8d shows the elemental composition of a mineral particle where www.nature.com/scientificreports/ a high Si signal is observed, which dominates the composition of mineral particles in this region. Additionally, the mineral particles were found to be mixed with small amounts of aluminosilicates, iron-rich dust, K, and minor Ca. These elements could show the presence of feldspars whose main source could be the erosion of the rocks or could be the result of volcanic emissions 53 . The volcanic ash emitted by the Popocatepetl is one of the main components of the soils that surround the AAO. Although during the sampling days there was no direct influence of the volcanic plumes, the soils surrounding the AAO are covered with material emitted previously. The elemental EDS map in Fig. S5 for a dust particle shows a homogeneous distribution of Si, Al, Mg, and Fe which demonstrates the geological origin of that particle. The presence of C and K with a uniform distribution also indicates that this particle is covered with organic material possibly originated from BB.
S-rich particles. Sulfur is one of the most frequently observed elements in the EDS spectra of particles measured at the AAO. The S present in the particles mainly comes from three sources: gas emissions from the Popocatepetl volcano, anthropogenic emissions from MC, and BB emissions. Volcanic emissions are the most important due to the proximity of the AAO to the volcano, which is known as one of the largest sulfur dioxide (SO 2 ) sources in the world 28,29 . Volcanic plumes consist of gases and sub-millimeter particles 28,29 . Figure 8e shows a TEM image of an individual S-rich aerosol, and Fig. 8f presents the EDS of this type of particle, both sampled at the AAO during HAP days. S-rich particles present differents morphologies with d p < 800 nm. Some particles presented www.nature.com/scientificreports/ crystalline structures 54 , but most of them suffered decomposition or evaporation as the microscope beam hit the surface. These particles were beam-sensitive and undergo some changes in their shape during analysis.
Tarballs. Figure 9a shows a TEM image of the spherical organic particles with d p < 600 nm, which was common in all aerosol samples collected at the AAO site. Those TB are particles with a special morphology (nearspherical) and composition (carbonaceous material) are quite abundant in biomass smoke plumes 50,55,56 . These particles in the Altzomoni mountain could have two origins: BB and anthropogenic urban emissions. Intense C peak in the EDS spectra confirmed the occurrence of TB at the AAO during the HAP episode (Fig. 9b).
Complex secondary particles. A significant number of secondary aerosol particles were also measured at the AAO. These particles are characterized by having complex elemental compositions and very irregular shapes (with d p between 0.5 and 1.6 µm). Figure 9c shows the TEM micrograph of a secondary particle whose morphology was affected by the incident beam of the microscope, and Fig. 9d presents a typical EDS spectrum of this type of particle where S and K signals predominate. The main sources of these particles can be anthropogenic emissions in urban areas and BB emissions. It was observed that most of these types of particles were affected by the strong electron beam of the TEM, evidencing the evaporation of some volatile compounds, which induced changes in their shape.
Most of the particles analyzed in this study were internally mixed, which evidences the presence of aged particles. However, it was also observed aerosol assemblies externally mixed, during the FT hours. Figure S6 shows an example of an externally mixed particle with a soot aggregate surrounded by other particles with different compositions (mainly S). The inset in Fig. S6 is a high magnification image of an internally mixed soot coated with S and K.
Elemental composition. A comparison of the elemental composition of PM 2.5 particles sampled at the AAO and MC is shown as pie charts in Fig. 10. The main elements detected with the XRF analysis were Al, Si, P, S, K,  www.nature.com/scientificreports/ Ca, Mn, Fe, Ni, Cu, and Zn. Figure 10a shows the elemental analysis of the particles sampled at the AAO, where the composition is dominated by S and K with 46% and 35%, respectively. The abundance of these elements suggested an important contribution of the emissions from wildfires, mainly the K, which is a tracer of BB 57,58 . The existence of dust-like aerosol could be evidence by the presence of Si and Fe with significant contributions of 5% and 10%, respectively, in addition to minor contributions of Al, Mn, and Ca. Although the AAO is located in a remote rural area, small amounts (< 1%) of Ni, Cu, and Zn were measured. These elements could be related to anthropogenic sources close to the sampling site (e.g., chimney of the TV-broadcast antenna facility). Figure 10b shows the elemental composition of particles sampled in MC, which presents values very similar to those measured at the AAO, i.e., with S and K as the elements contributing more to elemental composition with percentages of 37% and 38%, respectively. Although S can be the product of BB, it can also come from other anthropogenic sources (i.e., motor vehicles), being the main element responsible for the production of secondary aerosol particles 58,59 . Mineral-dust aerosol is generally a significant component during the dry-warm period 33,60 . The presence of Fe and Si with percentages of 10% and 7%, respectively, is the primary evidence of mineral dust particles. That mineral fraction in MC is partially a result of the MC semi-arid areas (e.g., former Texcoco and Chalco lakes), arid hills, and unpaved roads within the MC basin 38,61 . The significant Ca contribution could be due to fly ash emissions from two sources: construction activities and soil-dust resuspension 38 . Furthermore, the trace elements with percentages less than 1% (Ni, Cu, and Zn), have been associated with the emissions generated by high traffic in the urban area. For example, most of the Zn detected could originate from the wear of vehicle tires, while Ni and is a typical element in motor fuel additives 38 10 , who observed elevated sulfate at the higher altitudes above MC. Similar studies of aerosol chemical composition conducted at other high-altitude sites in different parts of the world (in the absence of BB emissions) reveal that aerosols in the FT contain a high fraction of sulfates 10,62,63 .
Air masses back-trajectories. BB smoke is one of the main atmospheric components affecting air quality and climate in Mexico due to massive plumes that can travel thousands of kilometers downwind 11,12 . Tracking of these plumes is only possible through satellite measurements, e,g, by the VIIRS radiometer 64 . The HYSPLIT model was used to identify the origin of the air masses reaching the sampling sites. Figure S7a-d show the HYSPLIT back-trajectories computed at different heights (100, 500, and 1000 m a.g.l.) during the HAP episode on May 13, 14, 16, and 17, respectively. The HYSPLIT simulations show that during the HAP episode the air masses mainly originate in western and southwestern Mexico where most of the active fires are concentrated (see fire distribution in Fig. 2) to reach the AAO. However, these air masses also cross over Morelos State, where the cities of Cuautla and Cuernavaca are located, possibly indicating not only volcanic but also urban emissions in the air mass reaching the AAO. Only on May 17, there was a possible transport from north of the AAO, carrying particles emitted in MC.

Conclusions and discussions
This study successfully combined remote sensing techniques, in-situ measurements, and laboratory analysis to characterize the atmospheric aerosol in MC and the AAO during the HAP period in May 2019. The high levels of PM 2.5 measured by the RAMA stations throughout the MCMA, evidenced the large amount of aerosol particles. Along May 2019 there was poor air quality, which got significantly reduced with the spread of the wildfires emissions. AERONET results suggest that the extinction of solar radiation in MC was mainly dominated by fine-mode particles. The GHI measured during the HAP was 17% lower than those calculated with the ESRA clear-sky model. It is possible that the high particle loads and dense smoke plumes contributed to the reduction in GHI, which affected both the horizontal and the vertical visibility during the VPAQ days. Although mineral dust particles and other sulfur-rich particles were observed in the collected aerosol samples, the predominant species were soot and TB. The latter two particles were observed simultaneously in the MC and the AAO. Likewise, although the XRF elemental composition of the particles shows similarities in both sampling sites, the higher percentage of S at the AAO particles should be noted. We believe that the higher percentage of S at the AAO suggests a contribution from the regional transport of particles in the FT. www.nature.com/scientificreports/