The Rare Earth Elements: Fundamentals and Applications

Geology, Geochemistry, and Natural Abundances of the Rare Earth Elements

Scott M. McLennan

State University of New York at Stony Brook, Stony Brook, NY, USA

and

Stuart Ross Taylor

Australian National University, Canberra, Australia

1 SUMMARY

The rare earth elements (REE) are trace elements in most geological settings and are of great utility in understanding a wide variety of geological, geochemical, and cosmochemical processes that take place on the Earth, other planets, and other planetary bodies (e.g., Moon, asteroids). The properties that lead to this importance include the following: REE are an extremely coherent group of trace elements, by geochemical standards, in terms of ionic radius, charge, and mineral site coordination, which makes them especially valuable for monitoring magmatic processes; slight variations in their overall refractory nature provides insights into early solar system high-temperature processes; the distinctive redox chemistries of europium and cerium result in unique insights into magmatic and aqueous processes, respectively; their generally insoluble character in geological settings and resistance to remobilization beyond the mineralogical scale during weathering, diagenesis, and metamorphism makes them important tracers for characterizing various geochemical reservoirs (e.g., planetary crusts and mantles).

In addition to being of great value to general geochemistry investigations, the REE have proven of increasingly great commercial value. Modern applications involve many that are useful in high technology, including some of strategic/military use. Accordingly, understanding the geological conditions leading to REE concentrations that are sufficient for economically viable extraction is also seen as increasingly important.

This chapter addresses geological and geochemical factors that control REE distributions in rocks and minerals, both in the Earth and on other planetary bodies, and the processes that give rise to economic concentrations of REE in the Earth’s crust. We begin with a discussion of the fundamental geochemistry and cosmochemistry of REE. This is followed by describing processes that influence the distribution of REE in rocks and minerals and the geological conditions that give rise to ore-grade concentrations. Finally, we characterize abundances and distributions of REE in various reservoirs, such as bulk solar system, bulk Earth, crust, oceans, and so forth, that are relevant to understanding the origin and evolution of the Earth.

2 INTRODUCTION

Geochemists have long recognized the misnomer associated with the REE, aptly captured in the title of one early paper, Dispersed and not-so-rare earths.¹ Although REE occur as trace elements in the vast majority of geological environments, their natural abundances in crustal rocks, mostly ranging from hundreds of parts per billion (terbium, holmium, thulium, lutetium) to tens of parts per million (lanthanum, cerium, neodymium), are not exceptionally low compared to many other elements. Thus, depending on the estimate, the most common REE, cerium, is approximately the 27th most abundant element in the continental crust of the Earth. Regardless of absolute amounts, the REE arguably are the single most important coherent suite of elements in nature for the purposes of interpreting a wide variety of geological processes for reasons discussed below. Accordingly, the absolute concentrations and embedded radiogenic isotopic systems (e.g., ¹⁴⁷Sm–¹⁴³Nd, ¹⁴⁶Sm–¹⁴²Nd, ¹⁷⁶Lu–¹⁷⁶Hf, ¹³⁸La–¹³⁸Ce) have been studied in exhaustive detail in a wide variety of rocks, minerals, and aqueous fluids on the Earth and other available solar system bodies.

Industrial uses of REE metals and compounds have expanded greatly over the past century, from the early application of mixing small amounts of cerium oxide with thorium oxide to produce incandescent gas light mantles, developed in the late nineteenth century, to being crucial components in a wide variety of cutting-edge technology applications.² Modern uses of the REE in high-technology applications include many of considerable strategic value.³ Accordingly, geological processes giving rise to ore-grade concentrations of REE are also of increasing interest.

The history of meaningful geological and geochemical research using REE dates from the pioneering work of Victor Goldschmidt and Eiiti Minami in 1935, who used X-ray spectrography to first determine REE abundances in rock samples—European and Japanese shale composites.⁴ At that time, most workers were of the opinion that relative REE distributions were not fractionated by geological processes and early differences in REE distributions noted between shales and meteorites were dismissed as analytical error. Modern REE geochemical research dates from the early 1960s with the development of rapid and precise instrumental techniques, notably radiochemical and instrumental neutron activation analyses pioneered by Larry Haskin, Roman Schmitt, and their colleagues.⁵,⁶ (Various rapid high-precision mass spectrometry methods such as thermal ionization isotope dilution mass spectrometry and spark source mass spectrometry soon followed.) The seminal breakthrough of Haskin’s work was that REE distributions in shales were indeed significantly fractionated from meteorites, having higher abundances and relative enrichment of the light rare earth elements (LREE), lanthanum through samarium, thus opening the door to the modern phase of REE geochemical research, summarized here.

In this chapter, we are concerned with four major issues: (i) general geochemistry and cosmochemistry of REE; (ii) geological conditions giving rise to normal concentrations of REE in rocks, minerals, and natural waters; (iii) geological conditions giving rise to ore-grade concentrations of REE; and (iv) the abundances and distributions of REE in various geochemical and cosmochemical reservoirs that are relevant to understanding the origin and evolution of the Earth.

3 GENERAL GEOCHEMISTRY OF THE RARE EARTH ELEMENTS

3.1 Geochemical and Cosmochemical Classification

The REE consist of the Group 3 transition elements 21Sc, 39Y, and 57La and the inner transition (lanthanide) elements 58Ce through 71Lu. As described in greater detail below, the REE are trivalent in all known geochemical systems with the exception of europium (which can also be divalent) and cerium (which can also be tetravalent). In geochemistry nomenclature, the term rare earth elements almost universally refers only to lanthanum, yttrium, and the lanthanides (i.e., La–Lu, Y), which differs from formal chemistry nomenclature, resulting in some confusion. Geochemists also subdivide REE into the light rare earths (LREE (La–Sm) and heavy rare earth elements (HREE; Gd–Lu) due to their contrasting geochemical behavior, with a natural break at the commonly anomalous europium (see below). In some geochemical literature, a group of middle REE, Nd–Tb, is also recognized.

Yttrium behavior is very similar to the HREE in the vicinity of Dy–Ho, which is why it is typically included with the other REEs in geochemical discussions. On the other hand, the geochemical behavior of scandium, especially in magmatic systems, is much more similar to the first row (ferromagnesian) transition elements, iron, vanadium, chromium, cobalt, and nickel, due to its smaller ionic radius and different coordination in mineral lattices.⁷ Because of this, many geochemists do not consider scandium as a REE, but rather as a ferromagnesian trace element. However, McLennan⁸, among others, has pointed out that in aqueous systems, scandium indeed has much more affinity for the other REE and so there is some inconsistency with REE terminology even within the geochemical literature (Figure 1).

Figure 1 Plot of ionic radius versus atomic number for the trivalent lanthanide elements (La–Lu). Also shown are the ionic radii for trivalent Y and Sc, for the Eu²+ and Ce⁴+, and for other selected cations. The regular decrease in the ionic radii of the trivalent lanthanides is part of the lanthanide contraction. Sc³+ is much smaller than the other rare earth elements and more similar in size to Fe²+ and Mg²+

The development of the lanthanide series by filling of inner 4f orbitals, which poorly shield outer electrons from the increasing nuclear charge, is the underlying cause of the remarkably coherent geochemical behavior of the group (Table 1). Among other things, this results in the trivalent state being especially stable (with two notable exceptions) and the ionic radius decreasing in a remarkably systematic manner, part of the "lanthanide contraction" (Figure 1; Table 1). The dominant controls on the geochemical and cosmochemical behavior of the REE are their size (ionic radius), mineral site coordination (CN), redox behavior, volatility, and complexing behavior. An additional influence, termed the tetrad effect, is more controversial and discussed in greater detail below. Although the lanthanides bear some chemical properties that are similar to the Group 13 elements (boron, aluminum, gallium, etc.), their geochemical and cosmochemical distributions are not significantly influenced by this.

With rare exception, REE are lithophile (rock-loving) in igneous systems and with the exception of Sc, are incompatible (i.e., bulk solid-melt partition coefficients, D < 1) with the degree of incompatibility increasing with increasing size (and decreasing atomic number). Their ionic potentials (nominal charge/ionic radius) are ≥4.0 and so they are large-ion lithophile, rather than high field strength, incompatible elements. Accordingly, the lanthanides (and yttrium) tend to be concentrated in magmatic liquids and late crystallizing mineral phases. Only sodium and calcium come close in size to the REE (apart from scandium) among the major mineral-forming cations in the Earth’s crust and mantle; however, substitution for these elements (especially sodium) leads to significant charge imbalances, thus limiting any such substitutions. In aqueous alteration systems, REE have very low fluid/rock partition coefficients and thus their primary abundances are only disturbed at relatively high fluid/rock ratios under most weathering, diagenetic, hydrothermal, and metamorphic conditions. This resistance to disturbance and remobilization is another important reason why REE are considered such valuable trace elements in geochemistry and cosmochemistry.

From a cosmochemical perspective, all of the REE are refractory elements (Table 1), with relatively high 50% condensation temperatures (≥1356 K at 10−4 bar), although significant variations in volatility are present (e.g., cerium, europium, and ytterbium being less refractory than other REE). These modest differences in volatility are important for interpreting certain early formed meteorite components (minerals and inclusions) in terms of the early thermal history of the solar nebula (see below).

Under aqueous conditions, the REE exist mostly in very low concentrations as a variety of complexes, with metals (REE³+), carbonate species (REECO3+), and bicarbonate species c01ie_003_001 dominating in seawater.⁹ For a number of rare earth complexes that in nature can exist in magmatic, hydrothermal, and other fluid systems, such as fluorides, chlorides, sulfates, hydroxides, and carbonates, stability constants tend to increase with increasing temperature and decrease with increasing pressure. In general, there also tends to be an increase in stability of complexes with the heavier (smaller) REE.¹⁰,¹¹ In aqueous fluids, REE concentrations tend to increase with decreasing pH.

Table 1 provides some basic data for selected REE properties of geological interest, compiled from several sources, and a thorough compilation of the full range of REE chemical and physical properties is available in Emsley.¹²

3.2 Normalization of Lanthanide Abundances

The absolute concentrations of REE in geological materials follow the Oddo–Harkins, or odd–even, effect such that even atomic number elements are of higher concentrations than their adjacent odd atomic number counterparts. Since REE concentrations and relative distributions are also highly variable among rocks and minerals, it is difficult to compare absolute abundances graphically. Accordingly, it is customary to display REE data as a plot of normalized values (on a logarithmic scale) versus atomic number or reverse-order ionic radius (on a linear scale), termed Coryell–Masuda plots (Figure 2). On such diagrams, REE in rocks and minerals tend to follow smooth patterns (but with important exceptions discussed below). The two most commonly used data sets for normalization are average CI chondritic meteorites (usually on a volatile-free basis), reflecting solar and bulk Earth abundances, and average shale reflecting upper continental crust abundances (see below). Commonly used values for these two normalization sets are provided in the data tables discussed in more detail below. It should be noted that there are several different sets of chondrite values in use that differ by up to about 15% in absolute abundances (but with negligible differences in relative abundances) and so some care must be taken when comparing diagrams among different workers.

Table 1 Selected rare earth element properties

In addition to these normalization standards, it is not uncommon to normalize REE data to other compositions that are especially relevant to a specific problem. For example, in an igneous rock suite, it might be useful to normalize samples to the least petrologically evolved magmatic rock in the series. When studying weathering processes, it might be useful to normalize weathered samples to the unweathered parent rock. For authigenic minerals, insight might be gained by normalizing the mineral to the fluid from which they precipitated.

3.3 Europium and Cerium Redox Geochemistry

The existence of europium and cerium in other than trivalent states (Eu²+/³+, Ce³+/⁴+) is of considerable importance to geochemistry. Reduction of europium occurs under highly reducing conditions. Such conditions typically exist only within certain magmatic or hydrothermal environments but rarely if ever in environments found at the surface of the Earth. The ionic radius of EuII is about 17% larger than EuIII and essentially identical to SrII (Figure 1). Accordingly, its substitution behavior differs greatly from the trivalent REE, resulting in anomalous REE patterns (Eu anomalies; see below). The most important geological example is that europium becomes highly concentrated in plagioclase feldspar (substituting into the calcium site). Plagioclase is only stable to about 10 kbars pressure or 40 km depth on Earth and so anomalous europium behavior in magmatic rocks is a clear sign of relatively shallow igneous partial melting or fractional crystallization processes. No example of europium reduction at surficial conditions has been convincingly documented but could potentially be present during diagenesis under highly reducing, high temperature, and alkaline conditions.¹⁶

In contrast, cerium is readily converted from CeIII to CeIV under oxidizing surficial conditions. Especially notable examples are during formation of manganese oxide particles in the oceans and under certain surficial weathering conditions. CeIV is about 15% smaller than CeIII and tends to form highly insoluble hydroxide complexes. These processes commonly lead to separation of cerium from the other trivalent REE, again resulting in anomalous REE patterns (Ce anomalies). The anomalous depletion of cerium in seawater (see below) as a result of manganese oxide formation is a direct reflection of such redox controls. During most continental weathering, cerium is readily oxidized to CeIV and forms highly insoluble cerium hydroxide which may be locally precipitated while other trivalent REE are more readily mobilized. The effect of this is that Ce anomalies may also occur in weathering profiles.⁸

Figure 2 (a) Rare earth element abundance data for average shale and chondritic meteorites illustrating the Oddo–Harkins effect. (b) Masuda–Coryell diagram showing the average shale normalized to average chondrites. From such normalization, it is possible to readily observe the relative LREE enrichment and negative Eu anomaly in the shale. Eu* is the expected value for Eu on a smooth chondrite-normalized pattern and used to quantify the size of the Eu anomaly. A similar formulation can be used to characterize Ce anomalies

Anomalous europium and cerium behavior can be quantified using the parameter Eu/Eu* and Ce/Ce*. Eu* is the expected value for europium for a smooth chondritenormalized REE pattern (Figure 2), such that

(1)

where the subscript "N" refers to the chondrite-normalized value. In the geochemical literature, arithmetic means (i.e., (SmN + GdN)/2) are sometimes used to calculate Eu* but this is incorrect because REE diagrams are plotted on a logarithmic scale and can lead to serious error especially for steep chondrite-normalized REE patterns. Cerium anomalies can be similarly calculated with

(2)

In some analytical methods, adjacent elements are not determined (e.g., gadolinium and/or praseodymium)¹⁷ and in these instances, they are estimated assuming smooth chondrite-normalized REE patterns apart from europium and cerium.

3.4 The Tetrad Effect

As the lanthanide series develops, there appears to be increased stability (manifested by a variety of observations) associated with quarter (neodymium–praseodymium), half (gadolinium), thREE–quarter (holmium–erbium), and completely (lutetium) filled 4f shells. Resulting anomalous behavior in REE distribution patterns has been termed the tetrad or double–double effect.¹⁸ The four tetrads are La–Ce–Pr–Nd, (Pm–)Sm–Eu–Gd, Gd–Tb–Dy–Ho, and Er–Tm–Yb–Lu. The effect was first noted in liquid/liquid partition coefficients and stability constants for organic compounds (e.g., log K of EDTA in aqueous solutions). A number of workers have also suggested that the effect can be seen in REE patterns of a wide variety of geological samples, especially those that have been influenced by aqueous interaction (e.g., seawater, marine phases) or by late-stage magmatic fluids (e.g., pegmatites). The observations that have been used to demonstrate the effect are apparent discontinuities in REE patterns at the tetrad boundaries generating either M-shaped or W-shaped REE patterns.¹⁹ Although not directly related to the phenomena, significant deviations of the Y/Ho ratio from chondritic values are also thought to commonly accompany samples with the tetrad effect.²⁰

McLennan²¹ reviewed the geochemical literature on the tetrad effect and concluded that many of the cited examples could be explained as artifacts related to a variety of factors including incomplete analyses, analytical error, inappropriate choices for normalization, and complex mixing processes that resulted in apparent discontinuities. In addition, for many sample varieties (e.g., seawater, shales), the apparent effect is observed by some laboratories but not by others. The most compelling examples of geological environments appear to involve igneous rocks associated with late-stage magmatic fluids (e.g., pegmatites, leucogranites). In our judgment, the question of whether or not the tetrad effect is truly a significant geochemical influence will remain clouded until there is a careful and systematic interlaboratory comparison of the same samples, preferably using a variety of analytical methods.

4 MINERALOGY AND GEOLOGY OF RARE EARTH ELEMENTS

Although REE are trace elements in most rocks and minerals, there are about 200 minerals in which REE are essential structural constituents, forming necessary structural cationic components of the mineral, or are major substituting cations in the structure. Reviews of REE mineralogy, geochemistry, and geology can be found in several multiauthored books edited by Henderson²², Lipin and McKay²³, Möller et al.²⁴, and Jones et al.²⁵, the latter including a comprehensive appendix of REE minerals known to that date. Taylor and McLennan²⁶ also provided a comprehensive review of REE geochemistry.

Cation sites in most of the common igneous rock-forming minerals, such as olivines, pyroxenes, iron-titanium-oxides, feldspars, and micas, are characterized by highly variable coordination number and charge but with overall cation site conditions that are not particularly favorable to substantial REE substitution. This, of course, is the reason why the REE are usually incompatible elements. The abundances of REE consequently tend to be low (mostly less than 100–200 times CI values) but highly varied, ranging over about five orders of magnitude, from LREE enrichment to HREE enrichment and with highly variable Eu/Eu* (Figure 3). The major influences on the REE patterns in such minerals are the mineral-melt partition coefficients (Kd), the bulk composition and REE content of the parent magma, the major element chemistry (and thus coordination of cation sites) of the mineral, and the pressure–temperature conditions at which the mineral formed.

It is this variation and the distinctive REE abundances and patterns in a wide variety of common rock-forming minerals that make the REE such useful trace elements for evaluating most igneous and metamorphic petrogenetic processes. For example, Eu anomalies in magmatic rocks commonly indicate the involvement of plagioclase feldspar fractionation and HREE depletion commonly indicates a role for garnet fractionation in the history of their parent magmas. The stabilities of such minerals are sensitive to pressure, temperature, and bulk composition and thus REE patterns in rocks and minerals, coupled with an understanding of the partition coefficients, can be used to quantitatively constrain the origin and history of the magmas.

Figure 3 Chondrite-normalized REE patterns for common igneous rock-forming minerals. The igneous rock type from which the mineral was extracted is also listed. Data from compilation provided by Taylor and McLennan.²⁶ The extreme range of REE and the distinctive patterns for certain minerals is one of the reasons why REE are valuable trace elements for evaluating petrogenesis

REE compounds tend to be relatively insoluble in aqueous fluids and fluid–rock partition coefficients tend to be very low during most fluid–rock interaction processes (e.g., weathering, diagenesis, hydrothermal activity). Accordingly, REE contents of most natural waters are very low, typically in the subpart per billion to subpart per trillion range. Only in high temperature, low pH hydrothermal environments do concentrations rise to near the part per million level.⁸–¹¹ REE patterns of aqueous fluids are also highly variable, reflecting among other things the ultimate crustal or mantle sources of the dissolved REE, the nature of fluid–rock interactions (e.g., fluid/rock ratio), redox conditions and history, the nature of REE complexing ligands, and the pressure-temperature-compositional history of the fluids. In turn, minerals precipitated directly from most natural waters (e.g., carbonates, silica, evaporites, sulfides) have very low REE contents. REE are also particle-reactive under marine conditions and accordingly are readily scavenged and adsorbed onto a number of marine sedimentary particles, such as clay minerals and iron-oxides.²⁷

REE are highly electropositive (Table 1) resulting mostly in the formation of ionic compounds. REE mineral types thus include a wide variety of silicates, carbonates, oxides, phosphates, borates, halides, arsenates, sulfates, and vanadates. Among the most significant for geochemistry are lanthanite [(La,Ce,Nd)2(CO3)3·8H2O], bästnasite [(Ce,La)(CO3)F], allanite [(Ce,Y,Ca)2(Al, Fe³+)3(SiO4)3OH], and the phosphates florencite [(Ce,La)Al3(PO4)2(OH)6], monazite [(La,Ce,Th)(PO4)], rhabdophane [(La,Ce)(PO4)·H2O], and xenotime [YPO4]. Of these, the most important REE ore minerals include bastnaesite, monazite, xenotime, as well as a rare form of REE-bearing clays.²⁸

4.1 REE Ore Geology

Industrial uses of pure REE metals and compounds have expanded greatly from the early applications in incandescent gas light mantles, as polishing agents, and as glass coloring to being crucial components in a wide array modern technologies (e.g., computers, magnets, lasers, petroleum refining, alloys).²,³ An underlying driver of industrial development was the fundamental research carried out on the REE because of their production as fission products in the nuclear fuel cycle. The past several decades have witnessed an explosion in industrial use, much of it having considerable strategic importance (e.g., precision guidance systems, stealth technology, night vision). Accordingly, the geological processes giving rise to ore-grade concentrations of REE are also of considerable and growing interest.

There are a number of recent reviews of the geology, geochemistry, and origin of REE ore deposits²⁸–³⁵. The most important types of REE ore deposits include sedimentary-hosted carbonate bodies of controversial origin, igneous carbonatite bodies, heavy mineral placer sands, and several types of regolith deposit. REE patterns of major REE ores are plotted in Figure 4. Minor REE ore deposits include those related to alkaline magmatism and to early Precambrian uraniferous/auriferous quartz-pebble conglomerate paleoplacers.

By far, the largest REE ore body in the world is the Bayan Obo REE-Nb–Fe deposit hosted in Early–Middle Proterozoic carbonate rocks in north China (inner Mongolia). The geology of the deposit is complex and there is no consensus on a genetic model. The REE ore, characterized by extreme LREE enrichment and no Eu anomalies, may be related either to carbonatite magmatism and/or hydrothermal activity. The dominant REE ore minerals are bastnäsite, a readily processed REE carbonate, and monazite. Ore grades are in the range of ~5–6% REE2O3.

Figure 4 Chondrite-normalized REE patterns for selected REE ores from Bayan Obo and Mountain Pass; ion absorption clay ores from Longnan and Xunwu, China; and heavy mineral placer concentrates (monazite from Queensland Australia and xenotime from Malaysia)³²,³⁵. Note that the ion absorption clay REE patterns are normalized to 100%.

The Middle Proterozoic Mountain Pass LREE-Ba deposit, located in eastern California, is the second largest known economical deposit and has been mined semicontinuously since 1954 (and currently under redevelopment). The deposit is closely related to mantle-derived carbonatite magmatism with the major REE ore mineral being igneous bastnäsite, with ore grades of about 9% REE2O3 (other minerals in the ore include calcite, dolomite, and barite). The ore is also extremely LREE enriched but with somewhat higher HREE abundances than at Bayan Obo.

Placers represent the third most important REE ore variety that has or is being mined and occur in widely dispersed unconsolidated Neogene and Quaternary beach sands, the most important being Ti-mineral-rich beach sands on the coasts of both western and eastern Australia. REE are by-products of these deposits and the main REE ore minerals are monazite ± xenotime. Xenotime is of special interest because unlike most other REE ore minerals that are highly LREE enriched, xenotime is enriched in the HREE (Figure 4).

The fourth variety of REE ore includes those associated with regoliths. One such deposit, exploited in China, is composed of ion absorption clays forming weathering profiles, up to ~10 m thick, on petrologically evolved igneous rocks such as granites. Deposits may be enriched in either LREE or HREE (and Y) and exhibit negative Eu anomalies and negative Ce anomalies. Negative Eu anomalies indicate an upper crustal source of REE and negative Ce anomalies suggest ion exchange from groundwaters depleted in cerium due to removal of insoluble CeIV-oxides during weathering. Although clays are enriched in REE, the overall grades are low (<0.5% REE2O3) and deposits are economical because of the ease of extraction and HREE enrichment. Another example is residual laterites that form on REE-rich rocks. The Mount Weld deposit of Western Australia, a laterite formed on a carbonatite, represents a major deposit that at the time of writing was about to go into production.³⁵

5 DISTRIBUTION OF REE IN MAJOR SOLAR SYSTEM RESERVOIRS

5.1 Solar System Abundances

The sun comprises approximately 99.87% of the mass of the solar system and its composition is the most direct estimate of the composition of the current solar system and primordial solar nebula from which the planets were derived. The REE composition of the sun can be determined directly from spectral data for the solar atmosphere using both photospheric absorption lines and coronal emission lines. Concentration data are determined relative to some standardized concentration (typically silicon = 10⁶ atoms or hydrogen = 10¹² atoms) and one recent estimate is given in Table 2.³⁶

The second approach to estimating the composition of the solar nebula is from average Ivuna-class carbonaceous chondritic meteorites, or CI chondrites (Type 1 carbonaceous chondrites using older terminology). This class of meteorites is the most volatile rich and thus considered to be the most primitive available. CI compositions are essentially identical to spectroscopically determined solar photosphere compositions for a broad array of elements, from the most refractory to the most volatile, within analytical uncertainty. Accordingly, the REE content of average CI chondrites is taken as the best estimate of the solid fraction (i.e., metal fraction) of the solar nebula. Table 2 also includes an estimate of average CI chondrites recalculated to a comparable concentration scale as the solar photosphere by assuming identical silicon atomic abundances. An advantage of meteorite data for determining the composition of the solar nebula is that abundances have been determined much more precisely than are direct spectral measurements of the sun (compare uncertainties in Table 2).

5.2 Meteorites

REE distributions in meteorites and their mineralogical-lithological components provide fundamental information about the origin and early history of the solar system. Reviews of meteorite chemistry, including REE chemistry, can be found in Brearley and Jones³⁷, Mittlefehldt et al.³⁸, and McSween and Huss³⁹.

Table 2 Log atomic concentration of rare earth elements in the solar photosphere and CI chondrites

REE patterns in bulk carbonaceous chondrites are fairly uniform, parallel to CI, and show no dependence on volatility (e.g., no Eu or Yb anomalies). This uniformity also applies to the ordinary (H, L, LL classes) and enstatite (EL, EH classes) chondrite classes that show significant loss of their moderately and highly volatile elements (e.g., potassium, lead) and/or variations in their metal/silicate ratios. Accordingly, REE abundances in chondritic meteorites indicate no substantial cosmochemical fractionation (i.e., volatile related; redox related) during their formation in the early stages of solar system evolution and indicate broad homogeneity in the solar nebula.

On the other hand, millimeter to centimeter scale calcium–aluminum refractory inclusions (CAI) and individual refractory mineral inclusions (e.g., hibonite (CaAl12O19) and perovskite) from the Allende and Murchison carbonaceous chondrites provide evidence for local heterogeneity in the early solar nebula related to very high-temperature processes.⁴⁰,⁴¹ CAIs have been radiometrically dated as the oldest known objects in the solar system (4.567 billion years old) and exhibit highly variable REE patterns with variable Eu and Yb anomalies, in the case of CAIs, and Ce, Eu, and Yb anomalies in the case of hibonite grains (Figure 5). Such anomalies cannot be related to magmatic processes but instead are due to complex evaporation-condensation processes that fractionated the least refractory cerium, europium, and ytterbium from the other more refractory REE (Table 1). This veritable zoo of REE patterns calls for very high local temperatures, very complex histories of evaporation and condensation, and local compositional heterogeneity.

Silicate-bearing differentiated meteorites from the asteroid belt, including achondrites and stony-iron meteorites, are thought to represent fragments from asteroidal parent bodies that were melted very early in the history of the solar system and differentiated into core, mantle, and crust. REE data from these materials show significant variations, compared to chondrites, which reflect the magmatic histories of their parent bodies. For example, the howardite–eucrite–diogenite (HED) meteorites appear to be petrologically interrelated and probably are derived from the large asteroid 4-Vesta. REE patterns of the basaltic eucrites display variable positive and negative Eu anomalies that can be related to partial melting histories within Vesta.

Meteorites that do not come from the asteroid belt include the so-called SNC meteorites (for shergottites–nakhlites–chassignites) derived from Mars and the lunar meteorites. These were ejected from the planetary surfaces during impact processes and their REE compositions are highly variable, reflecting the magmatic evolution of these planetary bodies (see below).

Figure 5 Chondrite-normalized REE patterns for selected calcium–aluminum inclusions (CAI) from the Allende carbonaceous chondrite and refractory mineral grains (perovskite, hibonite) from the Murchison carbonaceous chondrite.⁴⁰,⁴¹ The highly irregular REE patterns, including anomalies for the least refractory Ce, Eu, and Yb, are indicative of localized very high temperatures leading to complex REE evaporation/condensation processes

5.3 Planetary Compositions

As refractory lithophile elements, the REE play an important role in constraining the overall composition and history of the silicate fraction of planets, which for the terrestrial planets is also termed their primitive mantle (equivalent to the present-day crust plus mantle). Since there is no evidence for significant planetary-scale fractionation of refractory elements during the assembly and differentiation of planetary bodies, it is widely accepted that the primitive mantles of terrestrial planets and moon possess chondritic proportions of the REE. As such, the absolute concentrations of REE (and other refractory elements) in primitive mantles provide an important constraint on the proportions of volatile elements to refractory elements and on the oxidation state (i.e., metal/silicate ratio) of the body. To date, the only major planetary bodies for which REE data are directly available are the Earth, Moon, and Mars, and Taylor and McLennan⁴² recently reviewed these data.

There is an enormous body of knowledge upon which to base estimates of the composition of the primitive mantle of the Earth. There are basically two approaches: (i) those based on the composition of xenoliths and high-degREE partial melts from the upper mantle, and (ii) those based on fundamental cosmochemical principles. In practice, both lines of evidence are employed and models differ mostly in the relative weight given to each. As discussed above, it is generally assumed that refractory elements are not fractionated from each other in the primitive mantles of planetary bodies. Thus, for the Earth, the REE abundances can be derived by assuming that ratios such as REE/Ca and REE/Al are in chondritic proportions. In turn, absolute abundances of refractory elements depend on the diluting effects of (i) amount of relatively volatile elements in the primitive mantle and (ii) oxidation state of the planet, which controls the proportion of iron partitioned into the metal core and silicate primitive mantle. Table 3 lists estimates for the REE contents of the primitive mantles of Earth, Mars, and the Moon, and chondrite-normalized plots are given in Figure 6.

Table 3 Estimates of REE concentrations in the primitive mantles of Earth, Mars, and Moon and crusts of Mars and Moon⁴²

The terrestrial planets and other large silicate-rich planetary bodies (e.g., large asteroids, moons) typically are differentiated into metal cores, silicate mantles, and incompatible element-enriched silicate crusts. The mechanisms, scales, and timing of this process are extremely variable.⁴² For REE, crust-mantle differentiation is the most important process in controlling abundances and distributions; REE are essentially excluded from planetary metal cores due to their lithophile character. In Table 3 and Figure 6, estimates of the REE distributions in planetary crusts (Earth’s continental crust, lunar highland crust, Martian crust) are also given. In the following sections, we discuss the REE content of the Earth’s crust in much greater detail.

Figure 6 Chondrite-normalized REE patterns of the primitive mantles and crusts of Earth, Mars, and the Moon. For the Earth, the continental crust is shown; for the Moon, the highland crust is shown; and for Mars, the bulk crust is shown

6 DISTRIBUTION OF REE IN TERRESTRIAL RESERVOIRS

The major geochemical reservoirs of the Earth that are currently in existence—inner and outer core, upper and lower mantle, upper and lower continental crust, oceanic crust, sedimentary shell, oceans, and atmosphere—were established early in the planet’s history. On the other hand, the sizes and compositions of these reservoirs have changed over geological time through a variety of processes and on vastly differing timescales, but largely controlled by the history of crust-mantle evolution associated with plate tectonics.

For example, the continental crust is mostly very old (>2 billion years old on average) but has grown and internally differentiated (into upper and lower crust) episodically over geological time with the greatest activity occurring in the Late Archean (~2.5–3.2 billion years ago), resulting in a major change in composition (including for REE) at about that time. Today, the continental crust ultimately is derived from partial melting in the upper mantle initiated by fluids (depressing melting temperatures) that are released from the top of the downgoing plate during subduction. This magmatic activity expresses itself as the large island-arc and continental-arc volcanic chains (e.g., Japan, Andes) that form behind the great ocean trenches that mark the sites of subduction. Once formed, this continental crust rapidly differentiates into a relatively felsic upper crust and relatively mafic lower crust, a process that also causes a major fractionation of REE distributions. Recycling of this low-density crust back into the mantle is minimal and, accordingly, the composition of the mantle in turn evolves largely in response to this long-term extraction of continental crust. At present, the continents grow at ~1 km³ per year on average but crustal growth rates were more rapid in the geological past.

The oceanic crust in contrast forms from high degrees of partial melting of the upper mantle beneath the mid-oceanic ridges, at the site of upwelling convecting mantle, giving rise to mid-oceanic ridge basalts (MORB) at a rate of about 20 km³ per year. The oceanic lithosphere, comprising the oceanic crust and its attached rigid uppermost mantle, is recycled and lost back into the deeper mantle at subduction zones over short timescales, with no oceanic crust being older than ~0.2 billion years old. This continuous, geologically rapid recycling gives rise to a largely steady-state process that over geological time accounts for about 10% of the mantle being comprised of this recycled oceanic crust. Compositional changes in both oceanic crust and mantle, such as those resulting from the hydrothermal interaction of seawater with hot ocean-floor magmas, do occur but from the perspective of REE distributions are of second order importance.

Oceanic crust and, to a lesser degREE continental crust, is also influenced by intraplate volcanism that occurs when relatively deep stationary plumes rise to the surface to form basaltic volcanoes; the best example being the Hawaiian chain of oceanic islands. Intraplate volcanism, giving rise to what is termed oceanic island basalts (OIB), results from a long and complex history within the mantle, giving rise to an average composition (including REE) quite distinct from MORB. Overall, intraplate volcanism accounts for about 1.5 km³ per year and accordingly, while highly visible at the Earth’s surface, comprises <5–10% of the oceanic crust and an even small component of the continental crust.

Although fundamentally different types of crust characterize other planetary bodies (e.g., the lunar highland crust and mare basalt crust of the Moon), the Earth is unique in having a mostly ancient continental crust and a mostly very young oceanic crust that are almost completely separated laterally. Within our solar system, plate tectonic processes, which gave rise to these distinct crustal types, appear to have only taken place on Earth.

6.1 Present-Day Mantle

The primitive mantle, discussed above, provides insight into the bulk composition of the silicate fraction of the Earth. However, over geological time, mantle compositions have evolved in a very complex manner due largely to plate tectonic processes. Among those processes are extraction of continental crust, recycling of oceanic crust back into the mantle, recycling of sedimentary components into the mantle, complex interactions of subducting lithosphere with the upper mantle-lower mantle boundary (at ~660 km depth), and so forth. The overall effect is that the mantle is heterogeneous on scales ranging from thousands of kilometers to just a few kilometers. Thus, for REE, upper mantle is likely distinct from lower mantle; lithospheric mantle beneath continents is distinct from lithospheric mantle beneath oceanic crust; mantle sources of MORB are distinct from those of OIB; broad regions of the upper mantle, identified by geophysics, may be distinct from other regions; and even along single segments of mid-oceanic ridges, basalts from one volcano can have distinct mantle REE sources from volcanoes just a few kilometers away.

Accordingly, it is difficult to identify and quantify the scales of significant mantle reservoirs and estimating average compositions is equally difficult. Since the oceanic crust is dominated by MORB, one useful concept is to define a mantle reservoir equivalent to the depleted mantle source of MORB (DMM). Since MORB are derived from high degrees of partial melting and the REE are incompatible, the DMM REE composition is roughly parallel to MORB but at lower absolute concentrations proportional to the average degREE of partial melting. One such estimate is provided in Table 4 and plotted in Figure 7.

6.2 Oceanic Crust

The oceanic crust varies from 0 km (where magmas are erupting at the oceanic ridges) to >8 km thickness and averages ~7 km well away from the ridges. It is composed of three main parts:

1. Layer 1: sedimentary cover up to ~1 km thick but mostly ≤0.5 km thick. The ultimate source of most sediment is continental weathering and accordingly, the REE signature is similar to upper continental crust (see below). Biological activity supplies carbonate and siliceous sediment, derived from seawater, but these materials are very low in REE abundances.

2. Layer 2: MORB-type basalt and intrusive equivalents with intercalated sediment lenses that varies in thickness, but on average is ~2.5 km.

3. Layer 3: MORB-type basalt and intrusive equivalents of up to about 4.5 km in thickness.

Table 4 REE concentrations in major oceanic crust-related geochemical reservoirs⁴²

Figure 7 Chondrite-normalized REE patterns of selected rocks and reservoirs from oceanic crustal environments. Note that MORB and DMM have parallel REE patterns, a reflection that MORB is derived from DMM by high degrees (≥ 10%) of partial melting. Also note that pelagic clays have REE patterns similar to shales derived from the upper continental crust. The slight negative Ce anomaly is significant and reflects a small component of authigenic material derived from seawater

In Table 4, several relevant compositions are given, including average MORB, OIB, and pelagic clay. Any estimate of the bulk composition of the oceanic crust is dominated by the MORB component. The REE patterns of MORB basalts are typically depleted in the LREE with flat HREE patterns (Figure 7). OIB differ from MORB, being significantly enriched by large-ion lithophile elements, including the LREE (Figure 7). Thus, in addition to the depleted MORB, OIB contributes an uncertain amount of these elements to the overall composition of the oceanic crust that is eventually subducted back into the mantle. In addition to the OIB lavas, an additional minor component is variable amounts of deep-sea sediment, whose REE budget is dominated by deep-sea clays (Table 4; Figure 7).

6.3 Continental Crust

Continental crust (referred to hereafter simply as crust) comprises only about 0.4% of the mass of the Earth but its geochemical importance is far greater since it contains ~30–50% of the budget of the Earth’s incompatible elements, including LREE. The crust has an average thickness of 41 km, ranging from 10 to 80 km, and has grown episodically over geological time with ~60% being in place by 2700 million years ago. A major difference in the style of crustal formation-evolution started in the Late Archean, beginning about 3200 million years ago.⁴³

During the Archean, higher heat flow resulted in partial melting of the downgoing slab at relatively shallow depth (~50 km) in subduction zones. This produces the eruption of silica-rich magmas, leaving garnet as a residual phase. Garnet has REE patterns that are depleted in LREE and enriched in HREE (Figure 3). The erupting lavas in turn have a reciprocal pattern, enriched in LREE and depleted in HREE (Figure 8a), typically with no Eu anomaly. During the Late Archean (3.2–2.5 billion years ago), modern-style plate tectonics became established and oceanic crust was older and colder by the time it reached subduction zones. Under these conditions, the downgoing slab does not melt but does release fluids into the overlying mantle, which in turn partially melts to produce the present subduction-zone island-arc and continental-arc magma suite, which is added to the crust. Within 50–100 million years of crust formation, further partial melting within the crust results from high abundances of radioactive elements (potassium, thorium, uranium) and intrusions of basaltic plumes at the base of the crust. Partial melts produced in the lower crust rise and generate an upper crust dominated by granodiorites and granites (Figure 8a).

Figure 8 (a) Chondrite-normalized REE patterns of selected igneous rocks from the continental crust. Note the substantial negative Eu anomaly in K-rich granitic rocks and the very steep, HREE-depleted character of Archean Na-rich granites from the TTG suite discussed the text. (b) Chondrite-normalized REE patterns for various averages and composites of shales and for average glacial loess. The remarkable uniformity in these REE patterns is taken as compelling evidence that they reflect the REE pattern of the upper continental crust exposed to weathering and erosion. (Data from compilation given in Taylor and McLennan.⁴³)

In summary, continental crust has grown in an episodic manner through geological time with a major increase in growth rate during the Late Archean. At present, the crust continues to grow by island-arc and continental-arc magmatism, followed by episodes of intracrustal melting that differentiates the upper from lower crust.

The average REE abundances of the bulk continental crust is model dependent and can be determined from the average composition of modern island-arc magmas and the composition of distinctive Archean igneous suites (see Refs 42 and 43 for details). The composition, so determined, is characterized by modest LREE enrichment and no Eu anomalies (Table 5; Figure 6).

For the upper continental crust, roughly the upper 10–12 km, highly reliable estimates of the average REE composition can be determined from sediments and sedimentary rocks. From the earliest days of REE analyses, the remarkable uniformity in sedimentary REE patterns throughout the post-Archean has been noted (Figure 8b) and this has been interpreted to reflect the average composition of the upper continental crust that erodes to produce clastic sediment. Such an interpretation is also consistent with the very low levels of REE in natural waters (see Section 6.4) and in chemical sediments precipitated from water (carbonates, evaporites). A major advantage of this approach is that sedimentary rocks can be used to evaluate the evolution of the upper crustal REE composition over geological time (see below).

Accordingly, REE in average shale (for this work, based on PAAS, post-Archean average Australian shale) is equated to the upper continental crust after a minor adjustment in total REE to account for low REE abundance sediments (sandstones, limestones). The derived pattern (Figure. 8b) differs from the bulk crust in having higher REE abundances, greater LREE enrichment, and a distinctive negative Eu anomaly (see Figure 9 for comparison). Elevated levels of the most incompatible LREE are consistent with intracrustal partial melting. The Eu anomaly indicates that partial melting to form the upper crust resulted in plagioclase being a stable residual phase that sequestered europium into the lower crustal residue. Since plagioclase is only stable to depths of 40 km on Earth, the negative Eu anomaly provides compelling evidence that the crust differentiated through intracrustal (shallow) partial melting processes. This REE pattern is observed in virtually all sedimentary rocks dating back to the Archan–post-Archean boundary, at which time sedimentary patterns change (see below).

Table 5 REE concentrations in the major continental crust-related geochemical reservoirs⁴²

The simplest method for estimating the REE composition of the lower crust is through the mass balance of subtracting 25% of upper crustal composition from the bulk crust (Table 5; Figure 9). This composition is characterized by the complementary positive Eu anomaly. Although the upper crustal negative Eu anomaly indicates the importance of intracrustal partial melting in governing the composition of the lower crust, xenoliths and granulite facies rocks provide additional constraints. Lower crustal xenoliths are commonly mafic in composition, and frequently show a relative enrichment in Eu. However, this enrichment is mostly related to the accumulation of cumulate phases rather than being due to residual phases from partial melts and so these rocks may represent basaltic magmas that were later added to the base of the crust. Granulite facies regions also commonly possess positive Eu anomalies, but these are in the more felsic rocks rather than mafic rocks that could represent residues after partial melting. Many such terranes appear, on compositional grounds, to be upper crust that has been buried in Himalayan-type continental collisions, and so many regional granulites likely formed in mid-crust regions and are not a good model on which to base lower crustal compositions. In summary, the lower crust appears to be essentially the mafic residue left after extraction of the granodioritic upper crust together with additions from underplating by basaltic magmas.

Figure 9 (a) Chondrite-normalized REE patterns for the upper, bulk, and lower continental crust reservoirs. (b) Upper and lower continental crust normalized to the bulk continental crust, highlighting the fractionation of europium and the more incompatible LREE during intracrustal differentiation

How far back in geological time can these crustal compositions be traced? The sedimentary record provides the most insight. Sedimentary REE patterns remain constant throughout the post-Archean; however, an abrupt change takes place at the Archean-Proterozoic boundary. Archean sedimentary rocks, though more variable, have REE patterns that on average differ fundamentally from post-Archean sediments, being less enriched in LREE and lacking the negative Eu anomaly (Figure 10). This change reflects an episodic change in upper crustal composition and is related to large-scale emplacement of K-rich granitic rocks, depleted in europium, in the upper crust toward the close of the Archean. This process, termed cratonization, produces large volumes of granites derived from massive intracrustal melting, transfers heat-producing elements to the upper crust, and fundamentally stabilizes the crust. The changes in both upper crustal compositions and the REE patterns of derived sediment were nonsynchronous over the globe, and extended over several hundred million years.

This change is also consistent with the Archean igneous record. In the Archean, igneous rocks characteristic of modern island-arc magmatism are scarce and instead, a bimodal suite of Na-rich felsic igneous rocks (tonalites, trondhjemites, granodiorites or the TTG suite and their volcanic equivalents) and basaltic rocks dominate. Archean basalts typically have flat REE patterns, whereas the TTG suite is characterized by very steep REE patterns (see Figure 8a). Mixtures of these rocks bear a superficial resemblance to the REE patterns of island-arc magmatic rocks such as andesites. On average, the Archean bulk crust was probably similar to but slightly less LREE enriched than the post-Archean bulk continental crust, but this similarity in average REE patterns belies the very different geological processes that gave rise to Archean and post-Archean continental crust.

The Archean crust probably consisted of many small fast-spreading plates. As described above, the TTG suite was produced by subduction of young, warm basaltic crust. The steep REE patterns indicate that garnet was in the residue during partial melting, indicating a mantle origin since garnet is only stable in mafic–ultramafic systems at depths below about 40 km. The Archean crust thus formed as a mixture of piled-up basalts and related volcanic rocks and TTG intrusions and extrusives. Only minor intracrustal melting appears to have occurred in the earlier part of the Archean, since negative Eu anomalies are rare, and areas of the crust that underwent such melting (i.e., generating upper crustal negative Eu anomalies) formed only localized cratonic regions of limited global extent.

Figure 10 Comparison of chondrite-normalized REE patterns for the Archean andpost-Archean upper continental crust, estimated from the sedimentary rock data. The absence of a negative Eu anomaly in the Archean upper crust indicates that intracrustal differentiation processes were not widespread at that time

6.4 The Hydrosphere

Oceans comprise 96.8% of the Earth’s near-surface water and accordingly, completely dominate the REE mass balance in the global hydrosphere. A thorough review of the geochemistry of REE in natural waters can be found in Byrne and Sholkovitz.²⁷ Most REE in both terrestrial and marine waters are derived ultimately from the upper continental crust and accordingly, normalization to average shale is most informative. One notable exception is that REE in marine hydrothermal fluids are derived ultimately from interactions with oceanic basalts.

The abundances of REE in ocean water are vanishingly low, in the part per trillion range, due to their low solubility in natural waters and efficient scavenging by sediment particles (Table 6; Figure 11). For perspective, a column of ocean water of average depth (~4 km) contains about the same content of REE as 1 mm of clay-rich sediment on the seafloor. Relative to shales, the average ocean water REE pattern is characterized by a negative Ce anomaly and a trend of HREE enrichment. This pattern results from a variety of complex processes involving surface chemistry on sedimentary particles and solution chemistry. Ce anomalies form through a coupling with manganese redox chemistry. In the marine environment, MnII oxidizes to MnIV and forms Mn-oxide particles that sink; the process is biologically mediated. On Mn-oxide particles, cerium oxidation takes place and CeIV is adsorbed onto the particle surfaces and preferentially removed from the water column compared to the trivalent REE. One important sink for CeIV is in diagenetically formed Mn nodules in seafloor sediment. Enrichment of HREE results from the fact that stability constants for many REE complexes (mainly carbonate complexes in seawater) increase with increasing atomic number and accordingly, the LREE are preferentially scavenged by sediment particles.

In detail, REE abundances in marine waters are heterogeneous both laterally and vertically. This is largely a result of complex scavenging by particles as they sink through the ocean column and the low residence times of REE in seawater, ranging from about 50 years for cerium, 240–500 years for the LREE, and 520–2900 years for the HREE (Table 6). These residence times are less than or comparable to the ~ 1000 year mixing times of the oceans. It is this heterogeneity, especially for the LREE, which makes the use of the ¹⁴⁷Sm⇒¹⁴³Nd isotope system very useful for paleoceanography because it is possible to trace distinctive water masses on the basis of their neodymium isotopic composition (¹⁴³Nd/¹⁴⁴Nd).

Table 6 REE concentrations in selected natural waters and residence time in seawater⁸,⁴³–⁴⁶

Figure 11 Chondrite-normalized REE patterns of selected natural waters

The REE contents in river waters are about an order of magnitude greater than in seawater but are also variable in detail (Figure 11). Patterns are mostly HREE enriched relative to average upper continental crust (and average shale) and the degree of fractionation and absolute concentrations are pH dependent such that higher pH results in lower absolute concentrations and more fractionated (HREE-enriched) patterns. Accordingly, the major controls on REE distributions in river waters are the upper crustal sources of REE coupled with preferential scavenging of LREE by particles and colloids (especially Fe colloids). Ce anomalies, where present, are more muted than those observed in seawater.

The final natural waters to be considered are hydrothermal waters that form by the circulation of warm surface waters (e.g., ocean water, groundwater) into the subsurface due to deep heat sources, such as magma chambers. In the marine setting, hydrothermal waters are mostly derived from the circulation of seawater through the oceanic crust in the vicinity of the ridges. These waters are surprisingly uniform from place to place and characterized by very large positive Eu anomalies (Figure 11) thought to reflect the dominance of plagioclase alteration in the subsurface basalt. Once vented onto the seafloor, the REE are rapidly scavenged and deposited in the sediment around the ridges. In terrestrial settings, hydrothermal waters commonly have relatively flat shale-normalized REE patterns, reflecting the upper crustal sources, with absolute abundances being pH dependent (Figure 11).

7 CONCLUSIONS

From the earliest foundations of geochemistry and cosmochemistry, the REE have proven to be crucial for understanding a myriad of processes that have influenced the origin and evolution of the solar system and Earth. REE data provided new and important constraints on wide-ranging topics as broad as constraining the earliest history of the solar system using refractory inclusions in meteorites to tracing the evolution of the Earth’s crust using sedimentary rocks to understanding paleoceanography using ancient marine authigenic phases. After 50 years of concerted effort, REE geochemistry is now a fairly mature research field. For example, high-quality REE determinations in geological materials, once considered a major analytical challenge, can now be obtained fairly routinely mainly using argon plasma mass spectrometry. Most available geological and cosmochemical reservoirs and processes have now been evaluated at some significant level of understanding for their REE geochemistry.

Although high-quality data can now be relatively easily obtained, there are still issues with poor data that pass peer review and gets published, and data quality remains a lessening but still significant issue. Although there are many international standards available for comparisons (probably too many!), anomalous samples for which data quality may still be an issue (e.g., the tetrad effect) would benefit from international interlaboratory comparisons.

Research that is in need of increased attention by geologists and geochemists is in the field of the economic geology of REE. Increased use of REE in strategically sensitive applications, coupled with limited occurrences of ore deposits—in terms of both numbers and locations-indicates a need for further research in at least two areas. The first is better understanding of the geological relationships leading to ore-grade REE occurrences (e.g., the origin of the largest REE ore deposit on Earth remains controversial) and the second is to further study the fundamental geochemistry of REE, using experimental methods, to better understand processes that may lead to such occurrences.

8 GLOSSARY

Archean Eon: The period of time between the occurrence of the oldest rocks on the Earth (around 4000 million years ago; prior to which is called the Hadean Eon) and the base of the Proterozoic eon at about 2500 million years ago.

Authigenic: A mineral or other sedimentary rock component that forms in place, rather than having been transported. May form either at the time the sedimentary rock was deposited or sometime after deposition by diagenetic processes.

Carbonaceous Chondrites: Type of stony meteorite, characterized by the presence of hydrated, clay-type silicate minerals and abundant organic matter. They represent among the most primitive material known in the solar system, and are thought to approximate the bulk composition of the accreting solar nebula.

Carbonatite: A carbonate rock (calcite-and/or dolomite-rich) of magmatic origin typically associated with alkaline igneous rocks and kimberlites.

Crystal Fractionation: Separation of mineral crystals from a magma, typically by settling or flotation due to density differences, resulting in magmatic differentiation.

Diagenesis: Chemical, physical, mineralogical, and biological transformations undergone by a sediment between time of initial deposition and metamorphism.

Granulites: Metamorphic rocks, commonly with coarse grains and gneissic textures, formed at very high pressure and temperatures, typical of lower continental crust.

Leucogranite: A type of highly evolved granite characterized by light color due to a low content of dark mafic minerals.

Lithosphere: Cool outer rigid layer of the Earth, including crust and uppermost mantle. Separated from deeper convecting layers of the mantle by the asthenosphere, a relatively weak zone. Thickness varies from zero at active mid-ocean ridges to as much as 200 km or more beneath continents.

Partial Melting: Incomplete melting of a rock mass. The magma commonly separates from the site of melting.

Pegmatite: A very coarse grained igneous rock, typically of granitic composition, that forms during the latest stages of magma crystallization when volatile contents are high.

Placer and Paleoplacer: Mineral deposits resulting from mechanical concentration of mineral particles by sorting according to density during sedimentary transport (placer deposit), which may also be preserved in ancient sedimentary rock sequences (paleoplacer).

Primitive Mantle: The mantle of a planet prior to the extraction of any crust. Mostly used as a geochemical concept that is the sum of the present-day mantle and crust (but excluding the core) of a planet.

Proterozoic Eon: The period in Earth history between the Archean and the Phanerozoic eons (2500-542 million years ago).

Regolith: The general term for fragmental, unconsolidated material that forms a cover over more coherent bedrock. May be either a residual or transported deposit.

Solar Photosphere: The bright visible surface of the sun, resulting from a layer of strongly ionized gases.

Xenolith: An inclusion within an igneous rock that is genetically unrelated to the enclosing rock.

9 RELATED ARTICLES

Lanthanides: Luminescence Applications; Lanthanides in Living Systems; Lanthanide Oxide/Hydroxide Complexes; Lanthanides: Coordination Chemistry; Sustainability of Rare Earth Resources; The Electronic Structure of the Lanthanides; Variable Valency.

10 ABBREVIATIONS AND ACRONYMS