Particles collected above the salt lakes with the Sioutas impactor and open-faced filter holders were analyzed subsequently for their chemical composition to validate the potential formation pathways presented in section 3.2 and in Kamilli et al. (2015b). Impactor samples in the size range between 250 nm and 10 µm were examined using chemical imaging and electron microscopy by a combination of Raman microscopy and scanning electron microscopy with energy-dispersive x-ray spectroscopy (SEM-EDX). Results obtained from these methods are not discussed here, but can be found in Kamilli et al. (2015a, b). The following discussion is based on results obtained from ultrahigh resolution mass spectrometry.
The loaded filters from the open filter holders were analyzed with ultrahigh resolution mass spectrometry without size segregation (Kamilli et al., 2015a, b). Due to limitations in the sampling method chemical information about the freshly nucleated particles were not in the direct focus of examination, but conclusions about the formation may be drawn anyway. The filters were analyzed for (oxygenated) organics (CHO), chlorinated organics (CHOCl), sulfur-containing organics (CHOS) and nitrogen-containing organics (CHON) for field and lab samples, respectively. In the van Krevelen diagrams (Fig. 6a, b left panel) the H/C and the O/C ratios for the identified compounds are shown. The associated average carbon oxidation state (OSc) dependent on the carbon number (nc) are depicted in Fig. 6a, b in the right panels.
Figure 6 Van Krevelen diagrams of the ultrahigh resolution mass spectra (a, b left), and corresponding carbon oxidation states (a, b right) of filter samples taken during a lab experiment with the standard Fe (II) concentration (a) and in a field experiment at Lake Shot (b). The bubble size is proportional to the signal intensity in the mass spectra and the color code indicates the composition. The comparison of field and lab experiments regarding the number and total amount of compounds in percent is depicted in c).
For comparison, one lab sample with the normal amount of FeII in the salt lake mixture (Kamilli et al., 2015b), which is comparable to natural salt lake composition, is contrasted to a field sample taken at Lake Shot. Figure 6c shows the number of individual organic compounds found in both samples. It is obvious that in the lab sample, corresponding to a simulated salt lake, more individual compounds have been found, namely 2400 in contrast to about 850. Moreover, the compounds are more distributed in the
van Krevelen diagrams, particular the whole range of O/C ratios. More CHO compounds are identified at high H/C and low O/C ratio in the field than in the lab (Fig. 6a, b; left). Whereas in the field CHO compounds represent 2/3 of the total amount of organic compounds, they are about half of the compounds in the lab.
The reason is the higher number of other compounds formed in the lab. The CHOS compounds with an O/C ratio below 0.4 are indeed similarly distributed in both samples, but in the lab CHOS compounds with higher O/C ratios are found. In contrast to the field sample, the lab sample exhibits not only more long-chained CHO compounds, but also those compounds might be oxidized by sulfur more often resulting in a higher number of CHOS compounds in the lab. An additional group of compounds nearly absent in the field are CHONS. Even if they are identified with low intensities in the lab, they are evenly distributed between O/C ratio 0.1 and 0.3 in the lab sample. A total amount of 4 % in the lab is contrasted with 0.1 % of CHONS compounds in the field. Also, in the lab sample CHON are identified more frequently and with higher intensities in the van Krevelen diagrams (Fig. 6a, b; left). The oxidation of CHO compounds by nitrogen appears to occur more often in the lab, as CHON compounds are found at an OSc between -2 and 0 in a wide range of carbon numbers in the lab sample compared to only one distinct group of CHON compounds with 17 carbon atoms in the field sample (Fig. 6b; right). CHOCl is visible in the van Krevelen diagrams.
Obviously, the oxidation by halogens occurs more often in the field than in the lab, since CHOCl compounds are more abundant in the field sample.
Examinations in laboratory experiments above a simulated salt lake mixture by Kamilli et al. (2015b) identified that the Fenton reaction has a direct effect on the formation of sulfur-containing and halogenated organic compounds, which is in agreement with section 3.2. Without iron-addition, the ultrahigh resolution mass spectra identified less chemical diversity in the aerosol particles (Fig. 7a, b; C3). With FeII in the salt lake mixture the amount of CHOS and halogenated organic compounds increased, respectively (Fig. 7a, b; C1). By enhancing the FeII concentration, the relative abundance of CHOS and halogenated organic compounds increased further, while the relative abundance of CHO compounds decreased (Fig. 7a, b; C2).
Kamilli et al. (2015b) suggested that the bond of FeII to the organic matter in the aqueous phase results in less release of organic compounds into the gas phase, and therefore leads to reduced formation of particulate matter with increasing FeII concentration. This assumption is supported by the observed complexing of organic compounds by irradiated iron oxide in Voelker et al. (1997) and the chelating properties of FeII and 1,8-cineole-containing oils (Joshi et al., 2008).
While halogenated organic compounds were not only identified in the aerosol samples with ultrahigh resolution mass spectrometry, but also with SEM-EDX, Raman spectroscopy in both the field and lab samples, a mixture of (oxygenated) organic, sulfur- and nitrogen-containing organic compounds seems to play a major role in new particle formation in the salt lake environments. Beside the important effect of organic species in NPF, the interplay with inorganically dominated coarse mode particles by coating of the salts could also be shown with Raman microscopy (Kamilli et al., 2015a).
Figure 7 The increasing chemical diversity with increasing FeII concentration (C3 → C1 → C2) in the aerosol samples is depicted in a) Van Krevelen diagrams derived by ultrahigh resolution mass spectrometry and is summarized in b).
The color code indicates the type of compositions and the bubble size corresponds to the signal intensity in the mass spectra. Kamilli et al. (2015b).