**Aqueous-Phase Brown Carbon Formation from Aromatic Precursors under Sunlight Conditions**

**Kristijan Vidovi´c \* , Ana Krofliˇc , Martin Šala and Irena Grgi´c \***

Department of Analytical Chemistry, National Institute of Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia; ana.kroflic@ki.si (A.K.); martin.sala@ki.si (M.Š.)

**\*** Correspondence: kristijan.vidovic@ki.si (K.V.); irena.grgic@ki.si (I.G.)

Received: 22 November 2019; Accepted: 22 January 2020; Published: 24 January 2020

**Abstract:** At present, there are still numerous unresolved questions concerning the mechanisms of light-absorbing organic aerosol (brown carbon, BrC) formation in the atmosphere. Moreover, there is growing evidence that chemical processes in the atmospheric aqueous phase can be important. In this work, we investigate the aqueous-phase formation of BrC from 3-methylcatechol (3MC) under simulated sunlight conditions. The influence of different HNO2/NO<sup>2</sup> − concentrations on the kinetics of 3MC degradation and BrC formation was investigated. Under illumination, the degradation of 3MC is faster (k2nd(global) = 0.075 M−<sup>1</sup> ·s −1 ) in comparison to its degradation in the dark under the same solution conditions (k2nd = 0.032 M−<sup>1</sup> ·s −1 ). On the other hand, the yield of the main two products of the dark reaction (3-methyl-5-nitrocatechol, 3M5NC, and 3-methyl-4-nitrocatechol, 3M4NC) is low, suggesting different degradation pathways of 3MC in the sunlight. Besides the known primary reaction products with distinct absorption at 350 nm, second-generation products responsible for the absorption above 400 nm (e.g., hydroxy-3-methyl-5-nitrocatechol, 3M5NC-OH, and the oxidative cleavage products of 3M4NC) were also confirmed in the reaction mixture. The characteristic mass absorption coefficient (MAC) values were found to increase with the increase of NO<sup>2</sup> −/3MC concentration ratio (at the concentration ratio of 50, MAC is greater than 4 m<sup>2</sup> ·g <sup>−</sup><sup>1</sup> at 350 nm) and decrease with the increasing wavelength, which is characteristic for BrC. Yet, in the dark, roughly 50% more BrC is produced at comparable solution conditions (in terms of MAC values). Our findings reveal that the aqueous-phase processing of 3MC in the presence of HNO2/NO<sup>2</sup> −, both under the sunlight and in the dark, may significantly contribute to secondary organic aerosol (SOA) light absorption.

**Keywords:** brown carbon chromophores; biomass burning; photooxidation; nitration; nitrophenols; methylnitrocatechols; mass absorption coefficient

### **1. Introduction**

The existence of light-absorbing organic aerosols in the atmosphere, also termed as brown carbon (BrC), has become increasingly evident in recent years [1–3]. In contrast to black carbon (BC), which strongly absorbs light in the entire spectral range [4], BrC typically absorbs in the near-UV (300–400 nm) and blue-to-green visible ranges (400–550 nm) [1,2,5]. BrC impacts radiative forcing directly by absorbing solar irradiation and also indirectly by influencing the cloud formation [6,7]. In addition, its components can take part in different (photo)chemical reactions in the atmosphere and can also have harmful effects on the environment and living organisms [1,8].

BrC is largely observed in areas influenced by biomass and biofuel burning and is often attributed to primary emissions [1,9]. However, there is more and more evidence that secondary transformations of primarily emitted volatile organic compounds (VOCs), in the gas and/or in the atmospheric aqueous phase, also contribute to BrC formation [10–12]. Important precursors to BrC are aromatic compounds, which contribute significantly to the budget of atmospheric pollution [13]; among those, substituted

aromatics are especially important [14]. A class of aromatics with strong absorption in the near UV and visible are nitrophenols (NP), which are frequently determined in atmospheric particulate matter (PM), with elevated concentrations in winter PM and being primarily associated with biomass burning (BB) [15–19]. It has recently been shown that several nitroaromatic pollutants, such as NP, nitrocatechols (NC), and methylnitrocatechols (MNC), can be formed via (photo)chemical oxidation of preferentially gas-phase precursors in the atmosphere [20,21]. Moreover, evidence exists that their formation can take place in the atmospheric aqueous phase, which additionally contributes to the atmospheric abundance of these compounds [11,22,23].

At present, there are numerous open questions concerning the mechanisms of BrC formation, particularly those in cloud droplets and aqueous particles. However, there is rising evidence that chemical processes in the atmospheric condensed phase can efficiently contribute to organic aerosol aging, and thus eventually also to light-absorbing secondary organic aerosol (SOA) formation [24,25]. Yellow-colored MNC have been recognized as essential constituents of BB SOA and further also proposed as suitable tracer compounds [15,16,22,26]. Their complex formation mechanisms under nighttime conditions in the atmospheric aqueous phase have recently been extensively studied [27,28]. Moreover, it has been demonstrated that under sunlight conditions, the formation of nitroguaiacols (nitrated 2-methoxyphenols, NG) and their aging in the atmospheric aqueous phase are even more complex [11,23]. In general, nitrophenols have been shown to be susceptible to direct photolysis and to photo-oxidation [29–31]; as a result, the light-absorbing properties of BrC can change significantly during atmospheric aging [32,33]. Hems and Abbatt [33] have recently investigated the mechanism of color enhancement and fate of nitrophenols (NC, NG, and dinitrophenol) during aqueous-phase photo-oxidation. They have found out that the initially fast-formed nitrophenols, being functionalized with additional OH groups, likely lead to the increased absorption in the visible range, whereas further reactions lead to product formation with poor or no visible absorption (i.e., in the process called bleaching).

In this work, aqueous-phase transformations of 3-methylcatechol (3MC) as an important precursor to light-absorbing MNC were investigated under atmospheric sunlight conditions. The main goal was to find out how photochemical processing affects the kinetics of 3MC degradation at different HNO2/NO<sup>2</sup> − concentrations, and how the absorptive properties of the reaction mixture change during the aging. Although NO<sup>3</sup> − is the prevailing nitrogen species in atmospheric aerosols, HNO2/NO<sup>2</sup> − is known for its much higher reactivity at ambient conditions, at least in the dark, and is thus considered important for SOA formation and aging. Special attention was paid to the main dark BrC products, 3-methyl-5-nitrocatechol (3M5NC), and 3-methyl-4-nitrocatechol (3M4NC), and to the possible formation of absorbing second-generation products, such as hydroxy-3-methyl-5-nitrocatechol (3M5NC-OH). In addition, the influence of different reaction conditions on BrC formation/decomposition was quantified by a cumulative parameter, mass absorption coefficient (MAC in m<sup>2</sup> g −1 ), which was used to describe the evolution of BrC during the solution aging.

### **2. Materials and Methods**

### *2.1. Materials*

3-methylcatechol (3MC) (purity ≥98%) and sodium nitrite (NaNO2, ACS reagent, ≥97.0%), both from Sigma-Aldrich were used for the reaction mixture preparation in high purity water (18.2 MΩ cm), supplied by a Milli-Q water purification system. Standards of two reaction products, 3M5NC and 3M4NC, were prepared in the laboratory by the procedures adopted from Palumbo [34] and Kitanovski [35]. Sulfuric acid 98% (H2SO4, EMSURE, p.a. grade) was used for the pH adjustment.

For the quantification of 3MC, NaNO<sup>2</sup> and first-generation reaction products (3M5NC and 3M4NC), acetonitrile (Sigma-Aldrich, St. Louis, MO, USA, gradient grade, for HPLC >99.9%) and formic acid (Kemika, Zagreb, Croatia) were used for mobile phase preparation. For the second-generation product

identification, methanol (Fluka, Chromosolv LC-MS grade, ≥99.9%) and an ion-pair reagent for LC-MS Dibutylammonium Acetate-DBAA (Tokyo Chemical Industry, Tokyo, Japan, 0.5 mol·L −1 ) were used.

### *2.2. Laboratory Experiments*

The photolysis of 3MC in mildly acidic (pH = 4–5, adjusted with H2SO4) aqueous solutions of sodium nitrite (NaNO2;) was investigated at ambient temperature (25 ◦C). The reactions were carried out in a custom-built reactor, which is a modified rotating evaporator (Büchi, Flawil, Switzerland) equipped with a thermostated bath [36]. Only briefly, a low-volume condenser was installed to prevent significant evaporation from the solution ensuring a closed physical system. The reaction mixture (100 mL) was continuously mixed by rotation (50 rpm) of a round-bottom flask made of the DURAN glass (transparent one for daytime experiments and amber glass for dark reactions). For illumination, a LOT-QuantumDesign Europe solar simulator (L.O.T.-Oriel GmbH & Co. KG, Darmstadt, Germany) equipped with a xenon short-arc lamp (300 W, ozone free) was used. According to the specifications, the simulator produces irradiation equivalent to approximately one sun at the working distance of 180 mm. Moreover, the cut-off of the DURAN glass at 300 nm resembles the absorption of the stratospheric ozone, which allows us to mimic ambient conditions very well. The concentrations of 3MC and H2SO<sup>4</sup> were fixed at 0.1 and 0.05 mM, respectively. The influence of NaNO<sup>2</sup> concentration was investigated at 0.1, 0.5, 1.0, 2.0, and 5.0 mM. Samples (1.5 mL aliquots) were taken from the reaction mixture according to the predefined time protocol (usually each hour). In addition, a control experiment of the direct photolysis of 3MC without the addition of NaNO2/H2SO4, and the dark nitration of 3MC in acidic NaNO<sup>2</sup> (1 and 2 mM) solutions were performed.

An ultra-high pressure liquid chromatography (UltiMate 3000 UHPLC System; Thermo Fischer Scientific, Waltham, MA, USA) coupled with a diode array detector (DAD) or a triple quadrupole/linear ion trap mass spectrometer (4000 Q TRAP LC-MS/MS System; Applied Biosystems/MDS Sciex) was used for the detection of reactants and products. For the quantification of 3MC, NaNO2, and first-generation products (3M5NC and 3M4NC), an isocratic elution program with acetonitrile/0.1% formic acid (70/30, *<sup>V</sup>*/*V*) at the flow rate of 0.6 mL·min−<sup>1</sup> was applied, assuring the separation of components on an Atlantis T3 column (3.0 <sup>×</sup> 150 mm<sup>2</sup> , 3 µm particle size; Waters, Milford, MA, USA). The injection volume and the column temperature were 10 µL and 30 ◦C. The detection wavelengths used in the case of HPLC-DAD measurements were 275 nm for 3MC, 345 nm for MNC, and 355 nm for NaNO2. For the second-generation product identification (*m*/*z* 182 and 184), different chromatographic conditions were used: a Hypersil GOLD aQ column (2.1 <sup>×</sup> 150 mm<sup>2</sup> , 3 µm particle size; Thermo Scientific, Waltham, MA, USA) and an isocratic elution with methanol/50 mM DBAA (10/90, *V*/*V*) at the flow rate of 0.3 mL·min−<sup>1</sup> . The injection volume and the column temperature were again 10 µL and 30 ◦C. The detection wavelength used in this case was 388 nm.

Absorption spectra of the reaction solution were measured offline in a 1 cm quartz cuvette with a UV-Vis spectrometer (Lambda 25, PerkinElmer, Waltham, MA, USA) immediately after sampling. The absorbance was monitored in a broad spectral range from 200 to 700 nm.

### *2.3. Kinetic Analysis*

The experimentally obtained 3MC photolysis profiles were first treated by a pseudo-first-order kinetics:

$$\frac{\partial \mathbf{c}}{\partial t} = -k\_{\rm app} \cdot \mathbf{c}\_{\prime} \tag{1a}$$

where *c* is the concentration of 3MC, *t* denotes time, and *k*app is the apparent pseudo-first-order kinetic rate constant (in s−<sup>1</sup> ) of 3MC degradation. The integrated form of Equation (1a) was fitted to the experimental data points, *c*<sup>0</sup> being the initial concentration of 3MC in the reaction mixture.

$$
\mathfrak{c} = \mathfrak{c}\_0 \cdot e^{-k\_{\rm app} \cdot t}. \tag{1b}
$$

Referring to the direct photolysis of 3MC, which is negligible (see the control experiment), we postulated that the degradation of 3MC be dependent on the concentration of NaNO2.

$$\frac{\partial \mathbf{c}}{\partial t} = -k\_{\text{2nd}} \cdot \mathbf{c} (\text{NaNO}\_2) \cdot \mathbf{c}.\tag{2a}$$

Except for the condition where the initial concentrations of 3MC and NaNO<sup>2</sup> were equal (i.e., 0.1 mM; experiment 1), we further assumed constant NaNO<sup>2</sup> concentration in the reaction mixture throughout the course of experiment. Second-order kinetic rate constants (*k*2nd in M−<sup>1</sup> ·s −1 ) were thus estimated from *k*app as:

$$k\_{\rm app} = k\_{\rm 2nd} \cdot c(\rm NaNO\_2)\_0. \tag{2b}$$

*c*(NaNO2)<sup>0</sup> being the initial concentration of NaNO<sup>2</sup> in the reaction mixture. Note, however, that one cannot assume constant NaNO<sup>2</sup> concentration during experiment 1. Therefore, a second-order kinetic treatment was applied in this case, taking into account the measured *c*(NaNO2) at each time of the experiment.

$$\frac{\partial \mathbf{c}}{\partial t} = -k\_{\text{2nd}} \cdot \mathbf{c} (\text{NaNO}\_2)\_t \cdot \mathbf{c} \tag{3}$$

where *c*(NaNO2)<sup>t</sup> represents the concentration of NaNO<sup>2</sup> at the reaction time *t*.

The experimental data for which the assumption of a pseudo-first-order is valid (*c*(NaNO2) = 0.5, 1.0, 2.0, and 5.0 mM) were further treated simultaneously by the following function:

$$\frac{\partial \mathbf{c}}{\partial t} = -k\_{\text{2nd}} (\text{global}) \cdot \mathbf{c} (\text{NaNO}\_2)\_0 \cdot \mathbf{c}.\tag{4}$$

This gave us a global second-order kinetic rate constant, *k*2nd(global) representing all treated experimental conditions. Note that parameters obtained by accounting for the data at different experimental conditions at once tend to be more universally applied.

OriginPro 2018 was used for the fitting of the model functions to the experimental data (i.e., measured concentration and absorption profiles).

### *2.4. Determination of Absorption Properties*

The obtained *k*2nd(global) (Equation (4)) was further used to quantitatively describe the influence of the studied reaction conditions on BrC formation. In the case of experiment 1 (Table 1), *k*2nd(global) is not valid; therefore, *k*app (Equation (1a)) was used instead. According to the Beer–Lambert Law, the measured absorbance of a sample at a distinct wavelength, *A*tot is dependent on the characteristic mass absorption coefficient of contained species *i* at this wavelength, MAC*<sup>i</sup>* , and their concentrations, *ci* ; *l* is the absorption path length characteristic of the used cuvette.

$$A\_{\rm tot} = \sum\_{i} \mathbf{M} \mathbf{A} \mathbf{C}\_{i} \cdot \mathbf{c}\_{i} \cdot \mathbf{l}. \tag{5a}$$

**Table 1.** Ratios and yields of quantified 3-methyl-5-nitrocatchol (3M5NC) and 3-methyl-4-nitrocatechol (3M4NC) at different NaNO<sup>2</sup> concentrations under sunlight conditions after 8 h of reaction.


By definition, BrC absorbs in the near-UV and visible ranges, so we attributed *A*tot measured above 300 nm to the BrC formation (note that 3MC does not absorb in this range). Equation (5a) can be thus expressed as:

$$A\_{\rm tot}(>300\,\text{nm}) = \text{MAC}\_{\text{BrC}} \cdot \mathcal{m}\_{\text{BrC}} \cdot l. \tag{5b}$$

MACBrC (in m<sup>2</sup> ·g −1 ) being the mass absorption coefficient characteristic of the formed BrC and *m*BrC is its mass concentration in g m−<sup>3</sup> .

By combining Equation (5b) with the expression for the pseudo-first-order product formation and taking into account the law of conservation of mass (i.e., assuming that all consumed 3MC was converted to BrC; *m*<sup>0</sup> is the initial 3MC mass concentration):

$$m\_{\rm Br\overline{C}} = m\_0 \Big(1 - e^{-k\_{\rm 2nd}(\rm global) \cdot c(\rm NaNO\_2)\_0 \cdot t} \Big). \tag{6}$$

One can describe the evolution of *A*tot with time with the following function:

$$A\_{\rm tot} = \text{MAC}\_{\rm BrC} \cdot m\_0 \Big(1 - e^{-k\_{\rm rad} \left(\text{global}\right) \cdot c \left(\text{NaNO}\_2\right)\_0 \cdot t}\Big) \cdot l. \tag{7}$$

The derived function was fitted to the experimental data points, and characteristic MACBrC values for the investigated experimental conditions were obtained. Note that in the case of experiment 1 Table 1), *k*app was used instead of the product of *k*2nd(global) and *c*(NaNO2)0.

### **3. Results and Discussion**

### *3.1. Kinetic Analysis of 3MC Photolysis in Mildly Acidic NaNO<sup>2</sup> Solution*

In the dark reaction of 3MC and NaNO2, two main reaction products have been recently quantified, i.e., 3M5NC and 3M4NC; they accounted for 70–100% of the reacted 3MC at similar reaction conditions as used in this study [27]. In addition, the ratio of 3M5NC to 3M4NC was constant during the dark reaction and dependent only on the concentration of HNO<sup>2</sup> [27].

Under sunlight conditions, dark reaction mechanisms also exist; thus, 3M5NC and 3M4NC were expected to be formed in this study as well. The performed experiments under illumination are listed in Table 1 together with the corresponding 3M5NC/3M4NCratios and product yields determined after 8 h of reaction. The product yield was calculated as the sum of concentrations of target reaction products divided by the concentration of the reacted 3MC.

$$\text{Product yield} = \frac{\Sigma[\text{MNC}]\_{\text{8h}}}{[\text{3MC}]\_{\text{0h}} - [\text{3MC}]\_{\text{8h}}}.\tag{8}$$

Under the sunlight, the cumulative yield of identified MNC products was only up to about 45%. The ratio (and the yield) of 3M5NC to 3M4NC were additionally found to be dependent on the initial concentration of NaNO<sup>2</sup> (Table 1). The low product yield and the variable product ratio suggest the existence of different oxidation pathways of 3MC, and formation and/or degradation pathways of MNC products in comparison to the dark reaction.

During the photolysis of HNO2, diverse reactive species are formed (e.g., OH, NO, NO2), which are possible sinks of 3MC [37]. In general, OH radicals react with aromatics with high (near-diffusion controlled) rate constants in the order of 10<sup>10</sup> <sup>L</sup>·mol−<sup>1</sup> ·s <sup>−</sup><sup>1</sup> at 298 K [38,39]. The reactivity of aromatics with NO<sup>2</sup> is still relatively high (in the order of 108–10<sup>9</sup> L mol−<sup>1</sup> ·s −1 ), whereas NO seems to be too weak to react with aromatic compounds directly [23]. In summary, other degradation pathways of MNC, i.e., initiated by radicals, are very likely under illumination [33].

Under illumination, also the decay of 3MC was greatly dependent on the applied conditions (Table 2). In general, *k*app increased as the NaNO<sup>2</sup> concentration increased. Note, however, that in the cases where NaNO<sup>2</sup> was in excess, *k*2nd slightly decreased with NaNO<sup>2</sup> addition. This may be due to unaccounted 3MC degradation pathways, independent of (or indirectly dependent on) the NaNO<sup>2</sup> concentration. **Table 2.** Kinetic rate constants (apparent pseudo-first-order, kapp; second-order, k2nd; and global

*Atmosphere* **2020**, *11*, 131 6 of 17

second-order, k2nd(global)) of the photolysis of 3-methylcatechol (3MC) in aqueous solution (pH = 4.5–5,

**Table 2.** Kinetic rate constants (apparent pseudo-first-order, *k*app; second-order, *k*2nd; and global second-order, *k*2nd(global)) of the photolysis of 3-methylcatechol (3MC) in aqueous solution (pH = 4.5–5, T = 25 ◦C) at different initial concentrations of NaNO<sup>2</sup> under simulated sunlight conditions (YES) and in the dark (NO). The initial concentration of 3MC was always 0.1 mM. T = 25 °C) at different initial concentrations of NaNO2 under simulated sunlight conditions (YES) and in the dark (NO). The initial concentration of 3MC was always 0.1 mM. **Experiment NaNO2 [mM] Sunlight** *k***app (s−1)** *k***2nd (M−1s−1)** *k***2nd(global) (M−1·s−1)** 


3.1.1. Kinetic Analysis at Low NaNO<sup>2</sup> Concentrations 3.1.1. Kinetic Analysis at Low NaNO2 Concentrations

The time-dependent concentrations of 3MC (0.1 mM initial concentration) in 0.1 mM NaNO<sup>2</sup> solution under sunlight conditions are shown in Figure 1. In the investigated time range, *k*app (7.88 <sup>×</sup> <sup>10</sup>−<sup>6</sup> s −1 ) describes the behavior of this reaction reasonably well and is comparable with the reaction in the dark (8.33 <sup>×</sup> <sup>10</sup>−<sup>6</sup> s −1 ) [27]. However, the yield of identified MNC under sunlight is barely 17% after 8 h (Table 1), pointing to different degradation and/or formation pathways of 3MC and MNC, respectively. The time-dependent concentrations of 3MC (0.1 mM initial concentration) in 0.1 mM NaNO2 solution under sunlight conditions are shown in Figure 1. In the investigated time range, *k*app (7.88 × 10⁻<sup>6</sup> s⁻1) describes the behavior of this reaction reasonably well and is comparable with the reaction in the dark (8.33 × 10⁻6 s⁻1) [27]. However, the yield of identified MNC under sunlight is barely 17% after 8 h (Table 1), pointing to different degradation and/or formation pathways of 3MC and MNC, respectively.

**Figure 1.** 3-methylcatechol (3MC) degradation in aqueous solution at 0.1 mM initial concentrations of 3MC and NaNO2 (pH = 3.9, T = 25 °C). Experimental data (symbols) and calculated time-dependent **Figure 1.** 3-methylcatechol (3MC) degradation in aqueous solution at 0.1 mM initial concentrations of 3MC and NaNO<sup>2</sup> (pH = 3.9, T = 25 ◦C). Experimental data (symbols) and calculated time-dependent concentration profiles (lines) for the pseudo-first order (*k*app) and second order (*k*2nd) kinetics.

concentration profiles (lines) for the pseudo-first order (*k*app) and second order (*k*2nd) kinetics. Second-order kinetics was also applied to this set of experimental data, and the comparison of both models (Equations (1) and (3)) is shown in Figure 1. As *k*app describes experimental data adequately in Second-order kinetics was also applied to this set of experimental data, and the comparison of both models (Equations (1a) and (3)) is shown in Figure 1. As *k*app describes experimental data adequately in the investigated time range, we used it in further calculations with a fair amount of confidence.

the investigated time range, we used it in further calculations with a fair amount of confidence.

3.1.2. Kinetic Analysis at High NaNO2 Concentrations

Time-dependent concentrations of 3MC in the presence of different initial NaNO2 concentrations (i.e., from 0.5 to 5 mM) are shown in Figure 2a−d. The lines are the calculated concentration profiles
