Russian Geology and Geophysics © 2023, Novosibirsk State University

Vol. 64, No. 8, pp. 910–918, 2023       doi:10.2113/RGG20234561 Geologiya i Geofizika

Carbon Isotope Fractionation during Metal–Carbonate Interaction

at the Mantle Pressures and Temperatures
V.N. Reutskya, , Yu.M. Borzdova, Yu.V. Batalevaa, Yu.N. Palyanova,b
a
V.S. Sobolev Institute of Geology and Mineralogy, Siberian Branch of the Russian Academy of Sciences,
pr. Akademika Koptyuga 3, Novosibirsk, 630090, Russia
b
Novosibirsk State University, ul. Pirogova 1, Novosibirsk, 630090, Russia

Received 13 January 2023; accepted 22 February 2023

Abstract––Subduction of marine carbonates is accompanied by numerous transformations and interactions, including reactions with
reduced mantle rocks. At depths of 250–300 km, carbonates enter mantle zones where metallic iron can be stable. The interaction of car-
bonates with metals is one of the mechanisms of the release of elemental carbon and the formation of diamond. These processes are also
accompanied by carbon isotope fractionation and can result in a significant isotopic heterogeneity of mantle carbon. In this work we study
the partitioning of carbon isotopes between carbon and carbon-bearing phases obtained in experiments on the interaction of FeNi alloy with
(Mg,Ca)CO3, which simulates mantle–crust redox reactions in the temperature range 800–1550 °C and at a pressure of 6.3 GPa. It has been
established that at 800–1000 °C, the carbon of carbonate is reduced at the metal/carbonate interface and dissolves in the FeNi alloy. This
process leads to a 17–20‰ depletion of the metal in the heavy carbon isotope. At temperatures above 1330 °C, the fractionation of carbon
isotopes between carbonate and metal–carbon melts is reduced to 8.5‰, approaching the thermodynamic calcite–cohenite isotope equi-
librium. At temperatures above 1400 °C, diamond crystallizes from metal–carbon and carbonate melts, which leads to isotopic depletion
of the metal–carbon melt. As a result, the measured carbon isotope fractionation between the carbonate and metal–carbon melts increases
and moves away from the thermodynamic CaCO3–Fe3C equilibrium line. The carbonate–metal redox interaction is supposed to be one
of the probable mechanisms of the formation of isotopically light carbon in the mantle at the expense of the marine carbonate sediments
subducted into the mantle. This mechanism also provides the formation of anomalous isotopically heavy carbonates found in kimberlites
of the Siberian Platform.

Keywords: diamond; graphite; carbonate melt; metal–carbon melt; redox interaction; carbon isotopes; fractionation; experiment; high pressure; high tem-
perature

INTRODUCTION isotope fractionation during the interaction of metallic iron

with carbonates in the mantle P–T conditions permit consid-
The interaction of metals with carbonates is one of the ering the redox interaction to be the possible cause of sig-
mechanisms of the release of elemental carbon in the Earth’s nificant variations in the isotope ratios of the Earth’s mantle
mantle (Rohrbach and Schmidt, 2011). This process is ac- carbon. The isotopically light iron carbides resulted from the
companied by the formation of mantle minerals and carbon metal–carbonate interaction can further react with the man-
isotope fractionation (Palyanov et al., 2013; Luth et al., tle fluids or melts (Bataleva et al., 2017), releasing isotopi-
2022). It was experimentally shown that the interaction of cally light carbon and forming isotopically light minerals.
alkaline-earth car­bonates with metallic iron at high tempera- The self-diffusion of carbon isotopes in diamond is of ex-
tures and high pressures results in graphite/diamond and tremely low efficiency (Koga et al., 2003) and does not pro-
iron carbide crystallization (Palyanov et al., 2013). Owing to vide equilibration of the composition of neighboring growth
carbon isotope fractionation during this interaction, Fe3C zones several hundred microns in size in crystals that have
produced in the experiments at 1450–1550 °C and 6.5 GPa been in the Earth’s mantle for hundreds of thousands and
is, on average, 6.5‰ isotopically lighter than (Ca,Mg,Fe) millions of years (Harte et al., 1999, Schulze et al., 2004). If
CO3. Calculations of the reduced partition function ratios isotopically heavy diamonds crystallize together with car-
(β-factors) show that the equilibrium fractionation of carbon bides during redox interaction, then such diamonds keep
isotopes between СaCO3 and Fe3C at 1500 °С is 6.4‰ (Hor- their isotopic­composition unchanged. This might provide
ita and Polyakov, 2015). The established values of carbon the formation of carbon reservoirs of different isotopic com-
positions in the Earth’s mantle, whose existence is widely

Corresponding author. discussed in the scientific literature (Deines, 2004; Stachel
  E-mail address: [email protected] (V.N. Reutsky) and Harris, 2009; Cartigny, 2010). At the same time, the

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918 911

presence of impurities of other transition metals, such as site (Mg0.9Ca0.1CO3). The sample was placed in a platinum
nickel or cobalt, in iron hinders the formation of iron car- ampoule of relatively large volume. Figure 1a shows a
bides. The Fe–Ni–C system is considered to be the main me- scheme of the experimental ampoule assembly. The “sand­
tallic component of the mantle at depths of 250–300 km wich”-like initial sample assembly favors the formation of
(Rohrbach et al., 2014). Nickel not only prevents the forma- alternating reaction zones in the samples, which include
tion of iron carbides but also decreases the Fe–C eutectic newly formed carbon and carbon-bearing phases in various
temperature and reduces the solubility of carbon in the metal associations (Fig. 1b, c).
melt. The subducting metasomatized crust has low Ni con- The relatively large volume of the ampoules and the large
tents, but under cold-subduction conditions, carbonate-bear- amounts of initial materials made it possible to pick carbon
ing slabs can reach depths of 250 km or more, where the and carbon-bearing phases from different zones in amounts
nickel content is higher. The absence of iron carbides can sufficient for the analysis of the carbon isotope composition.
reduce the magnitude of carbon isotope fractionation three or The metal with alloying carbon from the central parts of the
four times, to 1.5–2.0‰, which is typical of fractionation be- ampoules was mechanically crushed and sampled under a
tween diamond/graphite and carbonate/carbonate fluid binocular. For separating diamond from the FeNi alloy, the
(Chacko et al., 1991; Deines and Eggler, 2009; Reutsky et samples were dissolved in a mixture of dilute acids (HCl
al., 2015). In this work we experimentally studied the frac- and HNO3) on boiling. Hot dilute hydrochloric acid was
tionation of carbon isotopes during the interaction of FeNi used to dissolve carbonate, and dilute nitric acid, to dissolve
alloy with Ca-magnesite at 800–1550 °C and 6.3 GPa, i.e., magnesiowüstite. Graphite was removed by heating the
under subduction conditions in which this interaction is most samples in the hot oxidizing mixture. Carbonate was
likely. crushed, and fragments free of graphite and diamond were
sampled under a binocular.
The carbon isotope composition of the experimentally
METHODS AND MATERIALS
obtained phases was analyzed at the Analytical Center for
Experiments on the reaction of FeNi alloy with (Mg,Ca) Multi-Elemental and Isotope Research SB RAS, Novosi-
CO3 at temperatures of 800 to 1550 ºC, a pressure of birsk, and at GFZ Potsdam. The isotope characteristics of
6.3 GPa, and duration of 35–105 h were carried out at the carbonate, graphite, diamond, and alloying carbon were
Institute of Geology and Mineralogy SB RAS, Novosibirsk, studied at the Institute of Geology and Mineralogy SB RAS,
using a multianvil high-pressure split-sphere apparatus Novosibirsk. For carbon isotope analysis of carbonates, a
(BARS). The technical details of these experiments are de- weighted sample was heated in a vacuum to 950 ºC; at the
scribed in Palyanov et al. (2023). The initial carbonate, the same time, CO2 produced during the thermal decomposition
only source of carbon in the system, had a carbon isotope of carbonate was collected in a trap cooled with liquid nitro-
composition of –0.2‰ VPDB. After the experiments, the gen. Carbon isotope ratios were measured using a Delta V
carbon isotope composition of the carbonate, newly formed Advantage gas mass spectrometer in the dual-inlet mode.
carbon (graphite, diamond), and carbon-bearing phases (al- The procedures were controlled using the isotope standards
loying carbon in solid metal, metal–carbon and carbonate NBS-18 (calcite, δ13C = –5.014‰ VPDB) and NBS-19
melts) was studied. (limestone, δ13C = +1.95‰ VPDB) prepared in the same
Experiments on the metal–carbonate interaction were way. Analysis of the carbon isotope composition of the oth-
carried out using the “sandwich” method elaborated by er phases was carried out after their thermal oxidation at
Palyanov et al. (2013). The sample for study comprised a 950 ºC in a vacuum reactor in the presence of vacuum-puri-
cylinder pressed from Fe and Ni powders (of bulk composi- fied copper oxide. The procedures were controlled using the
tion Fe90Ni10) and enclosed in a capsule made of Ca-magne- isotope standards USGS-24 (graphite, δ13C = –16.049‰

Fig. 1. Scheme of the ampoule assembly (a) in “sandwich” experiments on the metal–carbonate interaction and schemes of the interaction at the
lowest (b) and highest (c) temperatures. Dm – diamond, Gr – graphite.

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 912 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918

VPDB), GR-770 (graphite, δ13C = –24.65‰ VPDB), and Diamond and graphite crystals from some experiments
RM18 (graphite, δ13C = –7.01‰ VPDB). The reproducibil- were analyzed using both SIMS and the bulk-oxidation
ity of isotope analysis of the standards is better than 0.1‰ methods. The results of the two methods reveal a good
(2σ). The analytical error of single measurements does not agreement, which permits their direct comparison. At the
exceed 0.02‰. All isotope data are given in the VPDB same time, the high spatial resolution of SIMS made it pos-
(Vien­na Pee Dee Belemnite) scale. sible to study the variation in the carbon isotope composi-
At GFZ Potsdam, the carbon isotope composition of dia- tion in the direction of the growth of large diamond crystals
mond and graphite crystals was determined mostly for the from the metal zone. Before the analysis of the carbon iso-
experiments in which bulk oxidation of graphite and diamond tope composition, the crystals were polished to expose their
were impossible because of their small amounts. The crystals growth zones to a single plane. The results of carbon isotope
were studied on a Cameca IMS 1280-HR secondary-ion mass analysis of all phases after the experiments are presented in
spectrometer (SIMS) in the multicollector mode. Some of the Table 1. Since the phase formation varies in different tem-
experimentally obtained crystals and polished diamond plates perature intervals, it is worth considering the obtained iso-
were individually mounted in an indium holder and coated tope data for each interval.
with a conducting gold layer. Correction for instrumental 800–1000 °C. The isotopic composition of alloying car-
fractionation was made by parallel measurements for graph- bon depends on the experimental temperature, varying from
ite RM18 (δ13C = –7.01‰ VPDB) and diamond DM 150 –20.0‰ at 800 °C to –17.6‰ at 1000 °C. The bulk isotopic
(δ13C = –24.5‰ VPDB) (Reutsky et al., 2017). composition of carbonate carbon changes insignificantly in
these experiments, remaining close to the δ13C value of the
RESULTS initial carbonate. Thus, the difference in the δ13C value bet­
ween alloying carbon and carbonate becomes smaller with
A detailed analysis of phase formation and reconstruction
increasing temperature (Fig. 3). The δ13C of graphite from
of interaction processes in the system (FeNi)–(Mg,Ca)CO3
the magnesiowüstite reaction zone varies from –2.2 to
are presented in Palyanov et al. (2023). In this study we fo-
–11.3‰ and shows no correlation with the experimental
cus only on the main aspects of the phase formation and
sample structure. At 800–1000 ºC, a narrow reaction zone temperature. Graphite always has an intermediate carbon
consisting of magnesiowüstite ((Fe,Mg)O) and graphite isotope composition: It is isotopically lighter than carbonate
forms in the samples at the contact of the metallic cylinder but isotopically heavier than alloying carbon.
and carbonate (Fig. 2a–c). No melting of the sample occurs 1200–1330 °C. In the temperature range 1200–1330 °C,
in this temperature range. During the interaction, a solution the δ13C value of alloying carbon continues increasing from
of carbon in the metal forms. At higher temperatures –12.0‰ at 1200 °C to –9.0‰ at 1330 °C. The carbon iso-
(≥1200 ºС), the formation of a reaction zone of magnesio- tope composition of carbonate is –0.3 to –0.5‰. Graphite
wüstite and graphite is accompanied by the metal melting from the magnesiowüstite reaction zone is more steady in
(with the formation of FeNi–C melt); at 1330 ºC, carbonate isotopic composition as compared with the low-temperature
melting begins (Fig. 2d, e). In the experiments at 1400– experiments, δ13C = –6.0 and –5.6‰, respectively.
1550 ºC, the samples contain both metal–carbon and car- 1400–1550 °C. The isotopic composition of alloying car-
bonate melts, in which the formation of graphite and the bon in the experiments at 1470 and 1550 °C is –9.2 ± 1.0‰,
spontaneous crystallization of diamond are observed which is consistent with the δ13С of alloying carbon in the
(Fig. 2f–i). In all experiments, the carbon isotope composi- experiment at 1330 °С (–9.0‰). At 1400 °C, alloying car-
tion of carbonate, graphite, diamond (if present), and alloy- bon is slightly enriched in the heavy isotope (δ13С = –5.6‰),
ing carbon was analyzed. which, however, might be due to the possible presence of

Table 1. Results of isotopic analysis (‰, VPDB) of carbon phases obtained in the experiments on the metal–carbonate interaction at 6.3 GPa
Run Т, °С Metal zone Magnesiowüstite + graphite zone Carbonate zone
Alloying carbon Diamond Graphite Carbonate Graphite Diamond

1099 1550 –9.3 –2.7 to –4.7 –2.9 +0.1 – –1.8

1103 1470 –9.1 – –2.3 –0.6 –15.1 –5.3
1104 1400 –5.6 –1.4 to –2.9 –2.2 +0.2 –11.1 –2.0
–6.6
–9.5
1106 1330 –9.0 – –6.0 –0.5 – –
1024 1200 –12.0 – –5.6 –0.3 – –
1025 1000 –17.6 – –10.0 –0.3 – –
1090 900 –21.2 – –2.4 –1.1 – –
1089 800 –20.0 – –11.3 –0.5 – –

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918 913

Fig. 2. SEM microphotographs of fragments of the samples obtained in “sandwich” experiments in the system (FeNi)–(Mg,Са)CO3 at 6.3 GPa.
The relations among the carbon, carbon-bearing, and oxide phases in different reaction zones at 900 ºC (a, b), 1330 ºC (c), 1400 ºC (d, e), 1470 ºC
(f), and 1550 ºC (g–i). FeNi–C – metal with alloying carbon, Mws – magnesiowüstite (Mg,Fe)O, Ms – magnesite MgCO3, Dol – dolomite
CaMg(CO3)2, Arg – aragonite CaCO3, Carbmelt – carbonate melt, FeNi–Cmelt – metal–carbon melt, Gr – graphite, and Dm – diamond.

13
С–enriched diamond microcrystals in the analyzed mate- δ13С = +0.1 to +0.2‰. Graphite from the carbonate zone
rial (Table 1). Graphite from the magnesiowüstite reaction shows wide variations in δ13С (from –5 to –15‰) even
zone in these experiments is steady in the carbon isotopic within the same experiment. The δ13С value of diamond
composition, and its δ13С slightly decreases from –2.2 to from the carbonate zone also varies significantly, from
–2.9‰ with increasing temperature. Diamonds crystallized –1.8 to –5.3‰.
from the metal–carbon melt have average δ13С values of Diamond crystals grown in the metal zone reach 1.7 mm
–2.0‰ at 1400 °С and –3.3‰ at 1550 °С. in size, which made it possible to perform a thorough SIMS
The carbon isotope composition of the carbonate melt study of the δ13С change in the direction of their growth.
remains close to the initial value, demonstrating a tenden- The observed gradual 2.1‰ (Run 1099) and 1.5‰ (Run
cy toward its enrichment with the heavy isotope, up to 1104) decrease in δ13С in the growth direction is above the

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 914 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918

Fig. 3. Temperature dependence of the measured carbon isotope compositions in experiments on the FeNi–(Mg,Ca)CO3 interaction.

measurement accuracy. In the carbonate zone, the small size tally obtained δ13С values was 21.4‰, with the initial car-
of diamond crystals did not permit a thorough study of their bonate with δ13С = –0.2‰ VPDB being a single source of
carbon isotope composition (Fig. 4). carbon. The magnitude of equilibrium isotope fractionation
usually decreases with increasing temperature (Bottinga,
1969; Richet et al., 1977). This is because the β-factors of
DISCUSSION chemical compounds decrease with increasing temperature
and reach unity at an infinite temperature, which means no
The obtained isotope data unambiguously indicate that isotope fractionation among them (Bigeleisen and Mayer,
the redox interaction of metals with carbonate results in iso- 1947). The decrease in isotope fractionation between carbon
tope fractionation of carbon. The total range of experimen- and carbon-bearing phases with increasing temperature in

Fig. 4. Carbon isotope composition of diamonds from experi-

ments 1099 (1550 ºC) (a, c) and 1104 (1400 ºC) (b, d). a, b –
Diamond crystals from carbonate melt; c, d – diamond crystals
from metal–carbon melt. White dashed arrows show the crystal
growth directions.

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918 915

our experiments follows this regularity. Earlier, we estab- Without the active mass transport in the reaction volume,
lished similar dependences for carbon isotope fractionation metal interacts only with the part of carbonate that is in di-
in experiments on diamond growth in various model dia- rect contact with it. Carbon released during the interaction
mond-forming media (Reutsky et al., 2015). Among the had the same isotopic composition as carbon of the initial
known carbon compounds, metal carbides (FexCy, SiC) are carbonate and gradually diffused into the solid metal. Reach-
maximally depleted in the heavy carbon isotope because of ing a certain degree of saturation, it crystallized as graphite
their extremely low β-factors (Horita and Polyakov, 2015). within the reaction zone. The huge magnitude of isotope
This is confirmed by the systematics of the carbon isotope effects­ observed in the experiments at 800–1000 °C
compositions of natural minerals associated with carbides (δ13Сcarbonate–δ13СFeNi–C reaches 20‰) might result from the
(Trumbull et al., 2009; Mikhail et al., 2014; Shatsky et al., kinetic isotope fractionation of carbon during its diffusion
2020) and the experimental data on the carbon isotope com- into the solid metal rather than from the above interaction.
position of iron carbides produced in the HTHP experiments The isotope exchange in solids is of limited efficiency. The
(Palyanov et al., 2013). In all cases, crystalline FexCy and directed transport of material is a powerful mechanism of
SiC are the isotopically lightest carbon-bearing minerals of elemental and isotope fractionation. This explains the ob-
the studied assemblages. served isotope effects, which are larger than the calculated
In this work, we successfully avoided the crystallization equilibrium fractionation of carbon isotopes between car-
of iron carbides by adding nickel, which also lowers the bonate and iron carbide (Fig. 5). During carbon diffusion in
melting point of the metal and the solubility of carbon in it. metallic iron at 1.5 GPa and 1000 °C, the fractionation of
According to our data, alloying carbon in FeNi–metal is the carbon isotopes exceeds 20‰ (Mueller et al., 2014). This is
isotopically lightest in all experiments. On quenching, this close to the values established in our experiments at 800–
carbon forms flakes and globules with a low-ordered graph- 900 °C. By analogy, we assume that the different rates of
ite-like structure (Reutsky et al., 2008; Palyanov et al., diffusion of the heavy and light carbon isotopes in the solid
2023). In the molten metal, alloying carbon can form vari- metal in our low-temperature experiments might affect the
ous complexes with iron and nickel atoms. The equilibrium isotope fractionation. The wide variations in the δ13C values
isotope fractionation among compounds is controlled by of the newly formed graphite might also be due to the ki-
their β-factors, which depend on their structure. Therefore, netic isotope effects at 800–1000 °C. In case of graphite
the β-factor of alloying carbon can be similar to that of car- crystallization beyond the field of its thermodynamic stabil-
bides, i.e., be significantly smaller than the β-factor of other, ity, the fractionation is determined by kinetic rather than
more oxidized, carbon-bearing compounds. equilibrium isotope effects. Together with limited isotope
Carbonate with δ13С = –0.2‰ VPDB was the only source exchange, this might lead to the observed inhomogeneity of
of carbon in our experiments. The minor changes in the car- carbon isotope composition of graphite.
bon isotope composition of carbonate after the experiments The significant fractionation of carbon isotopes in our
are explained by several factors. In all experiments, the low-temperature experiments might also be due to the for-
amount of carbon in carbonate significantly exceeds the mation of a fluid phase capable of dissolving carbonate and
amount of all other carbon and carbon-bearing phases. The transporting it within the reaction volume. Light isotopo-
reduction of carbon during the interaction of carbonate with
metal, especially at temperatures insufficient for carbonate
melting, occurs directly at the carbonate–metal interface.
Self-diffusion of carbon isotopes in crystalline carbonate
does not provide carbon isotope re-equilibration even at
high experimental temperatures (Labotka et al., 2000). This
thesis is confirmed by the fact that in the experiments at
temperatures below 1400 °C, in which no carbonate melting
was observed, the δ13C value of carbonate does not differ
from that of the initial carbonate. In the experiments with
the interaction of carbonate and metal melts at ≥1400 °С, the
remained carbonate is enriched in the heavy carbon isotope
relative to the initial one (Table 1).
800–1000 °C. At low temperatures, local redox reactions
appear at the carbonate–metal interface to produce elemen-
tal carbon and magnesiowüstite. This process is considered
in detail by Palyanov et al. (2023) and can be schematically
expressed as
Fig. 5. Temperature dependence of carbon isotope fractionation bet­
(Mg,Ca)CO3 + (FeNi)metal → (Fe,Mg)O + C0graphite + ween carbonate and alloying carbon in metal. The circles mark the
measured values, and the line corresponds to the thermodynamic isoto-
(NiFe–С)metal + (Ca,Mg)CO3. pic calcite–cohenite equilibrium (Horita and Polyakov, 2015).

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 916 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918

logues of carbonate more readily enter into reactions and the alloying carbon isotopically lighter (relative to carbon-
easier dissolve in a limited volume of fluid than carbonate ate) than it must be at 1400–1550 °C in isotope equilibrium
with the heavy carbon isotope (Kueter et al., 2020). Thus, (Fig. 5).
the fluid can also provide enrichment of alloying carbon The diamond that crystallized in the metal zone is isoto-
with the light carbon isotope at temperatures below the pically lighter than the diamond that formed in the carbonate
melting point of carbonate. It is impossible to estimate the zone in the same experiment (Table 1). At the same time,
extent of this effect due to lack of data. In our low-tempera- both diamonds are highly heterogeneous in carbon isotope
ture experiments, both mechanisms might have contributed. composition, and the ranges of their δ13С variations gener-
1200–1330 °C. In the temperature range 1200–1330 °C, ally overlap. The kinetic isotope effects also significantly
the molten metal reacted with solid carbonate. The differ- contribute to this heterogeneity.
ence in δ13С values between carbonate and alloying carbon Possible causes of the high degree of carbon isotope
is reduced with increasing temperature, continuing the trend fractionation in our experiments. The higher values of
from the lower-temperature field, and approaches the cal- carbon isotope fractionation between carbonate and alloying
cite–cohenite isotope equilibrium line (Fig. 5). The carbon carbon, as compared with the equilibrium values, might
isotope composition of graphite was 5.3–5.4‰ lighter than partly be due to different β-factors of calcium and magne-
that of carbonate, which is approximately twice higher than sium carbonates. At 800 °C, magnesite is ca. 1‰ and dolo-
the thermodynamic equilibrium values (Chacko et al., 1991). mite is 0.3‰ isotopically heavier than calcite (Deines,
At the experimental temperatures, graphite is a metastable 2004), which increases the carbon isotope fractionation bet­
phase, and many factors affect its isotopic composition. ween cohenite and the corresponding carbonate. However, it
Nevertheless, the significantly smaller scatter of the δ13С is obvious that the carbonate composition effect is negligible
values of graphite indicates more effective carbon isotope to be the only cause of the observed difference in isotope
exchange in the system. In particular, diffusive fractionation fractionation.
of carbon isotopes is unlikely in the molten metal. The possible influence of the diffusion of carbon in a
1400–1550 °C. In the experiments at 1400–1550 °C, the solid metal on the isotope fractionation in the experiments at
metal and carbonate interacted in the molten state. The more 800–1000 °C is also of great interest. Mueller et al. (2014)
efficient isotope exchange provides stable δ13С values of the performed carbon diffusion experiments, using two blocks
alloying carbon and graphite from the magnesiowüstite re- of metallic iron with contrast carbon concentrations (11500
action zone. The crystalline phases formed from more isoto- and 150 mg/g). In our experiments, elemental carbon re-
pically homogeneous media. The significant enrichment of leased from carbonate during the redox reaction diffused
the metal melt in the heavy carbon isotope in the experiment into the solid metal. Thus, along with the carbon concentra-
at 1400 °C (Figs. 3 and 5) might be due to the presence of tion gradient, there is a boundary between different struc-
diamond microcrystals in the analyzed metal. In this experi- tures and chemical compositions in our experiments. The
ment, the diamond that crystallized from the metal–carbon presence of carbon in the metal after the experiments and
melt has a heavier isotopic composition of carbon (δ13С = the dependence of its isotopic composition on the experi-
–1.4 to –2.9‰) than the alloying carbon (Table 1). The dia- mental temperature indicate the successful transport of re-
mond crystallization might also have affected the fraction- leased carbon into solid FeNi alloy at 6.3 GPa. The observed
ation of carbon isotopes in the experiments at temperatures carbon isotope fractionation at 800–1000 °C significantly
above 1400 °C, because diamond removes the 13C isotope exceeds the equilibrium fractionation between CaCO3 and
from the metal–carbon melt and does not participate in iso- Fe3C (Horita and Polyakov, 2015) but is of the same scale
tope exchange after crystallization (Koga et al., 2003; Re- (20–30‰) as the diffusive fractionation of carbon isotopes
utsky et al., 2008). The SIMS-detected decrease in the δ13С in metallic iron (Mueller et al., 2014).
of diamond in the direction of growth (Fig. 4c, d) indicates Natural environments. During the subduction of car-
the progressive depletion of the metal-carbon melt in the bonate material into the Earth’s mantle, part of it inevitably
heavy carbon isotope. This is specific to diamond growth in reacts with reduced mantle rocks. The elemental carbon re-
metal–carbon systems, in which the preferable incorporation leased during this reaction has the carbon isotope composi-
of the heavy carbon isotope into the growing diamond leads tion of carbonate; if a fluid capable of dissolving carbonate
to a decrease in the δ13С of the alloying carbon and the fol- is generated, the released carbon might be slightly enriched
lowing formation of isotopically lighter diamond growth in the light carbon isotope. During the interaction without
zones (Reutsky et al., 2008). The selective fixation of the melting of rocks, this carbon diffuses into accessible crystal-
heavy carbon isotope by diamond from the metal–carbon line structures or forms its own crystalline phases. In the
melt, with the preserved bulk isotopic composition of car- case of fluid formation, the isotopically light (relative to car-
bonate, might be responsible for the higher values of carbon bonate) fluid can be removed from the interaction zone,
isotope fractionation between the alloying carbon and car- leaving the carbonate enriched in the heavy carbon isotope.
bonate as compared with the calcite–cohenite equilibrium Carbon is characterized by an extremely low solubility in
(Fig. 5). The depletion of the metal–carbon melt in the heavy silicate minerals (Keppler et al., 2003; Shcheka et al., 2006);
carbon isotope as a result of diamond crystallization makes therefore, in the presence of metals, carbon will accumulate

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918 917

in the metals. Such a directed distribution of carbon can be at a pressure of 6.3 GPa. It has been established that the re-
accompanied by significant isotope fractionation (carbon- dox reaction of initial carbonate, the only source of carbon,
ate–carbide isotope fractionation and diffusive fractionation yields various carbon and carbon-bearing phases: a solution
of carbon isotopes) and lead to the formation of local carbon of carbon in metal (alloying carbon), metastable graphite,
reservoirs with an isotopic composition from zero (δ13C of diamond, and metal–carbon and carbonate melts. The maxi-
marine carbonates) to –20‰ and lower (“organic matter” mum fractionation of carbon isotopes is reached under cold-
and the result of diffusive fractionation). With time, the car- subduction conditions at 800–900 °C, when carbon in the
bon isotope exchange leads to re-equilibration of the com- metal is depleted in the heavy isotope by more than 20‰. In
positions of neighboring minerals, but in the case of dia- general, the observed carbon isotope fractionation between
mond crystallization, the isotope ratio in the diamond will (Mg,Ca)CO3 and the formed carbon solution in FeNi melt is
remain unchanged. Thus, during a single geologic process, a well approximated by the thermodynamic calcite–cohenite
number of carbon-bearing minerals (including diamond) isotope equilibrium (Horita and Polyakov, 2015). The frac-
with contrasting isotope compositions can form from a sin- tionation does not change significantly for lack of crystalline
gle carbon source. The composition of the metal and the iron carbides, the formation of which is hindered by nickel
possibility of formation of iron carbides have little effect on impurity in the alloy. The above interaction is one of the
the extent of carbon isotope fractionation between carbonate most probable mechanisms of the formation of isotopically
and alloying carbon. light carbon in the mantle at the expense of the marine car-
During the melting, the fractionation of carbon isotopes bonate sediments subducted into the mantle. This mecha-
is limited by the redox interaction and crystallization of car- nism also ensures the formation of anomalous isotopically
bon phases. The interaction of metal and carbonate melts heavy carbonates found in kimberlites. In the cold-subduc-
leads to the crystallization of diamond in both the reduced tion zones, the carbon isotope fractionation can significantly
and oxidized parts of the system. The fractionation of car- exceed the thermodynamic equilibrium because of the dif-
bon isotopes between the melts corresponds to the fraction- fusive fractionation of carbon isotopes in solid state.
ation between iron carbide and carbonate, regardless of the
actual crystallization of carbides. Diamonds growing from
the metal–carbon melt are isotopically heavier than the ACKNOWLEDGEMENTS AND FUNDING
melt, and diamonds growing from the carbonate melt are
isotopically lighter than the carbonate. As a result, the aver- We are deeply grateful to the staff of the SIMS Labora-
age δ13С values of diamonds growing from the carbonate tory of the GFZ German Research Centre for Geosciences
and metal melts with contrasting carbon isotope ratios will (GFZ Potsdam), namely, Dr. Michael Wiedebeck, Dr. Fred-
differ slightly. Since isotope exchange in molten carbonate eric Couffignal, Dr. Alexander Rocholl, and Uwe Dittmann,
ensures bulk averaging of the isotopic composition of car- for providing the opportunity of high-spatial resolution anal-
bon, its interaction with metal and crystallization of dia- ysis of the carbon isotope composition of experimentally
mond can significantly enrich the carbonate with the heavy obtained graphite and diamond microcrystals.
carbon isotope. Kimberlites of the Udachnaya pipe contain This work was supported by grant 19-17-00075 from
carbonate autoliths abnormally enriched in the heavy car- the Russian Science Foundation, https://rscf.ru/proj-
bon isotope (δ13C of up to +31‰) (Galimov and Ukhanov, ect/19-17-00075/.
1989). Their morphology points to their magmatic xeno- Methods of stable isotope investigation and high-temper-
genic origin. The cause of the significant enrichment of ature/high-pressure experiments have been developed and
these carbonates in the heavy carbon isotope is still unclear. are supported on state assignment of IGM SB RAS
In our experiments, these autoliths might be residual melts (No. 122041400171-5 and No. 122041400159-3).
of subducted carbonate, whose carbon was fractionated
during the redox interaction with the reduced mantle mate- REFERENCES
rial, including metal melts. Many diamonds of the Siberian
Platform formed with the participation of subducted mate- Bataleva, Y.V., Palyanov, Y.N., Borzdov, Y.M., Bayukov, O.A., Zdro-
rial (Galimov, 1991; Shatsky et al., 2016; Zemnukhov et kov, E.V., 2017. Iron carbide as a source of carbon for graphite and
al., 2020). Inclusions of native metals and metal carbides diamond formation under lithospheric mantle P–T parameters. Lith-
os 286–287, 151–161, doi: 10.1016/j.lithos.2017.06.010.
were found in diamonds from the northeastern margin of Bigeleisen, J., Mayer, M.G., 1947. Calculation of equilibrium constants
the Siberian Platform (Shatsky et al., 2020). for isotopic exchange reactions. J. Chem. Phys. 15, 261–267, doi:
10.1063/1.1746492.
Bottinga, Y., 1969. Calculated fractionation factors for carbon and
CONCLUSIONS hydrogen isotope exchange in the system calcite–carbon dioxide–
graphite–methane–hydrogen–water vapor. Geochim. Cosmochim.
Acta 33, 49–64, doi: 10.1016/0016-7037(69)90092-1.
We obtained new experimental data on the interaction of
Cartigny, P., 2010. Mantle-related carbonados? Geochemical insights
FeNi alloy with (Mg,Ca)CO3, which simulates mantle–crust from diamonds from the Dachine komatiite (French Guiana). Earth
redox reactions in the temperature range 800–1550 °C and Planet. Sci. Lett. 296, 329–339, doi: 10.1016/j.epsl.2010.05.015.

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest
 918 V.N. Reutsky et al. / Russian Geology and Geophysics 64 (2023) 910–918

Chacko, T., Mayeda, T.K., Clayton, R.N., Goldsmith, J.R., 1991. Ox- interaction in the metal–carbonate system, conditions of crystalliza-
ygen and carbon isotope fractionations between CO2 and calcite. tion and indicator characteristics of diamond. Russ. Geol. Geophys.
Geochim. Cosmochim. Acta 55, 2867–2882, doi: 10.1016/0016- 64 (8), 892–909, doi: 10.2113.RGG20234559.
7037(91)90452-B. Reutsky, V.N., Borzdov, Yu.M., Palyanov, Yu.N., 2008. Carbon iso-
Deines, P., 2004. Carbon isotope effects in carbonate systems. Geochim. tope fractionation associated with HPHT crystallization of dia-
Cosmochim. Acta 68, 2659–2679, doi: 10.1016/j.gca.2003.12.002. mond. Diamond Relat. Mater. 17, 1986–1989, doi: 10.1016/j.dia-
Deines, P., Eggler, D.H., 2009. Experimental determination of carbon mond.2008.06.003.
isotope fractionation between CaCO3 and graphite. Geochim. Cos- Reutsky, V.N., Palyanov, Yu.N., Borzdov, Yu.M., Sokol, A.G., 2015.
mochim. Acta 73, 7256–7274, doi: 10.1016/j.gca.2009.09.005. Isotope fractionation of carbon during diamond crystallization
Galimov, E.M., 1991. Isotope fractionation related to kimberlite mag- in model systems. Russ. Geol. Geophys. 56 (1–2), 239–244, doi:
matism and diamond formation. Geochim. Cosmochim. Acta 55, 10.1016/j.rgg.2015.01.017.
1697–1708, doi: 10.1016/0016-7037(91)90140-Z. Reutsky, V.N., Shiryaev, A.A., Titkov, S.V., Wiedenbeck, M., Zudi-
Galimov, E.M., Ukhanov, A.V., 1989. The nature of the carbonate com- na, N.N., 2017. Evidence for large scale fractionation of carbon iso-
ponent of kimberlites. Geokhimiya, No. 3, 337–348. topes and of nitrogen impurity during crystallization of gem quality
Harte, B., Fitzsimons, I.C.W., Harris, J.W., Otter, M.L., 1999. Car- cubic diamonds from placers of North Yakutia. Geochem. Int. 55,
bon isotope ratios and nitrogen abundances in relation to cath- 988–999, doi: 10.1134/S001670291711009X.
odoluminescence characteristics for some diamonds from the Richet, P., Bottinga, Y., Javoy, M., 1977. A review of hydrogen, carbon,
Kaapvaal Province, S. Africa. Mineral. Mag. 63, 829–856, doi: nitrogen, oxygen, sulphur, and chlorine stable isotope enrichment
10.1180/002646199548961. among gaseous molecules. Annu. Rev. Earth Planet. Sci. 5, 65–110,
Horita, J., Polyakov, V.B., 2015. Carbon-bearing iron phases and the doi: 10.1146/annurev.ea.05.050177.000433.
carbon isotope composition of the deep Earth. PNAS 112 (1), 31–
Rohrbach, A., Schmidt, M.W., 2011. Redox freezing and melting in the
36, doi: 10.1073/pnas.1401782112.
Earth’s deep mantle resulting from carbon–iron redox coupling. Na-
Keppler, H., Wiedenbeck, M., Shcheka, S.S., 2003. Carbon solubility in
ture 472, 209–212, doi: 10.1038/nature09899.
olivine and the mode of carbon storage in the Earth’s mantle. Nature
Rohrbach, A., Ghosh, S., Schmidt, M.W., Wijbrans, C.H., Klemme, S.,
424, 414–416, doi: 10.1038/nature01828.
2014. The stability of Fe–Ni carbides in the Earth’s mantle: Evi-
Koga, K.T., Van Orman, J.A., Walter, M.J., 2003. Diffusive relaxation
of carbon and nitrogen isotope heterogeneity in diamond: a new dence for a low Fe–Ni–C melt fraction in the deep mantle. Earth
thermochronometer. Phys. Earth Planet. Inter. 139, 35–43, doi: Planet. Sci. Lett. 388, 211–221, doi: 10.1016/j.epsl.2013.12.007.
10.1016/S0031-9201(03)00141-9. Schulze, D.J., Harte, B., Valley, J.W., Channer, D.M.D., 2004. Evidence
Kueter, N., Schmidt, M.W., Lilley, M.D., Bernasconi, S.M., 2020. of subduction and crust–mantle mixing from a single diamond.
Kinetic carbon isotope fractionation links graphite and diamond Lithos 77, 349–358, doi: 10.1016/j.lithos.2004.04.022.
precipitation to reduced fluid sources. Earth Planet. Sci. Lett. 529, Shatsky, V.S., Zedgenizov, D.A., Ragozin, A.L., 2016. Evidence for a
115848, doi: 10.1016/j.epsl.2019.115848. subduction component in the diamond-bearing mantle of the Sibe-
Labotka, T.C., Cole, D.R., Riciputi, L.R., 2000. Diffusion of C and O rian craton. Russ. Geol. Geophys. 57 (1), 111–126, doi: 10.1016/j.
in calcite at 100 MPa. Am. Mineral. 85, 488–494, doi: 10.2138/am- rgg.2016.01.008.
2000-0410. Shatsky, V.S., Ragozin, A.L., Logvinova, A.M., Wirth, R., Ka-
Luth, R.W., Palyanov, Y.N., Bureau, H., 2022. Experimental petrology linina, V.V., Sobolev, N.V., 2020. Diamond-rich placer depos-
applied to natural diamond growth. Rev. Mineral. Geochem. 88, its from iron-saturated mantle beneath the northeastern margin
755–808, doi: 10.2138/rmg.2022.88.14. of the Siberian Craton. Lithos 364, 105514, doi: 10.1016/j.lith-
Mikhail, S., Guillermier, C., Franchi, I.A., Beard, A.D., Crispin, K., os.2020.105514.
Verchovsky, A.B., Jones, A.P., Milledge, H.J., 2014. Empirical evi- Shcheka, S.S., Wiedenbeck, M., Frost, D.J., Keppler, H., 2006. Carbon
dence for the fractionation of carbon isotopes between diamond and solubility in mantle minerals. Earth Planet. Sci. Lett. 245, 730–742,
iron carbide from the Earth’s mantle. Geochem. Geophys. Geosyst. doi: 10.1016/j.epsl.2006.03.036.
15, 855–866, doi: 10.1002/2013GC005138. Stachel, T., Harris, J.W., 2009. Formation of diamond in the Earth’s
Mueller, T., Watson, E.B., Trail, D., Wiedenbeck, M., Van Orman, J., mantle. J. Phys.: Condens. Matter 21, 364206, doi: 10.1088/0953-
Hauri, E.H., 2014. Diffusive fractionation of carbon isotopes in 8984/21/36/364206.
γ-Fe: Experiment, models and implications for early solar system Trumbull, R.B., Yang, J.S., Robinson, P.T., Di Pierro, S., Venne­mann, T.,
processes. Geochim. Cosmochim. Acta 127, 57–66, doi: 10.1016/j. Wiedenbeck, M., 2009. The carbon isotope composition of natural
gca.2013.11.014. SiC (moissanite) from the Earth’s mantle: New discoveries from
Palyanov, Y.N., Bataleva, Y.V., Sokol, A.G., Borzdov, Y.M., Kupriya- ophiolites. Lithos 113, 612–620, doi: 10.1016/j.lithos.2009.06.033.
nov, I.N., Reutsky, V.N., Sobolev, N.V., 2013. Mantle–slab inter- Zemnukhov, A.L., Reutsky, V.N., Zedgenizov, D.A., Ragozin, A.L.,
action and redox mechanism of diamond formation. PNAS 110, Zhelonkin, R.Y., Kalinina, V.V., 2020. Subduction related popula-
20408–20413, doi: 10.1073/pnas.1313340110. tion of diamonds in Yakutian placers, northeastern Siberian plat-
Palyanov, Yu.N., Bataleva, Yu.V., Borzdov, Yu.M., Kupriyanov, I.N., form. Cont­rib. Mineral. Petrol. 175, 98, doi: 10.1007/s00410-020-
Nechaev D.V., 2023. Experimental modeling of the mantle–slab 01741-w.

Downloaded from http://pubs.geoscienceworld.org/rgg/article-pdf/64/8/910/5934034/rgg_03_aug2023.pdf

by guest