Journal Archive

Basic working principle of all the filter-based instruments is the same. Briefly, ambient air is drawn at a certain flow rate and Particulate Matter (PM) is collected on a filter. Simultaneously, light of certain wavelength is made to pass through the deposited PM spot and attenuation of light intensity is used to calculate the light attenuation coefficient (β_abs) in units of m^-1. In order to ensure that the observed change in filter transmittance is not due to change in the intensity of the LED light source, a second filter spot is used as a reference. The equation used to calculate β_abs is derived from Beer-Lambert law and is written as

where A is the area of the filter sample spot in m² and V is the volume of air drawn during the given time period in m³. I₀ and I are the average filter transmittances during the prior time period and the current time period, respectively. Light attenuation coefficient can be converted into mass concentration of EBC mass using following equation

where,

σ_BC= Mass Absorption Cross section (MAC) of EBC (m² g^-1)
M_BC=Black Carbon Mass Concentration (μg m^-3)

MAC is an experimentally determined calibration constant and has unit of area per unit mass. MAC value measures the ability of a material to absorb intensity of light at a particular wavelength per unit gram. Although units are in area, it does not refer to actual size and can be considered analogous to nuclear cross section (a term often used in nuclear physics) which is quantified physically in terms of characteristic area, where larger the area larger is the probability of nuclear interaction. MAC is an experimentally determined calibration constant and have been derived for BC throughout the literature (Nordmann et al., 2013; Kondo et al., 2009). Known mass of BC (measured using other technique, instrument) is taken and then corresponding light attenuation coefficient (β_abs) is measured. Then the ratio of attenuation coefficient and known mass is used as MAC value.

The above stated formula (Equation (1)) is based on the assumption that the absorption is independent of non-linearity arising due to filter loading and also does not take into account the scattering due to filter medium as well scattering due to accumulated particulate matter. Further two correction factors corresponding to two effects (i) loading and (ii) multiple scattering effect are needed to be incorporated in equation to make filter-based measurements more accurate (Arnott et al., 2005; Weingartner et al., 2003). Loading effect refers to shadowing of freshly collected particles by previously collected particles and as a result, the measured BC becomes underestimated as the loading of particles on filter increases (Song et al., 2013). Second effect is caused by multiple scattering of the light beam at the filter fibres in the unloaded filter leading to an enhancement of the optical path and thus to enhanced light absorption of the deposited particles and ultimately leading to overestimation.

Various empirical correction functions and algorithm have been derived to account for these artifacts arising due to use of filter (Virkkula et al., 2015, 2007; Park et al., 2010; Kirchstetter and Novakov, 2007; Bond and Bergstrom, 2006; Schmid et al., 2006; Arnott et al., 2005; Weingartner et al., 2003; Bodhaine, 1995). Equation (1) is further modified by adding two further coefficients called as correction factors. New equation used to calculate absorption coefficient taking above explained effects into account can be written as

where, C is correction factor corresponding to multiple scattering and R corresponds to shadowing effect. Value of these coefficients depends on correction method applied. Immersion of the particle loaded filters in oil with a refracting index similar to the filter fibres can minimise the multiple scattering effect (Ballach et al., 2001). However, this technique cannot be applied to online monitoring instruments. Recently introduced dual spot measurement technique provides a good alternative solution for loading effect (Drinovec et al., 2015).

Furthermore, MAC value of carbonaceous aerosol is calculated experimentally and is specific to wavelength of light intensity used. Its stability depends on coating of particles by volatile compounds and co-existence of light-scattering particles (Kondo et al., 2009), thus is also a source of uncertainty for optical instruments. Increase in relative humidity of ambient air can also lead to increase in MAC value. Wu et al. (2016) found the rate of increase in MAC value per 10% increase in RH to be 0.20-0.24 m² g^-1. In order to overcome all the limitations described above, a variety of optical absorption and filter-based real-time measuring instruments are available commercially each having their advantages as well as limitations. We have tried to explain features, parameters, advantages of each instrument in the Table 1. Detailed working of instruments can be understood from references given along in the Table 1.

As mentioned above, filter-based techniques suffer from filter loading effects and also do not measure optical absorption by aerosol particles directly, none of the above methods can be considered as qualified to be a reference method. In-situ measurement of absorbance is a major advantage of the two techniques/instruments mentioned in this section, which are (i) Photoacoustic Spectroscopy (PAS), and (ii) Black Carbon Photometer (BCP).

(i) Photoacoustic Spectroscopy

PAS has been used widely for trace gas analysis as well as to study light absorption by aerosol and study the behaviour of materials (Harren and Cristescu, 2006; Sigrist, 2003). It is considered as a possible reference method for measurement of optical properties of aerosol in visible region (Petzold et al., 2013; Lack et al., 2006; Arnott et al., 2003). Use of difference between extinction and scattering coefficient as absorbance coefficient is another near contender (Singh et al., 2016; Sheridan et al., 2005) but it requires use of two separate instruments which can lead to significant errors due to potentially different measurement systems and conditions (Wang et al., 2017b). Accuracy of PAS system has been reported to be 5-10% depending on laboratory or field locations and methods of calibration (Lack et al., 2006). Also, in-situ measurement makes PAS a comparatively better choice. Detailed working of PAS as well as its use for measurement of optical absorbance of light by aerosol is given in various research articles published previously (Wang et al., 2017b; Lack et al., 2009, 2006; Arnott et al., 2006, 2003, 1999).

The working procedure for measuring aerosol optical properties is explained here briefly. Aerosol is passed through a chamber where a modulated light beam of certain wavelength is made to pass through the incoming aerosol, thus modulated light is absorbed by the particles. Light energy absorbed by the particles is transferred as heat energy to the surroundings, creating a pressure variation thus creating pressure/acoustic waves. These waves are further resonated in a cavity at characteristic radial and longitudinal frequencies. When laser beam power is modulated at the acoustic resonance frequency, the generated waves are further amplified and detected using microphones. As PAS offers various advantages (in-situ measurement and better accuracy) over other conventional methods, various photoacoustic sensors and analyser are developed individually with variable parameters (Wang et al., 2017b; Krämer et al., 2002). But despite its high relevance to EBC measurement, instruments based on this technique has not been fully explored and commercialised.

(ii) Black Carbon Photometer

This instrument is latest addition to the list of commercially available instruments which gives BC mass by measuring optical absorbance of aerosol. This instrument uses the technique of long-path photometry to quantify mass concentration of EBC by measuring extinction coefficient. It measures extinction coefficient of aerosol in-situ at two wavelengths 405 nm and 880 nm. Out of which, 880 nm channel is used to measure BC mass concentration. In situ-measurement of absorption coefficient using Beer-Lambert law approach (Equation 1) requires long path lengths of detection cells (where absorbance due to reference air and sample (ambient) air is measured in regular intervals, i.e. Io and I in Equation 1). However large detection cells are difficult to be rapidly flushed which leads to measurement error due to inherited variation in lamp intensity. The patented tubular design (Birks et al., 2020) of detection cell in BCP enables detection cell to be rapidly flushed, as a result sample (air containing aerosol) and reference measurement (clean air) can be switched quickly. Therefore, light attenuation through aerosol as well as clean air can be measured very rapidly. BCP’s tubular design is very similar to (extension of) the tubular design described in details in a study explaining folded tubular design for measurement of NO₂ (Birks et al., 2018). This technique can prove to be a good option for in-situ measurements. However, to the best of our knowledge, there is no published data of EBC measurement or inter-comparison available presently using this instrument. Therefore, no concluding remarks can be made about field performance of this technique.

Aethalometer is the most widely used instrument for EBC measurement in both countries. 71 out of 80 studies (89%) considered in this review from both countries (33 out of 40 (82%) from China and 38 out of 40 (96%) from India) have made use of different models of Aethalometer (AE-16, AE-21, AE-31, AE-42, AE-51, AE-33) Different models of Aethalometer have different features. AE-16 is the classic single wavelength (880 nm) Aethalometer which was introduced in 1997. In 1999, upgraded versions of AE-16, dual wavelength aethalometer AE-21 (370 nm and 880 nm) and seven wavelength Aethalometer AE-31 were introduced to provide options to study spectral dependency of optical properties of aerosol. AE-42 and AE-51 are the portable Aethalometers which were designed to provide ease of operation at remote location and on-board vehicle measurement. AE-42 uses the same aerosol collection, analysis and data systems as the ‘full-size’ standard models (AE-16, AE-21, AE-31) but is constructed in a smaller chassis and is equipped with an internal battery (Hansen and Schnell, 2005). AE-51 is pocket-size, light weight (230 g) and has a better resolution (1 ng m^-3) than other models (micro-Aeth, 2016). AE-33 is the latest versions which can measure absorbance at 7 wavelengths and offers improved measurement by making use of a patented technology, i.e. Dual-Spot Measurement (Drinovec et al., 2015). Since AE-31 is most commonly used model of Aethalometer and uncertainties evaluation of this model has been done in few studies, detail about AE-31 and its parameters and uncertainties is given in Table 1 and detailed working procedure can be found in references given along.

A recent review on the basis of 142 studies conducted in India from 2002 to 2018 also found that 96% of the studies used the various version of Aethalometer (Rana et al., 2019). This might be due to provision of multi wavelength channels as a result the ratio of absorbance of light at 370 nm and 880 nm can be used for source apportionment. For an example, absorbance at 370 nm is largely due to organic compounds produced by biomass burning. Therefore, the ratio of absorbance of light at 370 nm and 880 nm can be useful to find the contribution of biomass/wood burning in comparison to fossil fuels. Further details about “Aethalometer Model” used for source apportionment can be found in some of the studies included in this review (Dumka et al., 2018; Bisht et al., 2016). Another reason might be provision of measurement at 880 nm which is preferred wavelength for measurement of EBC. (Qin et al., 2019; Gupta et al., 2017; Srivastav et al., 2014; Tiwari et al., 2009). BC has been found to be primary absorber of light at 880 nm and absorbance due to other light absorbing aerosol is less significant at this wavelength (Bodhaine, 1995).

However, frequent use of Aethalometer is quite surprising considering the advantages of other filter-based instrument like COSMOS (provides stabilised MAC value due to removal of non-refractory light absorbing aerosol by using a heated inlet), Multi-Angle Absorption Photometer (MAAP) (provides correction for scattering due to aerosol and filter media) and Multi Spectrum Carbon Photometer (MSCP) (provides measurement at 10 wavelengths in comparison to 8 wavelength channels in Aethalometer). MAAP has been used in three studies (Zhang et al., 2020a; Xu et al., 2014; Hyvarinen et al., 2010) whereas COSMOS and MSCP have been used in only 1 study each (Xue et al., 2018; Wang et al., 2011). More surprisingly, in-situ measurement technique, i.e. of PAS has still not been used to report BC measurement. Some studies have used PAS technique as a reference method to check the performance of instrument used for measurement. Three unconventional instruments were used to report EBC concentrations. These are Spectrometer (Chen et al., 2016; Schleicher et al., 2013), Smoke Stain Reflectometer (SSR) (Sandeep et al., 2013) and Laser Integrating Plate Method (LIPM) (Genchen et al., 2005). All these instruments involve a common process of collecting particulate matter mass on a filter using a sampler and then using these samples to measure attenuation of light from different spots of filter using an instrument specific wavelength. These instruments give concentration of EBC in offline mode and hence are not much popular.

Wavelength used is the key parameter of any measurement which is based on measuring optical response (absorption, scattering) of species. Although optical response of EBC is almost similar throughout the visible spectrum but response of other light absorbing aerosol (especially Brown Carbon) has been found to be highly wavelength dependent (Liu et al., 2018; Shamjad et al., 2016). Measuring BC mass at lower wavelengths (towards UV spectrum) can lead to overestimation due to absorption by other light absorbing and scattering aerosol. Therefore, selection of wavelength used for measurement of EBC mass plays an important role to remove interferences due to presence of other light absorbing and scattering aerosol. Overall, 69 out of 80 studies (86%) from both countries (87% from China and 85% from India) have mentioned the wavelength used for measurement. Most of the studies used attenuation at 880 nm wavelength to measure BC concentrations which is obvious given that Aethalometer is the most widely used instrument whose 880 nm channel is considered as standard channel for reporting BC concentrations. Other wavelengths at which measurements have been done are 370 nm (Bisht et al., 2016), 565 nm (Wang et al., 2011), 650 nm (Chen et al., 2016; Schleicher et al., 2013), and 670 nm (Hyvarinen et al., 2010).

Simultaneous measurements of BC concentrations at 370 nm and 880 nm were done in Delhi (Bisht et al., 2016) and BC mean concentrations reported from 880 nm wavelength channel were 22% lower than that of 370 nm wavelength channel which is quite significant suggesting obvious bias among optics-based instrument using different wavelengths. One study (Xue et al., 2018) measured EBC concentration for two months at three wavelength bands (i) near Ultra-Violet (ii) Visible and (iii) near Infra-Red. Mean EBC concentration at near-IR band was found to be 2.6 μg m^-3 (ranging from 0.2 to 17.7 μg m^-3) which is only 56% of the mean EBC concentration of 4.6 μg m^-3 (ranging from 0.2 to 30 μg m^-3) measured at near-UV region. As mentioned above, high absorbance at lower wavelengths (370 nm and near-UV region) is due to presence of other light absorbing aerosol which is significant at lower wavelength as compared to absorbance due to EBC. Thus, use of different wavelengths in different instruments may cause discrepancy in results of different instruments. So, standardisation of wavelength used for measurement and reporting of EBC concentrations is necessary. The 880 nm wavelength can be considered a possible candidate for standard measurement as BC has been found to be primary absorber at this wavelength. However, MAC value corresponding to wavelength 880 nm shows large variation showing influence of other light absorbing aerosol at 880 nm wavelength also.

As per guidelines suggested by Petzold et al. (2013) for reporting BC measurements, MAC value corresponding to wavelength used should be mentioned when reporting EBC concentrations. Only 42 studies out of 80 (~52%) from both countries (52% from China and 52% from India) have mentioned the corresponding MAC value used for the conversion of attenuation coefficient to BC mass. About 93% of these MAC values used were manufacturer recommended. One study (Yan et al., 2008) conducted in rural area of China used MAC value which was derived by a linear regression of the BC concentrations at 880 nm from the Aethalometer against the light absorption coefficients at 532 nm simultaneously measured using PAS technique. The calculated value of MAC at 880 nm was 8.28 m² g^-1 which is almost half of the manufacturer recommended value of 16.6 m² g^-1 for 880 nm.

Similarly, other studies carried out around the world also found MAC value to be significantly different from that suggested by manufacturer for AE-31 corresponding to wavelength of 880 nm (Arnott et al., 2003; Bergin et al., 2001; Moosemuller et al., 1998). MAC value corresponding to 880 nm wavelength was found to be varying from 10 to 19 m² g^-1 (Lou et al., 2007). Manufacturer recommended value of MAC corresponding to 880 nm for new model of Aethalometer (AE-33) have been changed to 7.7 m² g^-1 (Kolhe et al., 2018; Liu et al., 2017) from previous value of 16.6 m² g^-1 used in its predecessors. These discrepancies in MAC value are due to different sources of EBC, aging and coating of primary particles over the time due to mixing states (Bond et al., 2013; Kondo et al., 2009; Lou et al., 2007).

MAC value of uncoated particles has been found to be highly stable having a value of 7.5±1.2 m² g^-1 at 550 nm (Bond and Bergstrom, 2006). MAC value of 10 m² g^-1 suggested for COSMOS remains stable due to use of heating inlet which removes coating of the particle (Kondo et al., 2009). As coating of particles is removed, MAC value becomes significantly independent of source and mixing states of the particles due to coating (Kondo et al., 2009). Stabilized MAC value provides a greater advantage to COSMOS over other filter-based instruments.

To reduce underestimation due to loading/shadowing effects various correction methods have been implemented in some of the studies which are included here. Virkkula (Virkkula et al., 2015, 2007) and Weingartnar (Weingartnar et al., 2003) method are the most common methods used to correct shadowing effect for Aethalometer. In these methods, empirically derived correction factor (R in Equation (3)) is used which is the function of instantaneous and threshold light attenuation through filter media. Recently introduced dual spot measurement technique (Drinovec et al., 2015) provides a good alternative to these conventionally used correction methods. Correction methods used to account for shadowing effect in COSMOS and MAAP are not mentioned in this review but they are explained in details in some other studies conducted worldwide (Miyazaki et al., 2008; Petzold and Schlonirrer, 2004).

Correction for multiple scattering effect have been done in very few studies (Begam et al., 2016; Ran et al., 2016; Dumka et al., 2010). Schmid method (Schmid et al., 2006) and Arnott method (Arnott et al., 2005) have been used in these studies to treat multiple scattering effects of filter media. A wavelength dependent correction factor (C in Equation (3)) derived by modelling analysis of basic scattering theory is used to account for multiple scattering in filter media. This correction factor can be further modified to account for scattering due to aerosol by adding another term which is a function of aerosol single scattering albedo.

Using MAAP to measure EBC concentration do not require corrections for scattering as it measures transmittance and reflectance at multiple angles and uses two-stream approximation in the radiative transfer budget. Inter-comparison of data with MAAP can be used to correct for the multiple scattering effect of the filter and scattering due to aerosols. This method has been implemented in two studies included in this review (Xue et al., 2018; Ran et al., 2016).

Study carried out in Beijing (Lou et al., 2007) analysed the contribution of EBC mass in TSP, PM₁₀ and PM_2.5. EBC mass detected using PM₁₀ inlet was found to be 90% of EBC mass detected simultaneously without using any inlet (i.e. TSP). Similarly, EBC mass detected using PM_2.5 inlet was found to be 82.6% of EBC mass detected simultaneously without using any inlet (i.e. TSP). Other studies conducted around world (Gatari and Boman, 2003; Moloi et al., 2002; Hitzenberger and Tohno, 2001) also found that most of the EBC is present in fine mode (PM_2.5) with a little mass existing in coarse mode (above PM_2.5). BC particles from combustion sources are always emitted in fine mode. However, internal mixing of BC with hygroscopic materials (sulphates and nitrates) can lead to particle growth at high humidity levels (Hitzenberger and Tohno, 2001).

Furthermore, EBC from non-exhaust emission and resuspension (Lugon et al., 2020) can also be present in coarse mode. Thus, using different source (TSP, PM₁₀, PM_2.5, PM₁) for EBC measurement can give slightly different results and thus should be mentioned while reporting EBC concentrations. 51 studies out of 80 (64%) from both countries (58% from India and 70% from China) have mentioned the inlet cutoff size. Out of these, 51% used PM_2.5 size segregation device (cyclone or impactor) and 25% percent measured EBC concentration in TSP. PM₁₀ and PM₁ inlets are used in 12% of studies reviewed here. Furthermore, cutoff points for different segregation device used can get altered due to flow rate of sampling. Two studies (Reddy et al., 2012; Lou et al., 2007) have clearly mentioned the changed cutoff size of cyclones used due to use of fixed flow rate of instrument of 3 lpm (Litre Per Minute).

Use of simplified term “Black Carbon” has become a common practice to describe various type of carbonaceous particles. On the basis of its chemical and optical properties, measurement technique used, the more general term Black Carbon has been dissected into three specific terms as Elemental Carbon (EC), Equivalent Black Carbon (EBC) and Refractory Black Carbon (rBC) (Lack et al., 2014; Petzold et al., 2013). As mentioned earlier, it is recommended to use term EBC for BC mass measuring optical absorbance of aerosol. Only 4 out of 47 studies published after 2013 (Kumar et al., 2020; Wang et al., 2020; Zhang et al., 2017; Liu et al., 2016) have made use of this terminology (EBC) to report measurement. Out of these, 3 studies were from China whereas there is only one study from India which has used EBC term instead of BC.

Uncertainty of a measurement is an important aspect which is needed to be mentioned in order to validate the measurement. Only a few studies (~30%) from both the countries have reported uncertainty quoting/citing the previous research or inter-comparisons performed as well as manufacturer specifications. Undefined contribution from above mentioned sources of uncertainty (shadowing, multiple scattering effects, and coating of EBC particles contributing to uncertainty in β_abs and MAC value, ultimately leading to uncertainty in M_BC) makes it difficult to evaluate uncertainty of EBC measuring instruments in a rigorous way. Several studies carried around the world have tried to in-corporate errors occurring due to these sources (Backman et al., 2017; Singh et al., 2010; Schmid et al., 2006; Arnott et al. 2005; Weingartener et al., 2003). As mentioned earlier, various correction methods are suggested to reduce these errors but magnitude of errors and the uncertainty arising from these effects are still unknown and are site and source specific. Apart from these sources, various other sources of uncertainty which can be defined and quantified are flow rate, area of the filter spot and drift in intensity of light source used.

Instrument specific uncertainty/accuracy based on the studies reviewed here and references therein are included in Table 1. Petzold and Schonlinner (2004) determined uncertainty in estimation of β_abs (12%) for MAAP by performing error analysis of scattering and absorbance due to particle loaded and particle free filter matrices. They also estimated uncertainty in M_BC calculation by MAAP to be 25% based on comparison with thermal technique (German reference method, Schmid et al., 2001). COSMOS accuracy (in M_BC) was determined to be 10% on the basis of results of COSMOS vs SP2 comparison summary given in table 2 of Ohata et al. (2019). However, later on Ohata et al. (2021) estimated the absolute accuracy of COSMOS to be 15% taking into account the 10% uncertainty of reference instrument (SP2) as well. Total root mean squared systematic error in absorption coefficient (β_abs) calculation due to quantifiable sources (flow rate, drift in reference and absorbance intensity, filter spot area) and correction factor has been shown to be 12% for COSMOS (Miyazaki et al., 2008). Using comparison with thermal method, Corrigan et al. (2006) reported uncertainty for Aethalometer (AE-31) to be ranging from 5-40%. Backman et al. (2017) evaluated relative uncertainty of Aethalometer (AE-31) to be 36% using equation of uncertainty propagation approach (Sharma et al., 2017a). However, manufacturer specified uncertainty of Aethalometer is mentioned to be 5% and 10% in few studies reviewed in current study (Zhou et al., 2018; Ji et al., 2017; Jose et al., 2016; Tiwari et al., 2013).

As mentioned in discussion above, some of the references given in included studies were based on inter-comparison of instruments (Ohata et al., 2019; Kondo et al., 2009; Corrigen et al., 2006; Arnott et al., 2005) and some were based on using error analysis and equation of uncertainty propagation (Backman et al., 2017; Miyazaki et al., 2008; Petzold and Schonlinner, 2004). However, instruments/techniques used as reference for inter-comparison are different for different studies. Inter-comparison technique can be much more effective for defining accuracy once a standard reference method for measurement of EBC is declared by the regulatory bodies. Furthermore, there is a considerable negligence in reporting uncertainty of the instruments as only very few studies have specified uncertainty of the instrument and most of them were based on manufacturer recommended value. Error analysis and equation of propagation of uncertainty approach can be applied whenever possible to further strengthen the reporting of EBC concentrations.