You are on page 1of 30

Chemical Geology 439 (2016) 13–42

Contents lists available at ScienceDirect

Chemical Geology

journal homepage: www.elsevier.com/locate/chemgeo

Review Article

Hydrothermal transport, deposition, and fractionation of the REE:


Experimental data and thermodynamic calculations
A. Migdisov a,⁎, A.E. Williams-Jones b, J. Brugger c, F.A. Caporuscio a
a
Earth and Environmental Division, Los Alamos National Laboratory, P.O. Box 1663, M.S. J535, Los Alamos, NM 87545, United States
b
Department of Earth and Planetary Sciences, McGill University, 3450 University Street, Montreal H3A 0E8, Canada
c
School of Earth, Atmosphere and Environment, Monash University, 9 Rainforest Walk, VIC 3800, Australia

a r t i c l e i n f o a b s t r a c t

Article history: For many years, our understanding of the behavior of the REE in hydrothermal systems was based on semi-
Received 29 February 2016 empirical estimates involving extrapolation of thermodynamic data obtained at 25 °C (Haas et al., 1995; Wood,
Received in revised form 4 June 2016 1990a). Since then, a substantial body of experimental data has accumulated on the stability of aqueous com-
Accepted 8 June 2016
plexes of the REE. These data have shown that some of the predictions of Haas et al. (1995) are accurate, but
Available online 11 June 2016
others may be in error by several orders of magnitude. However, application of the data in modeling hydrother-
Keywords:
mal transport and deposition of the REE has been severely hampered by the lack of data on the thermodynamic
Rare earth elements properties of even the most common REE minerals. The discrepancies between the predictions of Haas et al.
Hydrothermal (1995) and experimental determinations of the thermodynamic properties of aqueous REE species, together
Transport with the paucity of data on the stability of REE minerals, raise serious questions about the reliability of some
Deposition models that have been proposed for the hydrothermal mobility of these critical metals.
Fractionation In this contribution, we review a body of high-temperature experimental data collected over the past 15 years on
the stability of REE aqueous species and minerals. Using this new thermodynamic dataset, we re-evaluate the
mechanisms responsible for hydrothermal transport and deposition of the REE. We also discuss the mechanisms
that can result in REE fractionation during their hydrothermal transport and deposition. Our calculations suggest
that in hydrothermal solutions, the main REE transporting ligands are chloride and sulfate, whereas fluoride, car-
bonate, and phosphate likely play an important role as depositional ligands. In addition to crystallographic frac-
tionation, which is based on the differing affinity of mineral structures for the REE, our models suggest that the
REE can be fractionated hydrothermally due to the differences in the stability of the LREE and HREE as aqueous
chloride complexes.
© 2016 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2. Aqueous speciation of the REE at elevated temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1. Temperature and composition of REE-bearing fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2. Thermodynamic stability of aqueous species of the REE (III) at hydrothermal conditions . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1. Thermodynamic stability of simple hydrated ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.2. Thermodynamic stability of aqueous REE chloride complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.3. Thermodynamic stability of aqueous REE fluoride complexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.4. Thermodynamic stability of aqueous REE sulfate complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2.5. Thermodynamic stability of aqueous REE hydroxyl complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2.6. Thermodynamic stability of aqueous REE carbonate and phosphate complexes. . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3. Stability of Ce(IV) and Eu(II) species in hydrothermal solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4. Co-ordination chemistry of aqueous REE species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3. REE-bearing minerals involved in hydrothermal processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.1. Hydrothermally formed minerals of REE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

⁎ Corresponding author.
E-mail address: artas@lanl.gov (A. Migdisov).

http://dx.doi.org/10.1016/j.chemgeo.2016.06.005
0009-2541/© 2016 Elsevier B.V. All rights reserved.
14 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

3.2. Thermochemistry and solubility of hydrothermally formed REE-bearing minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24


3.2.1. REE oxides and hydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.2. REE halogenides (chlorides and fluorides) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.3. REE phosphates (monazite and xenotime) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.4. REE fluorocarbonates (bastnäsite-parisite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4. Hydrothermal mobility and depositional mechanisms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.1. Hydrothermal mobility of the REE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.2. Depositional mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5. Hydrothermal fractionation of REE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.1. REE fractionation through differences in mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.2. REE fractionation through differences in solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.3. Crystallographic control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

1. Introduction involving ligands that are common in nature, e.g., hydroxyl, chloride,
fluoride, carbonate and sulfate. Another factor controlling the hydrother-
Rare earth elements (REE; Y and 14 elements from La to Lu) are mal mobility of the REE, and, equally, the processes of their fixation
metals that are in high demand by new and emerging technologies, par- and concentration (depositional mechanisms), is the solubility of REE-
ticularly those focused on energy and environmental applications. Be- bearing minerals. These two factors are intimately linked in hydrothermal
cause their production is restricted largely to a single country (China) processes. Thus, for example, the high stability of an aqueous species
and their supply is therefore vulnerable, the REE are considered critical (e.g., REEF2+; see below) does not necessarily mean that this species
metals, and there is considerable pressure to find exploitable resources can contribute significantly to the hydrothermal transport of the metal
of them in other countries. This has resulted in a sharp increase in the of interest. If the solubility of the thermodynamically stable solid corre-
number of studies designed to understand the behavior of these ele- sponding to this species (e.g., REEF3(s); see below) is very low, the contri-
ments and thereby develop tools that can be used to predict the factors bution of the species in question to the metal of interest will be negligible.
controlling their mobilization and concentration in the Earth's crust. In The opposite may also be true. It is possible that a species having relatively
addition to their economic value, the REE offer researchers valuable low stability in aqueous solutions may be capable of playing a major role
tools for addressing fundamental issues relating to the origin and evolu- in the transport of the metal, if the ligand is present in high concentration
tion of our planet. The near universal occurrence of the REE at low, but in the hydrothermal fluid (e.g., Cl−) and the corresponding solids are
detectable levels, together with the slightly differing responses of sub- highly soluble. Reliable data on the thermodynamic properties of aqueous
sets of these elements to changing physico-chemical conditions, make species of the REE and for REE-bearing minerals are therefore pre-
them powerful tools for interpreting processes in a wide variety of geo- requisites for developing a quantitative understanding of the hydrother-
logical environments. Optimizing the application of these tools, howev- mal mobilization and concentration of the REE.
er, requires comprehensive understanding of the processes responsible The need for data on the thermodynamic properties of aqueous REE
for mobilization, fractionation, and concentration of the REE. species at elevated temperature has been widely appreciated since the
Our understanding of the geochemical behavior of the REE has ad- late 1980s. As experimental data on these species at elevated tempera-
vanced considerably during the last 30 years. For a long time, the REE ture were not available at this time, a number of studies attempted to
were thought by many to be effectively immobile in igneous rocks evaluate their thermodynamic properties using semi-empirical extrap-
that had been subjected to hydrothermal alteration. Indeed, in some olations. Prominent among them were the studies of Wood (1990a,
studies, the REE were even used to evaluate mass changes in elements 1990b), who used an isocolumbic approach, and Haas et al. (1995),
resulting from hydrothermal alteration. During the 1990s, however, who applied the HKF model (Shock and Helgeson, 1988) in predicting
mounting evidence from natural systems combined with theoretical the thermodynamic properties of aqueous complexes of the REE for a
evaluations of REE mobility and the first experimental studies of the sta- variety of ligands and a wide range of pressure and temperature.
bility of REE species in hydrothermal fluids completely discredited the These two studies, and especially that of Haas et al. (1995), became
notion of REE immobility (e.g., Williams-Jones and Wood, 1992; Gam- the main source of data for geochemical studies of REE mobility in hy-
mons et al., 1996). As a result, it is now almost universally agreed that drothermal systems. These data are still used extensively by researchers
hydrothermal fluids commonly mobilize the REE and, in some cases, in modelling the hydrothermal transport and deposition of the REE in
concentrate them to economic levels. The most compelling evidence nature (e.g., Jones et al., 1996; Lewis et al., 1998; Agangi et al., 2010).
of this is the occurrence of REE deposits of dominantly hydrothermal or- However, since the predictions of Haas et al. (1995) were published, a
igin, such as the giant Bayan Obo REE deposit in China (e.g., Chao et al., substantial body of experimental data has accumulated on the stability
1992; Smith et al., 2000). Hydrothermal mobilization and concentration of aqueous complexes of the REE (see below). While some of the predic-
of REE have also been documented for a number of other, smaller, but tions of Haas et al. (1995) have been confirmed, others differ by several
still potentially economic deposits ( e.g., Olivo and Williams-Jones, orders of magnitude from the experimentally determined values. The
1999; Lehmann et al., 1994; Williams-Jones et al., 2000; Cook et al., latter is not unexpected considering that the predictions were based
2013; Trofanenko et al., 2016). It is therefore becoming increasingly ap- on limited sets of data for ambient temperature. Indeed, it is quite sur-
parent that evaluation of the processes by which the REE are concen- prising that so many of the predictions returned values that were later
trated to economic levels in the Earth's crust requires a quantitative confirmed by experimental studies (e.g., those on the chloride com-
understanding of the factors responsible for their mobility and concen- plexes of the MREE, and monosulfate REE complexes). However, the
tration in/by hydrothermal solutions. discrepancies between the predictions of Haas et al. (1995) and exper-
The ability of aqueous solutions to mobilize REE from their host rocks imental determinations of the thermodynamic properties of other aque-
depends largely on the stability of REE complexes formed with ligands ous REE species, for example, fluoride and bisulfate complexes, raise
(anionic species), and the availability of these ligands in the solutions. questions about the reliability of a number of the models that have
Therefore, one of the major concerns is the stability of REE complexes been proposed for the hydrothermal mobility of these metals, which
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 15

are mostly based on the predictions of Haas et al. (1995) fluids are no exception. Fluid inclusion studies suggest that the Bayan
(e.g., Williams-Jones et al., 2000; Holtstam and Andersson, 2007; Obo deposit was formed from brines containing 7 to 10 wt.% NaCl equiv-
Ruberti et al., 2008; Xu et al., 2010). alent (Smith and Henderson, 2000). The fluids responsible for ore depo-
A potentially even more serious problem for evaluating the hydro- sition at Gallinas Mountains contained 12 to 18 wt.% NaCl equivalent
thermal mobility of the REE has been the lack of data on the thermody- (Williams-Jones et al., 2000), and the Karonge deposits formed from a
namic properties of even the most common REE minerals. Whereas brine containing N 25 wt.% NaCl equivalent (Lehmann et al., 1994). The
popular databases such as SUPCRT (Johnson et al., 1992) or Thermocalc study of the fluid inclusions from the Capitan Pluton, New Mexico by
contain thermodynamic data for a large number of minerals that are Banks et al. (1994) provides the most detailed characterization of REE-
known to control the mobility of the base and precious metals, these da- bearing fluids to date, and suggests that the mineralizing fluids carried
tabases are conspicuous by their absence of geologically relevant REE up to 44 wt.% Cl.
minerals. Recently, some thermodynamic data have become available In addition to chloride, the study of Banks et al. (1994) documented
for a few of these minerals and, although these require validation, appreciable concentrations of sulfate (up to 2.4 wt.%) and fluoride (up
they are showing that existing models of hydrothermal transport, depo- to 0.5 wt.%) in the fluids responsible for REE transport and deposition.
sition and fractionation of the REE may require re-evaluation. An important role for these ligands in the hydrothermal mobilization
The aim of this paper is to provide a comprehensive review of the of the REE is also evident from the occurrence of sulfate- and fluoride-
high temperature experimental data on the nature and stability of aque- bearing minerals, such as barite and fluorite, in a number of REE de-
ous REE species and on the stability of REE-bearing minerals that have posits (Smith et al., 2000; Gültekin et al., 2003; Fan et al., 2004; Liu,
become available since 1995. In light of these data, we conduct a prelim- 2004;) and the observation that fluorocarbonates [e.g., bastnäsite-
inary re-evaluation of models that have been proposed to explain the (Ce), parisite-(Ce) and synchysite-(Ce)] are commonly the dominant
behavior of the REE in hydrothermal systems. ore minerals (Smith and Henderson, 2000; Williams-Jones et al., 2000;
Fan et al., 2004). Two other important ligands for the hydrothermal geo-
2. Aqueous speciation of the REE at elevated temperature chemistry of the REE are carbonate and phosphate. Indeed, as noted
above, fluorocarbonates are important REE ore minerals
2.1. Temperature and composition of REE-bearing fluids (Williams-Jones et al., 2000; Long et al., 2010) and the same is true for
the phosphates, monazite-(Ce) and xenotime-(Y) (Smith and
The stability of aqueous REE species (both absolute values and rela- Henderson, 2000; Fan et al., 2004). Understanding the mechanisms
tive to each other) varies appreciably with temperature. For example, at governing interaction of the REE with these and the other ligands men-
ambient temperature, the heavy REE fluoride species, HREEF2 +, are tioned earlier is essential for quantifying the transport and deposition of
more stable than the light REE species, LREEF2+, whereas at high tem- these metals. Finally, hydroxyl is a constituent of any aqueous solution
perature the opposite is true (Migdisov et al., 2009). Based on fluid in- and is present in high concentrations in alkaline (high pH) fluids. Al-
clusion homogenization temperatures, the ores of the largest of though hydroxyl complexes are typically insignificant at acidic condi-
known REE deposit, Bayan Obo, China, are interpreted to have formed tions, at near-neutral and alkaline conditions their contribution to REE
at temperatures between 300 and 400 °C (Smith and Henderson, mobility in aqueous solutions could be considerable. There have been
2000). The same temperature range was interpreted for formation of very few studies that have documented the pH of hydrothermal fluids
the Gallinas Mountains deposits, USA, also based on fluid inclusion ho- responsible for REE ore formation. One of these is the study of Gysi
mogenization temperatures (Williams-Jones et al., 2000), and prelimi- and Williams-Jones (2013) who showed that the fluids responsible for
nary data for the Karonge deposits suggest that they formed at the mobilization of REE at Strange Lake, Canada, had a pH of ~2.
N420 °C (Lehmann et al., 1994). Finally, fluid inclusion homogenization
temperatures reported for the Capitan Pluton, USA, suggest that the as- 2.2. Thermodynamic stability of aqueous species of the REE (III) at hydro-
sociated REE mineralization formed between 260 and 480 °C (Banks thermal conditions
et al., 1994). This temperature range (250–500 °C) therefore corre-
sponds broadly to the temperature interval of primary relevance for The REE occur naturally in the trivalent form, except for Eu and Ce,
the hydrothermal transport and deposition of the REE (Table 1). which also can be bivalent (Eu) and tetravalent (Ce), respectively.
An important control on the hydrothermal mobility of the REE (and These exceptions will be discussed in Section 2.3. In aqueous solutions,
metals in general) is the availability and concentration of ligands with the REE are transported both as hydrated aqueous ions (e.g., Nd3+) and
which they can form stable aqueous complexes. Chloride is the domi- as aqueous complexes, e.g., NdCl2+, NdCl+ 2 , with the latter generally
nant ligand in natural aqueous fluids, and REE-forming hydrothermal predominating over the hydrated ions, particularly at elevated

Table 1
Some hydrothermal REE deposits and the chemistry of the associated fluids.

Name and location; type; references Major REE minerals Reserves Fluid homogenization temperature, salinity and other chemical
characteristics

Bayan Obo, China (Smith and Henderson, 2000; Monazite-(Ce) and 92 Mt total REE+Y (mainly Early monazite-(Ce) stage: N280–330 °C at P N 0.7 kbar, 1–5
Weng et al., 2015) bastnäsite-(Ce) La,Ce) out of 1540 Mt ore wt% NaCl, XCO2 from 0.3 to 0.45. Main stage bastnäsite-(Ce):
(12,620 ppm La, 26,012 Ce). N400 °C to 300 °C at P N .9 to 1.4 kbar. 6–10 wt.% NaCl, XCO2 0.3
due to carbonate dissolution.
The Gallinas Mountains, New Mexico Bastnäsite-(Ce) 15.84 kt total REE + Y out of ~400 °C, sulfate-rich NaCl-KCl brines having a salinity of ~15
(McLemore, 2011; Williams-Jones et al., 2000) 537,000 t of ore (2.95%) wt.% NaCl equivalent
Karonge, Burundi (Lehmann et al., 1994; van Bastnäsite-(Ce), Preliminary homogenization data suggest that they formed at
Wambeke, 1977) monazite-(Ce) N420 °C
Capitan Pluton, New Mexico (Banks et al., 1994) Allanite-(Ce), REE-rich Not economic Homogenization temperatures 260 to 480 °C; 80 wt.% total salt,
titanite including up to 44 wt.% Cl, 5245 ppm F, 24,210 ppm SO4
Olympic Dam, South Australia. IOCG deposit Bastnäsite-(Ce), 53 Mt total REE+Y (mainly Magnetite stage: N400 °C, medium-hyper saline(20–45 wt.%
(Bastrakov et al., 2007; Hitzman et al., 1992; florencite-(Ce), La,Ce) out of 9576 Mt ore NaCl), δ18O +7.7 to +12.8‰; Hematite stage: 150–300 °C, 1–8
Oreskes and Einaudi, 1992; Weng et al., 2015) monazite-(Ce), (1,705 ppm La, 2561 Ce). wt.% NaCl, δ18O-2.5 to +4.5‰
xenotime-(Y),
britholite-(Ce)
16 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

temperature. The stability of aqueous complexes in hydrothermal solu- and Helgeson (1988). For example, the stability of the Nd monochloride
tions is therefore one of the major factors controlling REE mobility in na- is conventionally reported by values of β for a reaction involving the for-
ture. As discussed above, owing to the lack of experimental data, mation of this complex from the hydrated Nd3 + and Cl− ions
information on the behavior of the REE at elevated temperature, until (e.g., Migdisov and Williams-Jones, 2002):
recently, has come from extrapolations based on experimental data col-
lected at ambient temperature. The most widely used sources of these Nd3 þ þ Cl− ¼ NdCl2þ ; logβ1Cl ðT; PÞ
data are Haas et al. (1995), who evaluated the stability of the most rel- ¼ logaðNdCl2 þÞ–logaðNd3 þÞ−logaðCl− Þ ð1Þ
evant aqueous complexes of the REE, and Shock and Helgeson (1988) The thermodynamic properties of an aqueous complex like NdCl2+
and Shock et al. (1997), who predicted the stability of simple hydrated are therefore typically derived with respect to the properties of the sim-
REE ions. In contrast, the behavior of REE aqueous complexes at ambient ple hydrated ions, in this example, Nd3+ and Cl−:
temperature is well documented as a result of a large number of exper-
imental studies dating back to the 1950s (reviewed in Wood, 1990b). ΔGf ;T;P ðNdCl2 þÞ ¼ ΔGf ;T;P ðNd3 þÞ þ ΔGf ;T;P ðCl− Þ−RTlnβ1Cl ðT; PÞ ð2Þ
The accumulation and refinement of these data is still in progress and
studies on the stability of aqueous REE complexes at room temperatures This is a matter of considerable concern because the reliability of the
still appear regularly (e.g., Soderholm et al., 2009; Smirnov and Grechin, experimentally determined stability constants is heavily dependent on
2013; Stepanchikova et al., 2014). A critical analysis of the available am- the reliability of sets of theoretically predicted values. The same concern
bient temperature data is beyond the scope of this paper. However, in also applies to a large proportion of the data available for the high tem-
our view, the most reliable data are those of Klungness and Byrne perature stability of REE-bearing minerals, as in many cases these data
(2000); Liu and Byrne (1995, 1997); Luo and Byrne (2000, 2001, are derived from solubility experiments (solubility products, Ks) and
2004, 2007); Schijf and Byrne (1999, 2004), and Luo and Millero therefore also rely on the data available for ligands and simple hydrated
(2004), which report formation constants of the REE at standard condi- ions (e.g., Poitrasson et al., 2004; Cetiner et al., 2005):
tions for all the naturally occurring inorganic ligands referred to above.    
−monazite−ðNdÞ 3þ 3þ
These studies systematically evaluated the stability of all the REE using NdPO4 þ PO3−
¼ Nd ; logKs ðT; PÞ ¼ loga Nd þ loga PO3− ;
4    4
the same experimental methods, and provide indispensable informa- ΔGT;P ðmonazite‐ðNdÞÞ ¼ ΔGT;P Nd

þ ΔGT;P PO3− −RT lnKs ðT; PÞ
4
tion on the trends in inter-element stability of the REE, which is essen-
ð3Þ
tial for understanding differences in the behavior of the various REE in
aqueous solutions (e.g., their fractionation). This information is not
available from other studies. If a refinement of the properties of simple hydrated ions at elevated
temperature were deemed necessary, this would likely require that a
2.2.1. Thermodynamic stability of simple hydrated ions large proportion of the data derived using solubility methods also be
Although experimentally determined stability constants have been re-evaluated. Since 1998, a number of studies have investigated the be-
reported for aqueous REE species at elevated temperature for over havior of simple hydrated REE ions (Mayanovic et al., 2002a, 2002b,
20 years, these constants (β) are presented using the stability of the 2007, 2009a, 2009b; Ragnarsdottir et al., 1998). These studies, however,
simple hydrated ions and their ligands as references, and the latter are have focused on the hydration structure and co-ordination of these ions,
based on the theoretical predictions of Shock et al. (1997) and Shock and have not attempted to revise the values recommended by Shock

Table 2
The HKF equation of state parameters for the simple hydrated REE ions and the main naturally occurring ligands.

Species ΔG0298 S0298 a1 ∗ 10 a2 ∗ 10−2 a3 a4 ∗ 10−4 c1 c2 ∗ 10−4 ω ∗ 10−5 Z Reference


(cal) (cal K−1)

La3+ −164000 −52 −2.788 −14.3824 10.9602 −2.1844 4.2394 −10.6122 2.1572 3 [1]
Ce3+ −161600 −49 −3.4833 −16.2789 12.1302 −2.1059 −0.355 −12.751 2.3265 3 [2]
Ce4+ −121300 −100.1 −4.2792 −18.2247 12.8996 −2.0255 40.279 −3.0141 3.6964 4 [2]
Pr3+ −162600 −50 −3.2061 −14.2894 8.5328 −2.1882 3.9607 −11.2844 2.3369 3 [2]
Nd3+ −160600 −49.5 −3.3707 −14.5452 8.3211 −2.1777 1.6236 −11.8344 2.255 3 [1]
Sm3+ −159100 −50.7 −3.2065 −15.6108 11.8857 −2.1337 1.9385 −11.8548 2.2955 3 [1]
Eu2+ −129200 −2.4 0.0407 −7.6776 8.7578 −2.4615 9.5539 −5.3567 1.0929 2 [2]
Eu3+ −137300 −53 −3.1037 −15.3599 11.7871 −2.144 6.0548 −10.49 2.3161 3 [1]
Gd3+ −158600 −49.2 −2.9771 −15.0506 11.6656 −2.1568 6.5606 −10.3474 2.3265 3 [1]
Tb3+ −159500 −54 −2.9245 −14.9162 11.5979 −2.1623 7.1853 −10.3677 2.4007 3 [2]
Dy3+ −158700 −55.2 −3.0003 −15.1074 11.6879 −2.1545 9.5076 −9.4919 2.3792 3 [1]
Ho3+ −161400 −54.3 −3.1198 −15.3992 11.8026 −2.1424 8.6686 −9.8178 2.3899 3 [1]
Er3+ −159900 −58.3 −3.3041 −15.8492 11.9794 −2.1238 8.2815 −10.0215 2.4115 3 [1]
Tm3+ −159900 −58.1 −3.2967 −15.8312 11.9724 −2.1245 8.4826 −10.0215 2.4333 3 [1]
Yb3+ −153000 −56.9 −3.4983 −16.3233 12.1658 −2.1042 7.3533 −10.4493 2.4443 3 [1]
Lu3+ −159400 −63.1 −3.563 −16.4812 12.2279 −2.0977 9.565 −9.716 2.4554 3 [1]
CO2° −92250 28.1 6.2466 7.4711 2.8136 −3.0879 40.0325 8.8004 −0.02 0 [3]
CO2−
3 −126191 −11.95 2.8524 −3.9844 6.4142 −2.6143 −3.3206 −17.1917 3.3914 −2 [3]
HCO− 3 −140282 23.53 7.5621 1.1505 1.2346 −2.8266 12.9395 −4.7579 1.2733 −1 [3]
3−
PO4 −243500 −53 −0.5258 −9.0576 9.2927 −2.4045 −15.1599 −28.4155 5.6114 −3 [2]
2− −
HPO4 260310 −8 3.6315 1.0857 5.3233 −2.8239 2.7357 −14.9103 3.3363 −2 [3]
H2PO− 4

270140 21.6 6.4875 8.0594 2.5823 −3.1122 14.0435 −4.4605 1.3003 −1 [3]
− − −
H3PO4° 273100 38 8.2727 12.4182 0.8691 3.2924 17.9708 1.7727 0.22 0 [3]
− − − − −
SO2−
4 177930 4.5 8.3014 1.9846 6.2122 2.697 1.64 17.998 3.1463 -2 [3]
HSO− 4

180630 30 6.9788 9.259 2.1108 −
3.1618 20.0961 −1.955 1.1748 −1 [2]

F −67340 −3.15 0.687 1.3588 7.6033 −2.8352 4.46 −7.488 1.787 −1 [3]
HF° −71662 22.5 3.4753 0.7042 5.4732 −2.8081 14.3647 −0.1828 −0.0007 0 [3]
Cl− −31379 13.56 4.032 4.801 5.563 −2.847 −4.4 −5.714 1.456 −1 [3]
HCl° −30411 0.42 16.1573 −11.4311 −46.1866 −2.3036 46.4716 −5.2811 0 0 [4]

References: [1]: (Shock and Helgeson, 1988) [2]: (Shock et al., 1997) [3]:(Johnson et al., 1992) [4]: (Tagirov et al., 1997)..
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 17

et al. (1997) and Shock and Helgeson (1988). In this contribution we spectroscopy (Migdisov and Williams-Jones, 2002, 2006;
therefore used the latter sources of data for the simple hydrated REE Stepanchikova and Kolonin, 2005; Migdisov et al., 2008), and in situ
ions; those for the ligands were also taken either from these sources EXAFS studies (Mayanovic et al., 2002a, 2002b, 2007, 2009a, 2009b).
or from Johnson et al. (1992). Table 2 lists the HKF parameters recom- The studies of Gammons et al. (2002) and Migdisov et al. (2009) in-
mended for these species by the above authors. In order to maintain in- volved simultaneous determination of the stability of a large number
ternal consistency of the data presented in this contribution, all of REE in chloride-bearing solutions, whereas the other studies were de-
derivations of the stability constants for aqueous species and solubility voted to single elements. These studies have demonstrated that in solu-
products for minerals were performed using the data listed in Table 2. tions containing up to 5 M NaCl at temperatures up to 300 °C, the
dominant REE chloride complexes are the mono- and dichloride species,
2.2.2. Thermodynamic stability of aqueous REE chloride complexes REECl2+ and REECl+ 2 . Fig. 1 shows the values of the formation constants
It is well-known that the REE form weak complexes with chloride for these species at 150, 200, and 250 °C recommended in the above
ions (Wood, 1990a; Haas et al., 1995; Luo and Byrne, 2001). However, publications, and compares them with the values calculated using the
as chloride is the dominant ligand in natural hydrothermal solutions predictions of Haas et al. (1995). Fig. 1A illustrates the inter-element de-
and some of the best studied REE deposits are known to have formed pendencies among the lanthanoids of the formation constants for
from brines with high salinity, chloride-bearing systems have attracted monochloride aqueous complexes (β1; REE3+ + Cl− = REECl2 +),
considerable attention. As a result, the stability of REE chloride com- Fig. 1B shows these dependencies for dichloride complexes (β2;
plexes at elevated temperature has been subjected to much greater REE3+ + 2Cl− = REECl+ 2 ), and, considering that Gammons et al.
scrutiny than REE complexes involving other ligands. The hydrothermal (2002) reported their data in the form of stepwise formation constants
speciation of the REE in chloride-bearing solutions has been investigat- (K2; REECl2+ + Cl− = REECl+ 2 ), Fig. 1C compares the data of Gammons
ed using a variety of experimental methods, including solubility et al. (2002) for 200 °C with other data re-calculated to the stepwise for-
(Gammons et al., 1996, 2002; Migdisov et al., 2009), in situ UV–Vis mation constants for the dichloride complexes. Despite some scatter,
the data available on the stability of chloride complexes of the REE are
in fairly good agreement, except for the data on the stability of REECl+ 2
4 A reported by Stepanchikova and Kolonin (2005), which we believe
were compromised by the methods employed by these researchers in
3.5
250 °C de-convoluting their spectra. For a more detailed discussion of the
3 data for REE chloride species, readers are referred to Migdisov et al.
200 °C (2009). The data show that the overall stability of REE chloride species
1
Cl

2.5
increases with increasing temperature. For the LREE, for example, this
log

2 stability rise is ~2 orders of magnitude for an increase in temperature


from 150 to 250 °C. The data also demonstrate that the LREE are more
1.5
150 °C stable, and, thus, more mobile than the HREE in chloride-bearing aque-
1 ous solutions, and that this difference in stability increases with increas-
ing temperature. It also can be seen that the model of Haas et al. (1995)
0.5 accurately predicts the stability of chloride complexes of the middle REE
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
(MREE), but underestimates the stability of these complexes for the
5.5 LREE and overestimates it for the HREE. Table 3 lists the HKF parameters
5.0
B recommended by Migdisov et al. (2009), based on the new experimen-
250 °C tal data for REE chloride complexes. Here and for other types of species
4.5
4.0
and minerals discussed in this contribution, the corresponding tabulat-
2

ed (for swp) apparent Gibbs free energies of formation are reported in


Cl

3.5
the electronic supplement to this paper. Note that Table 2 does not in-
log

3.0
200 °C clude the HKF parameters for chloride complexes of Y, since these spe-
2.5
cies have not yet been experimentally investigated at elevated
2.0 temperature, and neither Wood (1990a) nor Haas et al. (1995) evaluat-
1.5 ed their high-temperature properties.
1.0 Fig. 2 illustrates the distribution of Nd species as a function of chlo-
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu ride concentration calculated using the data of Migdisov et al. (2009)
2 for a system containing Nd3 +, NdCl2 +, and NdCl+ 2 at 100, 200 and
C 300 °C. The calculations were performed for NaCl solutions; concentra-
1.5 tions of chloride and Nd-bearing species were calculated using the ac-
200 °C
tivity model reported in Helgeson et al. (1981) and Oelkers and
1
log K2lC

Helgeson (1991). At 100 °C (Fig. 2A), the simple hydrated ion predom-
inates, and chloride complexes of Nd contribute to the mass of dissolved
0.5
Nd, but only in highly saline solutions. However, at higher temperature,
0 the importance of chloride complexes increases significantly, and at
300 °C (Fig. 2C) chloride species of Nd predominate except in highly di-
-0.5 lute solutions, which are not considered in the figure.
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Mayanovic et al. (2002a, 2007, 2009a, 2009b) determined the num-
Gammons et al. (1996) Stepanchikova and Kolonin (2005) Haas et al. (1995) ber of chloride ions bond to REE using in situ XAS. The data reported in
Migdisov et al. (2009) Gammons et al. (2002)
Table 3 were used to calculate average chlorination numbers values that
are compared with the XAS data in Table 4. The agreement is nearly per-
Fig. 1. Comparison of the values of the first formation constants (A), the second formation fect for Gd and Yb, but XAS data suggest a considerably higher stability
constants (B), and the stepwise formation constants (C) of REE chloride complexes
obtained experimentally and predicted theoretically from the data of Haas et al. (1995)
for the chloride complexes of the LREE (La, Nd), especially at elevated
(triangles). temperature. It is not clear what caused this disagreement. We do not
From Migdisov et al. (2009). exclude a problem with the XAS determination, considering that this
18 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Table 3
The HKF equation of state parameters for aqueous REE chloride species (Migdisov et al., 2009).

Species ΔG0298 S0298 a1 ∗ 10 a2 ∗ 10−2 a3 a4 ∗ 10−4 c1 c2 ∗ 10−4 ω ∗ 10−5 Z


(cal) (cal K−1)

LaCl+2 −196390 −27.49 −0.0557 −7.9173 8.8619 −2.4517 17.7012 −3.7964 1.5045 2
CeCl+2 −193910 −18.59 −0.9537 −10.11 9.7237 −2.3611 2.6518 −8.5997 1.3697 2
PrCl+2 −194950 − −0.6264 −9.3107 9.4096 −2.3941 12.6813 −5.2997 1.4284 2
NdCl+2 −192950 −24.08 −0.7662 −9.6522 9.5438 −2.38 9.3305 −6.5423 1.453 2
SmCl+2 −191390 −25.64 −0.5918 −9.2262 9.3764 −2.3976 9.4219 −6.5851 1.4765 2
EuCl+2 −169620 −30.1 −0.4595 −8.9032 9.2495 −2.4109 18.8766 −3.5133 1.5441 2
GdCl+2 −190950 −27.98 −0.3348 −8.5987 9.1298 −2.4235 19.5037 −3.1935 1.512 2
TbCl+2 −191790 −32.43 −0.2834 −8.4732 9.0804 −2.4287 20.0003 −3.2342 1.5792 2
DyCl+2 −191030 −35.34 −0.342 −8.6162 9.1367 −2.4228 26.0699 −1.2644 1.6233 2
HoCl+2 −193750 −33.16 −0.4913 −8.9808 9.28 −2.4077 23.6586 −1.9978 1.5903 2
ErCl+2 −192230 −36.48 −0.6871 −9.4589 9.4679 −2.388 22.7995 −2.4561 1.6407 2
TmCl+2 −192250 −40.31 −0.6665 −9.4087 9.4482 −2.39 23.3346 −2.4541 1.6987 2
YbCl+2 −185320 −40.04 −0.8966 −9.9705 9.669 −2.3668 20.5306 −3.4155 1.6945 2
LuCl+2 −191690 −50.13 −0.9155 −10.017 9.6871 −2.3649 26.6715 −1.7655 1.8473 2
LaCl+
2 −227450 −2.02 2.9168 −0.6596 6.0093 −2.7517 6.6097 −4.7029 0.5747 1
CeCl+
2 −224990 9.64 1.9073 −3.1244 6.9781 −2.6498 −21.971 −14.077 0.3982 1
PrCl+
2 −225970 1.82 2.2914 −2.1866 6.6095 −2.6886 −2.3592 −7.6362 0.5166 1
NdCl+2 −224000 −1.36 2.144 −2.5465 6.7509 −2.6737 −8.8885 −10.058 0.5647 1
SmCl+ 2 −222430 −3.21 2.3393 −2.0696 6.5635 −2.6934 −8.8795 −10.144 0.5927 1
+
EuCl2 −200660 −10.02 2.4959 −1.6871 6.4131 −2.7093 9.3035 −4.1509 0.6958 1
+
GdCl2 −221990 −7.61 2.634 −1.3499 6.2806 −2.7232 10.7695 −3.5255 0.6593 1
TbCl+
2 −222840 −12.4 2.6916 −1.2094 6.2254 −2.729 11.1974 −3.607 0.7319 1
DyCl+2 −222050 −17.33 2.6363 −1.3444 6.2785 −2.7234 22.9371 0.2368 0.8065 1
HoCl+2 −224770 −15.26 2.4712 −1.7476 6.4369 −2.7068 18.5365 −1.1932 0.7751 1
+
ErCl2 −223270 −20.65 2.2626 −2.2569 6.6371 −2.6857 16.7094 −2.0874 0.8569 1
TmCl+ 2 −223290 −20.2 2.2586 −2.2666 6.6409 −2.6853 16.6521 −2.0854 0.85 1
YbCl+2 −216360 −19.22 1.9985 −2.9017 6.8906 −2.659 11.1191 −3.9614 0.8351 1
+
LuCl2 −222750 −30.31 1.9757 −2.9572 6.9124 −2.6567 21.9175 −0.7409 1.0031 1

disagreement is appreciable for Nd chlorides at 150 °C, a temperature meaningful formation constants from them. Thus, only the data report-
for which formation constants have been independently determined ed by Migdisov and Williams-Jones (2007) and Migdisov et al. (2009)
by a variety of experimental techniques (Gammons et al., 1996; are considered reliable. The first of these studies investigated the solu-
Migdisov and Williams-Jones, 2002; Migdisov et al., 2009) and are in bility of NdF3 in HF-bearing solutions, whereas in the later study, the
very good agreement. On another hand, it should be noted that the solubility of all REE fluorides, except Y fluoride, were measured simulta-
data summarized in Table 3 are from experiments at temperatures up neously in HF-bearing solutions. As both studies were performed using
to 300 °C, whereas the XAS measurements were performed at temper- the same experimental method, it is not possible to check for systematic
atures up to 500 °C. It is therefore possible that 400 and 500 °C, the tem- experimental errors that may have been common to them.
peratures for which the disagreement is greatest for La and Nd, (for At ambient temperature, the dominant fluoride species are REEF2+
300 °C the agreement between the XAS and solubility/UV–visible and REEF+ 2 , and the stability of the LREE fluoride complexes is lower
methods is good for both La and Nd) are above the predictive capability than that of HREE complexes (Schijf and Byrne, 1999; Luo and Byrne,
of the model employed for the derivation of the data listed in Table 3. 2000, 2007; Luo and Millero, 2004). However, at higher temperature
We therefore believe that, whereas these data can be used reliably for both the speciation and inter-element stability relations are different.
modeling chloride speciation of HREE up to 500 °C, for the LREE the Firstly, Migdisov and Williams-Jones (2007) and Migdisov et al.
limit is ~ 350 °C. Calculations for temperatures above 350 °C should pro- (2009) did not find evidence for the contribution of REEF+ 2 to the
vide a conservative estimate of the stability of REE chloride complexes. mass balance of dissolved REE at temperatures above 100 °C, and con-
cluded that REEF2+ was the only REE-fluoride species predominating
2.2.3. Thermodynamic stability of aqueous REE fluoride complexes in solution Secondly, Migdisov et al. (2009) showed that, whereas
The REE form much stronger complexes with fluoride than with there is a continuous increase in the stability of REEF2+ with increasing
chloride ions ( Wood, 1990a, 1990b; Haas et al., 1995; Schijf and atomic number at low temperature, the opposite is the case at high tem-
Byrne, 1999; Luo and Byrne, 2000, 2007; Luo and Millero, 2004). How- perature, and the rate of this change increases with increasing temper-
ever, whereas the stability of fluoride complexes has been investigated ature. Fig. 3 compares values of the formation constants of REEF2 +
experimentally in numerous studies at ambient conditions using a vari- obtained from high temperature (Migdisov and Williams-Jones, 2007;
ety of methods (see references above), experimental data on the behav- Migdisov et al., 2009), at 5–40 °C (Luo and Millero, 2004), and the pre-
ior of these species at high temperature are much more limited. The dictions of Haas et al. (1995). From this figure, it is evident that the ex-
only experimental studies that have been published for hydrothermal trapolations of Haas et al. (1995) were based on the erroneous
conditions are those reporting results of solubility experiments at tem- assumption that at high-temperature the stability of these complexes
peratures up to 250 °C and saturated vapor pressure (Migdisov and should reflect the inter-element dependence observed at ambient tem-
Williams-Jones, 2007; Migdisov et al., 2009), and that of Bilal and perature. As a result, the study of Haas et al. (1995) significantly overes-
Langer (1987), who investigated the speciation of the REE with fluoride timates the stability of REEF2+, particularly at high atomic number, for
(up to 0.175 mM) in a 1 M solution of NaCl to 200 °C at 1000 bars. Un- which the difference reaches over two orders of magnitude. Table 5
fortunately, the conditions at which the measurements of Bilal and lists the corrected HKF parameters recommended by Migdisov et al.
Langer (1987) were made (high ionic strength and high pressure), to- (2009) for LnF2+.
gether with the very incomplete description of the methodology and The only study to have investigated the behavior of yttrium experi-
the failure to report numerical data (the authors only presented their mentally in F-bearing aqueous fluids is that of Loges et al. (2013). As
measurements graphically), make it extremely difficult to derive the theoretical studies of Wood (1990a) and Haas et al. (1995) did not
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 19

100% 100 °C temperatures up to 250 °C1 (Migdisov and Williams-Jones, 2008). In


the absence of other experimental data, this dataset can be compared
Nd3+ only with the theoretical predictions of Wood (1990a) (Nd, Er only)
80%
and Haas et al. (1995) (REESO+ 4 only).
Fig. 4 compares the values of the formation constants for the mono-
60%
and bi-sulfate complexes of the REE derived by Migdisov and Williams-
Jones (2008) with those predicted by Wood (1990a) and those calculat-
40% ed from Haas et al. (1995). The experimental data suggest that REESO+ 4
is slightly less stable than predicted theoretically, particularly at the
NdCl2+
20% higher end of the temperature range investigated. For temperatures
NdCl2+
below 200 °C, experimental values for β1 are very similar to those of
0% Wood (1990a) and indistinguishable, within the relatively large exper-
-4 -3 -2 -1 0 1 imental error, from the values calculated from Haas et al. (1995). For
higher temperature, however, the experimental data suggest that the
predictions of Haas et al. (1995) and Wood (1990a) overestimate the
100%
200 °C stability of mono-sulfate complexes, with differences reaching between
a half and one order of magnitude at 250 °C. Experimental data derived
80% for the stability of bi-sulfate complexes are very similar to those predict-
Nd3+ NdCl2+ ed for these species by Wood (1990a) for 150 °C. However, the experi-
60% mental data demonstrate that the bi-sulfate complexes are much more
stable at higher temperature than predicted by Wood (1990a), reaching
40% two orders of magnitude higher at 250 °C.
In contrast to the formation constants for REECl+ +
2 and REEF2 , those
for the mono- and bi-sulfate complexes of Nd, Sm, and Er determined
20% NdCl2+ experimentally by Migdisov and Williams-Jones (2008) are indistin-
guishable within the experimental uncertainty at all temperatures in-
0% vestigated. This effect was predicted by both Haas et al. (1995) and
-4 -3 -2 -1 0 1 Wood (1990a), but is in apparent disagreement with the finding for am-
bient temperature that the stability of REE mono-sulfate complexes de-
100% creases with increasing atomic number (Schijf and Byrne, 2004) (Fig. 5).
300 °C
It should be noted, however, that the differences determined by Schijf
80% and Byrne (2004) for the values of the formation constants are extreme-
NdCl2+ NdCl2+ ly small; their values of β1 are 3.60, 3.63, and 3.51, for Nd, Sm, and Er
60% mono-sulfate complexes, respectively. Considering that the uncer-
tainties associated with the derivation of the REE mono-sulfate forma-
tion constants range from 0.11 to 0.42, it is very likely that such small
40%
differences in the formation constants would not be detectable in exper-
Nd3+ iments at elevated temperature. It can be concluded, however, that in-
20% creasing temperature likely does not result in a significant increase in
the steepness of the inter-element stability dependence, as is the case
0% for chloride complexes, and, to a first approximation, that the variation
-4 -3 -2 -1 0 1 of the formation constants of mono- and bisulfate complexes are insig-
log a (Cl-) nificant for the lanthanoids at any temperature.
In view of the very limited experimental data available on the hydro-
Fig. 2. The distribution of Nd3+ species as a function of chloride activity among the thermal stability of REE sulfate complexes, it is premature to fit these
chloride complexes at 100, 200, and 300 °C based on thermodynamic data from Tables 1 data to the HKF model. However, as there is a need of thermodynamic
and 2. data to permit modeling of the hydrothermal transport of the REE as
sulphate complexes, we have made a preliminary fit (Table 6) of the ex-
perimentally derived formation constants of Migdisov and Williams-
consider Y fluoride species, this is also the only source of data on Y fluo- Jones (2008) using the simple Bryzgalin-Ryzhenko model, which re-
ride complexation. Loges et al. (2013) found that yttrium undergoes a quires only two empirical parameters plus the room-temperature for-
higher degree of fluorination than the lanthanoids, which leads to a pre- mation constants (Ryzhenko et al., 1985).
dominance of YF+ 2 over YF
2+
over the range of temperatures investigat-
ed (100 ° to 250 °C) for total concentrations of fluoride from 1 to 2.2.5. Thermodynamic stability of aqueous REE hydroxyl complexes
330 mmol/l (similar to those employed by Migdisov et al. (2009)). Hydroxyl complexes of the REE are among the least studied aqueous
Table 5 lists the HKF parameters derived by Loges et al. (2013) for YF+ 2 . complexes at hydrothermal conditions. Although a number of publica-
tions have reported experimental data on the solubility of solid REE hy-
2.2.4. Thermodynamic stability of aqueous REE sulfate complexes droxides and their solubility products (Deberdt et al., 1998; Diakonov
The speciation of REE in sulfate-bearing solutions has been studied et al., 1998; Wood et al., 2002; Louvel et al., 2015), the only study
extensively at ambient temperature using a variety of experimental reporting an experimental determination of the stability of aqueous
techniques (Wood, 1990b). Similarly to the fluoride complexes, the pre- REE hydroxyl complexes at hydrothermal conditions is that of Wood
dominant REE3+ sulfate species described in the literature are REESO+ 4 et al. (2002). Their experiments were performed at temperatures up
and REE(SO4)− 2 (see the review of Wood (1990b) and Schijf and Byrne to 290 °C and were devoted to Nd. However, Wood et al. (2002) were
(2004)). At hydrothermal conditions, the experimental data available
on the speciation of the REE in sulfate-bearing solutions is limited to a 1
The data reported earlier in Migdisov et al. (2006) were derived using an unreliable
single set of spectroscopic data reported for Nd, Sm, and Er for activity model and have been re-evaluated in Migdisov and Williams-Jones (2008).
20 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Table 4
Comparison of the number of chloride ions associated with the REE from EXAFS measurements and modeled using the data listed in Table 3.

T °C LaCl3 0.02 m; HCl 0.006 m NdCl3 0.05 m GdCl3 0.1 m GdCl3 0.006 m YbCl3 0.006 m; HCl 0.017

(Mayanovic et al., Model (Mayanovic et al., Model (Mayanovic et al., Model (Mayanovic et al., Model (Mayanovic et al., Model
2009a) 2009a) 2007) 2007) 2002b)

150 1.5 ± 0.6 0.464 1.2 ± 0.2 0.438 – – –


300 2.2 ± 0.7 1.332 1.4 ± 0.4 1.088 0.9 ± 0.7 0.988 0.4 ± 0.2 0.745 0.5 ± 0.3 0.626
400 2.6 ± 0.8 1.658 1.7 ± 0.5 1.299 1.3 ± 0.6 1.338 1.3 ± 0.3 1.132 1.6 ± 0.3 1.152
500 3.1 ± 0.7 1.837 2.0 ± 0.2 1.447 1.8 ± 0.7 1.593 1.7 ± 0.3 1.437 1.8 ± 0.2 1.486

(Nd3 +), a more modest contribution of NdOH2 + to the mass of dis-


9 log
F 2+
1 = log aREEdF
3+
- log aREE - log aF
-
250 °C solved Nd species, a narrower predominance field of Nd(OH)+ 2 , and

8 more alkaline conditions for the predominance of Nd(OH)3°.


200 °C
Unfortunately, except for Nd(OH)3°, the available data are not suffi-
7 150 °C cient to derive reliable parameters for an equation of state describing
the thermodynamic properties of hydroxyl complexes of the REE at ele-
250 °C
1

6
F

vated temperature. In Table 8, we provide the results of a preliminary fit


200 °C
log

5 of the data reported by Wood et al. (2002) to the Bryzgalin-Ryzhenko


150 °C equation of state. Note that, whereas the fit for Nd(OH)3° was based
40 °C
4 on hydrolysis constants at four temperatures, the fits for NdOH2+ and
Nd(OH)+ 2 are based on only three and two experimental data points (in-
3 5 °C
cluding 25 °C data from Haas et al. (1995)), respectively. The data for
2 these latter two species have a significantly lower level of reliability
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu than for Nd(OH)3° and an unknown level of uncertainty. We therefore
Migdisov et al. (2009) Haas et al., (1995) Luo and Millero, (2004) urge readers to use these data with caution and to not rely on them
for conditions significantly different from those studied by Wood et al.
Fig. 3. Comparison of the values of the first formation constants of REE fluoride complexes (2002). We also emphasize that understanding the behavior of REE in
obtained experimentally by Migdisov et al. (2009) at 150, 200, and 250 °C and Luo and
hydrothermal systems, for which hydroxyl complexation is believed
Millero (2004) at 5–40 °C and predicted theoretically from the data of Haas et al. (1995)
(triangles). to contribute significantly to the mass of dissolved REE, requires the ac-
cumulation of considerably more experimental data.

only able to determine the stability of a single complex, Nd(OH)3°, over 2.2.6. Thermodynamic stability of aqueous REE carbonate and phosphate
the full range of temperature investigated (30–290 °C); the stability of complexes
Nd(OH)2 + was determined at two temperatures (250 and 290 °C), Carbonate ligands are known to form among the strongest REE com-
and that of Nd(OH)+ 2 only at 290 °C. plexes at ambient temperature (Wood, 1990b; Luo and Byrne, 2004),
Table 7 lists the hydrolysis constants determined by Wood et al. but to our knowledge these species have not been investigated experi-
(2002), and compares these values to those calculated on the basis of mentally at hydrothermal conditions. The data currently available on
the predictions of Haas et al. (1995). The data obtained experimentally the high temperature stability of these complexes are therefore limited
by Wood et al. (2002) suggest that Haas et al. (1995) significantly to the theoretical predictions of Haas et al. (1995) and Wood (1990a).
overestimated the stability of hydroxyl complexes of Nd at elevated These two studies predict significantly different behavior for REE car-
temperature; Fig. 6 (Figure 10 from Wood et al., 2002) compares the bonate complexes at elevated temperature. In Fig. 7, we compare the
distribution of Nd hydroxyl species as a function of pH at 290 °C deter- values of the formation constants for the mono-carbonate complex of
mined from the experimental data with that predicted by Haas et al. Eu(III) as a function of temperature predicted by Wood (1990a) and cal-
(1995). In contrast to the theoretical predictions, the experimental culated based on the HKF parameters recommended by Haas et al.
data suggest a large field of predominance for the simple hydrated ion (1995). These two extrapolations return very different dependencies

Table 5
The HKF equation of state parameters for aqueous REE fluoride species (Loges et al., 2013; Migdisov et al., 2009).

Species ΔG0298 S0298 a1 ∗ 10 a2 ∗ 10−2 a3 a4 ∗ 10−4 c1 c2 ∗ 10−4 ω ∗ 10−5 Z


(cal) (cal K−1)

LaF+2 −236290 −28.9 −2.6763 −14.316 11.377 −2.1872 42.7211 4.8321 1.5259 2
CeF+2 −234190 −24.92 −3.5448 −16.436 12.2104 −2.0995 34.4101 2.1345 1.4656 2
PrF+2 −235200 −25.4 −3.2382 −15.688 11.9162 −2.1305 32.1968 1.3417 1.473 2
NdF+2 −233150 −26.48 −3.3812 −16.037 12.0534 −2.116 34.2449 2.0023 1.4892 2
SmF+2 −232090 −27.08 −3.2108 −15.621 11.8899 −2.1332 16.6719 −4.1344 1.4983 2
EuF+2 −210450 −28.85 −3.0953 −15.339 11.7791 −2.1449 13.5854 −5.2922 1.5251 2
GdF+2 −231710 −25.91 −2.9753 −15.046 11.6639 −2.157 12.7413 −5.4446 1.4806 2
TbF+2 −232790 −30.21 −2.9233 −14.919 11.614 −2.1622 7.7005 −7.4034 1.5458 2
DyF+2 −232030 −29.52 −3.0025 −15.113 11.69 −2.1543 4.2664 −8.5636 1.5353 2
HoF+2 −234580 −28.53 −3.1464 −15.464 11.8281 −2.1397 6.6315 −7.6938 1.5203 2
ErF+2 −233060 −32.54 −3.3383 −15.932 12.0123 −2.1204 5.1506 −8.4013 1.581 2
TmF+2 −233080 −31.84 −3.3423 −15.942 12.0161 −2.12 2.8337 −9.1731 1.5705 2
YbF+2 −226300 −30.41 −3.5804 −16.523 12.2446 −2.0959 −1.0405 −10.451 1.5488 2
LuF+2 −232530 −36.16 −3.6241 −16.63 12.2865 −2.0915 1.8202 −9.7329 1.6359 2
YF2+ −237220 −33.37 −1.9638 −12.5762 10.6931 −2.2591 53.0391 8.2035 1.5936 2
YF+
2 −308470 −14.33 −1.8049 −12.1883 10.5407 −2.2751 76.4665 18.9864 0.7611 1
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 21

9
100 °C Migdisov and 100 °C
8 Williams-Jones (2008) 13
Haas et al. (1995)

o
7

o
1

2
Wood (1990) 11

log
6

log
9
5
4 7

3 5
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

9 200 °C 13 200 °C
8
o

11

o
1

2
log

log
9
5
7
4
3 5
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

9
13
8
o

7 11
1

o
2
6
log

log
5
250 °C 7 250 °C
4
3 5
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Fig. 4. The distribution of values for the first and the second formation constants of REE sulfate complexes across the REE group of elements for 100, 200, and 250 °C.

for the stability of EuCO+3 on temperature, with differences in the pre- characterised by a strong negative chondrite-normalised Ce anomaly
dicted stability reaching six orders of magnitude at 250 °C. Given such owing to the fact that the cerium occurs dominantly as Ce(IV) and pre-
enormous differences, selection of a preferred dataset is pointless. Any cipitates as the highly insoluble mineral, cerianite-(Ce) [CeO2] (Alibo
quantitative evaluation of the behavior of the REE in carbonate- and Nozaki, 1999). The same is commonly also the case for
bearing solutions will therefore require the accumulation of new exper-
imental data at hydrothermal conditions. Table 6
Similarly, the stability of phosphate complexes of the REE has not The Bryzgalin-Ryzhenko equation of state parameters for aqueous REE sulfate species,
been investigated at elevated temperature. The only source of informa- based on a preliminary fit using the data of Migdisov and Williams-Jones (2008) and as-
tion on their behavior at hydrothermal conditions is therefore Haas et al. suming that the variation of the formation constants of mono- and bisulfate complexes
is insignificant along the lanthanoid group at any temperature.
(1995). It should be noted, however, that in view of the extremely low
solubility of solid REE phosphates (monazite and xenotime; see Species Basic species pK (298) A (zz/a) B (zz/a)
below), phosphate likely plays the role of a depositional, rather than a 1:1
transporting ligand in ore forming processes, and phosphate complexes LaSO+
4 La3+ SO2−
4 3.61 1.563 −193.7
are unlikely to contribute significantly to the mobility of REE in these CeSO+
4 Ce3+ SO2−
4 3.61 −0.266 744.8
+
systems. PrSO4 Pr3+ SO2−
4 3.62 −0.528 867.4
NdSO+4 Nd3+ SO2−
4 3.64 1.024 131.14
SmSO+ 4 Sm3+ SO2−
4 3.63 1.155 70.49
2.3. Stability of Ce(IV) and Eu(II) species in hydrothermal solutions +
EuSO4 Eu3+ SO2−
4 3.64 2.981 −757.7
+
GdSO4 Gd3+ SO2−
4 3.61 2.462 −508.3
Although Ce, like the other REE, is generally assumed to occur as TbSO+
4 Tb3+ SO2−
4 3.59 4.527 −1473
DySO+ Dy3+ SO2− 3.57 4.2 1300
Ce(III) in hydrothermal fluids, an assumption that is supported by the 4 4
HoSO+4 Ho3+ SO2−
4 3.54 3.157 −828.5
lack of chondrite-normalised anomalies in rocks in which the REE ErSO+
4 Er3+ SO2−
4 3.51 1.583 −105.83
have been hydrothermally mobilised (Bau, 1991), Ce(IV) can be impor- TmSO+ 4 Tm3+ SO2−
4 3.48 2.052 −319.6
tant in low temperature environments. For example, seawater is YbSO+
4 Yb3+ SO2−
4 3.46 2.679 −582.4
+
LuSO4 Lu3+ SO2−
4 3.44 −1.684 1474

3.8 1:2
La(SO4)−
2 La3+ SO2−
4 5.29 3.041 0
3.7 Ce(SO4)−
2 Ce3+ SO2−
4 5.217 3.087 0

Pr(SO4)2 Pr3+ SO2− 4.92 3.09 0
o

4
4 1

3.6 Nd(SO4)−2 Nd3+ SO2−


4 5.15 4.558 −800.4
log SO

Sm(SO4)− 2 Sm3+ SO2−


4 5.44 4.914 −1018
3.5 −
Eu(SO4)2 Eu3+ SO2−
4 5.42 3.018 0
Gd(SO4)−2 Gd3+ SO2−
4 5.2 3.039 0
3.4 Tb(SO4)− Tb3+ SO2− 5.13 3.019 0
2 4

Dy(SO4)2 Dy3+ SO2−
4 5.03 3.111 0
3.3
Ho(SO4)−2 Ho3+ SO2−
4 4.9 3.061 0
Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Er(SO4)−
2 Er3+ SO2−
4 5.2 4.198 −630.6
Tm(SO4)− 2 Tm3+ SO2−
4 5.15 3.067 0
Yb(SO4)−
2 Yb3+ SO2−
4 5.18 3.042 0
Fig. 5. Values for the first formation constant of REE sulfate complexes determined by
Lu(SO4)−
2 Lu3+ SO2−
4 5.39 3.078 0
Schijf and Byrne (2004) at 25 °C
22 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Table 7 Table 8
Comparison of hydrolysis constants predicted by Haas et al. (1995) and measured by The Bryzgalin-Ryzhenko equation of state parameters for aqueous REE hydroxyl species. A
Wood et al. (2002). preliminary fit based on Wood et al. (2002) and Haas et al. (1995).

T °C log K (Wood et al., 2002) log K (Haas et al., 1995) Species Basic species pK (298) A (zz/a) B (zz/a)

Nd3+ + H2O ↔ Nd(OH)2+ + H+ 1:1


250 −3.8 ± 0.3 −2.02 NdOH2+ Nd3+ OH− 5.906 1.266 0
290 −3.3 ± 0.3 −1.42
1:2
Nd3+ + 2H2O ↔ Nd(OH)2+ + 2H+ Nd(OH)2+ Nd3+ OH− 10.994 2.183 0
290 −8.4 ± 0.4 −5.41
1:3
Nd3+ + 3H2O ↔ Nd(OH)3° + 3H+ Nd(OH)3° Nd3+ OH− 15.34 3.034 0
30 −24.0 ± 0.4 −25.71
50 −24.3 ± 0.6 −23.33
100 −22.1 ± 0.4 −18.82 recently, has been limited to the theoretical study of Sverjensky
290 −13.9 ± 0.4 −10.51
(1984), which predicted that Eu2+ would be favored over Eu3+ with in-
creasing temperature, and the study of Haas et al. (1995) who reported
formation constants for Eu(II) chloride complexes based on the data for
groundwaters. In both cases, Ce(III) oxidation is mainly mediated by chloride complexes of divalent transitions metals. These data however,
surface catalysis, involving Mn(IV) minerals such are birnessite, or by have been largely superseded by a variety of recent experimental data
microbes (Moffett, 1990). Cerium(IV) is also important in silicate (e.g., Testemale et al., 2009; Mei et al., 2013), raising questions about re-
melts, and Ce(IV) can concentrate in the lattice of minerals such as zir- liability of earlier predictions of the stability of Eu2+ complexes. The XAS
con (Burnhama and Berry, 2014). However, there are very few quantita- study of Liu et al. (2016) provided the first experimental confirmation of
tive data on the stability of Ce(IV) minerals and Ce(IV) aqueous the increase in the stability of Eu2+ over Eu3+ with increasing temper-
complexes, and these data are of poor quality even for ambient temper- ature predicted by Sverjensky (1984), and demonstrated that Eu(II) is
ature, owing to the low solubility of Ce(IV) minerals, complex mineral- the prevalent oxidation state for Eu in most crustal fluids at tempera-
ogical relationships (poorly characterised hydroxides), and the tures ≥200 °C (Fig. 8). This study, however, did not confirm the stability
difficulty in accounting for minor reduction in aqueous systems ( of high-order Eu(II) chloride complexes (e.g., EuCl2− 4 ) in hydrothermal
e.g., Spahiu and Bruno, 1995; Um et al., 2011; Mioduski et al., 2015a). fluids. Instead, Liu et al. (2016) proposed that the Eu(II) aquo complex
At room temperature, Eu(II) is stable in aqueous solutions only at pH and EuCl+ are the main Eu(II) species in solution, based on the chemical
values above neutral and highly reducing conditions (fH2(g) close to 1 similarity of Eu(II) and Sr(II).
bar). For higher temperature, our understanding of its behavior, until
2.4. Co-ordination chemistry of aqueous REE species
290 °C; Wood et al. (2002)
With the advent of third generation synchrotron light sources and
100%
Nd(OH)3o peta-scale computing capabilities, considerable information is becoming
Nd(OH)2+ available on the molecular-level processes of hydration and complex-
Nd3+ formation under hydrothermal conditions (Sherman, 2010), including
80% for the REE. This information guides the interpretation of other experi-
ments (e.g., solubility studies), and is key to developing reliable extrapo-
60% lations beyond the conditions investigated experimentally.
The hydration of the lanthanoid(III) ions has long been a contentious
40% Nd(OH)2+ issue, which has been resolved only recently by combining results of high
quality X-ray absorption experiments with theoretical quantum chemical
simulations (Persson et al., 2008; Lundberg et al., 2010; Persson, 2010;
20% D'Angelo and Spezia, 2012). There is now good consensus that the aquo
ions of the light and medium REE3+ (La to Dy) mainly display nine-fold
0% symmetry, with more or less regular tricapped trigonal-prismatic config-
290 °C; Haas et al. (1995) urations (Fig. 9c,d). Owing to their smaller ionic radii, the heavy REE3+
100% Nd(OH)3o aquo ions (Ho-Lu) experience increasing disorder of the capping waters.
Nd(OH)2+ This results in a dynamic equilibrium between eight- and nine-

80% 18
Nd(OH)2+ Eu3+ + CO32- = EuCO3+
16
60% 14
Wood (1990a)
12
1

40% 10
Nd3+
log

8
20% 6 Haas et al. (1995)
4
0% 2
1 2 3 4 5 6 7 0
0 50 100 150 200 250 300 350
pH
T°C
Fig. 6. Plot of %Nd vs. pH at 290 °C showing the distribution of Nd3+ species among the
various hydroxide complexes based on thermodynamic data from: a) Wood et al. Fig. 7. Comparison of the values of the first formation constants of Eu carbonate complexes
(2002); and b) Haas et al. (1995). predicted by Wood (1990a) and Haas et al. (1995).
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 23

Fig. 8. Europium speciation in crustal fluids as a function of temperature and total dissolved chloride concentration (molal); thermodynamic properties for Eu(II) species as listed in Liu
et al. (2016), Eu(III) from the compilation in this paper. The isolines show Eu2+/Eutot (molal ratio). (a) Fluid buffered by quartz-muscovite-feldspar (QMF)-magnetite-hematite-pyrite;
(b) fluid buffered by quartz-muscovite-feldspar (AMF)-pyrite-pyrrhotite; (c) bisulfide-rich fluid at pH near pKa (H2S), (pHT 6.1–6.9) for a fluid containing 1 molal bisulfide, and a
bisulfide:sulfate ratio of ~1:0.1.

coordinate hydrated ions with di- and tri-capped trigonal prismatic con- 25 °C (note the relatively large errors compared to Persson et al.
figurations, and increasing proportions of eight-coordinated hydrated (2008), which reflects the use of the hydrothermal diamond anvil cell
ions with increasing REE3+ atomic number. at higher temperature) to 6 to 7 at 500 °C (Fig. 10b).
As reviewed above, REE3+ fluoride and chloride complexes become Mayanovic et al. (2009a) suggested that the decrease in the degree of
more stable with increasing temperature, and above 200 °C, the light chloride coordination of REE3+ from light to heavy REE (Fig. 10) is due to
REE3 +-fluoride and REE3 +-chloride complexes are more stable than a geometrical effect, which they refer to as ‘steric hindrance’; the smaller
their heavy REE3+ counterparts, whereas at low temperature, the situ- HREE3+ ions have less affinity for the large chloride ion. The same effect
ation is reversed for REE3 +-fluoride complexes, with the HREE com- is responsible for the isothermal decrease in hydration number of the
plexes being stronger than those of the LREE. Fig. 10a shows that in REE3+ ions. However, as shown in the case of the analogous REE3+ fluo-
situ XAS studies of RCl3 solutions (R = La3+, Nd3+, Gd3+, Yb3+) as a ride complexes, which exhibit a more complicated behavior, we current-
function of temperature (25–500 °C, P to 2.6 kbar) confirm these trends ly lack the theoretical and experimental data to fully decipher the
for chloride complexes (Mayanovic et al., 2002a, 2002b, 2007, 2009a). controls on REE3 + complex stability in hydrothermal fluids. Several
These in situ spectroscopic studies also demonstrate a general decrease studies, however, point to key factors (aside from simple geometrical
in coordination number with increasing temperature, from ~ 8 to 9 at considerations) that remain poorly quantified for hydrothermal systems.

(i) An important factor is probably the strong polarization of the


molecules within the first shell, a result of the high charge to
ionic radius ratio of the REE3+. For example, the molecular dy-
namic study of Terrier et al. (2010) showed that in the case of
La3 +, polarization of the water molecules located in the first
shell gives rise to dipole moments about 0.5 debye larger than
those of bulk water molecules (3.0 debye in pure water at
room temperature). Petit et al. (2008) noticed the same effect
for water molecules in the LaCl2(H2O)+ 6 complex, and also
showed that the dipole moment of chloride ions increases from
0.52 debye in the solution to 1.39 debye within the complex.
Given the strong correlation between dielectric properties and
hydration energy, and the strong change in dielectric properties
of solutions with temperature and pressure (Sherman, 2010),
this effect is expected to play a key role in controlling REE3+ spe-
ciation in hydrothermal fluids.
(ii) Petit et al. (2008) used first principle molecular dynamic simula-
tions to investigate the speciation of La3+ in a ~13 molal chloride
solution (35 wt.% LiCl equivalent) at ambient temperature. They
found that the most stable configuration consists of two Cl− and
six water molecules arranged in a square antiprism geometry
(Fig. 9a,b). This raises the possibility that the coordination of ha-
lide ions may favor one coordination geometry (i.e., square
antiprism) versus another (tri-capped trigonal prismatic) for
the aquo ion at room temperature; this geometric effect may
contribute to further decreasing the overall coordination num-
ber. We also note that such coordination changes are associated
Fig. 9. Coordination and hydration of REE3+ aqua ions (c,d) and LaCl+
with large entropy changes because of the release of free water
2 (a,b) at room
temperature according to experimental and first-principle molecular dynamic data molecules, and hence are strongly temperature-dependent
(Persson, 2010; Petit et al., 2008). (e.g., Sherman, 2010; Mei et al., 2013).
24 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Fig. 10. Complexing of REE in hydrothermal fluids. (a) Average numbers of chlorine (bottom surface) and oxygen (i.e., water molecules; top surface) atoms surrounding R3+ (R = La, Nd,
Gd, Yb) based on EXAFS studies of RCl3 solutions, and plotted as a function of temperature and REE3+ ionic radius (8-fold coordination; Shannon (1976)). (b) Total coordination (Cl + O).
The errors are the sums of individual errors, and result from correlations among the fitted O and Cl numbers. They should be considered as upper limits. Plots modified after Mayanovic
et al. (2009a, 2009b, 2007, 2002a, 2002b). Error bars are shows as striated cylinders.

Ohoa et al. (2015) performed in-situ 139La NMR spectroscopy at commonly contain significant proportions of REE. It is evident from
room temperature up to pressures of 16 kbar (i.e., in metastable this table that, almost without exception, these minerals are most abun-
water). They noted, based on the HKF parameters for LaCl2 + and dant in carbonatites, highly differentiated granitic bodies (pegmatites),
LaCl+2 derived by Migdisov et al. (2009), that the stability of the La(III) or highly evolved nepheline syenite intrusions. Thermodynamic prop-
chloride complexes decreases with increasing pressure, consistent erties of the vast majority of REE minerals, however, are unknown,
with high pressure favoring more highly solvated ions because solvent and therefore these minerals cannot be accounted for in thermodynam-
water molecules can be packed into a smaller volume. However, a ic modelling of hydrothermal REE transport. In the sections that follow,
large fraction of the La is predicted to remain in the form of chloride we review the thermodynamic data for the few REE minerals
complexes, even at a pressure of 16 kbar. Ohoa et al. (2015) found (highlighted in bold characters in the Table) for which such data have
that the longitudinal relaxation rates for 139La converged at ~ 12 kbar, been determined.
which most likely reflects rapid ligand substitutions diminishing the
local electric field gradients around La(III) at high pressure. Such molec- 3.2. Thermochemistry and solubility of hydrothermally formed REE-bearing
ular understanding of the effects of temperature and pressure on com- minerals
plex formation is essential for improving the reliability of our
extrapolation methods. The numerous applications of the REE in green and other technolo-
The coordination chemistry of Eu(II) differs from that of Eu(III), not gies has led recently to a large number of experimental studies aimed
only because of the lower charge to radius ratio, but also because of the at determining the thermodynamic properties of REE-bearing solids.
presence of a stereochemically active electron lone pair. The room tem- These properties have been determined using two main approaches:
perature EXAFS study of Moreau et al. (2002) concluded that Eu2+ oc- calorimetry and solubility. Calorimetric studies provide accurate deter-
curs in aqueous solutions as an equilibrium between a predominant minations of the enthalpy of formation and heat capacity, but, unless ca-
[Eu(H2O)7]2+ and a minor [Eu(H2O)8]2+ species. A comparison of the lorimetric measurements are made at temperatures near absolute zero,
hydration of the Eu2+ and Eu3+ ions in aqueous solution, based on the these studies can only recommend third-law entropies, which are com-
quantum mechanical charge-field molecular dynamics (QMCF-MD) ap- monly subject to large uncertainties. Solubility measurements return
proach (Canaval and Rode, 2015), yielded coordination numbers of 8.1 accurate values for the Gibbs free energy of formation of the dissolving
and 8.9 for Eu(II) and Eu(III), respectively, apparently overestimating solid (calculated from the solubility product), and, if the temperature
the hydration number of the Eu2+ ion. These simulations predicted dependence of the solubility product is determined, can provide a
that the 8-fold coordination sphere of Eu2 + has a very dynamic and ∂ΔG f
good estimate of the entropy of the dissolving phase (ð ∂T p
Þ ¼ −S).
distorted square antiprismatic geometry. They also revealed that the
mean residence time of first shell H2O differs by more than an order of As the heat capacity corresponds to the second derivative of the Gibbs
∂S C
magnitude among Eu2+ and Eu3+, the hydration shell of Eu2+ being free energy with temperature (ð∂T ÞP ¼ Tp ), the values of this parameter
more flexible as a result of a weaker ion-water bond strength. are considerably less reliable. Thus, a combination of calorimetric and
solubility measurements provides the most effective means of accurate-
3. REE-bearing minerals involved in hydrothermal processes ly determining the thermodynamic properties of minerals needed to
model REE transport and deposition at elevated temperature.
3.1. Hydrothermally formed minerals of REE For a detailed overview of the available calorimetric data for REE-
bearing solids, readers are referred to the recent review of Navrotsky
All minerals in the Earth's crust contain REE. Concentrations of REE et al. (2015). Among the studies reported, most are devoted to synthetic
in many minerals are commonly at the level of tens to hundreds of phases important for industrial applications but of little relevance to hy-
ppm. The substitution of REE into minerals is controlled dominantly drothermal processes. In the case of hydrothermally relevant minerals,
by the crystal chemistry of the phase (see Section 5.3). Table 9 lists most of the data are limited to their enthalpy of formation. Standard en-
the relatively common hydrothermal REE minerals and minerals that tropies have been evaluated for a small number of minerals, and the
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 25

Table 9
Relatively common and less common hydrothermal REE minerals.

Formula Geological setting

Oxides
Aeschynite (Ce,Ca,Fe,Th)(Ti,Nb)2(O,OH)6 Nepheline syenite
Loparite (Ce,Na,Ca)2 (Ti,Nb)2O6 Differentiated nepheline syenite
Fergusonite (Ce,La,Y)NbO4 Nepheline syenite
Samarskite (Yb,Y,REE,U,Th,Ca,Fe2+) (Nb,Ta,Ti)O4 Nepheline syenite

Silicates
Allanite A2M3Si3O12[OH] A = Ca, Sr, M = Fe, Mg, Mn) (Ce. La, Y) Nepheline syenite
Cerite (Ce,La,Ca)9(Mg,Fe+3)(SiO4)6(SiO3OH)(OH)3 Nepheline syenite
Gadolinite (Ce,La,Nd,Y)2FeBe2Si2O10 Syenite pegmatite
Stillwellite (Ce,La,Ca)BSiO5 Alkalic pegmatites in syenite
Mosandrite Na(Na,Ca)2(Ca,Ce,Y)4(Ti,Nb,Zr)(Si2O7)2(O,F)2F3 Nepheline syenite
Steenstrupine Na14Ce6Mn2+ Mn3+ Fe2+2(Zr,Th)(Si6O18)2(PO4)7·3(H2O) Nepheline syenite

Carbonates
Bastnäsite (Ce,La,Y)CO3F Carbonatites, alteration zones in alkalic rocks
Parisite Ca(Ce,La)2(CO3)3F2 Within calcite in hydrothermal deposits
Synchysite Ca(Ce,La,Nd,Y)(CO3)2F REE-bearing pegmatites and alpine veins
Burbankite (Na,Ca)3(Sr,Ba,Ce)3(CO3)5 Carbonatites
Ancylite Sr(La,Ce)(CO3)2(OH)·(H2O) Carbonatites
Cordylite Ba(Ce,La)2(CO3)3F2 Carbonatites

Phosphates
Apatite Ca10(PO4)6(OH,F,Cl)2 Ubiquitous
Brockite (Ca,Th,Ce)PO4·H2O Granite and granite pegmatites
Monazite (Ce,La,Nd,Th) PO4,/(Sm, Gd, Ce, Th)PO4 Granitic pegmatites
Xenotime (Y, or Dy,E,Tb,Yb)PO4 Acidic, alkalic igneous rocks and their pegmatites
Florencite CeAl3(PO4)2(OH)6 Sandstones, granites
Britholite (Ce,Ca,Th,La,Nd)5(SiO4,PO4)3(OH,F) Syenite pegmatite, nepheline syenite

Halides
Fluocerite (Ce,La)F3 Hydrothermal veins in granite
Gagarinite NaCaY(F,Cl)6 Granites

Vanadinate
Wakefieldite (La,Ce,Nd,Y)VO4 Rare

temperature dependence of heat capacity for an even smaller number. La(OH)3(s) (Fig. 11A) is ~0.5 orders of magnitude greater than that cal-
Most solubility experiments have been performed at ambient tempera- culated using calorimetric data, this deviation is of the same order of
ture; only a limited number of studies have considered hydrothermal magnitude as the deviation of their data from the calorimetric value
conditions. In the following sections, we review the available data for for Gd(OH)3(s) at 90 °C (Fig. 11B). We therefore consider these devia-
REE-bearing minerals relevant for hydrothermal systems, or for their tions as experimental error (the authors do not report uncertainties
end-members that can be used for modeling of REE-bearing solid for the derived values), and conclude that the solubility data of
solutions. Deberdt et al. (1998) are in reasonably good agreement with the calori-
metric determinations. In contrast, the solubility data collected by
3.2.1. REE oxides and hydroxides Wood et al. (2002) suggest that Nd(OH)3(s) is more stable than predict-
Thermodynamic properties of REE oxides have been reported in a ed by calorimetric data. The difference in the solubility products is 0.33
large number of publications dating back to the 1960s. These mostly in- of a log unit at ambient temperature, which is close to the experimental
volved calorimetric determinations. For detailed analyses of the data uncertainty (±0.27), but is systematic and increases with temperature.
available for REE oxides and the techniques used for their derivation, At 300 °C, the difference between the value obtained by Wood et al.
readers are referred to the reviews of Morss and Konings (2005); (2002) and the value calculated using the data listed in Table 10 is near-
Konings et al. (2014), and Navrotsky et al. (2015). Table 10 lists the ly an order of magnitude (0.93), whereas the experimental uncertainty
properties of REE oxides recommended by Konings et al. (2014) and is significantly lower (±0.22). The reason for this disagreement is not
Navrotsky et al. (2015). This table only considers simple REE oxides; clear. It could reflect a systematic error in the experiments of Wood
for the properties of complex oxides, such as, for example, NdFeO3, et al. (2002), but it is equally possible that the expression of the temper-
readers are referred to the review of Navrotsky et al. (2015). ature dependence of the heat capacity for Nd(OH)3(s) proposed by
The stability of the REE hydroxides has been investigated by both ca- Diakonov et al. (1998) does not yield the correct values at elevated tem-
lorimetric and solubility techniques. The thermochemistry of REE hy- perature. Table 10 therefore lists two alternative descriptions of the
droxides at room temperature has been recently reviewed by properties of Nd(OH)3(s), the first of which is based on the data of
Navrotsky et al. (2015), and the temperature dependence of the heat ca- Navrotsky et al. (2015) and Diakonov et al. (1998), and the second de-
pacity is summarized by Diakonov et al. (1998). These data are listed in rived from the data of Wood et al. (2002).
Table 10. In addition, the solubility products of La and Gd hydroxides The data listed in Table 10 suggest that at hydrothermal conditions,
have been determined by Deberdt et al. (1998) for temperatures up to the REE oxides hydrolyze to form REE hydroxides. Fig. 12 illustrates the
150 °C, and that for Nd hydroxide was determined by Wood et al. Gibbs free energy values for the reactions REE2O3(s) + 3H2O =
(2002) for temperatures up to 350 °C. Fig. 11 compares the experimen- 2REE(OH)3(s), calculated for oxide/hydroxide pairs, for which data are
tal values of the solubility products for the hydroxides with those calcu- currently available. All the calculated values are negative up to 500 °C,
lated based on the calorimetric information reported in Table 10 and the suggesting that none of the oxides considered are stable in contact
data for simple hydrated ions (Table 2). Although the value of the solu- with water. This conclusion is in apparent contradiction with the recent
bility product obtained by Deberdt et al. (1998) at 150 °C for study of Louvel et al. (2015), which reports XAS spectroscopic data and
26 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Table 10
Thermodynamic properties of REE oxides and hydroxides.

ΔG0f (298) S0(298) Cp = a + b ⋅ TK + c ⋅ TK−2 + d⋅ TK−1

J/mol J/K mol a b ∗ 100 c d Reference

La2O3 −1703890 127.32 120.6805 1.3424 −1413668 [1]


Ce2O3 −1709280 148.1 113.7360 2.8434 −641205 [2]
CeO2 −1026338 62.29 74.4814 0.5837 −1297100 [1]
Pr2O3 −1720040 152.7 121.6594 2.5561 −989420 [1]
Nd2O3 −1719840 158.7 117.1079 2.8137 −1258450 [1]
Sm2O3 −1734664 150.6 129.7953 1.9031 −1862270 [1]
Eu2O3 monocl −1555060 143.5 133.3906 1.6644 −1424350 [1]
Eu2O3-cubic −1565043 136.4 136.2978 1.4988 −1499300 [1]
EuO −564358 83.6 46.5453 0.7360 [1]
Gd2O3 monocl −1734203 157.1 114.6104 1.5234 −1249170 [1]
Gd2O3-cubic −1766570 150.6 114.8086 1.7291 −1283970 [1]
TbO2 −915111 86.9 73.2590 1.3202 −1042400 [1]
Tb2O3 −1777260 159.2 120.6682 2.2172 −1002610 [1]
Dy2O3 −1771740 149.8 121.2302 1.5276 −845800 [1]
Ho2O3 −1793280 156.38 121.9340 1.0116 −886280 [1]
Er2O3 −1810380 153.13 123.2921 0.8622 −1544330 [1]
Tm2O3 −1795080 139.7 128.4322 0.5232 −1178910 [1]
Yb2O3 −1726740 133.1 130.6438 0.3346 −1448200 [1]
Lu2O3 −1792670 126.79 122.4593 0.7290 −2034140 [1]
Y2O3 −1844070 98.96 122.9100 0.7430 −1931300 [2]
La(OH)3 −1284200 117.8 174.6003 1.1336 796396 −20746.01 [3], [4]
Pr(OH)3 −1284900 131.7 174.6003 0.9677 858653 −20698.87 [3], [4]
Nd(OH)3 −1281091 128.03 926.6786 −185.7060 −22453710 [5]
Nd(OH)3 −1283000 129.9 174.6003 1.0826 983534 −21241.87 [3], [4]
Eu(OH)3 −1180600 119.9 174.6003 2.7272 791828 −20557.44 [3], [4]
Gd(OH)3 −1276200 126.6 174.6003 1.4341 967638 −22925.02 [3], [4]
Tb(OH)3 −1281100 128.4 174.6003 1.2288 1137175 −22959.94 [3], [4]
Ho(OH)3 −1297400 130 174.6003 2.1297 1148144 −23562.31 [3], [4]
Y(OH)3 −1338500 99.2 174.6003 2.1297 1148144 −23562.31 [3], [4]

References: [1]: (Konings et al., 2014) [2]: (Morss and Konings, 2005) [3]: (Navrotsky et al., 2015) [4]: (Diakonov et al., 1998) [5]:(Wood et al., 2002).

solubility determinations for Yb2O3(s) at temperatures up to 400 °C. Silva and Queimado (1973) for HREE and MREE, but differ considerably
These data suggest that the Yb oxide is stable at these conditions. It is, from those for LREE. Migdisov et al. (2009) extrapolated the solubility
however, unclear from the paper whether or not the authors deter- products determined experimentally for 150, 200, and 250 °C down to
mined the state of the solid after interaction with water, and we there- 25 °C using the entropy and heat capacity data reported by Konings
fore cannot yet conclude that HREE hydroxides are less stable than and Kovács (2003), and the data listed in Table 2. These extrapolations
hydroxides of MREE and LREE. (Fig. 14) yield solubility products at 25 °C that are in good agreement
with the values determined by Itoh et al. (1984) and Frausto da Silva
3.2.2. REE halogenides (chlorides and fluorides) and Queimado (1973) for all REE. As the data reported in the summary
An appreciable amount of data has been accumulated on both the of Konings and Kovács (2003) satisfactorily predict the temperature de-
solubility of REE chlorides and their calorimetric thermochemical prop- pendence of the solubility products of all REE fluorides, we conclude
erties. An extensive review of the solubility data for REE chlorides has that the standard entropies and the temperature dependent heat capac-
been reported in Mioduski et al. (2008, 2009a, 2009b). However, the ity functions for all the REE fluorides reported by them are reliable and
high solubility of these compounds makes it difficult to derive standard that the reason for the disagreement over the solubility products of the
thermodynamic properties, since a reliable thermodynamic description LREE fluorides is errors in the calorimetrically determined standard en-
of solutions having multi-molar concentrations of REECl3 requires a re- thalpies of formation of the La, Ce, Pr, and Nd fluorides. The thermody-
liable activity model for the corresponding electrolyte. In this contribu- namic properties of the REE fluorides listed in Table 11 therefore
tion, we therefore list only thermochemical data for REE chlorides compile the values for the entropy and heat capacity expressions re-
obtained using calorimetric methods (Table 11). A review of the avail- ported by Konings and Kovács (2003), and the values of the standard
able thermochemical data for solid, liquid, and gaseous REE trihalides, Gibbs free energies obtained via extrapolation as reported in Migdisov
including REE chlorides, has been published by Konings and Kovács et al. (2009).
(2003); these data were refined later by Rycerz and Gaune-Escard
(2008). 3.2.3. REE phosphates (monazite and xenotime)
In contrast to the REE chlorides, REE fluorides are highly insoluble, The properties of the synthetic monazite end members (from La to
and consequently they occur as minerals, e.g., fluocerite-(Ce) Gd) have been extensively investigated using a variety of calorimetric
[(Ce,La)F3]. For the same reason, information on their thermodynamic techniques; a review of these data has been reported by Popa and
properties is available from both calorimetric and solubility studies. Konings (2006), and more recently by Navrotsky et al. (2015). In addi-
The calorimetric data have been summarized by Konings and Kovács tion, a number of studies have reported experimentally determined sol-
(2003), and an extensive review of the solubility data has been pub- ubility products for monazite. Liu and Byrne (1997) reported solubility
lished recently by Mioduski et al. (2015a, 2015b). Yet, the values of products for monazite at 25 °C (and reviewed the data available in the
the solubility products of REE fluorides for 25 °C recommended by dif- literature for this temperature at that time), Poitrasson et al. (2004),
ferent experimental studies vary by orders of magnitude (Fig. 13). The and Cetiner et al. (2005), reported solubility products for the Nd-
most consistent datasets are those of Itoh et al. (1984) and Frausto da endmembers for temperatures up to 300 and 150 °C, respectively, and
Silva and Queimado (1973). The solubility products calculated using Pourtier et al. (2010) investigated the solubility of NdPO4 at tempera-
the calorimetric data of Konings and Kovács (2003) compare reasonably tures ranging from 300 to 800 °C. In addition, Cetiner et al. (2005) re-
well with the experimental values of Itoh et al. (1984) and Frausto da ported solubility products for the Y, La, and Sm endmembers
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 27

A 20
cr +
24 La(OH)3 + 3H = La3+ + 3H2O REE2O3+ 3H2O = 2REE(OH)3 Nd (Wood et al., 2002)
0
22
Navrotsky et al, 2015 + -20 Nd
Diakonov et al., 1998
20 -40 Tb
Ho

Gr, kJ/mol
Deberdt et al., 1998
log Ks

18 -60 Gd
-80 Eu
16 Y
-100
14 La
-120
12 -140
10 -160
0 20 40 60 80 100 120 140
-180
T°C 0 100 200 300 400 500
T °C
20
cr + B
19 Gd(OH)3 + 3H =Gd3+ + 3H2O Fig. 12. The values of the Gibbs free energy for the interaction of REE oxides with water;
18 calculated based on the data reported in Table 10.
Navrotsky et al, 2015 +
17 Diakonov et al., 1998
16 without corresponding data on the temperature dependency of heat ca-
log Ks

Deberdt et al., 1998


pacity, cannot be used for calculations at elevated temperature.
15
The above notwithstanding, even for standard conditions (25 °C,
14 1 bar), there is poor agreement on the values of the thermodynamic
13 properties of both monazite and xenotime. This is illustrated, in the
case of the solubility products (Fig. 15) by the values reported for
12
25 °C in the studies discussed above, and those calculated based on
11 the thermodynamic properties of REEPO4 summarized in the review
10 of Navrotsky et al. (2015). As can be seen from this figure, although
0 20 40 60 80 100 120 140 there is agreement within an order of magnitude of the values of the sol-
T°C ubility products for the La, Ce, and Pr phosphates, there is considerable
disagreement on these values for the other REE (the disagreement
21
reaches four orders of magnitude for Tb). In our view, the most reliable
cr +
19
Nd(OH)3 + 3H =Hd3+ + 3H2O C dataset for REE phosphate solubility products at 25 °С is that published
by Liu and Byrne (1997). This dataset demonstrates excellent agree-
17 Navrotsky et al, 2015 + ment with that of Cetiner et al. (2005) for La and Nd monazite
Diakonov et al., 1998 endmembers, and that of Poitrasson et al. (2004) for monazite-(Nd). It
15
log Ks

Wood et al., 2002 is also in very good agreement with the extrapolations from high tem-
13 perature of Gysi et al. (2015) for the Dy, Er and Y xenotime
endmembers. We therefore recommended that the dataset of Liu and
11 Byrne (1997) be used to derive the standard properties of monazite
9 and xenotime.
We note that, in most cases, the calorimetric data presented in the
7 review of Navrotsky et al. (2015) fail to reproduce solubility products
of REE phosphates and, thus, likely misevaluate its standard Gibbs free
5
0 50 100 150 200 250 300 350 energies of formation. The source of this disagreement can be either in
erroneous values of the standard enthalpy of formation for REE phos-
T°C phate or in a low accuracy of the determination of its entropy (or both):

Fig. 11. Comparison of the values of the solubility products for REE hydroxides, obtained ΔG f ¼ ΔH f −TΔS f : ð4Þ
experimentally by A,B) Deberdt et al. (1998) and C) Wood et al. (2002), with those
calculated based on the calorimetric information reported in Table 10.
In order to identify the source of the error, we have calculated two
sets of hypothetical thermodynamic properties for NdPO4. In both
sets, ΔG0f and the temperature dependence of Cp were identical: the
determined at 25 °C. There are far fewer data for xenotime. With the ex- first was derived from the solubility data of Liu and Byrne (1997), and
ception of the study of Gysi et al. (2015), solubility products for the second was taken from Popa and Konings (2006). The difference be-
xenotime have only been measured at room temperature (Liu and tween the datasets was in the values of entropy: for the first dataset, we
Byrne (1997) and the review therein). Gysi et al. (2015) reported solu- derived a new value of SNdPO40 using Eq. (4) and used the value of ΔH0f
bility products for YPO4, ErPO4, DyPO4 and YbPO4 and a natural recommended in Navrotsky et al. (2015), whereas in the second dataset
xenotime-(Y) for temperatures up to 250 °C. Calorimetric determina- the value of entropy was not modified and was taken directly from
tions of the complete set of thermodynamic functions have been made Navrotsky et al. (2015), suggesting that the values of ΔH0f listed in this
for YPO4, TbPO4, DyPO4, ErPO4, YbPO4, LuPO4, by Gavrichev et al. review are unreliable. These two dataset were used to calculate the sol-
(2006, 2010, 2012, 2013a, 2013b) and Gysi et al. (2016). The data for ubility products for NdPO4 (using the data listed in Table 2) and the
Ho and Tm phosphates unfortunately are restricted to determinations values obtained were compared with the experimental data obtained
of the enthalpy of formation (Navrotsky et al., 2015) and therefore, using the solubility techniques (Cetiner et al., 2005; Poitrasson et al.,
28 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Table 11
Thermodynamic properties of REE chlorides and fluorides.

ΔG0f (298) S0(298) Cp = a + b⋅ TK + c ⋅ TK−2 + d ⋅ TK−0.5 + e ⋅ TK2

J/mol J/K mol a b ∗ 100 c d e ∗ 105 Reference

LaCl3 −995890 137.57 74.9288 5.1654 684520 [1]


CeCl3 −983610 151.42 90.9772 3.5812 −271530 [1]
PrCl3 −982720 153.30 85.6511 3.9524 134650 [1]
NdCl3 −965560 153.43 87.2834 3.8586 40210 [1]
SmCl3 −949550 150.12 95.3748 3.3444 −516135 [1]
EuCl3 −855390 144.06 100.9736 3.0092 −263620 [1]
GdCl3 −943280 151.42 88.7959 3.1444 −34750 [1]
TbCl3 −1010600 154.85 86.2920 3.8598 [1]
DyCl3 −923340 175.40 104.5279 −2.7019 4.531110 [1]
HoCl3 −928290 177.10 100.3820 0.5091 [1]
ErCl3 −925020 175.10 101.4247 −1.6327 3.625740 [1]
TmCl3 −926200 173.50 102.0423 −1.7956 3.725180 [1]
YbCl3 −892360 169.30 104.8985 −2.3840 4.060230 [1]
LuCl3 −917760 153.00 98.3259 −1.7950 4.101460 [1]
LaF3 −1656370 106.98 255.8744 −16.81500 −774231 −1972.378 8.481070 [1], [2]
CeF3 −1641370 115.23 103.2577 −1.29896 −720870 2.468810 [1], [2]
PrF3 −1641370 121.22 130.5994 −3.25026 −2655590 1.816890 [1], [2]
NdF3 −1634370 120.79 103.3867 0.16669 −1101170 1.039350 [1], [2]
SmF3 −1624370 116.50 169.0564 −7.68090 −4840760 [1], [2]
GdF3 −1621370 114.77 102.3403 0.60945 −1401620 [1], [2]
TbF3 −1618370 118.97 97.5769 1.98845 −1156100 [1], [2]
DyF3 −1616870 118.07 91.2338 2.82118 −926687 −0.315530 [1], [2]
HoF3 −1624370 120.34 131.7639 −6.50032 −2463830 4.425000 [1], [2]
ErF3 −1614370 116.86 121.3374 −3.03149 −2155640 2.273170 [1], [2]
TmF3 −1614370 114.98 115.6209 −1.71827 −1853060 1.291430 [1], [2]
YbF3 −1587370 111.84 103.7012 0.92366 −1516080 [1], [2]
LuF3 −1610370 94.83 89.0368 1.92857 −685980 [1], [2]

References: [1]: (Konings and Kovács, 2003) [2]: (Migdisov et al., 2009).

2004; Pourtier et al., 2010) (Fig. 16). As can be seen from Fig. 16a, the phosphates, thus, reports the values of ΔG0f derived from the solubility
data based on the modified entropy (black line; sH°f = −1967.900 kJ/ data of Liu and Byrne (1997), and the values of entropy and temperature
mol, ΔG°f = −1840.295 kJ/mol; S° = 95.9 J/mol K) return a tempera- dependence of heat capacity recommended in Popa and Konings (2006)
ture dependence, which differs greatly from that observed in the exper- and Navrotsky et al. (2015).
iments, whereas the other dataset (blue line) demonstrates fair Despite the fair agreement between the solubility products mea-
agreement with the data reported by Poitrasson et al. (2004) and with sured by Poitrasson et al. (2004) at T ≤ 300 °C and those calculated
one of the determinations of Cetiner et al. (2005). We therefore con- using calorimetric data, these two datasets suggest small differences
clude that the mismatch between the calorimetric and solubility data in the steepness of the temperature dependence of the NdPO4 solubility
is likely related to an incorrect evaluation of ΔH0f in the calorimetric product, which potentially may increase at elevated temperature. In
measurements. Table 12, which summarizes the data for REE order to select the most appropriate dataset, we compared these two
sets of data with the data collected for NdPO4 by Pourtier et al. (2010)
at 300–800 °C and 2000 bar. Pourtier et al. (2010) did not evaluate the
Itoh et al. (1984) Menon and Baryshnikov
solubility product. Instead, they determined values of the equilibrium
James, 1989 and Gol'shtein, constants for the reaction: NdPOmonazite
4 + 3H2O = Nd(OH)aq 3 + H3PO4.
Frausto Da Silva and (average values) 1972
Queimado (1973) 4 . In order to compare these data with the data of Poitrasson et al.
Amano et al., 2004
Konings and (2004) we re-calculated their solubility products to the constants of
Vasil'ev and
Kovacs (2003)
Kozlovskii, 1977
the above reaction using the formation constants for Nd(OH)aq 3 recom-

-12.0 mended by Wood et al. (2002) (Fig. 16b). It is clear from this figure
that the trend for the data of Pourtier et al. (2010) deviates strongly
from those based on solubility products and calorimetric data. We
-14.0
therefore did not consider this dataset further.
In contrast to monazite, for which high temperature solubility data
-16.0 are available only for its Nd-end member, there are considerably more
logKsp

data available for xenotime, as a result of the study of Gysi et al.


-18.0 (2015), which reports the solubility of four xenotime end members at
temperatures up to 250 °C. In selecting the dataset appropriate for
-20.0 these compounds, we used the same approach as for monazite. The
values of the standard Gibbs free energies were derived from the data
25 °C of Liu and Byrne (1997), the values for entropy were taken from
-22.0
Navrotsky et al. (2015), and the high-temperature heat capacity depen-
dencies were those determined by Gavrichev et al. (2006, 2010, 2012,
-24.0
2013a, 2013b,) and Gysi et al. (2016). This dataset was used to calculate
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
the solubility products for Y, Dy, Er, and Yb xenotime end-members at
Fig. 13. Published values of experimental and calculated solubility products for REE(III)
temperatures up to 300 °C, and the resulting values compared with
fluorides determined for 25 °C. The values for Konings and Kovács (2003) were the experimental determinations of Gysi et al. (2015). As can be seen
calculated based on the calorimetric data combined with the data reported in Table 2. from Fig. 17, the calculated values are in very good agreement with
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 29

-12 -12
-14 -14
La Nd
-16 -16
-18

logKsp
-18
-20
-20
-22
-24 -22

-26 -24
-28 -26
0 50 100 150 200 250 300 0 50 100 150 200 250 300

-12
-13
-14 Tb
Eu -15
-16
logKsp

-17
-18
-20 -19

-22 -21

-24 -23

-26 -25
0 50 100 150 200 250 300 0 50 100 150 200 250 300

-12 -12

-14 -14
Er Lu
-16 -16
logKsp

-18 -18

-20 -20

-22 -22

-24 -24
0 50 100 150 200 250 300 0 50 100 150 200 250 300
T °C T °C

1 2 3 4 5 6 7 8 9 10

Fig. 14. Examples of extrapolations of the values of solubility products for REE(III) fluorides from the experimental temperature (150, 200, and 250 °C; grey circles: Migdisov et al., 2009) to
25 °C. The extrapolations were performed using values of standard entropy and heat capacity (Cp) reported by Konings and Kovacs (2003). Also shown are the solubility products
experimentally determined or calculated for 25 °C. 1: Migdisov et al. (2009), 2: Itoh et al. (1984) 3: Frausto da Silva and Queimado (1973), 4: calculated values based on Konings and
Kovács (2003) 5: Menon and James (1989) 6: Amano et al. (2004) 7: Vasil'ev and Kozlovskii (1977) 8: Baryshnikov and Gol'shtein (1972)

the experimentally determined values. Table 12 therefore recommends 3.2.4. REE fluorocarbonates (bastnäsite-parisite)
this dataset for calculations of xenotime solubility at elevated To our knowledge, the only study which reports thermodynamic
temperature. data applicable to calculations of REE fluorocarbonate solubility at
high temperature is that of Gysi and Williams-Jones (2015). This
-20 study reports the results of differential scanning calorimetric experi-
REEPO4(s) = REE3++ PO43- Liu, Byrne (1997)
based on Navrotsky et al. (2015) ments on natural samples of bastnäsite-(Ce) and parisite-(Ce) conduct-
-22 Gisy et al. (2015)
ed at temperatures from 323 K to 1022 K. The stoichiometric formulae
Poitrasson et al. (2004)
determined for these minerals were Ce0.5La0.25Nd0.20Pr0.05CO3F for
-24 bastnäsite-(Ce) and CaCe0.95La0.6Nd0.35Pr0.10(CO3)3F2 for parisite-(Ce)
and represent compositions deemed to be typical of the naturally occur-
log Ks

-26 ring minerals. Data for a complete set of thermodynamic functions for
these minerals are listed in Table 13. These data permit evaluation of
-28 the stability relationships between these minerals and other carbonate
and fluoride-bearing phases (Fig. 18; Figure 11 from Gysi and Williams-
-30 Jones (2015)). These data also can also be used for approximate deter-
mination of the conditions, at which precipitation of bastnäsite-(Ce)
-32 or parisite-(Ce) occurs from natural solutions. It should be noted, how-
Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu ever, that, considering that these data correspond to only one particular
composition of bastnäsite and parisite solid solution, the relationships
Fig. 15. Published values of solubility products for REE(III) phosphates determined for observed on Fig. 19 may not accurately reflect the relationships for
25 °C these minerals in all natural systems. Changes in solid solution
30 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

-22 A Poitrasson et al. (2004)


members forming these solid solutions are available to test this as-
-24 Liu, Byrne (1997) sumption, Table 13 should be used with caution.
In addition to evaluating their thermodynamic properties, Gysi and
-26 Modified enthalpy
Modified entropy
Williams-Jones (2015) also reported the range of temperatures for
-28 which the two REE fluorocarbonate minerals are stable. Their experi-
log Ks

-30 ments indicate that the natural bastnäsite-(Ce) and parisite-(Ce)


decomposed at 1 bar above 528.15 K (255 °C) and 643.15 K (370 °C), re-
-32
spectively to form REE oxyfluorides and release CO2. These tempera-
-34 NdPO4= Nd3++ PO43- tures are considerably lower than the 1 bar temperatures (N 772 K/
-36 498 °C) reported by Haschke (1975) and Janka and Schleid (2009), for
the decomposition of bastnäsite-(La) and the 1 kbar temperature of
-38
0 50 100 150 200 250 300 350 1032 and 1072 K (758–798 °C) reported by Hsu (1992) for bastnäsite-
T °C (La) and bastnäsite-(Ce). A likely reason for the disagreement with
Haschke (1975) and Janka and Schleid (2009) is that, whereas Gysi
0 and Williams-Jones (2015) studied a natural bastnäsite-(Ce) solid solu-
Based on Table 11 B tion, Haschke and Janka and Schleid studied the bastnäsite-(La) end-
-5 Recalculated from Poitrasson et al. (2004) member. The very much higher temperature of Hsu (1992) could also
indicate the effect of pressure.
-10
log K

4. Hydrothermal mobility and depositional mechanisms


-15

-20 4.1. Hydrothermal mobility of the REE

-25 Although clearly not complete, the above datasets permit evaluation
NdPO4 + 3H2O = Nd(OH)3 + H3PO4 to a first approximation of the main factors controlling the transport and
-30 deposition of the REE in hydrothermal fluids. An essential factor in this
0 100 200 300 400 500 600 700
transport/deposition is the availability of suitable ligands in concentra-
T °C
tions sufficient for complexing the REE and depositing them in the
form of minerals. In evaluating REE mobilization processes, a starting
Fig. 16. A) Comparison of the values of the solubility products for Nd phosphate obtained
experimentally by Poitrasson et al. (2004); Cetiner et al. (2005), and Liu and Byrne (1997)
point is to review information on the concentrations of ligands in fluids
with those calculated based on different thermodynamic datasets (see the text) known to have been responsible for the transport and deposition REE
B) Comparison of the values of the dissolution equilibrium constant for Nd phosphate, (Table 1). Among studies reporting this information that of Banks
re-calculated from the data of Poitrasson et al. (2004), calculated using the data et al. (1994) is arguably the most detailed, providing information not
reported in Table 11, and experimentally determined by Pourtier et al. (2010)
only for the ligands, but also the concentrations of the REE. Much of
the following discussion is based on the fluid compositions reported in
composition lead inevitably to changes in the enthalpy and entropy of Banks et al. (1994) (see their Table 3) and on Migdisov and Williams-
mixing, and therefore to changes in the stability of these minerals. Jones (2014); for details not covered in the current review readers are
Nonetheless, given that the compositions of the minerals studied by referred to this publication.
Gysi and Williams-Jones (2015) are representative of those commonly The models evaluated below are not based on case studies, but have
occurring in nature, we speculate that the data reported in Table 13 been designed to highlight the main mechanisms that can be responsi-
will be applicable to most hydrothermal systems in which bastnäsite- ble for hydrothermal transport of REE, and therefore have been devel-
(Ce) and parisite-(Ce) occur. However, until data for the end- oped using highly simplified hypothetical scenarios. To achieve this,

Table 12
Thermodynamic properties of REE phosphates.

ΔGf0(298) S0(298) Cp = a + b⋅ TK + c ⋅ TK−2 + d ⋅ TK−0.5 + e ⋅ TK2

J/mol J/K mol a b ∗ 100 c d e ∗ 105

Monazites
LaPO4 −1848528 [1] 108.3 [2] 121.13 [4] 3.0116 [4] −2562500 [4]
CePO4 −1844480 [1] 120 [2] 125.21 [4] 2.7894 [4] −2408500 [4]
PrPO4 −1849976 [1] 123.2 [2] 124.50 [4] 3.0374 [4] −2449500 [4]
NdPO4 −1840296 [1] 122.9 [2] 132.96 [4] 2.2541 [4] −3100900 [4]
SmPO4 −1833449 [1] 122.5 [2] 133.13 [4] 2.3468 [4] −3068700 [4]
EuPO4 −1741096 [1] 117.2 [2] 137.56 [4] 1.7693 [4] −2785400 [4]
GdPO4 −1828503 [1] 124.6 [2] 133.24 [4] 1.2793 [4] −3097200 [4]

Xenotimes
TbPO4 −1831070 [1] 138.1 [2] 165.51 [8] 1.904 [8] −661298.2 [8] −1024.372 [8] 1.187 [8]
DyPO4 −1829100 [5] 185.5 [5] 185.5 [5] −3261000 [5] −751.9 [5]
DyPO4 −1826068 [1] 138.1 [2]
HoPO4 −1836794 [1] 142.3 [2]
ErPO4 −1831260 [1] 116.6 [2] 264.19 [6] −3.5430 [6] 491967.5 [6] −2752.696 [6] 0.942 [6]
ErPO4 −1832500 207.2 [5] −529000 [5] −1661 [5]
TmPO4 −1830518 [1] 138.1 [2]
YbPO4 −1801020 [1] 133.9 [2] 247.56 [7] −2.2910 [7] 256022.1 [7] −2463.215 [7] 0.687 [7]
LuPO4 −1826714 [1] 117.2 [2] 268.28 [9] −4.472 [9] 570634.2 [9] −2803.989 [9] 1.1187 [9]
YPO4 −1849118 [1] 108.8 [2] 134.06 [3] 1.8330 [3] −3490000 [3] −52.4841 [3] 0.0424 [3]

References: [1] – derived from (Liu and Byrne, 1997) [2] (Navrotsky et al., 2015) [3] (Gavrichev et al., 2010) [4] (Popa and Konings, 2006) [5] (Gysi et al., 2016) [6] (Gavrichev et al., 2012)
[7] (Gavrichev et al., 2013a) [8] (Gavrichev et al., 2013b) [9] (Gavrichev et al., 2006)
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 31

-20 -20
-22 -22
-24 -24
-26 -26

log Ks

log Ks
-28 -28
-30 -30
-32 -32
-34 -34
-36 YPO4 = Y3+ + PO43- -36 DyPO4 = Dy3+ + PO43-
-38 -38
-40 -40
0 100 200 300 400 0 100 200 300 400
TC T °C
-20 -20
Table 11
-22 -22
Gisy et al. (2015)
-24 -24
-26 -26
log Ks

log Ks
-28 -28
-30 -30
-32 -32
-34 -34
-36
ErPO4= Er3+ + PO43-
-36 YbPO4 = Yb3++ PO43-
-38 -38
-40 -40
0 100 200 300 400 0 50 100 150 200 250 300 350
T °C T °C

Fig. 17. Comparison of the values of the solubility products for xenotime obtained experimentally by Gysi et al. (2015), with those calculated based on the data reported in Table 11.

the original composition of the fluid reported in Banks et al. (1994) has primarily on the fact that fluoride species are among the strongest aque-
been modified as follows: ous complexes of the REE and the observation that fluorite is commonly
associated with REE mineralization, implying an elevated concentration
1) The solutions reported in Banks et al. (1994) are highly saline
of fluoride ions in the hydrothermal fluid ( Metz et al., 1985;
(Table 1). The concentration of chloride in these solutions reaches
Williams-Jones et al., 2000; Holtstam and Andersson, 2007; Ruberti
43 wt.%, which is significantly higher than that determined in
et al., 2008; Xu et al., 2010; Gysi and Williams-Jones, 2013). The models
other REE depositing hydrothermal systems. In addition, existing ac-
put forward for hydrothermal REE ore genesis proposed transport of the
tivity models are inadequate to describe such highly saline solutions
REE in the form of fluoride species and destabilization of the latter at a
(which are more akin to hydrated melts) at elevated temperature.
geochemical barrier due to interaction of the ore fluid with Ca-rich
The concentration of NaCl in the solutions used in our evaluations
rocks or mixing with Ca-rich fluids. In this model precipitation of highly
was therefore set at 10 wt.% (~3 times that of seawater).
insoluble fluorite strips fluorine from the fluid, which in turn causes pre-
2) The concentrations of fluoride reported by Banks et al. (1994) vary
cipitation of REE (Salvi and Williams-Jones, 1996; Smith and Hender-
from 358 to 5245 ppm. For our evaluations we used a conservative
son, 2000). These models largely ignored the potential contribution of
concentration of 500 ppm.
other ligands in transporting the REE, and did not account for the stabil-
3) The concentrations of sulfate reported by Banks et al. (1994) vary
ity and solubility of the corresponding solids, e.g., REE fluorides.
from 20,773 ppm to 24,210 ppm. For simplicity we have set the con-
Migdisov and Williams-Jones (2014) showed that although fluoride
centration of sulfate at 2 wt.%.
complexes of the REE can predominate over a range of conditions
4) Finally, in the examples provided in this contribution, we focus on
(Fig. 19), their predominance fields are significantly narrower than sug-
Nd-bearing systems, as these are the best studied with respect to
gested by the data of Haas et al. (1995). This is because the extrapola-
the stability of the aqueous species and minerals. The highest con-
tions to high temperature underestimated the stability of REE chloride
centration of Nd determined by Banks et al. (1994) is 190 ppm; in
complexes, and significantly overestimated the stability of REE fluoride
our calculations we used a concentration of 200 ppm. The effects ob-
complexes. Most importantly, the calculations reported in Migdisov and
served for Nd, however, have been confirmed for other REE; for de-
Williams-Jones (2014) demonstrated that the predominance fields of
tails the readers are referred to Migdisov and Williams-Jones (2014).
fluoride complexes correspond closely to the stability fields of solid
All calculations reported below are based on the dataset for REE min- REE fluorides, compounds that have extremely low and temperature-
erals and aqueous species presented in this contribution, and on the ac- retrograde solubility. In other words, if physico-chemical conditions
tivity model proposed for NaCl-based solutions (Helgeson et al., 1981; (pH in the case of the example shown in Fig. 19) shift the system into
Oelkers and Helgeson, 1991). the predominance field of REE fluoride species, it is inevitable that
Until recently, fluoride was considered to be the main transporting solid REE fluorides will precipitate, which effectively removes the REE
agent for REE in ore-forming hydrothermal processes. This was based from solution and buffers their concentrations at levels below 1 ppb.

Table 13
Thermodynamic properties of bastnäsite-(Ce) and parisite-(Ce) (Gysi and Williams-Jones, 2015).

ΔG0f (298) S0(298) Cp = a + b⋅ TK + c ⋅ TK−2 + d ⋅ TK−0.5

J/mol J/K mol a b c d

Bastnäsite Ce0.5La0.25Nd0.20Pr0.05CO3F −6838800 603.6 537.20 0 −8128000


Parisite CaCe0.95La0.6Nd0.35Pr0.10(CO3)3F2 −9143000 783.202 797.60 0 −8404000 −2096
32 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Fig. 18. Calculated mineral–fluid equilibria at 300 and 400 °C for the stability of bastnäsite-(Ce)-, parisite-(Ce)- and fluocerite-(Ce)-bearing mineral assemblages as a function of dissolved
ion activities (aF− and aCO2−
3 ) at 1 kbar (Gysi and Williams-Jones, 2015).

These calculations have demonstrated the practical impossibility of as a signature of fluoride transport of the REE. The formation of fluorite
transporting significant amounts of REE in the form of fluoride species, is subject to the same rules as those for the formation of solid REE
and shown that the models explaining REE ore formation by this pro- fluorides:
cess are unsupportable. A similar effect has been observed for alkaline
conditions with Nd(OH)3(s). We, therefore, interpret fluoride to play Å
Ca2 þ þ 2HF ¼ CaF2ðs Þ þ 2Hþ ; or ð6aÞ
the role of a depositional rather than transporting ligand in ore-
forming hydrothermal systems.
Other conclusions of the study of Migdisov and Williams-Jones
(2014) is that in fluorine-bearing solutions efficient hydrothermal
transport of REE is possible only under acidic conditions, and that the Ca2 þ þ 2F− ¼ CaF2ðsÞ ð6bÞ
main vehicles of this transport are REE chloride complexes. This is be-
cause at higher pH formation of REEF3(s) buffers REE concentration to As is the case for REE fluorides, formation of fluorite starts when the
exceptionally low levels. Moreover, in the case of appreciable concen- concentration of F- reaches a level corresponding to its solubility prod-
trations of fluoride, fluoride ion concentrations are too low at low pH uct, which occurs when HF° is significantly dissociated. In other
to play a role in REE transport because hydrofluoric acid is a weak acid words, the common occurrence of the REE and fluorite reflects the fact
(e.g., Hefter, 1984; Ryzhenko et al., 1991), and any fluorine is present that they share the same depositional control.
overwhelmingly as the ion pair, HF°. Thus, under acidic conditions pre- A potentially important limitation of the calculations reported above
cipitation of NdF3(s), for example, is most accurately described by the re- is that the solid controlling Nd concentration, NdF3(s) or fluocerite-
action: (Nd), is relatively uncommon in nature. Indeed, the dominant
fluorine-bearing REE mineral in REE deposits is bastnäsite (Salvi and
Å Williams-Jones, 1996; Holtstam and Andersson, 2007), although in
Nd3þ þ 3HF ¼ NdF3ðsÞ þ 3Hþ ð5Þ
some cases fluocerite is also present (Holtstam and Andersson, 2007;
Precipitation of NdF3(s) can, of course also be described by the reac- Gysi and Williams-Jones, 2013). Unfortunately, the lack of experimental
tion: data on the carbonate speciation of the REE precludes detailed analysis
of the transport and deposition of the REE in carbonate-bearing fluids.
Nd3þ þ 3F− ¼ NdF3ðsÞ : ð5aÞ Nevertheless, as shown below, the occurrence of bastnäsite may be
taken to indicate saturation concentrations of REE even lower than for
However, precipitation of NdF3(s) will begin only when the concen- fluocerite, and the concentrations of Nd-fluoride species reported in
tration of F- reaches a value determined by the solubility product our model should therefore be considered maximum values.
(which also depends on the concentration of Nd3 + in the solution). In the case of fluoride-free systems, the restriction of REE transport
This, in turn, requires an appreciable degree of dissociation of HF° and to acidic conditions does not apply. Although significantly weaker
thus higher pH. For the composition of our model solution, this concen- than other aqueous complexes of REE, chloride species can play a signif-
tration of F− is reached at pH values ranging from 2 to 3.5, depending on icant role in REE transport due to the high concentrations of chloride in
the temperature (Fig. 19). This effect also explains the common occur- natural solutions and the absence of solubility-limiting minerals (REE
rence of the REE with fluorite, which has been interpreted erroneously chlorides are highly soluble, see above). For a simple system involving
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 33

A A

B B

C C

Fig. 19. The speciation of Nd as a function of pH at 1000 bar and A) 200 °C, B) 300 °C and Fig. 20. The speciation of Nd as a function of pH at 1000 bar and A) 200 °C, B) 300 °C and
C) 400 °C. The initial composition of the solution before equilibration was 10 wt.% NaCl, C) 400 °C. The initial composition of the solution before equilibration was 10 wt.% NaCl and
500 ppm HF, and 200 ppm Nd. The stability field of NdF3(solid) is shown in pink, and 200 ppm Nd. The stability field of Nd(OH)3 (solid) is shown in grey.
that of field of Nd(OH)3 (solid) in grey. Distributions of subordinate Nd aqueous species
and Nd-free aqueous species are shown by dashed lines. The vertical red dashed line
corresponds to half the pK for the dissociation of water and neutral pH.
species to the mass of dissolved Nd are not very significant at low tem-
perature, but at higher temperature sulfate can significantly expand the
range of pH over which Nd can be stabilized in F-bearing solutions.
Whereas at 200 °C sulfate species of Nd are a minor (although signifi-
only Nd, NaCl, HCl, and water, the solubility-limiting phase is cant), at 300 °C, Nd(SO4)−s 2 predominates over NdF
2+
(and NdCl2 +)
Nd(OH)3(s), which starts forming at near-neutral conditions (for tem- over the full range of pH for which fluoride species predominate in
peratures of 200, 300, and 400 °C, neutrality corresponds to pHT values the sulfate-free system, and at 400 °C, Nd(SO4)−
2 not only predominates
of 5.64, 5.59, and 5.70, respectively, due to the changes in the dissocia- over NdF2+ (and NdCl2+) but is present in concentrations correspond-
tion constant of water). As can be seen from Fig. 20, a 10 wt.% NaCl so- ing to N1 ppm Nd at pH values between 3.5 and 5; at lower pH, NdCl2+
lution can stabilize 200 ppm Nd to reasonably high pH (pH ~ 5), i.e., to predominates. Complexes involving sulfate ions could therefore play a
conditions for which the effects associated with the hydrolysis of Nd be- significant role in Nd transport by weakly acidic hydrothermal fluids
come significant. Obviously, the speciation may differ significantly in even in the presence of fluoride ions. As is the case for chloride- and
more complex fluoride-free systems involving other REE-bearing min- fluoride-Nd complexes, the concentration of Nd(SO4)− 2 , however,
erals. However, these data demonstrate that transport of the REE as drops to insignificant values at near neutral pH due to saturation of
chloride complexes does not have internal limitations, like those of the fluid with solid NdF3.
fluoride-bearing systems, and can occur and be efficient at pH condi- In fluoride-free chloride-sulfate systems, the only solubility-limiting
tions typical of natural hydrothermal systems. phase is Nd(OH)3(s). Hence, appreciable concentrations of Nd can be
Sulfate is another important ligand shown to be present in apprecia- transported over a wide range of pH (Fig. 22). Sulfate complexes of Nd
ble concentrations in REE-bearing solutions Banks et al. (1994). The be- out-compete chloride species at most of the conditions for the simula-
havior of REE in sulfate-fluoride-bearing solutions is illustrated in Fig. 21 tions in Fig. 23: chloride complexes predominate only in highly acidic
(see Migdisov and Williams-Jones, 2014). Contributions of sulfate solutions, and the range of pH for which they predominate decreases
34 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

A A

B B

C
C

Fig. 21. The speciation of Nd as a function of pH at 1000 bar and A) 200 °C, B) 300 °C and
C) 400 °C. The initial composition of the solution before equilibration was 10 wt.% NaCl, Fig. 22. The speciation of Nd as a function of pH at 1000 bar and A) 200 °C, B) 300 °C and
2 wt.% Na2SO4, 500 ppm HF, and 200 ppm Nd. The stability field of NdF3(solid) is shown C) 400 °C. The initial composition of the solution before equilibration was 10 wt.% NaCl,
in pink, and that of Nd(OH)3 (solid) in grey. 2 wt.% Na2SO4, and 200 ppm Nd. The stability field of Nd(OH)3 (solid) is shown in grey.

form of chloride and sulfate complexes. Fluoride-bearing solutions are


with temperature. This is due to the fact that the stability of REE sulfate generally unable to transport significant concentrations of REE, except
complexes is 2.5 to 3.5 orders of magnitude higher than that of chloride under acidic conditions. However, even under acidic conditions, the
complexes. Moreover, this high stability expands the pH range over main vehicles of this transport are chloride and sulfate complexes. It is
which efficient transport of REE is possible relative to pure chloride- likely that at elevated temperature (400 °C) the “acidic” restrictions
only systems. Whereas chloride-only solutions can carry appreciable can be softened, if the solution contains appreciable concentrations of
concentrations of REE at pH values below 5 (Fig. 20), the presence of sulfate. The latter conclusion, however, should be treated with caution
2 wt.% sulfate expands this range to 6 at 200 and 300 °C, and as high considering that it is based on experimental data for the stability of sul-
as 7.5 at 400 °C (at this temperature, conditions are alkaline). As for fate complexes of the REE, for temperatures up to only 250 °C.
the chloride-only system, we note that the speciation in complex sys-
tems can differ significantly from that illustrated in Fig. 22. Neverthe- 4.2. Depositional mechanisms
less, the observed dependencies demonstrate that sulfate REE
complexes can provide an efficient vehicle for the transport of these el- An efficient mechanism for the deposition of the REE has already
ements in hydrothermal systems. been discussed, namely transport of the REE by acidic fluids containing
The main conclusion of the above modeling is that hydrothermal fluoride in the form of HF° followed by destabilization of the REE com-
transport of the REE in ore-forming systems occurs primarily in the plexes through neutralization of the solution acidity. This neutralization
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 35

6 interaction of the REE-F-bearing fluids with limestones or dolomites,


but also by mixing of these fluids with a CO2-rich fluid. It is also worth
4
pH mentioning that the modeled interaction results in extremely efficient
removal of REE from the solution (a decrease in REE concentration of
2
3 to 4 orders of magnitude), and therefore likely represents a particular-
logarithm of molality

CO2 total ly efficient depositional mechanism for the REE, comparable to that for
0
the formation of monazite. It should be noted that owing to the lack of
-2
experimental information (see above), the results of this modelling do
not consider the stability of carbonate species of REE, and are therefore
-4 preliminary. However, the extremely low solubility of bastnäsite per-
Deposition of bastnaesite-Ce mits us to speculate that even in the case of carbonate complexes having
-6 La total high stability, the REE-CO2-F system will behave similarly to the REE-F
Ce total system, and buffer REE concentrations to insignificant levels at realistic
-8 values of pH, thereby leaving the depositional mechanism unchanged.
Nd total In addition to carbonate rocks, many other rocks types contain min-
-10 eral assemblages that will buffer acidic fluids to high pH, e.g., all
0 0.05 0.1 0.15 0.2 0.25 0.3 feldspar-bearing rocks, which will do so by consuming H+ to form
Dolomite added to 1 kg of solution, mol micas. Other potential mechanisms for pH-driven REE mineral deposi-
tion are mixing with a higher pH fluid, e.g., meteoric or formational wa-
Fig. 23. Changes in the REE concentration in 1 kg of solution containing 10 wt.% NaCl, ters, or boiling/effervescence (because of the removal of acidic gases),
2 wt.% Na2SO4, 500 ppm HF, 120 ppm of Ce, 59.8 ppm of La, 49 ppm of Nd, and 12 ppm
of Pr during titration with dolomite at 300 °C and 500 bar.
both of which would be highly efficient mechanisms for producing fo-
cused precipitation of REE minerals.
Deposition of the REE, in principle, could also occur if the main car-
riers of REE, chloride or sulfate complexes, were destabilized due to pre-
triggers precipitation of fluoride-bearing minerals like fluocerite, for ex- cipitation of minerals containing the ligands, fluid-rock interaction, fluid
ample, through interaction of the fluid with host rocks that contain min- mixing, or a variety of redox processes. In the case of chloride, this po-
eral assemblages, which can buffer the pH to higher values. tentially could be achieved through mixing of brines with low salinity
The most obvious candidates for pH neutralization are carbonate metamorphic or meteoritic waters. Complexes of the REE with sulfate
rocks (e.g., dolomites and limestones). They also provide a source for could be destabilized through precipitation of highly insoluble sulfate-
the carbonate ion needed to precipitate bastnäsite and the calcium bearing phases, e.g., barite, which is a common gangue mineral in hy-
ion, which precipitates any excess fluorine as fluorite, the most impor- drothermal REE deposits (van Wambeke, 1977; Williams-Jones et al.,
tant gangue mineral in many hydrothermal REE deposits (Chao et al., 2000; Wang et al., 2001), or through processes that reduce the sulfur,
1992; Williams-Jones et al., 2000) or REE deposits that have been hy- and transform it into species, such as thiosulfates, sulfides, or native sul-
drothermally remobilized (Gysi and Williams-Jones, 2013). For an ex- fur. Migdisov and Williams-Jones (2014) evaluated a depositional sce-
ample of a REE deposit that formed in this manner, we need look no nario in which 1 kg of solution containing 10 ppm Nd, 500 ppm total
further than Bayan Obo, the World's richest REE deposit, which is hosted fluorine, 2 wt.% sulfate, and 10 wt.% NaCl is titrated by BaCl2 at 400 °C
dominantly by sedimentary rocks that were formerly dolomite (Chao and a pHT of 4.5. This model could represent mixing of REE-rich fluids
et al., 1992). With our current level of knowledge, we cannot model with a Ba-bearing fluid, or interaction of the REE fluid with a Ba-rich
the evolution of the fluid and minerals during this type of deposition wall rock. The aqueous speciation of Ba was modeled using Ba2 +,
due to the lack of thermodynamic data for carbonate aqueous species BaOH+, BaF+, and BaCl+ (Shock et al., 1997; Sverjensky et al., 1997),
of REE and a solid solution model for bastnäsite. However, the data of and the data for barite were taken from Robie and Hemingway
Gysi and Williams-Jones (2015) allow us to check if minerals with the (1995). The results show that deposition of barite can remove consider-
stoichiometry Ce0.5La0.25Nd0.20Pr0.05CO3F (bastnäsite-(Ce)) and able sulfate from solution, thereby destabilizing Nd(SO4)− 2 . This process,
CaCe0.95La0.6Nd0.35Pr0.10(CO3)3F (parisite-(Ce)) precipitated during the however, consumes large proportions of the Ba-bearing species Thus,
course of such interaction, and, in doing so, to evaluate to a first approx- ~17 g of BaCl2 were required to deposit 9.9 mg of Nd as NdF3(s) and re-
imation the efficiency of the proposed depositional mechanism. We duce the total dissolved Nd from 10 ppm to 200 ppb (Fig. 24a). The rea-
modeled the neutralization of 1 liter of REE-bearing fluid by reacting it son for the large consumption of Ba is the very high stability of
with dolomite. We assumed that the fluid contained stoichiometric pro- Nd(SO4)− 2 . Destabilization of this species requires almost complete re-
portions of Ce, La, Nd, and Pr, corresponding to the composition of the moval of sulfate from the solution, and thus addition of an equal amount
bastnäsite-(Ce), studied by Gysi and Williams-Jones (2015): 120 ppm of Ba. It should be noted that this model describes a very specific system,
of Ce, 59.8 ppm of La, 49 ppm of Nd, and 12 ppm of Pr. These composi- which was buffered with respect to solid NdF3, and in which the low
tions were set to be close to that of one of the samples analysed by Banks concentrations of chloride complexes of Nd were dictated by the pres-
et al. (1994) (sample W-3). The model was calculated for 300 °C and ence of this solid (see Fig. 21). In a fluoride-free solution, lacking the
500 bar. Interaction of a REE-F-bearing solution with dolomite results above buffering capacity, destabilization of REE sulfate complexes
in immediate deposition of bastnäsite (Fig. 23). This deposition is con- does not lead to precipitation of a Nd solid (Fig. 24b), but instead dis-
trolled primarily by the total CO2 concentration in the solution, and be- places the system to conditions for which there is a greater contribution
gins when the latter reaches ~ 0.05 mol/kg. The formation of REE of chloride complexes to REE transport. We therefore conclude that, al-
fluorides and parisite, was not detected: at all conditions bastnäsite is though the depositional mechanism described above can operate under
significantly more stable. Deposition of bastnäsite is likely not con- some conditions (e.g., fluoride-bearing fluids, solubility buffered by an
trolled by variations in the solution pH as its onset occurs significantly existing REE-bearing solid), the likelihood of these conditions occurring
ahead of the neutralization of the solution by dolomite. We therefore in nature is very low.
conclude that its deposition in response to interaction of REE-F-bearing A very efficient means of depositing REE from a fluid is to introduce
solutions with limestones and dolomites is not due to neutralization of phosphate, which interacts with the REE directly and deposits these
pH, but rather the introduction into the system of a new strong binding elements in the form of highly insoluble monazite or xenotime.
ligand, namely carbonate, which saturates the solution with bastnäsite. These minerals are two of the main ore minerals for the light REE
This suggests that bastnäsite deposition can be triggered, not only by and heavy REE, respectively, and, thermodynamic data suggest that
36 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

0 -3
A 400 °C, 1000 bar, pH=4.5 400 °C, 1000 bar, pH=4.5
-1 -4 10 ppm
Starting composition:
Total dissolved Ba Nd(SO 4) 2-
-2 NaCl 100 g/kg; Total dissolved SO 4 Starting composition:
-5 NaCl 100 g/kg; 1 ppm
Na2SO4 20 g/kg 2+
NdCl Na2SO4 20 g/kg
-3 HF 500 mg/kg 100 ppb
-6
log m

HF 500 mg/kg

log m
+ NdF3 solid 2+
-4 10 ppm NdF + NdF3 solid
-7 10 ppb
Nd(SO 4) 2- 1 ppm Nd(SO 4) 2-
-5 NdCl 2+
-8 1 ppb
-6 2+ 100 ppb
NdCl Nd
3+

-7 10 ppb -9

-8 -10
0 5 10 15 20 25 0 100 200 300 400 500
BaCl2 added to 1 kg of the solution, g Apatite added to 1 kg of the solution, mg

0 Fig. 25. Changes in the distribution of species in 1 kg of solution containing 10 wt.% NaCl,
B 2 wt.% Na2SO4, and 500 ppm HF during titration with apatite at 400 °C, 1000 bar, pH = 4.5,
-1 and Nd 10 ppm. Distributions of subordinate Nd aqueous species and Nd-free aqueous
species are shown by dashed lines.
-2 Total dissolved SO 4
2+
-3 NdCl
log m

Nd(SO 4) 2- + 5. Hydrothermal fractionation of REE


-4 NdCl 2

-5 Starting composition:
Nd
3+ Earlier we mentioned that the chemical properties of REE are very
NaCl 100 g/kg;
Na2SO4 20 g/kg similar, but change systematically with atomic number of the
-6
lanthanoids. From the datasets presented in this contribution, it is evi-
-7 dent that, in some cases, these changes are sufficient to lead to fraction-
ation of the REE during their hydrothermal transport and deposition.
-8
Fractionation may occur due to the variable stability of REE complexes
0 10 20 30 40
in hydrothermal solutions. Here, we consider two ligands, chloride
BaCl2 added to 1 kg of the solution, g
and sulfate, as the main vehicles for the hydrothermal transport of
REE. In Figs. 1 and 4, it can be seen that, whereas the stability of sulfate
Fig. 24. Changes in the distribution of species in A) 1 kg of solution containing 10 wt.% complexes is remarkably similar for the different REE at all tempera-
NaCl, 2 wt.% Na2SO4, and 500 ppm HF during titration with BaCl2 at 400 °C, 1000 bar,
tures investigated, the stability of REE chloride complexes decreases
pH = 4.5, and Nd 10 ppm (the concentration of the fluid saturated with NdF3) B) 1 kg
of fluoride-free solution containing 10 wt.% NaCl, and 2 wt.% Na2SO4, during titration with increasing atomic number and this decrease increases with in-
with BaCl2 at 400 °C, 1000 bar. creasing temperature. Thus at T ≥ 250 °C, chloride complexes of the
LREE are 1.5 orders of magnitude more stable (and, hence, more mo-
bile) than those of the HREE. A similar effect has been observed for
REE fluoride complexes. However, as discussed above, it is unlikely
both are very insoluble. Interaction of the REE-bearing fluid with that fluoride complexes contribute significantly to the hydrothermal
wall rocks containing phosphate (for example, in the form of apatite) mobility of the REE in ore-forming processes.
will lead to precipitation of monazite and/or xenotime, depending on Another potential cause of REE fractionation is the differential solu-
the relative proportions of light and heavy REE. Migdisov and bility of REE-bearing minerals. In any of the depositional processes de-
Williams-Jones (2014) evaluated a scenario involving the titration scribed above, minerals with lower solubility will be deposited before
by apatite, at 400 °C and a pH of 4.5, of 1 kg of solution containing those with higher solubility. If deposition occurs across physicochemical
10 ppm Nd, 500 ppm of total fluorine, 2 wt.% of sulfate, and 10 wt.% gradients that are spatially distributed, and if the minerals that deposit
of NaCl. The thermodynamic data for apatite were taken from Robie differ in their proportions of the REE, their differential solubility will
and Hemingway (1995), and those for non-REE phosphate aqueous lead to fractionation of the REE. This will also apply for a single mineral,
species were taken from Shock et al. (1997). Although these calcula- if the solubility of that mineral varies according to its content of the dif-
tions were performed for a fluorine-bearing system, the similar be- ferent REE. Unfortunately, there are only three minerals for which this
havior during titration of all species carrying Nd suggests that this control on fractionation currently can be evaluated, namely monazite,
scenario, in contrast to the scenario of destabilization of sulfate com- xenotime and fluocerite. From Fig. 26a, which shows that the solubility
plexes, is applicable to solutions of other compositions. At the condi- products of the different REE phosphate endmembers are very similar,
tions of the calculation, monazite is significantly less soluble than we can conclude that deposition of monazite would not lead to fraction-
apatite, and therefore all the apatite added to the system is con- ation of the REE. The same is also true for xenotime (Gysi et al., 2015). In
sumed to form monazite-(Nd) until the Nd content of the solution contrast, to monazite, the solubility product of fluocerite varies consid-
is exhausted (Fig. 25). Consequently, the molar mass of apatite re- erably with atomic number, increasing by 2 to 3 orders of magnitude
quired to deposit the dissolved Nd is the same as the molar mass of from Ce to Lu, depending on temperature (Fig. 26b), and thus, in princi-
Nd in the solution. The results of this calculation show that a mass ple, its deposition could lead to fractionation of the REE. In practice, this
of just a few hundred milligrams of apatite is sufficient to precipitate is unlikely to occur commonly because of prior precipitation of
virtually all of the Nd from the solution. This scenario is very similar bastnäsite, for which unfortunately there are no data with which to as-
to that described above for the deposition of bastnäsite-(Ce), and is sess whether or not it can play a role in REE fractionation.
perhaps one of the most efficient mechanisms for the deposition of Finally, fractionation of the REE can occur during the course of inter-
hydrothermally transported REE. action of REE-bearing solutions with a nominally REE-free host rock or
Finally, as can be seen from Figs. 19 and 21, precipitation of REE from hydrothermally formed minerals. Incorporation of REE in the crystallo-
fluoride-bearing solutions can be triggered by simple cooling. Indeed, graphic structure of these minerals is a complex process controlled by
our calculations show that an ore fluid will precipitate 99% of the dis- ionic radii of incorporated REE, their charges, and speciation of the
solved REE upon cooling from 400 °C (pH400 °C ~ 4) to 300 °C. REE in aqueous solutions. In the following sections, we discuss in
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 37

REEPO4 = REE3+ + PO43-


A A
-24 25 °C 1kg
wave n+1 1kg
1kg

200 °C
250 °C
-26 1kg

300 °C
1kg
150 °C

350 °C
-28 1kg

400 °C
-30
log Ks

250 °C 1kg
-32

-34 wave n
500 bar
-36
350 °C
-38

-40
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb
1.4E-03 B wave 10

REE phosphate redeposited, mol


B 1.2E-03
-18 Gd
-19
REEF3 = REE3++ 3F- 150 °C flow
1.0E-03
-20
8.0E-04 La
-21 200 °C
log Ks

-22 6.0E-04
-23 250 °C Y Er Pr
4.0E-04
-24 Yb
-25 Nd
2.0E-04 Dy
-26 Ce
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb 0.0E+00
450 440 430 420 410 400 390
T °C
Fig. 26. The distribution of values for the solubility products of REE phosphates (A) and
fluorides (B) at different temperatures.
REE phosphate redeposited, mol

1.4E-03 C wave 70
La
more detail the three types of hydrothermal fractionation described 1.2E-03
above and present some model examples for the purpose of illustration. Yb flow Pr
1.0E-03

5.1. REE fractionation through differences in mobility 8.0E-04 Er


6.0E-04 Gd
A preliminary model illustrating the effect of differential mobility on Nd
REE fractionation by hydrothermal chloride-bearing solutions was pre- 4.0E-04
Dy Sm
sented in Williams-Jones et al. (2012). This model, however, was devel- 2.0E-04
oped only for REE with atomic numbers between La to Gd, and was
based on a dataset, which did not include the corrections to the data 0.0E+00
450 400 350 300 250 200
of Popa and Konings (2006) discussed above. Here, we re-evaluate
T °C
this model with the revised thermodynamic data and expand it to in-
clude the HREE. The model reported here simulates interaction of a
Fig. 27. Fractionation of REE after interaction of a 10 wt.% NaCl solution carrying REE
10 wt.% NaCl REE-bearing solution with a pHT of 3.5 with a host rock (100 ppm) at pHT = 3.5 with a host rock containing 3 wt.% apatite. A) A sketch of the
containing 3 wt.% apatite, in order to trigger deposition of monazite/ model. B) Fractionation observed after interaction of 10 aliquots of the solution.
xenotime. To emphasize the controls of hydrothermal fractionation as- C) Fractionation observed after interaction of 70 aliquots of the solution. See text for
sociated with this interaction, we used highly idealized (and geochem- further detail.

ically unrealistic) compositions of the solution and minerals. For


example, the initial concentration of each of the REE was assumed to aliquot (second wave) that reacts with the rock and the minerals crys-
be 100 ppm, in order to permit direct comparison among them. Another tallized during the first wave. A further 98 aliquots (waves) follow.
assumption was that the rock containing apatite is chemically inert (in The first wave results in deposition of virtually all the dissolved REE as
practice fluid-rock interaction changes the composition of both compo- monazite/xenotime on contact and produces no fractionation. However
nents in a manner that depends highly on the composition of the latter after 10 aliquots have passed through the rock, separation of the REE is
(e.g., Seward et al., 2014)). This assumption was made in order to min- evident (Fig. 27b). The lightest REE, being more mobile, are re-
imize these effects of fluid-rock interaction and to address only those as- mobilized furthest along the fluid path. Thus, the peak of LaPO4 re-
sociated with the mechanisms of REE-chloride transport and monazite/ deposition is displaced to a location corresponding to a drop of temper-
xenotime deposition. It should be noted, however, that the mechanisms ature of between 40 and 50 °C, whereas the other REE phosphates have
discussed below can be easily adapted for natural systems using the peaks that, with increasing atomic number, are progressively closer to
data summarized in Tables 1-13. the location of fluid input. After 70 aliquots of fluid have passed through
In the scenario illustrated in Fig. 27a, the REE-bearing solution re- the rock, there is complete separation of LREE, MREE, and HREE
ferred to above comes in contact with the rock containing apatite at (Fig. 27c). The scenario discussed here is applicable to hydrothermal
450 °C and crystallizes monazite/xenotime by consuming apatite. re-mobilization of previously deposited masses of phosphates of either
After contact, the first aliquot of solution (1 liter) continues flowing hydrothermal or magmatic origin. It is also applicable to the case of slow
through the rock, gradually cools to 200 °C, and is followed by a second continuous deposition of hydrothermal REE phosphates, which after
38 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

deposition are immediately re-mobilized (partially or completely) by 4.0E-04


wave 1 A

REE fluoride deposited, mol


the next aliquot of fluid. The processes discussed above should be man- 3.5E-04 flow
ifested by differences in the relative proportions of the different REE in
3.0E-04 La
minerals located at the beginning of the column (leached minerals)
and minerals located at the re-mobilization front (re-deposited min- 2.5E-04 Gd Nd
erals). In the ideal case, whereas the leached phosphates should be Pr
2.0E-04
highly enriched in HREE, and most likely represented by xenotime, Sm
the minerals at the front of the re-mobilization wave should be enriched 1.5E-04 Yb
Ce
in LREE and dominated by monazite. It should be noted, however, that
1.0E-04
in natural solutions concentrations of the different REE may vary by or-
Er
ders of magnitudes and therefore the ideal distribution predicted by our 5.0E-05
model could be significantly altered.
0.0E+00
450 400 350 300 250 200
T °C
5.2. REE fractionation through differences in solubility 1.2E-03
B

REE fluoride deposited, mol


flow
In order to illustrate the role of solubility in controlling the frac- 1.0E-03
La
tionation of the REE, we re-calculated the model with the scenario wave 4
8.0E-04
shown in Fig. 27a, using a 2 wt.% sulfate solution free of chloride in
order to eliminate the effects associated with differences in the sta- Nd
6.0E-04 Ce Sm
bility of the different REE-chlorides. From Fig. 4, it is evident that Yb
Pr
the stability of the REE sulfate species is independent of the atomic 4.0E-04
number of the lanthanoids at any of the temperatures investigated. Gd Er
Therefore, the fractionation observed for such solutions depends en- 2.0E-04

tirely on the changing properties of the solids. As in the previous


0.0E+00
model, the solution contained equal concentrations of all the REE 450 400 350 300 250 200 150 100
(100 ppm) and the pH was set at 3.5. The rock interacting with the T °C
solution contained 3 wt.% fluorite instead of the 3 wt.% apatite used
in the previous model. Consumption of fluorite during interaction
Fig. 28. Fractionation of REE caused by deposition of REE fluorides from a 2 wt.% sulfate
with the dissolved REE, leads to the crystallization of solid REE fluo- solution carrying REE (100 ppm) at pHT = 3.5 and interacting with a host rock
rides, and therefore the distribution of the REE obtained in this containing 3 wt.% fluorite. A) Fractionation observed after interaction of one aliquot of
model only reflect changes in the solubility of these solids. the solution. B) Fractionation was observed after interaction of four aliquots of the
solution.
Fig. 28 illustrates the modelled distribution of the REE fluorides.
Contact of the rocks with the first aliquot of fluid resulted in immediate
fractionation of the REE. The first REE deposited from the solution are
La and Ce fluorides, which are the least soluble of the REE fluorides. 5.3. Crystallographic control
These LREE fluorides are followed along the path by MREE fluorides,
and, finally, the HREE fluorides (Er and Dy) which are deposited furthest Because the ionic radii of the REE (3 +, and 2 + in the case of Eu)
along the path at the lowest temperatures. Further flushing of the rock are similar to that of Ca2 + they can be incorporated in trace amounts
with the fluid leads to a more pronounced fractionation of REE. After in many common Ca-bearing major (e.g., feldspars) and minor
just four aliquots of fluid there is complete separation of the different (e.g., garnet, apatite, scheelite, fluorite, carbonates) rock-forming
REE (Fig. 28b). minerals (e.g., Oberti and Caporuscio, 1991). Therefore, in many sys-
A similar effect to that described above also can be observed, if tems, nominally REE-free minerals control the fractionation of REE
the main transporting ligand is chloride instead of sulfate. Fig. 29 during fluid-rock interaction. Depending on the size and geometry
shows this for the interaction of a solution containing 10 wt.% NaCl of the crystallographic sites, these minerals preferentially incorpo-
and 100 ppm of each REE at a pH of 3.5, with a rock containing rate light, middle, or heavy REE (“crystallographic control”; van
3 wt.% of fluorite. In this system, the effect of the solubility of the
REE fluorides, which is greatest for the HREE and least for the LREE,
competes with that of the stability of the REE chloride species, 3.5E-04
REE fluoride deposited, mol

which is highest for the LREE and lowest for the HREE. Fig. 29 Sm
3.0E-04
shows that the “solubility” effect largely out-competes the “mobili-
ty” effect for the modeled conditions, although the observed distri- 2.5E-04
bution of REE fluorides along the rock path shows a poorer Dy
2.0E-04 Pr Er
correlation with the atomic number of the REE than in the case of La
Gd
the sulfate solutions. This is not surprising considering that the solu- 1.5E-04 La Ce
bility of the HREE is 2.5–3 orders of magnitude greater than that of Yb
1.0E-04 Yb
LREE, whereas even at the highest temperature the stability of the Ce
LREE chloride species exceeds that of the HREE species by only ~ 1.5 5.0E-05 Nd
orders of magnitude. It should be emphasized that the fractionation
scenario proposed in this section is built on the assumption that pre- 0.0E+00
450 400 350 300 250 200 150
cipitation of the REE occurs in the form of fluocerite (REEF3). This
T °C
fractionation cannot be achieved, if these elements are deposited in
the form of monazite/xenotime, and the data needed to evaluate it
Fig. 29. Fractionation of REE caused by deposition of REE fluorides from a 10 wt.% NaCl
for REE fluorocarbonates like bastnäsite, which in most cases will solution carrying REE (100 ppm) at pHT = 3.5 and interacting with a host rock
precede and therefore preclude fluocerite precipitation, are not containing 3 wt.% fluorite. Fractionation was observed after interaction of four aliquots
available. of the solution
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 39

Hinsberg et al., 2010). A plot of the logarithm of the partition coeffi- following coupled substitution likely accounts for the incorporation of
cient versus the ionic radius of the REE produces a smooth parabolic Eu3+ in many hydrothermal minerals (Brugger et al., 2002):
curve, the maximum of which indicates the theoretical preferred ionic
þ þ
radius for the REE in the mineral of interest (lattice strain model of 2Ca2þ 3þ 2þ 3þ
Mineral þ Euaq þ Naaq ↔2Caaq þ EuMineral þ NaMineral ð11aÞ
Blundy et al., 1996). Extreme fractionation from convex upwards to
convex downward patterns can be obtained via closed system precipi- a2Ca2þ aEu3þ aNaþ
tation of a mineral favoring MREE (Brugger et al., 2000, 2002). K2 ðP; TÞ ¼ aq Mineral Mineral

The REE are commonly used in tracking fluid sources and chemistry a2 aEu3þ aNaþaq
Ca2þ aq

in hydrothermal systems, but their use has been hindered by the


Mineral
h i2 h i 
γ 2Ca2þ γ Eu3þ γ Naþ þ
Ca2þ
aq Eu3þMineral NaMineral
complexity of the controls on REE fractionation in such environ- ¼
Mineral Mineral
i : ð11bÞ
aq
h i2 h ih
ments (e.g., review by Gieré, 1996). Under equilibrium conditions, γ2 γ Eu3þ γ Naþaq
Ca2þ
Mineral aq Ca2þMineral Eu3þ
aq Naþaq
the REE contents of hydrothermal minerals depend upon (i) the
REE composition of the fluid, (ii) the pressure and temperature,
(iii) the nature of the REE aqueous complexes, and (iv) the fluid As Na and REE are trace elements substituting for a Ca2+ site, and ac-
redox and pH (e.g., Sverjensky, 1984). Non-equilibrium effects such as cording to the way the standard states have been fixed, aCa2+ Mineral
≈ 1,
sector zoning also cause fractionation among the REE ( e.g., Rakovan aNa+Mineral ≈ [Na+
Mineral], and aREE3+
Mineral
≈ [REE3+
Mineral] (Wagner and Schottky,
and Reeder, 1996). 1930), we obtain from the definition of the partition coefficient in
These equilibrium controls are illustrated by the following equations Eq. (8):
(details in Brugger et al., 2008). The oxidation state of Eu in the fluid h i
(i.e., Eu2+/Eu3+ ratio) depends not only on redox, but also on specia- γREE3þ γ Naþaq Naþ aq
Mineral=Fluid
¼ K2 ðP; TÞ h i ð12Þ
aq
tion, which depends on pH, the concentration of ligands, P and T; for ex- D 3þ  þ
Eu γ2 NaMineral Ca2þ
Ca2þ aq
ample aq

h i−

Eu2þ Clþ þ 0:25 O2 ðgÞ þ 3:5 H2 O ¼Eu3þ ðOHÞ4 þ Cl þ 3 Hþ ð7aÞ Thus, in contrast to the case where both the carrier and the trace el-
ement have the same valency, the partition coefficient resulting from a
logKP;T ¼ logKP;T ¼ a½Eu3þ ðOHÞ − þ aCl− coupled substitution such as Eq. (5) is dependent upon the composition
4
of the solution ([Na+ 2+ +
aq]/[Caaq ]) and of the solid (1/[NaMineral]), even
þ 3 aHþ −a Eu2þ Cl þ −0:25 aO2 ðgÞ þ 3:5 aH2 O ð7bÞ
½  when ideality governs the system. van Hinsberg et al. (2010) demon-
strated experimentally that in spite of this complexity, a simple ‘lat-
where KP,T is the equilibrium constant for the reaction, and is con- tice-strain model’ of REE incorporation can reproduce the partitioning
stant for a given pressure and temperature, and ai refers to the activity of REE between fluid and mineral. As expected, the partitioning depends
of the subscripted species. strongly on the nature of the REE complexes in solution, but this depen-
The partitioning of a trace element between a mineral and a fluid is dence is predictable.
described by the partition coefficient (DMineral/Fluid
Eu in the case of Eu), XANES studies have demonstrated preservation of Eu2+ and Eu3+ in
which is defined as: hydrothermal apatite and scheelite (Rakovan et al., 2001; Takahashi
et al., 2005; Brugger et al., 2006, 2008), providing support for the in-
Mineral=Fluid  tot     tot   tot  creased stability of Eu2+ in aqueous fluids at elevated temperature pre-
DEu ¼ EuMineral = Catot
Mineral = EuFluid = CaFluid ð8Þ
dicted by Sverjensky (1984). The thermodynamic analysis of Brugger
et al. (2008), based on the data for aqueous Eu species from Haas et al.
(1995) and Sverjensky (1984), shows that the variation in Eu2+/Eu3+
In the case of divalent europium, Eu2+ replaces Ca2+ via the follow- ratios in hydrothermal scheelite from an Archean orogenic Au deposit
ing isomorphous substitution: is consistent with fluid-rock interaction affecting mainly pH in these
systems (see also Fig. 8).
Ca2þ 2þ 2þ 2þ
Mineral þ Euaq ↔EuMineral þ Caaq : ð9aÞ
6. Conclusions

The corresponding mass action equation is written: The body of experimental data that has become available over the
last 15 years has allowed us to compile datasets on the thermodynamic
h i h i properties of aqueous species and REE minerals, which can be used to
aEu2þ aCa2þ Eu2þ
Mineral γ Eu2þ Ca2þ
aq γ Ca2þ
K1 ðP; TÞ ¼ Mineral aq
¼ h i Mineral
h i aq : ð9bÞ evaluate the behavior of the REE in hydrothermal processes. Although
aEu2þ aCa2þ 2þ
Euaq γ Eu2þ 2þ
CaMineral γCa2þ clearly incomplete, these datasets can be used to identify the mecha-
aq Mineral
aq Mineral
nisms mainly responsible for the transport, deposition, and fraction-
ation of these elements in hydrothermal systems. Models employing
where γi is the activity coefficient for species (i). From the definition these datasets suggest that the main ligands responsible for the trans-
of the partition coefficient (reaction (2)), we obtain: port of the REE at hydrothermal conditions are chloride and sulfate. Im-
γ Eu2þ γ Ca2þ portantly, they also suggest that fluoride, which previously was
Scheelite=Fluid considered to be the main ligand enabling hydrothermal transport of
DEu2þ ≈K1 ðP; TÞ aq Scheelite
ð10Þ
γCa2þ γEu2þ these elements, instead, is a depositional ligand. Indeed, the presence
aq Scheelite

of fluoride in hydrothermal solutions, rather than facilitating REE trans-


Hence, in the case of a homogeneous substitution, the partition coef- port, in most cases, considerably reduces the range of conditions for
ficient is relatively constant at a given pressure and temperature, espe- which the REE can be stabilized in hydrothermal solutions at significant
cially in relatively dilute solutions. concentrations.
However, the situation is different for coupled substitutions, as is the The presence of fluoride, especially combined with the neutraliza-
case for the substitution of REE3+ for Ca2+. Generally, charge balance tion of fluid acidity, can promote effective REE mineral deposition. How-
will be maintained via substitution of a monovalent ion; Na+ is the ever, the most effective means of inducing REE mineral deposition
most abundant monovalent ion in most crustal fluids, so that the involve the addition of phosphate or carbonate to the REE-bearing
40 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

solutions, which will trigger precipitation of monazite/xenotime and Blundy, J.D., Wood, B.J., Davies, A., 1996. Thermodynamics of rare earth element partitioning be-
tween clinopyroxene and melt in the system CaO-MgO-Al2O3-SiO2. Geochim. Cosmochim.
bastnäsite (in the presence of fluoride), respectively. In the case of REE Acta 60, 359–364.
transport as sulfate complexes, an additional means of precipitating Brugger, J., Etschmann, B., Chu, Y., Harland, C., Vogt, S., Ryan, C., Jones, H., 2006. In-situ XANES
spectroscopy of europium in hydrothermal scheelite: evidence for heterovalent europium.
REE minerals is to destabilize the REE sulfate complexes by removing Can. Mineral. 44, 1079–1087.
the sulfate through, for example, deposition of a relatively insoluble Brugger, J., Etschmann, B., Pownceby, M., Liu, W., Grundler, P., Brewe, D., 2008. Oxidation state of
mineral like barite. europium in scheelite: tracking fluid-rock interaction in gold deposits. Chem. Geol. 257,
26–33.
Fractionation of the REE can be triggered by both transport and de- Brugger, J., Lahaye, Y., Costa, S., Lambert, D., Bateman, R., 2000. Inhomogeneous distribution of
positional mechanisms. The former mechanism is related to the greater REE in scheelite and dynamics of Archaean hydrothermal systems (Mt. Charlotte and
Drysdale gold deposits, Western Australia). Contrib. Mineral. Petrol. 139, 251–264.
stability and thus higher solubility of LREE chloride species relative to
Brugger, J., Maas, R., Lahaye, Y., McRae, C., Ghaderi, M., Costa, S., Lambert, D., Bateman, R., Prince,
HREE chloride species. Thus the LREE are more easily leached and K., 2002. Origins of Nd-Sr-Pb isotopic variations in single scheelite grains from Archaean
transported than the HREE, and therefore will move further than the gold deposits, Western Australia. Chem. Geol. 182, 203–225.
Burnhama, A.D., Berry, A.J., 2014. The effect of oxygen fugacity, melt composition, temperature
less mobile HREE. Fractionation of the REE can also be effected by vari- and pressure on the oxidation state of cerium in silicate melts. Chem. Geol. 366, 52–60.
ations in the solubility of a REE mineral that depend on the relative pro- Canaval, L.R., Rode, B.M., 2015. The hydration properties of Eu(II) and Eu(III): an ab initio quan-
portions of the different REE endmembers in the solid solution. In the tum mechanical molecular dynamics study. Chem. Phys. Lett. 618, 78–82.
Cetiner, Z., Wood, S.A., Gammons, C.H., 2005. The aqueous geochemistry of the rare earth ele-
case of fluocerite, the solubility of HREE endmembers is significantly ments. Part XIV. The solubility of rare earth element phosphates from 23 to 150 °C.
higher than that of the LREE endmembers. Thus, progressive deposition Chem. Geol. 217, 147–169. http://dx.doi.org/10.1016/j.chemgeo.2005.01.001.
Chao, E.C.T., Back, J.M., Minkin, J.A., 1992. Host-rock controlled epigenetic, hydrothermal meta-
of fluocerite should result in crystallization of the LREE fluoride somatic origin of the Bayan Obo REE-Fe-Nb ore deposit, Inner Mongolia, P.R.C. Appl. Geo-
endmembers proximal to source and the HREE endmembers distal to chemistry 7, 443–458.
source. Finally, fractionation of the REE can occur during their incorpo- Cook, N.J., Ciobanu, C.L., O’Reilly, D., Wilson, R., Das, K., Wade, B., 2013. Mineral chemistry of rare
earth element (REE) mineralization, Browns Range, Western Australia. Lithos 172-173,
ration into the structure of nominally REE-free minerals. The latter 192–213.
mechanism is controlled by the crystallographic lattice of these min- D’Angelo, P., Spezia, R., 2012. Hydration of Lanthanoids(III) and Actinoids(III): an Experimental/
Theoretical Saga. Chem. Eur. J. 18 (36), 11162–11178.
erals and is mineral-specific.
Deberdt, S., Castet, S., Dandurand, J.L., Harrichoury, J.C., Louiset, I., 1998. Experimental study of La
The conclusions reached above are based on high temperature ex- (OH) 3 and Gd (OH) 3 solubilities (25 to 150 °C), and La-acetate complexing (25 to 80 °C).
perimental data and therefore were drawn with a reasonably high Chem. Geol. 151, 349–372. http://dx.doi.org/10.1111/j.1469-185X.1952.tb01392.x.
Diakonov, I.I., Ragnarsdottir, K.V., Tagirov, B.R., 1998. Standard thermodynamic properties and
level of confidence. It should be noted, however, that the data describing heat capacity equations of rare earth hydroxides: - II. Ce(III)-, Pr-, Sm-, Eu(III)-, Gd-, Tb-,
the behavior of the REE at elevated temperature are still very incom- Dy-, Ho-, Er-, Tm-, Yb-, and Y-hydroxides. Comparison of thermochemical and solubility
plete. In particular, there are no experimental data on the speciation data. Chem. Geol. 151, 327–347.
Fan, H.-R., Xie, Y.-H., Wang, K.-Y., Tao, K.-J., Wilde, S.a., 2004. REE Daughter Minerals Trapped in
of the REE in carbonate-bearing solutions, very little is known about Fluid Inclusions in the Giant Bayan Obo REE-Nb-Fe Deposit, Inner Mongolia, China. Int. Geol.
the high temperature stability of hydroxyl complexes of the REE, the Rev. 46, 638–645. http://dx.doi.org/10.2747/0020-6814.46.7.638.
Frausto da Silva, J.J.R., Queimado, M.M., 1973. Solubility products of lanthanide fluorides. Rev.
stability of the end-members of the REE fluorocarbonates (bastnäsite,
Port. Quim. 15, 29–34.
parisite), and the stability of many other potentially important REE Gammons, C.H., Wood, S.A., Li, Y., 2002. Complexation of the rare earth elements with aqueous
minerals. chloride at 200 °C and 300 °C and saturated water vapor pressure. Geochem. Soc. Spec.
Publ. 7, 191–207.
Gammons, C.H., Wood, S.A., Williams-Jones, A.E., 1996. The aqueous geochemistry of the rare
earth elements and yttrium: VI. Stability of neodymium chloride complexes from 25 to
Acknowledgments 300 °C. Geochim. Cosmochim. Acta 60, 4615–4630. http://dx.doi.org/10.1016/S0016-
7037(96)00262-1.
Gavrichev, K.S., Ryumin, M.A., Khoroshilov, A.V., Nikiforova, G.E., Tyurin, A.V., Gurevich, V.M.,
The authors are very grateful to J.B. Fein and T, Horscroft for the invi- Starykh, R.V., 2013b. Thermodynamic properties and phase transitions of tetragonal modi-
tation to submit this paper. The manuscript benefited from the reviews fication of terbium orthophosphate. Vestn. St. Petersburg State Univ. 4, 186–197.
Gavrichev, K.S., Ryumin, M.a., Tyurin, a.V., Gurevich, V.M., Khoroshilov, a.V., Komissarova, L.N., 2012.
of two anonymous referees. Thermodynamic functions of erbium orthophosphate ErPO4 in the temperature range of
0–1600 K. Thermochim. Acta 535, 1–7. http://dx.doi.org/10.1016/j.tca.2012.02.002.
Gavrichev, K.S., Ryumin, M.a., Tyurin, a.V., Gurevich, V.M., Komissarova, L.N., 2010. Heat capacity
Appendix A. Supplementary data and thermodynamic functions of xenotime YPO4(c) at 0–1600 K. Geochem. Int. 48,
932–939. http://dx.doi.org/10.1134/S0016702910090065.
Gavrichev, K.S., Ryumin, M.a., Tyurin, a.V., Gurevich, V.M., Nikiforova, G.E., Komissarova, L.N.,
Supplementary data to this article can be found online at http://dx. 2013a. Heat capacity and thermodynamic functions of YbPO4 from 0 to 1800 K. Inorg.
doi.org/10.1016/j.chemgeo.2016.06.005. Mater. 49, 701–708. http://dx.doi.org/10.1134/S0020168513070042.
Gavrichev, K.S., Smirnova, N.N., Gurevich, V.M., Danilov, V.P., Tyurin, a.V., Ryumin, M.a.,
Komissarova, L.N., 2006. Heat capacity and thermodynamic functions of LuPO4 in the
range 0–320 K. Thermochim. Acta 448, 63–65. http://dx.doi.org/10.1016/j.tca.2006.05.019.
References Gieré, R., 1996. Formation of rare earth minerals in hydrothermal systems. In: Jones, A.P., Wall,
F., Williams, T.C. (Eds.), Rare Earth Minerals: Chemistry, Origin and Ore Deposits. Chapman
Agangi, A., Kamenetsky, V.S., McPhie, J., 2010. The role of fluorine in the concentration and & Hall, London, pp. 105–150.
transport of lithophile trace elements in felsic magmas: insights from the Gawler Range Gültekin, A.H., Örgün, Y., Suner, F., 2003. Geology, mineralogy and fluid inclusion data of the
Volcanics, South Australia. Chem. Geol. 273, 314–325. http://dx.doi.org/10.1016/j. Kizilcaören fluorite–barite–REE deposit, Eskisehir, Turkey. J. Asian Earth Sci. 21, 365–376.
chemgeo.2010.03.008. http://dx.doi.org/10.1016/S1367-9120(02)00019-6.
Alibo, D.S., Nozaki, Y., 1999. Rare earth elements in seawater- particle association, shale- Gysi, A.P., Williams-Jones, A.E., 2013. Hydrothermal mobilization of pegmatite-hosted REE and
normalization, and Ce oxidation. Geochim. Cosmochim. Acta 63, 363–372. Zr at Strange Lake, Canada: A reaction path model. Geochim. Cosmochim. Acta 122,
Amano, O., Sasahira, A., Kani, Y., Hoshino, K., Aoi, M., Kawamura, F., 2004. Solubility of lanthanide 324–352. http://dx.doi.org/10.1016/j.gca.2013.08.031.
fluorides in nitric acid solution in the dissolution process of FLUOREX reprocessing system. Gysi, A.P., Williams-Jones, A.E., 2015. The thermodynamic properties of bastnäsite-(Ce) and
J. Nucl. Sci. Technol. 41, 55–60. http://dx.doi.org/10.3327/jnst.41.55. parisite-(Ce). Chem. Geol. 392, 87–101. http://dx.doi.org/10.1016/j.chemgeo.2014.11.001.
Banks, D., Yardley, B., Campbell, A., Jarvis, K., 1994. REE composition of an aqueous magmatic Gysi, A.P., Harlov, D., Filho, D.C., Williams-Jones, A.E., 2016. Experimental determination of the
fluid: a fluid inclusion study from the Capitan Pluton, New Mexico, U.S.A. Chem. Geol. high temperature heat capacity of a natural xenotime-(Y) solid solution and synthetic
113, 259–272. http://dx.doi.org/10.1016/0009-2541(94)90070-1. DyPO4 and ErPO4 endmembers. Thermochim. Acta 627–629, 61–67. http://dx.doi.org/10.
Baryshnikov, N.V., Gol’shtein, T.V., 1972. Solubility of yttrium and neodymium fluorides in aque- 1016/j.tca.2016.01.016.
ous solutions of nitric and hydrochloric acids. Nauch. Tr. Nauch.-Issled. Proekt. Inst. Gysi, A.P., Williams-Jones, A.E., Harlov, D., 2015. The solubility of xenotime-(Y) and other HREE
Redkometal. Prom. 45, 56–60. phosphates (DyPO4, ErPO4 and YbPO4) in aqueous solutions from 100 to 250 °C and psat.
Bastrakov, E.N., Skirrow, R.G., Didson, G.J., 2007. Fluid evolution and origins of iron oxide Cu-Au Chem. Geol. 401, 83–95. http://dx.doi.org/10.1016/j.chemgeo.2015.02.023.
prospects in the Olympic Dam district, Gawler craton, South Australia. Econ. Geol. 102, Haas, J., Shock, E.L., Sassani, D., 1995. Rare earth elements in hydrothermal systems: estimates of
1415–1440. standard partial molal thermodynamic properties of aqueous complexes of the rare earth
Bau, M., 1991. Rare-earth element mobility during hydrothermal and metamorphic fluid-rock elements at high. Geochim. Cosmochim. Acta 59, 4329–4350. http://dx.doi.org/10.1016/
interaction and the significance of the oxidation state of europium. Chem. Geol. 93, 0016-7037(95)00314-P.
219–230. Haschke, J.M., 1975. The lanthanum hydroxide fluoride carbonate system: the preparation of
Bilal, B.A., Langer, P., 1987. Complex formation of trace elements in geochemical systems: stabil- synthetic bastnaesite. J. Solid State Chem. 12, 115–121. http://dx.doi.org/10.1016/0022-
ity constants of fluorocomplexes of the lanthanides in a fluorite bearing model system up to 4596(75)90186-3.
200 °C and 1000 bar. Inorg. Chim. Acta 140, 297–298. http://dx.doi.org/10.1016/S0020- Hefter, G.T., 1984. Acidity constant of hydrofluoric acid. J. Solut. Chem. 13, 457–470. http://dx.
1693(00)81108-1. doi.org/10.1007/BF00647171.
A. Migdisov et al. / Chemical Geology 439 (2016) 13–42 41

Helgeson, H.C., Kirkham, D.H., Flowers, G.C., 1981. Theoretical prediction of the thermodynamic McLemore, V.T., 2011. Geology and mineral deposits of the Gallinas Mountains REE deposit, Lin-
behavior of aqueous electrolytes at high pressures and temperatures: IV. Calculation of ac- coln and Torrance Counties, NM; Preliminary report. SME Annual Meeting, Feb27-Mar02,
tivity coefficients, osmotic coefficients, and apparent molal and standard and relative par- Preprint 11-139 (Denver, CO).
tial molal properties to 600 °. Am. J. Sci. 281, 1249–1516. Mei, Y., Sherman, D., Liu, W., Brugger, J., 2013. Ab initio molecular dynamics simulation and free
Hitzman, M.W., Oreskes, N., Einaudi, M.T., 1992. Geological characteristics and tectonic setting of energy exploration of copper(I) complexation by chloride and bisulfide in hydrothermal
proterozoic iron-oxide (Cu-U-Au-Ree) Deposits. Precambrian Res. 58, 241–287. fluids. Geochim. Cosmochim. Acta 102, 45–64.
Holtstam, D., Andersson, U.B., 2007. The REE minerals of the Bastnäs-type deposits, south- Menon, M.P., James, J., 1989. Solubilities, solubility products and solution chemistry of lanthanon
central Sweden. Can. Mineral. 45, 1073–1114. trifluoride-water systems. J. Chem. Soc. Faraday Trans. 1 (85), 2683. http://dx.doi.org/10.
Hsu, L., 1992. Synthesis and stability of bastnaesites in a part of the system (Ce, La)-F-H-C-O. 1039/f19898502683.
Mineral. Petrol. 47, 87–101. Metz, M., Brookins, D., Rosenberg, P., Zartman, R., 1985. Geology and geochemistry of the Snow-
Itoh, H., Hachiya, H., Tsuchiya, M., Suzuki, Y., 1984. Determination of solubility products of rare bird deposit, Mineral County, Montana. Econ. Geol. 80, 394–409.
earth fluorides by fluoride ion-selective electrode. Bull. Chem. Soc. Jpn. 57, 1689–1690. Migdisov, A.A., Williams-Jones, A.E., 2002. A spectrophotometric study of neodymium (III) com-
Janka, O., Schleid, T., 2009. Facile synthesis of Bastnaesite-type LaF[CO3] and its thermal decom- plexation in chloride solutions. Geochim. Cosmochim. Acta 66, 4311–4323. http://dx.doi.
position to LaOF for bulk and Eu3+-doped samples. Eur. J. Inorg. Chem. 357–362. http://dx. org/10.1016/S0016-7037(02)00995-X.
doi.org/10.1002/ejic.200800931. Migdisov, A.A., Williams-Jones, A.E., 2006. A spectrophotometric study of erbium (III) speciation
Johnson, J.W., Oelkers, E.H., Helgeson, H.C., 1992. SUPCRT92: a software package for calculating in chloride solutions at elevated temperatures. Chem. Geol. 234, 17–27. http://dx.doi.org/
the standard molal thermodynamic properties of minerals, gases, aqueous species, and re- 10.1016/j.chemgeo.2006.04.002.
actions from 1 to 5000 bar and 0 to 1000 °C. Comput. Geosci. 18, 899–947. http://dx.doi. Migdisov, A.A., Williams-Jones, A.E., 2007. An experimental study of the solubility and speciation
org/10.1016/0098-3004(92)90029-Q. of neodymium (III) fluoride in F-bearing aqueous solutions. Geochim. Cosmochim. Acta 71,
Jones, A.P., Wall, F., Williams, C.T., 1996. Rare Earth Minerals: Chemistry, Origin and Ore De- 3056–3069. http://dx.doi.org/10.1016/j.gca.2007.04.004.
posits. Springer, Netherlands. Migdisov, A.A., Williams-Jones, A.E., 2008. A spectrophotometric study of Nd(III), Sm(III) and
Klungness, G.D., Byrne, R.H., 2000. Comparative hydrolysis behavior of the rare earths and yttri- Er(III) complexation in sulfate-bearing solutions at elevated temperatures. Geochim.
um: the influence of temperature and ionic strength. Polyhedron 19, 99–107. http://dx.doi. Cosmochim. Acta 72, 5291–5303. http://dx.doi.org/10.1016/j.gca.2008.08.002.
org/10.1016/S0277-5387(99)00332-0. Migdisov, A.A., Williams-Jones, A.E., 2014. Hydrothermal transport and deposition of the Rare
Konings, R.J.M., Kovács, A., 2003. Chapter 213 Thermodynamic properties of the lanthanide (III) Earth Elements by fluorine-bearing aqueous liquids. Mineral Deposits 49, 987–997.
halides. In: Gschneidner, K.A., Bünzli, J.-C.G., Pecharsky, V.K. (Eds.), Handbook on the Phys- Migdisov, A.A., Reukov, V.V., Williams-Jones, A.E., 2006. A spectrophotometric study of neodym-
ics and Chemistry of Rare Earths. Elsevier Science B.V., pp. 147–247. ium (III) complexation in sulfate solutions at elevated temperatures. Geochim. Cosmochim.
Konings, R.J.M., Beneš, O., Kovács, A., Manara, D., Sedmidubský, D., Gorokhov, L., Iorish, V.S., Acta 70, 983–992. http://dx.doi.org/10.1016/j.gca.2005.11.001.
Yungman, V., Shenyavskaya, E., Osina, E., 2014. The thermodynamic properties of the f- Migdisov, A.A., Williams-Jones, A.E., Normand, C., Wood, S.A., 2008. A spectrophotometric study
elements and their compounds. Part 2. The lanthanide and actinide oxides. J. Phys. Chem. of samarium (III) speciation in chloride solutions at elevated temperatures. Geochim.
Ref. Data 43, 013101. http://dx.doi.org/10.1063/1.3474238 (Citation). Cosmochim. Acta 72, 1611–1625. http://dx.doi.org/10.1016/j.gca.2008.01.007.
Lehmann, B., Nakai, S., Höhndorf, A., Brinckmann, J., Dulski, P., Hein, U., Masuda, A., 1994. REE Migdisov, A.A., Williams-Jones, A.E., Wagner, T., 2009. An experimental study of the solubility
mineralization at Gakara, Burundi: Evidence for anomalous upper mantle in the western and speciation of the Rare Earth Elements (III) in fluoride- and chloride-bearing aqueous
Rift Valley. Geochim. Cosmochim. Acta 58, 985–992. solutions at temperatures up to 300°C. Geochim. Cosmochim. Acta 73, 7087–7109. http://
Lewis, A.J., Komninou, A., Yardley, B.W.D., Palmer, M.R., 1998. Rare earth element speciation in dx.doi.org/10.1016/j.gca.2009.08.023.
geothermal fluids from Yellowstone National Park, Wyoming, USA. Geochim. Cosmochim. Mioduski, T., Gumiński, C., Zeng, D., 2008. IUPAC-NIST solubility data series. 87. Rare earth metal
Acta 62, 657–663. chlorides in water and aqueous systems. Part 1. Scandium group (Sc, Y, La). J. Phys. Chem.
Liu, Y., 2004. Re-Os dating of pyrite from Giant Bayan Obo REE-Nb-Fe deposit. Chin. Sci. Bull. 49, Ref. Data 37, 1765–1853. http://dx.doi.org/10.1063/1.2956740.
2627. http://dx.doi.org/10.1360/04wd0185. Mioduski, T., Gumiński, C., Zeng, D., 2009a. IUPAC-NIST Solubility Data Series. 87. Rare earth
Liu, X., Byrne, R.H., 1995. Comparative carbonate complexation of yttrium and gadolinium at metal chlorides in water and aqueous systems. Part 2. Light lanthanides (Ce–Eu). J. Phys.
25 °C and 0 . 7 mol dm−3 ionic strength. Mar. Chem. 51, 213–221. Chem. Ref. Data 38, 441–562. http://dx.doi.org/10.1063/1.3112775.
Liu, X., Byrne, R.H., 1997. Rare earth and yttrium phosphate solubilities in aqueous solution. Mioduski, T., Gumiński, C., Zeng, D., 2009b. IUPAC-NIST Solubility Data Series. 87. Rare earth
Geochim. Cosmochim. Acta 61, 1625–1633. http://dx.doi.org/10.1016/S0016- metal chlorides in water and aqueous systems. Part 3. Heavy lanthanides (Gd–Lu).
7037(97)00037-9. J. Phys. Chem. Ref. Data 38, 925–1011. http://dx.doi.org/10.1063/1.3212962.
Liu, W., Etschmann, B., Hazemann, J.-L., Testemale, D., Migdisov, A., Brugger, J., 2016. The hydro- Mioduski, T., Gumiński, C., Zeng, D., 2015a. IUPAC-NIST Solubility Data Series. 100. Rare
thermal geochemistry of europium(II/III): a review at the light of new in-situ XAS spectros- earth metal fluorides in water and aqueous systems. Part 2. light lanthanides (Ce–
copy results. Chem. Geol. (submitted for publication). Eu). J. Phys. Chem. Ref. Data 44. http://dx.doi.org/10.1063/1.4903362 (013102–1–
Loges, A., Migdisov, A.A., Wagner, T., Williams-Jones, A.E., Markl, G., 2013. An experimental study of 013102–55).
the aqueous solubility and speciation of Y(III) fluoride at temperatures up to 250 °C. Geochim. Mioduski, T., Gumiński, C., Zeng, D., 2015b. IUPAC-NIST Solubility Data Series. 100. Rare earth
Cosmochim. Acta 123, 403–415. http://dx.doi.org/10.1016/j.gca.2013.07.031. metal fluorides in water and aqueous systems. Part 3. Heavy lanthanides (Gd–Lu). J. Phys.
Long, K.R., Van Gosen, B.S., Foley, N.K., Cordier, D., 2010. The Principal Rare Earth Elements De- Chem. Ref. Data 44. http://dx.doi.org/10.1063/1.4918371 (023102–1–023102–50).
posits of the United States — A Summary of Domestic Deposits and a Global Perspective Gd Moffett, J.W., 1990. Microbially mediated cerium oxidation in seawater. Nature 345, 421–423.
Pr Sm Nd La Ce. US Geological Survey Scientific Investigations http://dx.doi.org/10.1007/ Moreau, G., Helm, L., Purans, J., Merbach, A.E., 2002. Structural investigation of the aqueous
978-90-481-8679-2_7. Eu2+ ion: Comparison with Sr2 + using the XAFS technique. J. Phys. Chem. A 106,
Louvel, M., Bordage, A., Testemale, D., Zhou, L., Mavrogenes, J., 2015. Hydrothermal controls on 3034–3043.
the genesis of REE deposits: insights from an in situ XAS study of Yb solubility and specia- Morss, L.R., Konings, R.J.M., 2005. Thermochemistry of binary rare earth oxides. In: Adachi, G.,
tion in high temperature fluids (T≤400 °C). Chem. Geol. 417, 228–237. http://dx.doi.org/10. Imanaka, N., Kang, Z.C. (Eds.), Binary Rare Earth Oxides. Kluwer Academic Publishers, Dor-
1016/j.chemgeo.2015.10.011. drecht, pp. 163–188 http://dx.doi.org/10.1007/1-4020-2569-6_7.
Lundberg, D., Persson, I., Eriksson, L., D’Angelo, P., De Panfilis, S., 2010. Structural study of the Navrotsky, A., Lee, W., Mielewczyk-Gryn, A., Ushakov, S.V., Anderko, A., Wu, H., Riman, R.C.,
N,N′-dimethylpropyleneurea solvated lanthanoid(III) ions in solution and solid state with 2015. Thermodynamics of solid phases containing rare earth oxides. J. Chem. Thermodyn.
an analysis of the ionic radii of lanthanoid(III) ions. Inorg. Chem. 49 (10), 4420–4432. 88, 126–141. http://dx.doi.org/10.1016/j.jct.2015.04.008.
Luo, Y., Byrne, R.H., 2000. The ionic strength dependence of rare earth and yttrium fluoride com- Oberti, R., Caporuscio, F.A., 1991. Crystal-chemistry of clinopyroxenes from mantle eclogites: a
plexation at 25 °C. Fluoride 29, 1089–1099. study of the key role of the M2 site population by means of crystal structure refinement.
Luo, Y., Byrne, R.H., 2001. Yttrium and rare earth element complexation by chloride ions at 25 °C. Am. Mineral. 76, 1141–1152.
J. Solut. Chem. 30, 837–845. Oelkers, E., Helgeson, H.C., 1991. Calculation of activity coefficients and degrees of formation of
Luo, Y.Y.-R., Byrne, R.H., 2004. Carbonate complexation of yttrium and the rare earth elements in neutral ion pairs in supercritical electrolyte solutions. Geochim. Cosmochim. Acta 55,
natural waters. Geochim. Cosmochim. Acta 68, 691–699. http://dx.doi.org/10.1016/S0016- 1235–1251. http://dx.doi.org/10.1016/0016-7037(91)90303-M.
7037(03)00495-2. Ohoa, G., Pilgrim, C.D., Martin, M.N., Colla, C.A., Klavins, P., Augustine, M.P., Casey, W.H., 2015. H2
Luo, Y.-R., Byrne, R.H., 2007. The Influence of Ionic Strength on Yttrium and Rare Earth Element and La139 NMR spectroscopy in aqueous solutions at geochemical pressures. Angew.
Complexation by Fluoride Ions in NaClO4, NaNO3 and NaCl Solutions at 25 °C. J. Solut. Chem. Chem. Int. Ed. 54, 15444–15447.
36, 673–689. http://dx.doi.org/10.1007/s10953-007-9141-6. Olivo, G.R., Williams-Jones, A.E., 1999. Hydrothermal REE-rich eudialyte from the Pilanesberg
Luo, Y., Millero, F.J., 2004. Effects of temperature and ionic strength on the stabilities of the first Complex, South Africa. Can. Mineral. 37, 653–663.
and second fluoride complexes of yttrium and the rare earth elements. Geochim. Oreskes, N., Einaudi, M.T., 1992. Origin of hydrothermal fluids at Olympic Dam - preliminary re-
Cosmochim. Acta 68, 4301–4308. http://dx.doi.org/10.1016/j.gca.2004.05.025. sults from fluid inclusions and stable isotopes. Econ. Geol. 87, 64–90.
Mayanovic, R.A., Anderson, A., Bassett, W., Chou, I., 2007. On the formation and structure of rare- Persson, I., 2010. Hydrated metal ions in aqueous solution: how regular are their structures?
earth element complexes in aqueous solutions under hydrothermal conditions with new Pure Appl. Chem. 82, 1901–1917.
data on gadolinium aqua and chloro complexes. Chem. Geol. 239, 266–283. http://dx.doi. Persson, I., D’Angelo, P., De Panfilis, S., Sandstrom, M., Eriksson, L., 2008. Hydration of lanthanoid(III)
org/10.1016/j.chemgeo.2006.10.004. ions in aqueous solution and crystalline hydrates studied by EXAFS spectroscopy and crystal-
Mayanovic, R.A., Anderson, A., Bassett, W., Chou, I.-M., 2009a. Steric hindrance and the enhanced lography: the myth of the “gadolinium break”. Chem. Eur. J. 14 (10), 3056–3066.
stability of light rare-earth elements in hydrothermal fluids. Am. Mineral. 94, 1487–1490. Petit, L., Vuilleumier, R., Maldivi, P., Adamo, C., 2008. Ab initio molecular dynamics study of a
http://dx.doi.org/10.2138/am.2009.3250. highly concentrated LiCl aqueous solution. J. Chem. Theory Comput. 4, 1040–1048.
Mayanovic, R.A., Anderson, A.J., Bassett, W.A., Chou, I.-M., 2009b. The structure and stability of Poitrasson, F., Oelkers, E., Schott, J., Montel, J.-M., 2004. Experimental determination of synthetic
aqueous rare-earth elements in hydrothermal fluids: New results on neodymium(III) NdPO4 monazite end-member solubility in water from 21 °C to 300 °C: implications for rare
aqua and chloroaqua complexes in aqueous solutions to 500 °C and 520 MPa. Chem. earth element mobility in crustal fluids. Geochim. Cosmochim. Acta 68, 2207–2221. http://
Geol. 259, 30–38. http://dx.doi.org/10.1016/j.chemgeo.2008.08.011. dx.doi.org/10.1016/j.gca.2003.12.010.
Mayanovic, R.A., Jayanetti, S., Anderson, A., Bassett, W.A., Chou, I.-M., 2002a. The Structure of Yb Popa, K., Konings, R.J.M., 2006. High-temperature heat capacities of EuPO4 and SmPO4 synthetic
3+ Aquo Ion and Chloro Complexes in Aqueous Solutions at Up to 500 °C and 270 MPa. monazites. Thermochim. Acta 445, 49–52. http://dx.doi.org/10.1016/j.tca.2006.03.023.
J. Phys. Chem. A 106, 6591–6599. http://dx.doi.org/10.1021/jp020140q. Pourtier, E., Devidal, J.-L., Gibert, F., 2010. Solubility measurements of synthetic neodymium
Mayanovic, R.A., Jayanetti, S., Anderson, A.J., Bassett, W.A., Chou, I.M., 2002b. The structure of Yb monazite as a function of temperature at 2 kbars, and aqueous neodymium speciation in
3+ and Y 3+ aquo ion and chloro complexes in hydrothermal solutions Constraining re- equilibrium with monazite. Geochim. Cosmochim. Acta 74, 1872–1891. http://dx.doi.org/
active transport models using mineralogical data. Goldschmidt Conf. Abstr. 2002 A496. 10.1016/j.gca.2009.12.023.
42 A. Migdisov et al. / Chemical Geology 439 (2016) 13–42

Ragnarsdottir, K.V., Oelkers, E.H., Sherman, D.M., Collins, C.R., 1998. Aqueous speciation of yttri- Stepanchikova, S.A., Biteikina, R.P., Shironosova, G.P., Kolonin, G.R., 2014. An experimental study
um at temperatures from 25 to 340 °C at Psat: an in situ EXAFS study. Chem. Geol. 151, of hydroxo complex formation in basic and near-neutral solutions of rare-earth elements
29–39. http://dx.doi.org/10.1016/S0009-2541(98)00068-0. and yttrium at 25 °C. Russ. Geol. Geophys. 55, 941–944.
Rakovan, J., Reeder, R., 1996. Intracrystalline rare earth element distribution in apatite - surface Sverjensky, D.A., 1984. Europium redox equilibria in aqueous solution. Earth Planet. Sci. Lett. 67
structural influences on incorporation during growth. Geochim. Cosmochim. Acta 60, (1), 70–78.
4435–4445. Sverjensky, D., Shock, E.L., Helgeson, H.C., 1997. Prediction of the thermodynamic properties of
Rakovan, J., Newville, M., Sutton, S., 2001. Evidence of heterovalent europium in zoned Llallagua aqueous metal complexes to 1000 °C and 5 kb. Geochim. Cosmochim. Acta 61, 1359–1412.
apatite using wavelength dispersive XANES. Am. Mineral. 86, 697–700. http://dx.doi.org/10.1016/S0016-7037(97)00009-4.
Robie, R.A., Hemingway, B.S., 1995. Thermodynamic properties of minerals and related sub- Tagirov, B.R., Zotov, A., Akinfiev, N., 1997. Experimental study of dissociation of HCl from 350 to
stances at 298.15 K and 1 Bar (105 Pascals) pressure and at higher temperatures. U.S. 500 °C and from 500 to 2500 bars: thermodynamic properties of HCl°(aq). Geochim.
Geol. Surv. Bull. 2131, 461. Cosmochim. Acta 61, 4267–4280. http://dx.doi.org/10.1016/S0016-7037(97)00274-3.
Ruberti, E., Enrich, G., Gomes, C., Comin-Chiaramonti, P., 2008. Hydrothermal REE Takahashi, Y., Kolonin, G.R., Shironosova, G.P., Kupriyanova, I.I., Uruga, T., Shimizu, H., 2005. De-
fluorocarbonate mineralization at Barra do Itapirapua, a multiple stockwork carbonatite, termination of the Eu(II)/Eu(III) ratios in minerals by X-ray absorption near-edge structure
Southern Brazil. Can. Mineral. 46, 901–914. (XANES) and its application to hydrothermal deposits. Mineral. Mag. 69, 179–190.
Rycerz, L., Gaune-Escard, M., 2008. Lanthanide(III) halides: thermodynamic properties and their Terrier, C., Vitorge, P., Gaigeot, M.P., Spezia, R., Vuilleumier, R., 2010. Density functional theory
correlation with crystal structure. J. Alloys Compd. 450, 167–174. http://dx.doi.org/10.1016/ based molecular dynamics study of hydration and electronic properties of aqueous
j.jallcom.2006.12.096. La(3+). J. Chem. Phys. 133, 044509.
Ryzhenko, B.N., Bryzgalin, O.V., Artamkina, I.Y., Spasennykh, M.Y., Shapkin, A.I., 1985. An electro- Testemale, D., Brugger, J., Liu, W., Etschmann, B., Hazemann, J., 2009. In-situ X-ray absorption
static model for the electrolytic dissociation of inorganic substances dissolved in water. study of iron(II) speciation in brines up to supercritical conditions. Chem. Geol. 264 (1-4),
Geochem. Int. 22, 138–144. 295–310.
Ryzhenko, B.N., Kovalenko, N.I., Mironenko, M.V., 1991. Ionization-constant of hydrofluoric acid Trofanenko, J., Williams-Jones, A.E., Simandl, G.J., Migdisov, A.A., 2016. The nature and origin of
at 500 °C, 1 kbar. Dokl. Akad. Nauk SSSR 317, 203–206. the REE mineralization in the Wicheeda carbonatite, British Columbia, Canada. Econ. Geol.
Salvi, S., Williams-Jones, A.E., 1996. The role of hydrothermal processes in concentrating high- 111 (1), 199–223.
field strength elements in the Strange Lake peralkaline complex, northeastern Canada. Um, N., Miyake, M., Hirato, T., 2011. Dissolution of cerium oxide in sulfuric acid. In: Yao, T. (Ed.),
Geochim. Cosmochim. Acta 60, 1917–1932. http://dx.doi.org/10.1016/0016- Zero-Carbon Energy Kyoto 2010 - Proceedings of the Second International Symposium of
7037(96)00071-3. Global COE Program “Energy Science in the Age of Global Warming—Toward CO2 Zero-
Schijf, J., Byrne, R.H., 1999. Determination of stability constants for the mono- and difluoro- Emission Energy System. Springer, Tokyo, Berlin, Heidelberg, New York, pp. 165–170.
complexes of Y and the REE, using a cation-exchange resin and ICP-MS. Polyhedron 18, van Hinsberg, V.J., Migdisov, A.A., Williams-Jones, A.E., 2010. Reading the mineral record of fluid
2839–2844. http://dx.doi.org/10.1016/S0277-5387(99)00205-3. composition from element partitioning. Geology 38, 847–850. http://dx.doi.org/10.1130/
Schijf, J., Byrne, R.H., 2004. Determination of SO4β1 for yttrium and the rare earth elements at G31112.1.
I = 0.66 m and t = 25 °C—implications for YREE solution speciation in sulfate-rich waters1. van Wambeke, L., 1977. The Karonge rare earth deposits, Republic of Burundi- new
Geochim. Cosmochim. Acta 68, 2825–2837. http://dx.doi.org/10.1016/j.gca.2003.12.003. mineralogical-geochemical data and origin of the mineralization. Mineral Deposits 12,
Seward, T.M., Williams-Jones, A., Migdisov, A., 2014. The Chemistry of Metal Transport and De- 373–380.
position by Ore-Forming Hydrothermal Fluids. In: Holland, H.D., Turekian, K.K. (Eds.), Trea- Vasil’ev, V.P., Kozlovskii, E.V., 1977. Solubility products of some rare earth element fluorides.
tise on Geochemistry, second ed. Elsevier Ltd., Oxford, pp. 29–57 http://dx.doi.org/10.1016/ Russ. J. Inorg. Chem. 22 (3), 853–856.
B978-0-08-095975-7.01102-5. Wagner, L.R., Schottky, V., 1930. Theory of ordered mixed phases. Z. Phys. Chem. 11B, 163–210.
Shannon, R.D., 1976. Revised effective ionic radii and systematic studies of interatomic distances Wang, D., Yang, J., Yan, S., Xu, J., Chen, Y., Pu, G., Luo, Y., 2001. A special orogenic-type rare earth
in halides and chalcogenides. Acta Crystallogr. A32, 751–767. element deposit in Maoniuping, Sichuan, China: geology and geochemistry. Resour. Geol.
Sherman, D.M., 2010. Metal complexation and ion association in hydrothermal fluids: insights 51, 177–188.
from quantum chemistry and molecular dynamics. Geofluids 10, 41–57. Weng, Z., Jowitt, S.M., Mudd, G.M., Haque, N., 2015. A detailed assessment of global rare earth
Shock, E.L., Helgeson, H.C., 1988. Calculation of the thermodynamic and transport properties of element resources- opportunities and challenges. Econ. Geol. 110, 1925–1952.
aqueous species at high pressures and temperatures: correlation algorithms for ionic spe- Williams-Jones, A.E., Wood, S.A., 1992. A preliminary petrogenetic grid for rare earth element
cies and equation of state predictions to 5 kb and 1000 °C. Geochim. Cosmochim. Acta fluorocarbonate and related minerals. Geochim. Cosmochim. Acta 56, 725–738.
52, 2009–2036. http://dx.doi.org/10.1016/0016-7037(88)90181-0. Williams-Jones, A.E., Migdisov, A.A., Samson, I.M., 2012. Hydrothermal mobilisation of the rare
Shock, E.L., Sassani, D.C., Willis, M., Sverjensky, D.A., 1997. Inorganic species in geologic fluids: earth elements – a tale of “ ceria ” and “ yttria. Elements 8, 355–360. http://dx.doi.org/10.
correlations among standard molal thermodynamic properties of aqueous ions and hydrox- 2113/gselements.8.5.355.
ide complexes. Geochim. Cosmochim. Acta 61, 907–950. Williams-Jones, A.E., Samson, I.M., Olivo, G.R., 2000. The genesis of hydrothermal fluorite-REE
Smirnov, P.R., Grechin, O.V., 2013. Coordination of ions in aqueous solutions of erbium chloride deposits in the Gallinas Mountains, New Mexico. Econ. Geol. 95, 327–341.
from X-ray diffraction data. Russ. J. Coord. Chem. 39, 685–688. http://dx.doi.org/10.1134/ Wood, S.A., 1990a. The aqueous geochemistry of the rare-earth elements and yttrium 2. Theoretical
S1070328413090078. predictions of speciation in hydrothermal solutions to 350 °C at saturation water vapor
Smith, M., Henderson, P., 2000. Preliminary fluid inclusion constraints on fluid evolution in the pressure. Chem. Geol. 88, 99–125. http://dx.doi.org/10.1016/0009-2541(90)90106-H.
Bayan Obo Fe-REE-Nb deposit, Inner Mongolia, China. Econ. Geol. 95, 1371. http://dx.doi. Wood, S.A., 1990b. The aqueous geochemistry of the rare-earth elements and yttrium. 1. Review
org/10.2113/95.7.1371. of available low-temperature data for inorganic complexes and the inorganic REE specia-
Smith, M., Henderson, P., Campbell, L., 2000. Fractionation of the REE during hydrothermal processes: tion of natural waters. Chem. Geol. 82, 159–186. http://dx.doi.org/10.1016/0009-
constraints from the Bayan Obo Fe-REE-Nb deposit, Inner Mongolia, China. Geochim. 2541(90)90080-Q.
Cosmochim. Acta 64, 3141–3160. http://dx.doi.org/10.1016/S0016-7037(00)00416-6. Wood, S., Palmer, D., Wesolowski, D., Bénézeth, P., 2002. The aqueous geochemistry of the rare
Soderholm, L., Skanthakumar, S., Wilson, R.E., 2009. Structures and energetics of erbium chlo- earth elements and yttrium. Part XI. The solubility of Nd (OH) 3 and hydrolysis of Nd3+
ride complexes in aqueous solution. J. Phys. Chem. A 113, 6391–6397. http://dx.doi.org/ from 30 to 290 °C at saturated water vapor pressure with in-situ pHm measurement. In:
10.1021/jp9012366. Hellmann, R., Wood, S.A. (Eds.), Water-Rock Interactions, Ore Deposits, and Environmental
Spahiu, K., Bruno, J., 1995. A selected thermodynamic database for REE to be used in HLNW per- Geochemistry: A Tribute to David A. Crerar, pp. 229–256 http://dx.doi.org/10.1016/0009-
formance assessment exercises. 2541(90)90080-Q.
Stepanchikova, S., Kolonin, G., 2005. Spectrophotometric study of Nd, Sm, and Ho complexation Xu, C., Kynicky, J., Chakhmouradian, A.R., Campbell, I.H., Allen, C.M., 2010. Trace-element model-
in chloride solutions at 100–250 °C. Russ. J. Coord. 31, 193–202. http://dx.doi.org/10.1007/ ing of the magmatic evolution of rare-earth-rich carbonatite from the Miaoya deposit, Cen-
s11173-005-0076-4. tral China. Lithos 118, 145–155. http://dx.doi.org/10.1016/j.lithos.2010.04.003.

You might also like