Global Biogeochemical Cycle of Lithium
Abstract
The total human mobilization of Li from the Earth's crust, >1,000 × 109 g/year, is much larger than Li mobilized by the natural processes of chemical and mechanical weathering (94 and 132 × 109 g/year, respectively), representing a ∼500% perturbation of the global cycle of Li by human activities. The anthropogenic perturbation of the global Li cycle shows enhanced releases to freshwaters from oil-produced water (46 × 109 g/year), leaching of coal ash (7–20 × 109 g/year), and extraction of groundwaters (29 × 109 g/year). The sum of these anthropogenic sources more than doubles the natural transport of dissolved Li to the sea in rivers (69 × 109 g/year). Currently, releases from the excretion of therapeutic drugs and disposal of lithium-ion batteries are a small component of the transport of Li in rivers, although the latter may increase markedly as lithium-ion batteries dominant global energy storage. Human emissions of Li particles to the atmosphere—55 × 109 g/year from coal combustion—comprise about 38% of the emission of Li to the atmosphere from various sources. The inputs to the atmosphere are more than the estimated deposition of Li from the atmosphere, which is poorly constrained by available data.
Key Points
-
Current human impact on the global lithium cycle has increased by a factor of 4 the mobilization of lithium to the Earth's surface
-
Major human releases of Li are found in coal ash leaching, oil-produced waters, and groundwater extraction
-
Mining and use of Li in industrial products will have large impacts on the global lithium cycle
Plain Language Summary
We quantified the extent to which the mining of lithium by humans, and other means by which lithium is mobilized at the Earth's surface, compare to the natural processes that deliver lithium to the environment. Given that lithium at high concentrations can have deleterious consequences for plants and animals, a comparison of natural to human sources allows us to evaluate the potential impacts of increasing production and use of lithium in the electronics industry and to evaluate the need for lithium recycling to maintain future supply for human use.
1 Introduction
Lithium is the lightest alkali metal and is one of a group of three elements (Li, Be, and B) that are anomalously rare in the Universe based on the standard cosmological model (Malaney & Fowler, 1988; Vangioni & Cassé, 2018). Due to its particular chemical properties (i.e., low atomic mass, high reactivity, and easy exchange of the outermost of its three electrons), Li has become a key element in modern batteries. A typical electric vehicle battery, for example, may contain as much as 12 kg of Li, which—combined with other industrial applications—contributes to the growing worldwide demand for Li (Bibienne et al., 2020).
Lithium is mined commercially from pegmatite minerals (e.g., lepidolite and spodumene) mostly in Australia and China. Large quantities of Li are also extracted from evaporite minerals and associated brines, with the largest production occurring in Chile, Argentina, and the United States. Current annual global production of Li is 77,000 mt/year (77 × 109 g/year; U.S. Geological Survey, 2020). At current use, proven reserves are expected to last for about 200 years, but demand for Li is estimated to increase 10-fold by 2050 (Sovacool et al., 2020). Junne et al. (2020) estimate that the ratio of cumulative Li demand to reserves may be as high as 6.5. Clearly, new deposits must be found, and large-scale recycling instituted, to accommodate increasing future demand.
Although Li has no known role in biochemistry, it has long been used to treat bipolar disorder and other problems of mental health (Geddes et al., 2004). The mechanism of its efficacy has not been fully elucidated (Curran & Ravindran, 2014). At high levels, Li is toxic to plants (McStay et al., 1980; Shahzad et al., 2017), aquatic organisms (Kszos & Stewart, 2003; Niemuth et al., 2019; Pinto-Vidal et al., 2021), and humans (Aral & Vecchio-Sadus, 2008). Although there is no official standard for Li in drinking water, the US Geological Survey suggests an upper limit of 60 μg/L and a health-based screening level of 10 μg/L (Lindsey et al., 2021).
There are two natural lithium isotopes, 6Li and 7Li, both of which are stable, representing 7.6% and 92.4% in natural abundance, respectively. The isotopic ratio is reported as δ7Li, which equals ([7Li/6Li] sample/[7Li/6Li] standard−1) × 1,000, where the standard is a synthetic carbonate designated as L-SVEC provided by the US National Institute of Standards and Technology. Early work by Lui-Heung Chan revealed substantial Li isotopic fractionation during water-rock interactions in the oceans (e.g., Chan & Edmond, 1988; Chan et al., 1992), and the Li isotope system is now widely used to study diverse earth processes (e.g., Penniston-Dorland et al., 2017; Tomascak et al., 2016).
Much attention has focused on the use of Li isotopes to study continental weathering, owing to the potential link between weathering rates and climate change. Numerous field and experimental studies have shown that during weathering of silicate rocks, 6Li partitions selectively into secondary minerals, particularly clays, leaving 7Li enriched in the aqueous phase (e.g., Huh et al., 1998, 2004; Kisakurek et al., 2005; Pistiner & Henderson, 2003; Rudnick & Gao, 2014; Wimpenny, Gilason, et al., 2010). Various researchers have used the Li isotope ratio in marine biogenic skeletons, such as foraminifera, which dominantly reflects the Li isotope composition in the ambient seawater, to estimate the rate of silicate weathering through geologic time (Hathorne & James, 2006). Misra and Froelich (2012) report that the δ7Li in seawater increased beginning 60 million years ago, implying large increases in continental weathering and the potential for enhanced sequestration of atmospheric CO2 by rock weathering (Pogge von Strandmann et al., 2020).
While studies of Li and its isotopes promise to provide valuable information on diverse earth processes (see reviews by Tomascak et al., 2016 and Penniston-Dorland et al., 2017), less attention has been paid to the escalating anthropogenic mobilization of Li. The purpose of this paper is to quantify the natural and anthropogenic global circulation of Li at the Earth's surface to provide context for the increasing flux of Li to nature as a result of human activities. There is also growing recognition that dispersal of Li from coal ash, wastewater from oil and gas exploration, and discarded electronic products present potential environmental hazards (Choi et al., 2019; Robinson et al., 2018), which will be quantified in this study.
2 Natural Biogeochemical Cycle
2.1 Terrestrial Rock Weathering
Lithium concentrations in most terrestrial rocks range from about 1 to 60 ppm (mg/kg) (e.g., Bradley et al., 2017; Teng et al., 2004). As a large ion lithophile element, Li increases in concentration during magmatic differentiation, reaching values greater than 8,000 mg/kg in granitic pegmatites. Lithium is also concentrated in evaporite deposits formed by precipitation of brines in desert lakes, as well as in lithium-rich clays, both of which commonly have concentrations ranging from 500 to 3,000 mg/kg (Bradley et al., 2017; Castor & Henry, 2020; SedDB, 2014). Chemical and mechanical weathering of terrestrial materials mobilizes Li, which is transported to the oceans in dissolved load, suspended load, and bed load (e.g., Huh et al., 1998).
About ∼90% of Li dissolved in river water originates from chemical weathering of silicate minerals (Kisakurek et al., 2005), recognizing that some of the dissolved Li may also derive from silicate clay minerals in sedimentary rocks (Penniston-Dorland et al., 2017) or continental hydrothermal activity (Henchiri et al., 2014; Millot, Scaillet, & Sanjuan, 2010; Pogge von Strandmann et al., 2016). A gross estimate of Li mobilization by chemical weathering can be derived using estimates of the global transport of total dissolved solids in rivers (3.9 × 1015 g/year; Garrels & MacKenzie, 1971) and the mean concentration of Li in the Earth's upper continental crust (24 mg/kg; Rudnick & Gao, 2014), yielding a value of 94 × 109 g Li/year. The mean concentration of dissolved Li in waters of Amazon River basin is 2.2 μg/L, with concentrations generally decreasing from headwaters to the mouth of the river (Dellinger et al., 2015). Gaillardet et al. (2014) calculate the global flux of Li transported as the dissolved load in rivers as 69 × 109 g Li/year, based on their estimate of the average dissolved Li concentration in rivers of 1.84 μg/L and river discharge of 3.74 × 104 km/year (Berner & Berner, 2012). This is quite similar to the value reported previously by Huh et al. (1998), 55 × 109 g Li/year, based on the more limited concentration data set available at that time, as well as a more recent estimate of Henchiri (2017). Additional Li may reach the oceans by submarine groundwater discharge in coastal areas, with estimated values ranging from 4 to 21 × 109 g Li/year (Mayfield et al., 2021; Pogge von Strandmann et al., 2014).
Continental rocks have δ7Li ranging from −4‰ to +8‰ (Teng et al., 2004). Enrichment of 7Li in river waters (δ7Li = +23‰) suggests that a portion of the Li released by chemical weathering is incorporated in 6Li-enriched clay minerals and carried in suspended load or bed load (Liu et al., 2015). Dellinger et al. (2017) show a range of concentrations of 10–100 mg/kg of Li in sediments of major world rivers. A recent estimate of the global average suspended sediment composition suggests a Li concentration of 8.5 mg/kg (Viers et al., 2009), significantly lower than an earlier estimate of 25 mg/kg (J.-M. Martin & Meybeck, 1979). Using the new value and estimates of total contemporary suspended sediment flux (12.6 × 1015 to 18.5 × 1015 g/year; Peucker-Ehrenbrink, 2010; Syvitski et al., 2005; Walling, 2006), we estimate that the flux of Li to the oceans in the suspended load ranges from 107 × 109 to 157 × 109 g/year; incorporating bed load might further raise this value by 10% (Syvitski et al., 2005) but is uncertain and not included here. The ratio of the Li flux in suspended to dissolved load is thus greater than 1 (1.6–2.3), consistent with previous conclusions that despite its solubility, Li is preferentially transported to the oceans in particulates (e.g., Gaillardet et al., 2014; Huh et al., 1998; Oelkers et al., 2011; Viers et al., 2009). For the Amazon River basin, Dellinger et al. (2015) report 40%–95% of the transported load of Li is carried in suspended sediments, with the highest values found in headwaters. In China, the Yellow River carries 73%–98% of its Li load in the suspended fraction (Guo et al., 2019). Not all of the Li in the suspended load reaches the ocean. In a study of processes occurring in estuarine environments, Pogge von Strandmann et al. (2008) suggested that continued weathering of suspended sediment and the associated formation of secondary minerals may remove as much as 15%–25% of the global riverine input of Li to the oceans.
2.2 Atmospheric Transport
The dominant natural sources of lithium in the atmosphere are volcanic emissions, sea-salt aerosols, and eolian dust. Li is removed from the atmosphere in deposited particulate matter and as a dissolved constituent in rainfall.
To our knowledge, there are no published estimates of the annual flux of lithium to the atmosphere in volcanic emissions. We therefore estimate this flux by drawing upon the better-studied volcanic flux of Fe (8,800 × 109 g/year; Rauch & Pacyna, 2009, and references therein), which, like Li, is moderately incompatible during the relevant magmatic processes. We determine a global average Li/Fe ratio for arc volcanics of 0.00126 (PetDB, 2021), which is almost identical to an average arc andesite ratio (0.00141) reported by Keleman et al. (2014). Multiplying the annual volcanic Fe flux by the Li/Fe ratio yields an estimate of 11 × 109 g/year of Li released to the atmosphere in volcanic emissions, a relatively small contribution to the movement of Li at the Earth's surface.
We estimate the transport of Li into the atmosphere by sea-salt aerosols, by multiplying estimates of such aerosol production (5–10 × 1015 g/year; Andreae & Rosenfeld, 2008; Lewis & Schwartz, 2004), by an estimate of the mass fraction of Li in sea salt (0.005; Bruland et al., 2014), assuming that the production of sea-salt aerosol is nonfractionating with respect to elemental Li in seawater (Glass & Matteson, 1973). This yields an estimate of Li mobilization of 26–52 × 109 g/year in sea-salt aerosols. For comparison, Stoffyn-Egli and Mackenzie (1984) calculate this flux as 1 × 109 g Li/year, using the Na content of seawater and a much lower estimate of sea-salt production.
A number of studies have examined Li mobilized to the atmosphere as eolian dust, in part to explore continental weathering and its influence on climate (e.g., Sauzéat et al., 2015). Changing deposition of Li in the Antarctic and Greenland snowpacks, for example, has been used to infer changing rates of global dust transport over glacial to interglacial periods, with greater rates of dust deposition during the last glacial epoch (Siggaard-Andersen et al., 2007). Lithium and Li isotopes are also used to constrain the relative proportions and sources of natural and anthropogenic dusts in various settings around the world (e.g., Clergue et al., 2015; W. Li et al., 2020; Y. F. Li et al., 2020; Millot, Petelet-Giraud et al., 2010). While the Li concentration in eolian dust varies by source, available data range from ∼17 to 41 mg/kg (e.g., Liu et al., 2013; Teng et al., 2004), with an average similar to that of the upper continental crust (24 mg/kg; Rudnick & Gao, 2014). We estimate the amount of Li mobilized to the atmosphere in eolian dusts by multiplying the average upper crustal Li concentration by an estimate of global dust production (1,600 × 1012 g/year; Andreae & Rosenfeld, 2008), yielding 38.4 × 109 g/year. Thus, sea salt and eolian dusts comprise the largest natural Li fluxes to the atmosphere.
Deposition of Li from the atmosphere, by rainfall and dryfall, has been measured infrequently and available data vary widely (Table 1). Generally, higher Li concentrations are measured in continental than in maritime or coastal precipitation, and values vary due to local and seasonal influences, including human activities (e.g., Clergue et al., 2015; Y. F. Li et al., 2020; Millot, Petelet-Giraud et al., 2010; see Section 3.2). Lithium in young sections of Antarctic and Greenland ice cores provide the lowest global values, primarily due to their distance from dust sources (Siggaard-Andersen et al., 2002, 2007). The concentration of Li in rainfall collected at maritime and coastal sites ranges from 0.04 to 1 μg/L, with the higher values likely reflecting dissolved Li from local dusts (Millot, Petelet-Giraud et al., 2010). Using the lower, more conservative value of 0.04 μg/L, and 3.73 × 1017 L/year of oceanic precipitation (Trenberth et al., 2007), yields a global estimate of 15 × 109 g/year for the deposition of Li over the oceans. Measured concentrations of Li in continental precipitation extend from 0.035 μg/L to as high as 5.5 μg/L, the latter for unfiltered samples from the Tibetan plateau, but in general range from about 0.4 to 0.9 μg/L. Using this range, and an estimate of terrestrial rainfall of 1.13 × 1017 L/year globally (Trenberth et al., 2007), would indicate deposition of 45 × 109 to 102 × 109 g Li/year on the land surface worldwide. Thus, total atmospheric deposition, 60 × 109 to 117 × 109 g Li/year, is less than the total of our estimated flux of Li to the atmosphere of 143 × 109 g/year (Figure 1). Particular attention should be given to particles that are deposited near sites of emission, thus avoiding global atmospheric transport. The atmospheric deposition of Li on the land surface could contribute a small fraction to the transport of dissolved Li in river waters (Gou et al., 2019).
| µg/L | Reference | ||
|---|---|---|---|
| Polar | |||
| Greenland (ice cores) | |||
| ∼400 ybp | Filtered | 0.0021 | Siggaard-Andersen et al. (2002) |
| 75,000 ybp | Filtered | 0.0055 | Siggaard-Andersen et al. (2002) |
| East Antarctica (ice core) | |||
| 0–12 ka BP | Filtered | 0.007 | Siggaard-Andersen et al. (2007) |
| Unfiltered | 0.006 | Siggaard-Andersen et al. (2007); Gabrielle et al. (2005) | |
| 20–45 ka BP | Filtered | 0.011 | Siggaard-Andersen et al. (2007) |
| Unfiltered | 0.041 | Siggaard-Andersen et al. (2007); Gabrielle et al. (2005) | |
| Maritime or coastal | |||
| France | |||
| Guadeloupe (Lesser Antilles) | Filtered | <0.04 | Clergue et al. (2015) |
| Brest | Filtered | 0.4 | Millot, Petelet-Giraud, et al. (2010) |
| Dax | Filtered | 0.45 | Millot, Petelet-Giraud, et al. (2010) |
| The Azores | |||
| Pico Island | Filtered | 0.051 | Louvat and Allegre (1998) |
| Sao Miguel | Filtered | 0.11 | Pogge von Strandmann et al. (2010) |
| USA | |||
| Hawaii | Filtered | 0.15 | W. Li et al. (2020) |
| Filtered | 0.075 | Pistiner and Henderson (2003) | |
| Unfiltered | 1.01 | Pistiner and Henderson (2003) | |
| Continental | |||
| France | |||
| Orleans | Filtered | 0.36 | Millot, Petelet-Giraud, et al. (2010) |
| Clermont-Ferrand | Filtered | 0.39 | Millot, Petelet-Giraud, et al. (2010) |
| Strengbach | Filtered | 0.035 | Lemarchand et al. (2010) |
| China and Tibet | |||
| Yellow River, Longman Station | Filtered | 0.87 | Gou et al. (2019) |
| Yangtze River Basin (snowmelt) | 2.62 | Ma et al. (2020) | |
| Tibetan Plateau (snowpack) | Filtered | 3.83 | Y. F. Li, Huang, et al. (2020) |
| Unfiltered | 5.54 | Y. F. Li, Huang, et al. (2020) | |
The global biogeochemical cycle of lithium, showing annual flux in 109 g/year. Estimates from the text are rounded to the nearest integer.
2.3 Terrestrial Biogeochemistry
Despite having no known biochemical role in plants, Li is found at concentrations between 0.05 and 8.0 mg/kg in plant tissues (Clergue et al., 2015; Lemarchand et al., 2010). The annual uptake and return of Li in terrestrial ecosystems are estimated at 24 × 109 g/year, on the basis of the Li content in foliage (0.2 mg/kg) and net primary production of dry matter on land (120 × 1015 g/year; Schlesinger & Bernhardt, 2020). The circulation of Li in terrestrial ecosystems is largely passive. For comparison, whereas potassium, another alkali metal, actively accumulates in the soil beneath plants, Li shows no obvious accumulations in the islands of fertility under desert shrubs (Schlesinger et al., 1996).
2.4 Ocean Li Balance
Investigations of Li cycling in the oceans span more than four decades (e.g., Edmond et al., 1979; Stoffyn-Egli & MacKenzie, 1984), with recent interest spurred by the potential for Li isotopes to shed light on continental weathering and climate change. The dominant inputs of Li to seawater are river fluxes and high-temperature hydrothermal venting, and the dominant sinks occur through incorporation of Li in clays and other minerals in authigenic sediments and altered ocean crust.
Estimates of the average flux of Li from hydrothermal vents to the ocean vary by more than a factor of 2, from 41 × 109 to 90 × 109 g/year (Elderfield & Schultz, 1996; Misra & Froelich, 2012; Pogge von Strandmann et al., 2020). Lithium is also added to seawater in fluids expelled in the forearc of subduction zones. A recent estimate by Kastner et al. (2014) suggests a global Li flux of ∼0.9 × 109 g/year for this process, within the range of an earlier estimate by Hathorne and James (2006), and in keeping with the conclusion of Stepanov (2021) that Li exhibits only minor metamorphic devolatilization. These findings contrast significantly those of Misra and Froelich (2012). Adding the forearc flux of Kastner et al. (2014) to the range of values from hydrothermal vents, the flux of Li dissolved in rivers (69 × 109 g/year), and the lesser contribution of atmospheric deposition over the oceans (15 × 109 g/year), suggests a total input of dissolved Li to the world's oceans ranging from 126 × 109 to 175 × 109 g/year.
Stoffyn-Egli and MacKenzie (1984) first noted that the total input of dissolved Li to the oceans was in excess of known sinks. They suggested that the formation of authigenic clay minerals in marine sediments that incorporate Li might resolve this imbalance—a process now known as reverse weathering. Recognizing that the δ7Li of marine sediments, which ranges from −4.3‰ to +14.5‰ (Chan et al., 2006), is substantially lower than that of seawater (+31‰), Misra and Froelich (2012) balanced the global budget for Li in the oceans by assuming a substantial uptake of Li via reverse weathering and basalt alteration, which fractionates against 7Li by 15‰ relative to seawater. Estimates of the global sink of Li by these processes range from 97 × 109 to 201 × 109 g/year (Hathorne & James, 2006; G. Li & West, 2014; Misra & Froelich, 2012). Andrews et al. (2020) find that a proportion of this flux (80 × 109 g/year) is found in authigenic clays associated with carbonate-rich sediments. Independent evidence supports preferential incorporation of 6Li by clay minerals, which presumably begins with the formation of secondary clay minerals in soils and continues in rivers and marine sediments (Lemarchand et al., 2010; Ma et al., 2020; Steinhoefel et al., 2021; J.-W. Zhang et al., 2021). Biotic uptake of Li accounts for removal of only 2–6 × 109 g/year from the oceans (Hathorne & James, 2006).
These calculations suggest an approximate Li balance in the oceans between the total dissolved inputs (126 × 109 to 175 × 109 g/year) and sinks (99 × 109 to 207 × 109 g/year), yielding a mean residence time of 1.0–1.6 million years for Li in seawater (Huh et al., 1998; Marschall et al., 2017; Misra & Froelich, 2012; Whitfield & Turner, 1979). With a mean seawater concentration of 0.179 mg/L (Bruland et al., 2014), Li is a well-mixed, conservative ion in the oceans (Nozaki, 1997).
In order to estimate the total flux of Li sequestered to the ocean floor, the flux of Li transported to the oceans in terrigenous suspended sediment, estimated above as 107 × 109 to 157 × 109 g/year, is added to the sink of dissolved Li+ incorporated into authigenic sediments and altered basalt crust. This yields a total Li sink of 206 × 109 to 364 × 109 g Li/year in ocean sediments. It is important to note, however, that studies of diverse river systems suggest that the bulk of riverine Li may be deposited near shore. Pogge von Strandmann et al. (2008) suggested that estuaries may confine 15%–25% of riverine Li, and Brunskill et al. (2003) and Wimpenny, James, et al. (2010) suggested that as little as 10% of the Li in rivers may be transported to the open ocean.
Of the total Li accumulation in ocean sediments, we estimate that 51–58 × 109 g/year are subducted and the remainder is buried in sedimentary accumulations along passive margins. The estimates of the subducted flux of Li are calculated from estimates of the total mass of subducted sediments (11.3 × 1015 g/year [Plank, 2014] to 13 × 1015 g/year [Lipp et al., 2020]), each multiplied by the mean concentration of Li in subducting sediments (45 mg/kg; Plank, 2014). Earlier, Chan et al. (2006) estimated the Li flux into subduction zones as 38 × 109 g/year. All of these estimates are compatible with the observation that globally about 10% of marine sediments are subducted and the rest accumulate in passive margins (Schlesinger & Bernhardt, 2020, p. 138).
3 Human Perturbation of the Li Cycle
3.1 Mining of Lithium Ores
The high electrochemical potential of Li and its use for rechargeable batteries in electric cars and in other appliances has raised Li demand over time (Figure 2). Rechargeable lithium-ion batteries are critical in replacing fossil fuels to power cars and trucks from renewable sources of energy (Bradley et al., 2017). In addition to the use of Li for batteries (56% of global consumption), Li is used in glass and ceramics (23% of global consumption; G. Martin et al., 2017; Sterba et al., 2019) and various other industrial products. Since mid-1990s, the global production of Li ores has increased by 15-fold; from 6 × 109 to 77 × 109 g in 2019 (Figure 2). While the major global Li deposits are located in Chile (50% of estimated global reserves), Australia (16.5%), Argentina (10%), and China (5.9%), in 2019 the major Li ore producers were Australia (58% of global production), Chile (22%), China (9.7%), and Argentina (8.3%) (British Petroleum (BP), 2020). Lithium is mined from pegmatites (e.g., Australia) that are used mostly for glasses and ceramics, as well as extracted from Li-rich brines (160–1,400 mg/L), mostly in the Salar of Uyuni, Salar de Atacama, and Salar de Hombre Muerto in South America. Lithium is delivered in the form of lithium carbonate, lithium chloride, and lithium hydroxide—all soluble and potentially mobile in the environment (Bradley et al., 2017). Given that the current global production of Li consists of only 0.1%–0.5% of the potential global Li reserves estimated as 15.5–62 × 1012 g (Bowell et al., 2020; BP, 2020), the rapid increasing demands for Li products suggest that the anthropogenic Li flux from Li mining will substantially increase during the next decades.
Variations of global Li ore production (data from USGS, 2020).
3.2 Air-Borne Emissions From Coal Combustion and Retention of Lithium in Coal Combustion Residuals
The average concentration of Li in global coals is estimated as 12 mg/kg (Ketris & Yudovich, 2009), with higher average values in coals from China (31.8 mg/kg; Dai et al., 2012) and lower values from the United States (16 mg/kg; Palmer et al., 2015). To evaluate the global Li flux from coal combustion, we use the Li contents in coals and the relative coal production in different countries compared to global production (Table 2). China, for example, produced 3,486 × 1012 g in 2019, which equivalent to 43% of the global coal production (BP, 2020). Given that 70% of China coal is mined in the northern basins of Shanxi and Inner Mongolia (Bai et al., 2018), the relatively high Li content of coal from these basins (165 mg/kg; Qin et al., 2015) has an overall effect on the calculated global Li concentration in coal, which is estimated as 68 mg/kg (Table 2). Given the increase in the global production of coal, up to 8,129 × 1012 g in 2019 (data from BP, 2020), coal combustion has doubled global Li flux from coal from ∼250 × 109 g/year in early 1980s to 550 × 109 g/year in 2019 (Figure 3).
| Country | Tons (×106)a | Percent of global coal production | Li in coal (mg/kg) | Ash yield in coals (%) | Enrichment factor in coal ash | Expected Li in fly ash (mg/kg) | Sources |
|---|---|---|---|---|---|---|---|
| China-northern basin | 2,692 | 33.1 | 165 | 21 | 4.8 | 786 | Qin et al. (2015) |
| Rest of China | 1,154 | 13.2 | 32 | 20 | 5 | 159 | Dai et al. (2012) |
| India | 756 | 9.3 | 24 | 30 | 3.3 | 79 | Tewalt et al. (2010) |
| USA | 640 | 7.9 | 16 | 11 | 9.1 | 145 | Tewalt et al. (2010) |
| Indonesia | 610 | 7.5 | 10 | 7 | 14.3 | 141 | Tewalt et al. (2010) |
| Australia | 507 | 6.2 | 26 | 9 | 10.8 | 275 | Tewalt et al. (2010) |
| Russia | 440 | 5.4 | 14 | 10 | 10 | 137 | Tewalt et al. (2010) |
| South Africa | 254 | 3.1 | 37 | 20 | 5 | 185 | Tewalt et al. (2010) |
| Rest of the world | 1,076 | 13.2 | 12 | 20 | 5 | 60 | Dai et al. (2012) |
| Global | 8,129 | 100 | 68 | 18 | 5.4 | 372 | This study |
- a Based on BP (2020) for annual production in 2019.
Based on Li variations in fly ash generated from combustion of coal in India, Bhangare et al. (2011) defined Li as a semivolatile element that is volatilized during coal combustion. In order to evaluate the magnitude of Li capture on fly ash, we analyzed the Li concentrations in fly ash samples originating from combustion of coals from the three major basins in the United States (see provenance of investigated coal ash samples, analytical methods, and their lithium concentrations in Supporting Information). A comparison of the median Li concentrations in coals from the three regions (Palmer et al., 2015) to those of the corresponding fly ashes shows a systematic enrichment of ∼10-fold, with Li concentrations of 79–170 mg/kg in fly ash (Figure 4). The ∼10-fold enrichment is consistent with the theoretical Li enrichment in fly ash, given that US coals typically have ash contents of 9%–10% (Palmer et al., 2015). Similar dependence of the Li enrichment in fly ash on the ash content of the feed coals has been shown for feed coal-fly ash comparisons from India (Bhangare et al., 2011) and China (Dai et al., 2014; J. Li et al., 2012; Wang et al., 2019; Wei & Song, 2020), suggesting almost complete partitioning of the Li in the combusted coal into the resulting fly ash.
Box and whisker plot of Li concentrations in the three major basins (Eastern, Central, and Western) in the United States (data retrieved from Palmer et al., 2015) as compared to Li in fly ash originated from corresponding feed coals (left panel); the direct relationships between Li in fly ash-feed coal pairs from the United States (right panel). The data show ∼10-fold enrichment of Li in the fly ash, which is consistent with the ash yield percent of the US coals. The boxes in boxplot represent the 25th to 75th percentiles, lines are minimum and maximum values, and circles are outliers of the data distribution.
The annual worldwide production of coal combustion residuals (CCRs), including fly ash and bottom ash, is estimated at 1,222 × 1012 g/year, which corresponds to ∼15% of the total global mass of coal burned (Harris et al., 2019). In order to calculate the expected Li concentration in global CCRs, we estimated the percentage contribution to global coal combustion for each of the major coal-producing regions (Table 2). For each region, we used averages of the Li concentrations in their parent coals, and their % ash yields, and assumed that Li in their CCRs is proportionally enriched as a function of % ash yield (as demonstrated for the US fly ash; Figure 4). This calculation yields an estimated Li concentration of 372 mg/kg in global CCRs. This estimate suggests a global enrichment factor of 5.4, which integrates the different % ash yields and production contributions from the major world coal-producing regions (Table 2).
The capture of Li on fly ash from the flue gas depends on the availability and efficiency of the electrostatic precipitators and fabric filters installed in coal plants; the lack of this infrastructure would result in atmospheric Li emission. While in many countries the installation of these scrubber systems is mandatory to prevent the fugitive emission of fly ash particles, many studies have reported release of fly ash particles to the atmosphere and deposition on surface soils, as demonstrated in India (Mandal & Sengupta, 2006; Praharaj et al., 2003), China (Huang et al., 2017; Lu et al., 2013; Tang et al., 2013; Y. Zhang et al., 2020), and the United States (Sears & Zierold, 2017; Zierold & Odoh, 2020). Consequently, we estimate that 10% of the Li in CCRs is released to the atmosphere in particles, inferring a global Li flux of 55 × 109 g/year in 2019 (i.e., 10% of the Li flux from coal combustion escapes as fly ash aerosols). For comparison, we estimate that the losses of other elements to the atmosphere during coal combustion range from 1% to 2% for V (Schlesinger et al., 2017) to 27% for F (Schlesinger et al., 2020).
Given the large volume of CCR waste that is generated from coal combustion, estimated as 1,222 × 1012 g/year (Harris et al., 2019), CCR in many countries is reused, particularly for concrete, cement, and structural fills (Gollakota et al., 2019; Harris et al., 2019; Luo et al., 2020). The relative fraction of utilized CCR varies among countries; from 38% in India to 70% in China (Gollakota et al., 2019; Yao et al., 2015). Based on the differential reuse of coal ash in the major coal combustion countries, we estimate that in 2019, 64% of global CCRs (782 × 1012 g) were reused, while the remaining 440 × 1012 g of CCRs was disposed in coal ash impoundments and landfills.
The interaction of CCRs with water mobilizes trace elements into leachates of the CCRs, which upon release to the environment through permitted releases, leaking, or spills can contaminate associated rivers, lakes, and groundwater (e.g., Brandt et al., 2018; Harkness et al., 2016; L. Ruhl et al., 2012; Vengosh et al., 2019). Elevated Li concentrations have been reported for effluents discharged from coal ash ponds in North Carolina, with concentrations up to 445 μg/L, which are over 100-fold higher than Li in upstream waters that feed these plants (L. Ruhl et al., 2012). We conducted leaching experiments to evaluate the water-soluble Li extracted from fly ash originating from combustion of the major coal sources in the United States, following a modified version of EPA Method 1316 reported in Wang et al. (2020) (see Methods in Supporting Information). In the United States, the concentration of water-soluble Li in fly ash is between 3.9 and 13.2 mg/kg, which reflects mobilization of 4%–12% of the total Li in fly ash. This can be compared to the magnitude of boron mobilization from fly ash (12%–29%), which is a known tracer for coal ash contamination of water resources (Harkness et al., 2016; L. Ruhl et al., 2012; R. L. Ruhl et al., 2014). Using the predicted Li concentration in global CCRs (372 mg/kg; Table 2) and the estimated global quantity of disposed CCRs (440 million tons), we calculate that 164 × 109 g Li/year is temporarily stored in CCR impoundments or released to the environment. Assuming that 4%–12% of fly ash Li is mobilized into the aqueous phase; this suggests that 6.5–19.6 × 109 g Li/year is mobilized into the aquatic systems (Table 3).
| Process | Total | Dissolved lithium delivered to freshwaters |
|---|---|---|
| Rock weathering | 201 | 69 |
| Ore mining | 77 | |
| Coal combustion | 550 | |
| Fugitive emission to the atmosphere | 55 | |
| Disposed coal ash | 163 | |
| Mobilization from CCR to aquatic phase | 6.5–19.5 | |
| Oil-produced water | 660 | |
| Released to the oceans | 200 | |
| Reinjected on land | 460 | |
| Released to freshwaters | 46 | |
| Natural gas produced water | 0.038 | |
| Global groundwater utilization | 29 | |
| Release to the environment from therapeutic drugs | 0.84 | |
| Other releases from products and chemicals (estimated) | 1 | |
| Sum | 145–165 |
- Abbreviation: CCR, coal combustion residual.
3.3 Oil-Produced Water
The production of oil and gas is typically accompanied by produced water that becomes oil and gas wastewater. High concentrations of Li have been reported in oilfield brines from Arkansas, North Dakota, Oklahoma, Texas, Wyoming, and Gulf Coast region, with Li concentrations in some cases as high as 700 mg/L (Bradley et al., 2017; Collins, 1976). A systematic survey of available data of Li content in produced waters from conventional and unconventional oil and gas wells in the United States and China shows a wide range of Li concentrations, up to 300 mg/L (e.g., Silurian produced water from the Appalachian Basin), with a median value of 40 mg/L for Li in produced water.
Clark and Veil (2009) and Grubert and Sanders (2018) estimated an average produced water-to-oil ratio in the United States is ∼5:1 and 8:1, respectively, but global estimates are lower, typically 3:1 (Ahmadun et al., 2009). Using the global water-to-oil ratio and the median Li concentration of 40 mg/L in produced water, the increase in oil production from 3.5 × 109 m3 in early the 1980s to 5.5 × 109 m3 in 2019 (BP, 2020) indicates an increase of produced water from 10.4 × 109 to 16.6 × 109 m3, with the Li flux from 400 × 109 to 660 × 109 g/year (Figure 5).
Annual changes in global crude oil production (BP, 2020) and Li flux in oil-produced water, assuming a ratio of produced water to oil of 3:1 and a Li concentration of 40 mg/L in produced water.
For natural gas, the ratio of produced water to conventional gas production in the United States is estimated as 2.4 × 10−4, using data from Grubert and Sanders (2018). Assuming that this ratio applies globally, the increase of global natural gas production from 1,450 × 109 m3 in early 1980s to 4,000 × 109 m3 in 2019 (BP, 2020) suggests an increase in the volume of produced water from 0.35 × 109 to 0.96 × 109 m3, ∼17-fold less than that of oil-produced water. Using the same Li concentration of 40 mg/L in produced water, the Li flux from produced water coextracted with natural gas is estimated to range from 0.014 × 109 to 0.038 × 109 g/year and thus insignificant among anthropogenic mobilizations.
Because off-shore produced water comprises about 30% of global produced water (Dal Ferro & Smith, 2007), and recognizing the customary lack of any treatment of produced water generated offshore, we suggest an anthropogenic release of ∼200 × 109 g Li/year to the oceans (based on 2019 oil production data). The Li isotope composition of oilfield water is typically high and overlaps with the δ7 Li of seawater (+31‰). We therefore expect off-shore discharge of oil and gas wastewater to have only a small impact on the Li isotope composition in seawater. On-shore, most of the oil and gas wastewater is disposed in the subsurface through deep injection wells and/or reused for oil recovery enhancement (e.g., >95% in the United States; Veil, 2015). Assuming that 10% of the global on-shore produced water is discharged to the surface suggests a flux of ∼46 × 109 g/year to freshwaters (Table 3).
In many parts of the world, oil and gas exploration has resulted in wastewater spills. In the Appalachian Basin (Pennsylvania) and Williston Basin (North Dakota) in the United States, wastewater spills and the frequency of spills have been directly linked to the density of unconventional oil and gas wells (Lauer et al., 2016; Vengosh et al., 2014). In the Williston Basin in North Dakota, for example, the volume of oil wastewater spills increased from 2,600 m3 in 2007 to 17,700 m3 in 2015, resulting in lithium concentrations up to 3.5 mg/L in contaminated waters (Lauer et al., 2016). In some parts of the United States, oil and gas wastewater is released to the environment without adequate treatment. The estimated annual volume of produced water generated on-shore in the United States in 2012 was 3.17 × 109 m3, of which 96 × 106 m3 (3%) was discharged into surface water (Veil, 2015), causing local river contamination as demonstrated in several cases in Pennsylvania (Harkness et al., 2015; Warner et al., 2012, 2013). While the global magnitude of spilled and discharged oil and gas wastewater is unknown, the common high concentrations of lithium in these wastewaters infer additional, yet unquantified, of lithium flux to the hydrosphere.
3.4 Lithium in Agricultural Soil Additives
Millot, Petelet-Giraud et al. (2010) suggested that some differences in the concentration of Li in precipitation across France might be due to regional differences in fertilizer use, including applications of lime. Some Li added to agricultural soils may be emitted as fugitive dust and some adds to the load of dissolved Li in runoff waters. Senesi et al. (1999) suggest a typical concentrations of <1 mg Li/kg in various fertilizers (N, P, and K) and Tomascak et al. (2016) report ∼1 mg/kg of Li in typical limestones. If we multiply the annual global application of fertilizers (144 × 1012 g N; 222 × 1012 g phosphate rock; 43 × 1012 g potash; and 430 × 1012 g lime; USGS, 2020) by 1 mg/kg, the total Li added to agricultural fields sums to <1 × 109 g/year—a trivial value in its global cycle. However, additional Li emissions to freshwaters may accompany its future use as a catalyst in the industrial fixation of nitrogen for fertilizer (e.g., Suryanto et al., 2021).
3.5 Water Used in Geothermal Power Generation
Typically, geothermal power plants use high-temperature fluids extracted from the subsurface to generate electricity. Depending on the geologic setting and type of power plant, waste fluids can contain high concentrations of dissolved solids that must be condensed and removed prior to reinjection into the subsurface (for improved fluid circulation and/or waste disposal), or released or stored on the surface (e.g., Bayer et al., 2013; Baysal & Gunduz, 2016; Clark et al., 2013; Kaya et al., 2011). A number of studies have documented elevated Li contamination of local surface water or groundwater associated with geothermal power plants (e.g., Aksoy et al., 2009; Dogdu & Bayari, 2005). With growing worldwide demand for Li (G. Martin et al., 2017), there is increasing interest in extracting the Li in geothermal wastewaters, although to date this has proven economically viable in only limited cases (Finster et al., 2015; Neupane & Wendt, 2017; Paranthaman et al., 2017). Paranthaman et al. (2017) estimate that the high saline fluids from the Salton Sea Geothermal Area in the United States could potentially yield 120,000 tons of Li carbonate annually (23 × 109 g Li/year), representing ∼30% of 2019 global annual Li consumption (77 × 109 g/year; USGS, 2020). We do not include the flux of Li from geothermal waste waters in our compilation because quantitative global information is not available. Should efforts to extract Li from these wastes increase in coming years, this may represent a significant flux of Li to industrial products and the environment.
3.6 Global Groundwater Withdrawal
Recent estimates of global groundwater withdrawal vary between 734 and 982 km3/year (Margat & van der Gun, 2013; Wada et al., 2010). The highest extraction rate occurs in the Indo-Gangetic Basin of northwestern India and Pakistan, which accounts for ∼25% of worldwide groundwater withdrawal (Gleeson et al., 2012; MacDonald et al., 2016; Wada et al., 2012).
Lithium concentrations in global groundwaters are highly variable; the median Li concentration of groundwater in the United States is 8 μg/L (n = 1,464), whereas groundwater from the sedimentary basins in arid zones of the western United States is about 50 μg/L (Ayotte et al., 2011; Lindsey et al., 2021). Groundwaters extracted from metamorphic and igneous rocks of the Piedmont region of North Carolina have a median Li concentrations of 4 μg/L (n = 596; Coyte et al., 2020). In northwestern India, Li concentrations vary between 14 and 28 μg/L in sedimentary alluvium and limestone aquifers, and 20 and 30 μg/L in mostly felsic igneous rocks (Coyte et al., 2019).
To estimate the flux of Li from extraction of groundwater to the surface, we assume an average global groundwater withdrawal rate of 900 km3/year and that 50% of the extracted global groundwater contains a Li concentration of 10 μg/L, while the other half of the extracted groundwater has a concentration of 50 μg/L. Combining the two types of groundwater yields a global Li flux of 29 × 109 g/year from groundwater withdrawal. Presumably, all of this Li is discharged into stream waters (Table 3).
3.7 Lithium as a Therapeutic Drug
The USGS (2020) indicates that ≤1% of the global annual production of Li is used in therapeutic drugs, amounting to about 0.8 × 109 g/year worldwide. As an alternative estimate, about 2.3 × 106 people in the United States suffer from some form of bipolar disorder that might respond to a Li regimen, often 1 g/day (Reports and Data Inc., 2020), and perhaps accounting for as much as 0.84 × 109 g Li/year escaping to the environment in wastewater. Inasmuch as Li is not metabolized, we assume that all therapeutic Li makes its way to treated or untreated wastewaters. These rough estimates of the contribution of Li drugs to wastewaters are each about 1% of the annual transport of Li in the world's rivers, although certainly some rivers carry much more than others. Choi et al. (2019) document a doubling of Li concentrations in the Han River as it flows through Seoul, South Korea. Using the δ7Li of potential source waters, they conclude that lithium-ion batteries and therapeutic drugs were the likely sources of Li enrichment in the river.
3.8 Current and Potential Recycling of Lithium
Many industrial uses of Li, including in ceramics and glass, are likely to retain Li in long-lived products and waste materials, which act as a sink for Li at the surface of the Earth. Lithium-ion batteries dominate the industrial use of Li. Until recently, most of these were small batteries, widely dispersed in the economy, and without organized efforts for recycling. For instance, each year about 1 billion cell phones are produced, each containing about 1–3 g of Li in batteries, for a total up to 3 × 109 g/year. The content of Co and Ni in Li batteries is of greater economic value than Li, so currently only about 1% of the industrial use of Li is recycled.
The vast majority of appliances containing Li batteries are still in active use, so the amount of material available for recycling is relatively small. With the proliferation of electric vehicles, with batteries containing up to 12 kg of Li, the scale and economics of recycling Li batteries are likely to increase rapidly in the coming years. Novel approaches to recovering Li from battery wastes are under active development (Zhao et al., 2020), and we can expect that recovery of Li will increasingly supply a large fraction of the current annual demand (Junne et al., 2020).
4 Conclusions
Our calculations of the anthropogenic perturbation of the global Li cycle show enhanced releases to freshwaters from oil-produced water (46 × 109 g/year), leaching of coal ash (7–20 × 109 g/year), and extraction of groundwaters (29 × 109 g/year; Table 3). Currently, releases from the excretion of therapeutic drugs and disposal of lithium-ion batteries are a small component of the transport of Li in rivers, although the latter may increase markedly as lithium-ion batteries continue to grow in use without concomitant recycling. Our values for the global flux of dissolved Li in rivers are largely derived from data taken several decades ago, and we can anticipate increasing flux in the future (e.g., Gou et al., 2019). The sum of the dominant anthropogenic inputs to rivers (ranging from 82 × 109 to 95 × 109 g/year) exceeds the measured natural transport of dissolved Li to the sea in rivers (69 × 109 g/year) by 19%–38%. The simplest explanation for this is that most of the anthropogenic input to freshwaters is rapidly precipitated or adsorbed onto clay and other minerals and not measured in the flux of dissolved Li to the oceans, as is also concluded for the Li flux in suspended and bed load sediments. All these estimates will improve with more extensive data.
Human emissions of Li particles to the atmosphere, 55 × 109 g/year from coal combustion, comprise about 38% of the emission of Li to the atmosphere from various sources (Figure 1). The inputs to the atmosphere are greater than the estimated deposition of Li from the atmosphere, which is poorly constrained by available data. Particular attention should be given to particles that are deposited near sites of emission, thus avoiding global atmospheric transport.
The total human mobilization of Li from the Earth's crust, >1,000 × 109 g/year (Table 3), is much larger than Li mobilized from the natural processes of chemical and mechanical weathering (94 and 132 × 109 g/year, respectively), representing a ∼500% perturbation of the global cycle of Li by human activities. This estimate is much larger than the magnitude of the perturbation of the cycle calculated by Sen and Peucker-Ehrenbrink (2012) in their compilation of human impacts to various elemental cycles. This difference reflects our inclusion of Li mobilized in the waste waters and solids associated with fossil fuel use, in addition to that mobilized by mining Li resources. Anthropogenic activities more than double the flux of Li to freshwaters (anthropogenic/natural = 114%; Table 3), exceeding the magnitude of the human perturbation of N (59%) and P (100%) transport to the sea each year (Schlesinger & Bernhardt, 2020). This perturbation is likely to increase in the coming years. The increase in dissolved Li and in δ7Li in river waters can be expected to alter the Li and δ7Li of global mean seawater over long time periods.
Acknowledgments
The authors thank Terry Plank for helpful conversations about lithium partitioning in various environments, Jérôme Gaillardet and Philip Pooge von Strandmann for helpful reviews of our draft manuscript, and Stan Coffman for drafting of Figure 1.
Open Research
Data Availability Statement
Being a review paper, most of the data used in this review and synthesis are accessed via the citations in the manuscript, linked to the references below. The Supporting Information contains data on the measurements of Li in coal ash leachates.
