Atmospheric aging promotes internal mixing of black carbon
(BC), leading to an enhancement of light absorption and radiative forcing.
The relationship between BC mixing state and consequent absorption
enhancement was never estimated for BC found in the
Arctic region. In the present work, we
aim to quantify the absorption enhancement and its impact on radiative
forcing as a function of microphysical properties and mixing state of BC
observed in situ at the Zeppelin Arctic station (78

Single-particle soot photometer (SP2) measurements showed a mean mass
concentration of refractory black carbon (rBC) of 39 ng m

In the late winter, favorable transport pathways and scarce removal
mechanisms lead to an enhancement of aerosol concentration in the Arctic,
well known as the Arctic haze (Barrie, 1986;
Shaw, 1995). The aerosol population of the Arctic haze is mainly composed of
sulfate, organic matter, ammonium, nitrate, mineral dust and black carbon
(BC) (Quinn et al., 2007). BC, emitted by
incomplete combustion of fossil fuels, biofuels and biomass, is of particular
interest as it is mainly of anthropogenic origin and dominates light
absorption by atmospheric aerosols, causing a positive radiative forcing
(Bond et al., 2013) on a global scale. In the Arctic, BC influences the energy budget
by altering the radiative properties of clouds, absorbing the solar radiation
in the atmosphere and darkening the snow surface (i.e.,
Flanner, 2013; Quinn et al., 2015; Mahmood et al., 2016; Sand et al., 2016).
The combination of these three forcing mechanisms makes the Arctic more
vulnerable to climate change and contributes to what is now called “Arctic
amplification”. However the aforementioned effects depend on the absolute
atmospheric BC mass concentration, which varies between 20 and
80 ng m

Presently, the optical properties of BC in the Arctic atmosphere and its size distribution and mixing state are poorly characterized. During the CLIMSLIP (Climate impacts of short-lived pollutants in the Arctic) project, we addressed this gap with dedicated in situ measurements of BC properties including mixing state and mass absorption cross section during Arctic haze conditions at the Zeppelin station during springtime. This allowed quantification of the absorption enhancement via comparison with observationally constrained optical modeling. Finally, the resulting impact of BC mixing state on its radiative forcing was assessed using a 1-D radiative transfer model.

The CLIMSLIP field experiment took place between 22 March and
11 April 2012. The instrumentation was deployed at the Zeppelin
station (475 m a.s.l.; 78

The single-particle soot photometer (SP2, eight-channel, Droplet Measurement
Technologies, Longmont, CO, USA) was used to determine concentration, size
distribution and coating thickness of BC at the Zeppelin site.
The operation principles are given by Stephens et al. (2003), Schwarz et al. (2006), and Moteki and
Kondo (2010). Briefly, the SP2 is based on the laser-induced incandescence
technique: the particles are directed through an intra-cavity Nd:YAG
continuous-wave laser beam at a wavelength of 1064 nm, in which
light-absorbing particles are heated. BC-containing particles reach
incandescence, and the peak intensity of the emitted thermal radiation, which
occurs when the boiling point temperature of BC is reached, is proportional
to the BC mass contained in the particle. Hereafter we follow the
recommendation of Petzold et al. (2013)
and use the term refractory black carbon (rBC) whenever referring to BC quantified with laser-induced incandescence and use the term

Optical particle sizing is based on the collection of elastically scattered
laser light at 1064 nm. For BC-free particles, which do not evaporate in the
laser beam, the peak scattering intensity is translated to an optical
diameter using a refractive index (RI) of 1.5 and assuming spherical particle
shape. The optical diameter (

Thereby refractive indices of 1.5–0i and 2.26–1.26i, specific for the Nd

The SP2 calibration for the incandescence signal was performed in situ using size-selected fullerene particulate (Alfa Aesar; no. FS12S011). The scattering detector was calibrated in situ with spherical polystyrene latex size standards of 200, 220 and 269 nm in diameter (Thermo Scientific, formerly Duke Scientific). The scattering signal at incandescence signal onset, which is after evaporation of coatings but before onset of BC evaporation, was compared against the incandescence signal to verify that the measured coating thickness is unbiased for uncoated BC when applying the calibrations and BC RI as described above. A complete description of the calibration setup, calibration materials and operation principles can be found in Moteki and Kondo (2010), Gysel et al. (2011), Baumgardner et al. (2012), and Laborde et al. (2012a, b).

The continuous soot monitoring system (COSMOS, Kanomax, Osaka, Japan) is a
single-wavelength photometer measuring the light attenuation through a filter
collecting the aerosol sample (Miyazaki et al., 2008). A key
difference to other aerosol absorption photometers is the heated inlet, which
is operated at a temperature of 400

List of instruments and measured parameters.

Flow chart from
observations to radiative forcing:

Data from several instruments were used to characterize the optical
properties of the total aerosol: a nephelometer (model 3563, TSI Inc., St.
Paul, MN, USA), a seven-wavelength Aethalometer (model AE31, Magee Scientific
Corporation, Berkeley, CA, USA) and a sun precision filter radiometer. The
complete list of instruments used in this work with measured and derived
parameters is presented in Table 1 and schematized
in Fig. 1a, b. The nephelometer was used to
monitor the aerosol total scattering (

The Aethalometer was used to monitor the aerosol light absorption
coefficient (

Note that an updated

The single-scattering albedo (SSA), which is defined as

The aerosol optical depth (AOD) at wavelengths of 368, 412, 500 and 862 nm was monitored by means of a sun precision filter radiometer. Each year,
during wintertime, the instrument is calibrated at the World Optical Depth
Research and Calibration Center of Davos. The AOD is used to assess the
total aerosol load integrated over the vertical column. Additionally,
information about the aerosol size distribution can be derived from the
wavelength dependence of AOD. This dependency is parameterized using the
Ångström exponent of the AOD (

The mass absorption cross section (MAC) of a certain component of particulate matter is defined as the contribution of this component to the aerosol absorption coefficient divided by its mass concentration, which translates to

The influence of observed coatings of BC-containing particles on MAC was
investigated from a theoretical point of view using Mie theory. Calculations
were performed by means of the “BHCOAT” code (Bohren and Huffman, 1998), which is
a numerical implementation of Mie theory. The BC-containing particles are
assumed to have concentric-sphere morphology with a spherical BC core
embedded in a shell of the internally mixed non-absorbing material. The
coating thickness is defined as the thickness of the coating layer, i.e., the
difference between the radii of the whole particle and the BC core
(consistent with the definition in Sect. 2.2.1). The RI of
the BC core was assumed to be 1.95–0.79i at a wavelength of 550 nm (Bond and
Bergstrom, 2006). For the same wavelength, an RI of 1.55–

Radiative transfer simulations were conducted in order to quantify the
effects of different BC mixing state scenarios on atmospheric radiation
fluxes following the schematic shown in Fig. 1d.
The radiative forcing due to the aerosol–radiation interaction (RF

Note, this definition of radiative forcing agrees with the definition by Stamnes et al. (2017) whereas it differs from the definition of
the Intergovernmental Panel on Climate Change (Myhre et al., 2013). More details
on this topic will be provided in Sect. 3.4.2. We used the Atmospheric
Radiative Transfer Database for Earth Climate Observation model (ARTDECO),
which is developed and maintained at the Laboratoire d'Optique
Atmospheìrique (LOA), distributed by the data center AERIS/ICARE
(

Histograms of main aerosol properties calculated from 2 h averages.

ARTDECO further requires various aerosol optical properties as input (see schematic in Fig. 1). The total aerosol burden is provided in the form of AOD at 550 nm wavelength, which was obtained by interpolating the sun radiometer AOD measurements at 368, 412, 500 and 862 nm. Then, ARTDECO internally adjusts the AOD provided at 550 nm to other wavelengths using the wavelength dependence of aerosol extinction. Extinction was calculated as the sum of the absorption- and humidity-corrected scattering coefficients and provided as model input at the wavelengths 370, 550 and 880 nm. The aerosol population was assumed to be confined between 0 and 1 km above ground and chosen to match the AOD. The SSA and asymmetry parameter are provided as input to ARTDECO for the wavelengths 370, 550 and 880 nm, as inferred from interpolated or extrapolated aerosol measurements and Mie calculations. More detail is provided in Sect. 3.4.1, specifically on the relative humidity (RH) dependence of aerosol optical properties and the approach to simulating the effects of different BC mixing state scenarios.

Here we present an optical characterization of the total aerosol at the
Zeppelin station during the Arctic haze 2012 period. All measurements,
excluding AOD, were made at RH

Qualitative information on the aerosol size distribution shape can be
obtained from

The dry aerosol SSA, inferred from absorption coefficient and light-scattering measurements interpolated to a wavelength of 550 nm, was observed
to be

Statistical analysis of black carbon particle properties for the full campaign.

The SP2 quantitatively detects rBC mass in single particles with rBC mass-equivalent core diameters in the range of 80 nm

Frequency distributions of single-particle properties inferred from
the SP2 measurement during the whole campaign:

As summarized by Petzold et al. (2013),
BC can be measured with different techniques. The heterogeneity of
measurement approaches may lead to discrepancies between different types of
operationally defined BC mass concentrations, especially in
pristine areas where BC loadings are close to the limit of detection of many
instruments. The AMAP report (2015) underlined the need for
comparable BC-measuring techniques in the Arctic region in order to
accurately monitor the consequences of anthropogenic activities on BC load
and estimating the subsequent climatic impacts. In order to quantify the
potential inconsistencies between BC mass concentrations measured
by the SP2 and the COSMOS, an intercomparison study was carried out at the
Zeppelin station from 30 March to 11 April when
these two instruments were operated in parallel (more than 200 h of
simultaneous measurements). The COSMOS raw data were analyzed using the mass
absorption cross section reported in Kondo et al. (2009).
The two instruments showed a good correlation (Pearson correlation
coefficient of 0.89; Fig. 5) and the agreement of absolute BC mass values
was good (slope of the regression line is 1.14). When only considering the BC
mass concentration data at values higher than the limit of detection of the
COSMOS instrument reported in the literature
(50 ng m

Comparison of black carbon mass concentrations measured by COSMOS and SP2. Each individual data point represents the 2 h mean while error bars indicate the standard deviation.

The mixing state of BC-containing aerosol was inferred from single-particle
measurements performed with the SP2 using the approach described in Sect. 2.2.1. This was possible for all BC cores having a BC core diameter in the
range of 220 nm

The SP2 makes it possible to distinguish two distinct types of particle morphology for individual internally mixed BC particles (Sedlacek et al., 2012; Dahlkötter et al., 2014; Moteki et al., 2014): (i) BC is only a minor volume fraction and fully embedded in the coating material somewhere near the particle center, and (ii) BC is attached to or at least near the surface of the coating material. We used the method introduced by Moteki et al. (2014) to show that at Svalbard, where the dominant fraction of BC-containing particles was found to have a small BC volume fraction, only around 2 % of the particles containing BC cores in the mass range of 6–10 fg exhibited the SP2 signal features corresponding to the attached geometry. While the exact value is subject to uncertainty, it is a robust result that the embedded type morphology clearly dominates over the attached type morphology for the BC particles. The fact that the dominant fraction of BC particles has substantial coatings with embedded type morphology supports using the simplified assumption of concentric core–shell geometry for inferring the mixing state based on SP2 data and for estimating the effect of the coatings on particle properties.

The above discussion of BC mixing state focused on a narrow BC core size
range and thus on BC-containing particles only. Alternatively, particle
mixing state can be discussed for all particles within a certain optical
particle diameter range including both BC-free and BC-containing particles.
In the following we discuss particles with an overall optical diameter in the
range of 200–260 nm. In this size range, less than 5 % of the particles
contained a detectable amount of BC, while more than 95 % were BC free.
These numbers show that most of the non-BC particulate matter is externally
mixed from BC, by both number and volume. It is important to emphasize that
the reported number fraction of BC-containing particles is a lower limit of
the true value as BC cores with

The aerosol light absorption coefficient was measured at seven wavelengths between 370 and 950 nm. However, from here on, we only discuss results at a wavelength of 550 nm, as this choice will allow direct comparison to results shown in previous literature.

Mass absorption cross section (MAC

The mass absorption cross section of BC was calculated from daily averaged
values of the total absorption coefficient (

The observed MAC

The absorption cross section of BC-containing particles was calculated with
the BHCOAT implementation of Mie theory assuming concentric-sphere geometry
for coated BC cores (Sect. 2.4.1). The refractive
indices at the 550 nm wavelength were assumed to be RI

Figure 6 shows the resulting relationship between
the BC mass absorption cross section and the shell-to-core diameter ratio,
the latter chosen as parameter to indicate the coating thickness in relative
terms. The ensemble of simulations gives an overview on the potential
absorption enhancement at Zeppelin. For bare BC
(

The coated sphere model is a simplification of the actual BC particle
morphology, which might not provide an accurate representation of the actual
mixing geometry of BC particles, with consequent effects for the
estimation of the optical properties (Adachi
et al., 2010). However, it might be considered a fair approximation for
highly aged BC particles at Svalbard, which are embedded in coatings and
have a low BC volume fraction as discussed in Sect. 3.2.3. Indeed, China et al. (2015) also found that aged BC is predominantly embedded in the coating
material and that the Mie approach is suitable for estimating the absorption
of aged BC in such a case. More recently, Liu et al. (2017) confirmed that Mie theory
with assuming spherical core–shell geometry realistically describes the
optical behavior of embedded BC cores when the coating mass is greater than
around 3 times the mass of the BC core. In our work, the volume of the
coating material was converted to mass using a density of 1100 kg m

Campaign average and standard deviation of optical properties of
black carbon and total aerosol for different BC coating thickness scenarios
and resulting radiative implications. The base case BC coating thickness
scenario (“medium”) reflects the median of all measured values, while the
“thin” and “thick” scenarios reflect the 25th and 75th
percentiles, respectively. The scenarios “global model MAC low” and
“global model MAC high” reflect extreme values applied in previous global
model simulations. The black carbon mass absorption cross section (MAC

The most relevant MAC

The absorption enhancement factor (

The fact that the measured and predicted MAC values of BC agree with each other, as shown in Sect. 3.3.2, is the starting point for a sensitivity analysis investigating the effects of light absorption enhancement, due to transparent particulate matter internally mixed with BC, on the optical properties and the radiative forcing of the total aerosol in the Arctic.

Based on the finding discussed in Sect. 3.3.2, we
used the measured rBC mass concentration and the modeled MAC for different
mixing degrees to calculate the aerosol light absorption coefficient
(

Hereby, the approximation is made to use the dry aerosol absorption
coefficient, i.e., MAC

The dry scattering coefficient is taken from the nephelometer measurement
(Fig. 1). Hygroscopic growth of the aerosol at
ambient RH does cause a substantial increase in the scattering cross section
compared to the dry aerosol properties (e.g., Xia et al., 2007). The RH
effects on scattering coefficient and aerosol radiative efficiency were also
shown by Rastak et al. (2014) for aerosol in the European Arctic. At the Zeppelin site Zieger
et al. (2010) measured the aerosol scattering enhancement factor
(

From the ambient scattering coefficient and the dry aerosol absorption
coefficient it was possible to estimate the aerosol SSA
at ambient RH (SSA

As for

Eventually, the absorption enhancement induced by a coating on BC cores
discussed in Sect. 3.3.1 leads to a decrease in
the SSA of the total aerosol (averaged SSA

The obtained SSA

The AOD at 550 nm includes the contribution of hygroscopic
growth and was used to parameterize the aerosol load. As shown in Sect. 3.3.1, the MAC

The Legendre moments of the light-scattering phase function, which are
required for the radiative transfer simulations, were derived within ARTDECO
from the asymmetry parameter and kept unchanged for all simulations. The use
of a constant

Dependence of radiative forcing difference (

In this section we present the results concerning the impact of absorption
enhancement of BC on the total aerosol radiative forcing. The radiative
fluxes at the TOA were quantified with the ARTDECO
package, which makes use of the one-dimensional discrete model 1-D – DISORT2.1 as a radiative transfer equation solver. In addition to aerosol properties,
the equation solver needs other environmental variables such as the aerosol
vertical distribution, surface albedo and solar zenith angle. We define the
radiative forcing difference,

RF

To characterize aerosol and in particular BC physical and optical properties
and to understand the radiative impact of the lensing effect caused by
internal mixing of BC with transparent particulate matter, an intensive field
experiment was conducted during the 2012 Arctic spring at the Zeppelin
station in Svalbard, Norway. An optical survey of the aerosol showed thin
aerosol optical depth, negative values of the Ångström exponent
difference and high SSA, indicating that typical Arctic
haze conditions prevailed with an aerosol population dominated by fine and
non-absorbing particles, while no extreme smoke events occurred. A single-particle soot photometer (SP2) was used to infer the key properties of BC-containing particles. Low rBC mass concentrations of 39 ng m

The effect of absorption enhancement on the aerosol radiative forcing
(RF

The data from the present work are open source and are available
at

The supplement related to this article is available online at:

MZ conceived the paper. MZ and MG interpreted the data and treated the SP2 data. KE and SV provided the Aethalometer data; YK provided the COSMOS data; SK provided the sun radiometer data; PT provided the nephelometer data. PL encouraged investigation of the radiative implications, and PD and VW supervised the use of the ARTDECO model. MZ analyzed all other data, prepared the figures, and drafted the paper. PL and UB supervised the entire work development. HWJ initiated and coordinated the study. All authors actively discussed the results and contributed to the final paper.

The authors declare that they have no conflict of interest.

This work was supported by the Agence Nationale de la Recherche under the
contract ANR 2011 Blanc SIMI 5-6 021 04 and a grant from Labex OSUG@2020
(Investissements d'avenir – ANR10 LABX56). The research project no. 1030
(CLIMSLIP-NyA) was performed at the AWIPEV station Ny-Ålesund and was
supported by the French Polar Institute (IPEV). Marco Zanatta and Martin
Gysel received financial support from the ERC under grant
ERC-CoG-615922-BLACARAT. We acknowledge the support by SFB/TR 172 “ArctiC
Amplification: Climate Relevant Atmospheric and SurfaCe Processes, and
Feedback Mechanisms (AC)