Book
Recent Lore
Lore with Images
Search
World Map

Geochemistry of Surface Waters Around Four Hard-Rock Lithium Deposits in Central Europe

Jonas Toupal1,*, David R. Vann1, Chen Zhu2 and Reto Gieré1,3

1 Department of Earth and Environmental Science, University of Pennsylvania, Philadelphia, PA, 19104-6316, USA

2 Department of Geological Sciences, Indiana University, Bloomington, IN, 47405, USA

3 Center of Excellence in Environmental Toxicology, University of Pennsylvania, Philadelphia, PA, 19104-6316, USA

* Correspondence: toupal@sas.upenn.edu

Disclaimer: This is a not-the-most-up-to-date manuscript of the work and not formatted perfectly. Additionally, figures are not included. The most up to date and correct version is found at here.

Abstract:

There are several Li-mica and spodumene deposits in Central Europe. The Cínovec deposit on the border of the Czech Republic and Germany is the largest known hard-rock Li deposit in Europe and is currently being explored. Weathering and mining of such deposits could release Li and F to the environment. Both elements are associated with potential health effects when ingested via drinking water, yet little is known about the aqueous geochemistry of streams and creeks around Li hard-rock deposits. In this study, we sampled surface waters (n = 47) near three Li-mica and one spodumene deposit to examine a potential public health risk resulting from the weathering of Li-minerals. At the Cínovec and Homolka (also in the Czech Republic) sites, several water samples have elevated Li contents relative to those in typical surface waters, but they are all below the U.S. EPA-recommended threshold of 0.7 mg/L. Three of the ten samples from Cínovec contain F in concentrations that are above World Health Organization’s 1.5 mg/L drinking water limit, and thus these waters (max. concentration: 3.8 mg/L) may represent a public health hazard. Based on geochemical modeling and statistical analysis, the lithium in the water comes from Li-micas and the fluoride is derived from a combination of Li-micas and fluorite, depending on the specific site analyzed. Lithium seems to be scavenged by clay minerals. Keywords: Lithium; zinnwaldite; spodumene; fluorosis; PHREEQC; mining.

1. Introduction

Lithium (Li) is a strategic metal, important to the current and future energy infrastructure. Lithium-ion batteries are utilized in mobile phones, computers, and electric vehicles (EVs), as well as for electric energy storage technologies, such as large-scale batteries to store electricity produced by solar and wind power plants (Li et al., 2019; Liu et al., 2019). The use of these appliances and devices has increased significantly in the past years and is likely to keep boosting the demand for Li in the near future (Greim et al., 2020; Martin et al., 2017). There are other applications for Li beyond electronics: it is incorporated, for example, in glass, ceramics, glass ceramics, lubricating greases, pharmaceuticals (to treat bipolar disorder, manic depression, and Wilson’s disease; (Loganathan et al., 2008; Mayo Clinic, n.d.; Swain, 2017)), air treatment, polymer production, and aluminum production, amongst others (e.g., (Bradley et al., 2017a; Martin et al., 2017)). Additionally, the nuclear sector uses Li to produce tritium (3H), a fuel for nuclear fusion. Even if the success of EVs penetrating the market is limited, Li might still be a sought-after metal in the future should nuclear fusion become economically viable (Kovari et al., 2018). It has also been suggested that liquid Li may keep the fusion process stable, which would create immense demand for Li (Ono et al., 2017). The amount of Li these various sectors consume today is not changing dramatically, except for the battery industry. In 2014, rechargeable batteries were responsible for only about 29 wt.% of the Li produced worldwide, whereas glass and ceramics accounted for 35 wt.% (Bradley et al., 2017b); in 2019, however, batteries consumed 65 wt.% of the Li produced, and glass and ceramics only 18 wt.%, reflecting the increase in popularity of portable electronics and EVs (USGS, 2020).

The majority of Li (58% by weight) is supplied from brine deposits located in the so-called “ABC triangle” (the border area of Argentina, Bolivia, and Chile), in SW China (e.g., Zhabuye Lake, Tibet), and in the southwestern USA (e.g., Clayton Valley, NV) (Bradley et al., 2017a; Swain, 2017). Even though production of Li from brine deposits is, under current economic and technological conditions, relatively cheap (approximately 50% cheaper than production from hard rocks), interest is now also directed at deposits hosted by pegmatites and granites (Flexer et al., 2018; USGS, 2015). In Europe, for example, there is a strong momentum to create a vertically integrated EV supply chain, from mining to battery manufacture to vehicle assembly, as evidenced by the latest ERA-MIN3 (ERA-MIN3, 2021, p. 3) project or in other publications (Gourcerol et al., 2019; Lebedeva et al., 2017). With no brine deposits on the European continent, the hard-rock deposits are being considered for exploitation (Gourcerol et al., 2019). Several Li-bearing minerals of economic value occur in hard-rock deposits, with the most important being: the Li-pyroxene spodumene (LiAlSi2O6), the phyllosilicates petalite (LiAlSi4O10), “lepidolite” (KLi2Al(Si4O10)(F,OH)2) and “zinnwaldite” (KLiFe2+Al(AlSi3O10)F1.5(OH)0.5), and the phosphate amblygonite (Li,Na)AlPO4(F,OH). The terms “lepidolite” and “zinnwaldite” are not valid mineral species, but rather refer to a series of compositions ranging between the species polylithionite (KLi2Al(Si4O10)(F,OH)2) and trilithionite (K(Li1.5Al1.5)(AlSi3O10)(F,OH)2) and between siderophyllite (KFe2+2Al(Al2Si2O10)(OH)2) and polylithionite, respectively. Most current operations focus on obtaining Li from spodumene (e.g., Greenbushes Deposit, Western Australia; Pervomaisky Mine, Southeastern Russia; Tanco Mine, Canada; (MIR, 2008)). Other mineral deposits currently explored include a zinnwaldite deposit in Cínovec, Czech Republic and a lepidolite deposit in Central Vietnam (Breiter et al., 2019; Hien-Dinh et al., 2017). The Cínovec site is unusual, as historically, it was mined for tin (“cín” in Czech) and tungsten, and many underground tunnels persist. The first records of mining in this area are from the second half of the 15th century, and the oldest known underground shaft dates to 1686 (Richter, 2018). Mining on the German side was halted after World War II, on the Czech side in the year 1990 (Richter, 2018). During the previous mining activities Li was not a valuable commodity, but now, companies are considering re-opening the mining operations specifically for the extraction of this metal (European Metals, 2019).

Information on the environmental impacts of obtaining Li from granite and pegmatite deposits is relatively scarce (Bradley et al., 2017a). Landscape alteration, such as open pit mines, and related silicate dust plumes are major physical impacts, typical of any type of surface mining. Some projects, such as the exploitation of the Cínovec deposit, are proposed as underground operations (European Metals, 2019). In both situations, leaching of tailings and wind dispersion of tailing dust will be a major environmental and health concern. Leaching of both tailings and ore-containing rocks, induced or facilitated by weathering and mining, is likely to impact water geochemistry. For example, a water quality study of tailing ponds around mined spodumene pegmatites in South Dakota reported pH values of 8 – 10, which were interpreted as reflecting equilibration with orthoclase and plagioclase (Rahn et al., 1996). The authors of this investigation further documented low concentrations of heavy metals, but did not analyze the waters for Li. There are, to the best of our knowledge, only two other studies that examined the geochemistry of waters near Li hard-rock deposits and also included data for Li concentrations. Near a spodumene deposit in Southeastern Ireland, researchers analyzed 115 samples of both surface and ground water (Kavanagh et al., 2017). The Li concentration in the surface waters ranged up to 91 µg/L, with an average of 20 µg/L, whereas that in the groundwaters ranged up to 97 µg/L, with an average of 23 µg/L. These scientists further reported that in general, typical Li concentrations in fresh waters are between 1 and 20 µg/L(Kavanagh et al., 2017), consistent with the global riverine mean of 1.9 µg/L (Huh et al., 1998). In the second study, researchers focusing on the environmental impacts of the C57 Gonçalo spodumene mine in Portugal (Rodrigues et al., 2019) collected ten water samples with a Li concentration ranging from 6.9 to 74.1 µg/L. They further stated that the typical concentration of Li in freshwaters in Portugal ranges between 1 to 10 µg/L. Both of these studies, thus, show that Li concentration is elevated near spodumene deposits. Another study aimed to quantify the environmental impacts of Li hard-rock mining in the largest Li mine in Asia, the Jiajika mine in China (Gao et al., 2021). This investigation, however, focused on As and some heavy metals, and neither Li nor fluoride (F-) were analyzed. The authors were able to show that the concentrations of the elements of interest in the nearby streams adhered to the World Health Organization (WHO) drinking water limits, except for As, Pb, and Mn in some samples, and that pH was higher and dissolved oxygen lower near tailings (Gao et al., 2021).

Lithium is considered a dangerous metal by the Danish Environmental Protection Agency (Kjølholt et al., 2003) and is listed as such in the Australian Inventory of Chemical Substances (AICS, 2007). The US Environmental Protection Agency (EPA) recommends a threshold of 0.7 mg/L for Li in drinking water based on a Lowest Observable Adverse Effect Level resulting in nephrotoxicity in humans receiving long-term Li therapy for treatment of manic depressive disorders (Tripathi, 2011), but does not define a maximum contaminant level (US EPA, 2021). While Li is used in pharmaceuticals to treat depression, bipolar disorder, and mania (Loganathan et al., 2008; Mayo Clinic, n.d.; Swain, 2017), there is a suite of known side effects resulting from Li intake, including poor memory, unusual weakness, dizziness, slow heartbeat, renal failure, and interference with several biological processes (Mayo Clinic, n.d.; Tanveer et al., 2019; US EPA, 2008). Patients receiving lithium orally take between 300 – 1200 mg of Li two to three times a day (Mayo Clinic, n.d.). Therapeutic serum dosage is on the order of 10 mg/L Li in blood; at 20 mg/L, the concentration can be fatal to humans (Aral and Vecchio-Sadus, 2008). In plants, Li phytotoxicity manifests itself as necrosis, or breakdown of chlorophyll (Tanveer et al., 2019).

Table 1. Average bulk chemical composition and main minerals of the four studied hard-rock deposits (wt.%). Mineral abbreviations are as follows: Ab = albite; Kfs = K-feldspar; Qz = quartz; Znw = zinnwaldite; Tpz = topaz; Fl = fluorite; (Li)-Ms = (Li)-muscovite; Ap = apatite; Amb = amblygonite; Tur = tourmaline; Pl = plagioclase; Spd = spodumene. Cínovec Homolka Podlesí Rappold Complex* Granite Greisen Granite Stock Granite Dike Granite Pegmatite SiO2 73.93 69.25 72.93 73.80 70.10 74.05 TiO2 0.02 0.01 0.04 0.06 0.02 0.02 Al2O3 14.13 13.71 14.80 14.58 15.77 14.64 Fe2O3 0.30 2.76 0.84 0.30 0.11 1.89 FeO 0.64 1.81 0.88 0.67 MgO 0.10 0.02 0.16 0.04 0.02 0.10 MnO 0.08 0.38 0.05 0.02 0.04 0.47 CaO 0.60 0.44 0.50 0.39 0.43 0.51 Li2O 0.21 0.92 0.07 0.13 0.33 0.07 Na2O 3.79 0.08 3.93 3.19 4.10 5.04 K2O 3.04 3.28 4.14 4.46 4.25 1.75 P2O5 0.01 0.02 0.56 0.38 1.03 0.03 F 1.05 4.45 0.34 1.28 1.31 Major Minerals Qz, Ab, Znw Qz, Ab, Znw Qz, Ab, Li-Ms, Or Qz, Ab, Znw, Tpz Qz, Kfs, Ab, Tpz, Znw Kfs, Qz, Pl, Ms, Spd Minor Li-/F-minerals Tpz, Fl Tpz, Fl Tpz, Ap Amb, Tur, Ap Tur, Ap, Amb Tur Source (Breiter et al., 2017) (Breiter et al., 2017) (Breiter and Scharbert, 1995) (Breiter, 2002) (Breiter, 2002) (Knoll et al., 2018)

*Sankt Radegund bei Graz pegmatites are part of the Rappold Complex, but no geochemical composition of those specific outcrops was available. Aral and Vecchio-Sadus reviewed the available literature on the toxicity of Li and concluded that at natural background Li concentrations, its toxicity is relatively low for humans (Aral and Vecchio-Sadus, 2008). Additionally, researchers have also examined the possibility that elevated Li concentrations in drinking water may lead to a reduction of crime and suicide rates, but not all studies support this hypothesis (Kabacs et al., 2011; Ohgami et al., 2009; Oliveira et al., 2019). While there are no regulatory limits on the amount of Li in drinking water aside from the EPA-recommendation, the Australian Capital Territory set a limit on the amount of Li allowed to enter waterways (2.5 mg/L (EPR, 2005)). Based on the available information in the literature, we suggest that the Li distribution in the environment near Li deposits should be monitored. Apart from Li, Li-micas also contain significant amounts of fluorine (F), which may be released to the environment during weathering. High concentrations of F- in water may cause fluorosis (Khairnar et al., 2015; Li et al., 2016; Perumal et al., 2013; Rango et al., 2012). The WHO’s (WHO, 1999) threshold for F- concentration in drinking water is 1.5 mg/L. Ingesting water with F- concentrations between 1.5 and 3.0 mg/L leads to dental fluorosis, symptoms of which range from dental lesions and tooth discoloration to broken teeth (Yadav et al., 2009). Bones begin to show signs of fluorosis above 3.0 mg/L, with symptoms, such as, skeletal lesions, osteosclerosis, and changes in the structures of joints and limbs (Comba and Cinar, 2016). Severe skeletal fluorosis has been shown to interfere with the ability to breathe, as structural ribcage deformation affects proper lung function (Mandal et al., 2014). Fluoride can replace the hydroxyl group in hydroxyapatite, which makes up bones (Godebo et al., 2020; Kurdi, 2016; Srivastava and Lohani, 2015). The transformation of hydroxyapatite into fluorapatite has a negative impact on bone density, strength, collagen content and composition, and microstructure (Godebo et al., 2020). Globally, India is suffering the most from elevated F- concentrations in drinking water, but as elevated concentrations have also been observed elsewhere, F- levels are monitored in many parts of the world (Ali et al., 2016; Banerjee, 2015; Beltrán-Aguilar et al., 2010; Fordyce et al., 2007). Because both lepidolite and zinnwaldite, as well as amblygonite may contain considerable amounts of F, water sources near deposits that host these minerals should be analyzed for F- contents.

The goal of this research was to establish background levels of Li, F-, and other ionic species in waters around four hard-rock Li deposits in Central Europe, one of which is being actively explored and scheduled for production in the near future. If mining were to take place in one of these four deposits, this study provides pre-mining concentrations of these aqueous species and offers a benchmark for potentially required remediation. Additionally, this study aims to understand whether natural weathering of these deposits can lead to dangerous amounts of Li or F- in surface and drinking waters and to a potential public health concern. Finally, the geochemical data could also be used in exploration for Li deposits elsewhere.

2. Materials and Methods

2.1. Geological Setting

We collected water samples near three Li-mica granite/greisen bodies and one spodumene pegmatite in Central Europe (Figure A1). The granites are products of the Variscan orogeny, which resulted from the collision between Gondwana and Laurasia (Kroner and Romer, 2013). The Variscan subduction ceased at around 340 Ma, but the subsequent isothermal exhumation of the subducted continental crust is thought to be responsible for Late Variscan high-temperature metamorphism and voluminous granitic magmatism (Kroner and Romer, 2013). On the other hand, the Rappold Complex, which includes the Sankt Radegund bei Graz spodumene pegmatites, resulted from of an Eo-Alpine intra-continental collision during the Cretaceous (Gaidies et al., 2008) and is a prominent unit in the Upper Austroapline basement nappes between the Wölz and Saualpe-Koralpe Complexes (Gaidies et al., 2008).

2.1.1 Cínovec/Zinnwald, Czech Republic/Germany

Cínovec is located in the northwestern part of the Czech Republic (Figure A1), on the border with Germany, where the site is known as Zinnwald. The granite/greisen deposit stretches across the border (Figure 1). Geologically, these rocks belong to the Ore Mountains, also known by their German name “Erzgebirge” or Czech name “Krušné hory”. Historically, Cínovec has been mined for tin and tungsten, but the mines are no longer in operation. Large quantities of Li, however, are hosted by the granites and greisen, representing the largest known Li deposit in Europe and the fourth largest non-brine deposit in the world (European Metals, 2019). Therefore, the site is currently being explored for Li mining. The deposit is rich in the Li-mica zinnwaldite, named after this locality. The host rock, a slightly peraluminous A-type granite, was extensively greisenized, i.e., altered due to interaction with volatiles during solidification of the magma (Breiter et al., 2019, 2017). It is rich in Li2O (0.095 – 0.962 wt.%) and F (0.62 – 4.18 wt.%), but poor in P2O5 (≤ 0.023 wt.%, Table 1, (Breiter et al., 2017)), and contains elevated amounts of Nb, Sn, Ta, and W. The major minerals present are quartz, albite, and zinnwaldite; minor minerals include topaz, fluorite, cassiterite, and zircon (Breiter, 2012). Zinnwaldite, occurring down to a depth of 735 m below the surface, is rich in Fe (8.9 – 15.2 wt.% FeO), Li2O (2.0 – 4.4 wt.%), and F (7.0 – 8.6 wt.%; (Breiter et al., 2019)). The granites were emplaced into the Teplice Rhyolite (Breiter et al., 2019) (see Fig. 1). This site is especially interesting for its history of underground Sn and W mining, which has created tunnels that are now abandoned and drain percolating water. In addition, sand representing old tailings, has been deposited near the topographical peak of the area, where it is exposed to rain and vegetation. This sand is composed of zinnwaldite, quartz, and albite, and contains 0.27 wt.% Li2O (Czech Geological Survey, 2018).

2.1.2. Homolka, Czech Republic

The Homolka granite, occurring in the southern Czech Republic (Figures A1, A2), close to the border with Austria, is in the NW part of the Moldanubian pluton of the South-Bohemian batholith (Breiter and Scharbert, 1995). It is a peraluminous granite rich in Li2O (0.16 wt.%), Rb (400 – 1500 mg/kg), F (0.5 – 1.0 wt.%), P2O5 (0.5 – 1.0 wt.%), Sn, Nb, and Ta, and poor in Ti, Mg, Fe, Ca, Sr, Ba, Zr, and REE (Table 1, (Breiter and Scharbert, 1995)). The major minerals are quartz, albite, orthoclase, and a Li-muscovite, whereas the accessory minerals are topaz, apatite, cassiterite, ferrocolumbite, and rutile (Breiter and Scharbert, 1995). The Li-muscovite is rich in Li2O (1.5 – 2.15 wt.%) and F (1.7 – 2.6 wt.%, (Breiter and Scharbert, 1995)). The Homolka granite has not been mined and is not currently being considered an economic resource of Li, but it forms a large (6 km2), well defined outcrop, surrounded by two-mica granites: Lásenice to the West (“Two-Mica Granite” in Fig. A2) with no available information on F, Li2O, or P2O5 contents (René et al., 2003), and Eisgarn to the East and South (“Porphyritic Two-Mica Granite” in Fig. A2), which contains an average of 0.22 wt.% F, 0.03 wt.% Li2O, and 0.37 wt.% of P2O5 (Breiter and Scharbert, 1998).

2.1.3. Podlesí, Czech Republic

The Podlesí granite, located in the western Czech Republic (Figures A1, A3) in the Ore Mountains, comprises two types of Li-enriched rocks, the “Podlesí stock granite” and a dike granite (Breiter, 2002). Both rock types are relatively rich in incompatible elements, such as Li2O (0.033 – 0.445 wt.%), Rb (< 2754 mg/kg), Cs (< 253 mg/kg), Sn (< 65 mg/kg), and W (< 99 mg/kg), as well as F (0.6 – 2.4 wt%) and P2O5 (0.4 – 1.5 wt%), whereby the stock is generally more enriched (Breiter, 2002). One of these dikes (the thickest found, 7 m across) outcrops in an abandoned quarry, from which granite was extracted (Breiter, 2004) and which floods during major rain events. The Li is hosted by zinnwaldite that is polylithionite-rich in the stock granite, but siderophyllite-rich in the dike granite (Breiter, 2002). The other major minerals are quartz, albite and topaz in the stock granite, and quartz, K-feldspar, albite, and topaz in the dike Additionally, amblygonite and tourmaline are also reported as Li-bearing minerals (Breiter, 2002). Whereas the Podlesí granite stock and dike granites make up a relatively small area, the surrounding albite-biotite granites have above-average Li2O (~0.033 wt.% (Breiter et al., 2005)) and F (0.5 – 1.0 wt.% (Breiter et al., 2006)) contents as well. There is also significant tin mineralization (cassiterite) in many parts of the area. The country rock of the granites consists of phyllite.

2.1.4. Sankt Radegund bei Graz, Austria

There are over 20 known spodumene pegmatites in the Eastern Alps of Austria, which formed during the Cretaceous (Gaidies et al., 2008) and are spatially associated with leucogranites and barren pegmatites (Knoll et al., 2018). In addition to spodumene, the pegmatites contain K-feldspar, quartz, plagioclase, and muscovite, as well as minor garnet and tourmaline (Knoll et al., 2018). The spodumene pegmatites occur as dikes that were intruded into metamorphic rocks. We sampled stream water close to several known pegmatite outcrops near the town of Sankt Radegund bei Graz, Styria (Figures A1, A4). Their lengths are at least 15 – 20 m each, known to be at least 1.5 – 3 m thick, and geologically are part of the Rappold Complex (Schuster et al., 2019). The bulk-rock chemical composition given in Table 1 is from a different pegmatite within the Rappold Complex, because the outcrops near which we sampled did not have geochemical data available. The host rocks of the spodumene pegmatite outcrops in the sampling area are paragneiss and mica-schist; other prominent metamorphic rocks occurring within the studied watershed include marble and occasional dolomite, which both belong to the Schöckelkalk Formation (Flügel et al., 2011; Knoll et al., 2018).

2.2. Sampling

Hydrology (Figures 1 and A2-A4) was mapped in ArcGIS (Version 10.5) using the Hydrology toolset with the 30-Meter Shuttle Radar Topography Mission digital elevation model (DEM) downloaded from https://dwtkns.com/srtm30m/. Water-sampling points were then selected to best reflect the environment near a given deposit by aiming for sample collection at stream intersections, where available, to represent the largest possible area. Samples were collected within a given deposit, downstream from the deposit, as well as upstream where possible. The latter was not achievable in the cases of Cínovec and Homolka, where the deposits represent the highest-elevation peaks in the area (see Fig. 1, 2). The general sampling density ranged from 1 sample per 0.5 km2 to 1 sample per 3 km2, depending on the nature and area of the deposit and the number and size of nearby streams. Where possible, we sampled streams that were similar in terms of water flow, slope, and shape to minimize erosion effects on the geochemical signatures but, as expected, streams closer to the deposits and at higher elevations tended to be smaller than those downstream. Additionally, some streams could only be accessed at a few places as they were running through private properties, therefore limiting our efforts in this realm. Sampling took place in July and August 2020.

Our sampling procedure followed (Hem, 1985), except for the filtration step. We chose not to filter our samples in order to not only quantify the dissolved load, but also the exchangeable fraction to obtain a better understanding of the mobility of Li in these environments, as suggested by the US EPA (Puls and Barcelona, 1989). Water samples were collected into 250 mL polyethylene bottles. Temperature, pH, dissolved oxygen, and electric conductivity were determined directly in the field using a handheld Vernier digital probe (LabQuest 2). Alkalinity was measured in the field using an Hach 2444301 Alkalinity Test Kit, Model AL-AP. The samples were subsequently acidified to pH 2 with nitric acid and stored in a refrigerator at ~4°C until further analysis; the only exception was the air travel from the Czech Republic to Philadelphia, USA, during which the samples were in the cargo hold of the airplane, where the temperature typically is ~7°C. Average precipitation over the area of the Czech Republic for July of 2020 was 68 mm, lower than the 88 mm average for July between 1981 and 2010 (Czech Hydrometeorological Service, 2020a). August was wetter than normal, with 111 mm compared with the 80 mm average (Czech Hydrometeorological Service, 2020a). Mean daily temperatures were normal for July and hotter for August, at 17.7°C and 18.8°C, respectively, compared to long-term average values between 1981 and 2020 of 17.8°C and 17.3°C, respectively (Czech Hydrometeorological Service, 2020b). In general, the Czech Republic, especially the western and northwestern (Podlesí and Cínovec sites) regions, was experiencing a drought in the summer of 2020, apparent in the field as many desired sampling-point streams were dried out (Gvozdikova et al., 2020; Svabenicka et al., 2020). Our labeling scheme follows the first letter of the deposit (e.g., C for Cínovec) and numbers from 1 up, skipping numbers where we aimed to sample, but the site was dry.

The water from the Cínovec Deposit was sampled on August 11, 2020. This site is on the eastern flank of the Erzgebirge. The flow of the nearest monitored river, the Bílina, was much lower in July 2020 than in previous years (Czech Hydrometeorological Service, 2020c). This was evident in the field, as four sites we aimed to sample were dried up. The weather alternated between cloudy and sunny, and the temperature ranged from 18 to 24°C. A total of 10 samples were collected, one of which was an underground sample (labeled CMine), collected on the German side of the deposit in an old tin mine, which now serves as a tourist attraction.

The streams around the Homolka Granite deposit were sampled on July 13 and July 14, 2020. The flow rate of the Lásenice river (connecting H10 and H11 in Figure A2) shows slightly above-average values relative to historical data (Czech Hydrometeorological Service, 2020c). During sampling, the weather was sunny, ranging from 14 – 20°C over the course of the day. A total of 16 samples were collected.

The Podlesí Granite water was collected on July 29, 2020. The site is on the eastern flank of the Erzgebirge, in the rain shadow; in 2020, it received even less rainfall than normal (Svabenicka et al., 2020). Recorded flow for July 2020 in the nearest monitored river, the Ohře, was significantly below the historic mean or median values (Czech Hydrometeorological Service, 2020c). Several sampling sites were dry here as well. The weather was warm (20 – 25°C) and sunny during the sampling. A total of 11 samples were collected.

The water samples from the Sankt Radegund Pegmatite field were collected on August 8, 2020. The water level of the nearest monitored river, the Raab river, was slightly above average values during the sampling (Austrian Hydrometeorological Service, 2020). The day was sunny, with temperatures ranging from 19 to 24 °C. A total of 10 samples were collected.

2.3. Geochemical Analysis

The water samples were transported to the Department of Earth and Environmental Science (EES) at the University of Pennsylvania in order to perform chemical analyses by Inductively-Coupled Plasma Optical Emission Spectrometry (ICP-OES) and Ion Chromatography (IC). The following elements were analyzed by ICP-OES: Li, Na, K, Rb, Cs, Mg, Ca, Mn, Fe, Al, Si, and Cu. Using IC, the following anion concentrations were determined: SO42-, Cl-, F-, and PO43-. The HCO3- content was calculated from the alkalinity. The excess nitrate concentration, which resulted from the acidification of our samples, led to an overlap with the sulfate peak on the ion chromatograms. Therefore, sulfur was subsequently also analyzed by ICP-OES. We used six preliminary non-acidified samples from the Cínovec area, collected in the summer of 2019, that we analyzed on both IC and ICP-OES and the sulfate concentrations agreed well (within 92%).

2.4. Statistical Analysis

One-way cross correlation of all of our analyzed elements and anions as well as physical parameters (elevation, dissolved oxygen, pH, temperature, alkalinity, conductivity) was performed in Python (version 3), using the numpy as well as scipy.stats packages. Linear regressions of several of the more correlated variable pairs were also calculated.

2.5. Geochemical Modeling

Speciation and mineral saturation indices were modeled using PHREEQC (version 3.5, run on a Mac version 10.14, (Parkhust and Appelo, 2013)). Saturation indices allow us to make thermodynamic predictions about certain mineral precipitation/dissolution processes. For these calculations, we used the Lawrence Livermore National Laboratory thermodynamic database (llnl.dat), distributed together with PHREEQC, and augmented it by the thermodynamic data (log K = 15.6847) for zinnwaldite from (Ogorodova et al., 2010). We used the following dissolution equation of zinnwaldite: KLiFeAl2Si3O10F2 + 8 H+ = K+ + Li+ + 2 Al3+ + Fe2+ + 2 F- + 3 SiO2 + 4 H2O (1) modeled with the theoretical composition as presented in (Ogorodova et al., 2010) and shown in equation (1).

Additionally, inverse modeling with PHREEQC was carried out to complement the saturation indices and examine, using a different approach, which minerals in the Cínovec area have most likely been dissolved. The method was described in (Zhu and Anderson, 2002). We assume the initial water is rainwater with a composition of that sampled at the Bedřichov, CZE monitoring site (Czech Hydrometeorological Service, 2007). The final water is that of the collected stream samples. The increase of the chemical constituents in the final water, compared to the rainwater, was assumed to be caused by the weathering of local rocks. Specifically, we assume that the Li is derived from the dissolution of zinnwaldite, F- from the dissolution of zinnwaldite and fluorite, Cl- from biotite (in the Teplice Rhyolite) and atmospherically dry-deposited halite (Drever, 1988), Mg from biotite, and sulfate from gypsum. The mass balance of Al and Si was achieved from the dissolution of feldspars (albite, K-feldspar and anorthite), biotite (Mg-endmember phlogopite), and zinnwaldite, and from precipitation of the three clay minerals kaolinite (Al2(Si2O5)(OH)4), allophane (Al2O3)(SiO2)1.3-2·2.5-3H2O, amorphous), and imogolite (Al2SiO3(OH)4). Apatite (with a theoretical composition of Ca5(PO4)3(F0.8,(OH)0.2) was also included as a potential source of Ca and F-. Goethite was included as a sink of Fe. Ion exchange between clays and Li, Na, K, and Ca was allowed, and is presented as LiX, NaX, KX, and CaX2 (Carroll, 1959); Li is scrubbed by clays very rapidly (Li and Liu, 2020; von Strandmann et al., 2020), making clay adsorption of Li essential to realistic modeling efforts. As the streams are open to the atmosphere, CO2 (g) and O2 (g) were used as phases as well. As there can be many combinations of mineral-dissolution processes resulting in the same chemical composition of water, the standard approach is to only consider minimal models (Parkhust and Appelo, 2013), i.e., the fewest number of phases involved in the model, leading to the most essential geochemical reactions considered.

3. Results

3.1. Geochemical Analysis

The entire set of geochemical data for all water samples can be found in the Appendix, Table A1. Generally, the samples containing the highest amounts of total dissolved solids (TDS) are from the Cínovec site, with sample C14 having the highest values. The highest Li concentrations are observed in the waters from the Cínovec and Homolka sites (Figure 2). Fluoride is highly concentrated in the Cínovec sites, which are also most enriched in Na and Cl. The Homolka site is richer in Al and SO42- than the other sites. Generally, iron and phosphate ions are more concentrated in the Podlesí site. Sankt Radegund bei Graz has the highest arithmetic mean concentrations of Mg, Ca, SiO2, and HCO3-. Copper showed almost no variation across the four samples and was mostly near the ICP-OES detection limit. The major cation and anion distributions of all samples are shown in a Piper diagram (Figure 3). The dominant cations are Ca2+ and Na+, whereas the three major anions are more evenly represented. Waters draining the pegmatite at Sankt Radegund bei Graz are distinct from the other sites, being Ca-HCO3 waters (Figure 3). Streams around the three Li-mica deposits range from calcic to sodic waters, with a few in the Podlesí site being Mg-rich. Although individual stream samples from these three deposits span the range between sulfate and chloride waters, those from the Homolka site are mostly sulfate-dominated, those from the Cínovec site include the most chloride-enriched, leaving the waters from Podlesí plotting in between. Upstream samples/samples closer to the source rocks are generally more concentrated in Li and F relative to downstream samples. The mine-drainage water (CMine) is a Ca-Cl water type, and so is sample C14, which is richest in Li and F-. An important result to note is that the stoichiometric ion imbalance (IB) of some of the samples is considerably non-zero (arithmetic mean is 12.9%, with a standard deviation of 6.5%), the highest being 57% for sample C12 (see below). This imbalance drops by about a half for most samples when the speciated imbalance is calculated (using PHREEQC) rather than the stoichiometric imbalance; for C12, for example, the speciated IB has a value of 33.3%.

3.2. Statistical Analysis

The arithmetic mean, standard deviation, and maximum concentrations of Li and F- in the waters from the four studied areas are listed in Table 2. Of note are the relatively large standard deviations of Li concentrations for the Cínovec and Homolka sites as well as the large standard deviation for the Cínovec F- concentrations. The Cínovec and Homolka sites show nearly identical maximum and arithmetic mean Li concentrations, yet very different F- concentrations. Correlations between variable element pairs of interest are shown in Table 3, with the cross-correlations of all parameters displayed in the Appendix (Tables A2-A5). The pairs presented in Table 3 were chosen based on chemical makeup of the Li-minerals and magnesium based on its similar ionic radius, resulting in similar environmental behavior (Ogawa et al., 2014). Several strong and significant correlations are observed: between Li and F- in Cínovec and Homolka, Li and Al in the same deposits, Li and Mg in Cínovec and Podlesí, albeit of opposite signs, Li and Si in Homolka, Li and K in Cínovec and Sankt Radegund bei Graz, and F- and PO43- in Podlesí. Amongst ions not presented in Table 3 (Tables A2-A5), strong and significant correlations (r > 0.8, p < 0.05) observed in Cínovec are: Li and Ca, Li and K, Li and Cl, Na and Ca, Na and K, Na and Cl, Mg and Ca, Mg and Al, Mg and Cl, Mg and SO42-, Mg and HCO3-, K and Cl-, Al and Cl-, Al and SO42-, Fe and Mn, Cl- and HCO3-, and SO42- and HCO3-. In Homolka, there are significantly fewer strong and significant correlations: Na and Mg, Na and Ca, and Mg and Ca. In Podlesí, strong and significant correlations are observed between Na and Ca, Na and K, Na and Cl-, Ca and K, Ca and Cl-, K and Cl-, and Fe and Mn. Finally, Sankt Radegund bei Graz presents only Ca and HCO3- as a strong and significant correlation.

Table 2. Descriptive statistics of Li and F ions in the waters at our four studied sites (mg/L). Lithium Fluoride Deposit Mean Std Dev Max Mean Std Dev Max n Cínovec 0.011 0.012 0.036 1.365 1.408 3.806 10 Homolka 0.013 0.009 0.035 0.321 0.112 0.544 16 Podlesí 0.006 0.001 0.009 0.103 0.022 0.140 11 Sankt Radegund 0.005 0.002 0.007 0.062 0.016 0.088 10

Table 3. Correlation coefficients (r) between Li concentration and that of several ions of interest and between F- and phosphate ions. Star () highlights significant correlations (p < 0.05). Deposit F vs. Li Si vs. Li Fe vs. Li Al vs. Li Mg vs. Li K vs. Li F vs. PO4 Cínovec 0.679 0.485 0.146 0.764* 0.776* 0.802* -0.110 Homolka 0.701* 0.853* 0.162 0.792* 0.458 0.231 -0.452 Podlesí 0.532 0.319 0.043 0.448 -0.799* -0.049 0.604* Sankt R. -0.270 0.567 0.066 0.140 0.507 0.738* 0.519

Special interest was placed on the strong correlation, especially at Cínovec and Homolka, between Li and F- (Fig. 4). In the Cínovec dataset, sample C12, with its low Li (0.003 mg/L) but high F- (3.566 mg/L) contents, is an outlier, which affects the slope of the regression line, causing it to not go through the population of samples with lower Li concentrations (solid green line in Fig. 4). Therefore, we calculated a second correlation for the Cínovec samples without sample C12, yielding a much better correlation (r = 0.985; dotted green line in Fig. 4). The molar F:Li values in zinnwaldite are about 1.5, depending on the exact composition (a microprobe analysis of a Zinnwald mica shows a Li2O content of 3.9 wt.%, and a F content of 6.5 wt.%, yielding a molar F:Li ratio of 1.31 (Tischendorf, 1997)), yet the slope of correlation lines between the corresponding aqueous ionic species is around 30 (36 without the C12 sample) for the two correlation lines (Fig. 4). The Homolka and Podlesí sites have comparable slopes of ~3, but the correlation is not significant (p = 0.061) in the Podlesí site. On the other hand, the correlation for the Homolka samples is better (r = 0.701) and highly significant (p = 0.005). The Sankt Radegund bei Graz site shows no correlation (r = -0.270, p = 0.451).

The Al vs. Li relationship adds additional insight, as it is strong and significant in both the Cínovec and the Homolka datasets (Table 3). Similar to the comparison of the F:Li values between the water samples and micas, we would expect the molar Al:Li values to be between 1 and 2, if the two elements were derived exclusively from the micas, but the slopes of the best-fit lines are about 16 and 14 for Cínovec and Homolka, respectively. The relationship between Si and Al is strong and significant in the Homolka site (r = 0.785, p = 0.001, Table A3) and the slope of the best fit line is 4.5, close to pure feldspar dissolution which would plot near 3. The strong (r = 0.853) and significant (p = 0.000) relationship between Si and Li in the Homolka site leads to a best-fit line slope for the correlation line between the two elements of about 98, much higher than an expected value of 2-4 should both elements be released from micas only. The strong and significant Mg vs. Li correlations in Cínovec (r = 0.776, p = 0.008) and Podlesí (r = -0.799, p = 0.002) are interesting as some Mg could likely be substituted into zinnwaldite and some Li into biotite. The strong and significant correlation of K vs. Li in the Cínovec site leads to a best-fit line slope of 15, significantly above the 1:1 ratio expected should they be leached from the same Li-micas, but nearly the same value as for Al vs. Li. The strong and significant correlation between F- and PO43- at the Podlesí site seems to reflect the presence in the rocks of amblygonite and apatite. 3.3. Geochemical Modeling

In general, the water samples examined in greater detail with geochemical modeling tools are undersaturated with respect to zinnwaldite, fluorite, and the feldspars (Figure 5) and supersaturated with respect to kaolinite, allophane, and imogolite (not shown in Figure 5 for easier visualization of the other minerals of interest). In our inverse modeling inputs, this result allowed us to force the undersaturated minerals to dissolve only and the supersaturated phases to precipitate only.

To examine the outlier in the Cínovec deposit (sample C12) in greater detail, saturation indices (SIs) of minerals known to be present in the area were modeled. Special interest was placed on fluorite because of the high F-, but low Li contents determined for the water at the C12 site (see Fig. 4). Zinnwaldite, albite and K-feldspar are undersaturated at site C12, whereas quartz is near saturation (Fig. 5); fluorite and fluorapatite are also undersaturated. Calcite is included in the model to compare to fluorite at the C12 site. The C14 site features the largest SIs for all minerals (except fluorapatite, at C5), whereby the water appears to be oversaturated with respect to K-feldspar, albite, quartz and fluorapatite. The SIs for the mine water (CMine) are elevated as well, but not significantly higher than, for example, the waters at C3 or C5 (with the exception of albite), neither of which are rich in F- or Li. In general, the waters are undersaturated with respect to the modeled minerals, except for occasionally K-feldspar, albite, and fluorapatite, and they are near equilibrium for quartz and fluorite at all sites.

The results of the inverse modeling for the water sample from the mine shaft (CMine) and for the fluoride-enriched samples C12 and C14 are provided in Table 4. The composition of the initial rainwater is given in Table A6. We were able to produce several minimal inverse models in each case. We used an uncertainty of 0.05 for CMine, but had to increase it to 0.10 for C12 and 0.15 for C14 to produce balanced models. An uncertainty of 1 means that any initial water is allowed to evolve into any final water; as the number increases, it means that there are likely processes at play that were not accounted for in the modeling (e.g., absorption by vegetation, kinetics). In C14, two models are produced with the only major difference being kaolinite precipitation as opposed to allophane precipitation, which then changes the other phases accordingly. In both models, six-times more fluorite is dissolved compared to zinnwaldite. None of the feldspars nor apatite are predicted to supply ions into the solution. The modeling results further suggest that monovalent cations (Li, Na, and K) are predicted to be scrubbed from the solution by clays in favor of Ca, which would be released from the clays into the solution. In C12, five models are produced, with models 1,2,3, and 5 distinguished mostly by the types of clays that precipitate and feldspars that dissolve. Model 4 is different from the other models as significantly more zinnwaldite (99x) is predicted to dissolve. In Models 1,2,3, and 5, fluorite supplies nearly ~232x more F-; in Model 4, this ratio drops to 1.35. Ion exchange suggests Li and K being bound to clays, thus releasing Na and Ca. At the CMine site, nine minimal inverse models were produced. There are mostly two groups of models: Models 1,7, and 8, and Models 2,3,4,5,6, and 9. According to the first group of models, the same amounts of zinnwaldite and fluorite are dissolved. The other group predicts no fluorite dissolution, suggesting that all F- is derived from zinnwaldite. The major differences within the two groups of models are the Si-Al balance of which feldspars dissolve and which clays precipitate. The ion exchange is similar to C14 in that monovalent cations are scrubbed from solution and Ca released into the solution. At all sites and in all models, CO2 and O2 are removed from the atmosphere and brought into the solution.

Table 4. Precipitation (-)/dissolution (+) of phases (in mmol) predicted to explain the observed water composition via inverse modeling. Phases are presented in bold. Rainwater from the Bedřichov (Table A6, Czech Hydrometeorological Service, 2007) station was used as initial water. Uncertainty used in the models is 15% for C14, 10% for C12, and 5% for CMine. C14 Zinnwaldite Fluorite Albite Anorthite K-Feldspar Apatite Biotite Gypsum Kaolinite Goethite Halite Allophane Imogolite LiX NaX KX CaX2 CO2(g) O2(g) Model 1 0.0140 0.0855 0.0000 0.0000 0.0000 0.0000 0.1524 0.4770 -0.0360 0.0000 2.7544 0.0000 0.0000 -0.0088 -1.1377 -0.0750 0.6108 1.1323 0.0029 Model 2 0.0140 0.0855 0.0000 0.0000 0.0000 0.0000 0.1457 0.4770 0.0000 0.0000 2.7611 -0.0327 0.0000 -0.0088 -1.1444 -0.0683 0.6108 1.1323 0.0029 C12 Zinnwaldite Fluorite Albite Anorthite K-Feldspar Apatite Biotite Gypsum Kaolinite Goethite Halite Allophane Imogolite LiX NaX KX CaX2 CO2(g) O2(g) Model 1 0.0004 0.0927 0.0000 0.1259 0.0000 0.0000 0.0030 0.2648 -0.0861 -0.0004 0.0000 0.0000 -0.0392 0.0000 0.8234 -0.4557 0.0880 0.6189 0.0001 Model 2 0.0004 0.0927 0.0000 0.1161 0.0196 0.0000 0.0030 0.2648 -0.1253 -0.0004 0.0000 0.0000 0.0000 0.0000 0.8234 -0.4459 0.0684 0.6189 0.0001 Model 3 0.0004 0.0927 0.0196 0.1161 0.0000 0.0000 0.0030 0.2648 -0.1253 -0.0004 0.0000 0.0000 0.0000 0.0000 0.8039 -0.4459 0.0880 0.6189 0.0001 Model 4 0.0396 0.0536 0.0000 0.1259 0.0000 0.0000 0.0030 0.2648 -0.1644 -0.0396 0.0000 0.0000 0.0000 -0.0392 0.8234 -0.4166 0.0489 0.6189 0.0099 Model 5 0.0004 0.0927 0.0000 0.1259 0.0000 0.0000 0.0030 0.2648 -0.0298 -0.0004 0.0000 -0.0955 0.0000 0.0000 0.8234 -0.4557 0.0880 0.6189 0.0001 CMine Zinnwaldite Fluorite Albite Anorthite K-Feldspar Apatite Biotite Gypsum Kaolinite Goethite Halite Allophane Imogolite LiX NaX KX CaX2 CO2(g) O2(g) Model 1 0.0345 0.0345 0.0495 0.0000 0.0348 0.0000 0.0245 0.1443 0.0000 -0.0337 1.6906 0.0000 -0.0823 -0.0302 -0.7614 0.0000 0.3958 0.5425 0.3055 Model 2 0.0691 0.0000 0.0000 0.0000 0.0842 0.0000 0.0245 0.1443 0.0000 -0.0683 1.6906 -0.1169 0.0000 -0.0648 -0.7119 -0.0840 0.4303 0.5425 0.3141 Model 3 0.0690 0.0000 0.0498 0.0171 0.0000 0.0000 0.0245 0.1443 0.0000 -0.0682 1.6906 0.0000 -0.1168 -0.0647 -0.7617 0.0000 0.4132 0.5425 0.3141 Model 4 0.0690 0.0000 0.0842 0.0000 0.0000 0.0000 0.0245 0.1442 0.0000 -0.0682 1.6906 -0.1169 0.0000 -0.0647 -0.7961 0.0000 0.4304 0.5425 0.3141 Model 5 0.0725 0.0000 0.0000 0.0000 0.0994 0.0000 0.0232 0.1374 -0.1276 -0.0717 1.6898 0.0000 0.0000 -0.0682 -0.7048 -0.1014 0.4372 0.5425 0.3150 Model 6 0.0706 0.0000 0.1004 0.0000 0.0000 0.0000 0.0232 0.1374 -0.1261 -0.0697 1.6889 0.0000 0.0000 -0.0662 -0.8082 0.0000 0.4372 0.5425 0.3145 Model 7 0.0345 0.0345 0.0000 0.0000 0.0843 0.0000 0.0245 0.1443 0.0000 -0.0337 1.6906 0.0000 -0.0823 -0.0302 -0.7119 -0.0495 0.3958 0.5425 0.3055 Model 8 0.0345 0.0345 0.0843 0.0000 0.0000 0.0000 0.0245 0.1443 0.0000 -0.0337 1.6906 0.0000 -0.0823 -0.0302 -0.7962 0.0348 0.3958 0.5425 0.3055 Model 9 0.0690 0.0000 0.0000 0.0173 0.0498 0.0000 0.0245 0.1443 0.0000 -0.0682 1.6906 0.0000 -0.1169 -0.0647 -0.7119 -0.0495 0.4131 0.5425 0.3141

4. Discussion

The Li concentrations measured in this study are elevated (over 0.02 mg/L) in five samples, and on average (for Cínovec and Homolka), they tend to plot above average values reported in other studies that measured Li in surface waters (Figure 6). For Podlesí and Sankt Radegund bei Graz, the Li values resemble typical background values reported in the literature (~0.01 mg/L, (Aral and Vecchio-Sadus, 2008; Kavanagh et al., 2017; Rodrigues et al., 2019)).

The Cínovec and Homolka mean and median Li concentrations are above those of known drinking water concentrations in Japan and England (Kabacs et al., 2011; Ohgami et al., 2009), the average (which average not specified in the publication) of background Czech waters (n = 5,765), and above the Mackenzie river (Millot et al., 2010), but lower relative to the waters around spodumene deposits in SE Ireland and Portugal ((Kavanagh et al., 2017; Rodrigues et al., 2019), Figure 6). It is important to note that in the SE Ireland study, the Li concentrations measured in the month of July were the lowest (median < 0.01 mg/L) amongst March, May, July, and September, with no correlation to the total amount of precipitation during those months (Kavanagh et al., 2017). However, we believe it is also important to point out that during the collection of our water samples, it was a relatively dry period (see above), which may have affected the Li concentrations not only by lowering the weathering intensity but also the dilution effects, therefore the net impact of precipitation is difficult to predict. The maximum concentrations we observed are not nearly as elevated as the outliers in most other studies shown in Figure 6, including the nationwide geochemical study of Czech surface waters, which reported the highest Li concentration (0.114 mg/L) from an area of “Permian-Carbonian, Mesozoic, and Tertiary sediments”, without any additional details (Chuman et al., 2013). The highest surface water Li concentrations are known from northern Chile near Li-brine deposits, where the concentrations can reach 20 mg/L (Figueroa et al., 2012), about three orders of magnitude larger than the arithmetic mean concentrations reported in our study and in the studies of waters near spodumene deposits in SE Ireland and Portugal (see above). Aral and Vecchio-Sadus reported on communities in Chile using water with Li concentrations approaching 10 mg/L, with no reported adverse health effects, although acquired tolerance may be a factor (Aral and Vecchio-Sadus, 2008). The concentrations we determined are much lower and, moreover, considerably below the EPA-recommended 0.7 mg/L threshold. We therefore conclude that there is likely no public health risk resulting from the Li concentrations of the waters examined for this paper, but additional sampling during spring/fall should be performed for a more complete assessment. The stream water F- concentrations were very high around the Cínovec deposit, with a maximum value of 3.8 mg/L (Table 2), well above the WHO drinking water limit of 1.5 mg/L (WHO, 1999). At these concentrations, long-term exposure to this water could result in skeletal fluorosis (Mandal et al., 2014). The arithmetic mean F- concentrations at Cínovec, however, are slightly below the WHO drinking water limit of 1.5 mg/L (see Fig. 7). The arithmetic mean F- concentrations at Cínovec, on the other hand, are much higher than the average concentrations from the Czech Republic (Fig. 7). Whereas the Homolka waters are not nearly as enriched in F- as those at the Cínovec site – likely due to the smaller proportion of F in the respective host rocks (Table 1, (Breiter, 2012; Breiter and Scharbert, 1995)) – they are still considerably richer in F- than the typical concentrations in Czech waters (Fig. 7). The mean and median F- concentrations at the Podlesí site are comparable to the overall concentrations of the Czech Republic, and the waters at the Sankt Radegund bei Graz site did not contain elevated amounts of F- (Table 2, Fig. 7).

The arithmetic mean of the absolute stoichiometric ion imbalance over all of our samples (IB = 12.86%) is above the generally accepted 5%, but this result is not uncommon in the literature (Fritz, 1994). It can likely be explained by the addition of nitric acid, which resulted in the following four possible errors: 1) it made us unable to measure nitrate and nitrite concentrations and therefore, our anion concentration sum is lower than the true anion concentration sum; in fact, the anion sum is lower than the cation sum in most samples (37 out of 47 samples); 2) we measured sulphate by ICP-OES due to the nitrate peak being too dominant over the sulphate one seen by IC; when analyzing a non-acidified set of preliminary samples from the same area collected in the summer of 2019 (Table A7, arithmetic mean IB = 4.08%) using both the ICP-OES and the IC, the ICP-OES data for sulphate agreed with the IC data at 92%, therefore reporting slightly lower sulphate concentrations than on the IC; cation concentrations were very comparable between the two datasets (compare Tables A1 and A7); 3) alkalinity, used to calculate bicarbonate and carbonate concentrations, was measured with a titration kit in the field, potentially leading to these values also being slightly incompatible with what we would see on the IC; large IB values are in fact commonly blamed on the field alkalinity measurement (Fritz, 1994); and 4) there is a possibility that an important cation for some samples has not been measured, which could explain the 10 samples (e.g., C12, H2) where the cation sum was lower than the anion sum. Additionally, the choice to not filter our samples could also have led to larger IBs due to the possible presence of colloids in our water samples, which may dissolve during acidification (Deutsch and Siegel, 1997). Finally, the speciated IB is typically about half that of the stoichiometric IB and, in fact, it is <5% for the majority of our samples, and in our most imbalanced sample (C12) it has a value of 33.3%.

The highest concentrations of both Li and F- are found in sample C14. An old mining shaft, “Starý Martin”, near the town of Krupka, is located about one kilometer north of this sampling site and at higher elevation (Figure 1). Breiter et al. (Breiter et al., 2017) mapped Krupka as a “hidden stock of post caldera zinnwaldite granites”. The proximity of this granite stock and the drainage to this underground shaft could be causing the very high concentrations of Li and F-, which are higher than water sampled within the actual Cínovec mine (CMine). At the Homolka site, the three samples most enriched in Li – H3, H4, and H7 (label hidden below H4) – all came from small streams (< 1.0 m wide) directly within the deposit itself (Fig. A2). Not all small streams within the deposit were significantly enriched in Li, however (e.g., H1, H2), although most do plot above at least 0.01 mg/L (e.g., H5, H6, H8). Samples H12 and H13 are from a catchment area in which there are no outcrops of the Homolka granite, and plot at background values (0.006 mg/L). Because two of the three downstream samples (H11, H14) did not show elevated Li concentrations, there seems to exist some Li-removal process(es) further downstream, possibly via uptake by vegetation and/or clay adsorption ((Aral and Vecchio-Sadus, 2008)), or via precipitation of known, natural Li compounds with low solubilities, such as Li-fluoride, -phosphate, or -carbonate (Kavanagh et al., 2017). Alternatively, dilution by mixing with water from other drainage areas could also explain the lower Li concentrations in the downstream samples. In fact, dilution may have a considerable effect on ion concentrations. In another study, for example, significant impacts on the concentrations of Fe, Cu, Zn, and Co due to dilution have been observed only 300 meters away from the point source of these elements (Moon, 1999). The magnitude of dilution can be approximated by analyzing stream sediment and taking into account the catchment area of the stream (Moon, 1999). This approach, however, was beyond the scope of our study but might provide valuable insights in a future investigation. Inverse geochemical modeling showed that the evolution of water chemistry from rainwater to streamflow samples can be accounted for by dissolving zinnwaldite, fluorite, biotite, gypsum, halite (dry deposition) and CO2, and precipitation of Al-Si phases such as allophane and kaolinite. The decrease of Li in downstream samples was modelled as ion-exchange reactions involving Li species. This approach presents an uncertainty, and, in the future, lithium isotope analysis could represent another constraint to interpreting the lower Li concentrations in downstream samples.

Weathering and associated Li behavior in river basins have been studied increasingly by using Li isotopes as a proxy for silicate weathering (Penniston-Dorland et al., 2017; Tomascak, 2004; Tomascak et al., 2016; von Strandmann et al., 2020). Typically, congruent weathering (primary mineral dissolution without significant secondary mineral precipitation, (Dellinger et al., 2015)) takes place upstream and leads to larger concentrations of aqueous Li and isotopically light waters (i.e., low 𝛿7Li values), similar to the parent rock material (von Strandmann et al., 2020). Further downstream, where secondary mineral precipitation is significant, weathering is termed incongruent and Li concentrations tend to be lower because Li is removed from solution by incorporation into the crystal structure and/or the exchangeable layer of clays (von Strandmann et al., 2020). Accommodation of Li in the exchangeable layer does not lead to much isotopic fractionation, but the Li that is incorporated into the octahedral site of the clays is preferentially the lighter 6Li, leaving the waters enriched in 7Li (Dellinger et al., 2015; Tomascak et al., 2016; von Strandmann et al., 2020). Studying the isotopic composition of our aqueous samples in the future would allow us to evaluate whether the lower Li concentrations downstream are due to clay precipitation (high 𝛿7Li) or due to dilution (similar 𝛿7Li as upstream).

Near the Cínovec site, higher Li concentrations were observed in samples collected closer to the deposit than further downstream, likely pointing to dilution or the mentioned Li-removal processes. Other sites showed no upstream/downstream distinction. Low concentrations of both Li and F- at the Podlesí site could be explained by the physical size of the Li-enriched granite bodies – their area is relatively small compared to the watershed sampled and, thus, dilution could be effective at lowering the concentration. In the Sankt Radegund bei Graz site, similarly, the pegmatite outcrops are likely too small to have a major impact on the chemical composition of the surface water. The SE Ireland pegmatites form a long chain of pegmatite outcrops (Kavanagh et al., 2017), likely having a larger impact on the local aqueous geochemistry. The samples from the Gonçalo mine in Portugal were collected within an active mining area and in a landscape dominated by Li-aplite-pegmatite-sills (Rodrigues et al., 2019), also likely having larger impact on the water geochemistry as a result of the larger area covered. The dominant signature of Mg2+, Ca2+, and HCO3- in the Sankt Radegund waters most likely reflect the weathering of the marble/dolomite Schöckelkalk Formation, as indicated by the arithmetic mean SIs for calcite and dolomite, which are near equilibrium (-0.52 and -0.60, respectively) for these waters.

Lithium is strongly correlated with F- in the three Li-mica deposits, albeit not significantly in the Podlesí site (Table 3, Figure 4). The strong correlations point to these elements being leached from the same minerals, which would be the Li-micas. The slopes of the F- vs. Li linear regression lines for the Podlesí and Homolka waters are similar (~3). We would expect the slope to be around 2 if all F- were derived from the micas, based on the theoretical KLiFeAl2Si3O10F2 composition of the model zinnwaldite (see equation (1)). The Cínovec correlation coefficient is comparable to Homolka’s (0.679 and 0.701, respectively), but the slope of the correlation line is an order of magnitude larger for the Cínovec data (Fig. 4). This result implies dissolution of an additional F-bearing phase. The water composition at site C12 is an outlier among the Cínovec samples, likely because it represents drainage from a fluorite-rich and zinnwaldite-poor vein. Fluorite, and possibly fluorapatite, are the only other minerals in the area that contain F (Breiter et al., 2017) and both are undersaturated at the C12 site (Figure 5). Based on four out of five minimal inverse models (Table 4), the C12 site is most likely explained by near exclusive fluorite dissolution (i.e., very little zinnwaldite); the remaining model (Model 4) predicts fluorite and zinnwaldite to dissolve in comparable amounts, which would require the Li supplied by the zinnwaldite dissolution to be scrubbed by clay minerals, which would be kaolinite in this specific model. At the C14 site, both inverse models explain the excess F- as produced by dissolving six-times more fluorite than zinnwaldite. At the CMine site, the excess F- is due to exclusive dissolution of zinnwaldite, or alternatively, it is caused by dissolution of equal amounts of fluorite and zinnwaldite. No model for any site requires apatite to dissolve or to precipitate, even though it is thermodynamically undersaturated in C12 and oversaturated in C14 (Figure 5). If our inverse modeling results were correct, it would appear that each site is quite different and it is difficult to state conclusively whether zinnwaldite or fluorite are the main phases supplying F-, as the inverse models point in different directions at the different sites. It does appear, however, that fluorite dissolution is more important at sites C14 and C12, whereas zinnwaldite dissolution is more important at the CMine site. A dissolution experiment of the rocks would allow for a better explanation in regard to which minerals are providing the ions we observed in the water, as it would take into account the actual geochemical concentrations in the rock, mineral makeup, kinetics, as well as thermodynamics. We are currently working on such an experiment. Lithium also correlates strongly with Al and Si (Table 2), the core building blocks of the mica structure, although only one Si (Homolka) and two Al (Cínovec and Homolka) correlations are significant. While the Si vs. Li correlation is very strong and significant at the Homolka site, the slope of the line approaches 100, i.e., most of the silicon is derived from other sources. At Cínovec and Homolka, the slopes of the Al vs. Li regression lines are ~16 and ~14, respectively, suggesting significant release of aluminum from minerals other than the Li-micas, most likely the feldspars or biotite, also suggested by the nearly identical K vs. Li regression line slope in Cínovec of ~15. The strong and significant Si vs. Al regression in Homolka yields a slope of ~4.5, above the expected value for pure feldspar dissolution (~3), suggesting significant input from the micas. Even though Fe is present in zinnwaldite, it does not correlate with Li at any of the sites (Tables A2-A5). Therefore, Fe is likely derived from more common Fe-bearing minerals, such as biotite (predicted to dissolve, for example, at all modelled Cínovec sites; see Table 4), or hematite, which is present in minor amounts in the pink-colored granites. We analyzed the correlation between Mg and Li in view of their nearly identical ionic radii (0.72 Å and 0.74 Å, respectively). In Li-brine processing, it is generally difficult to separate Li+ from Mg2+, and as a result, brines with large molar Mg/Li values are not being considered economical Li-sources (Ogawa et al., 2014). The arithmetic mean Mg/Li molar values in Cínovec waters are 69.7, in Homolka 33.0, in Podlesí 91.5, and in Sankt Radegund 167.7. The seawater Mg/Li value is 7000, which is the major reason why seawater is currently not a source of Li. The molar Mg/Li value of the Great Salt Lake, which is currently being explored as a potential source of Li (Marthi and Smith, 2019), is 133.33, whereas it is 6.6 for the Salar de Atacama, the most productive, albeit not the largest, brine deposit (Kavanagh et al., 2018). The relatively low Mg/Li values of the waters studied here suggest it might be feasible to recover leached Li from the waterways or tailing ponds around hard-rock deposits, especially should the concentration of Li increase as a result of mining operations. Both elements also tend to be incorporated into the clay interlayer or precipitate as part of secondary minerals during incongruent weathering (von Strandmann et al., 2020; Wimpenny et al., 2014), suggesting that perhaps it is this process that leads to the strong correlation (Table 3) rather than primary mineral dissolution. Clay uptake may also be why the highest Li concentrations reported in the current Czech waters are found near sediment deposits (Chuman et al., 2013). We see a strong positive correlation between Mg and Li at the Cínovec site and a comparably strong negative correlation between these species at the Podlesí site. While we cannot say much about clay precipitation/uptake of ions without sequential extraction of the sediment/clay minerals or without a Li-isotope study (Li-isotope fractionation most pronounced during secondary mineral formation (von Strandmann et al., 2020)), our inverse models do suggest significant Li removal from solution via clay adsorption and/or precipitation. Our models only consider ion-exchange as we do not have data on how much Li is/would be present in the clays in our study area, which would allow us to model the Li removed from solution by clay precipitation. The best approach to studying the ratio of how much Li is in the exchangeable site compared to the octahedral site of the clays in our samples would be via determination of Li isotopic compositions (von Strandmann et al., 2020). Either process (clay precipitation or adsorption) would explain why we find highly Li-concentrated samples near the deposits and low Li concentrations in downstream waters. This lends itself to a future investigation assessing the Li movement in these environments. Adsorption of Li onto clays, as documented in experiments (up to 99% of Li scrubbed by kaolinite in laboratory conditions within 1000 minutes of the experiment, (Li and Liu, 2020)), points to a potential remediation technique for high-Li waters under natural conditions. This method would also work better in lower-Mg/Li waters, as there would be less competition between the ions. We also examined the phosphate correlations due to the presence of amblygonite and fluorapatite at the Podlesí site, and indeed, there is a strong and significant positive correlation between F- and phosphate at this location. On the other hand, we also observed a relatively strong correlation between F- and phosphate (not significant, however) in the Sankt Radegund bei Graz deposit where no F- or phosphate-bearing minerals have been identified. No strong correlation between Li and phosphate is found at any of the studied sites (Tables A2-A5). Finally, there are several very strong and significant correlations in the Cínovec site, mostly involving Cl-, Na, K, Mg (Table A2). Those are presented as additional clues into the environment examined and could prove important to future investigations of similar deposits, but these strong correlations are much fewer in number at our other studied locations. Lithium correlates very strongly with some of these ions in the Cínovec waters as well (e.g., Ca, Cl-, K). Our inverse modeling results suggest that these correlations could be the result of clay interaction, as Li, Ca, K, and Na are all affected by clay adsorption/precipitation. It is further important to note that some of these correlations, especially between Si, Al, and Mg, may be an artifact of not filtering the water samples and introducing contributions of microminerals from the areas examined.

5. Conclusions

Weathering of Li-mica deposits can result in elevated Li and F- levels in the surface waters in close proximity to these deposits. The most important characteristics impacting the surface water chemistry appear to be: 1) Li and F contents of the rocks; 2) presence or absence of accessory fluorite; 3) the size of the Li-rich granite body relative to the drainage area; and 4) how close to the surface a given Li-rich granite is to have an impact on surface waters. In the case of the studied Cínovec deposit, markedly high F- levels have been observed, which are likely a product of fluorite and zinnwaldite dissolution, depending on the exact site and rock type being drained. The amount of F- present in three of the ten Cínovec samples is considerably above the WHO drinking water limit and represents a public health risk. This risk could be heightened by the development and mining of this deposit as more of these Li- and F-rich rocks would be exposed to the atmosphere, which would result in weathering. However, Li-ores with no or only minimal fluorite may not result in the increased leaching of F- into the waterways, as suggested by our statistical and geochemical modeling work. Dissolution experiments should be carried out on the leaching of Li and F from these rocks to confirm our prediction and pinpoint which mineral is most responsible for the F- input. The Li concentrations in five of the samples analyzed here are above those of average surface waters around the world, but not as high as in two spodumene deposits examined in other studies, likely due to clay uptake of the Li. The measured Li concentrations in waters around the studied Li deposits do not currently pose a public health threat based on the EPA-recommended Li concentration in drinking water.

Funding:

This research was funded in part by the Water Center at Penn Student Support Program.

Acknowledgments:

For guidance early in the conceptualization phase and suggestion of field sites, we thank Dr. Karel Breiter. We also thank Dr. Rainer Sennewald for access to the old Zinnwald tin mine, his insights on local geology, and assistance with sampling. Finally, we thank the two anonymous reviewers for their invaluable feedback and editor Dr. Stefano Albanese for handling our manuscript.

References

AICS, 2007. Australia Inventory of Chemical Substances.

Ali, S., Thakur, S.K., Sarkar, A., Shekhar, S., 2016. Worldwide contamination of water by fluoride. Environmental Chemistry Letters 14, 291–315.

Aral, H., Vecchio-Sadus, A., 2008. Toxicity of lithium to humans and the environment - A literature review. Ecotoxicology and Environmental Safety 70, 349–356.

Austrian Hydrometeorological Service, 2020. Hydrology of Austria. Banerjee, A., 2015. Groundwater fluoride contamination: A reappraisal. Geoscience Frontiers 6, 277–284.

Beltrán-Aguilar, E.D., Barker, L., Dye, B.A., 2010. Prevalence and Severity of Dental Fluorosis in the United States, 1999-2004. NCHS Data Brief 53.

Bradley, D.C., McCauley, A.D., Stillings, L.M., 2017a. Mineral-deposit model for lithium-cesium-tantalum pegmatites (No. 2010-5070– 0), Mineral deposit models for resource assessment, Chapter O. U.S. Geological Survey, Reston, Virginia.

Bradley, D.C., Stillings, Lisa M., Jaskula, B.W., Munk, L., McCauley, Andrew D., 2017b. Lithium, in: Schulz, K.J., DeYoung, J.H., Jr., Seal, R.R., II, and Bradley, D.C., Eds., Critical Mineral Resources of the United States - Economic and Environmental Geology and Prospects for Future Supply. pp. K1–K21.

Breiter, K., 2012. Nearly contemporaneous evolution of the A- and S-type fractionated granites in the Krušné Hory/Erzgebirge Mts., Central Europe. Lithos 151, 105–121.

Breiter, K., 2004. Podlesí Quarry.

Breiter, K., 2002. From explosive breccia to unidirectional solidification textures: magmatic evolution of a phosphorus- and fluorine-rich granite system (Podlesí, Krušné hory Mts., Czech Republic). Bulletin of the Czech Geological Survey 77, 67–92.

Breiter, K., Ďurišová, J., Hrstka, T., Korbelová, Z., Hložková Vaňková, M., Vašinová Galiová, M., Kanický, V., Rambousek, P., Knésl, I., Dobeš, P., Dosbaba, M., 2017. Assessment of magmatic vs. metasomatic processes in rare-metal granites: A case study of the Cínovec/Zinnwald Sn–W–Li deposit, Central Europe. Lithos 292–293, 198–217.

Breiter, K., Förster, H.-J., Škoda, R., 2006. Extreme P-, Bi-, Nb-, Sc-, U- and F-rich zircon fractionated perphosphorous granites: The peraluminous Podlesí granite system, Czech Republic. Lithos 88, 15–34.

Breiter, K., Hložková, M., Korbelová, Z., Vašinová Galiová, M., 2019. Diversity of lithium mica compositions in mineralized granite-greisen system: Cínovec Li-Sn-W deposit, Erzgebirge. Ore Geology Reviews 106, 12–27.

Breiter, K., Müller, A., Leichmann, J., Gabašová, A., 2005. Textural and chemical evolution of a fractionated granitic system: the Podlesí stock, Czech Republic. Lithos 80, 323–345.

Breiter, K., Scharbert, S., 1998. Latest intrusions of the Eisgarn Pluton (South Bohemia - Northern Waldviertel). Jahrbuch der Geologischen Bundesanstalt 141, 25–37.

Breiter, K., Scharbert, S., 1995. The Homolka magmatic centre - an Example of Late Variscan ore bearing magmatism in the Southbohemian Batholith (Southern Bohemia, Northern Austria). Jahrbuch der Geologischen Bundesanstalt 138, 9–25.

Carroll, D., 1959. Ion exchange in clays and other minerals. Bulletin of the Geological Society of America 70, 749–780.

Chuman, T., Hruska, J., Oulehle, F., Gürtlerova, P., Majer, V., 2013. Does stream water chemistry reflect watershed characteristics? Environmental Monitoring and Assessment 185, 5683–5701. https://doi.org/10.1007/s10661-012-2976-3

Comba, B., Cinar, A., 2016. Investigation of effects of fluorosis on some minerals and hormones in sheep. Ankara Üniversitesi Veteriner Fakültesi Dergisi 63, 223–227.

Czech Geological Survey, 2018. Povrchová těžba lithia (“Surface mining of lithium”). Cínovec.

Czech Hydrometeorological Service, 2020a. Average monthly precipitation in 2020. Czech Hydrometeorological Service.

Czech Hydrometeorological Service, 2020b. Average monthly temperatures in 2020. Czech Hydrometeorological Service.

Czech Hydrometeorological Service, 2020c. Daily and historical average and median flowrates of monitored streams. Czech Hydrometeorological Service.

Czech Hydrometeorological Service, 2007. Average concentrations in rainwater. CHMU.

Dellinger, M., Gaillardet, J., Bouchez, J., Calmels, D., Louvat, P., Dosseto, A., Gorge, C., Alanoca, L., Maurice, L., 2015. Riverine Li isotope fractionation in the Amazon River basin controlled by the weathering regimes. Geochemica et Cosmochimica Acta 164, 71–93.

Deutsch, W.J., Siegel, R., 1997. Groundwater Geochemistry: Fundamentals and Applications to Contamination, 1st ed. CRC Press.

Drever, J., 1988. The geochemistry of natural waters, 2nd ed. Prentice Hall Inc., Englewood Cliffs, New Jersey, USA.

EPR, 2005. Environment Protection Regulation 2005, made under Environment Protection Act 1997 (No. SL2005-38). Updated 2019, Australian Capital Terriroty.

ERA-MIN3, 2021. ERA-MIN3. Network of 27 Institutions, Europear Union Horizon.

European Metals, 2019. Quarterly activities report - September 2019.

European Metals Holdings Limited, West Perth, WA, Australia.

Figueroa, L., Barton, S., Schull, W., Razmilic, B., Zumaeta, O., Young, A., Hoskins, J., Ilgren, E., 2012. Environmental lithium exposure in the North of Chile - I. Natural Water Sources. Biological Trace Element Research.

Flexer, V., Baspineiro, C.F., Galli, C.I., 2018. Lithium recovery from brines: A vital raw material for green energies with a potential environmental impact in its mining and processing. Science of the Total Environment 639, 1188–1204.

Flügel, H.W., Nowotny, A., Gross, M., 2011. Geologische Karte der Republik Österreich 1:50,000, 164 Graz.

Fordyce, F.M., Vrana, K., Zhovinsku, E., Povoroznuk, V., Toth, G., Hope, B.C., Iljinsky, U., Baker, J., 2007. A health risk assessment for fluoride in Central Europe. Environmental Geochemistry and Health 29, 83–102.

Fritz, S.J., 1994. A survey of chsarge-balance errors on published analyses of potable ground and surface waters. Ground Water 32, 539–547.

Gaidies, F., Krenn, E., De Capitani, C., R. Abart, 2008. Coupling forward modelling of garnet growth with monazite geochronology: an application to the Rappold Complex (Austroalpine crystalline basement). Journal of Metamorphic Geology 26, 775–793.

Gao, J.-Q.G., Yu, Y., Wang, D.-H., Wang, W., Wang, C.-H., Dai, H.-Z., Hao, X.-F., Cen, K., 2021. Effects of lithium resource exploitation on surface water at Jiajika mine. Environmental Monitoring and Assessment 193.

Godebo, T.R., Jeuland, M., Tekle-Haimanot, R., Shankar, A., Alemayehu, B., Assefa, G., Whitford, G., Wolfe, A., 2020. Bone quality in fluoride-exposed populations: A novel application of the ultrasonic method. Bone reports 12, 100235.

Gourcerol, B., Gloaguen, E., Tuduri, J., Melleton, J., 2019. Typology of hard-rock Li-hosted deposits in Europe. Presented at the 15th Biennal Meeting SGA 2019, Glasgow, UK.

Greim, P., Solomon, A.A., Breyer, C., 2020. Assessment of lithium criticality in the global energy transition and addressing policy gaps in transportation. Nature Communications 11, 4570. https://doi.org/doi.org/10.1038/s41467-020-18402-y

Gvozdikova, B., Sadkova, E., Fatka, O., Lamacova, A., 2020. Monthly report on the hydrometeorological situation and draught in the Czech Republic: August 2020 (No. #8). Czech Hydrometeorological Service.

Hem, J.D., 1985. Study and interpretation of the chemical characteristics of natural water, Third Edition. ed. USGS.

Hien-Dinh, T.T., Dao, D.A., Tran, Tam, Wahl, M., Stein, E., Gieré, Reto, 2017. Lithium-rich albite-topaz-lepidolite granite from Central Vietnam: a mineralogical and geochemical characterization. European Journal of Mineralogy 29, 35–52.

Huh, Y., Chan, L.C., Zhang, L., Edmond, J.M., 1998. Lithium and its isotopes in major world rivers: implications for weathering and the oceanic budget. Geochemica et Cosmochimica Acta 62, 2039–2051.

Kabacs, N., Memon, A., Obinawa, T., Stochl, J., Perez, J., 2011. Lithium in drinking water and suicide rates across the East of England. The British Journal of Psychiatry 198, 406–407.

Kavanagh, L., Keohane, J., Cabellos, G.G., Lloyd, A., Cleary, J., 2018. Global lithium sources - Industrial use and future in the electric vehicle industry: A review. Resources 7, 1–29.

Kavanagh, L., Keohane, J., Cleary, J., Cabellos, G.G., Lloyd, A., 2017. Lithium in the natural waters of the South East of Ireland. International Journal of Environmental Research and Public Health 14, 561.

Khairnar, M.R., Dodamani, A.S., Jadhav, H.C., Naik, R.G., Deshumkh, M.A., 2015. Mitigation of fluorosis - A review. Journal of Clinical and Diagnostic Research 9, ZE05–ZE09.

Kjølholt, J., Stuer-Lauridsen, F., Mogensen, A.S., Havelund, S., 2003. The Elements in the second rank - An environmental problem now or in the future? (No. 770 2003). Danish Environmental Protection Agency, Danish Ministry of the Environment.

Knoll, T., Schuster, R., Huet, B., Mali, H., Onuk, P., Horschinegg, M., Ertl, A., Giester, G., 2018. Spodumene pegmatites and related leucogranites from the Austroalpine Unit (Eastern Alps, Central Europe): Field relations, petrography, geochemistry, and geochronology. The Canadian Mineralogist 56, 489–528.

Kovari, M., Coleman, M., Cristescu, I., Smith, R., 2018. Tritium resources available for fusion reactors. Nuclear Fusion 58, 1–10.

Kroner, U., Romer, R.L., 2013. Two plates - Manu subduction zones: The Variscan orogeny reconsidered. Gondwana Research 24, 298–329.

Kurdi, M.S., 2016. Chronic fluorosis: The disease and its anaesthetic implications. Indian Journal of Anaesthesia 60, 157–162.

Lebedeva, N., Di Persio, F., Boon-Brett, L., 2017. Lithium ion battery value chain and related opportunities for Europe (No. JRC105010). Joint Research Centre, The European Commission, Luxembourg.

Li, H., Eksteen, J., Kuang, G., 2019. Recovery of lithium from mineral resources: State-of-the-art and perspectives - A review. Hydrometallurgy 189, 105–129.

Li, W., Liu, X.-M., 2020. Experimental investigation of lithium isotope fractionation during kaolinite adsorption: Implications for chemical weathering. Geochemica et Cosmochimica Acta 284, 156–172.

Li, X., Wu, P., Han, Z., Shi, J., 2016. Sources, distributions of fluoride in waters and its influencing factors from an endemic fluorosis region in central Guizhou, China. Environmental Earth Sciences 75, 981–995.

Liu, D., Gao, X., An, H., Qi, Y., Sun, X., Wang, Z., Chen, Z., An, F., Jia, N., 2019. Supply and demand response trends of lithium resources driven by the demand of emerging renewable energy technologies in China. Resources, Conservation & Recycling 145, 311–321.

Loganathan, S., Nayak, R., Sinha, S., Taly, A., Math, S., Varghese, M., 2008. Treating mania in Wilson’s disease with lithium. Journal of Neuropsychiatry and Clinical Neuroscience 20, 487–489.

Mandal, A., Ray, S., Mandal, S.K., Roy, T., Mallik, R., Deoghuria, D., 2014. A study on spirometric evaluation of lung volume restriction in prediagnosed cases of skeletal fluorosis. Journal of Evolution of Medical and Dental Sciences 3, 7558–7562.

Marthi, R., Smith, Y.R., 2019. Selective recovery of lithium from the Great Salt Lake using lithium manganese oxide-diatomaceous earth composite. Hydrometallurgy 186, 115–125.

Martin, G., Rentsch, L., Höck, M., Bertau, M., 2017. Lithium market research - global supply, future demand and price development. Energy Storage Materials 6, 171–179.

Mayo Clinic, n.d. Lithium (Oral Route). Mayo Foundation for Medical Education and Research.

Millot, R., Vigier, N., Gaillardet, J., 2010. Behaviour of lithium and its isotopes during weathering in the Mackenzie Basin, Canada. Geochemica et Cosmochimica Acta 74, 3897–3912.

MIR, 2008. The trouble with Lithium 2: Under the microscope. Meridian International Research, Martainville, France.

Moon, C.J., 1999. Towards a quantitative model of downstream dilution of point source geochemical anomalies. Journal of Geochemical Exploration 65, 111–132.

Ogawa, Y., Koibuchi, H., Suto, K., Inoue, C., 2014. Effects of the chemical compositions of Salars de Uyuni and Atacama brines on lithium concentration during evaporation. Resource Geology 64, 91–101.

Ogorodova, L.P., Kiseleva, I.A., Melchakova, L.V., 2010. Thermodynamic Properties of Lithium Micas. Geochemistry International 48, 415–418.

Ohgami, H., Terao, T., Shiotsuki, I., Ishii, N., Iwata, N., 2009. Lithium levels in drinking water and risk of suicide. British Journal of Psychiatry 194, 464–465.

Oliveira, P., Zagalo, J., Madeira, N., Neves, O., 2019. Lithium in public drinking water and suicide mortality in Portugal: Initial approach. Acta médica portuguesa 32, 47–52.

Ono, M., Majeski, M.A., Jaworski, M.A., Hirooka, Y., Kaita, R., Gray, T.K., Maingi, R., Skinner, C.H., Christenson, M., Ruzic, D.N., 2017. Liquid lithium loop system to solve challenging technology issues for fusion power plant. Nuclear Fusion 57, 1–9.

Parkhust, D.L., Appelo, C.A.J., 2013. Description of input and examples for PHREEQC version 3 - A computer program for speciation, batch-reaction, one-dimensional transport, and Inverse geochemical calculations. USGS, Colorado, USA.

Penniston-Dorland, S., Liu, X.-M., Rudnick, R.L., 2017. Lithium Isotope Geochemistry. Reviews in Mineralogy and Geochemistry 82, 165–217.

Perumal, E., Paul, V., Govindarajan, V., Panneerselvam, L., 2013. A brief review on experimental fluorosis. Toxicology Letters 223, 236–251.

Puls, R.W., Barcelona, M.J., 1989. Ground Water Sampling for Metals and Analyses (No. EPA/540/4-89/001), Superfund Ground Water Issue. United States Environmental Protection Agency.

Rahn, P.H., Davis, A.D., Webb, C.J., Nichols, A.D., 1996. Water quality impacts from mining in the Black Hills, South Dakota, USA. Environmental Geology 27, 38–53.

Rango, T., Kravchenko, J., Atlaw, B., McCornick, P.G., Jeuland, M., 2012. Groundwater quality and its health impact: An assessment of dental fluorosis in rural inhabitants of the Main Ethiopian Rift. Water for Food Faculty Publications 13.

René, M., Matejka, D., Nosek, T., 2003. Geochemical constraints on the origin of a distinct type of two-mica granites (Deštná - Lásenice type) in the Moldanubian Batholith (Czech Republic). Acta Montana IRSM AS CR Series A No.23, 59–76.

Richter, J., 2018. Competent Person’s Report on the Zinnwald Lithium Ore Deposit (No. 30130130). G.E.O.S. Ingenieurgesellschaft mbH. Rodrigues, P.M.S.M., Antão, A.M.M.C., Rodrigues, R., 2019. Evolution of the impact of lithium exploitation at the C57 mine (Gonçalo, Portugal) on water, soil, and air quality. Environmental Earth Sciences 78, 533.

Schuster, R., Knoll, T., Mali, H., Huet, B., Griesmeier, G.E.U., 2019. Field trip guide: A profile from migmatites to spodumene pegmatites (Styria, Austria) (No. Berichte der Geologischen Bundesanstalt, 134). Geologische Bundesanstalt, St. Radegund, Austria.

Srivastava, A., Lohani, M., 2015. Fluorine, a Dreaded Element: A Review on Occurrence of Fluorine in Environment and its Standard Methods of Analysis. International Journal of Environmental Research and Development 5, 7–21.

Svabenicka, J., Stepnakova, B., Fatka, O., Lamacova, A., 2020. Monthly report on the hydrometeorological situation and draught in the Czech Republic: July 2020 (No. #7). Czech Hydrometeorological Service.

Swain, B., 2017. Recovery and recycling of lithium: A review. Separation and Purification Technology 172, 388–403.

Tanveer, M., Hasanuzzaman, M., Wang, L., 2019. Lithium in environment and potential targets to reduce lithium toxicity in plants. Journal of Plant Growth Regulation 38, 1574–1586. https://doi.org/10.1007/s00344-019-09957-2 Tischendorf, G., 1997. On Li-bearing micas: estimating Li from electron microprobe analyses and an improved diagram for graphical representation. Mineralogical Magazine 61, 809–834.

Tomascak, P.B., 2004. Developments in the Understanding and Application of Lithium Isotopes in the Earth and Planetary Sciences. Reviews in Mineralogy and Geochemistry 55, 153–195.

Tomascak, P.B., Magna, T., Dohmen, R., 2016. Advances in Lithium Isotope Geochemistry. Springer, Switzerland.

Tripathi, R.K., 2011. Lithium. Virginia Department of Health.

US EPA, 2021. National primary drinking water regulations. US EPA.

US EPA, 2008. Provisional Peer Reviewed Toxicity Values for Lithium (No. CASRN 7439-93-2). U.S. Environmental Protection Agency, Cincinnati, OH, USA. USGS, 2020. Mineral commodity summaries 2020. U.S. Geological Survey, Reston, Virginia.

USGS, 2015. Mineral commodity summaries 2015. U.S. Geological Survey, Reston, Virginia.

von Strandmann, P.A.E.P., Kasemann, S.A., Wimpenny, J.B., 2020. Lithium and lithium isotopes in Earth’s surface cycles. Elements 16, 253–258.

WHO, 1999. Guidelines for Drinking Water Quality, part 1, vol. 2. World Health Organization, Geneva, Switzerland.

Wimpenny, J., Colla, C.A., Yin, Q.-Z., Rustad, J.R., Casey, W.H., 2014. Investigating the behaviour of Mg isotopes during the formation of clay minerals. Geochemica et Cosmochimica Acta 128, 178–184.

Yadav, J.P., Lata, S., Katarja, S.K., Keemer, S., 2009. Fluoride distribution in ground water and survey of dental fluorosis among school children in the villages of Jhajjar District of Haryana, India. Environmental Geochemistry and Health 31, 431–438.

Zhu, C., Anderson, G., 2002. Environmental Applications of Geochemical Modeling. Cambridge.

Entered by: 0x5b20…D3B2

Archmagus George of the Sun (#4766)
Owner: 0x5b20…D3B2