Air Quality, Atmosphere & Health https://doi.org/10.1007/s11869-018-0607-z Aqueous chemistry of airborne hexavalent chromium during sampling M. Amouei Torkmahalleh 1 & M. Karibayev 1 & D. Konakbayeva 1 & M. M. Fyrillas 2 & A. M. Rule 3 Received: 27 April 2018 / Accepted: 22 July 2018 # Springer Nature B.V. 2018 Abstract Cr(III) is an essential micronutrient for the proper function of human being, while Cr(VI) is a carcinogenic chemical, which has been one of the hazardous air pollutants defined by US Environmental Protection Agency (US EPA) in 2004. Accurate measurements of atmospheric hexavalent chromium concentration are required to evaluate its toxicity. In the present study, a simulation tool using MATLAB program was developed to evaluate soluble and insoluble chromium species formed during the Cr(VI) field sampling (500 ml, 0.12 M HCO3− buffer, pH = 9, 24 h, cellulose filter) which will assist us to better quantify the hexavalent chromium concentration. In this study, Cr(VI) was found to be dominant in soluble form as CrO42− and in precipitated form as (NH4)2CrO4, CaCrO3, BaCrO4, and PbCrO4 at pH = 9 cellulose filter. Secondly, reduction of Cr(VI) to Cr(III) was higher than the oxidation of Cr(III) to Cr(VI). Basic pH solutions retard the conversion of Cr(VI) in the presence of Fe(II) and As(III) and facilitate the precipitation of Cr(III). The presence of the NaHCO3 as buffer on the cellulose filters and also in the filter extraction solution may add to the precipitation of Cr(VI) as NaCrO4. This study provides new insights to improve cellulose sampling filters, and the filter extraction solutions to either prevent Cr(VI) precipitation during the wet analysis of Cr(VI) or improve the Cr(VI) analysis methods to quantify total Cr(VI) (soluble and insoluble Cr(VI)). Keywords Cr . Modeling . Sampling . MATLAB . Particulate matter . Sodium bicarbonate Introduction There are many studies in the literature addressing total Cr concentration in airborne particulate matter (RogulaKozlowska 2016; Khan et al. 2017). However, limited studies are found to quantify different oxidation states and chemistry of Cr in ambient PM. Primarily, there are two main oxidation states of chromium found in atmospheric particulate matter: trivalent chromium (Cr(III)) and hexavalent chromium Electronic supplementary material The online version of this article (https://doi.org/10.1007/s11869-018-0607-z) contains supplementary material, which is available to authorized users. * M. Amouei Torkmahalleh mehdi.torkmahalleh@nu.edu.kz 1 Chemical and Aerosol Research Team, Department of Chemical Engineering, School of Engineering, Nazarbayev University, Astana, Kazakhstan 010000 2 Department of Mechanical Engineering, Frederick University, 1303 Nicosia, Cyprus 3 Department of Environmental Health and Engineering, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD 21209, USA (Cr(VI)). While hexavalent chromium is highly toxic, trivalent chromium is an essential microelement for living organisms and plays a vital role in the control of lipid, glucose, and protein metabolism (Kotas and Stasicka 2000). Cr(VI) is considered a BGroup A carcinogen^ by US Environmental Protection Agency (US EPA), and exposure to Cr(VI) leads to asthma, lung cancer, nasal damage, and bronchitis depending on the duration of the exposure (Park et al. 2004). The solubility of Cr(VI) designates its toxicity. Insoluble Cr(VI) could be more toxic than soluble Cr(VI) compounds. Soluble hexavalent chromium can be reduced to Cr(III) in the blood of living organisms (ATSDR 2000). Highly insoluble Cr(VI) such as ZnCrO4, BaCrO4, and PbCrO4 induced tumor responses during experiments performed on animals (IARC 1990). Soluble and insoluble atmospheric chromium chemistry has been poorly understood in particular during Cr(VI) sampling (Guertin et al. 2005). Accurate sampling and analysis techniques are required to better quantify the soluble and insoluble hexavalent chromium in the atmosphere to appropriately understand its health effects. Hexavalent chromium exists in solution as CrO42−, HCrO4 − , and H2CrO4 (Palmer and Puls 1994). At pH around 1, chromic acid (H2CrO4) is the dominant form, whereas at pH values between 1 and 5, HCrO4− is more favored. Dichromate Air Qual Atmos Health (CrO42−) becomes more dominant at pH values greater than 6 (Davis and Oslen 1995). The dominant form of trivalent chromium in solutions with pH < 4 is Cr3+ ions (Rai et al. 1987). However, for pH values higher than 4, trivalent chromium could be found as Cr(OH)2+ and Cr(OH)4−. In the pH range of 6 to 12, Cr(OH)3 precipitates (Rai et al. 1987). The interconversion between Cr(III) and Cr(VI) can take place with atmospheric reductants and oxidants under aqueous and solid-gas phase reactions. The HNO3, HCHO, SO2, NO2, O3, V2+, Fe2+, m-xylene, and benzene reactions with hexavalent chromium in a laboratory chamber were investigated by Grohse et al. (1988). Due to a wide range of reactant concentrations, the percentage of conversion between Cr(VI) and Cr(III) ranged from 5 to 99% after 24 h. The oxidation of Cr(III) to Cr(VI) through reduction with MnO2 is mainly a heterogonous reaction, and its reaction rate increases as pH values decrease (Guertin et al. 2005). The Fe(II) in NaCl solution can reduce Cr(VI) into Cr(III) with higher rates in the pH range of 5 to 8.7 (Pettine et al. 1998). Seigneur and Constantinou (1995) developed a computer model to predict chromium chemistry and speciation in two sets of ambient atmospheric PM matrixes: liquid-coated particles at pH 1 and liquid droplets at pH 4. The results of the computer simulation model showed that overall conversion of Cr(VI) to Cr(III) both in liquid-coated particles and droplets is favorable. However, their investigation did not consider soluble and insoluble chromium chemistry and speciation. Moreover, their model did not implement field data either as the model input or for the model validation. The bicarbonate impregnated cellulose filters at pH 9 are utilized to sample atmospheric chromium, as Teflon filters do not absorb water during the impregnation of the filters. Eastern Research Group (ERG) has shown that when filters are left in the National Air Toxic Trends Stations (NATTS) samplers for 33 to 105 h at the temperature above 15 °C, the reduction in Cr(VI) ranged from 30 to 58% (ERG 2007). Tirez et al. (2011) determined 75 ± 39% Cr(VI) recovery, and the Cr(III) conversion was 1.7 ± 1.2% during the sampling in Flemish Region of Belgium. The observed conversion of Cr(VI) during sampling in previous studies could be the result of the deliquescence of collected ambient particulate matter that provides aqueous reaction media (Amouei Torkmahalleh et al. 2012, and the subsequent reactions with organic matter, SO2, and other reductants (Huang et al. 2013; Grohse et al. 1988). The conversion of Cr(III) during sampling could be attributed to the reaction of Cr(III) with dissolved Mn (Seigneur and Constantinou 1995; Nico and Zasoski 2000), water-soluble organic carbons (WSOC) that contain secondary organic aerosol (SOA) (Huang et al. 2013) and also reactions with gaseous oxidants such as O3 and reactive oxygen species (ROS) (Werner et al. 2006; Amouei Torkmahalleh et al. 2012, 2013a). Cellulose and Teflon filters have lower Cr background in comparison with other filters such as PVC and quartz (ERG 2007). ERG (2007) suggested that the cellulose filters are more suitable compared to other types of filters for the preservation of Cr(VI). The reported concentrations and conversion of Cr(VI) in the atmosphere are concerned with soluble Cr(VI), and very little information exists for the insoluble Cr(VI) concentration. This limitation is due to the analytical methods employed in previous studies that utilize slightly acidic or basic filter extraction, which is unable to dissolve all insoluble Cr(VI). Our understanding of soluble and insoluble chromium chemistry during and after 24 h sampling of chromiumcontaining particles, and the subsequent extraction needs to be improved. The objective of the present investigation is to model Cr chemistry in airborne particulate matter at pH 9 applicable to the Cr(VI) sampling, and filter extraction to identify the soluble and insoluble compounds of Cr. This identification will then help us to better design Cr(VI) monitoring systems including the sampling and filter extraction. Materials and methods Simulation tool The proposed model in our study is illustrated schematically in Fig. 1. Inside the sampling cellulose filters, solid particles including insoluble and soluble Cr compounds (PbCrO 4 , K 2 CrO 4 , BaCrO 4 , CaCrO 4 , Na 2 CrO 4 , and (NH4)2CrO4) exist. A layer of water is formed around the solid particles due to the deliquescence of the particles at or beyond the deliquescence relative humidity (DRH). The deliquescence results in the dissolution of soluble Cr species and other soluble species into the aqueous layer. The dissolved species are present in a basic aqueous solution covering the solid-core particles in the ionic forms. Fig. 1 Schematic representation of the simulation model. The dark blue region shows the particle core where solid compounds exit; the middle blue zone shows the aqueous solution; the light blue area presents the ambient air Air Qual Atmos Health The pH of the solution is maintained at 9 due to impregnating of the filters using NaHCO3 prior to sampling. The infinite amount of gaseous species including O2, SO2, NH3, and O3 with constant concentration surrounds the basic solution. The solid Cr compounds in the particles are assumed to be non-limiting. The ambient air temperature was set as (24 °C) in the model. Interconversion of Cr(III) and Cr(VI) Seventeen equilibrium reactions (Table 1) and five kinetic reactions (Table 2) representing the chemistry of Cr species at basic aqueous solutions are considered for the model. Initial concentrations of species The ambient PM concentrations were obtained from the literature. Concentrations of the dissolved gases in the aqueous layer were calculated by Henry’s law to obtain the initial concentrations of those species in the model. Table 3 presents the initial concentration values in the present study. Atmospheric trace elements and SO 2 concentrations were obtained from the daily air measurements that were performed in Meadowlands, New Jersey by the Department of Environmental Protection in 2012 (NJDEPA 2012). It is understood that the concentrations of the different oxidation states of a given element vary with the source and the location. Since no information is available for the speciation of the trace elements in Table 1 ambient PM in NJ, USA, we assumed that the total concentration of the given element is equivalent to its reactive oxidation state. Thus, with this assumption, the upper limit of the effect of the reactive trace elements on Cr speciation is considered in this study. For example, the VO2+ concentration was assumed to be the same as V concentration. Total ambient atmospheric concentrations of As were assumed to be the As(III) concentration since it is the reactive form of As and As (V) concentration was assigned to be zero. The following equation can estimate the absorption of atmospheric SO2 by water: ½SO2 Šaq ¼ H1 ½SO2 Šg ð1Þ where H 1 is 3.33 μg SO 2 . ml −1 solution /mg SO 2 /m 3 air (Terraglio and Manganelli 1967). The saturation concentrations estimated for oxidants in aqueous solutions including O3, O2, HO2, and HO3 were obtained from the Werner et al. (2006). Equation (2) was used to find the solution concentration of atmospheric species from their ambient concentrations. CðaqÞ ¼ K Cair =W ð2Þ where C(aq) is the given species concentration in the aqueous layer formed around atmospheric PM due to the deliquescence (mol/l solution), and K is a conversion factor. Cair is the given species concentration in the ambient air (μg m−3air), and BW^ is the liquid water content of the ambient atmospheric PM (μg water m−3air) which is equal to 7.0 μg m−3air. Equilibrium reactions (M is mol/L of solution) Number Reactions Equilibrium parameter Reference 1 Cr(OH)3 (s) + 3H+(aq) ⇌ Cr3+(aq) + 3 H2O(aq) Cr2(SO4)3 (s) ⇌ 2Cr3+(aq) + 3 SO42−(aq) 2HCrO4−(aq) ⇌ Cr2O72−(aq) + H2O(aq) HSO4−(aq) + HCrO4−(aq) ⇌ CrSO72−(aq) + H2O(aq) HSO4−(aq) ⇌ SO42−(aq) + H+(aq) HCrO4−(aq) ⇌ CrO42−(aq) + H+(aq) H2O(aq) + SO2 (aq) ⇌ H2SO3 (aq) H2SO3 (aq) ⇌ HSO3−(aq) + H+(aq) HSO3−(aq) ⇌ SO32−(aq) + H+(aq) Keq1 = 2 × 109 M−2 Keq2 = 1.3 × 10−7 M5 Keq3 = 98 M−1 Keq4 = 4.1 M−1 pKeq5 = 1.987 pKeq6 = 6.51 Keq7 = 0.966 Keq8 = 1.7 × 10−2 M Keq9 = 5 × 10−6 M Rai et al. 1987 Seigneur 1985 Beattie and Haight 1972 Beattie and Haight 1972 Hummel et al. 2002 Sadiq 1992 Terraglio and Manganelli 1967 Terraglio and Manganelli 1967 Terraglio and Manganelli 1967 PbCrO4(s) ⇌ Pb2+(aq) + CrO42−(aq) BaCrO4(s) ⇌ Ba2+(aq) + CrO42−(aq) (NH4)2CrO4(s) ⇌ 2NH4+(aq) + CrO42−(aq) CaCrO4(s) ⇌ Ca2+(aq) + CrO42−(aq) K2CrO4(s) ⇌ 2 K+(aq) + CrO42−(aq) Na2CrO4(s) ⇌ 2Na+(aq) + CrO42−(aq) Cr(OH)3(s) + H2O(aq) + 3O3 ⇌ H+(aq) + HCrO4−(aq) + 3HO3(aq) Cr(OH)3(s) + H2O(aq) + 3O2 ⇌ H+(aq) + HCrO4−(aq) + 3HO2(aq) Keq10 = 2.8 × 10−13 M2 Keq11 = 1.2 × 10−10 M2 Keq12 = 75.7 M3 Keq13 = 7.1 × 10−4 M2 Keq14 = 136 M3 Keq15 = 626 M3 Keq16 = 1.23 × 10−25 M2 Keq17 = 3.24 × 10−86 M2 Sillen and Martell 1964 Huang et al. 2014 Huang et al. 2014 Huang et al. 2014 Huang et al. 2014 Huang et al. 2014 Werner et al. 2006 Werner et al. 2006 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Air Qual Atmos Health Table 2 Kinetic reactions Number Reactions Equilibrium parameter 180½ 2þ Š½ Š½ Reference Š VO HCrO−4 HCrO−4 = VO2þ 18 Cr(VI) + 3VO2+ → Cr(III) + 3VO2+ k ¼ 1þ320 19 Cr(VI) + 3 Fe(II) → Cr(III) + 3 Fe(III) 2HCrO4−(aq) + 4HSO3−(aq) + 6H+ → 2Cr3+ + 2SO42− + S2O62− + 6H2O(aq) 3Cr(VI) + 2As(III) → 3Cr (III) + 2As(V) 2Cr(III) + 3MnO2 → 2Cr(VI) + 3Mn(II) k = (4.4 × 103[H+] + 3 × 103[H+]2)[Fe2+][HCrO−4 ] k = 3.9 × 106[Cr(VI)][HSO−3 ][H+] Espenson 1970 Beattie and Haight 1972 ð710−5 þ0:02½Hþ Šþ2:5½Hþ Š2 Þ½CrðVIފ k¼ 1þ22½As1ðIIIފ k = 0.92[MnO2][Cr(III)] Beattie and Haight 1972 Schroeder and Lee 1975 20 21 22 Computer simulation Espenson 1964 Š illustrated the schematic algorithms for solving these equilibrium and kinetic reactions using MATLAB 2016a software. The aqueous layer of basic solution was considered to be a well-mixed batch reactor with constant volume, where there is no inlet and outlet flow. Within this well-mixed reaction mixture, equilibrium and kinetic reactions took place. The equilibrium reactions were simulated using their equilibrium constants (Table 1), while rate expressions were utilized for simulation of the kinetic reactions. The equilibrium reactions were considered to proceed instantaneously. A loop was created to combine the equilibrium and kinetic reactions that illustrated schematically in Fig. 2. In this created loop, initial concentrations of Table 3 were used to run the equilibrium reactions. Then, the concentrations obtained after the equilibrium reactions were considered as the initial concentrations for the kinetic reactions. The results of the kinetic reactions provided the initial concentrations for the equilibrium reactions. This loop continued until the concentrations of all atmospheric species stabilized such that their concentrations were at least 99% of the concentrations found in a previous loop. Figure 2 Table 3 Initial Concentrations in the computer simulation ½ Numerical methods To obtain the concentrations of the compounds after equilibrium, the extent of reaction was estimated. For example, for two equilibrium reactions, there is a two system of equations where εi is the extent of the reaction, which is unknown: aA þ bB→cC þ dD Keq1 ¼ ð3Þ ½C þ ε1 þ ε2 Š½D þ ε1 Š ½A −ε2 −ε1 Š½B−ε1 Š ð4Þ aA þ eE→cC þ hH Keq2 ¼ ð5Þ ½C þ ε1 þ ε2 Š½H þ εŠ ½A−ε1 −ε2 Š½E−ε2 Š ð6Þ Using the initial concentrations and equilibrium constants, the extent of reactions will be estimated by solving Eqs. (4) Species Ambient initial concentration Species Ambient initial concentration HCrO4− 2.43 × 10−2 (μg/m3air) 2.43 × 10−2 (μg/m3air) 0 7.0 (μg/m3air) 0.9254 (μg/m3air) 5.88 × 10−3 (μg/m3air) 0 Pb2+ Na+ K+ Ca2+ Ba2+ As(III) As(IV) 1.93 × 10−3(μg/m3air) 6.99 × 10−1(μg/m3air) 4.776 × 10−1(μg/m3air) 3.838 (μg/m3air) 3.47 × 10−3(μg/m3air) 5.0 × 10−4(μg/m3air) 0 0.16259 (μg/m3air) 0 2.6 × 10−4(mol/Lsolution) 1.0 × 10−15(mol/Lsolution) 1.0 × 10−8 (mol/Lsolution) 8.0 × 10−9(mol/Lsolution) 14.0 (μg/m3air) 0 MnO2 Mn2+ S2O62− SO2(g) H2SO3 HSO3− SO32− CrSO72− 2.51 × 10−3(μg/m3air) 0 0 1.0 × 10−3 ppm 0 0 0 0 +3 Cr CrO4− SO42− HSO4− VO2+ VO2+ Fe(II) Fe(III) O2 HO3 (aq) O3 HO2 (aq) NH4+ Cr2O72− Air Qual Atmos Health In this study, equilibrium refers to the condition when the concentration of a given atmospheric species does not change with time. A Cr compound is defined as soluble at a given pH, if its net transfer direction was from the solid core to the aqueous layer of solution. After dissolution, soluble compound dissociates to its constitutive ions. A Cr compound is considered insoluble, if the net transfer of the Cr compound is from the aqueous layer to the solid core. An insoluble compound precipitates at a given pH (pH 9 in our study), and thus is present both in solution as ions and in the solid core when the equilibrium is established. Results Fig. 2 Schematic diagram of the simulation tool Cr(VI) solution equilibrium and (6). The extent of reaction is applied to update concentrations for the next step each time. A positive value of the extent of reaction shows that the reaction occurs in the forward direction, while a negative value indicates that the reaction progressed in the reverse direction. The production or consumption rate of the individual reactant or product is calculated through the overall reaction rate as shown in Eqs. (7) and (8). Figure 3 illustrates total ambient soluble Cr(VI) concentrations over time, where total Cr(VI) is a sum of HCrO4−, Cr2O72−, and CrO42−. The total ambient soluble Cr(VI) concentration sharply decreased and reached steady state (8.08 × 10−5 μg m−3air) after 0.005 s. Figures S1 and S2 in the Electronic Supplementary Material and Fig. 4 present the ambient concentrations of HCrO4−, Cr2O72−, and CrO42− with time, respectively. From the beginning of the simulation until steady state, HCrO4− concentration sharply decreased to 2.67 × 10−10 μg m−3air, while CrO42− and Cr2O72− concentrations dramatically increased from zero to 8.05 × 10 −5 μg m−3air and 1.55 × 10−10 μg m−3air, respectively. The most dominant soluble Cr(VI) species in the solution was CrO42−. The production of CrO42− took place through reaction 6 (Table 1) given HCrO4− consumption. This observation is consistent with reductions in HCrO4− concentration. Moreover, the results of our simulation demonstrated that K2CrO4(s) and Na2CrO4(s) dissolved into the solution (Fig. S3) resulting in the formation of the considerable amount of CrO42− through reactions (14) and (15), respectively. The dissolution was indicated using the negative concentrations of K2CrO4(s) and Na2CrO4(s). This observation is consistent with the early dramatic increases in concentrations of Na+ (Fig. S4) and K+ (Fig. S5) in the basic solution. However, a significant portion of the produced CrO42− precipitated as (NH 4) 2CrO4(S) and CaCrO4(s) (Fig. S6) given the NH4+ and Ca2+ present in the solution (reactions (12) and (13)), and a small portion of CrO42− precipitated as PbCrO4 and BaCrO4 (Fig. S7) in the presence of Pb2+ and Ba2+ ions (reactions 10 and 11). The increased positive concentrations of (NH4)2CrO4(S), CaCrO4(s), PbCrO4, and BaCrO4 indicate the precipitation of these species (Fig. S6). This precipitation is consistent with the sharp aA þ bB→cC r¼− 1 d½AŠ 1 d ½ BŠ 1 d ½ CŠ ¼− ¼ a dt b dt c dt ð7Þ ð8Þ In Eq. (9), Bdt^ is considered to be as small as possible (10−6 s), thereby allowing the concentrations to be estimated forward in time, according to Euler’s method (Hamming 2012) by repeatedly applying this approximation for each time step interval (dt). To calculate the concentrations of the atmospheric species in the aqueous layer with time, a dynamic mass balance equation for each species was derived taking into account the total consumption or production of a given species through the multiple reactions and equilibria (Eq. (9)). d½AŠ ¼ r 1 þ r2 þ r3 þ … þ rN ; dt ð9Þ The rate expression sign depends on the role of the element in the reaction such that a negative sign refers to the reactants and positive sign refers to the products. In Eq. (9), rN represents the reaction rate of component A in reaction number N in which component A is either a reactant or product. The total reaction rate of each compound was calculated using Eq. (9) which was then multiplied by Bdt^ to obtain its updated concentration. Air Qual Atmos Health Fig. 3 The total ambient soluble Cr(VI) concentration over time (available HCrO4−, Cr2O72−, and CrO42−) Fig. 5 The ambient VO2+ and VO2+ concentration over time decreases in the concentrations of Pb+, NH4+, and Ba+ (Figs. S8, S9, and S10, respectively). Seigneur and Constantinou (1995) found that HCrO4− is a dominant form of soluble Cr(VI) in the PM matrix at pH 4, while in our study CrO42− was found to be the most dominant Cr species at pH 9. PbCrO 4 , BaCrO 4 , CaCrO 4 , and (NH4)2CrO4 are insoluble, while K2CrO4 and NaCrO4 are soluble at pH 9 (the Cr sampling condition) under our simulation condition. Conversion of Cr(VI) to Cr(III) Figure S1 illustrates a sharp decrease in HCrO4− concentration over 24 h. The Cr(VI) conversion to Cr(III) occurs through various reactions in the presence of atmospheric VO2+, Fe(II), and As(III). Figure 5 shows that the major Cr(VI) reduction to Cr(III) is observed in the presence of VO2+ (Table 2, reaction 18). The ambient VO2+ concentration dramatically dropped from 5.8 × 10−5 μg m−3air to 0 after 35 min. The dramatic decrease in VO2+ concentration resulted in the formation of VO2+ from 0 to 6.3 × 10−3 μg m−3air. Moreover, it was found that the formation of S2O62− increased (from 0 to 9.19 × 10−3 μg m−3air (Fig. 6) indicating the conversion of Cr(VI) to Cr(III) (reaction 20 in Table 2). The ambient concentrations of Fe(II) and As(III) did not change as shown in Figs. S11 and S12, respectively, indicating that the basic impregnated cellulose filters can prevent reduction of Cr(VI) in the presence of As(III) and Fe(II). This observation is also understood from the reaction rate expressions for reactions 19 and 21. The reaction rates depend on the H+ concentration. Thus, at basic pH that offers low H+ concentrations, the reaction rate significantly decreases. The HO3 and HO2 concentrations (Figs. S13 and S14, respectively) sharply decreased suggesting the formation of Cr(OH)3 from HCrO4− (reactions 16 and 17). The O3 and O2 concentrations were constant (Figs. S15 and S16, respectively). Cr(III) solution equilibrium Fig. 4 The ambient CrO42− concentration over time Cr(OH)3 dissociated (negative value) into Cr3+ and OH− ions in the aqueous solution (Fig. 7) within 5 ms, approximately. Air Qual Atmos Health Fig. 6 The ambient S2O62− concentration over time However, most of the generated Cr3+ precipitated (positive value) as Cr2(SO4)3 in the presence of sulfate ions consistent with a sharp drop in sulphate concentration (Fig. S17). Figure 8 shows a profound drop in Cr3+ concentration from 2.4 × 10−2 μg m−3air to 2.8 × 10−5 μg m−3air. In conclusion, the majority of soluble Cr3+ precipitated as Cr2(SO4)3 at pH 9. Conversion of Cr(III) to Cr(VI) The conversion of Cr(III) into Cr(VI) can occur through kinetic reaction 22 in the presence of MnO2 oxidant. Fig. S18 Fig. 7 The ambient Cr2(SO4)3 and Cr(OH)3 concentration over time presents the MnO2 concentration which is constant over 24 h, as most of the generated Cr3+ precipitated (positive value) as Cr2(SO4)3. The Mn(II) ambient concentration remained as zero over 24 h suggesting the inert effect of MnO2 on Cr(III) under the current condition. Discussions Cr(VI) exposure and sampling In this study, our estimated ambient Cr(VI) concentration is in good agreement with daily measurements done in industrial locations. For instance, typical ambient Cr(VI) concentrations measured in various industrial locations were 50 to 400 μg m − 3 a i r for stainless steel welding, 100 to 500 μg m−3air for chromate production, 5 to 25 μg m−3air for chrome plating, 10 to 140 μg m−3air for ferrochrome alloy production, and 60 to 600 μg m−3air for chromate pigment production (Khlystov and Ma 2006). In our study, the estimated K2CrO4 and Na2CrO4 concentrations were − 23.85 μg m−3air and − 42.50 μg m−3air at equilibrium respectively (− 66.35 μg m−3air in total). Negative values suggest that at least 23.85 μg m−3air K2CrO4 and 42.50 μg m−3air Na2CrO4 are needed to reach equilibrium. Thus, total ambient soluble Cr(VI) concentration requires to be at least 66.35 μg m−3air which is 2 to 4 orders of magnitude higher than urban areas reported by Amouei Torkmahalleh et al. (2013b). Amouei Torkmahalleh et al. (2013b) reported Cr(III) to Cr(VI) conversions collected by cellulose filters during sampling to be 8.6 ± 0.2% in summer and 8.4 ± 5.9% in winter, respectively. Air Qual Atmos Health Fig. 8 The ambient Cr3+ concentration over time Our study shows that these small conversions of Cr(III) to Cr(VI) could be due to the HCrO4− to Cr(OH)3 conversion. Impregnated cellulose filter When deliquesce continues to happen at RH values beyond the deliquescence relative humidity (DRH), NaHCO3 coated on the filter can dissolve into the deliquesced layer formed around the solid core of the particles establishing the basic pH. As a result, the initial concentration of Na+ can be relatively high (saturated level) in the deliquesced layer due to the unlimited presence of NaHCO3. However, in the current simulation, the initial concentration of Na+ in the solution was set by the Na concentration in ambient PM rather than the NaHCO3 as buffer. To further investigate the Cr chemistry during the sampling, we have conducted further simulation analyses using higher initial concentrations of Na+ in the deliquesced layer. Figs. S19 and Fig. 9 show the concentrations of Na2CrO4 and K2CrO4 at steady state when the initial Na+ concentration was 10 and 100 times higher than the given value in Table 3. As can be seen, when the initial Na+ concentration is 10 times higher, the solution chemistry is in favor of the dissolution of Na2CrO4 similar to the base case simulation. However, when the initial concentration is 100 times higher, the chemistry shifts toward the precipitation of Na2CrO4, while K2CrO4 is being still dissolved into the deliquesced layer. This observation suggests that the dissolution of the bicarbonate sodium into the aqueous layer may significantly increase the Na+ concentration and cause further precipitation of Cr(VI) as Na2CrO4. The chemistry for other compounds including (NH 4 ) 2 CrO 4 , CaCrO 4 , BaCrO 4 , and PbCrO 4 remained independent of the initial Na+ concentration. In conclusion, it is likely that impregnating the cellulose filter adds to the precipitation of Cr(VI) which will then lead to underestimate the soluble Cr(VI) concentrations using typical basic extraction solution. Furthermore, when extraction is applied using a 20 mM bicarbonate solution, it is possible to further precipitate Cr(VI) compounds during the extraction due to the presence of additional bicarbonate sodium in the solution. When extraction is performed, the concentration of species on the sampling filters is diluted while Na+ concentration may increase. A future study is required to quantify the soluble and insoluble compounds of Cr(VI) during the filter extraction process. The increasing concentration of Na+ in the deliquesced layer presented in this study, excludes the effect of the NaHCO3 dissolution on the concentration of Na+ in the deliquesced layer. Thus, the real concentration of Na+ must remain unchanged in the solution due to the unlimited presence of the NaHCO3 in the system maintaining its buffer role in the deliquesced layer. Cr(VI) analysis Fig. 9 Ambient air concentrations of Na2CrO4 and K2CrO4 with time when initial Na+ concentration is 100 times higher than the base case simulation A major fraction of Cr(VI) compounds (69%) in bag filter dust particles originating from ferrochrome production were reported to be water insoluble (Du Preez et al. 2017). Huang et al. (2014) found that insoluble Cr(VI) accounts for most of the Cr(VI) in the atmosphere during a sampling campaign in NJ, USA which is consistent with the model prediction in this Air Qual Atmos Health study. Thus, it is likely that the current analytical method that implements ion chromatography and the 20 mM NaHCO3 extraction solution (Amouei Torkmahalleh et al. 2012) only quantifies soluble Cr(VI) at pH 9. Thus, it underestimates the amount of Cr(VI) in the atmosphere as the majority of the Cr(VI) is found to be insoluble. Future studies need to be conducted to develop analytical methods to quantify total Cr(VI) including soluble and insoluble Cr(VI) in ambient PM. Model uncertainty The current model can be further improved by incorporating the speciation and solution chemistry of other metals including As, V, Mn, and Fe. Also, addition of the reaction kinetics and equilibrium constants for the reaction of organic carbons in ambient PM with Cr(VI) that are rarely found in the literature can further improve the model prediction. However, since the chemistry of Cr is mainly driven by its solution chemistry, the addition of the speciation of other metals is unlikely to significantly change the results. These modifications are the subject of our future study. Conclusion It was found that basic solutions on the cellulose filters preserve the conversion of Cr(VI) to Cr(III) and vice versa by preventing the reactions of Fe(II) and As(III) with Cr(VI) and MnO2 with Cr(III), respectively. However, cellulose filters cannot prevent Cr(VI) conversion in the presence of VO2+ and the formation of S2O62−. Moreover, CrO42− was the soluble form of Cr(VI). PbCrO4, BaCrO4, CaCrO4, (NH4)2CrO4 precipitated. Cr3+ was the dominant soluble form of Cr(III), while majority of Cr(III) precipitates during the sampling on the cellulose filters as Cr2(SO4)3. The system reached steady state within 5 ms which is comparable with the time scale of the molecular diffusion within the deliquesced particles. Shah et al. (2017) reported time scale for diffusion of ozone and volatile organic compounds within Cr-containing deliquesced particles to be from 1 ns to 1 ms for particles ranging from 50 nm to 50 μm, respectively. The presence of the NaHCO3 coated to the cellulose filters and also in the extraction solution adds to the insoluble Cr(VI) compounds, which will then cause further errors in the estimation of soluble Cr(VI) concentration using 20 mM bicarbonate sodium extraction solution and IC. These results suggest future studies on Cr(VI) sampling and analyses to improve cellulose sampling filters and the filter extraction solution to prevent Cr(VI) precipitation during the wet analysis of Cr(VI), or improve the Cr(VI) analysis methods to quantify total Cr(VI) (soluble and insoluble Cr(VI)). Funding Information This study was funded by the Nazarbayev University under the Faculty Small Grant (No. 090118FD5315). References Amouei Torkmahalleh M, Lin L, Hoslen TM, Rasmussen DH, Hopke PK (2012) The impact of deliquescence and pH on Cr speciation in ambient PM samples. Aerosol Sci Technol 46(6):690–696 Amouei Torkmahalleh M, Lin L, Hoslen TM, Rasmussen DH, Hopke PK (2013a) Cr speciation changes in the presence of ozone and reactive oxygen species at low relative humidity. Atmos Environ 71:92–94 Amouei Torkmahalleh M, Yu C, Lin L, Fan Z, Swift J, Bonanno L, Rasmussen D, Holsen T, Hopke PK (2013b) Improved atmospheric sampling of hexavalent chromium. J Air Waste Manag Assoc 63: 1313–1323 ATSDR (2000) Final report of toxicological profile for chromium. Agency for Toxic Substantances and Disease Registry, Atlanta Beattie JK, Haight GP (1972) Inorganic reaction mechanisms, Part II. Prog Inorg Chem 17:93–145 Davis A, Oslen RL (1995) The geochemistry of chromium migration and remediation in the subsurface. Groundwater 33:759–768 Du Preez SP, Beukes JP, Van Dalen WPJ, Van Zyl PG, Paktunc D, LoockHattingh MM (2017) Aqueous solubility of Cr(VI) compounds in ferrochrome bag filter dust and the implications thereof. Water SA 43(2):298–309 Eastern Research Group (2007) Collection and analysis of hexavalent chromium in ambient air, Morrisville, NC Espenson JH (1964) Mechanisms of the reaction of vanadium(VI) and Cromium(VI) and of the induced oxidation of iodide ion. J. Am Chem Soc 86:1883–5101 Espenson JH (1970) Rate studies on the primary step of the reduction of chromium (VI) by iron(II). J Am Chem Soc 92:1880–1883 Grohse PM, Hodson WFL, Wilson BM. CARB Contract No. A6-096– 32. Research Triangle Institute (1988) The fate of hexavalent chromium in the atmosphere, California Air Resources Board Sacramento Guertin J, Jacobs JA, Avakian CP (2005) Chromium. Chromium (VI) handbook 103:132–140 Hamming R (2012) Numerical methods for scientists and engineers. Courier Corporation, Chelmsford Huang L, Zhihua F, Chang HY, Hopke PK, Lioy PG, Buckley B, Lin L, Ma Y (2013) Interconversion of chromium species during air sampling: effects of O3, NO2, SO2, particle matrices, temperature and humidity. Environ Sci Technol 47:4408–4415 Huang L, Yu CH, Hopke PK, Shin JY, Fan Z (2014) Trivalent chromium solubility and its influence on quantification of hexavalent chromium in ambient particulate matter using EPA method 6800. J Air Waste Manag Assoc 64:1439–1445 Hummel W, Berner U, Curti E, Pearson FJ, Thoenen T (2002) Nagra/PSI chemical thermodynamic data base 01/01. Radiochim Acta 90:509 IARC (1990) IARC monographs on the evaluation of carcinogenic risks to humans, nickel and welding. Food Chem Toxicol 49:257–445 Khan MF, Hwa SW, Hou LC, Mustaffa NIH, Amil N, Mohamad N, Sahani M, Jaafar SA, Nadzir MSM, Latif MT (2017) Influences of inorganic and polycyclic aromatic hydrocarbons on the sources of PM2.5 in the southeast Asian urban sites. Air Qual Atmos Health 8: 999–1013 Khlystov A, Ma Y (2006) An on-line instrument for mobile measurements of the spatial variability of hexavalent and trivalent chromium in urban air. Atmos Environ 40:8088–8093 Kotas J, Stasicka Z (2000) Chromium occurrence in the environment and methods of its speciation. Environ Pollut 107:263–283 Nico PS, Zasoski RJ (2000) Importance of Mn (II) availability of Cr (III) oxidation on birnessite. Environ Sci Technol 38:5253–5260 Air Qual Atmos Health NJDEPA (2012) New Jersey chromium emission inventory. New Jersey Department of Environmental Protection Agency, Trenton Palmer CD, Puls RW (1994) Natural attenuation of hexavalent chromium in groundwater and soils. US Environmental Protection Agency, Washington, DC Park RM, Bena JF, Stayner LT, Smith RJ, Gibb HJ, Lees PS (2004) Hexavalent chromium and lung cancer in the chromate industry: a quantitative risk assessment. Risk Anal 24:1099–1108 Pettine M, D’Ottone L, Campanella L, Millero FJ, Passino R (1998) The reduction of chromium (VI) by iron (II) in aqueous solution. Geochim Cosmochim Acta 62:1509–1519 Rai D, Sass BM, Moore DA (1987) Chromium (III) hydrolysis constants and solubility of chromium (III) hydroxide. Inorg Chem 26:345–349 Rogula-Kozlowska W (2016) Size-segregated urban particulate matter: mass closure, chemical composition, and primary and secondary matter content. Air Qual Atmos Health 9:533–550 Sadiq M (1992) Toxicometal chemistry in marine environments. Marcel Dekker Inc, New York Schroeder DC, Lee GF (1975) Potential transformations of chromium in natural waters. Water Air Soil Pollut 4:355–365 Seigneur C (1985) A theoretical study of the atmospheric chemistry of chromium. Environmental Protection Agency. Office of Air Quality Planning and Standards, Durham Seigneur C, Constantinou E (1995) Chemical kinetic mechanism for atmospheric chromium. Environ Sci Technol 29:222–231 Shah D, Aldamzharov B, Bukayeva B, Amouei Torkmahalleh M, Ahmadi G (2017) Intermolecular interactions and its effect within Cr3+-containing atmospheric particulate matter using molecular dynamics simulations. Atmos Environ 166:334–339 Sillen GL, Martell AE (1964) Stability constants of metal ion complexes. Special Publication, London Terraglio FP, Manganelli RM (1967) The absorption of atmospheric sulfur dioxide by water solutions. J Air Pollut Control Assoc 17:403– 406 Tirez K, Silversmit G, Bleux N, Adriaensens E, Roekens E, Servaes K, Vanhoof C, Vineze L, Berghmans P (2011) Determination of hexavalent chromium in ambient air: a story of method induced Cr (III) oxidation. Atmos Environ 45:5332–5341 Werner M, Nico P, Guo B, Kennedy I, Anastasio C (2006) Laboratory study of simulated atmospheric transformations of chromium in ultrafine combustion aerosol particles. Aerosol Sci Technol 40:545–556