Geochemistry and Cosmochemistry of Potassium Stable Isotopes

Abstract

Stable potassium isotopes are one of the emerging non-traditional isotope systems enabled in recent years by the advance of Multi-Collector Inductively-Coupled-Plasma Mass-Spectrometry (MC-ICP-MS). In this review, we first summarize the geochemical and cosmochemical properties of K, its major reservoirs, and the analytical methods of K isotopes. Following this, we review recent literature on K isotope applications in the fields of geochemistry and cosmochemistry. Geochemically, K is a highly incompatible lithophile element, and a highly soluble, biophile element. The isotopic fractionation of K is relatively small during magmatic processes such as partial melting and fractional crystallization, whereas during low-temperature and biological processes fractionation is considerably larger. This resolvable fractionation has made K isotopes promising tracers for a variety of Earth and environmental processes, including chemical weathering, low-temperature alteration of igneous rocks, reverse weathering, and the recycling of sediments into the mantle during subduction. Sorption and interactions of aqueous K with different clay minerals during cation exchange and clay formation are likely to be of fundamental significance in generating much of the K isotope variability seen in samples from the Earth surface and samples carrying recycled surface materials from the deep Earth. The magnitude of this fractionation is process- and mineral-dependent. Comprehensive quantification of pertinent K isotope fractionation factors is currently lacking and urgently needed. Significant fractionation during biological activities, such as plant uptake, demonstrates the potential utility of K isotopes in the study of the nutrient cycle and its relation to the climate and various ecosystems, enabling new and largely unexplored avenues for future research.

Of significant importance to the cosmochemistry community, K is a moderately volatile element with large variations in K/U ratio observed among chondrites and planetary materials. As this indicates different degrees of volatile depletion, it has become a fundamental chemical signature of both chondritic and planetary bodies. This volatile depletion has been attributed to various processes such as solar nebula condensation, mixing of volatile-rich and -poor reservoirs, planetary accretional volatilization via impacts, and/or magma ocean degassing. While K isotopes have the potential to distinguish these different processes, the current results are still highly debated. A good correlation between the K isotope compositions of four differentiated bodies (Earth, Mars, Moon, and Vesta) and their masses suggests a ubiquitous volatile depletion mechanism during the formation of the terrestrial planets. It is still unknown whether any of the K isotopic variation among chondrites and differentiated bodies can be attributed to inherited signatures of mass-independent isotopic anomalies.

1. Introduction

The alkali metal potassium (K) is a minor constituent of the Solar System, but a major component of the Earth’s crust. Its abundance is ranked 20^th out of all elements in the entire Solar System (Palme et al., 2014a), and 8^th in the Earth's crust and seawater (Millero, 2014; Rudnick and Gao, 2014). There are three naturally-occurring isotopes of potassium, ³⁹K, ⁴⁰K, and ⁴¹K, of which ³⁹K and ⁴¹K are stable isotopes, accounting for 93.2581% (atomic percentage) and 6.7302% of all potassium, respectively (Garner et al., 1975b). The isotope ⁴⁰K is radioactive, accounting for only 0.0117% of all potassium. It has a long half-life (1.248(3) × 10⁹ years [Kossert and Günther, 2004]) and its branching decay products are ⁴⁰Ca (89.25%) and ⁴⁰Ar (11.55%). The decay of ⁴⁰K was one of the primary sources of radiogenic heating (after ²⁶Al and ⁶⁰Fe) during the Earth’s early differentiation, and at the present day, still accounts for nearly 20% of the total radiogenic heat production of the Earth, following ²³²Th and ²³⁸U (McDonough et al., 2020). Since the 1950s, the ⁴⁰K-⁴⁰Ca-⁴⁰Ar isotope systems (K-Ar, Ar-Ar, etc.) have been extensively applied to radioactive dating (Marshall and DePaolo, 1982; McDougall and Harrison, 1999; Dickin, 2018). Potassium is a representative Group IA alkali metal element of the periodic table, with other metal elements belonging to the same group including Li, Na, Rb, Cs and Fr. Sodium and K are major alkali metals, and their geochemical cycles are closely related to a wide range of low- and high-temperature processes because of their common presence in nature. However, our current knowledge on the cycling of these two elements is considerably limited compared to our understanding of many other major elements in the periodic table (e.g., Berner and Berner, 2012). For example, fluxes of K inputs and outputs in the modern ocean are still poorly quantified (Bloch and Bischoff, 1979; Jarrard, 2003; Holland, 2005; S. Li et al., 2019a). Stable isotope ratios are known to be valuable tools that can provide unique information on geochemical cycles of an element invisible to its concentration alone. Sodium is monoisotopic (²³Na). Lithium and Rb have two stable isotopes, but both are trace elements and have properties that are slightly different from Na and K. As a result, the K stable isotope system holds a particularly significant role in elucidating the geochemical behavior of alkali elements in natural systems.

For a long time, K stable isotopes had relatively few applications in geochemistry and cosmochemistry due to the significant analytical challenges in achieving a precision high enough to resolve natural isotopic variability (e.g., Humayun and Clayton, 1995a, 1995b). However, since 2012 high-precision K stable isotope analyses have become achievable as a result of advancements in MC-ICP-MS techniques (Li et al., 2016; Wang and Jacobsen, 2016a; Hu et al., 2018a; Morgan et al., 2018; Chen et al., 2019b, 2021; Moynier et al., 2021). Recent studies have shown that K stable isotopes have many applications in geochemistry such as surface weathering, hydrothermal alteration, oceanic plate subduction, and global elemental cycling, as well as in cosmochemical investigations such as solar nebula condensation, planetary formation, and magma ocean evolution (e.g., Wang and Jacobsen, 2016b; Parendo et al., 2017; Li et al., 2019; Sun et al., 2019). Teng et al. (2017) first briefly summarized K isotopes together with other “emerging” stable isotope systems (i.e., Ti, V, Rb, Sr, Cd, Sn, Sb, Te, Ba, and Nd). Recently, we also published the first K isotope review article in Mandarin Chinese (Wang et al., 2020b), which was primarily intended to introduce this new isotope system to the geochemistry community in China. Nevertheless, the field of K isotopes is growing fast, and many exciting new results have emerged since the publication of our previous review. This updated review highlights additional new developments and discoveries, while also being written for a broader international audience. The intent of this review is to provide the most up-to-date information on the nucleosynthetic origin of K, the geochemical and cosmochemical properties of K, the history of K isotope research, K isotope analytical methods, and the application of K isotopes to low-temperature geochemistry, biogeochemistry, and high-temperature geochemistry and cosmochemistry.

2. Stellar nucleosynthetic and cosmogenic origins of potassium

Both the stable isotopes of K (³⁹K and ⁴¹K) are produced by oxygen burning in massive stellar explosions (e.g., Type II supernovae; Clayton, 2003). While ³⁹K is made directly during this process, ⁴¹K is synthesized first as radioactive ⁴¹Ca, which then shortly decays to ⁴¹K. The half-life of ⁴¹Ca is ~0.1 Ma, the shortest among the radionuclides used as chronometers (e.g., ²⁶Al, ⁵³Mn, ⁶⁰Fe). The half-life was initially estimated as 1.04±0.05 × 10⁵ years (Kutschera et al., 1992), but was recently redetermined as 0.994±0.015 × 10⁵ years (Jörg et al., 2012). Based on the observation of excess ⁴¹K in Ca-Al-rich refractory inclusions (CAIs), it was revealed that not all ⁴¹Ca decayed to ⁴¹K during the earliest stage of the Solar System prior to the formation of the first solids (Srinivasan et al., 1994). Due to the short half-life of ⁴¹Ca, this observation has been interpreted as an injection of freshly synthesized ⁴¹Ca (along with other short-lived radionuclides such as ²⁶Al and ⁶⁰Fe) from a nearby asymptotic giant branch (AGB) star, which may have triggered the formation of the Solar System (Wasserburg et al., 1995). Alternatively, CAIs could have been irradiated by the young Sun, causing the fossil ⁴¹Ca (now excess ⁴¹K) to be generated locally by the cosmic-ray effect within our Solar System (see below). A third possible explanation is that the fossil ⁴¹Ca in CAIs was inherited from parental molecular clouds (Liu et al., 2012; Young, 2014). The radioactive ⁴⁰K can be produced by two stellar processes (Clayton, 2003): 1) oxygen burning in massive stellar explosions, which is the same formation process as the stable K isotopes; and 2) slow neutron-capture process (s-process) via ³⁹K (n, γ)⁴⁰K reaction. The latter process can occur at any time when free neutrons are available in stars (Clayton, 2003). Both processes are extremely inefficient in the production of ⁴⁰K, making it the least abundant isotope among the three naturally occurring K isotopes.

Galactic and solar cosmic-ray irradiation can also produce all three natural-occurring isotopes of K. This “cosmic-ray effect” includes two types of reactions: 1) cosmic-ray-induced spallation by primary particles; and 2) neutron capture reactions by secondary thermal neutrons, which were first generated by the primary spallation. In the primary spallation reaction, high-energy cosmic-ray particles (e.g., proton) collide with a target nucleus (mainly Fe and Ni due to their large spallation cross-sections) to induce nuclear reactions and produce cosmogenic nuclides. For example, ⁴¹K can be produced via the spallation reaction: ⁵⁶Fe(p, ¹²Cα)⁴¹K. For the neutron capture reaction, secondary thermal neutrons that were generated by primary cosmic-ray spallation can be captured by other nuclides. For example, ⁴¹Ca can be produced via the neutron capture reaction: ⁴⁰Ca(n, γ)⁴¹Ca. This newly formed ⁴¹Ca radionuclide subsequently decays to ⁴¹K. All three K isotopes produced by cosmic-ray irradiation have been found in iron meteorites due to their high abundances of the target elements (Fe, Ni), negligible indigenous K contents, and often long exposure-ages (e.g., Voshage, 1978; Voshage et al., 1983). Cosmogenic live ⁴¹Ca has been observed in stony meteorites (Bogard et al., 1995; Herzog et al., 2011); however, no excess ⁴¹K has yet been detected due to the high indigenous K contents in stony meteorites.

3. Chemical properties of potassium

3.1. Cosmochemical properties

In the cosmochemical classification, elements can be classified as refractory, major, moderately volatile, volatile, and highly volatile elements according to their 50% condensation temperature (T_c) under solar nebular conditions (Palme et al., 2014a). Refractory elements are those with T_c greater than 1355 K, such as Al, Ti, Ca, U and Sr. Major elements have T_c between 1355 K and 1250 K, such as Mg, Si, Ni and Fe. Moderately volatile elements are those with T_c between 1250 K and 665 K, such as K, Na, Cl, Rb, Zn and Cu. Volatile elements and highly volatile elements are elements with T_c less than 665 and 252 K, respectively, such as H, O, N, C and the noble gas elements. Potassium is classified as a moderately volatile element as its T_c is 1006 K (Lodders, 2003).

As first shown by Wasserburg et al. (1964), although K and U concentrations vary dramatically among different igneous rocks on Earth, the K/U ratios remain essentially constant. Because both K and U are large ion lithophile elements, the K/U ratio does not change readily during magmatic processes (e.g., metal-silicate differentiation, partial melting). However, K and U have different cosmochemical properties (K is a moderately volatile and U is a refractory element), hence the K/U ratio is susceptible to change during planetary processes such as solar nebula condensation, planetary volatilization, and impact-induced degassing. The K/U ratios of igneous rocks, therefore, are characteristic for each planetary body. Additionally, the K/U ratio facilitates the evaluation of degrees of volatile loss from planetary bodies relative to the solar initial, as defined by the solar photosphere composition (see Figure 1). Table 1 summarizes K contents, K/U ratios, and Rb/Sr ratios in major meteorite types and planetary materials in the Solar System. Similar to the K/U ratio, the Rb/Sr ratio (Rb is moderately volatile whereas Sr is refractory) also serves as an indicator of volatile loss and correlates well with the K/U ratio (see Figure 2). Ivuna-like carbonaceous chondrites (CI) resemble the solar photosphere composition. The K/U ratios of other carbonaceous chondrites decrease gradually relative to CI chondrites from CI to CK groups. Earth, the Moon, Vesta and the angrite parent body (APB) are also depleted in the moderately volatile element K (and other moderately volatile elements such as Na, Cl, Cu, Zn and Rb) relative to CI chondrites. For example, Earth has a K/U ratio of 13,800 (Arevalo et al., 2009), which is about five-fold smaller than the K/U ratio of CI carbonaceous chondrites (67,400; Palme et al., 2014b). Mars has been estimated to have a higher K/U ratio (20,000) than Earth (Yoshizaki and McDonough, 2020), which has been argued as an indication that Mars is a more volatile-rich planet (Dreibus and Wanke, 1985). The Moon has a K/U ratio of 2,500 (Taylor, 1982), which is a depletion of about five times relative to the Earth. To date, however, the causes and mechanisms of moderately volatile element loss among planetary bodies (Earth, Moon, Vesta, and angrite parent body [APB]) relative to the bulk composition of the Solar System remains unknown and is a topic of vigorous research.

Figure 1.

The K/U ratios in different chemical groups of chondrites and planetary bodies (Earth, Venus, Mars, Mercury, Moon, Vesta and APB) normalized to the K/U ratio in CI chondrites. The ratios of the solar photosphere are also shown. All data are from the compilation (see Table 1).

Table 1.

Potassium concentrations in meteorites and planetary materials

		K concentration (ppm)	K/U (ppm/ppm)	Rb/Sr (ppm/ppm)	Reference
Solar photosphere			>64,000	0.426	(Lodders, 2003, 2021)
Carbonaceous chondrites	CI	546	67,400	0.298	(Palme et al., 2014a; Alexander, 2019a)
	CM	403	40,300	0.170	(Wasson et al., 1988; Friedrich et al., 2002; Hewins et al., 2014; Alexander, 2019a)
	CO	345	26,500	0.118	(Wasson et al., 1988; Alexander, 2019a)
	CR	303	23,300	0.110	(Lodders and Fegley, 1998; Stracke et al., 2012; Alexander, 2019a)
	CK	285	19,000	0.098	(Kallemeyn et al., 1991; Lodders and Fegley, 1998; Alexander, 2019a; Lodders, 2021)
	CV	310	18,200	0.083	(Jarosewich et al., 1987; Wasson et al., 1988; Stracke et al., 2012; Alexander, 2019a)
Ordinary chondrites	H	780	65,000	0.290	(Wasson et al., 1988; Alexander, 2019b)
	L	825	63,500	0.279	(Wasson et al., 1988; Alexander, 2019b)
	LL	790	60,800	0.279	(Wasson et al., 1988; Alexander, 2019b)
Enstatite chondrites	EH	800	88,900	0.334	(Wasson et al., 1988; Barrat et al., 2014; Alexander, 2019b)
	EL	735	88,600	0.286	(Wasson et al., 1988; Barrat et al., 2014; Alexander, 2019b)
Earth		280	13,800	0.0528	(Kargel and Lewis, 1993; Arevalo et al., 2009)
Venus			7,000		(Surkov et al., 1987)
Mars		360	20,000	0.0682	(Yoshizaki and McDonough, 2020)
Mercury			12,800		(Peplowski et al., 2011)
Moon		83	2,500	0.0093	(Taylor, 1982)
Vesta		83	3,800	0.0105	(Jones, 1984; Kitts and Lodders, 1998; Lodders and Fegley, 1998)
Angrite		28	176	0.0010	(Tera et al., 1970; Philpotts and Schnetzler, 1971; Ma et al., 1977; Mittlefehldt and Lindstrom, 1990; Prinz et al., 1990; Nyquist et al., 1994; Warren et al., 1995; Riches et al., 2012; Baghdadi et al., 2015; Sanborn et al., 2015)

Figure 2.

The K/U ratios versus Rb/Sr ratios in different chemical groups of chondrites and bulk planetary bodies (Earth, Mars, Moon, Vesta and APB) normalized to the K/U ratio and Rb/Sr in CI chondrites. All data are from the compilation (see Table 1).

3.2. Geochemical properties

As a lithophile element, K is preferentially incorporated into the silicate Earth (mantle + crust) during mantle-core differentiation. The metal-silicate partition coefficient for K (D=wt.%K_metal/wt.%K_silicate) is generally low; for example, D has been experimentally determined to be 0.0026 at 1450°C and 1.5 GPa, and becomes smaller at lower temperatures and pressures (Murrell and Burnett, 1986). Due to uncertainties associated with large extrapolation of laboratory results to relevant higher temperatures and pressures, and the varied dependence of D not only on T-P but also the amount of other light elements (e.g., O and Si) in the system, precise constraints on D under mantle-core differentiation conditions are less certain (e.g., Ito et al., 1993; Gessmann and Wood, 2002; Hirao et al., 2006; Watanabe et al., 2014). Earlier studies indicated that a considerable amount of K (~250 ppm) could be present in the core, and radioactive decay of ⁴⁰K from such quantities could serve as a significant heat source that sustains the geodynamo (e.g., Gessmann and Wood, 2002; Murthy et al., 2003; Bouhifd et al., 2007). However, an increasing body of evidence based on laboratory experiments and theoretical calculations suggest only a few ppm to up to ~40 ppm K is likely in the core (e.g., Corgne et al., 2007; Watanabe et al., 2014; Blanchard et al., 2017; Xiong et al., 2018). This amount is negligible compared to the average concentration of 160 ppm K in the bulk Earth (McDonough, 2014) and 240 ppm in the bulk silicate Earth (McDonough and Sun, 1995) (Table 2). The detection of geoneutrinos emitted from the decay of ⁴⁰K as well as subtracting the signals from the crust and mantle would help to constrain the amount of the ⁴⁰K in the core; however, presently any signal from ⁴⁰K is still below the current detection limit (Gando et al., 2011; Agostini et al., 2020). As such, K could be one of the potential light elements in the core, but the effect of ⁴⁰K radioactivity on the core heat flow is negligible (Corgne et al., 2007).

Table 2.

Potassium elemental concentrations in major Earth’s reservoirs

Reservoirs		Mass fraction (%)a	Concentration of K (ppm)	Fraction of total K (%)
	Riverb	0	2.3	0
	Oceanc	0.023	399	0.049
CRUST		0.435	12930	30.3
	Continental crust	0.345	16133	30.0
	continental sedimentd	0.013	17765	1.2
	upper continental cruste	0.113	23244	14.2
	middle continental cruste	0.114	19093	11.7
	lower continental cruste	0.105	5064	2.9
	Oceanic crust f	0.090	651	0.32
	mid-ocean ridge basalt f		1237
	oceanic island basaltg		6890
MANTLE		67.7	191	69.6
	upper mantleh	17.5	50	4.7
	lower mantlei	50.2	240	64.9
CORE j		32.3	0–40	0

^a(Yoder, 1995; Schubert et al., 2001; Peterson and Depaolo, 2007)

^b(Livingstone, 1973)

^c(Culkin and Cox, 1966)

^d(Wedepohl, 1995)

^e(Rudnick and Gao, 2014)

^f(White and Klein, 2014)

^g(Taylor and McLennan, 2003)

^h(Workman and Hart, 2005)

^I(McDonough and Sun, 1995)

^j(McDonough, 2014)

Potassium is an incompatible element with an ionic radius much larger than that of the major cations (i.e., Mg and Fe) in olivine and pyroxene. Its partition coefficient between olivine (or pyroxene) and melt is generally less than 0.01 (GERM database: https://kdd.earthref.org/KdD; Philpotts and Schnetzler, 1970; Hart and Dunn, 1993). During mantle partial melting, K is preferentially partitioned into the melt. As a result, K is much more abundant in Earth's crust than the mantle. As shown in Table 2, the average content of K in the lower mantle is estimated to be 240 ppm (McDonough and Sun, 1995), whereas the average K content in the (melt depleted) upper mantle is estimated to be 50–60 ppm (Salters and Stracke, 2004; Workman and Hart, 2005). In contrast, K is enriched in the crust, e.g., up to an average of 2.80% (K₂O) in the upper continental crust (Rudnick and Gao, 2014). Further, as magma evolves, K becomes gradually concentrated in felsic rocks (e.g., granite, pegmatites), forming K-rich minerals such as potassium feldspar, biotite, and muscovite. Table 3 summarizes common K-bearing minerals and their occurrences.

Table 3.

Common K-bearing minerals and their bond lengths

Mineral		Formula	Common Occurrence	K stoichiometric concentration (wt.%)	Force constant (N/m)a	Mean bond length (Å)a	CNb
Silicates	microcline	KAlSi3O8	Granite	14.05	48.92	2.895	7
	leucite	KAlSi2O6	Alkalic rock	17.91	35.31	2.955	6
	nepheline	Na3KAl4Si4O16	Alkalic rock	6.69	51.90	2.992	9
	muscovite	KAl2(AlSi3O10)(OH)2	Granite	9.82	59.93	2.871	6
	lepidolite	KLi2AlSi4O10(OH)2	Pegmatites	10.12	43.59	3.022	6
	phlogopite	KMg3AlSi3O10(OH)2	Ultramafic/metamorphic	9.37	44.85	2.961	6
Carbonates	potassium carbonate	K2CO3		56.58	63.82	2.832	9
	potassium bicarbonate	KHCO3		39.05	66.15	2.848	8
Halide	sylvite	KCl	Evaporite	52.45	44.06	3.144	6
Nitrate	niter	KNO3	Soil	38.67	89.51	2.876	9
Sulfate	alunite	KAl3(SO4)2(OH)6	Sulfate ore	9.44	112.32	2.825	12
Sulfide	djerfisherite	K6CuFe24S26Cl	Meteorite	9.38	34.76	3.343	9

^aData from (Y. Li et al., 2019c)

^bCoordination Number

Potassium is a soluble element. It is highly mobile during chemical weathering, metasomatism, and hydrothermal alteration. Potassium is a major cation in seawater with a modern seawater K content of ~399 ppm (Culkin and Cox, 1966). Potassium has a prolonged residence time (τ_K) in seawater of up to 11 million years (Goldberg, 1961), meaning K in seawater is conservative. Figure 3 illustrates the major sources and sinks of K into and out of the ocean (Bloch and Bischoff, 1979; Elderfield and Schultz, 1996; Jarrard, 2003; Holland, 2005; Miller et al., 2011; Berner and Berner, 2012; Staudigel, 2014). The two major sources of K into the ocean are river water and hydrothermal fluids, while the major two sinks are authigenic clay formation and low-temperature alterations of basalts. The amount of K in river water varies substantially depending on the lithology of the basin and the intensity of weathering, ranging from 0.1 to 238 ppm (Cocco et al., 1978). Estimates of the average global river water K content vary between 2.3 ppm (Livingstone, 1973) and 1.4 ppm (Berner and Berner, 2012). Most studies indicate that the majority (60–87%) of dissolved K in the river can be attributed to silicate weathering while a smaller fraction (~17%) comes from evaporite weathering (Meybeck, 1987; Gaillardet et al., 1999; Berner and Berner, 2012). Yet, in contrast, one study has argued that a major proportion (26–61%) of the dissolved K in rivers is instead from plant sourced K through the leaching of dead vegetation on the surface and in soils (Chaudhuri et al., 2007). Other minor (<1%) contributions to the dissolved K in rivers include atmospheric inputs (sea-salt spray or soil dust) and anthropogenic inputs (fertilizer) (Stallard and Edmond, 1981; Gaillardet et al., 2014).

Figure 3.

The major sources and sinks of the K in the ocean and their K isotope compositions. All K flux estimates are from the previous studies (Bloch and Bischoff, 1979; Elderfield and Schultz, 1996; Jarrard, 2003; Holland, 2005; Miller et al., 2011; Berner and Berner, 2012; Staudigel, 2014). The K isotope data sources: seawater (Hille et al., 2019; Wang et al., 2020), river water (S. Li et al., 2019a; Wang et al., 2021), hydrothermal fluids (Zheng et al., 2019), marine sediments and altered oceanic crusts (Santiago Ramos et al., 2018, 2020; Hu et al., 2020, 2021a).

Adsorption and ion exchange are the two major mechanisms responsible for removing K from the ocean (e.g., Berner and Berner, 2012), as K ions (K⁺) in aqueous solutions are both readily adsorbed by hydroxides such as Al(OH)₃ and Fe(OH)₃, and susceptible to cation exchange with clay minerals. In addition to adsorption and ion exchange, the K ions dissolved in seawater may react with detrital aluminosilicate weathering products to form authigenic aluminosilicates in situ in the ocean:

$${\text{K}}^{+}{\text{ + HCO}}_{3^{-}}+{\text{H}}_4{\text{SiO}}_4+\text{Al-silicate}\to \text{KAl-silicate}+{\text{CO}}_2+{\text{H}}_2\text{O}$$

This authigenic clay formation process is often referred to as reverse weathering (Mackenzie and Garrels, 1966), and is another important mechanism in removing K from seawater. Marine sediments are highly enriched in K (Plank and Langmuir, 1998) and their subsequent subduction is an important source of K for arc volcanos as well a contributor to mantle heterogeneity.

Potassium is also an essential element for life. Living organisms are generally enriched in K, especially in plant cells, where K is the most abundant metal (Sardans and Peñuelas, 2015). Because of the high K content in plant cells, the name of potassium, in fact, derives from “potash” meaning ashes in a pot after combustion of plant matter (Arevalo, 2016). Potassium is involved in a number of cellular biochemical processes, including photosynthesis and the regulation of osmotic pressure balance (Szczerba et al., 2009). Crucial among all processes is the “sodium-potassium pump” mechanism of the cell membrane. Organisms have developed specific enzymes that can drive the active exchange of Na and K inside and outside the cell membrane by consuming ATP (Adenosine triphosphate), achieving complex biological functions such as neural signaling (Voet et al., 2003). In normal animal cells, 30% of ATP is used to maintain the “sodium-potassium pump”, whereas in nerve cells the amount of ATP consumed in such process can be up to 70% (Voet et al., 2003). The 1997 Nobel Prize in Chemistry was awarded to Jens Christian Skou for the discovery of the “sodium-potassium pump” mechanism.

4. History of potassium stable isotope studies and early measurements

The K stable isotopes ³⁹K and ⁴¹K were first discovered in 1921 by F. W. Aston at the Cavendish Laboratory in Cambridge using the first velocity-(single)-focusing mass spectrograph (Aston, 1921). The radioisotope ⁴⁰K only accounts for 0.0117 % of all K, thus was not discovered until 1935 by Alfred Nier using an improved double-focusing mass spectrometer at the University of Minnesota (Nier, 1935). In the 1930s, Brewer and Kueck first measured ³⁹K/⁴¹K ratios in natural samples including rocks, minerals, plants, and animals (Brewer and Kueck, 1934). They determined a natural ³⁹K/⁴¹K ratio of 13.9 ± 0.4. Within typical analytical precision of a few percent in early studies, no resolvable differences in K isotope ratios were detected among geological samples, or in products produced during K metal evaporation experiments (Brewer, 1936a; 1936b). However, a series of research reported resolvable K isotope variations among certain biological and biomedical samples; for example, cancer tissues were reported to be enriched in light K isotopes compared to normal tissues (Lasnitzki and Brewer, 1938, 1940, 1941a, 1941b, 1942). In addition, a laboratory study showed that large K isotope fractionation could occur during ion exchange process where K solution was passed through a zeolite column (Taylor and Urey, 1938). Early attempts at applying K stable isotopes to natural samples waned after a study showed that the reproducibility of ³⁹K/⁴¹K ratio measurements on an early variant of a Thermal Ionization Mass Spectrometer (TIMS) were limited by variable isotope fractionation during thermal ionization of K on filaments, and the best achievable precision at the time, 1–2%, was insufficient to resolve K isotope variations in most natural samples (Cook, 1943).

In the 1970s, with continuous improvements in ion optics and the electronic systems equipped on TIMS, the analyses of ³⁹K/⁴¹K ratios could reach a precision of down to ~1‰. The absolute ³⁹K/⁴¹K ratio of the reference material NIST SRM 985 was measured as 13.8566 ± 0.0063 (Garner et al., 1975b), a value still being used today by the International Union of Pure and Applied Chemistry (IUPAC) (Meija et al., 2016). During this time, a wide variety of geological samples were surveyed, including basalts, granites, mantle peridotites, chondritic meteorites, and Apollo lunar samples (Burnett et al., 1966; Schreiner and Welke, 1971; Barnes et al., 1973; Begemann and Stegmann, 1976; Church et al., 1976; Bhattacharjee and Venkatasubramanian, 1977; Stegmann and Begemann, 1979). A major discovery from these studies was that ⁴¹K/³⁹K ratios in lunar soils could be enriched in ⁴¹K up to 12.58‰ compared to terrestrial and lunar igneous rock samples, likely owing to the volatilization of K caused by solar wind bombardments and micrometeorite impacts on the lunar surface (Barnes et al., 1973; Garner et al., 1975a; Church et al., 1976).

A research group at the University of Chicago first used Secondary Ion Mass Spectrometry (SIMS) to measure K isotopes (Humayun and Clayton, 1995b, 1995a). To avoid matrix effects, samples were first purified through column chemistry and then fused into barium borate glass that was mounted and polished prior to the SIMS analysis. This SIMS method avoided variable fractionation associated with the thermal ionization of samples on filaments during TIMS measurements, and a significantly improved precision of 0.5‰ therefore could be achieved (Humayun and Clayton, 1995b, 1995a). Using this improved capability for K isotope analysis, the Chicago group focused on investigating the cause(s) related to varying degrees of volatile element depletions in planetary bodies in the Solar System (e.g., Earth, Moon, Vesta) relative to CI carbonaceous chondrites (Figure 1). One possible cause is partial volatilization, and this hypothesis, if true, would predict varying degrees of K isotope variations in material from these planetary bodies due to the Rayleigh fractionation effect. However, with the exception of lunar soils, K isotope compositions of all geological and extraterrestrial samples analyzed at the time did not differ substantially (Humayun and Clayton, 1995b, 1995a), arguing against the partial volatilization hypothesis. Instead, Humayun and Clayton proposed volatile inventories in parent bodies were a consequence of incomplete condensation of the solar nebula. Since no K isotope variation could be resolved within analytical precision of 0.5‰ in this comprehensive survey of terrestrial and extraterrestrial samples by Humayun and Clayton, interest in K isotope research largely diminished. In the following 20 years after 1995, only a few K isotope studies were undertaken, including studies on tektites (Humayun and Koeberl, 2004; Herzog et al., 2008), chondrules in ordinary chondrites (Alexander et al., 2000; Alexander and Grossman, 2005), and on K isotope fractionation during laboratory volatilization and diffusion experiments (Yu et al., 2003; Richter et al., 2011, 2014). In addition to SIMS analysis, high-precision analyses of K isotope ratios on new generation (Triton) TIMS instruments have been demonstrated using internal normalization for instrumental mass bias correction (Wielandt and Bizzarro, 2011; Naumenko et al., 2013). These methods, however, are intended for the study of K isotope anomalies, so they erase natural mass-dependent K isotope variations in samples during mass bias correction.

The field of K isotope geochemistry was revived after recent analytical advances that have made it possible to measure natural mass-dependent K isotope variations at high precision (~0.1‰ or better) using multi-collector ICP-MS (MC-ICP-MS). Richter et al. (2011, 2014) achieved a precision of ~0.3‰ on ⁴¹K/³⁹K measurements using a single focusing MC-ICP-MS (Isoprobe). Morgan et al. (2012) first reported a double focusing MC-ICP-MS (Neptune) method with improved precision. Since 2016, several research groups have successively developed various MC-ICP-MS high-precision K isotope analysis methods to achieve precision of as small as 0.05‰ (Li et al., 2016; Wang and Jacobsen, 2016a; Hu et al., 2018a; Morgan et al., 2018), which is an order of magnitude better than the best precision obtained by the previous SIMS method. This new level of analytical precision has brought into view significant but previously unresolvable differences in K isotopic compositions among terrestrial and extraterrestrial samples for the first time, opening a new chapter in applying K stable isotopes as tracers for various geological processes.

5. High precision analysis method for K stable isotopes

5.1. Chemical separation

Potassium stable isotope analysis of bulk samples by MC-ICP-MS typically requires sample dissolution and purification, followed by sample-standard bracketing measurements. In general, silicate rocks can be dissolved in hydrofluoric acid (HF) and other strong acids (nitric and hydrochloric acids) at high temperatures. The presence of matrix elements can lead to erroneous K isotope measurements on MC-ICP-MS (i.e., matrix effects), so separation of K from sample matrices is necessary to ensure a high precision and high accuracy analysis. The majority of recent studies adopted an ion-exchange chromatographic protocol based on Strelow et al. (1970), and a summary is provided in Table 4 (Strelow et al., 1970; Humayun and Clayton, 1995a; Wang and Jacobsen, 2016a; Li et al., 2016, 2020; Hu et al., 2018a; Morgan et al., 2018; Chen et al., 2019b, 2021; Ku and Jacobsen, 2020; Moynier et al., 2021). Briefly, Bio-Rad AG50W cation exchange resin and a weak (0.5–1.5 N) nitric acid (HNO₃) solution are commonly used to elute K from matrix elements. Unpublished results from the University of Minnesota have demonstrated that weak hydrochloric acid (HCl) works equally well as an eluent. Using HCl may also minimize potential degradation of resin, as HNO₃ is an oxidant, thereby using HCl instead may prolong the lifespan of resin for repeated use. Because heavy K isotopes are first eluted from the ion exchange column relative to the light ones (Morgan et al., 2018; Chen et al., 2019b), 100% recovery of K is required to eliminate potential K isotopic fractionation during column separation. It is important to note that the existing methods based on AG50W resin typically cannot fully separate K from several matrix elements such as Cr, Rb, and V. In most cases, this incomplete separation does not produce perceptible matrix effects that would affect data quality (Chen et al., 2019b), because K is often a major element in the sample, and the proportion of matrix elements after columns are still comparably low (less than 5% of the K content). However, in low-K rocks (e.g., mantle peridotites and heavily chemically-weathered rocks), the proportion of matrix elements remaining in purified solutions could be much higher than 5%, resulting in severe matrix effects (>0.1‰). Therefore, repeated separation using the same ion-exchange chromatography column is required, which, however, may increase blank contamination and reduce the column recovery. Monitoring the total recovery rates and final matrix element abundances is recommended to ensure negligible column fractionation and matrix effects.

Table 4.

Potassium ion-exchange separation procedure comparison

	Strelow et al. (1970)	Humayun and Clayton (1995b)	Wang and Jacobsen (2016b)	Li et al. (2016)	Morgan et al. (2018)	Hu et al. (2018a)	Chen et al. (2019b)	University of Minnesota (unpublished)	Li et al. (2020)	Ku and Jacobsen (2020)	Moynier et al. (2021)	Chen et al. (2021)
Resin	Bio-Rad AG50W-X8 (200–400 mesh)	Bio-Rad AG50W-X8 (100–200 mesh)	Bio-Rad AG50W-X8 (100–200 mesh)	Bio-Rad AG50W-X12 (100–200 mesh)	Dionex CS-16 cation exchange column	Bio-Rad AG50W-X8 (200–400 mesh)	Bio-Rad AG50W-X8 (100–200 mesh)	Bio-Rad AG50W-X8 (200–400 mesh)	Bio-Rad AG50W-X12 (200–400 mesh)	Bio-Rad AG50W-X8 (100–200 mesh)	Bio-Rad AG50W-X8 (200–400 mesh)	Bio-Rad AG50W-X8 (200–400 mesh)
Column	Borosilicate glass	Pyrex or quartz	Quartz	Quartz		Disposable Bio-Rad Poly-Prep polyethylene column	Disposable Poly-Prep Econo-Pac polyethylene column	Disposable Bio-Rad Poly-Prep polyethylene column	Savillex Standard micro-column	Quartz	Disposable Bio-Rad Poly-Prep polyethylene column	Disposable Bio-Rad Poly-Prep polyethylene column
I.D.	2.5 cm	2.2 cm	1 cm	0.4 cm		0.8 cm	1.5 cm	0.8 cm	0.4 cm	1 cm	1.5 cm
Resin volume	90 mL	90 mL	13 mL	1 mL		2 mL	17 mL	2 mL	15 mL	13 mL	2 mL	2 mL
Cleaning	3 N HNO3	4 N HNO3 or 6 N HCl	4 N HCl	4.5 N HNO3		6 N HCl	6 N HCl	8 N HCl	6 N HCl	4 N HCl	6 N HCl	6 N HCl
Eluting	0.5 N HNO3	0.5 N HNO3	0.5 N HNO3	1.5 N HNO3	0.2% HNO3	0.5 N HNO3	0.7 N HNO3	0.4 N HCl	2 N HCl + 0.1 N HF	0.5 N HNO3	0.5 N HNO3	0.45 N HCl
Elution rate	3±0.5 mL/min	3–5 mL mL/min	0.4–0.6 mL/min
Potassium fraction	600–850 mL	700–1000 mL	180–340 mL	5–17 mL		14–36 mL	88–195 mL	23–51 mL	11–21 mL		13–35 mL	20–45 mL
Recovery		≥99.8%	>99%	99.4±2.1%	>99%		>99%	>99%	99.5±0.6%	>99%		>99%
Blank	0.5–1 μg	0.2%	0.82 μg	3–8 ng	0.06–1.06%		0.26±0.15 μg	~10 ng	<10 ng			0.13 μg

5.2. Mass spectrometry

The major challenge of high-precision K isotope measurements on MC-ICP-MS results from severe interferences from Ar-related species on K isotopes (e.g., ⁴⁰ArH⁺ on ⁴¹K⁺), because Ar is the primary working gas (carrier, auxiliary and cooling gas) by conventional ICP-MS. Two methods have now been demonstrated to be effective in removing Ar-related interferences during K isotope analysis (see Figure 4 and Table 5): the collision cell method (Feldmann et al., 1999a, 1999b; Bourg et al., 2010; Richter et al., 2011, 2014; Wang and Jacobsen, 2016b, 2016a; Li et al., 2016, 2017; Li, 2017; Parendo et al., 2017; S. Li et al., 2019a; Ku and Jacobsen, 2020; Chen et al., 2021; Moynier et al., 2021) and the “cold plasma” method (Jiang et al., 1988, 2019; Morgan et al., 2012, 2013a, 2013b, 2014, 2018; Hu et al., 2018a; Chen et al., 2019b; Sun et al., 2019; Li et al., 2020). The collision cell method uses collision/reaction gases (e.g., H₂ or He) to dissociate molecular ions (⁴⁰ArH⁺ and ³⁸ArH⁺) into monoatomic ions (⁴⁰Ar⁺, ³⁸Ar⁺ and H⁺) or neutral atoms/molecules through a series of physical and chemical reactions in the cell (e.g., Li et al., 2016), so that Ar and Ar-related species can be reduced to a negligible level. Deuterium gas (D₂) can be used in place of H₂ as the collision/reaction gas to further reduce ⁴⁰ArH⁺ formation (i.e., forming ⁴⁰ArD⁺ instead)(Li et al., 2016), but H₂ is generally sufficient (Wang and Jacobsen, 2016a). Earlier high-precision K isotope analyses using the collision cell method were exclusively achieved using GV Instruments Isoprobe, which was the only commercially available MC-ICP-MS equipped with a collision cell (Li et al., 2016; Wang and Jacobsen, 2016a). However, recently the Sapphire from Nu Plasma, which is a new generation collision cell equipped MC-ICP-MS, has become available. This new instrument has demonstrated much improved capabilities relative to earlier instruments; K isotope analysis with a precision of ~0.05‰ or better can be achieved with comparably less effort from operators and considerably reduced sample consumption (Ku and Jacobsen, 2020; Chen et al., 2021; Moynier et al., 2021). Undoubtedly, the study of K isotope geochemistry will gain further traction from this improved capability in instrumentation.

Figure 4.

Comparison of the peak scans using the two methods of K isotope measurements by MC-ICP-MS. Figure is reproduced from Chen et al. (2019b).

Table 5.

Potassium isotope analysis method comparison

	Wang and Jacobsen (2016b)	Li et al. (2016)	Morgan et al. (2018)	Hu et al. (2018a)	Chen et al. (2019b)	Li et al. (2020)	Ku and Jacobsen (2020)	Moynier et al. (2021)	Chen et al. (2021)
Vendor	GV Instruments IsoProbe P	GV Instruments IsoProbe P	Thermo Scientific Neptune Plus	Nu Instruments Nu Plasma II	Thermo Scientific Neptune Plus	Nu Instruments Nu Plasma III	Nu Sapphire	Nu Sapphire	Nu Sapphire
Single/ Double-focusing	Single	Single	Double	Double	Double	Double	Double	Double	Double
Ar-removing	collision cell	collision cell	cold plasma	cold plasma	cold plasma	cold plasma	collision cell	collision cell	collision cell
RF power	1350 W	1350 W	500–600 W	700–1125 W	600 W	700 W	1300 W	1300 W	1300 W
Mass resolution	Low ~400	Low ~400	High ~10,000	High ~10,000	High ~10,000	High >10,000	Low ~400	Low ~300	Low ~400
Introduction system	Dry plasma APEX IR+ ACM	Wet plasma spray chamber	Wet plasma spray chamber	Dry plasma DSN-100	Dry plasma APEX Ω	Dry plasma Aridus III	Dry plasma APEX Ω	Dry plasma APEX Ω	Dry plasma APEX Ω
Bracketing standard	Merck KGaA Suprapur 99.995% potassium nitrate (KNO3)	NIST SRM 3141a K standard solution	NIST SRM 999b 99.977 % potassium chloride (KCl)	NIST SRM 3141a K standard solution	NIST SRM 3141a K standard solution	NIST SRM 3141a K standard solution	Merck KGaA Suprapur 99.995% potassium nitrate (KNO3)	NIST SRM 3141a K standard solution	NIST SRM 3141a K standard solution
Other standards	Bulk Silicate Earth (BSE)	in-house standard UW-K	NIST SRM 70b NIST SRM 918 NIST SRM 985	NIST SRM 999c SRM 918b SRM 193	Merck KGaA Suprapur	GBW(E) 081590 K solution	NIST SRM 3141a K standard solution
Typical precision (2SD)	0.20	0.15	0.15	0.05	0.10	0.06	0.03	0.04	0.07

The “cold plasma” method uses the pseudo high mass resolution capability of MC-ICP-MS (m/Δm > 10,000) to separate the ⁴¹K peak from ⁴⁰ArH⁺ peak, with measurements being made on the interference-free left shoulder of the ⁴¹K peak (see Figure 3). The ³⁸ArH⁺ interference on ³⁹K is typically negligible by MC-ICP-MS because of the low natural abundance of ³⁸Ar and high abundance of ³⁹K. “Cold” plasma refers to operation of ICP-MS at a reduced RF power (~600–800 W) relative to the typical RF power at 1350 W during analysis. Under the normal operation conditions with an RF power of 1350 W (i.e., “hot plasma”), Ar⁺ and ArH⁺ are the main background ions. In contrast, the use of lower RF power (“cold plasma”) inhibits the production of these Ar-based ion species that require high ionization energy, and shifts the main background ions to NO⁺ (Jiang et al., 1988). As a result, ⁴⁰ArH⁺ intensity is greatly reduced, and an interference-free plateau on the left side of the ⁴¹K peak can be obtained and used for accurate K isotope analysis. Compared to the collision cell method, cold plasma analysis generally requires higher concentrations of K as this method sacrifices ion transmission in exchange for sufficient mass resolving power to resolve the ⁴⁰ArH⁺ interference on ⁴¹K⁺.

Table 5 summarizes high-precision K isotope analytical methods published in recent years. Of these instruments, the now obsolete GV Instruments IsoProbe, the latest Nu Instruments Sapphire and the Thermo Scientific Proteus are models equipped with a collision cell. The Thermo Scientific Neptune Plus and Nu Instruments Nu Plasma II/III/1700 do not have collision cell technology but have high mass resolution capability sufficient for the “cold plasma” method. Among these instruments, the Nu Instruments Sapphire has a unique design that includes both a low-energy ion path equipped with a collision cell and a traditional high-energy ion path bypassing the collision cell, so both the collision cell and “cold plasma” methods can be realized on this machine. A preliminary head-to-head comparison of the two methods using the same Sapphire MC-ICP-MS at the University of Minnesota has shown that a long-term reproducibility of ~0.07‰ (2SD) can be achieved using the “cold plasma” method with ~5 ppm K solutions, whereas an improved precision of <0.05‰ (2SD) is achievable using the low-energy collision cell path with only ~150 ppb K solutions. Overall, the reported precisions acquired from different laboratories are similar regardless the method and instrument used (see Figure 4). Importantly, no systematic difference in K isotope results is observed from different laboratories so far.

5.3. Potassium isotope standards

The isotopic composition of K is generally expressed as δ⁴¹K in parts per thousand (‰), with δ⁴¹K defined as δ⁴¹K = [(⁴¹K/³⁹K)_sample/(⁴¹K/³⁹K)_standard – 1]×1000. Several different reference standards have been used to define δ⁴¹K by different research groups. Currently, the only international standard with certified K isotope abundance is NIST SRM 985, which is representative of the natural K isotope abundance by IUPAC (Garner et al., 1975b). This standard is a high purity (99.9 wt.%) KCl (J. T. Baker Chemical Company, Phillipsburg, New Jersey). However, the production of NIST SRM 985 has long ceased making it increasingly difficult to obtain. Instead, Humayun and Clayton (1995a, 1995b) chose a Suprapur® 99.995% high purity KNO₃ solution from Merck KGaA as the 0‰ reference (δ⁴¹K_Suprapur). Later, the Harvard group (Wang and Jacobsen, 2016a, 2016b; Parendo et al., 2017) used the same Suprapur® solution as a working standard and then converted the measured δ⁴¹K_Suprapur values to a Bulk Silicate Earth (BSE) scale (δ⁴¹K_BSE) where 0‰ was defined by the average K isotope compositions of three basaltic rocks analyzed at the time (Wang et al., 2016a). This conversion can be done through a simple relation: δ⁴¹K_BSE = δ⁴¹K_Suprapur + 0.479. Recently, several groups have begun referencing their K isotope data to a high-purity KNO₃ solution–NIST SRM 3141a (δ⁴¹K_NIST3141a) (Li et al., 2016, 2017; Hu et al., 2018a; Chen et al., 2019b, 2019a; Tuller-Ross et al., 2019b, 2019a; Xu et al., 2019; Jiang et al., 2019; S. Li et al., 2019a; Sun et al., 2019; Tian et al., 2019; Zhao et al., 2020; Huang et al., 2020). The Princeton group references their data to the NIST SRM 999b standard (δ⁴¹K_NIST999b) (Morgan et al., 2018; Santiago Ramos et al., 2018), whereas the Berkeley group uses their own internal K standard (ULTRA Scientific) (δ⁴¹K_ULTRA) (Christensen et al., 2018). It is worth noting that, despite different reference frames, the reported δ⁴¹K values of basalts and other igneous rocks are similar: −0.5‰ (δ⁴¹K_ULTRA) (Christensen et al., 2018), −0.48±0.03‰ (δ⁴¹K_Suprapur) (Wang and Jacobsen, 2016a), −0.52‰ (δ⁴¹K_NIST3141a) (Li et al., 2017) and −0.54±0.06‰ (δ⁴¹K_NIST999b) (Morgan et al., 2018), indicative of similar K isotope compositions in the four reference K standards. Two groups have directly cross-calibrated NIST SRM 3141a, 999c, and Suprapur standards, and they confirmed that there is no difference in ⁴¹K/³⁹K between these three standards within analytical uncertainties (~0.05‰) (Hu et al., 2018a; Chen et al., 2019b). However, with a high analytical precision, Ku and Jacobsen (2020) reported that NIST SRM 3141a is +0.047±0.003‰ relative to the Suprapur standard. It is therefore likely that the other standards mentioned above also have small differences in their K isotope compositions. Nevertheless, an excellent agreement in the δ⁴¹K offset between seawater and basalts (i.e., ~0.6‰) has emerged from different laboratories capable of high-precision K isotope analysis to date, demonstrating an internal data consistency independent of the reference standard used. It should be noted that NIST SRM 985 is enriched in heavy K isotopes by +0.26‰ (±0.07‰) when compared to NIST SRM 3141a, 999b, and 999c (Morgan et al., 2018). Thus, Chen et al. (2019b) proposed to the IUPAC to refine the atomic weight and isotope abundance of K, both of which were based on previous analyses of NIST SRM 985 (Garner et al., 1975b).

As K isotope geochemistry remains in its infancy, it is imperative and valuable for the community to follow a consistent reference frame at this early stage so that δ⁴¹K data from different laboratories in future research can be compared without ambiguities. Considering the current K isotopic standard information together, we consider NIST SRM 3141a to be the best reference standard candidate for the following reasons (Hu et al., 2018a): first, NIST SRM 3141a is derived from NIST SRM 999a and therefore has the same K isotope composition as NIST SRM 999a, 999b and 999c. Second, NIST SRM 3141a is distributed as a solution form instead of powders, isotope homogeneity is therefore ensured. Third, because SRM 3141a is the most recent K standard produced by NIST, it is easy to obtain and stock is expected to last long into the foreseeable future. Although it was proposed to use seawater as the 0‰ reference for K isotopes (Morgan et al., 2018), we consider that it is more convenient to adopt the commercially available NIST SRM 3141a for the sake of general accessibility and quality control. Also, NIST SRM 3141a is of high purity, so its K isotope composition can be measured directly on MC-ICP-MS, whereas a seawater reference frame either requires purifying of seawater through column chemistry before analysis or a scale conversion through a secondary high-purity K standard that is calibrated against seawater. The NIST SRM 3141a reference frame is used in all the following discussions of this review.

There are however two potential shortcomings of using NIST SRM 3141a as a universal standard for K isotopes, which require continued evaluation by the community. Firstly, NIST SRM 3141a was developed as a K concentration standard, rather than a certified isotopic standard. It is not clear if each batch or aliquot has exactly the same K isotopic composition. Secondly, it is unknown if the borosilicate glass ampules used to pack the NIST SRM 3141a solution contains a significant amount of K. It is unlikely that the packing material for a K concentration standard would be K rich, but if it is, the weak nitric acid in this standard could potentially leach K from the glass. Leaching of K to varying extents may cause variations in the K isotopic composition of this standard solution. To avoid any of the two potential pitfalls mentioned here, it is therefore always necessary for each lab to measure not only NIST SRM 3141a but also seawater and common geological reference materials (see Section 5.4 and Table 6) for inter-laboratory comparisons. So far, based on published data from different laboratories, there is no evidence indicating any measurable K isotope heterogeneity in different aliquots of NIST SRM 3141a.

Table 6.

Average K isotope compositions of common geological reference materials in literatures (see Supplementary Data for the data compilation)

Name	Type		δ41KNIST3141a	2SD	n
	Mafic
BCR-1	basalt	USGS	−0.43 ± 0.06		70
BCR-2	basalt	USGS	−0.49 ± 0.18		86
BHVO-1	basalt	USGS	−0.43 ± 0.07		118
BHVO-2	basalt	USGS	−0.46 ± 0.09		108
W-1	diabase	USGS	−0.39 ± 0.01		22
W-2a	diabase	USGS	−0.48 ± 0.20		9
	Intermediate
AGV-1	andesite	USGS	−0.45 ± 0.02		66
AGV-2	andesite	USGS	−0.47 ± 0.11		29
	Felsic
G-2	granite	USGS	−0.45 ± 0.04		79
GS-N	granite	ANRT	−0.46 ± 0.04		27
GSP-1	granodiorite	USGS	−0.48 ± 0.07		65
GSP-2	granodiorite	USGS	−0.49 ± 0.05		18
QLO-1	quartz latite	USGS	−0.44 ± 0.18		26
RGM-1	rhyolite	USGS	−0.37 ± 0.03		38
RGM-2	rhyolite	USGS	−0.38 ± 0.06		9
	Sedimentary
MAG-1	marine mud	USGS	−0.45 ± 0.12		48
SCo-1	shale	USGS	−0.38 ± 0.07		51
SGR-1	oil shale	USGS	−0.27 ± 0.05		53

5.4. Potassium isotope compositions of geological reference materials

Since high-precision K isotope measurements first became available in 2016, multiple research groups have extensively surveyed the K isotope compositions of a series USGS reference materials. Although instruments and methods differ, δ⁴¹K values measured for the same reference material in different laboratories are, in general, consistent within analytical uncertainties (see Figure 5). Here, we compile the published δ⁴¹K results for some commonly analyzed USGS standards (Li et al., 2016; Wang and Jacobsen, 2016a; Hu et al., 2018a; Morgan et al., 2018; Santiago Ramos et al., 2018; Chen et al., 2019b; Xu et al., 2019; Huang et al., 2020) and calculate the averages to serve as a useful benchmark for quality control of any future K isotope measurements (see Table 6).

Figure 5.

Comparison of K isotope compositions of common geological reference materials analyzed in different laboratories. The dotted line represents the average value of global oceanic basalts (–0.43‰; Tuller-Ross et al., 2019a). All error bars shown are 2σ as reported in the literature.

Previously, different batches of USGS reference materials from the same location (e.g., BHVO-1 and BHVO-2) have been found to have slightly different isotopic compositions for several elements other than K (e.g., Weis et al., 2005). Here, we show in Table 6 that there are no measurable K isotopic differences between batches of the same USGS reference materials under current analytical uncertainties. For example, the average K isotopic composition of BHVO-1 is −0.43 ± 0.07‰, which agrees well with that of BHVO-2 (–0.46 ± 0.09‰). This conclusion also applies to BCR-1 and BCR-2, W-1 and W-2a, AGV-1 and AGV-2, GSP-1 and GSP-2, RGM-1 and RGM-2 (see Table 6).

6. Potassium stable isotope application in low-temperature geochemistry and biogeochemistry

Over the past few years significant progress has been made in low-temperature K isotope geochemistry that has shed new light on the global biogeochemical cycle of K. Although K is a major element in Earth’s surface environments, our current knowledge on the global K cycle lags behind the understanding of many other major elements such as Ca and Mg (Berner and Berner, 2012). One particular interest of the global K cycle is its close tie to the silicate rock cycle, because K primarily resides in silicate minerals rather than carbonates due to its large ionic radius making it largely incompatible to carbonate crystal lattices. As a result, it is conceivable that the K cycle is tightly related to silicate weathering and hence the global carbon cycle.

Available high-precision δ⁴¹K data have provided novel insights into some critical features of the global K cycle that were previously unknown from K concentration measurements. In 2016, two studies independently reported that seawater is enriched in heavy K isotopes by ~0.6‰ relative to igneous rocks (Li et al., 2016; Wang and Jacobsen, 2016a). Recent studies confirmed the homogeneity of δ⁴¹K in the ocean and have suggested seawater has an average δ⁴¹K value of +0.12 ± 0.07‰ (2SD)(Hille et al., 2019; Wang et al., 2020a). This first-order δ⁴¹K contrast between the ocean and BSE remains incompletely explained as of today, but it demonstrates significant K isotope fractionation during K cycling processes at the Earth surface and has stimulated continuous investigations to better characterize K isotope variations in different surface K reservoirs and during relevant surface processes.

One important process that regulates the K cycle in the ocean is hydrothermal alteration. Low-temperature (< ~70°C) hydrothermal alteration is a seawater K sink (e.g., Seyfried and Bischoff, 1979), whereas high-temperature hydrothermal alteration is a major K source in the ocean (e.g., Elderfield and Schultz, 1996). Parendo et al. (2017) first investigated the behavior of K isotopes during hydrothermal alteration as recorded in ophiolite samples of Ordovician ages, and they reported sample δ⁴¹K values varying between −0.01‰ and −0.67‰. This significant K isotopic variation was interpreted to largely reflect a simple mixing of K derived from seawater and unaltered basalts without significant K isotope fractionation (Parendo et al., 2017). Nevertheless, the underlying assumption of this interpretation was a similar δ⁴¹K value of Ordovician seawater relative to the modern ocean, and this assumption was contended by later studies that argued for secular changes in seawater K isotope compositions over geological history (S. Li et al., 2019a; W. Li et al., 2019a). More recently, Santiago Ramos et al. (2020) and Hu et al., (2021a) analyzed a large number of altered oceanic crust samples of various younger ages and argued that isotopically light K isotopes are preferentially removed from seawater during low-temperature alteration of oceanic crust (see Figure 6).

Figure 6.

Potassium isotope compositions of biological samples and sedimentary rocks. Data sources (Li et al., 2016; Li, 2017; Parendo et al., 2017; Christensen et al., 2018; Morgan et al., 2018; Santiago Ramos et al., 2018; S. Li et al., 2019a; W. Li et al., 2019a; Huang et al., 2020).

Rivers are another major K input to the ocean, so it is critical to characterize δ⁴¹K in river waters in order to assess the influence of this input on the seawater K isotope composition. Li et al. (2019a) reported δ⁴¹K values between −0.44‰ to +0.12‰ in water samples from several major Chinese rivers and their tributaries. In this study, sediments and river waters from two catchments of granitic bedrock were also studied, which provided the first direct evidence that light K isotopes are preferentially retained in secondary clays produced by silicate weathering with an estimated K isotope fractionation factor of ~−0.55‰ (S. Li et al., 2019a). They also observed a significant negative correlation between river water δ⁴¹K values and catchment weathering intensity, implying that K isotopes could be a promising tracer for silicate weathering. This study estimated that the average δ⁴¹K value of the global riverine input into the ocean is −0.22‰ (±0.04‰), indicating that the heavy δ⁴¹K value of seawater cannot be fully explained by the riverine input. Wang et al. (2021) recently reported the K isotopic composition of 24 major world rivers and found a range from −0.59 ± 0.04‰ to −0.08 ± 0.04‰, which overall agrees with the range (−0.44‰ to +0.12‰) reported by S. Li et al. (2019a). As such, this new study is consistent with previous work in suggesting that riverine input alone cannot explain the heavy δ⁴¹K value of seawater (see Figure 3). Wang et al. (2021) proposed a new estimate of global riverine input into the ocean of −0.38 ± 0.04‰ δ⁴¹K based on the flux-weighted and regionally-adjusted mean compositions of global rivers.

Variations in K isotope compositions have also been characterized for a range of other natural samples from surface environments, such as loess, shale, and marine sediments. Loess samples analyzed so far show relatively small variability in δ⁴¹K, ranging from −0.58 to −0.35 ‰ (W. Li et al., 2019a; Huang et al., 2020). Huang et al. (2020) found that δ⁴¹K values of their loess samples decrease with increasing weathering intensity, implying that light K isotopes are retained in residues from chemical weathering of silicates. This inference is consistent with the study of rivers introduced earlier (Li et al., 2019). The δ⁴¹K values of analyzed shale samples are more variable (ranging from −0.69‰ to −0.08 ‰) and show a negative correlation with both Fe₂O₃/Al₂O₃ and Fe₂O₃/K₂O ratios of the samples (Huang et al., 2020). Such correlations are interpreted to stem from the higher Fe₂O₃/Al₂O₃ and Fe₂O₃/K₂O ratios in clay minerals that have lower δ⁴¹K values (Huang et al., 2020). Furthermore, there are two different trends of δ⁴¹K versus K/Na for shales. The trend of decreasing δ⁴¹K values with increasing K/Na ratios is interpreted to reflect K-Na exchange with clay minerals, whereas the opposite trend is interpreted to be related to the neoformation of glauconite (W. Li et al., 2019a). Hu et al. (2021a) provided a set of K isotope data for <100 Ma marine sediments from drill cores and confirmed the occurrence of significant K isotope fractionation associated with the formation of these marine sediments.

Behavior of K isotope systematics during continental weathering has also been investigated through weathering profiles (see Figure 6). Chen et al. (2020) studied K isotope variability within two weathering profiles where basalts were intensely weathered to bauxites. Bauxites were found to have δ⁴¹K values as low as −0.94‰, which was interpreted to reflect preferential removal of heavy K isotopes into aqueous solution during chemical weathering. Teng et al. (2020) reported K isotope data from two weathering profiles, one mainly composed of diabase and the other one composed of granite. Distinct K isotope fractionation behavior was observed; the diabase weathering profile showed very limited K isotope variability (−0.48‰ to −0.41‰), whereas the granite weathering profile showed significant K isotope variability that ranges from −0.49‰ in the fresh rock to −0.63‰ in the most intensely weathered material (Teng et al., 2020). Similarly, the δ⁴¹K decrease in the granite weathering profile was attributed to the preferential release of isotopically heavy K into aqueous solution during weathering. Existing studies showed that weathering products (residue) tend to be enriched in light K isotopes (Figure 6), but controls on δ⁴¹K variability in weathering profiles are not always well identified and further studies are needed.

Interesting results have also emerged from biological systems, paving the way for new applications of K isotopes. Terrestrial plants have so far shown a large δ⁴¹K spread ranging from −1.7‰ to +0.1‰, with the majority of samples showing K isotopic values below that of igneous rocks (Li et al., 2016; Li, 2017; Christensen et al., 2018; Morgan et al., 2018). Potassium in plants is sourced from water in soils, which is enriched in heavy K isotopes due to the chemical weathering of silicates (S. Li et al., 2019a). Therefore, the enrichment of light K isotopes seen in plant tissue is best explained by the preferential uptake of light K isotopes through the root system (Li, 2017). This interpretation was confirmed by plant growth experiments that directly measured δ⁴¹K values in both plants and K fertilizer used (Christensen et al., 2018). In contrast, δ⁴¹K values of various types of marine algea vary from 0.57‰ to 1.11‰, which is considerably higher than the δ⁴¹K of modern seawater (Li, 2017). This has been hypothesized to reflect preferential excretion of isotopically light K by cells of these algea (Li, 2017). For comparison, marine animals have an even greater range in δ⁴¹K from −0.56‰ to 0.97‰ (Li, 2017; Morgan et al., 2018). Among the animals analyzed, sea clams have δ⁴¹K higher than seawater, whereas fish and jelly fish show δ⁴¹K values lower than seawater. These relationships are hypothesized to be caused by different K isotope fractionation associated with physiology (such as osmotic pressure regulation) of different species (Li, 2017), and/or with their trophic levels in a food web (Clementz et al., 2003; Li, 2017).

There are also several experimental and theoretical studies intended to quantify K isotope fractionation factors necessary to interpret data from various low temperature environments. Li et al. (2017) performed recrystallization experiments for eight K-bearing salts at room temperature and measured K isotope fractionation factors (⁴¹ΔK_min-aq) between the salts and the respective saturated K-salt solution. These experiments revealed a varying magnitude of K isotope fractionation between K-salts and K solutions, with Δ⁴¹K_min-aq ranging from −0.5 ‰ for KI to +0.5 ‰ for K₂CO₃*1.5H₂O. Notably, there is negligible K isotope fractionation between sylvite and KCl solution (0.02 ±0.12‰), indicating that sylvite could be a potential archive for K isotopes of brine (e.g., ancient seawater). The results of Li et al. (2017) showed that K isotope fractionation factors are controlled by the K bond length and the valence of anions surrounding K in the crystal structure. Equilibrium K isotope fractionation factors between aqueous solutions and a range of K-bearing minerals have also been predicted by theoretical calculations (Zeng et al., 2019). In addition to equilibrium K isotope fractionation, large kinetic K isotope fractionation by ion diffusion has also been demonstrated by laboratory experiments at low temperatures (Bourg et al., 2010; Christensen et al., 2019), and inferred from field observations on pore waters in marine sediments (Santiago Ramos et al., 2018).

Laboratory experiments have also been conducted to study the mechanisms of K isotopic fractionation during chemical weathering, including the dissolution of silicate rocks and adsorption on clay minerals. Li et al. (2021a) carried out a systematic investigation on proton-driven and ligand-driven dissolutions of basalt and granite samples at both room and hydrothermal temperatures. They showed that lighter K isotopes are enriched in the solutions at the beginning of the experiments (<30 mins), but as the experimental system approaches both chemical and isotopic steady state no significant K isotopic fractionation occurs. This study indicates that the large isotopic fractionation observed during continental weathering (e.g., Chen et al., 2020; Teng et al., 2020) is unlikely due to primary mineral dissolution, but instead due to the formation of secondary clay minerals through either adsorption or the incorporation of K onto/into clay minerals during soil formation. The batch experiments of K adsorption on kaolinite and smectite by Li et al. (2021b) showed heavy K isotopic enrichment on clay surfaces when the system reached chemical steady state, which is the opposite to what has been observed in field observations (e.g., weathering profiles and river waters; S. Li et al., 2019a; Chen et al., 2020; Teng et al., 2020; Wang et al., 2021). The authors suggested, based on K K-edge XANES data, that the incorporation of K into clay minerals plays a more important role than the adsorption of K onto clay minerals in regulating K isotope behavior during silicate weathering. However, currently no systematic clay synthesis experiments have been undertaken, which is highly needed to better understand the mechanisms of K isotopic fractionation during chemical weathering.

7. Potassium stable isotope application in high-temperature geochemistry

The principle of isotope fractionation predicts that the higher the temperature, the smaller the isotope fractionation under equilibrium conditions (Urey, 1947). Theoretical calculations show that equilibrium fractionation of K isotopes between minerals at high temperatures are small (Y. Li et al., 2019b, 2019c) and typically below current analytical uncertainties. The equilibrium isotopic fractionation between minerals is related to both their bonding strengths and coordination numbers (see Table 2; Schauble, 2004; Y. Li et al., 2019c). Table 3 shows a compilation of the force constants for K-containing bonds (e.g., K-O, K-S, and K-Cl bonds) for each mineral calculated by first principles (Y. Li et al., 2019c). The equilibrium fractionation factors of K isotopes between minerals at different temperatures can be calculated on basis of the following simplified formula.

$${\Delta }^{41}{\text{K}}_{\text{A-B}}\simeq 5718\times （F_A–F_B）/T^2$$

where F_A and F_B are the average force constants of the K-bearing bonds of minerals A and B (see Table 3) and T is the temperature (in Kelvin). For example, the mean force constant for K-bearing bonds in dolomite is 59.93 N/m, whereas the mean force constant for K in microcline is 48.92 N/m. At a temperature of 1000 K, the calculated equilibrium fractionation factor of K isotopes between dolomite and microcline is 0.06‰, which is difficult to detect at the current precision of K isotope measurements (~0.05‰). At higher temperatures, this fractionation factor will be even smaller. Thus, the isotopic fractionation of K produced during magmatic processes under high temperature conditions will be relatively small and difficult to measure.

The conclusion drawn from the theoretical calculation is substantiated with observations of igneous rock systems in nature. As shown in Figure 7 and Table 6, a large number of igneous rocks from different regions and tectonic settings on Earth (e.g., ultramafic, mafic, intermediate, and felsic) yield the same K isotopic composition of about −0.5 – −0.4‰ (Li et al., 2016; Wang and Jacobsen, 2016a; Hu et al., 2018a; Morgan et al., 2018; Santiago Ramos et al., 2018; Chen et al., 2019b; Xu et al., 2019; Huang et al., 2020). Crucial in this assessment is the recently measured the K isotope compositions of a series of igneous rocks (from basalt to rhyolite) from the Hekla volcano in Iceland, which are all derived from the same source but at different stages of magmatic evolution (Tuller-Ross et al., 2019b). They found no significant difference (–0.53 – −0.41‰) in the K isotope compositions of these igneous rocks, suggesting that no significant isotopic fractionation of K isotopes is generated in the high temperature magmatic environment. A similar observation has also been made among a series of samples produced by fractional crystallization from the cooling Kilauea Iki lava lake, Hawaii (Hu et al., 2021b). Both studies suggest that high-temperature igneous processes do not significantly fractionate K isotopes. This is very different from the sedimentary rocks discussed in the previous section, which show significant K isotope fractionation.

Figure 7.

Potassium isotope compositions of igneous rocks. Data sources: MORB (mid-ocean ridge basalt), OIB (ocean island basalt) and BABB (back-arc basin basalt) from Tuller-Ross et al. (2019a); AOC (altered oceanic crust) from Santiago Ramos et al. (2020); Ophiolite from Parendo et al. (2017) and Santiago Ramos et al. (2020); Eclogite from Liu et al. (2020); Continental basaltic lava from Sun et al. (2020); I-type, A-type, and S-type granites from Huang et al. (2020); Pegmatites from Morgan et al. (2018); Weathering profiles from Chen et al. (2020) and Teng et al., (2020). See Supplementary Data for the complete data compilation.

Among igneous rocks, the only samples that exhibit substantial K isotope fractionations are pegmatites and the minerals separated therein (e.g., Amazonite) (Morgan et al., 2018). Compared to the restricted δ⁴¹K range in other igneous rocks (−0.5 – −0.4‰), these pegmatites have very large K isotopic variations (–1.36 – −0.11‰). Pegmatites are the last product of magma evolution and are recrystallized under the influence of hydrothermal fluids, resulting in large K isotope fractionations (Morgan et al., 2018).

Potassium is a large ion lithophile and incompatible element. During partial melting of the mantle, K has a strong tendency to partition into the melt phase, thus generating no significant isotopic fractionation. Thus, mid-ocean ridge basalts (MORB) and ocean island basalts (OIB) theoretically inherit the K isotopic compositions of their mantle sources. Tuller-Ross et al. (2019a) measured MORB and OIB samples collected from the Atlantic Mid-ocean Ridge, East Pacific Rise, Galapagos Ridge, Gulf of Aden, Red Sea etc. They found that the MORB and OIB samples from different regions have indistinguishable K isotope compositions and suggested that the upper mantle of the Earth is a homogeneous reservoir in term of K isotopes (see Figure 7). The mean value calculated from globally distributed oceanic basalts is −0.43 ± 0.17‰ (2SD). This value can be used as the best available K isotope composition of the Bulk Silicate Earth (BSE) to date. This new value is consistent with the previously proposed BSE value of −0.48 ± 0.03‰ derived from a much smaller sample set (Wang and Jacobsen, 2016a), and with measurements of basaltic samples from other studies (Li et al., 2016; Hu et al., 2018a; Morgan et al., 2018; Xu et al., 2019).

Although fresh oceanic basalts on a global scale have similar K isotopic compositions and the mantle (at least the upper mantle) reservoir appears homogeneous in K isotopes, as discussed in Section 6, K isotopic fractionation occurs during ocean floor hydrothermal alteration (Parendo et al., 2017; S. Li et al., 2019a; W. Li et al., 2019a; Hu et al., 2020). With some exceptions, most sediments are enriched in lighter K isotopes and most AOC in heavier K isotopes. Therefore, the subduction of oceanic crust will bring both the lighter K isotopes in the sediment and the heavier K isotopes in the AOC into the mantle, generating heterogeneity of K isotopes in the mantle (Hu et al., 2018b; Sun et al., 2019). Given the limited K isotope data to date, K isotopic compositions for many lithologies (e.g., mantle peridotites) are still not available. Therefore, the heterogeneity of K isotopes in the mantle has not yet been observed. Nevertheless, preliminary results show that the K isotopic compositions of peridotite xenoliths vary by ~2‰ (Ionov and Wang, 2020), which is most likely due to metasomatic processes and melt refertilization in the mantle. K isotopes could be used to constrain the contribution from the subducted oceanic crust in the sources of OIB or other mantle-derived rocks (Sun et al., 2019). However, caution is urged since the K isotope compositions of AOC could be additionally fractionated by high-temperature high-pressure metamorphic dehydration processes in the subduction zone (Liu et al., 2020). Presently it is not entirely clear what the K isotopic signatures of subducted AOC are. In contrast to AOC, the major K host phase in sediments is phengite, which can resist breakdown to 300 km in depth at a cold subduction zone (Schmidt, 1996). Thus, the K isotopic signatures in sediments are most likely preserved during subduction metamorphism with no significant K isotopic fractionation found so far in metasediments.

Hydrothermal alteration processes associated with magmatic differentiation can also cause significant K isotope fractionation. A study on the Jiangxi Dexing porphyry copper mine showed that the K isotopic compositions of rocks underwent severe hydrothermal alteration, varying substantially from −1.02 to +0.38‰. Furthermore, most of the altered rocks have higher δ⁴¹K values than average igneous rocks (~ −0.5‰), indicating that the alteration products, especially phyllosilicate minerals, preferentially enrich heavy K isotopes during hydrothermal alteration (W. Li et al., 2019b). The study also found no correlation between the δ⁴¹K values of the altered rocks and their corresponding K contents (or mineralogical characteristics), suggesting that the δ⁴¹K values of the altered rocks are mainly controlled by the conditions and isotopic compositions of the hydrothermal fluids. Preliminary studies on the Dexing porphyry copper mine have shown that K isotopes can be used to trace different geological fluids in complex hydrothermal systems (W. Li et al., 2019b). Therefore, K isotopes may potentially probe the formation of hydrothermal mineral deposits.

8. Potassium stable isotope application in cosmochemistry

The K isotope composition of meteorites, Apollo lunar rocks and Apollo lunar soils begun being studied as early as the 1960s (Burnett et al., 1966; Barnes et al., 1973; Garner et al., 1975a; Church et al., 1976). In 1995, Humayun and Clayton (1995a, 1995b) extensively surveyed the K isotopic variations across the planetary materials in the Solar System. However, with the exception of lunar soils, the K isotope compositions of all extraterrestrial materials (meteorites and lunar rocks) were not resolvable from that of the Bulk Silicate Earth. Nevertheless, with the improved analytical precision available since 2016, it has been shown that the bulk silicate Moon (BSM) is enriched in heavy K isotopes (–0.07 ± 0.09‰; 2SD) compared to the BSE (–0.43 ± 0.17‰; 2SD; see Figure 8; Wang and Jacobsen, 2016b; Tian et al., 2020). The ~0.4‰ K isotopic difference between the Earth and Moon has been interpreted as a consequence of the Moon-forming Giant Impact (Wang and Jacobsen, 2016b), which is consistent with the Moon’s overall enrichment in heavy isotopes of other moderately volatile elements such as Cl, Rb, Zn and Cu (Herzog et al., 2009; Sharp et al., 2010; Paniello et al., 2012; Pringle and Moynier, 2017). Together, these isotopic signatures from moderately volatile elements provide new constraints for theories on the origin of the Moon. Furthermore, a systematic survey on a more geochemically/petrologically diverse set of lunar samples showed a heterogeneous redistribution of K isotopes in the Moon introduced during its magmatic evolution (Tian et al., 2020). In contrast to lunar mare basalts, which show a narrow range of K isotopic variation (–0.15 to –0.01‰) and are all significantly heavier than the BSE, lunar non-mare rocks span a substantial range, from −2.60‰ to +0.51‰. Lunar meteorites with KREEP signatures are substantially enriched in ⁴¹K relative to lunar mare basalts, whereas some non-mare rocks, especially those dominated by low-K feldspathic materials, show extremely light δ⁴¹K values (Tian et al., 2020). This recent observation of extremely light K isotope enrichment in lunar non-mare rocks agrees well with the lunar anorthosite (–3.9 ± 0.9‰) previously measured by Humayun and Clayton (1995b). Lunar anorthosites (or other lunar materials that compose a significant feldspathic component) are low in initial K abundances and thus more susceptible to the incorporation of ³⁹K-enriched vapor phases in the context of urKREEP degassing (Tian et al., 2020), which is consistent with the anorthosite light-isotope enrichments seen in other MVE isotopes (e.g., Cl, Cu, Zn, and Ga) (Moynier et al., 2006; Boyce et al., 2015; Kato et al., 2015; Kato and Moynier, 2017).

Figure 8.

Potassium isotope compositions of extraterrestrial samples. Data sources: (Wang and Jacobsen, 2016b; Tian et al., 2019; Bloom et al., 2020; Ku and Jacobsen, 2020; Tian et al., 2020a; Zhao et al., 2020). See Supplementary Data for the data compilation.

The second largest asteroid, Vesta, like the Moon, is also depleted in both the moderately volatile element K and other moderately volatile elements (see Figures 1 and 2). Recent measurements of the K isotope compositions of HED (howardite-eucrite-diogenite) meteorites (presumably originated from asteroid 4-Vesta [Binzel and Xu, 1993]) have shown that the enrichments of heavy K isotopes are more extreme than the BSE and BSM (see Figure 8) (Tian et al., 2019). Vesta's significant loss of moderately volatile elements and its associated heavy isotope enrichment are interpreted as the results of either 1) inherited signatures from its precursor materials; 2) vapor loss during the accretionary growth of planetesimals; 3) degassing during a global magma ocean; or 4) a combination of the above (Tian et al., 2019).

The K isotopes of various types of martian meteorites (e.g., shergottites, nakhlites, Chassigny, and NWA 7034) have also been analyzed (Tian et al., 2019, 2020b; Ku and Jacobsen, 2020). The martian meteorites define an average δ⁴¹K of −0.28 ± 0.18‰, n = 30, and can be directly used as a proxy of the K isotopic composition of bulk silicate Mars. Compared to the BSE (–0.43 ± 0.17‰), bulk silicate Mars is slightly enriched in heavy K isotopes. This difference between Earth and Mars is small but statistically resolvable (Tian et al., 2020b). Intriguingly, as seen in Figure 9, a strong correlation is found in four of the inner Solar Systems planetary bodies (Earth, Mars, Moon, and 4-Vesta) between average bulk silicate K isotopic composition and the planetary bodies surface gravity and other related parameters (body radius, mass, and escape velocity etc.). Such a correlation provides strong evidence for planetary K depletion originating from vapor loss during accretionary growth. Tian et al. (2020b) further interpreted this correlation as the ramification of a late-stage evaporation event, which occurred universally for terrestrial planets (Elkins-Tanton, 2012). They suggest that large bodies tend to retain volatiles once their size exceeds a certain threshold, thus becoming less isotopically fractionated; whereas small bodies have insufficient gravity to prevent vapor loss. They further correlated K isotopes with estimates of bulk water abundances (e.g., McCubbin and Barnes, 2019) and proposed K isotopes as a new conservative isotopic tracer for comparing highly and moderately volatile inventories of bulk planetary bodies.

Figure 9.

The average K isotopic compositions of four differentiated terrestrial planetary bodies (The Earth, Mars, Moon, and asteroid 4-Vesta) versus their corresponding radius, mass, surface gravity, and escape velocity in log scale. See Supplementary Data for the data compilation. Figure is reproduced from Tian et al. (2020b).

In contrast to differentiated planetary bodies (e.g., Earth, Moon, Mars, and Vesta), undifferentiated planetary bodies (e.g., carbonaceous, ordinary, and enstatite chondrites; CC, OC and EC) exhibit larger K isotope fractionations of up to 2‰ (Bloom et al., 2020; Ku and Jacobsen, 2020; Zhao et al., 2020) (see Figure 8). Such large K isotopic variations are proposed to have been formed by a combination of solar nebular processes and parent-body alterations (thermal metamorphism, aqueous alteration and impact volatilization) (Bloom et al., 2020; Zhao et al., 2020). Notably, although the K isotope range of carbonaceous chondrites overlap with those of ordinary chondrites (see Figure 8), a dichotomy of K isotopic compositions between the two groups is observed, with the vast majority of CCs showing heavier δ⁴¹K than most OCs. Carbonaceous and ordinary chondrites are thought to have formed in two regions of the Solar System, presumably separated by Jupiter (the inner Solar System vs. the outer Solar System). This K isotope dichotomy between OCs and CCs is consistent with the non-mass dependent isotope systems (e.g., Δ¹⁷O, ε⁵⁰Ti, ε⁵⁴Cr, and ε⁹²Mo) (Warren, 2011a, 2011b; Kruijer et al., 2017, 2020; Scott et al., 2018). Such K isotope dichotomy between OCs and CCs has been recently proposed as being inherited from the heterogenous distribution of ⁴¹Ca (decay to ⁴¹K) in the presolar molecular cloud (Ku and Jacobsen, 2020). Whether the three K isotopes (³⁹K, ⁴⁰K, and ⁴¹K) follow mass-dependent fractionation among chondrites or not is still unknown. Thus, this K isotopic anomaly hypothesis remains to be tested.

The K isotope systematics among chondrites can be compared to the isotopes of other moderately volatile elements such as Cu, Zn, and Rb in order to shed light on the mechanisms producing the progressive depletion pattern of moderately volatile elements among chondrites (see Figure 1 in Bloom et al., 2020). Numerous hypotheses have been proposed to explain such depletion patterns including: incomplete condensation from the solar nebula (Wasson and Chou, 1974; Wai and Wasson, 1977; Humayun and Clayton, 1995b; Cassen, 1996; Ciesla, 2008); partial evaporation of interstellar dusts prior to incorporation into the solar nebula (Yin, 2005) or in the solar nebula (Huss et al., 2003; Huss, 2004); and the mixing of chondrite components from two different reservoirs (e.g., Alexander et al. 2001; Anders 1964; Shu et al. 1997). Generally speaking, equilibrium condensation from the solar nebula would generate little/negligible isotopic fractionation, while partial evaporation would enrich heavier isotopes in the more volatile depleted samples (Humayun and Clayton, 1995b). Previously, the variations in Cu, Zn, and Rb isotopes among chondrites have shown positive trends between elemental abundances and their isotopic compositions, which is the opposite of what would be expected from partial evaporation (Luck et al., 2003, 2005; Moynier et al., 2007; Pringle and Moynier, 2017; Pringle et al., 2017). Therefore this observation has been interpreted to be the result of mixing between two solar nebular reservoirs (Luck et al., 2003, 2005; Moynier et al., 2007; Pringle and Moynier, 2017; Pringle et al., 2017). In contrast to Cu, Zn, and Rb isotopes, there is no clear trend between K abundances and K isotopes among chondrites (Wang and Jacobsen, 2016b; Bloom et al., 2020; Ku and Jacobsen, 2020; Zhao et al., 2020). Also there are no correlations between K isotopes and other moderately volatile element (Cu, Zn, and Rb) isotopes (see Figure 7 in Bloom et al., 2020). Several possibilities could contribute to this decoupling. First of all, K is fluid mobile and prone to thermal diffusion; both aqueous alteration and thermal metasomatism tend to migrate K (Grossman et al., 2000; Grossman and Brearley, 2005) and fractionate K isotopes (Jiang et al., 2020; Koefoed et al., 2020). Aqueous alteration and thermal metasomatism generate large isotopic heterogeneity even at the centimeter scale, which makes it challenging to estimate bulk K isotope compositions of chondrites based on small sample sizes (~0.1 g). Secondly, the volatilities of K, Cu, Zn, and Rb (1006 K, 1037 K, 726 K, and 800 K, respectively; Lodders, 2003) are different and span a range of ~300 K in term of 50% condensation temperature. In addition, experiments show temperature and oxygen fugacity also modify the relative volatilities of K, Cu, Zn, and Rb during evaporation (e.g., Sossi et al., 2019; Zhang et al., 2021). Conclusively determining if the isotopes of K and the other moderately volatile elements are decoupled among chondrites, and if so, what caused this decoupling, is another area that needs further investigation.

In addition to bulk chondrite analysis, high-precision K isotopic analysis has recently started being undertaken on individual chondrules within both OCs and CCs (Jiang et al., 2020; Koefoed et al., 2020). Nevertheless, due to the effects of thermal alteration in the LL4 chondrite Hamlet (Koefoed et al., 2020) and aqueous alteration the CV3 chondrite Allende (Jiang et al., 2020), any nebular effects on their K isotopic systems were unable to be conclusively understood. Both samples did show δ⁴¹K variations of ~1‰, with the CC chondrules showing heavier compositions than the OCs, in line with the dichotomy observed in the bulk chondrite samples. Additionally, Hamlets chondrules showed a striking correlation between δ⁴¹K and chondrule mass. However, earlier in situ SIMS K isotopic studies found variations of over 20‰ and no correlation between δ⁴¹K and any other parameter in chondrules from the unequilibrated OCs Semarkona and Bishupur (Alexander et al., 2000; Alexander and Grossman, 2005). Yet, as these SIMS analysis had significantly larger analytical uncertainties (1–10 ‰) compared to the recent high-precision analysis (~0.05 ‰) and possibly suffered from analytical artifacts, it is difficult to directly compare these chondrule data. More study is needed to understand the K isotopic systematics of chondrules.

The stable K isotope fractionation among iron meteorites has not yet been systematically studied and is largely unknown. In the metal phase, the dominating phase of iron meteorites, cosmogenic K overwhelms native (non-cosmogenic) K. Nearly all previous studies on K isotopes in iron meteorites focus on using cosmogenic K to determine their cosmic-ray exposure ages (Herzog and Caffee, 2014). The measured raw ⁴¹K/³⁹K ratios among iron meteorites vary from 0.100 to 0.717 (Voshage, 1978), which corresponds to a δ⁴¹K_NIST3141a of ~400 to 9,000‰. This large variation is beyond what has been observed, or is expected, for mass-dependent fractionation, and thus should be attributed to the cosmogenic spallation effect (see Section 2). Some iron meteorites do contain silicate inclusions; for example, Colomera IIE iron contains large K-feldspars (Wasserburg et al., 1968). Humayun and Clayton (1995a) analyzed the K isotope composition of one such K-feldspar inclusion in Colomera IIE iron, which yielded −2.2 ± 0.9‰. As shown above, cosmic-ray spallation produces significantly higher ⁴¹K/³⁹K ratios than normal (i.e., 0‰), leading to extremely heavy isotope compositions (e.g., ~400 to 9,000‰; Voshage, 1978). As such, the light K isotope composition of the K-feldspar inclusion in Colomera IIE iron cannot be explained by the cosmogenic effect. The IIE iron meteorites have been linked to a H chondrite parent-body based on the oxygen isotope analysis of silicate inclusions (e.g., McDermott et al., 2016). This K isotope composition of the K-feldspar inclusion in Colomera IIE appears slightly lighter than those of H chondrites (see Figure 8 and Supplementary Data; Bloom et al., 2020; Ku and Jacobsen, 2020), but considering analytical uncertainties, it is still reasonably close. It is interesting to note that the isotopes of Ge, another moderately volatile element, also show similarly low values for IIE irons and H chondrites (Luais, 2007; Florin et al., 2020), which are unique when compared to the other types of irons and chondrites. Nevertheless, overall, there is still too little K isotope data to make a strong connection between IIE and H chondrites. This would require a systematic study into the K isotope compositions of silicate inclusions in both IIE and other iron meteorites.

Tektites are natural glasses formed after meteorites impact the Earth’s surface. The major components of tektites are derived from the Earth's upper continental crust (UCC) with only trace amounts of extraterrestrial materials incorporated (Koeberl, 2007). Tektites have attracted the attention of planetary scientists as their impact formation process is somewhat comparable (albeit under much lower energy) to the Moon forming impact (Taylor, 2014). Compared to their source rocks (i.e., average UCC composition), tektites are significantly depleted in volatile elements, similar to the depleted volatile inventories of the Moon relative to the Earth (see Figures 1 and 2). Jiang et al. (2019) measured various types of tektites around the world and found no difference between their K isotopes and the average of the Earth's crust, being consistent with previous reports (Herzog et al., 2008; Li et al., 2016). This suggests that the expected K loss and K isotopic fractionation did not occur during the melting and volatilization of tektites, which is markedly different from that of lunar rocks (Wang and Jacobsen, 2016b; Tian et al., 2020a). Nevertheless, the temperature, pressure, and especially the oxygen fugacity during the formation of tektites are very different from the conditions experienced during the formation of the Moon. Thermodynamic calculations indicate that the volatilities of K and other volatile elements are influenced by these P-T-fO₂ conditions to varying degrees (Jiang et al., 2019). Trinitites, which are melt residuals formed during a transient heating event (i.e., a nuclear detonation), are natural analogs to tektites and volatile-depleted planetary materials such as lunar samples (Day et al., 2017). Similar to tektites, most trinitites analyzed show the same K isotopic composition as the BSE. Only one trinitite, the closest one to the center of the nuclear detonation, shows a depletion in K and an enrichment (~0.2‰) in heavy K isotopes (Chen et al., 2019a), which also correlates with a heavy Zn isotope enrichment. This trinitite is the first “experimental” product showing that the stable isotopes of K can be fractionated by evaporation from a melt under the extreme conditions of a nuclear detonation and that K and Zn isotopic fractionation are correlated during evaporation.

Potassium isotopes, together with isotopes of other volatile elements, such as Rb, Zn and Cu, can be used to constrain the temperature, pressure, and oxygen fugacity conditions during impact-melting and volatilization. Zhang et al. (2021) recently conducted laboratory evaporation experiments on basaltic material in a vacuum and analyzed the K and Rb isotopic compositions of the residues. From this, they calibrated the isotopic fractionation factors for K and Rb in a vacuum. As expected, the observed fractionation factors for both isotopic systems agreed well with their fractionation factors calculated using the Rayleigh distillation equation, which is the square root of the mass of the lighter isotope over the heavier isotope [e.g., (38.964/40.962)^0.5 for K]. Nevertheless, these experimentally-determined (also theoretically-predicted) isotopic fractionation factors under vacuum indicate that highly volatile depleted samples (e.g., HED, angrites) should display significant isotopic fractionation (~10s-100s ‰), which is inconsistent with observations in nature. As such, it appears that even within a solar nebular or planetary setting, volatile evaporation rarely occurs under vacuum. The relatively small (~1‰) K isotopic fractionation widely observed among planetary materials (see Figure 8) indicate a ubiquitous near-equilibrium vapor pressure during the volatile evaporation of all terrestrial planets.

9. Conclusions and outlooks

The K stable isotope system is one of the emerging non-traditional isotope systems (Teng et al., 2017). By reviewing the history and current states of K isotope research, we have the following conclusions and suggestions.

1. The current high-precision analytical method for K stable isotopes by MC-ICP-MS can routinely achieve precision better than 0.05‰. This opens up a wide range of possible applications of K stable isotopes in geochemistry and cosmochemistry research.

2. Measurements of K isotopes for the same geological reference materials are consistent within margin of error by different research groups despite using different methods (the collision cell method or the “cold” plasma method). In addition, despite different reference standards used, high-precision K isotopes data from different groups are in a good internal agreement.

3. Most igneous rocks on the Earth are indistinguishable in terms of their K isotopic compositions (−0.5 – −0.4‰). The best available K isotope composition of the Bulk Silicate Earth (BSE) derived from global oceanic basalts is −0.43 ± 0.17‰.

4. Fractionation of K isotopes is more significant during low-temperature geochemical and biological processes. Seawater and river waters have δ⁴¹K values higher than that of BSE, and one cause of these higher values is preferential enrichment of light K isotopes in secondary clays produced during silicate weathering and/or during ion exchange between clays and aqueous solutions. The first-order δ⁴¹K contrast between seawater and BSE, however, is still not fully explained, and requires further investigations on major K sources and sinks in the ocean.

5. Because of the large K isotope differences between the altered oceanic crust and overlying sediments relative to fresh mid-ocean ridge basalts, oceanic crust subduction will potentially introduce heterogeneity of K isotopes in the mantle.

6. The Moon and Vesta are significantly enriched in heavy K isotopes relative to the Earth, providing new constraints on the theories of their origins. Carbonaceous and ordinary chondrites are generally different in their K isotopes, further validating their formation in different regions of the solar nebula.

Potassium is a major element. It is a highly incompatible lithophile soluble element in terms of its geochemical behavior and is categorized as moderately volatile under cosmochemical classifications. Therefore, its stable isotope ratio has applications in a wide range of research fields. Significant breakthroughs in K stable isotope research have been made since 2016 with the advancement of high-precision analyses. With the arrival of the new generation collision-cell equipped multi-collector ICP mass spectrometers, the analytical precision of K isotopes will be further improved. Future directions in this field include:

1. further exploration and surveying of natural variations in K isotope compositions and identifying natural processes to which the K isotope tool is/is not sensitive to. For example, because of the different K isotopic signatures of subducting sediments and AOCs, K isotopes can be used to trace subducted oceanic crustal materials and mantle heterogeneity. Potassium isotopes could also be used to further explore the global K cycle and quantify the sources and sinks of oceanic K.

2. fundamental laboratory research to calibrate the fractionation factors between important minerals and fluids; For example, clay formation seems to be a major mechanism for fractionating K isotopes in low-temperature aqueous environments; however, there is still a lack of clay formation and cation-exchange experiments, limiting our understanding of the fundamental constraints on K isotope fractionation during these processes.

3. identifying key major natural science questions for which the K isotope tool can be applied, thus attracting more researchers from outside the field to consider K stable isotopes in their own research. For example, K is an essential nutrient and vital for plant growth and development. As K isotopes are fractionated during K transport within plant components, K isotope analyses could be a vital tool in plant biology and agriculture studies. In addition, K is also important for animal life as it regulates fluid balance, muscle contraction, and nerve transmission. As such, K isotopes could have significant potential in future medical studies.