On the Use of Brown Carbon Spectra as a Tool to Understand Their Broader Composition and Characteristics: A Case Study from Crop-residue Burning Samples.

This study proposes a novel approach to the use of brown carbon (BrC) absorption spectra as a tool to understand their broader composition and characteristics. The ratios of absorption coefficient (b abs) spectra over a wavelength range (310-600 nm) for water-soluble and methanol-soluble BrC were used to quantify the relative contribution of water-soluble and water-insoluble chromophores to total BrC. The same ratios for the samples collected during the day versus night were used to assess the diurnal variability in BrC composition and concentrations. Ratios of b abs at different wavelengths with respect to that at 365 nm were used to understand whether BrC is predominantly composed of one type of chromophore, that is, humic-like substances, or different chromophores (e.g., nitroaromatic compounds) with the understanding that different chromophores absorb predominantly at different wavelengths. As a case study, day/night pairs of PM2.5 samples collected from Patiala (30.33°N, 76.4°E) during paddy residue burning were used, and results are discussed. A majority of BrC from paddy residue burning were found to be water-insoluble, and the fraction of water-soluble BrC to total BrC showed a decreasing trend with increasing wavelength. During the burning period, night-time water-soluble nitrogenous organic species were found to be more absorbing than daytime water-soluble nitrogenous species. The proposed method will be very useful for BrC studies over the globe.

Introduction

Carbonaceous aerosol consists of organic matter (OM) and black carbon (BC). BC is known to exert a strong warming effect on global and regional climate.[1] Some assessments suggest that the positive radiative forcing by BC is second only to carbon dioxide.[2] Recent studies found that certain types of organic carbon (OC), termed “brown carbon (BrC)”, also absorb radiation efficiently in near-ultraviolet (UV) and visible regions and contribute ∼35% to direct radiative forcing by carbonaceous aerosol.[3] BrC is a light-absorbing part of OC and characterized by an absorption spectrum that smoothly increases from the visible to UV wavelengths.[4−7] Sources of BrC in the atmosphere are a variety of primary emissions and secondary formation processes.[7,8] The primary sources of BrC are incomplete combustion of fossil fuels (related to traffic), industrial processes, domestic heating, and biomass burning (BB).[7] Further, several laboratory chamber experiments and ambient measurements showed the formation of secondary BrC chromophores through various mechanisms such as photo-oxidation of aromatic volatile organic compounds (VOCs), ozonolysis of terpenes, and subsequent aging in the presence of ammonium ions and humidity.[9−14] Emissions from BB are shown to be a large source of primary BrC as well as precursors of secondary BrC on regional and global scales.[15−19] Atmospheric humic-like substances (HULIS), major primary BrC, are an important fraction in organic aerosol.[20]

A BrC consists of a variety of organic compounds with different absorbing characteristics. It is very difficult to identify individual BrC species and their optical properties, and therefore, they are often presented as the absorption coefficient at a certain wavelength. Furthermore, optical properties of BrC are reported to change with various atmospheric processes such as oxidation, solar irradiation, and changes in temperature and relative humidity.[13,21,22] These factors result in highly variable composition and concentration of BrC in time and space, which in turn results in substantial uncertainties in predicting their climate effects.[23]

This paper mainly focuses on the use of BrC spectra of water-soluble and methanol-soluble OC as a tool to understand their broader composition and optical characteristics as well as to assess the relative contribution of water-soluble and water-insoluble BrC to total BrC.

Materials and Methods

As a case study for the proposed approach, particulate matter smaller than 2.5 μm aerodynamic diameter (PM2.5) samples were used. The sampling site was located on the terrace of the Department of Physics, Punjabi University, Patiala (30.33°N, 76.4°E, 250 m above mean sea level), a semi-urban city surrounded by agricultural area within a few kilometers. Patiala is situated in the northwestern part of the Indo-Gangetic Plain. Day/night (10 h integration time) pairs of ambient PM2.5 samples (n = 69) were collected every day before (n = 21), during (n = 36), and after (n = 8) a large-scale paddy residue burning from October to November 2014. The samples were collected on precombusted (at 450 °C for 10–12 h) Tissuquartz filters (PALLFLEX, 25 × 20 cm2) using a high-volume air sampler (Thermo-Anderson, HVS) with the flow rate of ∼1.13 m3 min–1. Soon after their collection, the filters were wrapped in aluminum foils, sealed in plastic ziplock bags, and stored in a deep freezer (−19 °C) until the time of analysis. Further, a wide variety of chemical species were analyzed in these filters. They included water-soluble OC (WSOC) and water-soluble organic nitrogen (WSON = water-soluble total nitrogen–inorganic nitrogen (IN), where IN is the sum of NH4+–N and NO3––N), and were analyzed using a TOC–TN analyzer (Shimadzu, model-TOC–LCPH with AS–L autosampler). Elemental carbon (EC) and OC were analyzed with the EC–OC analyzer (Sunset Inc), and inorganic cations (NH4+, Na+, K+, Mg2+, and Ca2+) and anions (Cl–, NO3–, and SO42–) were analyzed using ion chromatography (Dionex, ICS-5000). A detailed description of analytical procedures is given elsewhere.[24−26] Gas-phase markers such as carbon monoxide (CO) and nitrogen oxides (NO = NO + NO2) were measured using gas analyzers (Ecotech Serinus) with 2 min integration time.

For BrC spectra, a 3 cm2 punch of sample filters was extracted with water, and another 1.5 cm2 punch with methanol using ultrasonication. The extract was introduced into the liquid waveguide capillary cell (World Precision Instrument, Sarasota, FL, 2 m path length) coupled to a portable UV–visible spectrophotometer (model USB-4000) and deuterium and tungsten halogen lamps (DT-Mini-2, Ocean Optics) via a syringe filter (Acrodisc Syringe filters with GHP Membrane, 0.45 μm, 25 mm) to remove insoluble particles (including BC and dust), and the high-resolution absorption spectra were saved from 300 to 700 nm for each filter extract using a portable UV–visible spectrophotometer. A spectrum of water extracts was referenced with water and that of methanol extracts with methanol to remove any possible effect of absorbance from solvent on the observed spectrum of samples. This also enables one to compare the spectra of different solvents. Several blank filters were also treated like samples, and reported values are corrected for blanks. The analysis of about 20% samples was also repeated for absorption spectra and found to be within 8%. The detailed analytical protocol for the assessment of optical properties of BrC was described elsewhere.[27,28] The absorption coefficient at a given wavelength (babs_λ) was estimated as follows:Here, Aλ is the absorbance at wavelength λ, A700 is an absorbance at the reference wavelength (700 nm) to account for any baseline drift, Vext = volume of water (or methanol) used for the extraction of the aerosol filter punch, Va = volume of air filtered during sampling, and l = path length (2 m).

Results

The PM2.5 concentration ranged from ∼90 to 500 μg m–3 during the study period with the average values of 154 ± 57, 271 ± 122, and 156 ± 18 μg m–3 during pre-burning (T1: 11th–22nd October, n = 25), burning (T2: 23rd October to 10th November, n = 36), and post-burning (T3: 12th–15th November, n = 8) periods, respectively, reflecting the effect of BB on ambient PM2.5 concentrations. For the total daytime PM2.5 samples, the contributions of OM (2.1 times the measured concentration of OC, Rajput and Sarin, 2014)[29] were 36, 46, and 46%; water-soluble inorganic species (WSIS = sum of NH4+ + Na+ + K+ + Mg2+ + Ca2+ + Cl– + NO3– + SO42–) were 14, 41, and 15%; and EC were 5, 4, and 5%, during T1, T2, and T3, respectively. Similarly, for nighttime samples, the contributions of OM were 46, 63, and 46%; WSIS were 11, 13, and 18%; and EC were 5, 3, and 5% during T1, T2, and T3, respectively. The observation suggests that OM was dominating in all the three periods, and nighttime OM concentrations were always higher than those observed during the daytime. Similar observations were reported by Rastogi et al. (2014)[26] from the same sampling site. Highest OM concentrations (OM ≈ 63%) were observed in nighttime samples collected during T2 period, and they were slightly higher compared to the previously reported value (OM ≈ 56%).[26]

Proposed Concept of Using BrC Spectra to Assess Their Composition and Characteristics

BrC species may consist of a variety of organic chromophores with variable concentrations and different absorbing characteristics. The absorption coefficient (babs) spectra of water extracts or methanol extracts of aerosol represent the bulk optical property of water-soluble or methanol-soluble (assumed to be total) BrC, respectively, as a function of wavelength. If the BrC composition is uniform from sample to sample, then the ratio of babs spectra for different samples shall be uniform (constant) as a function of wavelength. The magnitude of this constant would suggest the relative abundance of BrC chromophores in different samples. However, if the BrC composition is not uniform from sample to sample, then one would expect variable ratio at different wavelengths depending on the change in composition and concentrations of different chromophores. It is known that different types of chromophores absorb strongly at different wavelengths, for example, HULIS-type BrC (usually primary BrC) absorbs strongly at around 365 nm, whereas nitroaromatic-type BrC (usually secondary BrC) absorbs strongly at a relatively higher wavelength (>400 nm).[12,20,23,30−34] The wavelengths at which non-uniformity in babs spectra ratio is observed would indicate the change in the type of BrC in a given sample. It is relevant to state here that the proposed method is not the replacement of doing molecular speciation of BrC, but it shall be very useful in understanding the possible dominant BrC species present in a given sample.

Further, the same concept can be applied to water extracts and methanol extracts of the same sample. One can calculate the ratio of babs at different wavelengths for water extract to that for methanol extract and use it to quantify the relative contribution of water-soluble and water-insoluble chromophores to total BrC as a function of wavelength in a given sample presuming that methanol-soluble BrC represents total BrC, which is a valid assumption as per literature.[31,35,36] These concepts are used in the subsequent sections on the samples collected from Patiala as a case study.

BrC Composition and Characteristics through babs Spectra

Figure shows a comparison of the babs spectra for BrC extracted with water (babs_Water) and methanol (babs_Methanol). They exhibit very different light absorption properties during the T2 period compared to those observed during the T1 and T3 periods. The variability in the night/day (N/D) ratio of babs_Water spectra during the T2 period was large compared to those observed during the T1 and T3 periods (Figure a–c). The observation during T2 suggests that BB emits a variety of BrC chromophores with the highest concentrations of chromophores absorbing in 310–400 nm range. Magnitudes of N/D ratio were always >1, with highest values during T2 (1.8–3) followed by T3 (1.3–1.5) and T1 (1–1.3) over 310–600 nm range, reflecting the changes in composition and concentrations of BrC chromophores which absorb at different wavelengths. Here, the absolute magnitude reflects the N/D change in chromophore concentrations, and the variability in magnitude indicates the N/D change in composition, as conceptualized in Section . During T1 and T3, a slightly lower N/D ratio with mild but noticeable variability in 450–600 nm range is suggestive of the presence of daytime secondary BrC in water extracts. The babs_Water showed a sharp increase in absorption at 310–400 nm range during the nights of T2 period, which is likely due to HULIS from BB.[15,28,30] In the same set of samples, a low absorption at >500 nm was also noticed, which was not observed during the T1 and T3 periods. Similar spectra were observed by Xie et al.[33] for the laboratory-generated secondary BrC formed by the reaction between benzene and m-cresol under high NO conditions.[33] Further, a spectral enhancement was observed in 410–490 nm range of babs_Water during the T2 period (as reflected in N/D ratio, Figure b). This observation suggests that the observed spectral band was likely associated with the presence of a variety of absorbing nitrogenous organic species, which were likely generated through secondary organic aerosol (SOA) in the presence of NO.[33] High NO was also observed in this study during the T2 period (figure not shown). This inference is attested by a significant correlation (r2 = 0.55) between WSON/WSOC ratio and absorption coefficient measured at 450 nm for water extracts (babs_450_Water) in the samples collected during nighttime of T2 period (Figure b). Higher WSON/WSOC is an indicative of a larger fraction of water-soluble nitrogenous organic species in the total water-soluble organic species. Further, absorbing properties of nitrogenous organic species in samples collected during daytime and nighttime were likely different, as for similar WSON/WSOC ratios, the babs_450_Water values were very different (Figure a,b). Daytime WSON/WSOC ratios were also poorly correlated with babs_450_Water (Figure a). Satish et al.[22] also documented that absorbing nitrogenous species are volatile and/or photosensitive, and therefore, their abundance reduces during daytime.

Figure 1

babs spectra during day and night and their N/D ratio for water-soluble BrC during (a) pre-burning (T1), (b) during burning (T2), and (c) during post-burning (T3) periods, and for methanol-soluble BrC during (d) pre-burning (T1), (e) during burning (T2), and (f) during post-burning (T3) periods. Note that y-axis range is different for the figures (a–c) and (d–f).

Figure 2

Linear relationship between WSON to WSOC ratio and absorption coefficient at 450 nm for water extracts (babs_450_Water) in (a) daytime samples and (b) nighttime samples. Higher nighttime WSON/WSOC ratios were associated with higher babs_450_Water, suggesting that nighttime nitrogenous compounds were more absorbing.

A recent study by Cheng et al.[36] suggested that majority (∼85%) of organic compounds in the ambient PM could be methanol-soluble. In the present study, the average WSOC/OC ratio is about 0.60, that is, ∼40% of OC is water-insoluble. Figure d–f shows the babs spectra and their N/D ratio for methanol extracts during T1, T2, and T3, respectively. The babs_Methanol values were much higher than the babs_Water values, suggesting that water-insoluble BrC species are relatively more absorbing during all the three periods. During T2, both babs_Water and babs_Methanol show higher values in shorter-wavelength range (310–400 nm); however, babs_Methanol showed the higher value in the visible region too (Figure e). This observation suggests that babs_Methanol contains a significant fraction of water-insoluble chromophores which can absorb radiation in the visible region. This can be ascribed to differences in solubility of the chromophores in water and methanol.[16,37,38] Some studies suggest that the molecules comprising more aromatic rings (i.e., a higher degree of conjugation) have higher absorption that extends to longer wavelengths.[34] Further, Zhang et al. (2013)[37] had documented BrC chemical speciation, which showed that larger molecular weight polycyclic aromatic hydrocarbons (PAHs) absorbed more toward the visible range and have lower solubility in water. Sun et al.[35] suggested two kinds of OC: a water-soluble organic mixture with some UV–visible absorption and a water-insoluble OC with higher absorption.[16,30] Further, quinoid compounds are strong candidates for such absorption.[35] This type of absorption spectra was present in the nights of the T2 period; however, it disappeared in day samples (Figure b,e). This could be due to various reasons such as the evaporation of higher-volatility compounds (e.g., nitrophenols), photobleaching, and so forth. A similar observation is reported in Lin et al.,[40] which suggests that the decrease in absorbance does not follow a single exponential decay because different chromophores decompose at different rates.[39] It is important to state here that these statements are postulations from the measured absorption spectra of water-soluble and methanol-soluble extracts and reported maximum absorption by different chromophores at different wavelengths in literature. To make firm conclusions, more work such as comparison of the actually measured molecular speciation with absorbance spectra is needed.

In general, the absorption in the visible range is more important to the energy balance than the absorption in the UV range as nearly 40% of solar energy is found at wavelengths between 400 and 600 nm. UV absorption affects photolysis, but wavelengths below 400 nm provide only about 4% to solar energy.[35]

Relative Contribution of Water-Soluble and Water-Insoluble Fraction to Total BrC

To investigate the contribution of water-soluble and water-insoluble fraction to total BrC during different periods, the ratio of babs (water/methanol) was calculated (Figure ). The babs_Methanol represents total BrC, babs_Water denotes water-soluble BrC, and their difference reflects water-insoluble BrC. This approach helps in assessing the fraction of water-soluble BrC in the total BrC. It was observed that on average (integrated over 310–600 nm), the water-soluble BrC contributes ∼33, 27, and 26% in daytime samples, whereas 26, 19, and 23% in nighttime samples to total BrC in the samples collected during T1, T2, and T3 periods, respectively. Observations clearly suggests that contributions of water-soluble BrC were more in daytime samples compared to that in nighttime samples, and its fraction was minimum during T3, maximum during T1, and in-between during T2. These observations also suggest that water-insoluble BrC dominates total BrC over the study regions during the study period. However, this water-soluble fraction of BrC was not uniform at all wavelengths. It decreases with increasing wavelength and goes to as low as ∼10% at 600 nm.

Figure 3

babs (water/methanol) ratio versus wavelength and % of water-insoluble fraction versus wavelength during (a) pre-burning (T1), (b) during burning (T2), and (c) during post-burning (T3) periods. Note that shaded area is overlapped, that is, daytime W/M (blue) was always higher than that during nighttime (red); however, the absolute value of ratio varied as a function of wavelength. Green and black lines exhibit % water-insoluble BrC fraction as a function of wavelength during day and night, respectively.

Further, % water-insoluble BrC in both day-and-night samples was also estimated using the eq .

The babs (water/methanol) ratios for shorter (310–400 nm) and longer-wavelength region (400–600 nm) were averaged. At the shorter-wavelength region, water-insoluble fractions during the day (59, 67, and 67%) and night (67, 72, and 70%) samples were significant in T1, T2, and T3 periods, respectively. Similarly, at the longer-wavelength region, water-insoluble fractions during the day (72, 77, and 79%) and night (79, 86, and 81%) were dominant in T1, T2, and T3 periods, respectively. During T2 period, night samples influenced by BB emissions showed highest fraction (∼86%) of water-insoluble BrC and day/night differences at the shorter-wavelength region were less than those observed at the longer-wavelength region (Figure b). Lee et al.[11] reported that chemical composition and optical properties of BrC cannot be viewed as static, as they may change with time. Light-absorbing compounds (responsible for the color of BrC) can be potentially photobleached in sunlight and lose their ability to absorb visible radiation.[11] Our observations suggest that BrC absorbing at higher wavelengths is predominantly water-insoluble, and they are affected by daytime atmospheric processes such as photolysis/oxidation of BrC chromophores and other transformation reactions.[13,40,41]

Temporal Variability in BrC Composition through Absorption Spectra

In order to understand the day-to-day variability in BrC composition through absorption spectra, the absorptions at 400, 405, 420, 450, 500, and 550 nm wavelength (representing absorption by nitroaromatics)[23,37,39,41] were normalized to the absorption at 365 nm (representing absorption by HULIS compounds).[15,28,30] WSON comprised different nitrogen-containing WSOC generated through SOA and NO.[13,23] Linear regression analysis between WSON and NO showed good correlation (r2 = 0.64, figure not shown) during the T2 period, suggesting the possibility of the presence of water-soluble nitrogen-bearing BrC compounds. Recent laboratory studies have documented that anthropogenic VOCs (benzene, toluene, and phenols) and PAHs react with nitrogen oxides and produce light-absorbing nitroaromatics.[12,23,32,33,38,42] These classes of nitroaromatics absorb in the visible region such as 400, 405, 420, 450, 500, and 550 nm.[33,40,42]

Figure shows the temporal variability in babs_400, babs_405, babs_420, babs_450, babs_500, and babs_550 normalized to babs_365 for both water-soluble and methanol-soluble BrC. The observed ratios were higher at shorter wavelengths and decreased toward the visible region. It is interesting to note that methanol-soluble BrC showed relatively less variability during all the three periods. However, water-soluble samples showed higher day–night variability during the T2 period and lesser variability in T1 and T3 periods. This further suggests that BB emits a noticeable fraction of water-soluble chromophore, which absorbs at the visible region. At 405 and 420 nm, the variability was very high during the T2 period. It could be the presence of nitroaromatics BrC chromophores. At 405 and 420 nm during the T2 period, enhanced absorbance (>20%) was observed in night samples compared to day samples. Nitrophenolic compounds such as 2,4-dinitrophenol, 3-nitrophenol, and 2,5-dinitrophenol compounds absorb at 405, 420 nm, and 450 nm, respectively.[23] These compounds could be an important fraction of BrC in the urban atmosphere. Further, the overall normalized ratios for babs_Methanol were higher compared to those for babs_Water, further suggesting that there was always a significant insoluble fraction of BrC. However, this insoluble fraction was not uniform and showed maximum contribution during the burning period.

Figure 4

Temporal variability in babs_400, babs_405, babs_420, babs_450, babs_500, and babs_550 normalized to babs_365 for both (a) water-soluble and (b) methanol-soluble BrC.

Conclusions

A novel approach has been proposed to use only babs spectra of water-soluble and methanol-soluble OC as a tool to understand the broader composition and characteristics of water-soluble and water-insoluble BrC, which otherwise need copious amount of work and expensive facilities. As a case study, babs spectra of PM2.5 samples collected before (T1), during (T2), and after (T3) a large-scale paddy residue burning over Patiala were used. Magnitudes of N/D babs spectra ratios were always >1, with highest values during T2 (1.8–3) followed by T3 (1.3–1.5) and T1 (1–1.3) for 310–600 nm wavelength range, reflecting the changes in composition and concentrations of BrC chromophores. BrC abundances and composition were found to be very different during T2 nights, with significant contributions of chromophores absorbing in 310–400 nm range. The contribution of water-soluble BrC to total BrC decreases with wavelength, and water-insoluble BrC dominates total BrC composition during the whole study period. On average (integrated over 310–600 nm), the water-soluble BrC contributes ∼33, 27, and 26% in daytime samples, whereas 26, 19, and 23% in nighttime samples to total BrC during T1, T2, and T3 periods, respectively. Further, temporal variability in babs at different wavelengths normalized to that at 365 nm in day and night samples suggests that BB likely emits nitroaromatics, which are photosensitive and/or volatile. The proposed approach shall be useful in BrC studies over different regions.