Phase 3 of the Integrated Campaign for Aerosols, gases, and Radiation Budget cruise-based experiment (hereafter referred to as the ICARB-2018) was carried out during the winter period (16 January–13 February 2018) along the track shown by the solid black line in Fig. 1, covering different parts of the south-eastern Arabian Sea (SEAS), the northern Indian Ocean (NIO), and the equatorial Indian Ocean (EIO), as highlighted by the different boxes about the track. More details about the experiment and sampling conditions are available in earlier publications (Gogoi et al., 2019; Nair et al., 2020; Kompalli et al., 2020a). Briefly, the measurements were made from the specially configured aerosol laboratory on the top deck of the ship, ∼15 m above the sea level, and the instruments sampled air from a community aerosol inlet set up with an upper size cut-off at 10 µm at a flow rate of 16.67 litres per minute (LPM). Membrane-based dryers were installed in the sampling lines to remove the excess moisture (to limit the sampling relative humidity (RH) to <40 %). Proper care was taken to avoid contamination from the ship's emissions by aligning the bow of the ship against the wind direction, and any spurious data were removed during post-processing (as has been done in earlier such campaigns; Moorthy et al., 2008).

The South Asian region is known for its seasonally contrasting synoptic meteorology associated with the Asian monsoon and north–south excursion of the inter-tropical convergence zone (ITCZ) and monsoonal circulations (Das, 1986; Asnani, 1993). During the winter (December to February), calm north-easterly winds prevail over the Indian landmass, which facilitates extensive transport of continental air mass to the surrounding ocean. The synoptic conditions during the campaign period were quite similar to the climatological pattern, as revealed by Fig. S1a in the Supplement, which shows the prevailing synoptic mean wind vectors at 925 hPa derived from ERA-Interim wind data from ECMWF (European Centre for Medium range Weather Forecasting; https://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/, last access: 25 February 2020) data. The spatial distribution of fire counts over the continental landmass lying upwind of the campaign area, as derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) fire radiative power (MODIS Thermal Anomalies/Fire locations, Collection 6 product obtained from https://earthdata.nasa.gov/firms, last access: 24 February 2020) for the period 10 January to 14 February 2018, is shown Fig. S1b in the Supplement. It reveals a significant number of fire events in the upwind regions. Monthly mean tropospheric NO2 column abundances obtained from TROPOspheric Monitoring Instrument (TROPOMI) (http://www.temis.nl/airpollution/no2col/no2regio_tropomi.php, last access: 27 February 2020) data, shown in the bottom panels of the same figure (for January 2018 see Fig. S1c and for February 2018 see Fig. S1d in the Supplement), show significant emissions over the continental areas upwind, a part of which would be transported to the oceanic regions during the ICARB-2018. Earlier, based on the observations using the optical attenuation technique (aethalometer) over the upwind locations (Kharagpur, Bhubaneswar, Vizag, Trivandrum), Kompalli et al. (2013, 2014b) reported that the highest equivalent black carbon (EBC; optically measured BC) mass concentrations throughout the year are seen during the winter period (mean values ranging from ∼5389±1245 ngm-3 over Trivandrum to 11691±4457 ngm-3 over Kharagpur), which highlighted the source strength during this season. Also, the east coast of India is more industrialized compared to the west coast and/or peninsular India (Fig. S1c and d in the Supplement; Moorthy et al., 2005; Kompalli et al., 2013).

Air mass back trajectories derived using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) (https://www.arl.noaa.gov/hysplit/ready/, last access: 20 September 2019) and shown in Fig. 1 highlight the potential long-range transport of the continental emissions to different oceanic regions covered during the campaign. Accordingly, the cruise region is divided into five distinct sub-regions: (i) south-eastern Arabian Sea (SEAS), which encountered direct outflow from the strong source regions in the western coastal and peninsular India region (shown with a blue box and air mass trajectories); (ii) northern Indian Ocean-east (NIO-E) (eastern leg of the cruise covering the NIO region) that experienced air masses from the east coast of India and the Bay of Bengal regions (red); (iii) equatorial Indian Ocean (EIO), where mostly marine air masses originated and/or confined within the north-eastern Arabian Sea and without any direct influence of continental outflow (green); (iv) northern Indian Ocean-west (NIO-W), which experiences outflow mainly from western coastal regions of Peninsular India after considerable transit over the Sea (magenta); and (v) southern Arabian Sea (SAS), the unshaded region of the track where widespread rainfall associated with the passage of a large scale meteorological system was encountered. We have not included the data collected over the SAS region in the overall analysis of the present study, and a brief discussion about it is provided in the Supplement. During the rest of the cruise period, calm winds (<5 ms-1) and clear sky conditions prevailed with no significant variation in air temperature (mean ∼28±0.8 ∘C) and relative humidity (mean ∼73±5 %) conditions. Table 1 gives the details of the regional mean (avg), maximum (max) and minimum (min) values of meteorological variables (air temperature (AT), relative humidity (RH), wind speed (WS), wind direction (WD), total accumulated rainfall (RF) amount) for different regions covered during the ICARB-2018.

Of the several measurements made, the measurements of the BC single particle microphysical properties were carried out using a single-particle soot photometer (SP2) (Model: SP2-D; Droplet Measurement Technologies, Boulder, USA), which was operated at a flow rate of 0.08 Lmin-1. The SP2 employs a 1064 nm Nd:YAG intracavity laser and by using a laser-induced incandescence technique, it characterizes the physical properties of refractory BC (rBC) at the individual particle level (Moteki and Kondo, 2007; Schwarz et al., 2008, 2013; Laborde et al., 2012; Liu et al., 2014; Shiraiwa et al., 2007; Kompalli et al., 2020b). It provides information about mass and number concentrations and size distributions of rBC. The amplitude of the incandescence signal is proportional to the rBC mass present in the BC-containing particles, and the mass equivalent diameter (the diameter of a sphere containing the same mass of rBC as measured), or BC core diameter (Dc), is obtained from the measured rBC mass by assuming a density, ρ∼1.8 gcm-3, for atmospheric BC (Bond and Bergstrom, 2006; Moteki and Kondo, 2007, 2010; McMeeking et al., 2011). Further, the amplitude of the scattering signal provides information about the scattering cross section of the particles, which is used to determine the optical sizing of the particles. In the case of BC-containing particles, the scattering signal gets distorted as it passes through the laser beam because of the intense thermal heating of the particle and evaporation of the coating. Thus, the scattering signal of the BC particle is reconstructed using a leading-edge only (LEO) fitting technique, as described in earlier publications (Gao et al., 2007; Liu et al., 2010, 2014, 2017), and this scattering cross section is matched with the modelled values in a Mie lookup table to derive the optical diameter of a BC particle or the coated BC size (Dp). Here, the total particle is treated as an ideal two-component sphere with a concentric core–shell morphology, with a core (rBC) refractive index value of 2.26-1.26i (Moteki et al., 2010; Liu et al., 2014; Taylor et al., 2015) and a coating refractive index of 1.5+0i (which is an optimum value and in the range of refractive indices of inorganic salts ((NH4)2SO4=1.51; NaCl=1.53) and secondary organic aerosols (∼1.44–1.5) at λ=1064 nm (Schnaiter et al., 2005; Metcalf et al., 2012; Lambe et al., 2013; Laborde et al., 2013; Taylor et al., 2015). These two diameters (Dp and Dc) are used to infer the coating thickness. Before the experiment, the SP2 was calibrated using Aquadag^® black carbon particle standards (Aqueous Deflocculated Acheson Graphite, manufactured by Acheson Inc., USA), and a correction factor of 0.75 is applied to address the difference between Aquadag^® standards and ambient BC (e.g., Moteki and Kondo 2010; Laborde et al., 2012). A detailed description of the instrument, data interpretation procedures, uncertainties, and caveats involved can be found elsewhere (Liu et al., 2010, 2014; McMeeking et al., 2010; Sedlacek et al., 2012, 2018; Kompalli et al., 2020b). It is recognized that the SP2 cannot provide the details of rBC aggregate morphology or the relative position of the BC within the particle, which can be determined better through microscopy-based studies (e.g., Adachi et al., 2010; Ueda et al., 2018). However, the intensity of the incandescence signal detected by the SP2 is proportional to the refractory black carbon mass present in the particle and is independent of particle morphology and mixing state (Slowik et al., 2007a; Moteki and Kondo, 2007; Schwarz et al., 2008). Again, though the SP2 has limited detection sensitivity towards pure scatterers because of the limited size range it covers, the light scattering information at 1064 nm has been widely used to accurately derive the size of the coated particle (Gao et al., 2007; Moteki et al., 2010; Shiraiwa et al., 2008, 2010; Laborde et al., 2013; Taylor et al., 2015; Liu et al., 2017).

Supplementing the above, we used the information on the mass concentration of non-refractory PM1.0 aerosols (organics, sulfate, ammonium, nitrate, and chloride) from a collocated aerosol chemical speciation monitor (ACSM; Model: 140; Aerodyne, USA) (Ng et al., 2011). The objective here is to identify the possible coating material on rBC particles. The ACSM consists of a particle sampling inlet, three vacuum chambers (differentially pumped by turbopumps, backed by the main diaphragm pump), and a residual gas analyser (RGA) mass spectrometer (Pfeiffer Vacuum GmbH). The particles are drawn to an aerodynamic lens assembly having D50 limits (50 % transmission range) of 75–650 nm and 30 % to 40 % transmission efficiency at 1 µm (Liu et al., 2007) through a 100 µm critical orifice. These particles are focused into a narrow beam and transmitted to a vacuum environment where they are flash vaporized by the thermal capture vaporizer (Xu et al., 2017; Hu et al.,  2017a, b) operating at 525 ∘C. Subsequently, these vapours are ionized via 70 eV electron impact ionization and detected with a quadrupole mass spectrometer. The data are processed as per the prescribed methodology (Ng et al., 2011; Middlebrook et al., 2012; Kompalli et al., 2020b). We used software provided by the manufacturer (Aerodyne Research, ACSM Local, version 1.6.0.3, within IGOR Pro version 7.0.4.1) for processing and analysis of data. Using the default fragmentation table (Allan et al., 2004), the measured fractions of unit mass resolution spectra signals were apportioned to individual aerosol species. The required corrections for the instrument performance for the varied inlet pressures and N2 signal were performed (Ng et al., 2011; Sun et al., 2012). Mass-dependent ion transmission efficiency correction of the residual gas analyser was carried out using the signals from the internal diffuse naphthalene source (m/z 128). The calibrations of ionization efficiency (IE) and relative IE (RIE) calibrations were performed prior to the experiment by using monodisperse (300 nm) particles of NH4NO3 and (NH4)2SO4 (Jayne et al., 2000; Allan et al., 2003; Jimenez et al., 2003; Canagaratna et al., 2007). The present ACSM consists of a capture vaporizer with an inner cavity to reduce the particle bounce (Xu et al., 2017), resulting in a higher collection efficiency (about unity) (Hu et al., 2017a, b). Therefore, the composition-dependent collection efficiency correction prescribed by Middlebrook et al. (2012), applicable to standard vaporizer instruments, was not applied to our data. More than 1200 quality checked individual observations with a time resolution of ∼30 min formed the database for this study.

Continuous measurements of the particle number size distributions in the size range of 10 to 414 nm were also carried out aboard, at the 5 min interval, using a scanning mobility particle sizer spectrometer (SMPS; TSI Inc., USA) during the campaign (Kompalli et al., 2020a). The SMPS consists of an electrostatic classifier (TSI 3080), a long differential mobility analyser to size segregate the particles based on their electrical motilities (Wiedensohler, 1988) that are subsequently counted by using a water-based condensation particle counter (TSI 3786). Concurrent measurements of the particle number size distributions in the aerodynamic diameters range of 542 to 19 800 nm (which can be converted to stokes diameters using an effective particle density) have also been made using the aerodynamic particle sizer (APS; Make: TSI, Model: 3321) that works based on the “time-of-flight” technique (Leith and Peters, 2003). Though the contribution from the particles in the sizes measured by the APS to the overall aerosol number concentrations are found to be <2 %; combining both these measurements gives the total particle number concentrations covering a wide size range (10–10 000 nm). We have used the particle number size distributions (and total particle number concentrations) from 10–1000 nm from the SMPS+APS measurements, along with the number concentration of rBC from the SP2 to estimate the fraction of rBC-containing particles.

The extent of coating on rBC particles is quantified in terms of the bulk relative coating thickness (RCT) defined as Dp/Dc, where Dp and Dc are coated and core BC particle diameters, respectively. It is estimated by dividing the total volume of coated BC with that of rBC cores in a given time window (5 min in our case) following Liu et al. (2014, 2019): 1DpDc=∑iDp,i3∑iDc,i33, where Dp,i and Dc,i are the diameters of coated and core rBC, respectively, for each single particle i. In addition to RCT, we have used bulk volume-weighted absolute coating thickness (ACT, in nm), defined as (Dp-Dc)/2 (both Dp and Dc used here are volume averaged diameters) based on the assumption of a concentric core–shell morphology, as another diagnostic of the coating on the population of rBC particles (Gong et al., 2016; Cheng et al., 2018; Brooks et al., 2019). More details about the parameters bulk RCT and ACT, the methodology used here, and uncertainties associated were described elsewhere (Liu et al., 2017, 2019; Sedlacek et al., 2018; Brooks et al., 2019; Kompalli et al., 2020b).

The means of the mass median diameter (MMD) and number median diameter (NMD) were determined from the size distributions of BC cores for each time window by least squares fitting to an analytical monomodal log-normal distribution (Liu et al., 2010, 2014; Kompalli et al., 2020b) of the following form: 2dXdln⁡Dq=∑X02πln⁡σmexp⁡-ln⁡Dq-ln⁡Dm22ln⁡σm. Here X0 corresponds to mass/number concentration of the mode, Dm is the mass/number median diameter, Dq is particle diameter, dX is mass/number concentrations in an infinitesimal diameter interval dln⁡Dq, and σm is the geometric standard deviation (of the median diameter).

Using the bulk RCT and MMD of the BC cores, the volume-weighted coated BC size (Dp,v) is calculated as below to indicate the mean coated BC size: 3Dp,v=DpDc×MMD. Since the ratio of the mass of non-absorbing coating material to the rBC core is an important parameter in determining the degree of absorption enhancement of BC, we quantified their mixing in terms of the bulk mixing ratio of coating mass over rBC mass (MR,bulk) derived by assuming densities for the bulk coating (ρcoating) and rBC core (ρrBCcore∼1.8 gcm-3) (Liu et al., 2019) as below: 4MR,bulk=DpDc3-1×ρcoatingρrBCcore. Here we have used the effective dry density (ρcoating) of ambient NR-PM1.0 based on the measured near real-time chemical composition (Budisulistiorini et al., 2016) by assuming densities of organics and inorganics as 1.4 (Hallquist et al., 2009) and 1.77 gcm-3 (Park et al., 2004), respectively.

To explore the distribution of BC core-coatings, a parameter of scattering enhancement (Esca) for each single particle is determined using the expression (Liu et al., 2014, 2019): 5Esca=Smeasured,coatedBCScalculated,uncoatedBC, where the term in the numerator is the scattered light intensity of the coated rBC particle measured by the scattering detector of the SP2 and reconstructed using the LEO technique, while the denominator is the scattering intensity of the uncoated BC calculated using Mie single particle scattering solutions assuming sphericity (Liu et al., 2014, 2019; Taylor et al., 2014; Brooks et al., 2019). For this purpose, the measured rBC mass and a refractive index of BC of ∼2.26±1.26i (Moteki et al., 2010) at the SP2 laser wavelength, ∼1064 nm, were used. For an uncoated rBC particle, Esca is equal to 1, and Esca increases with an increased coating at a given core size. Combined with rBC core diameters and coating parameters, Esca is helpful in identifying the nature of sources, though under the assumption that no material loss via oxidative and/or photochemistry occurs, which can either alter overall particle size and/or the refractive index of the coating.

The spatial variations of the number and mass concentrations of refractory BC core particles during the cruise are shown in Fig. 2, along with the statistics over different regions depicted using the box-and-whisker plots.

The highest values and variabilities (standard deviation) in the rBC mass (mean ∼938±293 ngm-3) and number (∼378±137 cm-3) concentrations are noticed over the SEAS region, which is in the proximity of the source regions and/or outflow from the western coast and peninsular India, where industrialized cities and major ports are located (Figs. 1 and S1 in the Supplement). The concentrations decreased gradually by half as the ship headed towards the NIO-E (eastern leg of the NIO), where it received outflow mostly from the east coast and/or the Bay of Bengal regions. The lowest concentrations (which were 4–5 fold lower than the values seen over the SEAS) are observed over the remote EIO region. The lower concentrations (<200 ngm-3) highlighted the cleaner nature of this region, which encountered mostly oceanic air masses. The concentrations increased again (almost 3-fold compared to the values seen over the EIO region) as the ship traversed to the NIO-W (western leg of the NIO) region, which experienced continental outflow air masses from the west coast and/or peninsular India, similar to the SEAS but farther from the coast and SEAS region. Thus, the concentrations in the NIO-W were lower than those seen in the SEAS and comparable to those seen in its eastern counterpart (NIO-E). This also indicated varied amounts of BC in the outflow from different parts of the peninsula, apparently due to different source strengths and transit times involved. A similar spatial variability pattern was also reported for other aerosol parameters during the present campaign (Gogoi et al., 2019; Nair et al., 2020; Kompalli et al., 2020a).

The size distributions of rBC are influenced by the source, sinks, and transformation processes taking place during advection and are known to be important in assessing the light absorption characteristics (Reddington et al., 2013). The MMD values of the rBC size distributions are strongly influenced by the source of BC emissions (e.g., Ko et al., 2020; Cheng et al., 2018). Recently, Ko et al. (2020) compared the MMD and NMD values of rBC size distributions from different dominant sources. Previous studies report that the MMD (and NMD) values over the regions dominated by fresh fossil fuel emissions are smaller (MMD∼100 to 178 nm and NMD∼30 to 80 nm) compared to the areas with dominant solid-fuel sources (biomass, biofuel, coal-burning) (MMD∼130 to 210 nm and NMD∼100 to 140 nm), whereas well-aged and background BC particles in outflow regions have MMD values in between (MMD∼180 to 225 nm and NMD∼90 to 120 nm) (McMeeking et al., 2010; Liu et al., 2010, 2014; Kondo et al., 2011; Cappa et al., 2012; Sahu et al., 2012; Metcalf et al., 2012; Laborde et al., 2013; Reddington et al., 2013; Gong et al., 2016; Raatikainen et al., 2017; Krasowsky et al., 2018; Brooks et al., 2019; Kompalli et al., 2020b; Ko et al., 2020). The spatial distribution of mass median diameter and number median diameters during the ICARB-2018 shown in Fig. 3 was interpreted based on this backdrop. The observed mean NMD (0.10–0.11 µm) and MMD (0.19–0.20 µm) values over the entire study region (Fig. 3c and d) are within the range of values reported in earlier studies for chemically aged continental outflow and a combination of sources. Chemical ageing of BC is another important factor affecting the rBC core sizes owing to transformation processes (such as collapsing of the BC cores and/or due to coagulation) taking place during the long-range transport (Shiraiwa et al., 2008; Bond et al., 2013; Ko et al., 2020). Freshly produced BC particles comprise fractal-like aggregates of spherical graphitic monomers with diameters of 10–50 nm (Köylü et al., 1995; Bond and Bergstrom, 2006; Bond et al., 2013; Petzold et al., 2013). However, as they evolve in the atmosphere, restructuring of these aggregates occurs due to the above processes and/or condensation of vapours. Compaction can be induced by capillary forces while vapour condensation fills the voids of the aggregates (capillary condensation) (Weingartner et al., 1995; Pagels et al., 2009; Khalizov et al., 2009; Chen et al., 2018, 2016; Ivanova et al., 2021, and references therein) and/or restructuring driven by surface tension forces at the solid–liquid interfaces during condensation of coating material (Kutz and Schmidt-Ott, 1992; Slowik et al., 2007b; Zhang et al., 2008, 2016; Schnitzler et al., 2017). Recently, Ivanova et al. (2021) presented a detailed account of the above processes. As such, increased ageing (temporal and/or chemical) is more likely to result in compact cores (Liu et al., 2019; Laborde et al., 2013); however, the effectiveness of a condensable vapour to cause restructuring also depends on its chemical composition (Xue et al., 2009; Chen et al., 2016). The observed MMD values in this study reflected such transformation processes.

Notably, the NIO-E region depicted a slightly larger mean MMD (∼0.20 µm) due to frequent larger values (35 % of the measurements showed MMD>0.20 µm) compared to all the other regions (Fig. 3d). This is a result of the following possibilities: (i) Self-coagulation of rBC cores due to enhanced atmospheric ageing during their transport from the source regions on the east coast to the adjacent marine regions (at the same time, sedimentation of larger particles resulting in a large reduction in number concentration and mass concentration). It may be noted that coagulation, even though it increases the rBC core diameters and reduces number concentrations, is a slow process. The coagulation rate depends on the square of the particle number concentrations and is the lowest between particles of the same size. Thus, the coagulation rates would be higher near source regions of the nascent aerosols and drop off gradually at farther distances. (ii) The second and most important possibility is associated with the cloud processing of rBC particles. The less-soluble BC particles remain within a non-precipitating cloud as interstitial particles. A cloud undergoes multiple evaporation–condensation cycles before it transforms into a precipitating system. During such cycles, interstitial BC in cloud droplets can grow larger (especially following the evaporation of cloud droplets containing multiple rBC particles) due to agglomeration with other interstitial rBC aerosols. (iii) The third possibility is the varied nature of dominant sources. A sizable increase in the contribution from solid-fuel sources (biomass/crop residue/coal burning) in the upwind regions (the eastern coast of India) through the transported air masses can lead to larger BC cores (Brooks et al., 2019; Kompalli et al., 2020b). Interestingly, the EIO region showed the largest variability with a non-negligible contribution (∼8 %) from smaller BC cores (MMD<0.18 µm). Over the NIO-W, the MMD values remained between 0.18–0.20 µm suggesting advection of BC originating from mixed sources over peninsular India and/or the west coast. It may be noted that the exact sources cannot be identified from the MMD value of rBC size distributions alone. More details on source apportionment are provided in Sect. 3.3.

The spatial distribution of NMD also showed a similar picture to that of MMD over all the regions. We compared the MMD values of rBC observed in our campaign with the values reported from selected locations with distinct dominant sources in different environments in Table 2.

As evident from Table 2, the MMD values during ICARB-2018 mostly fall in the category of BC from the continental outflow and originated from mixed sources (McMeeking et al., 2010, 2011; Ueda et al., 2016; Cheng et al., 2018). Recently, Kompalli et al. (2020b) also reported mean MMD values of 0.18–0.19 µm over Bhubaneswar (located on the east coast of India) during the winter when urban continental outflow with mixed sources from the Indo-Gangetic Plain prevailed. Similarly, Liu et al. (2019) have reported MMD∼0.19–0.21 µm in the urban environment of Beijing with mixed sources. The mean MMD values (Table 2) and mass size distributions over different regions covered in this study (Fig. S2 in the Supplement) revealed that although the peak amplitudes varied in proportion to the magnitude of the BC loading, which decreased with increasing distance from the peninsula, the modal diameters (0.19–0.20 µm) showed little variability, which is also underlined by similar geometric standard deviation values ∼1.55–1.59. This is also consistent with the widespread nature of the continental outflow to the northern Indian Ocean (from west to east) and mixed sources for rBC particles in the outflow (McMeeking et al., 2010).

The above discussions have established that i.

The extent of the coating as measured by the coating thickness and the bulk mixing ratio of coating mass over rBC mass, and hence the mixing of BC with condensable species, is highest closer to the coast where the outflow is strong and decreases in the farther oceanic regions.

The main features in the figure are a.

A reasonable amount of moderately coated BC particles having ACT of ∼50–100 nm with scattering enhancements in the range of Esca∼2–10 were seen, and a significant proportion of moderately coated BC particles with large scattering enhancement (Esca∼10–100), which increased with BC core sizes during the ICARB cruise. This highlighted the substantial contribution from a combination of mixed sources that co-emit BC and condensable material (Liu et al., 2014, 2019).

The high proportion of thick coatings on BC particles may result in significant increases in absorption by the BC. As reported by Brooks et al. (2019) from measurements over the IGP, a significant fraction of moderately coated BC particles with increased scattering enhancement were found to have higher mass absorption coefficient values over the Indo-Gangetic Plain. They attributed this to vigorous mixing between various sources due to the high amounts of secondary aerosol formation and photochemical ageing across northern India. The present scenario, with a steady pollutant outflow containing contributions from diverse sources, is akin to it. Such alterations to the mixing state of BC can contribute to significant enhancement in the absorbing characteristics of BC aerosols over the marine regions as they underdo chemical ageing while transiting from upwind source regions on the west and east coasts of India, which may have noteworthy regional climate implications.

The large uncoated rBC particles (core diameters>0.18 µm and thin coatings of ACT<50 nm) with low scattering enhancements were also found during our measurements but in smaller quantities, consistent with the findings of Brooks et al. (2019) for the Indian region during the pre-monsoon and monsoon seasons. Liu et al. (2019) described that large uncoated rBC particles are indicative of coal-burning emissions and suggested that larger core BC particles with no or thin coatings display minimal Esca values. This can be due to significant bias introduced by the sphericity assumption used in the Mie calculations and a possible difference in the refractive index of BC produced from coal-burning from the value assumed (∼2.26±1.26i) in the Mie calculations used to derive Esca. However, this aspect cannot be addressed in the present study in the absence of measurements of BC morphology or source-dependent refractive index of rBC. As such, a more detailed in situ analysis is necessary to address this.

Information on the nature of the coating material along with the state of mixing of BC particles gives insight into the magnitude of the mixing-induced absorption enhancement for BC (Cappa et al., 2012, 2018; Peng et al., 2016; Liu et al., 2017). We have used the concurrent measurements of non-refractory PM1.0 (NR-PM1.0) aerosol mass and chemical composition data from the ACSM to infer the nature of the coating material during the ICARB-2018. Though it may be possible for BC to be mixed externally with coarse mode aerosols like dust or sea-salt aerosols in the real atmosphere, Fig. 6 directly dispels the importance of this notion.

General features of NR-PM1.0 chemical composition are shown in Fig. 7, which shows the sub-regional mean mass concentrations and mass fractions (MF) of different species over the oceanic regions covered during the cruise.

In line with the pattern for rBC, the overall NR-PM1.0 aerosol mass concentration was also highest over the SEAS (mean∼23.44±5.07 µgm-3), decreasing steadily further away from the continent to reach the lowest concentrations (∼5.48±2.83 µgm-3) over the EIO, where mostly oceanic air masses prevailed. Sulfate and organics were the two major species dominating the NR-PM1.0 composition. Of these two, sulfate dominated most of the oceanic sub-regions with a mass fraction ≥0.4 except for the SEAS region, where organics (∼0.39) were also equally important. The mass fraction of ammonium, which is formed through gas-phase and aqueous-phase chemical reactions of NH3 produced from animal wastes, fertilizers, ocean, and soil, was in the range of ∼0.20–0.23, while nitrate (0.01–0.02) and chloride are negligible in all the regions. To summarize, the South Asian outflow plumes consist of more or less equal proportions of organics and sulfate aerosols in the vicinity of the west coast and/or peninsular India, gradually changing to a sulfate-rich aerosol system in the remote oceanic regions. This marked change from the prominent presence of organics plus sulfate over coastal regions to a strong sulfate dominance over the remote oceanic regions emphasized the atmospheric processes (formation, transformation, and removal) that are at play in determining the life cycle of these species during their transport. Possible oxidation of primary particulate organic matter due to heterogeneous reactions involving oxidants such as OH, O3, and NO3 during long-range transport can result in their volatilization, thus restricting their lifetime (Molina et al., 2004; Donahue et al., 2006; DeCarlo et al., 2010). However, enhanced sulfate production is possible through the gas to particle conversion in the SO2-rich air masses (especially when ambient relative humidity is higher) (Unger et al., 2006; Meng et al., 2016) that originate over the Indian region or through the dimethyl sulfide (DMS) pathway from the marine emissions (Zorn et al., 2008; Shank et al., 2012). During long-range transport, in situ oxidation of SO2 by OH radicals in the gas phase followed by condensation onto pre-existing particles or reaction of S(IV) via H2O2 and O3 in the aqueous phase can lead to enhanced sulfate concentrations. All such processes could have contributed to observed spatial heterogeneity in the observed organics to sulfate concentrations. Detailed investigation of these processes is not within the scope of the present study.

Further, the association between ammonium and sulfate (Fig. S3 in the Supplement) indicated an NH4+ deficit environment (Aswini et al., 2020). In Fig. S4 in the Supplement, the ratio (expressed as %) of non-refractory coating mass on BC to the total NR-PM1.0 mass concentrations (from the ACSM) are shown. The mean ratios (varying between 23 %–35 %) for different sub-regions are also mentioned in the figure.

We examined the association between bulk absolute coating thickness (inferences did not change even if we use RCT) and mass concentrations of NR-PM1.0 organic aerosols, sulfate, and ammonium during the campaign (Fig. 8) when ACT is low (ACT<50 % of MMD) (Fig. 8a–c) and high (ACT>50 % of MMD) (Fig. 8d–f). The colour bar indicates the values of the corresponding MMDs. The solid line is the linear least squares fit between the variables and the corresponding correlation coefficient (Pearson's r) is also shown in the figure.

Regression analysis suggested two distinct regression lines in the sulfate and ammonium plots. The bulk ACT showed a reasonable association (r>0.6; p<0.01) with NR-PM1.0 sulfate and ammonium and a weaker association with organics (r∼0.41; p<0.01) during low (ACT<50 % of MMD) observations. Corresponding slopes indicated that ACT on BC intensified more steeply with an enhanced mass concentration of sulfate (slope∼2.14) compared to organics (slope∼1.18), whereas the steepest enhancement in the ACT was noticed for ammonium (slope∼7.15). In contrast, there is no significant association between bulk ACT and mass concentrations of NR-PM1.0 species during high ACT (ACT>50 % of MMD) observations. In this case, higher ACT values prevailed for all values of sulfate (or organics) and ammonium. The low and high ACT groups of observations are distinctly separated by the corresponding MMD values, where more frequent higher MMD values were seen for the BC population having a higher coating.

Thus, Fig. 8 suggested the possibility of complex coating on BC in the proximity of continental outflow where higher ACT (also higher MMD), significant amounts of organics, and ammonium were observed. In the far oceanic regions, sulfate components can act as a relatively dominant coating material on BC (compared to organics). This is not unexpected considering the chemically aged BC particles that were present in the remote marine boundary layer dominated by the sulfate aerosol system. Previously, Ueda et al. (2018) reported morphological features of soot-containing particles over the Southern Ocean and suggested that aged soot-containing particles were transformed by soluble materials derived from dimethyl sulfide (DMS) oxidation during the summer. Though this DMS oxidation pathway may not be strong for in situ production during the winter, it may be important when the reactions during entrainment of free tropospheric air masses into remote marine atmospheric boundary layer are considered (e.g., Clarke and Kapustin, 2002). Zhang et al. (2008) suggested that soot particles acquire a large mass fraction of sulfuric acid during atmospheric ageing. It is well known that semivolatile organic compounds are lost during atmospheric ageing due to evaporation following the dilution and oxidation of primary organics due to heterogeneous reactions (Donahue et al., 2006; Lambe et al., 2013). Such chemical ageing (the degree of oxygenation) of organics has previously been identified by increasing f44 and decreasing f43, which are the ratios between the mass-to-charge ratios at 44 and 43 and the total organics signal in the component mass spectrum, respectively (Ng et al., 2011). During this cruise, gradual increases in f44 and decreases in f43 were observed with distance from the coast. However, as described earlier, sulfate concentrations can increase due to the heterogeneous in situ oxidation of SO2 and condensation of sulfuric acid onto pre-existing particles after gas-phase oxidation of SO2 during long-range transport (Kompalli et al., 2020b, and references therein). These processes can contribute to the alteration of the mixing state of BC, which is a relatively longer-lived species, and determine its dominant coating material. Another interesting possibility is organosulfates acting as a coating material on BC, but examining it is beyond the scope of the present study.

The MMD values of rBC showed no particular pattern with changing NR-PM1.0 species mass concentrations. Further, the rBC mass concentrations showed a good association with organics (r∼0.78) and sulfate (r∼0.71) (not shown), corroborating the presence of multiple sources that co-emit particulate organics and sulfate along with BC aerosols. As described earlier, the mean fraction of rBC-containing particles remained similar (∼8 %) over different regions (SEAS, NIO-E, and EIO). However, frequent new particle formation events (Kompalli et al., 2020a) resulted in lower FBC (<2 %), which is reflected in the amounts of NR-PM1.0 mass that are bound to rBC particles. It was observed that new particle formation and subsequent growth to larger sizes preceded an enhancement in the mass concentrations of submicron aerosols (mostly organics and occasionally sulfate). This suggests another possibility of a reduction in the primary particulate pollution, and the available secondary condensate promotes nucleation in such cases. Also, the portion of the non-refractory submicron aerosol mass bound to the BC particles varied over a wide range of 10 %–40 % over different regions (Fig. S4 in the Supplement) and lower values were especially seen during new particle formation (NPF) periods (not shown here). Figure S5 in the Supplement shows the association between the sum of mass concentrations of organics, sulfate, and ammonium aerosols and bulk absolute coating thickness over different sub-regions covered during the ICARB-2018. This reveals a very good association (r∼0.90) over the remote EIO and a weak association over the NIO-E regions, whereas no association was found over the SEAS and NIO-W regions. It also advocates that variation in the submicron aerosol composition can explain alterations to rBC mixing state over the remote regions, whereas the rBC mixing state would be complex in the vicinity of the source regions. All these suggested the presence of distinct sources and atmospheric processes involving a range of condensing gaseous species contributing to the BC coating, altering its mixing state by enhancing the extent of the coating. However, it is not possible to precisely demarcate the sources and coating substances at the individual particle level with the present methodology using the SP2 and ACSM in tandem. It can be achieved through an instrument like a soot particle aerosol mass spectrometer (SP-AMS) (e.g., Gong et al., 2016; Liu et al., 2018). Alternatively, in situ morphological analysis techniques (like transmission electron microscope (TEM) equipped with an energy dispersive X-ray analyser) (Adachi et al., 2010, 2014; Ueda et al., 2016, 2018) are needed for the identification of particle-level morphology.

The present study focused on the characterization of refractory BC microphysical properties, including its mixing state information over the south-eastern Arabian Sea and the northern and equatorial Indian Ocean regions, which were influenced by distinct outflow air masses (that originated over the east coast and/or Bay of Bengal, the west coast/peninsular India and marine regions and traversed over the oceanic regions). The sensitivity of the optical, hygroscopic properties of BC to these observed characteristics and the estimation of regional radiative and climatic implications due to such highly chemically aged, thickly coated large BC particles form the scope of future study.