Changes in Chemical Composition of Dissolved Organic Matter in Solar Ponds From Salt Lake Brine

The abundance and chemical composition of dissolved organic matter (DOM) in the brine of solar ponds inuence eciency of mineral extraction and rates of brine evaporation, and cause undesired odor and color of the products. In this paper, we report an investigation on changes of DOM compositions in solar ponds from salt lake brine through different approaches. The results showed that the DOM was primarily composed of carbohydrates, aliphatic and aromatic compounds. Dissolved organic carbon (DOC) analysis revealed that the concentration of DOC in solar pond increased with exposure time, and up to 15−fold upon evaporation/irradiation of salt lake brine. Analyses with the elemental composition, Fourier transform infrared spectroscopy (FTIR) and cross polarization magic angle spinning (CP/MAS) 13 C nuclear magnetic resonance (NMR) indicated that the relative abundance of aliphatic compounds (including functionalized ones) increased in solar pond process, while an opposite phenomenon was observed for carboxylic acid moieties, aromatics and carbohydrates. Pyrolysis−gas chromatography−mass spectrometry (Py−GC−MS) revealed that most of the DOM in salt lake brine contained methylene chain, terpenoid-like, carbohydrate and/or aromatic structures. The presence of some sulfur-containing organics implied some anaerobic biotic decays, but microbiological processes were probably subordinate to photo-induced DOM transformations. In the salt lake brine, exposure-driven decay decreased the abundance of polysaccharides and increased that of mono- and polyaromatic pyrolysis products. Finally, the implications and guidelines for removing DOM from brine in the process of brine resource exploitation were discussed.


Introduction
A solar pond is a system of trapping and storing solar energy for extensive applications, such as heating of buildings (Shah et al., 1981;Newell et al., 1990), desalination (Hull and Nielsen, 1989), power production (Tabor and Doron, 1990; Sherman and Imberger, 1991), the addition of heat from external sources (Ganguly et al., 2018) and hydrogen production (Karakilcik et al., 2018). Also, several studies have been made in mineral processing with solar pond technology to achieve the temperature requirements of the otation stage in last decades (Nie et  Brine water usually contains raw materials such as potassium chloride, lithium carbonate, boron, sodium and magnesium for mineral extraction (Zheng, 2011;Bian et al., 2017;Peng et al., 2017). Due to the technically and economically viable, solar pond has been used successfully in mineral extraction from brine water (Nie et al., 2011;Yu et al., 2015). Dissolved organic matter (DOM), the most active fraction in solar pond, often has adverse effects in the process of mineral extraction related to an evaporation rate and a degree of brine evaporation, and further affects quality of products (e.g. smell, color and purity) due to its absorption for most of the sunlight as well as its complexed behaviors with inorganic components (Scully and Lean, 1994;Zhao, 2008Zhao, , 2010Shalev et al., 2018). To more e ciently utilize brine resources, it is important to remove their DOM to meet market demands for high value products nowadays such as battery-grade lithium carbonate.
Various physical, chemical and biological technologies have been proposed for DOM removal from brine. Zhao (2010) used a modi ed ultra ltration membrane to remove more than 90% of the total organic carbon (TOC) of salt lake brine, and generated a superior grade of BaSO 4 . A hydrogen peroxide oxidation method (Zhang et al., 2008) allowed for a removal of colored DOM in salt lake brine from the Yuncheng. This method yielded 60% removal e ciency and the whiteness of products was obviously enhanced. Nevertheless, treatments targeting a DOM removal often fail to produce a desired result. This is mainly attributed to the presence of an inherently recalcitrant DOM fraction. Research showed that aromatic, humi ed and higher molecular weight fractions of DOM were more e ciently removed by photochemical degradation than by biological treatments (Xu et al., 2018a), while the reverse was true for carbohydrates (Xu et al., 2016). Moreover, chemical oxidation methods were not suitable for alkane series (Peter and Yoshimoto, 2018), and it was di cult to remove low molecular weight DOM by membrane technology. Thus, knowledge on the chemical structures of brine DOM is required to develop treatment strategies for an e cient removal of DOM.
The Qinghai − Tibet Plateau is the largest plateau in China and the highest plateau in the world, and mainly consists of east − west oriented mountain range, wide valleys, and at highland > 4000 m altitude (Zheng, 2011). Salt lake brine is widespread this region. The brine is rich in mineral resources (e.g. halite, potash and borate) and some of them reach the cut-off grade of exploitation. Due to lack of mineral energy, construction of solar pond has become a widely used technique in the production of minerals from brine for their capacity to capture and store solar energy, especially in Qinghai − Tibet Plateau region, with excellent solar energy resources (Nie et al., 2011;Yu et al., 2015), which provides a partial solution to mineral energy shortages. In addition to lithium, boron, potassium, bromide and other useful trace elements that are of great economic value, DOM are also an important part of brine. The chemical evolution of different types of inorganic constituents has been extensively investigated (Cui et al., 2015;Li et al., 2018;Zhang et al., 2018). However, a lack of knowledge on the composition, distribution and variation of DOM in this system negatively affects further development of mineral extraction strategies.
In this study, the changes of chemical composition of DOM derived from solar ponds were investigated using Da Qaidam Salt Lake brine, located in Qaidam basin, Qinghai − Tibet Plateau region. Solid-phase extraction with PPL cartridges, which reportedly produced high DOC recovery and representative DOM fractions (Yang et al., 2017a;Yang et al., 2017b), was used to isolate DOM from the hypersaline water. The DOM was characterized using elemental composition, Fourier transform infrared spectroscopy (FTIR), 13 C nuclear magnetic resonance spectrometry ( 13 C NMR) and pyrolysis-GC − MS (Py − GC − MS). This work is an initial step in extending our knowledge on the roles of DOM played in solar ponds, which is also essential to guide development of treatment strategies for these brines.

Sites description and sampling
The brine of salt lake, located in the central of Qaidam basin, Qinghai−Tibet plateau, contains huge amounts of useful mineral elements (e.g. potassium, lithium, boron, sodium and magnesium), which have been well described in our previous studies (Yang et al., 2017a). Due to the shortage of fossil fuel in this remote area, solar pond technology is the main solution to separate and extract minerals from brine. Industrial mineral exploitation in Da Qaidam Salt Lake is mainly conducted by the mining company Dahua Industry Company Ltd., which has built a production plant and has produced potassium chloride, potassium sulfate and lithium chloride since 2003.
The solar ponds were constructed near the salt lake (Da Qaidam Salt Lake, 37°50′50″N 95°14′42″E). The solar pond of Da Qaidam Salt Lake ( Fig. 1), used to obtain primary products of potash, includes the initial brine pond (SIP), sodium pond (SSP), regulation pond (SRP), carnallite pond (SCP, the main pond for primary commodity extraction) and bittern pond (SBP). The brine samples of different stage of solar pond were collected in two periods (2017 and 2018). Each sample was ltered immediately with 0.7 μm GF/F glass ber (Whatman, pre-combusted at 450 ℃ for 5 h) in the eld. Subsequently, the <0.7 μm samples were covered with ice packs in dark, and transported as soon as possible to lab for DOM isolation and puri cation.

Physical and chemical characterization
The dissolved fraction of total, inorganic, and organic carbon concentrations of the brine samples were determined using a total organic carbon analyzer (AnalytikJena multi N/C 3100, Germany) with a high-temperature catalytic oxidation method. The contents of major cations and anions such as Na + Mg 2+ and Clwere determined by traditional titration methods. The concentration of Li + analyzed performed on inductively coupled plasma optical emission spectroscopy (ICAP 6500 Duo, Thermo Scienti c). All the mentioned data were obtained by three analytical replicates.

Isolation and puri cation of DOM
The DOM samples were isolated and puri ed following a procedure established in our previous studies (Yang et al., 2017a;Yang et al., 2017b). Brie y, the PPL cartridges (Agilent Technologies, USA) was activated with methanol. Next, aliquots of acidi ed samples (pH 2) were passed through the cartridges using a pump (Gast Company, USA) at a ow rate of 3 mL/min. Before elution, the cartridges were rinsed with 0.01 mol/L HCl to remove excess salts. The DOM were eluted with methanol, and were dried and stored in a desiccator for further analysis. Blank controls were conducted with Milli-Q water acidi ed using HCl (pH 2) as the same procedure.

Elemental analysis
The contents of C, H and N of the puri ed DOM samples was investigated in duplicate performed on an elemental analyzer equipment (EL CUBE, Germany). The percentage of oxygen was calculated by subtracting the relative contents of C, H and N from 100%, and trace fractions of S and P were ignored herein. The carbon stable isotopic determination (δ 13 C) was determined using a Delta plus XP isotope ratio mass spectrometer (Thermo Finnigan, USA), and its equation following as: in which R is the ratio of 13 C/ 12 C.

FTIR analysis
The spectra of FTIR were obtained using a FTIR spectrometer (Thermo Nicolet NEXUS 670, USA) from 400 to 4000 cm -1 with 64 scans. Pellets were prepared by pressing sample and KBr (spectroscopy grade, 1:100) under vacuum. Prior to analysis, the FTIR spectra and second-derivative spectra were normalized, and the reconstructed data matrix was then progressed using the drawing software.

Solid-state 13 C NMR analysis
Cross polarization/magic angle spinning solid (CP/MAS) nuclear magnetic resonance ( 13 C NMR) spectroscopy were recorded on a Bruker Advance DSX 300 MHz, and rotor spin speed at 4500 Hz; the optimum relaxation delay was 3 s, and the contact time was 2 ms.

Pyrolysis (Py−GC−MS) analysis
The pyrolysis of DOM samples were conducted on a pyrolysis instrument (Py, CDS Pyroprobe 2000, USA), which heated by 5 • C/ms from 250 to 610 • C and then kept at 610 • C for 10s. Pyrolysis products were separated and detected on a GC−MS (QP2010, Shimadzu Corporation, Japan). The split mode was 1:40; helium carrier gas speed was 1 mL/min; and the capillary column was HP-5MS (length 30 m× thickness 0.25 mm× diameter 0.25 μm). The MS detector was selected at 70 eV electron ionization, and the m/z scan ranged from 45 to 650.

Physical and chemical characterization
The pH, total dissolved solid (TDS) and major inorganic content of the ve samples were listed in Table 1. The pH was slightly basic in the initial stages of solar pond (7.4 in SIP and 7.2 in SSP) and mildly acidic in the late stages of solar pond samples (6.6 in SRP, 4.6 in SCP, and 4.6 in SBP). The contents of TDS ranged from 273.5 to 479.6 g/L, which increased 1.7 times resulted from the chemical evolution of brine along the course of natural evaporation] with solar pond technology. The dominated inorganic constituents were Li + Na + Mg 2+ K + Cl − B 2 O 3 and SO 4 2− in the salt lake brine, however, the varied trends differed. The  The contents of dissolved of total, inorganic, organic carbon and total nitrogen (TN) were listed in Table 1. The dissolved carbon (DC) concentrations exhibited an increasing trend which elevated from 57.0 (SIP) to 349.3 g/L (SBP), while dissolved inorganic carbon (DIC) has no systematic trend which ranged from 8.3 (SCP) to 40.9 g/L (SSP). The DOC and TDS obviously increased (r = 0.97, P < 0.01) with exposure time increasing. This implied that the DOM loss by degradation was compensated by an increased concentration of DOM due to evaporation in the solar pond and/or microbial growth. The DOC and total TN contents were strongly correlated (r = 0.99), suggesting that the vast majority of the N existed in an organic form.

Elemental analysis
The values of elemental compositions, atomic ratios (H/C, O/C and N/C), and δ 13 C of the puri ed DOM samples derived from solar pond in this study were listed in   and 190-220 ppm (carbonyl C). The corresponding normalized distributions of these regions were estimated using the instrument's software (Table 3).  (Table 3). In general, the aliphatic C values and methoxyl/N-alkyl C showed an increasing trend in the solar pond during an evaporation/irradiation treatment (r = 0.91; P < 0.001), while the O-alkyl C (r = − 0.84, P < 0.001), total aromatics (r = − 0.79; P < 0.05) and carboxyl/amide C (r = − 0.77; P < 0.001) exhibited an opposite trend. The results are well agreed with the observation of elemental composition and FTIR analysis.  (Table S2). The products were grouped into 11 subclasses based on chemical structures (Table 4). Remaining compounds, including unidenti ed ones, were labelled as "other compounds".  On the whole, the most abundant group (24.99 − 28.08%) was 1-ring unsaturated alicyclic compounds (ALICYCL) based on cyclopentene and cyclohex(adi)enes groups such as cyclohexadiene, C 1 -cyclohexene, C 1 -cyclohexadienes (4 isomers), C 2 -cyclohexadienes (10 isomers), C 3 -cyclohexadienes (7 isomers) and C 4 -cyclohexene (limonene). They represented a cyclic aliphatic component of the DOM. They could be terpenoid-derived, and shared similar features to the DOM from the Great Salt Lake (Leenheer et al., 2004). It is argued that the terpenoid-like DOM was probably degradation products of algal and bacterial precursors (pigments and steroids therein). Poorly-identi ed terpenoid-like structures were often found in the DOM sample, sometimes they contained abundant carboxylic groups. Such carboxylic groups could be present in the source of the alicyclic compounds, and have been eliminated during analytical pyrolysis (decarboxylation).
The The (linear) n-alkanes, n-alkenes, isoprenoid alkanes and alkenes with a chain length ranging from C 6 to C 23 accounted for 11.21 − 15.69%. They were grouped as methylene chain compounds (MCC). Other MCC such as fatty acids, methylketones, alkylnitriles or alkylamides were not detected. These patterns were probably related to aliphatic material in microbial sources. Lack of phytadienes excluded a signi cant source in fresh phytoplankton, but intense photo-oxidation could e ciently eliminate such moieties. The isoprenoid compounds included C 9 -alkadiene (diemethylheptadiene compound), a C 12 -isoprenoid alkanone and the other unidenti ed isoprenoid ketones. These were uncommon pyrolysis products and highlighted the idiosyncratic nature of the DOM in hypersaline water. They might originate either from a speci c kind of diatom or bacteria, or represent oxidation products of any isoprenoid MCC precursors.
The detected phenols (PHEN) including phenol, methylphenols, C 2 -alkylphenols (4 isomers) and a C 3 -alkylphenol accounted for 3.28 − 3.88%. These compounds could originate from many sources such as microbial ones. A partial confusion with C 9isoprenoid alkatriene groups (m/z 107 and 122) for the dimethylphenols could not be excluded.
The sum of compounds with a halogen atom (HALOGENCOMP) in its structure (0.47 − 1.65 %) was detected in our study. The compounds included bromomethane (MeBr) and iodomethane (MeI), but other organohalogens could be among the unidenti ed products. Polycyclic aromatic hydrocarbons (PAHs) were mostly indenes (C 0 -C 3 -alkylindenes) and naphthalenes (C 0 -C 2 ). The extensive methylation was probably indicative of a terpenoid-like or asphaltene source, even though a contamination from sample treatment could not be excluded.
Compounds with nitrogen (NCOMP) in their structures were scarce (1.43 − 1.84%) and dominated by pyrroles (N-methylpyrrole, 2methylpyrrole and 3-methylpyrrole), with traces of benzonitrile and C 3 /C 4 -alkylanilines. These compounds should probably be ascribed to microbial DOM. The lack of indoles, acetamides, cyanobenzenes and diketopiperazines indicated that the N-rich DOM was strongly affected by decay processes, as intact proteins, chitins or peptidoglycans would generate a more diverse N-product ngerprint upon Py − GC − MS. This decay process could also explain the rather low abundance of organic N in spite of a prevailing microbial source of the DOM: N-rich biopolymers tended to be relatively labile components. The only sulphur-containing products (SCOMP) were benzothiazole, S 2 and/or SO 2 (0.66 − 2.19%). Identi cation of thiophene was tentative and was added to the "other compounds".
A peak with m/z 81, 109 and 124 at the expected retention time of guaiacol was the only possible sign of lignin (LIG), but a source in an isoprenoid alkadiene or trimethylcyclopentenone could not be excluded. Finally, the other compounds (4.89 − 6.20%) included numerous unidenti ed peaks, peaks from plasticizers (phthalic anhydride), C 0 -C 3 alkylbenzofurans and a dioxane.
In summary, there were several classes of compounds with an aromatic character (MAH, PAH, PHEN, benzofurans), others with a cyclic aliphatic character (alicyclic compounds, carbohydrate products) and compounds with an acyclic structure (linear alkanes and alkenes, isoprenoid alkanes and alkenes, and pentanoic acid derivatives). Remaining compounds contained atoms such halogen, sulphur or nitrogen.

Comparison of pyrolysis products of DOM with different origins
In our previous studies, the DOM samples isolated from solar ponds of oil eld-produced brine (OSDOM) had been investigated by Py − GC − MS analysis (Table S1,  Da Qaidam Salt Lake, was of modern age (kerogen could be a source of the defunctionalized terpenoid structures, but this was not the case). We believe the DOM studied here was also of recent, but nevertheless strongly degraded source.
The same phenomenon was observed for the polysaccharides: they were slightly more abundant in the salt lake brine and were preferentially degraded in the solar pond process. The difference with the alicyclic aliphatic compounds was that, for the carbohydrate products, a decrease due to exposure was more pronounced in the salt lake brine. Hence, the degradation occurring in the salt lake brine eliminated polysaccharides, whereas in the oil eld-produced brine the terpenoid-like component of the DOM was more severely affected. The results were in agreement with those of NMR and FTIR analyses.
Oxygen lacking aromatic DOM products (MAHs and PAHs) were more abundant in the salt lake brine than in the oil eld-produced brine, and the opposite was observed for the O-containing aromatic compounds (in particular, phenols). Their abundance did not change signi cantly with sunshine exposure time, suggesting that these compounds re ected DOM types of intermediate degradability. Nevertheless, they appeared to correspond to DOM of relatively recalcitrant nature in the salt lake brines (increased in nal exposure phases). If the oil eld-produced brine contained signi cant amounts of petroleum-derived DOM, we would expect higher PAH levels than in the salt lake brine. Hence, it seems that the oil eld-produced brine DOM did not contain oil-derived DOM.
Again, a source in microbial terpenoid-like structures was one of the more plausible explanations of the presence of the poly-alkyl substituted PAHs.
The value of the MAHs and PAHs as indications of aromatic DOM in the brine samples was compromised by their possible release from the exchange resins (Daignault et al., 1988). On the other hand, they have often been identi ed in DOM from (hyper) saline lakes, and are ubiquitous products of pyrogenic and geological OM (kerogen) as well (Witter and Jones, 1999). Many kinds of su ciently degraded sources of organic sources, including humic acids, could produce aromatic groups lacking oxygen after pyrolysis (Song and Peng, 2010).
Chitins (e.g. from brine shrimps) and peptidoglycans (bacterial cell walls) were not recognized. If the DOM were indeed largely bacterial/planktonic, the relatively low nitrogen content (or low abundance of N-containing pyrolysis products), in comparison with fresh aquatic microbial OM (C/N < 10), could be explained by preferential decay of proteinaceous groups.
The organohalogen MeI (methyliodide) was much more abundant in the OSDOM than SSDOM and vice versa for MeBr. It also seemed to con rm the high halogen content in the oil eld-produced brine (Zhang, 1987;Tan H et al., 2007).

Implications
This study showed the change of the DOM in the solar ponds, which could be used to evaluate their in uences on the cycling of local carbon and nitrogen. Furthermore, the characters of autochthonous-DOM sources in salt lake brine, may give rise to a high potential risk in complexing with organic or heavy metal ion pollutants (Xu et al., 2018b), especially in the salt lake with a high concentration as found in salt lake in Tibet region. This study contributed to a suitable management of brine resources and in particular showed important aspects of the DOM composition and their changes during application of solar pond technology.
Identi cation of DOM in salt lake brine was the rst step to develop methods of a separation or removal of brine DOM.
It appeared that the DOC concentration increased during sunlight exposure and that the DOM composition of both brine sources also changed in the solar ponds, probably due to a combination of photodegradation and biodegradation. The FTIR spectra of DOM samples isolated from solar pond process (a), and their corresponding second derivative spectra (b).